The following changes since commit c52d69e7dbaaed0ffdef8125e79218672c30161d: Merge remote-tracking branch 'remotes/cschoenebeck/tags/pull-9p-20211027' into staging (2021-10-27 11:45:18 -0700) are available in the Git repository at: https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20211027 for you to fetch changes up to 820c025f0dcacf2f3c12735b1f162893fbfa7bc6: tcg/optimize: Propagate sign info for shifting (2021-10-27 17:11:23 -0700) ---------------------------------------------------------------- Improvements to qemu/int128 Fixes for 128/64 division. Cleanup tcg/optimize.c Optimize redundant sign extensions ---------------------------------------------------------------- Frédéric Pétrot (1): qemu/int128: Add int128_{not,xor} Luis Pires (4): host-utils: move checks out of divu128/divs128 host-utils: move udiv_qrnnd() to host-utils host-utils: add 128-bit quotient support to divu128/divs128 host-utils: add unit tests for divu128/divs128 Richard Henderson (51): tcg/optimize: Rename "mask" to "z_mask" tcg/optimize: Split out OptContext tcg/optimize: Remove do_default label tcg/optimize: Change tcg_opt_gen_{mov,movi} interface tcg/optimize: Move prev_mb into OptContext tcg/optimize: Split out init_arguments tcg/optimize: Split out copy_propagate tcg/optimize: Split out fold_call tcg/optimize: Drop nb_oargs, nb_iargs locals tcg/optimize: Change fail return for do_constant_folding_cond* tcg/optimize: Return true from tcg_opt_gen_{mov,movi} tcg/optimize: Split out finish_folding tcg/optimize: Use a boolean to avoid a mass of continues tcg/optimize: Split out fold_mb, fold_qemu_{ld,st} tcg/optimize: Split out fold_const{1,2} tcg/optimize: Split out fold_setcond2 tcg/optimize: Split out fold_brcond2 tcg/optimize: Split out fold_brcond tcg/optimize: Split out fold_setcond tcg/optimize: Split out fold_mulu2_i32 tcg/optimize: Split out fold_addsub2_i32 tcg/optimize: Split out fold_movcond tcg/optimize: Split out fold_extract2 tcg/optimize: Split out fold_extract, fold_sextract tcg/optimize: Split out fold_deposit tcg/optimize: Split out fold_count_zeros tcg/optimize: Split out fold_bswap tcg/optimize: Split out fold_dup, fold_dup2 tcg/optimize: Split out fold_mov tcg/optimize: Split out fold_xx_to_i tcg/optimize: Split out fold_xx_to_x tcg/optimize: Split out fold_xi_to_i tcg/optimize: Add type to OptContext tcg/optimize: Split out fold_to_not tcg/optimize: Split out fold_sub_to_neg tcg/optimize: Split out fold_xi_to_x tcg/optimize: Split out fold_ix_to_i tcg/optimize: Split out fold_masks tcg/optimize: Expand fold_mulu2_i32 to all 4-arg multiplies tcg/optimize: Expand fold_addsub2_i32 to 64-bit ops tcg/optimize: Sink commutative operand swapping into fold functions tcg/optimize: Stop forcing z_mask to "garbage" for 32-bit values tcg/optimize: Use fold_xx_to_i for orc tcg/optimize: Use fold_xi_to_x for mul tcg/optimize: Use fold_xi_to_x for div tcg/optimize: Use fold_xx_to_i for rem tcg/optimize: Optimize sign extensions tcg/optimize: Propagate sign info for logical operations tcg/optimize: Propagate sign info for setcond tcg/optimize: Propagate sign info for bit counting tcg/optimize: Propagate sign info for shifting include/fpu/softfloat-macros.h | 82 -- include/hw/clock.h | 5 +- include/qemu/host-utils.h | 121 +- include/qemu/int128.h | 20 + target/ppc/int_helper.c | 23 +- tcg/optimize.c | 2644 ++++++++++++++++++++++++---------------- tests/unit/test-div128.c | 197 +++ util/host-utils.c | 147 ++- tests/unit/meson.build | 1 + 9 files changed, 2053 insertions(+), 1187 deletions(-) create mode 100644 tests/unit/test-div128.c From: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr> Addition of not and xor on 128-bit integers. Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr> Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org> Message-Id: <20211025122818.168890-3-frederic.petrot@univ-grenoble-alpes.fr> [rth: Split out logical operations.] Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/qemu/int128.h | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/include/qemu/int128.h b/include/qemu/int128.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/int128.h +++ b/include/qemu/int128.h @@ -XXX,XX +XXX,XX @@ static inline Int128 int128_exts64(int64_t a) return a; } +static inline Int128 int128_not(Int128 a) +{ + return ~a; +} + static inline Int128 int128_and(Int128 a, Int128 b) { return a & b; @@ -XXX,XX +XXX,XX @@ static inline Int128 int128_or(Int128 a, Int128 b) return a | b; } +static inline Int128 int128_xor(Int128 a, Int128 b) +{ + return a ^ b; +} + static inline Int128 int128_rshift(Int128 a, int n) { return a >> n; @@ -XXX,XX +XXX,XX @@ static inline Int128 int128_exts64(int64_t a) return int128_make128(a, (a < 0) ? -1 : 0); } +static inline Int128 int128_not(Int128 a) +{ + return int128_make128(~a.lo, ~a.hi); +} + static inline Int128 int128_and(Int128 a, Int128 b) { return int128_make128(a.lo & b.lo, a.hi & b.hi); @@ -XXX,XX +XXX,XX @@ static inline Int128 int128_or(Int128 a, Int128 b) return int128_make128(a.lo | b.lo, a.hi | b.hi); } +static inline Int128 int128_xor(Int128 a, Int128 b) +{ + return int128_make128(a.lo ^ b.lo, a.hi ^ b.hi); +} + static inline Int128 int128_rshift(Int128 a, int n) { int64_t h; -- 2.25.1 From: Luis Pires <luis.pires@eldorado.org.br> In preparation for changing the divu128/divs128 implementations to allow for quotients larger than 64 bits, move the div-by-zero and overflow checks to the callers. Signed-off-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20211025191154.350831-2-luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/hw/clock.h | 5 +++-- include/qemu/host-utils.h | 34 ++++++++++++--------------------- target/ppc/int_helper.c | 14 +++++++++----- util/host-utils.c | 40 ++++++++++++++++++--------------------- 4 files changed, 42 insertions(+), 51 deletions(-) diff --git a/include/hw/clock.h b/include/hw/clock.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/clock.h +++ b/include/hw/clock.h @@ -XXX,XX +XXX,XX @@ static inline uint64_t clock_ns_to_ticks(const Clock *clk, uint64_t ns) return 0; } /* - * Ignore divu128() return value as we've caught div-by-zero and don't - * need different behaviour for overflow. + * BUG: when CONFIG_INT128 is not defined, the current implementation of + * divu128 does not return a valid truncated quotient, so the result will + * be wrong. */ divu128(&lo, &hi, clk->period); return lo; diff --git a/include/qemu/host-utils.h b/include/qemu/host-utils.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/host-utils.h +++ b/include/qemu/host-utils.h @@ -XXX,XX +XXX,XX @@ static inline uint64_t muldiv64(uint64_t a, uint32_t b, uint32_t c) return (__int128_t)a * b / c; } -static inline int divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) +static inline void divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) { - if (divisor == 0) { - return 1; - } else { - __uint128_t dividend = ((__uint128_t)*phigh << 64) | *plow; - __uint128_t result = dividend / divisor; - *plow = result; - *phigh = dividend % divisor; - return result > UINT64_MAX; - } + __uint128_t dividend = ((__uint128_t)*phigh << 64) | *plow; + __uint128_t result = dividend / divisor; + *plow = result; + *phigh = dividend % divisor; } -static inline int divs128(int64_t *plow, int64_t *phigh, int64_t divisor) +static inline void divs128(int64_t *plow, int64_t *phigh, int64_t divisor) { - if (divisor == 0) { - return 1; - } else { - __int128_t dividend = ((__int128_t)*phigh << 64) | (uint64_t)*plow; - __int128_t result = dividend / divisor; - *plow = result; - *phigh = dividend % divisor; - return result != *plow; - } + __int128_t dividend = ((__int128_t)*phigh << 64) | (uint64_t)*plow; + __int128_t result = dividend / divisor; + *plow = result; + *phigh = dividend % divisor; } #else void muls64(uint64_t *plow, uint64_t *phigh, int64_t a, int64_t b); void mulu64(uint64_t *plow, uint64_t *phigh, uint64_t a, uint64_t b); -int divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor); -int divs128(int64_t *plow, int64_t *phigh, int64_t divisor); +void divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor); +void divs128(int64_t *plow, int64_t *phigh, int64_t divisor); static inline uint64_t muldiv64(uint64_t a, uint32_t b, uint32_t c) { diff --git a/target/ppc/int_helper.c b/target/ppc/int_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/ppc/int_helper.c +++ b/target/ppc/int_helper.c @@ -XXX,XX +XXX,XX @@ uint64_t helper_divdeu(CPUPPCState *env, uint64_t ra, uint64_t rb, uint32_t oe) uint64_t rt = 0; int overflow = 0; - overflow = divu128(&rt, &ra, rb); - - if (unlikely(overflow)) { + if (unlikely(rb == 0 || ra >= rb)) { + overflow = 1; rt = 0; /* Undefined */ + } else { + divu128(&rt, &ra, rb); } if (oe) { @@ -XXX,XX +XXX,XX @@ uint64_t helper_divde(CPUPPCState *env, uint64_t rau, uint64_t rbu, uint32_t oe) int64_t rt = 0; int64_t ra = (int64_t)rau; int64_t rb = (int64_t)rbu; - int overflow = divs128(&rt, &ra, rb); + int overflow = 0; - if (unlikely(overflow)) { + if (unlikely(rb == 0 || uabs64(ra) >= uabs64(rb))) { + overflow = 1; rt = 0; /* Undefined */ + } else { + divs128(&rt, &ra, rb); } if (oe) { diff --git a/util/host-utils.c b/util/host-utils.c index XXXXXXX..XXXXXXX 100644 --- a/util/host-utils.c +++ b/util/host-utils.c @@ -XXX,XX +XXX,XX @@ void muls64 (uint64_t *plow, uint64_t *phigh, int64_t a, int64_t b) *phigh = rh; } -/* Unsigned 128x64 division. Returns 1 if overflow (divide by zero or */ -/* quotient exceeds 64 bits). Otherwise returns quotient via plow and */ -/* remainder via phigh. */ -int divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) +/* + * Unsigned 128-by-64 division. Returns quotient via plow and + * remainder via phigh. + * The result must fit in 64 bits (plow) - otherwise, the result + * is undefined. + * This function will cause a division by zero if passed a zero divisor. + */ +void divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) { uint64_t dhi = *phigh; uint64_t dlo = *plow; unsigned i; uint64_t carry = 0; - if (divisor == 0) { - return 1; - } else if (dhi == 0) { + if (divisor == 0 || dhi == 0) { *plow = dlo / divisor; *phigh = dlo % divisor; - return 0; - } else if (dhi >= divisor) { - return 1; } else { for (i = 0; i < 64; i++) { @@ -XXX,XX +XXX,XX @@ int divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) *plow = dlo; *phigh = dhi; - return 0; } } -int divs128(int64_t *plow, int64_t *phigh, int64_t divisor) +/* + * Signed 128-by-64 division. Returns quotient via plow and + * remainder via phigh. + * The result must fit in 64 bits (plow) - otherwise, the result + * is undefined. + * This function will cause a division by zero if passed a zero divisor. + */ +void divs128(int64_t *plow, int64_t *phigh, int64_t divisor) { int sgn_dvdnd = *phigh < 0; int sgn_divsr = divisor < 0; - int overflow = 0; if (sgn_dvdnd) { *plow = ~(*plow); @@ -XXX,XX +XXX,XX @@ int divs128(int64_t *plow, int64_t *phigh, int64_t divisor) divisor = 0 - divisor; } - overflow = divu128((uint64_t *)plow, (uint64_t *)phigh, (uint64_t)divisor); + divu128((uint64_t *)plow, (uint64_t *)phigh, (uint64_t)divisor); if (sgn_dvdnd ^ sgn_divsr) { *plow = 0 - *plow; } - - if (!overflow) { - if ((*plow < 0) ^ (sgn_dvdnd ^ sgn_divsr)) { - overflow = 1; - } - } - - return overflow; } #endif -- 2.25.1 From: Luis Pires <luis.pires@eldorado.org.br> Move udiv_qrnnd() from include/fpu/softfloat-macros.h to host-utils, so it can be reused by divu128(). Signed-off-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20211025191154.350831-3-luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/fpu/softfloat-macros.h | 82 ---------------------------------- include/qemu/host-utils.h | 81 +++++++++++++++++++++++++++++++++ 2 files changed, 81 insertions(+), 82 deletions(-) diff --git a/include/fpu/softfloat-macros.h b/include/fpu/softfloat-macros.h index XXXXXXX..XXXXXXX 100644 --- a/include/fpu/softfloat-macros.h +++ b/include/fpu/softfloat-macros.h @@ -XXX,XX +XXX,XX @@ * so some portions are provided under: * the SoftFloat-2a license * the BSD license - * GPL-v2-or-later * * Any future contributions to this file after December 1st 2014 will be * taken to be licensed under the Softfloat-2a license unless specifically @@ -XXX,XX +XXX,XX @@ this code that are retained. * THE POSSIBILITY OF SUCH DAMAGE. */ -/* Portions of this work are licensed under the terms of the GNU GPL, - * version 2 or later. See the COPYING file in the top-level directory. - */ - #ifndef FPU_SOFTFLOAT_MACROS_H #define FPU_SOFTFLOAT_MACROS_H @@ -XXX,XX +XXX,XX @@ static inline uint64_t estimateDiv128To64(uint64_t a0, uint64_t a1, uint64_t b) } -/* From the GNU Multi Precision Library - longlong.h __udiv_qrnnd - * (https://gmplib.org/repo/gmp/file/tip/longlong.h) - * - * Licensed under the GPLv2/LGPLv3 - */ -static inline uint64_t udiv_qrnnd(uint64_t *r, uint64_t n1, - uint64_t n0, uint64_t d) -{ -#if defined(__x86_64__) - uint64_t q; - asm("divq %4" : "=a"(q), "=d"(*r) : "0"(n0), "1"(n1), "rm"(d)); - return q; -#elif defined(__s390x__) && !defined(__clang__) - /* Need to use a TImode type to get an even register pair for DLGR. */ - unsigned __int128 n = (unsigned __int128)n1 << 64 | n0; - asm("dlgr %0, %1" : "+r"(n) : "r"(d)); - *r = n >> 64; - return n; -#elif defined(_ARCH_PPC64) && defined(_ARCH_PWR7) - /* From Power ISA 2.06, programming note for divdeu. */ - uint64_t q1, q2, Q, r1, r2, R; - asm("divdeu %0,%2,%4; divdu %1,%3,%4" - : "=&r"(q1), "=r"(q2) - : "r"(n1), "r"(n0), "r"(d)); - r1 = -(q1 * d); /* low part of (n1<<64) - (q1 * d) */ - r2 = n0 - (q2 * d); - Q = q1 + q2; - R = r1 + r2; - if (R >= d || R < r2) { /* overflow implies R > d */ - Q += 1; - R -= d; - } - *r = R; - return Q; -#else - uint64_t d0, d1, q0, q1, r1, r0, m; - - d0 = (uint32_t)d; - d1 = d >> 32; - - r1 = n1 % d1; - q1 = n1 / d1; - m = q1 * d0; - r1 = (r1 << 32) | (n0 >> 32); - if (r1 < m) { - q1 -= 1; - r1 += d; - if (r1 >= d) { - if (r1 < m) { - q1 -= 1; - r1 += d; - } - } - } - r1 -= m; - - r0 = r1 % d1; - q0 = r1 / d1; - m = q0 * d0; - r0 = (r0 << 32) | (uint32_t)n0; - if (r0 < m) { - q0 -= 1; - r0 += d; - if (r0 >= d) { - if (r0 < m) { - q0 -= 1; - r0 += d; - } - } - } - r0 -= m; - - *r = r0; - return (q1 << 32) | q0; -#endif -} - /*---------------------------------------------------------------------------- | Returns an approximation to the square root of the 32-bit significand given | by `a'. Considered as an integer, `a' must be at least 2^31. If bit 0 of diff --git a/include/qemu/host-utils.h b/include/qemu/host-utils.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/host-utils.h +++ b/include/qemu/host-utils.h @@ -XXX,XX +XXX,XX @@ * THE SOFTWARE. */ +/* Portions of this work are licensed under the terms of the GNU GPL, + * version 2 or later. See the COPYING file in the top-level directory. + */ + #ifndef HOST_UTILS_H #define HOST_UTILS_H @@ -XXX,XX +XXX,XX @@ void urshift(uint64_t *plow, uint64_t *phigh, int32_t shift); */ void ulshift(uint64_t *plow, uint64_t *phigh, int32_t shift, bool *overflow); +/* From the GNU Multi Precision Library - longlong.h __udiv_qrnnd + * (https://gmplib.org/repo/gmp/file/tip/longlong.h) + * + * Licensed under the GPLv2/LGPLv3 + */ +static inline uint64_t udiv_qrnnd(uint64_t *r, uint64_t n1, + uint64_t n0, uint64_t d) +{ +#if defined(__x86_64__) + uint64_t q; + asm("divq %4" : "=a"(q), "=d"(*r) : "0"(n0), "1"(n1), "rm"(d)); + return q; +#elif defined(__s390x__) && !defined(__clang__) + /* Need to use a TImode type to get an even register pair for DLGR. */ + unsigned __int128 n = (unsigned __int128)n1 << 64 | n0; + asm("dlgr %0, %1" : "+r"(n) : "r"(d)); + *r = n >> 64; + return n; +#elif defined(_ARCH_PPC64) && defined(_ARCH_PWR7) + /* From Power ISA 2.06, programming note for divdeu. */ + uint64_t q1, q2, Q, r1, r2, R; + asm("divdeu %0,%2,%4; divdu %1,%3,%4" + : "=&r"(q1), "=r"(q2) + : "r"(n1), "r"(n0), "r"(d)); + r1 = -(q1 * d); /* low part of (n1<<64) - (q1 * d) */ + r2 = n0 - (q2 * d); + Q = q1 + q2; + R = r1 + r2; + if (R >= d || R < r2) { /* overflow implies R > d */ + Q += 1; + R -= d; + } + *r = R; + return Q; +#else + uint64_t d0, d1, q0, q1, r1, r0, m; + + d0 = (uint32_t)d; + d1 = d >> 32; + + r1 = n1 % d1; + q1 = n1 / d1; + m = q1 * d0; + r1 = (r1 << 32) | (n0 >> 32); + if (r1 < m) { + q1 -= 1; + r1 += d; + if (r1 >= d) { + if (r1 < m) { + q1 -= 1; + r1 += d; + } + } + } + r1 -= m; + + r0 = r1 % d1; + q0 = r1 / d1; + m = q0 * d0; + r0 = (r0 << 32) | (uint32_t)n0; + if (r0 < m) { + q0 -= 1; + r0 += d; + if (r0 >= d) { + if (r0 < m) { + q0 -= 1; + r0 += d; + } + } + } + r0 -= m; + + *r = r0; + return (q1 << 32) | q0; +#endif +} + #endif -- 2.25.1 From: Luis Pires <luis.pires@eldorado.org.br> These will be used to implement new decimal floating point instructions from Power ISA 3.1. The remainder is now returned directly by divu128/divs128, freeing up phigh to receive the high 64 bits of the quotient. Signed-off-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20211025191154.350831-4-luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/hw/clock.h | 6 +- include/qemu/host-utils.h | 20 ++++-- target/ppc/int_helper.c | 9 +-- util/host-utils.c | 133 +++++++++++++++++++++++++------------- 4 files changed, 108 insertions(+), 60 deletions(-) diff --git a/include/hw/clock.h b/include/hw/clock.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/clock.h +++ b/include/hw/clock.h @@ -XXX,XX +XXX,XX @@ static inline uint64_t clock_ns_to_ticks(const Clock *clk, uint64_t ns) if (clk->period == 0) { return 0; } - /* - * BUG: when CONFIG_INT128 is not defined, the current implementation of - * divu128 does not return a valid truncated quotient, so the result will - * be wrong. - */ + divu128(&lo, &hi, clk->period); return lo; } diff --git a/include/qemu/host-utils.h b/include/qemu/host-utils.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/host-utils.h +++ b/include/qemu/host-utils.h @@ -XXX,XX +XXX,XX @@ static inline uint64_t muldiv64(uint64_t a, uint32_t b, uint32_t c) return (__int128_t)a * b / c; } -static inline void divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) +static inline uint64_t divu128(uint64_t *plow, uint64_t *phigh, + uint64_t divisor) { __uint128_t dividend = ((__uint128_t)*phigh << 64) | *plow; __uint128_t result = dividend / divisor; + *plow = result; - *phigh = dividend % divisor; + *phigh = result >> 64; + return dividend % divisor; } -static inline void divs128(int64_t *plow, int64_t *phigh, int64_t divisor) +static inline int64_t divs128(uint64_t *plow, int64_t *phigh, + int64_t divisor) { - __int128_t dividend = ((__int128_t)*phigh << 64) | (uint64_t)*plow; + __int128_t dividend = ((__int128_t)*phigh << 64) | *plow; __int128_t result = dividend / divisor; + *plow = result; - *phigh = dividend % divisor; + *phigh = result >> 64; + return dividend % divisor; } #else void muls64(uint64_t *plow, uint64_t *phigh, int64_t a, int64_t b); void mulu64(uint64_t *plow, uint64_t *phigh, uint64_t a, uint64_t b); -void divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor); -void divs128(int64_t *plow, int64_t *phigh, int64_t divisor); +uint64_t divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor); +int64_t divs128(uint64_t *plow, int64_t *phigh, int64_t divisor); static inline uint64_t muldiv64(uint64_t a, uint32_t b, uint32_t c) { diff --git a/target/ppc/int_helper.c b/target/ppc/int_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/ppc/int_helper.c +++ b/target/ppc/int_helper.c @@ -XXX,XX +XXX,XX @@ uint64_t helper_divdeu(CPUPPCState *env, uint64_t ra, uint64_t rb, uint32_t oe) uint64_t helper_divde(CPUPPCState *env, uint64_t rau, uint64_t rbu, uint32_t oe) { - int64_t rt = 0; + uint64_t rt = 0; int64_t ra = (int64_t)rau; int64_t rb = (int64_t)rbu; int overflow = 0; @@ -XXX,XX +XXX,XX @@ uint32_t helper_bcdcfsq(ppc_avr_t *r, ppc_avr_t *b, uint32_t ps) int cr; uint64_t lo_value; uint64_t hi_value; + uint64_t rem; ppc_avr_t ret = { .u64 = { 0, 0 } }; if (b->VsrSD(0) < 0) { @@ -XXX,XX +XXX,XX @@ uint32_t helper_bcdcfsq(ppc_avr_t *r, ppc_avr_t *b, uint32_t ps) * In that case, we leave r unchanged. */ } else { - divu128(&lo_value, &hi_value, 1000000000000000ULL); + rem = divu128(&lo_value, &hi_value, 1000000000000000ULL); - for (i = 1; i < 16; hi_value /= 10, i++) { - bcd_put_digit(&ret, hi_value % 10, i); + for (i = 1; i < 16; rem /= 10, i++) { + bcd_put_digit(&ret, rem % 10, i); } for (; i < 32; lo_value /= 10, i++) { diff --git a/util/host-utils.c b/util/host-utils.c index XXXXXXX..XXXXXXX 100644 --- a/util/host-utils.c +++ b/util/host-utils.c @@ -XXX,XX +XXX,XX @@ void muls64 (uint64_t *plow, uint64_t *phigh, int64_t a, int64_t b) } /* - * Unsigned 128-by-64 division. Returns quotient via plow and - * remainder via phigh. - * The result must fit in 64 bits (plow) - otherwise, the result - * is undefined. - * This function will cause a division by zero if passed a zero divisor. + * Unsigned 128-by-64 division. + * Returns the remainder. + * Returns quotient via plow and phigh. + * Also returns the remainder via the function return value. */ -void divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) +uint64_t divu128(uint64_t *plow, uint64_t *phigh, uint64_t divisor) { uint64_t dhi = *phigh; uint64_t dlo = *plow; - unsigned i; - uint64_t carry = 0; + uint64_t rem, dhighest; + int sh; if (divisor == 0 || dhi == 0) { *plow = dlo / divisor; - *phigh = dlo % divisor; + *phigh = 0; + return dlo % divisor; } else { + sh = clz64(divisor); - for (i = 0; i < 64; i++) { - carry = dhi >> 63; - dhi = (dhi << 1) | (dlo >> 63); - if (carry || (dhi >= divisor)) { - dhi -= divisor; - carry = 1; - } else { - carry = 0; + if (dhi < divisor) { + if (sh != 0) { + /* normalize the divisor, shifting the dividend accordingly */ + divisor <<= sh; + dhi = (dhi << sh) | (dlo >> (64 - sh)); + dlo <<= sh; } - dlo = (dlo << 1) | carry; + + *phigh = 0; + *plow = udiv_qrnnd(&rem, dhi, dlo, divisor); + } else { + if (sh != 0) { + /* normalize the divisor, shifting the dividend accordingly */ + divisor <<= sh; + dhighest = dhi >> (64 - sh); + dhi = (dhi << sh) | (dlo >> (64 - sh)); + dlo <<= sh; + + *phigh = udiv_qrnnd(&dhi, dhighest, dhi, divisor); + } else { + /** + * dhi >= divisor + * Since the MSB of divisor is set (sh == 0), + * (dhi - divisor) < divisor + * + * Thus, the high part of the quotient is 1, and we can + * calculate the low part with a single call to udiv_qrnnd + * after subtracting divisor from dhi + */ + dhi -= divisor; + *phigh = 1; + } + + *plow = udiv_qrnnd(&rem, dhi, dlo, divisor); } - *plow = dlo; - *phigh = dhi; + /* + * since the dividend/divisor might have been normalized, + * the remainder might also have to be shifted back + */ + return rem >> sh; } } /* - * Signed 128-by-64 division. Returns quotient via plow and - * remainder via phigh. - * The result must fit in 64 bits (plow) - otherwise, the result - * is undefined. - * This function will cause a division by zero if passed a zero divisor. + * Signed 128-by-64 division. + * Returns quotient via plow and phigh. + * Also returns the remainder via the function return value. */ -void divs128(int64_t *plow, int64_t *phigh, int64_t divisor) +int64_t divs128(uint64_t *plow, int64_t *phigh, int64_t divisor) { - int sgn_dvdnd = *phigh < 0; - int sgn_divsr = divisor < 0; + bool neg_quotient = false, neg_remainder = false; + uint64_t unsig_hi = *phigh, unsig_lo = *plow; + uint64_t rem; - if (sgn_dvdnd) { - *plow = ~(*plow); - *phigh = ~(*phigh); - if (*plow == (int64_t)-1) { + if (*phigh < 0) { + neg_quotient = !neg_quotient; + neg_remainder = !neg_remainder; + + if (unsig_lo == 0) { + unsig_hi = -unsig_hi; + } else { + unsig_hi = ~unsig_hi; + unsig_lo = -unsig_lo; + } + } + + if (divisor < 0) { + neg_quotient = !neg_quotient; + + divisor = -divisor; + } + + rem = divu128(&unsig_lo, &unsig_hi, (uint64_t)divisor); + + if (neg_quotient) { + if (unsig_lo == 0) { + *phigh = -unsig_hi; *plow = 0; - (*phigh)++; - } else { - (*plow)++; - } + } else { + *phigh = ~unsig_hi; + *plow = -unsig_lo; + } + } else { + *phigh = unsig_hi; + *plow = unsig_lo; } - if (sgn_divsr) { - divisor = 0 - divisor; - } - - divu128((uint64_t *)plow, (uint64_t *)phigh, (uint64_t)divisor); - - if (sgn_dvdnd ^ sgn_divsr) { - *plow = 0 - *plow; + if (neg_remainder) { + return -rem; + } else { + return rem; } } #endif -- 2.25.1 From: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20211025191154.350831-5-luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tests/unit/test-div128.c | 197 +++++++++++++++++++++++++++++++++++++++ tests/unit/meson.build | 1 + 2 files changed, 198 insertions(+) create mode 100644 tests/unit/test-div128.c diff --git a/tests/unit/test-div128.c b/tests/unit/test-div128.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/tests/unit/test-div128.c @@ -XXX,XX +XXX,XX @@ +/* + * Test 128-bit division functions + * + * Copyright (c) 2021 Instituto de Pesquisas Eldorado (eldorado.org.br) + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, see <http://www.gnu.org/licenses/>. + */ + +#include "qemu/osdep.h" +#include "qemu/host-utils.h" + +typedef struct { + uint64_t high; + uint64_t low; + uint64_t rhigh; + uint64_t rlow; + uint64_t divisor; + uint64_t remainder; +} test_data_unsigned; + +typedef struct { + int64_t high; + uint64_t low; + int64_t rhigh; + uint64_t rlow; + int64_t divisor; + int64_t remainder; +} test_data_signed; + +static const test_data_unsigned test_table_unsigned[] = { + /* Dividend fits in 64 bits */ + { 0x0000000000000000ULL, 0x0000000000000000ULL, + 0x0000000000000000ULL, 0x0000000000000000ULL, + 0x0000000000000001ULL, 0x0000000000000000ULL}, + { 0x0000000000000000ULL, 0x0000000000000001ULL, + 0x0000000000000000ULL, 0x0000000000000001ULL, + 0x0000000000000001ULL, 0x0000000000000000ULL}, + { 0x0000000000000000ULL, 0x0000000000000003ULL, + 0x0000000000000000ULL, 0x0000000000000001ULL, + 0x0000000000000002ULL, 0x0000000000000001ULL}, + { 0x0000000000000000ULL, 0x8000000000000000ULL, + 0x0000000000000000ULL, 0x8000000000000000ULL, + 0x0000000000000001ULL, 0x0000000000000000ULL}, + { 0x0000000000000000ULL, 0xa000000000000000ULL, + 0x0000000000000000ULL, 0x0000000000000002ULL, + 0x4000000000000000ULL, 0x2000000000000000ULL}, + { 0x0000000000000000ULL, 0x8000000000000000ULL, + 0x0000000000000000ULL, 0x0000000000000001ULL, + 0x8000000000000000ULL, 0x0000000000000000ULL}, + + /* Dividend > 64 bits, with MSB 0 */ + { 0x123456789abcdefeULL, 0xefedcba987654321ULL, + 0x123456789abcdefeULL, 0xefedcba987654321ULL, + 0x0000000000000001ULL, 0x0000000000000000ULL}, + { 0x123456789abcdefeULL, 0xefedcba987654321ULL, + 0x0000000000000001ULL, 0x000000000000000dULL, + 0x123456789abcdefeULL, 0x03456789abcdf03bULL}, + { 0x123456789abcdefeULL, 0xefedcba987654321ULL, + 0x0123456789abcdefULL, 0xeefedcba98765432ULL, + 0x0000000000000010ULL, 0x0000000000000001ULL}, + + /* Dividend > 64 bits, with MSB 1 */ + { 0xfeeddccbbaa99887ULL, 0x766554433221100fULL, + 0xfeeddccbbaa99887ULL, 0x766554433221100fULL, + 0x0000000000000001ULL, 0x0000000000000000ULL}, + { 0xfeeddccbbaa99887ULL, 0x766554433221100fULL, + 0x0000000000000001ULL, 0x0000000000000000ULL, + 0xfeeddccbbaa99887ULL, 0x766554433221100fULL}, + { 0xfeeddccbbaa99887ULL, 0x766554433221100fULL, + 0x0feeddccbbaa9988ULL, 0x7766554433221100ULL, + 0x0000000000000010ULL, 0x000000000000000fULL}, + { 0xfeeddccbbaa99887ULL, 0x766554433221100fULL, + 0x000000000000000eULL, 0x00f0f0f0f0f0f35aULL, + 0x123456789abcdefeULL, 0x0f8922bc55ef90c3ULL}, + + /** + * Divisor == 64 bits, with MSB 1 + * and high 64 bits of dividend >= divisor + * (for testing normalization) + */ + { 0xfeeddccbbaa99887ULL, 0x766554433221100fULL, + 0x0000000000000001ULL, 0x0000000000000000ULL, + 0xfeeddccbbaa99887ULL, 0x766554433221100fULL}, + { 0xfeeddccbbaa99887ULL, 0x766554433221100fULL, + 0x0000000000000001ULL, 0xfddbb9977553310aULL, + 0x8000000000000001ULL, 0x78899aabbccddf05ULL}, + + /* Dividend > 64 bits, divisor almost as big */ + { 0x0000000000000001ULL, 0x23456789abcdef01ULL, + 0x0000000000000000ULL, 0x000000000000000fULL, + 0x123456789abcdefeULL, 0x123456789abcde1fULL}, +}; + +static const test_data_signed test_table_signed[] = { + /* Positive dividend, positive/negative divisors */ + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0x0000000000000000LL, 0x0000000000bc614eULL, + 0x0000000000000001LL, 0x0000000000000000LL}, + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0xffffffffffffffffLL, 0x0000000000000000LL}, + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0x0000000000000000LL, 0x00000000005e30a7ULL, + 0x0000000000000002LL, 0x0000000000000000LL}, + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0xffffffffffffffffLL, 0xffffffffffa1cf59ULL, + 0xfffffffffffffffeLL, 0x0000000000000000LL}, + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0x0000000000000000LL, 0x0000000000178c29ULL, + 0x0000000000000008LL, 0x0000000000000006LL}, + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0xffffffffffffffffLL, 0xffffffffffe873d7ULL, + 0xfffffffffffffff8LL, 0x0000000000000006LL}, + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0x0000000000000000LL, 0x000000000000550dULL, + 0x0000000000000237LL, 0x0000000000000183LL}, + { 0x0000000000000000LL, 0x0000000000bc614eULL, + 0xffffffffffffffffLL, 0xffffffffffffaaf3ULL, + 0xfffffffffffffdc9LL, 0x0000000000000183LL}, + + /* Negative dividend, positive/negative divisors */ + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0x0000000000000001LL, 0x0000000000000000LL}, + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0x0000000000000000LL, 0x0000000000bc614eULL, + 0xffffffffffffffffLL, 0x0000000000000000LL}, + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0xffffffffffffffffLL, 0xffffffffffa1cf59ULL, + 0x0000000000000002LL, 0x0000000000000000LL}, + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0x0000000000000000LL, 0x00000000005e30a7ULL, + 0xfffffffffffffffeLL, 0x0000000000000000LL}, + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0xffffffffffffffffLL, 0xffffffffffe873d7ULL, + 0x0000000000000008LL, 0xfffffffffffffffaLL}, + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0x0000000000000000LL, 0x0000000000178c29ULL, + 0xfffffffffffffff8LL, 0xfffffffffffffffaLL}, + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0xffffffffffffffffLL, 0xffffffffffffaaf3ULL, + 0x0000000000000237LL, 0xfffffffffffffe7dLL}, + { 0xffffffffffffffffLL, 0xffffffffff439eb2ULL, + 0x0000000000000000LL, 0x000000000000550dULL, + 0xfffffffffffffdc9LL, 0xfffffffffffffe7dLL}, +}; + +static void test_divu128(void) +{ + int i; + uint64_t rem; + test_data_unsigned tmp; + + for (i = 0; i < ARRAY_SIZE(test_table_unsigned); ++i) { + tmp = test_table_unsigned[i]; + + rem = divu128(&tmp.low, &tmp.high, tmp.divisor); + g_assert_cmpuint(tmp.low, ==, tmp.rlow); + g_assert_cmpuint(tmp.high, ==, tmp.rhigh); + g_assert_cmpuint(rem, ==, tmp.remainder); + } +} + +static void test_divs128(void) +{ + int i; + int64_t rem; + test_data_signed tmp; + + for (i = 0; i < ARRAY_SIZE(test_table_signed); ++i) { + tmp = test_table_signed[i]; + + rem = divs128(&tmp.low, &tmp.high, tmp.divisor); + g_assert_cmpuint(tmp.low, ==, tmp.rlow); + g_assert_cmpuint(tmp.high, ==, tmp.rhigh); + g_assert_cmpuint(rem, ==, tmp.remainder); + } +} + +int main(int argc, char **argv) +{ + g_test_init(&argc, &argv, NULL); + g_test_add_func("/host-utils/test_divu128", test_divu128); + g_test_add_func("/host-utils/test_divs128", test_divs128); + return g_test_run(); +} diff --git a/tests/unit/meson.build b/tests/unit/meson.build index XXXXXXX..XXXXXXX 100644 --- a/tests/unit/meson.build +++ b/tests/unit/meson.build @@ -XXX,XX +XXX,XX @@ tests = { # all code tested by test-x86-cpuid is inside topology.h 'test-x86-cpuid': [], 'test-cutils': [], + 'test-div128': [], 'test-shift128': [], 'test-mul64': [], # all code tested by test-int128 is inside int128.h -- 2.25.1 Prepare for tracking different masks by renaming this one. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 142 +++++++++++++++++++++++++------------------------ 1 file changed, 72 insertions(+), 70 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct TempOptInfo { TCGTemp *prev_copy; TCGTemp *next_copy; uint64_t val; - uint64_t mask; + uint64_t z_mask; /* mask bit is 0 if and only if value bit is 0 */ } TempOptInfo; static inline TempOptInfo *ts_info(TCGTemp *ts) @@ -XXX,XX +XXX,XX @@ static void reset_ts(TCGTemp *ts) ti->next_copy = ts; ti->prev_copy = ts; ti->is_const = false; - ti->mask = -1; + ti->z_mask = -1; } static void reset_temp(TCGArg arg) @@ -XXX,XX +XXX,XX @@ static void init_ts_info(TCGTempSet *temps_used, TCGTemp *ts) if (ts->kind == TEMP_CONST) { ti->is_const = true; ti->val = ts->val; - ti->mask = ts->val; + ti->z_mask = ts->val; if (TCG_TARGET_REG_BITS > 32 && ts->type == TCG_TYPE_I32) { /* High bits of a 32-bit quantity are garbage. */ - ti->mask |= ~0xffffffffull; + ti->z_mask |= ~0xffffffffull; } } else { ti->is_const = false; - ti->mask = -1; + ti->z_mask = -1; } } @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) const TCGOpDef *def; TempOptInfo *di; TempOptInfo *si; - uint64_t mask; + uint64_t z_mask; TCGOpcode new_op; if (ts_are_copies(dst_ts, src_ts)) { @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) op->args[0] = dst; op->args[1] = src; - mask = si->mask; + z_mask = si->z_mask; if (TCG_TARGET_REG_BITS > 32 && new_op == INDEX_op_mov_i32) { /* High bits of the destination are now garbage. */ - mask |= ~0xffffffffull; + z_mask |= ~0xffffffffull; } - di->mask = mask; + di->z_mask = z_mask; if (src_ts->type == dst_ts->type) { TempOptInfo *ni = ts_info(si->next_copy); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } QTAILQ_FOREACH_SAFE(op, &s->ops, link, op_next) { - uint64_t mask, partmask, affected, tmp; + uint64_t z_mask, partmask, affected, tmp; int nb_oargs, nb_iargs; TCGOpcode opc = op->opc; const TCGOpDef *def = &tcg_op_defs[opc]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) /* Simplify using known-zero bits. Currently only ops with a single output argument is supported. */ - mask = -1; + z_mask = -1; affected = -1; switch (opc) { CASE_OP_32_64(ext8s): - if ((arg_info(op->args[1])->mask & 0x80) != 0) { + if ((arg_info(op->args[1])->z_mask & 0x80) != 0) { break; } QEMU_FALLTHROUGH; CASE_OP_32_64(ext8u): - mask = 0xff; + z_mask = 0xff; goto and_const; CASE_OP_32_64(ext16s): - if ((arg_info(op->args[1])->mask & 0x8000) != 0) { + if ((arg_info(op->args[1])->z_mask & 0x8000) != 0) { break; } QEMU_FALLTHROUGH; CASE_OP_32_64(ext16u): - mask = 0xffff; + z_mask = 0xffff; goto and_const; case INDEX_op_ext32s_i64: - if ((arg_info(op->args[1])->mask & 0x80000000) != 0) { + if ((arg_info(op->args[1])->z_mask & 0x80000000) != 0) { break; } QEMU_FALLTHROUGH; case INDEX_op_ext32u_i64: - mask = 0xffffffffU; + z_mask = 0xffffffffU; goto and_const; CASE_OP_32_64(and): - mask = arg_info(op->args[2])->mask; + z_mask = arg_info(op->args[2])->z_mask; if (arg_is_const(op->args[2])) { and_const: - affected = arg_info(op->args[1])->mask & ~mask; + affected = arg_info(op->args[1])->z_mask & ~z_mask; } - mask = arg_info(op->args[1])->mask & mask; + z_mask = arg_info(op->args[1])->z_mask & z_mask; break; case INDEX_op_ext_i32_i64: - if ((arg_info(op->args[1])->mask & 0x80000000) != 0) { + if ((arg_info(op->args[1])->z_mask & 0x80000000) != 0) { break; } QEMU_FALLTHROUGH; case INDEX_op_extu_i32_i64: /* We do not compute affected as it is a size changing op. */ - mask = (uint32_t)arg_info(op->args[1])->mask; + z_mask = (uint32_t)arg_info(op->args[1])->z_mask; break; CASE_OP_32_64(andc): /* Known-zeros does not imply known-ones. Therefore unless op->args[2] is constant, we can't infer anything from it. */ if (arg_is_const(op->args[2])) { - mask = ~arg_info(op->args[2])->mask; + z_mask = ~arg_info(op->args[2])->z_mask; goto and_const; } /* But we certainly know nothing outside args[1] may be set. */ - mask = arg_info(op->args[1])->mask; + z_mask = arg_info(op->args[1])->z_mask; break; case INDEX_op_sar_i32: if (arg_is_const(op->args[2])) { tmp = arg_info(op->args[2])->val & 31; - mask = (int32_t)arg_info(op->args[1])->mask >> tmp; + z_mask = (int32_t)arg_info(op->args[1])->z_mask >> tmp; } break; case INDEX_op_sar_i64: if (arg_is_const(op->args[2])) { tmp = arg_info(op->args[2])->val & 63; - mask = (int64_t)arg_info(op->args[1])->mask >> tmp; + z_mask = (int64_t)arg_info(op->args[1])->z_mask >> tmp; } break; case INDEX_op_shr_i32: if (arg_is_const(op->args[2])) { tmp = arg_info(op->args[2])->val & 31; - mask = (uint32_t)arg_info(op->args[1])->mask >> tmp; + z_mask = (uint32_t)arg_info(op->args[1])->z_mask >> tmp; } break; case INDEX_op_shr_i64: if (arg_is_const(op->args[2])) { tmp = arg_info(op->args[2])->val & 63; - mask = (uint64_t)arg_info(op->args[1])->mask >> tmp; + z_mask = (uint64_t)arg_info(op->args[1])->z_mask >> tmp; } break; case INDEX_op_extrl_i64_i32: - mask = (uint32_t)arg_info(op->args[1])->mask; + z_mask = (uint32_t)arg_info(op->args[1])->z_mask; break; case INDEX_op_extrh_i64_i32: - mask = (uint64_t)arg_info(op->args[1])->mask >> 32; + z_mask = (uint64_t)arg_info(op->args[1])->z_mask >> 32; break; CASE_OP_32_64(shl): if (arg_is_const(op->args[2])) { tmp = arg_info(op->args[2])->val & (TCG_TARGET_REG_BITS - 1); - mask = arg_info(op->args[1])->mask << tmp; + z_mask = arg_info(op->args[1])->z_mask << tmp; } break; CASE_OP_32_64(neg): /* Set to 1 all bits to the left of the rightmost. */ - mask = -(arg_info(op->args[1])->mask - & -arg_info(op->args[1])->mask); + z_mask = -(arg_info(op->args[1])->z_mask + & -arg_info(op->args[1])->z_mask); break; CASE_OP_32_64(deposit): - mask = deposit64(arg_info(op->args[1])->mask, - op->args[3], op->args[4], - arg_info(op->args[2])->mask); + z_mask = deposit64(arg_info(op->args[1])->z_mask, + op->args[3], op->args[4], + arg_info(op->args[2])->z_mask); break; CASE_OP_32_64(extract): - mask = extract64(arg_info(op->args[1])->mask, - op->args[2], op->args[3]); + z_mask = extract64(arg_info(op->args[1])->z_mask, + op->args[2], op->args[3]); if (op->args[2] == 0) { - affected = arg_info(op->args[1])->mask & ~mask; + affected = arg_info(op->args[1])->z_mask & ~z_mask; } break; CASE_OP_32_64(sextract): - mask = sextract64(arg_info(op->args[1])->mask, - op->args[2], op->args[3]); - if (op->args[2] == 0 && (tcg_target_long)mask >= 0) { - affected = arg_info(op->args[1])->mask & ~mask; + z_mask = sextract64(arg_info(op->args[1])->z_mask, + op->args[2], op->args[3]); + if (op->args[2] == 0 && (tcg_target_long)z_mask >= 0) { + affected = arg_info(op->args[1])->z_mask & ~z_mask; } break; CASE_OP_32_64(or): CASE_OP_32_64(xor): - mask = arg_info(op->args[1])->mask | arg_info(op->args[2])->mask; + z_mask = arg_info(op->args[1])->z_mask + | arg_info(op->args[2])->z_mask; break; case INDEX_op_clz_i32: case INDEX_op_ctz_i32: - mask = arg_info(op->args[2])->mask | 31; + z_mask = arg_info(op->args[2])->z_mask | 31; break; case INDEX_op_clz_i64: case INDEX_op_ctz_i64: - mask = arg_info(op->args[2])->mask | 63; + z_mask = arg_info(op->args[2])->z_mask | 63; break; case INDEX_op_ctpop_i32: - mask = 32 | 31; + z_mask = 32 | 31; break; case INDEX_op_ctpop_i64: - mask = 64 | 63; + z_mask = 64 | 63; break; CASE_OP_32_64(setcond): case INDEX_op_setcond2_i32: - mask = 1; + z_mask = 1; break; CASE_OP_32_64(movcond): - mask = arg_info(op->args[3])->mask | arg_info(op->args[4])->mask; + z_mask = arg_info(op->args[3])->z_mask + | arg_info(op->args[4])->z_mask; break; CASE_OP_32_64(ld8u): - mask = 0xff; + z_mask = 0xff; break; CASE_OP_32_64(ld16u): - mask = 0xffff; + z_mask = 0xffff; break; case INDEX_op_ld32u_i64: - mask = 0xffffffffu; + z_mask = 0xffffffffu; break; CASE_OP_32_64(qemu_ld): @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) MemOpIdx oi = op->args[nb_oargs + nb_iargs]; MemOp mop = get_memop(oi); if (!(mop & MO_SIGN)) { - mask = (2ULL << ((8 << (mop & MO_SIZE)) - 1)) - 1; + z_mask = (2ULL << ((8 << (mop & MO_SIZE)) - 1)) - 1; } } break; CASE_OP_32_64(bswap16): - mask = arg_info(op->args[1])->mask; - if (mask <= 0xffff) { + z_mask = arg_info(op->args[1])->z_mask; + if (z_mask <= 0xffff) { op->args[2] |= TCG_BSWAP_IZ; } - mask = bswap16(mask); + z_mask = bswap16(z_mask); switch (op->args[2] & (TCG_BSWAP_OZ | TCG_BSWAP_OS)) { case TCG_BSWAP_OZ: break; case TCG_BSWAP_OS: - mask = (int16_t)mask; + z_mask = (int16_t)z_mask; break; default: /* undefined high bits */ - mask |= MAKE_64BIT_MASK(16, 48); + z_mask |= MAKE_64BIT_MASK(16, 48); break; } break; case INDEX_op_bswap32_i64: - mask = arg_info(op->args[1])->mask; - if (mask <= 0xffffffffu) { + z_mask = arg_info(op->args[1])->z_mask; + if (z_mask <= 0xffffffffu) { op->args[2] |= TCG_BSWAP_IZ; } - mask = bswap32(mask); + z_mask = bswap32(z_mask); switch (op->args[2] & (TCG_BSWAP_OZ | TCG_BSWAP_OS)) { case TCG_BSWAP_OZ: break; case TCG_BSWAP_OS: - mask = (int32_t)mask; + z_mask = (int32_t)z_mask; break; default: /* undefined high bits */ - mask |= MAKE_64BIT_MASK(32, 32); + z_mask |= MAKE_64BIT_MASK(32, 32); break; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) /* 32-bit ops generate 32-bit results. For the result is zero test below, we can ignore high bits, but for further optimizations we need to record that the high bits contain garbage. */ - partmask = mask; + partmask = z_mask; if (!(def->flags & TCG_OPF_64BIT)) { - mask |= ~(tcg_target_ulong)0xffffffffu; + z_mask |= ~(tcg_target_ulong)0xffffffffu; partmask &= 0xffffffffu; affected &= 0xffffffffu; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) vs the high word of the input. */ do_setcond_high: reset_temp(op->args[0]); - arg_info(op->args[0])->mask = 1; + arg_info(op->args[0])->z_mask = 1; op->opc = INDEX_op_setcond_i32; op->args[1] = op->args[2]; op->args[2] = op->args[4]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } do_setcond_low: reset_temp(op->args[0]); - arg_info(op->args[0])->mask = 1; + arg_info(op->args[0])->z_mask = 1; op->opc = INDEX_op_setcond_i32; op->args[2] = op->args[3]; op->args[3] = op->args[5]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) /* Default case: we know nothing about operation (or were unable to compute the operation result) so no propagation is done. We trash everything if the operation is the end of a basic - block, otherwise we only trash the output args. "mask" is + block, otherwise we only trash the output args. "z_mask" is the non-zero bits mask for the first output arg. */ if (def->flags & TCG_OPF_BB_END) { memset(&temps_used, 0, sizeof(temps_used)); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) /* Save the corresponding known-zero bits mask for the first output argument (only one supported so far). */ if (i == 0) { - arg_info(op->args[i])->mask = mask; + arg_info(op->args[i])->z_mask = z_mask; } } } -- 2.25.1 Provide what will become a larger context for splitting the very large tcg_optimize function. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 77 ++++++++++++++++++++++++++------------------------ 1 file changed, 40 insertions(+), 37 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct TempOptInfo { uint64_t z_mask; /* mask bit is 0 if and only if value bit is 0 */ } TempOptInfo; +typedef struct OptContext { + TCGTempSet temps_used; +} OptContext; + static inline TempOptInfo *ts_info(TCGTemp *ts) { return ts->state_ptr; @@ -XXX,XX +XXX,XX @@ static void reset_temp(TCGArg arg) } /* Initialize and activate a temporary. */ -static void init_ts_info(TCGTempSet *temps_used, TCGTemp *ts) +static void init_ts_info(OptContext *ctx, TCGTemp *ts) { size_t idx = temp_idx(ts); TempOptInfo *ti; - if (test_bit(idx, temps_used->l)) { + if (test_bit(idx, ctx->temps_used.l)) { return; } - set_bit(idx, temps_used->l); + set_bit(idx, ctx->temps_used.l); ti = ts->state_ptr; if (ti == NULL) { @@ -XXX,XX +XXX,XX @@ static void init_ts_info(TCGTempSet *temps_used, TCGTemp *ts) } } -static void init_arg_info(TCGTempSet *temps_used, TCGArg arg) +static void init_arg_info(OptContext *ctx, TCGArg arg) { - init_ts_info(temps_used, arg_temp(arg)); + init_ts_info(ctx, arg_temp(arg)); } static TCGTemp *find_better_copy(TCGContext *s, TCGTemp *ts) @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) } } -static void tcg_opt_gen_movi(TCGContext *s, TCGTempSet *temps_used, +static void tcg_opt_gen_movi(TCGContext *s, OptContext *ctx, TCGOp *op, TCGArg dst, uint64_t val) { const TCGOpDef *def = &tcg_op_defs[op->opc]; @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_movi(TCGContext *s, TCGTempSet *temps_used, /* Convert movi to mov with constant temp. */ tv = tcg_constant_internal(type, val); - init_ts_info(temps_used, tv); + init_ts_info(ctx, tv); tcg_opt_gen_mov(s, op, dst, temp_arg(tv)); } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) { int nb_temps, nb_globals, i; TCGOp *op, *op_next, *prev_mb = NULL; - TCGTempSet temps_used; + OptContext ctx = {}; /* Array VALS has an element for each temp. If this temp holds a constant then its value is kept in VALS' element. @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) nb_temps = s->nb_temps; nb_globals = s->nb_globals; - memset(&temps_used, 0, sizeof(temps_used)); for (i = 0; i < nb_temps; ++i) { s->temps[i].state_ptr = NULL; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) for (i = 0; i < nb_oargs + nb_iargs; i++) { TCGTemp *ts = arg_temp(op->args[i]); if (ts) { - init_ts_info(&temps_used, ts); + init_ts_info(&ctx, ts); } } } else { nb_oargs = def->nb_oargs; nb_iargs = def->nb_iargs; for (i = 0; i < nb_oargs + nb_iargs; i++) { - init_arg_info(&temps_used, op->args[i]); + init_arg_info(&ctx, op->args[i]); } } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(rotr): if (arg_is_const(op->args[1]) && arg_info(op->args[1])->val == 0) { - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], 0); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (partmask == 0) { tcg_debug_assert(nb_oargs == 1); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], 0); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); continue; } if (affected == 0) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(mulsh): if (arg_is_const(op->args[2]) && arg_info(op->args[2])->val == 0) { - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], 0); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(sub): CASE_OP_32_64_VEC(xor): if (args_are_copies(op->args[1], op->args[2])) { - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], 0); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = arg_info(op->args[1])->val; tmp = dup_const(TCGOP_VECE(op), tmp); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_dup2_vec: assert(TCG_TARGET_REG_BITS == 32); if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], + tcg_opt_gen_movi(s, &ctx, op, op->args[0], deposit64(arg_info(op->args[1])->val, 32, 32, arg_info(op->args[2])->val)); break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_extrh_i64_i32: if (arg_is_const(op->args[1])) { tmp = do_constant_folding(opc, arg_info(op->args[1])->val, 0); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = do_constant_folding(opc, arg_info(op->args[1])->val, op->args[2]); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { tmp = do_constant_folding(opc, arg_info(op->args[1])->val, arg_info(op->args[2])->val); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) TCGArg v = arg_info(op->args[1])->val; if (v != 0) { tmp = do_constant_folding(opc, v, 0); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); } else { tcg_opt_gen_mov(s, op, op->args[0], op->args[2]); } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = deposit64(arg_info(op->args[1])->val, op->args[3], op->args[4], arg_info(op->args[2])->val); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = extract64(arg_info(op->args[1])->val, op->args[2], op->args[3]); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = sextract64(arg_info(op->args[1])->val, op->args[2], op->args[3]); - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = (int32_t)(((uint32_t)v1 >> shr) | ((uint32_t)v2 << (32 - shr))); } - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = do_constant_folding_cond(opc, op->args[1], op->args[2], op->args[3]); if (tmp != 2) { - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->args[1], op->args[2]); if (tmp != 2) { if (tmp) { - memset(&temps_used, 0, sizeof(temps_used)); + memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); op->opc = INDEX_op_br; op->args[0] = op->args[3]; } else { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(s, &temps_used, op, rl, (int32_t)a); - tcg_opt_gen_movi(s, &temps_used, op2, rh, (int32_t)(a >> 32)); + tcg_opt_gen_movi(s, &ctx, op, rl, (int32_t)a); + tcg_opt_gen_movi(s, &ctx, op2, rh, (int32_t)(a >> 32)); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(s, &temps_used, op, rl, (int32_t)r); - tcg_opt_gen_movi(s, &temps_used, op2, rh, (int32_t)(r >> 32)); + tcg_opt_gen_movi(s, &ctx, op, rl, (int32_t)r); + tcg_opt_gen_movi(s, &ctx, op2, rh, (int32_t)(r >> 32)); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (tmp != 2) { if (tmp) { do_brcond_true: - memset(&temps_used, 0, sizeof(temps_used)); + memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); op->opc = INDEX_op_br; op->args[0] = op->args[5]; } else { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) /* Simplify LT/GE comparisons vs zero to a single compare vs the high word of the input. */ do_brcond_high: - memset(&temps_used, 0, sizeof(temps_used)); + memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); op->opc = INDEX_op_brcond_i32; op->args[0] = op->args[1]; op->args[1] = op->args[3]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) goto do_default; } do_brcond_low: - memset(&temps_used, 0, sizeof(temps_used)); + memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); op->opc = INDEX_op_brcond_i32; op->args[1] = op->args[2]; op->args[2] = op->args[4]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->args[5]); if (tmp != 2) { do_setcond_const: - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); } else if ((op->args[5] == TCG_COND_LT || op->args[5] == TCG_COND_GE) && arg_is_const(op->args[3]) @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (!(tcg_call_flags(op) & (TCG_CALL_NO_READ_GLOBALS | TCG_CALL_NO_WRITE_GLOBALS))) { for (i = 0; i < nb_globals; i++) { - if (test_bit(i, temps_used.l)) { + if (test_bit(i, ctx.temps_used.l)) { reset_ts(&s->temps[i]); } } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) block, otherwise we only trash the output args. "z_mask" is the non-zero bits mask for the first output arg. */ if (def->flags & TCG_OPF_BB_END) { - memset(&temps_used, 0, sizeof(temps_used)); + memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); } else { do_reset_output: for (i = 0; i < nb_oargs; i++) { -- 2.25.1 Break the final cleanup clause out of the main switch statement. When fully folding an opcode to mov/movi, use "continue" to process the next opcode, else break to fall into the final cleanup. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 190 ++++++++++++++++++++++++------------------------- 1 file changed, 94 insertions(+), 96 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) switch (opc) { CASE_OP_32_64_VEC(mov): tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); - break; + continue; case INDEX_op_dup_vec: if (arg_is_const(op->args[1])) { tmp = arg_info(op->args[1])->val; tmp = dup_const(TCGOP_VECE(op), tmp); tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; case INDEX_op_dup2_vec: assert(TCG_TARGET_REG_BITS == 32); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tcg_opt_gen_movi(s, &ctx, op, op->args[0], deposit64(arg_info(op->args[1])->val, 32, 32, arg_info(op->args[2])->val)); - break; + continue; } else if (args_are_copies(op->args[1], op->args[2])) { op->opc = INDEX_op_dup_vec; TCGOP_VECE(op) = MO_32; nb_iargs = 1; } - goto do_default; + break; CASE_OP_32_64(not): CASE_OP_32_64(neg): @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = do_constant_folding(opc, arg_info(op->args[1])->val, 0); tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(bswap16): CASE_OP_32_64(bswap32): @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = do_constant_folding(opc, arg_info(op->args[1])->val, op->args[2]); tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(add): CASE_OP_32_64(sub): @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = do_constant_folding(opc, arg_info(op->args[1])->val, arg_info(op->args[2])->val); tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(clz): CASE_OP_32_64(ctz): @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } else { tcg_opt_gen_mov(s, op, op->args[0], op->args[2]); } - break; + continue; } - goto do_default; + break; CASE_OP_32_64(deposit): if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->args[3], op->args[4], arg_info(op->args[2])->val); tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(extract): if (arg_is_const(op->args[1])) { tmp = extract64(arg_info(op->args[1])->val, op->args[2], op->args[3]); tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(sextract): if (arg_is_const(op->args[1])) { tmp = sextract64(arg_info(op->args[1])->val, op->args[2], op->args[3]); tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(extract2): if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) ((uint32_t)v2 << (32 - shr))); } tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(setcond): tmp = do_constant_folding_cond(opc, op->args[1], op->args[2], op->args[3]); if (tmp != 2) { tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - break; + continue; } - goto do_default; + break; CASE_OP_32_64(brcond): tmp = do_constant_folding_cond(opc, op->args[0], op->args[1], op->args[2]); - if (tmp != 2) { - if (tmp) { - memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); - op->opc = INDEX_op_br; - op->args[0] = op->args[3]; - } else { - tcg_op_remove(s, op); - } + switch (tmp) { + case 0: + tcg_op_remove(s, op); + continue; + case 1: + memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); + op->opc = opc = INDEX_op_br; + op->args[0] = op->args[3]; break; } - goto do_default; + break; CASE_OP_32_64(movcond): tmp = do_constant_folding_cond(opc, op->args[1], op->args[2], op->args[5]); if (tmp != 2) { tcg_opt_gen_mov(s, op, op->args[0], op->args[4-tmp]); - break; + continue; } if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { uint64_t tv = arg_info(op->args[3])->val; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (fv == 1 && tv == 0) { cond = tcg_invert_cond(cond); } else if (!(tv == 1 && fv == 0)) { - goto do_default; + break; } op->args[3] = cond; op->opc = opc = (opc == INDEX_op_movcond_i32 @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) : INDEX_op_setcond_i64); nb_iargs = 2; } - goto do_default; + break; case INDEX_op_add2_i32: case INDEX_op_sub2_i32: @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) rh = op->args[1]; tcg_opt_gen_movi(s, &ctx, op, rl, (int32_t)a); tcg_opt_gen_movi(s, &ctx, op2, rh, (int32_t)(a >> 32)); - break; + continue; } - goto do_default; + break; case INDEX_op_mulu2_i32: if (arg_is_const(op->args[2]) && arg_is_const(op->args[3])) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) rh = op->args[1]; tcg_opt_gen_movi(s, &ctx, op, rl, (int32_t)r); tcg_opt_gen_movi(s, &ctx, op2, rh, (int32_t)(r >> 32)); - break; + continue; } - goto do_default; + break; case INDEX_op_brcond2_i32: tmp = do_constant_folding_cond2(&op->args[0], &op->args[2], op->args[4]); - if (tmp != 2) { - if (tmp) { - do_brcond_true: - memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); - op->opc = INDEX_op_br; - op->args[0] = op->args[5]; - } else { + if (tmp == 0) { do_brcond_false: - tcg_op_remove(s, op); - } - } else if ((op->args[4] == TCG_COND_LT - || op->args[4] == TCG_COND_GE) - && arg_is_const(op->args[2]) - && arg_info(op->args[2])->val == 0 - && arg_is_const(op->args[3]) - && arg_info(op->args[3])->val == 0) { + tcg_op_remove(s, op); + continue; + } + if (tmp == 1) { + do_brcond_true: + op->opc = opc = INDEX_op_br; + op->args[0] = op->args[5]; + break; + } + if ((op->args[4] == TCG_COND_LT || op->args[4] == TCG_COND_GE) + && arg_is_const(op->args[2]) + && arg_info(op->args[2])->val == 0 + && arg_is_const(op->args[3]) + && arg_info(op->args[3])->val == 0) { /* Simplify LT/GE comparisons vs zero to a single compare vs the high word of the input. */ do_brcond_high: - memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); - op->opc = INDEX_op_brcond_i32; + op->opc = opc = INDEX_op_brcond_i32; op->args[0] = op->args[1]; op->args[1] = op->args[3]; op->args[2] = op->args[4]; op->args[3] = op->args[5]; - } else if (op->args[4] == TCG_COND_EQ) { + break; + } + if (op->args[4] == TCG_COND_EQ) { /* Simplify EQ comparisons where one of the pairs can be simplified. */ tmp = do_constant_folding_cond(INDEX_op_brcond_i32, @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (tmp == 0) { goto do_brcond_false; } else if (tmp != 1) { - goto do_default; + break; } do_brcond_low: memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->args[1] = op->args[2]; op->args[2] = op->args[4]; op->args[3] = op->args[5]; - } else if (op->args[4] == TCG_COND_NE) { + break; + } + if (op->args[4] == TCG_COND_NE) { /* Simplify NE comparisons where one of the pairs can be simplified. */ tmp = do_constant_folding_cond(INDEX_op_brcond_i32, @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } else if (tmp == 1) { goto do_brcond_true; } - goto do_default; - } else { - goto do_default; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (tmp != 2) { do_setcond_const: tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); - } else if ((op->args[5] == TCG_COND_LT - || op->args[5] == TCG_COND_GE) - && arg_is_const(op->args[3]) - && arg_info(op->args[3])->val == 0 - && arg_is_const(op->args[4]) - && arg_info(op->args[4])->val == 0) { + continue; + } + if ((op->args[5] == TCG_COND_LT || op->args[5] == TCG_COND_GE) + && arg_is_const(op->args[3]) + && arg_info(op->args[3])->val == 0 + && arg_is_const(op->args[4]) + && arg_info(op->args[4])->val == 0) { /* Simplify LT/GE comparisons vs zero to a single compare vs the high word of the input. */ do_setcond_high: @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->args[1] = op->args[2]; op->args[2] = op->args[4]; op->args[3] = op->args[5]; - } else if (op->args[5] == TCG_COND_EQ) { + break; + } + if (op->args[5] == TCG_COND_EQ) { /* Simplify EQ comparisons where one of the pairs can be simplified. */ tmp = do_constant_folding_cond(INDEX_op_setcond_i32, @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (tmp == 0) { goto do_setcond_high; } else if (tmp != 1) { - goto do_default; + break; } do_setcond_low: reset_temp(op->args[0]); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->opc = INDEX_op_setcond_i32; op->args[2] = op->args[3]; op->args[3] = op->args[5]; - } else if (op->args[5] == TCG_COND_NE) { + break; + } + if (op->args[5] == TCG_COND_NE) { /* Simplify NE comparisons where one of the pairs can be simplified. */ tmp = do_constant_folding_cond(INDEX_op_setcond_i32, @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } else if (tmp == 1) { goto do_setcond_const; } - goto do_default; - } else { - goto do_default; } break; - case INDEX_op_call: - if (!(tcg_call_flags(op) + default: + break; + } + + /* Some of the folding above can change opc. */ + opc = op->opc; + def = &tcg_op_defs[opc]; + if (def->flags & TCG_OPF_BB_END) { + memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); + } else { + if (opc == INDEX_op_call && + !(tcg_call_flags(op) & (TCG_CALL_NO_READ_GLOBALS | TCG_CALL_NO_WRITE_GLOBALS))) { for (i = 0; i < nb_globals; i++) { if (test_bit(i, ctx.temps_used.l)) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } } } - goto do_reset_output; - default: - do_default: - /* Default case: we know nothing about operation (or were unable - to compute the operation result) so no propagation is done. - We trash everything if the operation is the end of a basic - block, otherwise we only trash the output args. "z_mask" is - the non-zero bits mask for the first output arg. */ - if (def->flags & TCG_OPF_BB_END) { - memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); - } else { - do_reset_output: - for (i = 0; i < nb_oargs; i++) { - reset_temp(op->args[i]); - /* Save the corresponding known-zero bits mask for the - first output argument (only one supported so far). */ - if (i == 0) { - arg_info(op->args[i])->z_mask = z_mask; - } + for (i = 0; i < nb_oargs; i++) { + reset_temp(op->args[i]); + /* Save the corresponding known-zero bits mask for the + first output argument (only one supported so far). */ + if (i == 0) { + arg_info(op->args[i])->z_mask = z_mask; } } - break; } /* Eliminate duplicate and redundant fence instructions. */ -- 2.25.1 Adjust the interface to take the OptContext parameter instead of TCGContext or both. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 67 +++++++++++++++++++++++++------------------------- 1 file changed, 34 insertions(+), 33 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct TempOptInfo { } TempOptInfo; typedef struct OptContext { + TCGContext *tcg; TCGTempSet temps_used; } OptContext; @@ -XXX,XX +XXX,XX @@ static bool args_are_copies(TCGArg arg1, TCGArg arg2) return ts_are_copies(arg_temp(arg1), arg_temp(arg2)); } -static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) +static void tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) { TCGTemp *dst_ts = arg_temp(dst); TCGTemp *src_ts = arg_temp(src); @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) TCGOpcode new_op; if (ts_are_copies(dst_ts, src_ts)) { - tcg_op_remove(s, op); + tcg_op_remove(ctx->tcg, op); return; } @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) } } -static void tcg_opt_gen_movi(TCGContext *s, OptContext *ctx, - TCGOp *op, TCGArg dst, uint64_t val) +static void tcg_opt_gen_movi(OptContext *ctx, TCGOp *op, + TCGArg dst, uint64_t val) { const TCGOpDef *def = &tcg_op_defs[op->opc]; TCGType type; @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_movi(TCGContext *s, OptContext *ctx, /* Convert movi to mov with constant temp. */ tv = tcg_constant_internal(type, val); init_ts_info(ctx, tv); - tcg_opt_gen_mov(s, op, dst, temp_arg(tv)); + tcg_opt_gen_mov(ctx, op, dst, temp_arg(tv)); } static uint64_t do_constant_folding_2(TCGOpcode op, uint64_t x, uint64_t y) @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) { int nb_temps, nb_globals, i; TCGOp *op, *op_next, *prev_mb = NULL; - OptContext ctx = {}; + OptContext ctx = { .tcg = s }; /* Array VALS has an element for each temp. If this temp holds a constant then its value is kept in VALS' element. @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(rotr): if (arg_is_const(op->args[1]) && arg_info(op->args[1])->val == 0) { - tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); + tcg_opt_gen_movi(&ctx, op, op->args[0], 0); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (!arg_is_const(op->args[1]) && arg_is_const(op->args[2]) && arg_info(op->args[2])->val == 0) { - tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (!arg_is_const(op->args[1]) && arg_is_const(op->args[2]) && arg_info(op->args[2])->val == -1) { - tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (partmask == 0) { tcg_debug_assert(nb_oargs == 1); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); + tcg_opt_gen_movi(&ctx, op, op->args[0], 0); continue; } if (affected == 0) { tcg_debug_assert(nb_oargs == 1); - tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); continue; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(mulsh): if (arg_is_const(op->args[2]) && arg_info(op->args[2])->val == 0) { - tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); + tcg_opt_gen_movi(&ctx, op, op->args[0], 0); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(or): CASE_OP_32_64_VEC(and): if (args_are_copies(op->args[1], op->args[2])) { - tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(sub): CASE_OP_32_64_VEC(xor): if (args_are_copies(op->args[1], op->args[2])) { - tcg_opt_gen_movi(s, &ctx, op, op->args[0], 0); + tcg_opt_gen_movi(&ctx, op, op->args[0], 0); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) allocator where needed and possible. Also detect copies. */ switch (opc) { CASE_OP_32_64_VEC(mov): - tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); continue; case INDEX_op_dup_vec: if (arg_is_const(op->args[1])) { tmp = arg_info(op->args[1])->val; tmp = dup_const(TCGOP_VECE(op), tmp); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_dup2_vec: assert(TCG_TARGET_REG_BITS == 32); if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - tcg_opt_gen_movi(s, &ctx, op, op->args[0], + tcg_opt_gen_movi(&ctx, op, op->args[0], deposit64(arg_info(op->args[1])->val, 32, 32, arg_info(op->args[2])->val)); continue; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_extrh_i64_i32: if (arg_is_const(op->args[1])) { tmp = do_constant_folding(opc, arg_info(op->args[1])->val, 0); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = do_constant_folding(opc, arg_info(op->args[1])->val, op->args[2]); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { tmp = do_constant_folding(opc, arg_info(op->args[1])->val, arg_info(op->args[2])->val); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) TCGArg v = arg_info(op->args[1])->val; if (v != 0) { tmp = do_constant_folding(opc, v, 0); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); } else { - tcg_opt_gen_mov(s, op, op->args[0], op->args[2]); + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[2]); } continue; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = deposit64(arg_info(op->args[1])->val, op->args[3], op->args[4], arg_info(op->args[2])->val); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = extract64(arg_info(op->args[1])->val, op->args[2], op->args[3]); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (arg_is_const(op->args[1])) { tmp = sextract64(arg_info(op->args[1])->val, op->args[2], op->args[3]); - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = (int32_t)(((uint32_t)v1 >> shr) | ((uint32_t)v2 << (32 - shr))); } - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = do_constant_folding_cond(opc, op->args[1], op->args[2], op->args[3]); if (tmp != 2) { - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) tmp = do_constant_folding_cond(opc, op->args[1], op->args[2], op->args[5]); if (tmp != 2) { - tcg_opt_gen_mov(s, op, op->args[0], op->args[4-tmp]); + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[4-tmp]); continue; } if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(s, &ctx, op, rl, (int32_t)a); - tcg_opt_gen_movi(s, &ctx, op2, rh, (int32_t)(a >> 32)); + tcg_opt_gen_movi(&ctx, op, rl, (int32_t)a); + tcg_opt_gen_movi(&ctx, op2, rh, (int32_t)(a >> 32)); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(s, &ctx, op, rl, (int32_t)r); - tcg_opt_gen_movi(s, &ctx, op2, rh, (int32_t)(r >> 32)); + tcg_opt_gen_movi(&ctx, op, rl, (int32_t)r); + tcg_opt_gen_movi(&ctx, op2, rh, (int32_t)(r >> 32)); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->args[5]); if (tmp != 2) { do_setcond_const: - tcg_opt_gen_movi(s, &ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); continue; } if ((op->args[5] == TCG_COND_LT || op->args[5] == TCG_COND_GE) -- 2.25.1 This will expose the variable to subroutines that will be broken out of tcg_optimize. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct TempOptInfo { typedef struct OptContext { TCGContext *tcg; + TCGOp *prev_mb; TCGTempSet temps_used; } OptContext; @@ -XXX,XX +XXX,XX @@ static bool swap_commutative2(TCGArg *p1, TCGArg *p2) void tcg_optimize(TCGContext *s) { int nb_temps, nb_globals, i; - TCGOp *op, *op_next, *prev_mb = NULL; + TCGOp *op, *op_next; OptContext ctx = { .tcg = s }; /* Array VALS has an element for each temp. @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } /* Eliminate duplicate and redundant fence instructions. */ - if (prev_mb) { + if (ctx.prev_mb) { switch (opc) { case INDEX_op_mb: /* Merge two barriers of the same type into one, @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) * barrier. This is stricter than specified but for * the purposes of TCG is better than not optimizing. */ - prev_mb->args[0] |= op->args[0]; + ctx.prev_mb->args[0] |= op->args[0]; tcg_op_remove(s, op); break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_qemu_st_i64: case INDEX_op_call: /* Opcodes that touch guest memory stop the optimization. */ - prev_mb = NULL; + ctx.prev_mb = NULL; break; } } else if (opc == INDEX_op_mb) { - prev_mb = op; + ctx.prev_mb = op; } } } -- 2.25.1 There was no real reason for calls to have separate code here. Unify init for calls vs non-calls using the call path, which handles TCG_CALL_DUMMY_ARG. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 25 +++++++++++-------------- 1 file changed, 11 insertions(+), 14 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static void init_ts_info(OptContext *ctx, TCGTemp *ts) } } -static void init_arg_info(OptContext *ctx, TCGArg arg) -{ - init_ts_info(ctx, arg_temp(arg)); -} - static TCGTemp *find_better_copy(TCGContext *s, TCGTemp *ts) { TCGTemp *i, *g, *l; @@ -XXX,XX +XXX,XX @@ static bool swap_commutative2(TCGArg *p1, TCGArg *p2) return false; } +static void init_arguments(OptContext *ctx, TCGOp *op, int nb_args) +{ + for (int i = 0; i < nb_args; i++) { + TCGTemp *ts = arg_temp(op->args[i]); + if (ts) { + init_ts_info(ctx, ts); + } + } +} + /* Propagate constants and copies, fold constant expressions. */ void tcg_optimize(TCGContext *s) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (opc == INDEX_op_call) { nb_oargs = TCGOP_CALLO(op); nb_iargs = TCGOP_CALLI(op); - for (i = 0; i < nb_oargs + nb_iargs; i++) { - TCGTemp *ts = arg_temp(op->args[i]); - if (ts) { - init_ts_info(&ctx, ts); - } - } } else { nb_oargs = def->nb_oargs; nb_iargs = def->nb_iargs; - for (i = 0; i < nb_oargs + nb_iargs; i++) { - init_arg_info(&ctx, op->args[i]); - } } + init_arguments(&ctx, op, nb_oargs + nb_iargs); /* Do copy propagation */ for (i = nb_oargs; i < nb_oargs + nb_iargs; i++) { -- 2.25.1 Continue splitting tcg_optimize. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 22 ++++++++++++++-------- 1 file changed, 14 insertions(+), 8 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static void init_arguments(OptContext *ctx, TCGOp *op, int nb_args) } } +static void copy_propagate(OptContext *ctx, TCGOp *op, + int nb_oargs, int nb_iargs) +{ + TCGContext *s = ctx->tcg; + + for (int i = nb_oargs; i < nb_oargs + nb_iargs; i++) { + TCGTemp *ts = arg_temp(op->args[i]); + if (ts && ts_is_copy(ts)) { + op->args[i] = temp_arg(find_better_copy(s, ts)); + } + } +} + /* Propagate constants and copies, fold constant expressions. */ void tcg_optimize(TCGContext *s) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) nb_iargs = def->nb_iargs; } init_arguments(&ctx, op, nb_oargs + nb_iargs); - - /* Do copy propagation */ - for (i = nb_oargs; i < nb_oargs + nb_iargs; i++) { - TCGTemp *ts = arg_temp(op->args[i]); - if (ts && ts_is_copy(ts)) { - op->args[i] = temp_arg(find_better_copy(s, ts)); - } - } + copy_propagate(&ctx, op, nb_oargs, nb_iargs); /* For commutative operations make constant second argument */ switch (opc) { -- 2.25.1 Calls are special in that they have a variable number of arguments, and need to be able to clobber globals. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 63 ++++++++++++++++++++++++++++++++------------------ 1 file changed, 41 insertions(+), 22 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static void copy_propagate(OptContext *ctx, TCGOp *op, } } +static bool fold_call(OptContext *ctx, TCGOp *op) +{ + TCGContext *s = ctx->tcg; + int nb_oargs = TCGOP_CALLO(op); + int nb_iargs = TCGOP_CALLI(op); + int flags, i; + + init_arguments(ctx, op, nb_oargs + nb_iargs); + copy_propagate(ctx, op, nb_oargs, nb_iargs); + + /* If the function reads or writes globals, reset temp data. */ + flags = tcg_call_flags(op); + if (!(flags & (TCG_CALL_NO_READ_GLOBALS | TCG_CALL_NO_WRITE_GLOBALS))) { + int nb_globals = s->nb_globals; + + for (i = 0; i < nb_globals; i++) { + if (test_bit(i, ctx->temps_used.l)) { + reset_ts(&ctx->tcg->temps[i]); + } + } + } + + /* Reset temp data for outputs. */ + for (i = 0; i < nb_oargs; i++) { + reset_temp(op->args[i]); + } + + /* Stop optimizing MB across calls. */ + ctx->prev_mb = NULL; + return true; +} + /* Propagate constants and copies, fold constant expressions. */ void tcg_optimize(TCGContext *s) { - int nb_temps, nb_globals, i; + int nb_temps, i; TCGOp *op, *op_next; OptContext ctx = { .tcg = s }; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) available through the doubly linked circular list. */ nb_temps = s->nb_temps; - nb_globals = s->nb_globals; - for (i = 0; i < nb_temps; ++i) { s->temps[i].state_ptr = NULL; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) uint64_t z_mask, partmask, affected, tmp; int nb_oargs, nb_iargs; TCGOpcode opc = op->opc; - const TCGOpDef *def = &tcg_op_defs[opc]; + const TCGOpDef *def; - /* Count the arguments, and initialize the temps that are - going to be used */ + /* Calls are special. */ if (opc == INDEX_op_call) { - nb_oargs = TCGOP_CALLO(op); - nb_iargs = TCGOP_CALLI(op); - } else { - nb_oargs = def->nb_oargs; - nb_iargs = def->nb_iargs; + fold_call(&ctx, op); + continue; } + + def = &tcg_op_defs[opc]; + nb_oargs = def->nb_oargs; + nb_iargs = def->nb_iargs; init_arguments(&ctx, op, nb_oargs + nb_iargs); copy_propagate(&ctx, op, nb_oargs, nb_iargs); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (def->flags & TCG_OPF_BB_END) { memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); } else { - if (opc == INDEX_op_call && - !(tcg_call_flags(op) - & (TCG_CALL_NO_READ_GLOBALS | TCG_CALL_NO_WRITE_GLOBALS))) { - for (i = 0; i < nb_globals; i++) { - if (test_bit(i, ctx.temps_used.l)) { - reset_ts(&s->temps[i]); - } - } - } - for (i = 0; i < nb_oargs; i++) { reset_temp(op->args[i]); /* Save the corresponding known-zero bits mask for the @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st8_i32: case INDEX_op_qemu_st_i64: - case INDEX_op_call: /* Opcodes that touch guest memory stop the optimization. */ ctx.prev_mb = NULL; break; -- 2.25.1 Rather than try to keep these up-to-date across folding, re-read nb_oargs at the end, after re-reading the opcode. A couple of asserts need dropping, but that will take care of itself as we split the function further. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 14 ++++---------- 1 file changed, 4 insertions(+), 10 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) QTAILQ_FOREACH_SAFE(op, &s->ops, link, op_next) { uint64_t z_mask, partmask, affected, tmp; - int nb_oargs, nb_iargs; TCGOpcode opc = op->opc; const TCGOpDef *def; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } def = &tcg_op_defs[opc]; - nb_oargs = def->nb_oargs; - nb_iargs = def->nb_iargs; - init_arguments(&ctx, op, nb_oargs + nb_iargs); - copy_propagate(&ctx, op, nb_oargs, nb_iargs); + init_arguments(&ctx, op, def->nb_oargs + def->nb_iargs); + copy_propagate(&ctx, op, def->nb_oargs, def->nb_iargs); /* For commutative operations make constant second argument */ switch (opc) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(qemu_ld): { - MemOpIdx oi = op->args[nb_oargs + nb_iargs]; + MemOpIdx oi = op->args[def->nb_oargs + def->nb_iargs]; MemOp mop = get_memop(oi); if (!(mop & MO_SIGN)) { z_mask = (2ULL << ((8 << (mop & MO_SIZE)) - 1)) - 1; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } if (partmask == 0) { - tcg_debug_assert(nb_oargs == 1); tcg_opt_gen_movi(&ctx, op, op->args[0], 0); continue; } if (affected == 0) { - tcg_debug_assert(nb_oargs == 1); tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); continue; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } else if (args_are_copies(op->args[1], op->args[2])) { op->opc = INDEX_op_dup_vec; TCGOP_VECE(op) = MO_32; - nb_iargs = 1; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) op->opc = opc = (opc == INDEX_op_movcond_i32 ? INDEX_op_setcond_i32 : INDEX_op_setcond_i64); - nb_iargs = 2; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (def->flags & TCG_OPF_BB_END) { memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); } else { + int nb_oargs = def->nb_oargs; for (i = 0; i < nb_oargs; i++) { reset_temp(op->args[i]); /* Save the corresponding known-zero bits mask for the -- 2.25.1 Return -1 instead of 2 for failure, so that we can use comparisons against 0 for all cases. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 145 +++++++++++++++++++++++++------------------------ 1 file changed, 74 insertions(+), 71 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool do_constant_folding_cond_eq(TCGCond c) } } -/* Return 2 if the condition can't be simplified, and the result - of the condition (0 or 1) if it can */ -static TCGArg do_constant_folding_cond(TCGOpcode op, TCGArg x, - TCGArg y, TCGCond c) +/* + * Return -1 if the condition can't be simplified, + * and the result of the condition (0 or 1) if it can. + */ +static int do_constant_folding_cond(TCGOpcode op, TCGArg x, + TCGArg y, TCGCond c) { uint64_t xv = arg_info(x)->val; uint64_t yv = arg_info(y)->val; @@ -XXX,XX +XXX,XX @@ static TCGArg do_constant_folding_cond(TCGOpcode op, TCGArg x, case TCG_COND_GEU: return 1; default: - return 2; + return -1; } } - return 2; + return -1; } -/* Return 2 if the condition can't be simplified, and the result - of the condition (0 or 1) if it can */ -static TCGArg do_constant_folding_cond2(TCGArg *p1, TCGArg *p2, TCGCond c) +/* + * Return -1 if the condition can't be simplified, + * and the result of the condition (0 or 1) if it can. + */ +static int do_constant_folding_cond2(TCGArg *p1, TCGArg *p2, TCGCond c) { TCGArg al = p1[0], ah = p1[1]; TCGArg bl = p2[0], bh = p2[1]; @@ -XXX,XX +XXX,XX @@ static TCGArg do_constant_folding_cond2(TCGArg *p1, TCGArg *p2, TCGCond c) if (args_are_copies(al, bl) && args_are_copies(ah, bh)) { return do_constant_folding_cond_eq(c); } - return 2; + return -1; } static bool swap_commutative(TCGArg dest, TCGArg *p1, TCGArg *p2) @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; CASE_OP_32_64(setcond): - tmp = do_constant_folding_cond(opc, op->args[1], - op->args[2], op->args[3]); - if (tmp != 2) { - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); + i = do_constant_folding_cond(opc, op->args[1], + op->args[2], op->args[3]); + if (i >= 0) { + tcg_opt_gen_movi(&ctx, op, op->args[0], i); continue; } break; CASE_OP_32_64(brcond): - tmp = do_constant_folding_cond(opc, op->args[0], - op->args[1], op->args[2]); - switch (tmp) { - case 0: + i = do_constant_folding_cond(opc, op->args[0], + op->args[1], op->args[2]); + if (i == 0) { tcg_op_remove(s, op); continue; - case 1: + } else if (i > 0) { memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); op->opc = opc = INDEX_op_br; op->args[0] = op->args[3]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; CASE_OP_32_64(movcond): - tmp = do_constant_folding_cond(opc, op->args[1], - op->args[2], op->args[5]); - if (tmp != 2) { - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[4-tmp]); + i = do_constant_folding_cond(opc, op->args[1], + op->args[2], op->args[5]); + if (i >= 0) { + tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[4 - i]); continue; } if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; case INDEX_op_brcond2_i32: - tmp = do_constant_folding_cond2(&op->args[0], &op->args[2], - op->args[4]); - if (tmp == 0) { + i = do_constant_folding_cond2(&op->args[0], &op->args[2], + op->args[4]); + if (i == 0) { do_brcond_false: tcg_op_remove(s, op); continue; } - if (tmp == 1) { + if (i > 0) { do_brcond_true: op->opc = opc = INDEX_op_br; op->args[0] = op->args[5]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (op->args[4] == TCG_COND_EQ) { /* Simplify EQ comparisons where one of the pairs can be simplified. */ - tmp = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[0], op->args[2], - TCG_COND_EQ); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_brcond_i32, + op->args[0], op->args[2], + TCG_COND_EQ); + if (i == 0) { goto do_brcond_false; - } else if (tmp == 1) { + } else if (i > 0) { goto do_brcond_high; } - tmp = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[1], op->args[3], - TCG_COND_EQ); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_brcond_i32, + op->args[1], op->args[3], + TCG_COND_EQ); + if (i == 0) { goto do_brcond_false; - } else if (tmp != 1) { + } else if (i < 0) { break; } do_brcond_low: @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (op->args[4] == TCG_COND_NE) { /* Simplify NE comparisons where one of the pairs can be simplified. */ - tmp = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[0], op->args[2], - TCG_COND_NE); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_brcond_i32, + op->args[0], op->args[2], + TCG_COND_NE); + if (i == 0) { goto do_brcond_high; - } else if (tmp == 1) { + } else if (i > 0) { goto do_brcond_true; } - tmp = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[1], op->args[3], - TCG_COND_NE); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_brcond_i32, + op->args[1], op->args[3], + TCG_COND_NE); + if (i == 0) { goto do_brcond_low; - } else if (tmp == 1) { + } else if (i > 0) { goto do_brcond_true; } } break; case INDEX_op_setcond2_i32: - tmp = do_constant_folding_cond2(&op->args[1], &op->args[3], - op->args[5]); - if (tmp != 2) { + i = do_constant_folding_cond2(&op->args[1], &op->args[3], + op->args[5]); + if (i >= 0) { do_setcond_const: - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); + tcg_opt_gen_movi(&ctx, op, op->args[0], i); continue; } if ((op->args[5] == TCG_COND_LT || op->args[5] == TCG_COND_GE) @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (op->args[5] == TCG_COND_EQ) { /* Simplify EQ comparisons where one of the pairs can be simplified. */ - tmp = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[1], op->args[3], - TCG_COND_EQ); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_setcond_i32, + op->args[1], op->args[3], + TCG_COND_EQ); + if (i == 0) { goto do_setcond_const; - } else if (tmp == 1) { + } else if (i > 0) { goto do_setcond_high; } - tmp = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[2], op->args[4], - TCG_COND_EQ); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_setcond_i32, + op->args[2], op->args[4], + TCG_COND_EQ); + if (i == 0) { goto do_setcond_high; - } else if (tmp != 1) { + } else if (i < 0) { break; } do_setcond_low: @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (op->args[5] == TCG_COND_NE) { /* Simplify NE comparisons where one of the pairs can be simplified. */ - tmp = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[1], op->args[3], - TCG_COND_NE); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_setcond_i32, + op->args[1], op->args[3], + TCG_COND_NE); + if (i == 0) { goto do_setcond_high; - } else if (tmp == 1) { + } else if (i > 0) { goto do_setcond_const; } - tmp = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[2], op->args[4], - TCG_COND_NE); - if (tmp == 0) { + i = do_constant_folding_cond(INDEX_op_setcond_i32, + op->args[2], op->args[4], + TCG_COND_NE); + if (i == 0) { goto do_setcond_low; - } else if (tmp == 1) { + } else if (i > 0) { goto do_setcond_const; } } -- 2.25.1 This will allow callers to tail call to these functions and return true indicating processing complete. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool args_are_copies(TCGArg arg1, TCGArg arg2) return ts_are_copies(arg_temp(arg1), arg_temp(arg2)); } -static void tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) +static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) { TCGTemp *dst_ts = arg_temp(dst); TCGTemp *src_ts = arg_temp(src); @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) if (ts_are_copies(dst_ts, src_ts)) { tcg_op_remove(ctx->tcg, op); - return; + return true; } reset_ts(dst_ts); @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) di->is_const = si->is_const; di->val = si->val; } + return true; } -static void tcg_opt_gen_movi(OptContext *ctx, TCGOp *op, +static bool tcg_opt_gen_movi(OptContext *ctx, TCGOp *op, TCGArg dst, uint64_t val) { const TCGOpDef *def = &tcg_op_defs[op->opc]; @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_movi(OptContext *ctx, TCGOp *op, /* Convert movi to mov with constant temp. */ tv = tcg_constant_internal(type, val); init_ts_info(ctx, tv); - tcg_opt_gen_mov(ctx, op, dst, temp_arg(tv)); + return tcg_opt_gen_mov(ctx, op, dst, temp_arg(tv)); } static uint64_t do_constant_folding_2(TCGOpcode op, uint64_t x, uint64_t y) -- 2.25.1 Copy z_mask into OptContext, for writeback to the first output within the new function. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 49 +++++++++++++++++++++++++++++++++---------------- 1 file changed, 33 insertions(+), 16 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct OptContext { TCGContext *tcg; TCGOp *prev_mb; TCGTempSet temps_used; + + /* In flight values from optimization. */ + uint64_t z_mask; } OptContext; static inline TempOptInfo *ts_info(TCGTemp *ts) @@ -XXX,XX +XXX,XX @@ static void copy_propagate(OptContext *ctx, TCGOp *op, } } +static void finish_folding(OptContext *ctx, TCGOp *op) +{ + const TCGOpDef *def = &tcg_op_defs[op->opc]; + int i, nb_oargs; + + /* + * For an opcode that ends a BB, reset all temp data. + * We do no cross-BB optimization. + */ + if (def->flags & TCG_OPF_BB_END) { + memset(&ctx->temps_used, 0, sizeof(ctx->temps_used)); + ctx->prev_mb = NULL; + return; + } + + nb_oargs = def->nb_oargs; + for (i = 0; i < nb_oargs; i++) { + reset_temp(op->args[i]); + /* + * Save the corresponding known-zero bits mask for the + * first output argument (only one supported so far). + */ + if (i == 0) { + arg_info(op->args[i])->z_mask = ctx->z_mask; + } + } +} + static bool fold_call(OptContext *ctx, TCGOp *op) { TCGContext *s = ctx->tcg; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) partmask &= 0xffffffffu; affected &= 0xffffffffu; } + ctx.z_mask = z_mask; if (partmask == 0) { tcg_opt_gen_movi(&ctx, op, op->args[0], 0); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - /* Some of the folding above can change opc. */ - opc = op->opc; - def = &tcg_op_defs[opc]; - if (def->flags & TCG_OPF_BB_END) { - memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); - } else { - int nb_oargs = def->nb_oargs; - for (i = 0; i < nb_oargs; i++) { - reset_temp(op->args[i]); - /* Save the corresponding known-zero bits mask for the - first output argument (only one supported so far). */ - if (i == 0) { - arg_info(op->args[i])->z_mask = z_mask; - } - } - } + finish_folding(&ctx, op); /* Eliminate duplicate and redundant fence instructions. */ if (ctx.prev_mb) { -- 2.25.1 Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) uint64_t z_mask, partmask, affected, tmp; TCGOpcode opc = op->opc; const TCGOpDef *def; + bool done = false; /* Calls are special. */ if (opc == INDEX_op_call) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) allocator where needed and possible. Also detect copies. */ switch (opc) { CASE_OP_32_64_VEC(mov): - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); - continue; + done = tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); + break; case INDEX_op_dup_vec: if (arg_is_const(op->args[1])) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - finish_folding(&ctx, op); + if (!done) { + finish_folding(&ctx, op); + } /* Eliminate duplicate and redundant fence instructions. */ if (ctx.prev_mb) { -- 2.25.1 This puts the separate mb optimization into the same framework as the others. While fold_qemu_{ld,st} are currently identical, that won't last as more code gets moved. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 89 +++++++++++++++++++++++++++++--------------------- 1 file changed, 51 insertions(+), 38 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_call(OptContext *ctx, TCGOp *op) return true; } +static bool fold_mb(OptContext *ctx, TCGOp *op) +{ + /* Eliminate duplicate and redundant fence instructions. */ + if (ctx->prev_mb) { + /* + * Merge two barriers of the same type into one, + * or a weaker barrier into a stronger one, + * or two weaker barriers into a stronger one. + * mb X; mb Y => mb X|Y + * mb; strl => mb; st + * ldaq; mb => ld; mb + * ldaq; strl => ld; mb; st + * Other combinations are also merged into a strong + * barrier. This is stricter than specified but for + * the purposes of TCG is better than not optimizing. + */ + ctx->prev_mb->args[0] |= op->args[0]; + tcg_op_remove(ctx->tcg, op); + } else { + ctx->prev_mb = op; + } + return true; +} + +static bool fold_qemu_ld(OptContext *ctx, TCGOp *op) +{ + /* Opcodes that touch guest memory stop the mb optimization. */ + ctx->prev_mb = NULL; + return false; +} + +static bool fold_qemu_st(OptContext *ctx, TCGOp *op) +{ + /* Opcodes that touch guest memory stop the mb optimization. */ + ctx->prev_mb = NULL; + return false; +} + /* Propagate constants and copies, fold constant expressions. */ void tcg_optimize(TCGContext *s) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; + case INDEX_op_mb: + done = fold_mb(&ctx, op); + break; + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: + done = fold_qemu_ld(&ctx, op); + break; + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st8_i32: + case INDEX_op_qemu_st_i64: + done = fold_qemu_st(&ctx, op); + break; + default: break; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (!done) { finish_folding(&ctx, op); } - - /* Eliminate duplicate and redundant fence instructions. */ - if (ctx.prev_mb) { - switch (opc) { - case INDEX_op_mb: - /* Merge two barriers of the same type into one, - * or a weaker barrier into a stronger one, - * or two weaker barriers into a stronger one. - * mb X; mb Y => mb X|Y - * mb; strl => mb; st - * ldaq; mb => ld; mb - * ldaq; strl => ld; mb; st - * Other combinations are also merged into a strong - * barrier. This is stricter than specified but for - * the purposes of TCG is better than not optimizing. - */ - ctx.prev_mb->args[0] |= op->args[0]; - tcg_op_remove(s, op); - break; - - default: - /* Opcodes that end the block stop the optimization. */ - if ((def->flags & TCG_OPF_BB_END) == 0) { - break; - } - /* fallthru */ - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st8_i32: - case INDEX_op_qemu_st_i64: - /* Opcodes that touch guest memory stop the optimization. */ - ctx.prev_mb = NULL; - break; - } - } else if (opc == INDEX_op_mb) { - ctx.prev_mb = op; - } } } -- 2.25.1 Split out a whole bunch of placeholder functions, which are currently identical. That won't last as more code gets moved. Use CASE_32_64_VEC for some logical operators that previously missed the addition of vectors. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 271 +++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 219 insertions(+), 52 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static void finish_folding(OptContext *ctx, TCGOp *op) } } +/* + * The fold_* functions return true when processing is complete, + * usually by folding the operation to a constant or to a copy, + * and calling tcg_opt_gen_{mov,movi}. They may do other things, + * like collect information about the value produced, for use in + * optimizing a subsequent operation. + * + * These first fold_* functions are all helpers, used by other + * folders for more specific operations. + */ + +static bool fold_const1(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1])) { + uint64_t t; + + t = arg_info(op->args[1])->val; + t = do_constant_folding(op->opc, t, 0); + return tcg_opt_gen_movi(ctx, op, op->args[0], t); + } + return false; +} + +static bool fold_const2(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { + uint64_t t1 = arg_info(op->args[1])->val; + uint64_t t2 = arg_info(op->args[2])->val; + + t1 = do_constant_folding(op->opc, t1, t2); + return tcg_opt_gen_movi(ctx, op, op->args[0], t1); + } + return false; +} + +/* + * These outermost fold_<op> functions are sorted alphabetically. + */ + +static bool fold_add(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_and(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_andc(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + static bool fold_call(OptContext *ctx, TCGOp *op) { TCGContext *s = ctx->tcg; @@ -XXX,XX +XXX,XX @@ static bool fold_call(OptContext *ctx, TCGOp *op) return true; } +static bool fold_ctpop(OptContext *ctx, TCGOp *op) +{ + return fold_const1(ctx, op); +} + +static bool fold_divide(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_eqv(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_exts(OptContext *ctx, TCGOp *op) +{ + return fold_const1(ctx, op); +} + +static bool fold_extu(OptContext *ctx, TCGOp *op) +{ + return fold_const1(ctx, op); +} + static bool fold_mb(OptContext *ctx, TCGOp *op) { /* Eliminate duplicate and redundant fence instructions. */ @@ -XXX,XX +XXX,XX @@ static bool fold_mb(OptContext *ctx, TCGOp *op) return true; } +static bool fold_mul(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_mul_highpart(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_nand(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_neg(OptContext *ctx, TCGOp *op) +{ + return fold_const1(ctx, op); +} + +static bool fold_nor(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_not(OptContext *ctx, TCGOp *op) +{ + return fold_const1(ctx, op); +} + +static bool fold_or(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_orc(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + static bool fold_qemu_ld(OptContext *ctx, TCGOp *op) { /* Opcodes that touch guest memory stop the mb optimization. */ @@ -XXX,XX +XXX,XX @@ static bool fold_qemu_st(OptContext *ctx, TCGOp *op) return false; } +static bool fold_remainder(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_shift(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_sub(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + +static bool fold_xor(OptContext *ctx, TCGOp *op) +{ + return fold_const2(ctx, op); +} + /* Propagate constants and copies, fold constant expressions. */ void tcg_optimize(TCGContext *s) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(not): - CASE_OP_32_64(neg): - CASE_OP_32_64(ext8s): - CASE_OP_32_64(ext8u): - CASE_OP_32_64(ext16s): - CASE_OP_32_64(ext16u): - CASE_OP_32_64(ctpop): - case INDEX_op_ext32s_i64: - case INDEX_op_ext32u_i64: - case INDEX_op_ext_i32_i64: - case INDEX_op_extu_i32_i64: - case INDEX_op_extrl_i64_i32: - case INDEX_op_extrh_i64_i32: - if (arg_is_const(op->args[1])) { - tmp = do_constant_folding(opc, arg_info(op->args[1])->val, 0); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - CASE_OP_32_64(bswap16): CASE_OP_32_64(bswap32): case INDEX_op_bswap64_i64: @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(add): - CASE_OP_32_64(sub): - CASE_OP_32_64(mul): - CASE_OP_32_64(or): - CASE_OP_32_64(and): - CASE_OP_32_64(xor): - CASE_OP_32_64(shl): - CASE_OP_32_64(shr): - CASE_OP_32_64(sar): - CASE_OP_32_64(rotl): - CASE_OP_32_64(rotr): - CASE_OP_32_64(andc): - CASE_OP_32_64(orc): - CASE_OP_32_64(eqv): - CASE_OP_32_64(nand): - CASE_OP_32_64(nor): - CASE_OP_32_64(muluh): - CASE_OP_32_64(mulsh): - CASE_OP_32_64(div): - CASE_OP_32_64(divu): - CASE_OP_32_64(rem): - CASE_OP_32_64(remu): - if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - tmp = do_constant_folding(opc, arg_info(op->args[1])->val, - arg_info(op->args[2])->val); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - CASE_OP_32_64(clz): CASE_OP_32_64(ctz): if (arg_is_const(op->args[1])) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; + default: + break; + + /* ---------------------------------------------------------- */ + /* Sorted alphabetically by opcode as much as possible. */ + + CASE_OP_32_64_VEC(add): + done = fold_add(&ctx, op); + break; + CASE_OP_32_64_VEC(and): + done = fold_and(&ctx, op); + break; + CASE_OP_32_64_VEC(andc): + done = fold_andc(&ctx, op); + break; + CASE_OP_32_64(ctpop): + done = fold_ctpop(&ctx, op); + break; + CASE_OP_32_64(div): + CASE_OP_32_64(divu): + done = fold_divide(&ctx, op); + break; + CASE_OP_32_64(eqv): + done = fold_eqv(&ctx, op); + break; + CASE_OP_32_64(ext8s): + CASE_OP_32_64(ext16s): + case INDEX_op_ext32s_i64: + case INDEX_op_ext_i32_i64: + done = fold_exts(&ctx, op); + break; + CASE_OP_32_64(ext8u): + CASE_OP_32_64(ext16u): + case INDEX_op_ext32u_i64: + case INDEX_op_extu_i32_i64: + case INDEX_op_extrl_i64_i32: + case INDEX_op_extrh_i64_i32: + done = fold_extu(&ctx, op); + break; case INDEX_op_mb: done = fold_mb(&ctx, op); break; + CASE_OP_32_64(mul): + done = fold_mul(&ctx, op); + break; + CASE_OP_32_64(mulsh): + CASE_OP_32_64(muluh): + done = fold_mul_highpart(&ctx, op); + break; + CASE_OP_32_64(nand): + done = fold_nand(&ctx, op); + break; + CASE_OP_32_64(neg): + done = fold_neg(&ctx, op); + break; + CASE_OP_32_64(nor): + done = fold_nor(&ctx, op); + break; + CASE_OP_32_64_VEC(not): + done = fold_not(&ctx, op); + break; + CASE_OP_32_64_VEC(or): + done = fold_or(&ctx, op); + break; + CASE_OP_32_64_VEC(orc): + done = fold_orc(&ctx, op); + break; case INDEX_op_qemu_ld_i32: case INDEX_op_qemu_ld_i64: done = fold_qemu_ld(&ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_qemu_st_i64: done = fold_qemu_st(&ctx, op); break; - - default: + CASE_OP_32_64(rem): + CASE_OP_32_64(remu): + done = fold_remainder(&ctx, op); + break; + CASE_OP_32_64(rotl): + CASE_OP_32_64(rotr): + CASE_OP_32_64(sar): + CASE_OP_32_64(shl): + CASE_OP_32_64(shr): + done = fold_shift(&ctx, op); + break; + CASE_OP_32_64_VEC(sub): + done = fold_sub(&ctx, op); + break; + CASE_OP_32_64_VEC(xor): + done = fold_xor(&ctx, op); break; } -- 2.25.1 Reduce some code duplication by folding the NE and EQ cases. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 145 ++++++++++++++++++++++++------------------------- 1 file changed, 72 insertions(+), 73 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_remainder(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_setcond2(OptContext *ctx, TCGOp *op) +{ + TCGCond cond = op->args[5]; + int i = do_constant_folding_cond2(&op->args[1], &op->args[3], cond); + int inv = 0; + + if (i >= 0) { + goto do_setcond_const; + } + + switch (cond) { + case TCG_COND_LT: + case TCG_COND_GE: + /* + * Simplify LT/GE comparisons vs zero to a single compare + * vs the high word of the input. + */ + if (arg_is_const(op->args[3]) && arg_info(op->args[3])->val == 0 && + arg_is_const(op->args[4]) && arg_info(op->args[4])->val == 0) { + goto do_setcond_high; + } + break; + + case TCG_COND_NE: + inv = 1; + QEMU_FALLTHROUGH; + case TCG_COND_EQ: + /* + * Simplify EQ/NE comparisons where one of the pairs + * can be simplified. + */ + i = do_constant_folding_cond(INDEX_op_setcond_i32, op->args[1], + op->args[3], cond); + switch (i ^ inv) { + case 0: + goto do_setcond_const; + case 1: + goto do_setcond_high; + } + + i = do_constant_folding_cond(INDEX_op_setcond_i32, op->args[2], + op->args[4], cond); + switch (i ^ inv) { + case 0: + goto do_setcond_const; + case 1: + op->args[2] = op->args[3]; + op->args[3] = cond; + op->opc = INDEX_op_setcond_i32; + break; + } + break; + + default: + break; + + do_setcond_high: + op->args[1] = op->args[2]; + op->args[2] = op->args[4]; + op->args[3] = cond; + op->opc = INDEX_op_setcond_i32; + break; + } + return false; + + do_setcond_const: + return tcg_opt_gen_movi(ctx, op, op->args[0], i); +} + static bool fold_shift(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - case INDEX_op_setcond2_i32: - i = do_constant_folding_cond2(&op->args[1], &op->args[3], - op->args[5]); - if (i >= 0) { - do_setcond_const: - tcg_opt_gen_movi(&ctx, op, op->args[0], i); - continue; - } - if ((op->args[5] == TCG_COND_LT || op->args[5] == TCG_COND_GE) - && arg_is_const(op->args[3]) - && arg_info(op->args[3])->val == 0 - && arg_is_const(op->args[4]) - && arg_info(op->args[4])->val == 0) { - /* Simplify LT/GE comparisons vs zero to a single compare - vs the high word of the input. */ - do_setcond_high: - reset_temp(op->args[0]); - arg_info(op->args[0])->z_mask = 1; - op->opc = INDEX_op_setcond_i32; - op->args[1] = op->args[2]; - op->args[2] = op->args[4]; - op->args[3] = op->args[5]; - break; - } - if (op->args[5] == TCG_COND_EQ) { - /* Simplify EQ comparisons where one of the pairs - can be simplified. */ - i = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[1], op->args[3], - TCG_COND_EQ); - if (i == 0) { - goto do_setcond_const; - } else if (i > 0) { - goto do_setcond_high; - } - i = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[2], op->args[4], - TCG_COND_EQ); - if (i == 0) { - goto do_setcond_high; - } else if (i < 0) { - break; - } - do_setcond_low: - reset_temp(op->args[0]); - arg_info(op->args[0])->z_mask = 1; - op->opc = INDEX_op_setcond_i32; - op->args[2] = op->args[3]; - op->args[3] = op->args[5]; - break; - } - if (op->args[5] == TCG_COND_NE) { - /* Simplify NE comparisons where one of the pairs - can be simplified. */ - i = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[1], op->args[3], - TCG_COND_NE); - if (i == 0) { - goto do_setcond_high; - } else if (i > 0) { - goto do_setcond_const; - } - i = do_constant_folding_cond(INDEX_op_setcond_i32, - op->args[2], op->args[4], - TCG_COND_NE); - if (i == 0) { - goto do_setcond_low; - } else if (i > 0) { - goto do_setcond_const; - } - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(shr): done = fold_shift(&ctx, op); break; + case INDEX_op_setcond2_i32: + done = fold_setcond2(&ctx, op); + break; CASE_OP_32_64_VEC(sub): done = fold_sub(&ctx, op); break; -- 2.25.1 Reduce some code duplication by folding the NE and EQ cases. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 159 +++++++++++++++++++++++++------------------------ 1 file changed, 81 insertions(+), 78 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_andc(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_brcond2(OptContext *ctx, TCGOp *op) +{ + TCGCond cond = op->args[4]; + int i = do_constant_folding_cond2(&op->args[0], &op->args[2], cond); + TCGArg label = op->args[5]; + int inv = 0; + + if (i >= 0) { + goto do_brcond_const; + } + + switch (cond) { + case TCG_COND_LT: + case TCG_COND_GE: + /* + * Simplify LT/GE comparisons vs zero to a single compare + * vs the high word of the input. + */ + if (arg_is_const(op->args[2]) && arg_info(op->args[2])->val == 0 && + arg_is_const(op->args[3]) && arg_info(op->args[3])->val == 0) { + goto do_brcond_high; + } + break; + + case TCG_COND_NE: + inv = 1; + QEMU_FALLTHROUGH; + case TCG_COND_EQ: + /* + * Simplify EQ/NE comparisons where one of the pairs + * can be simplified. + */ + i = do_constant_folding_cond(INDEX_op_brcond_i32, op->args[0], + op->args[2], cond); + switch (i ^ inv) { + case 0: + goto do_brcond_const; + case 1: + goto do_brcond_high; + } + + i = do_constant_folding_cond(INDEX_op_brcond_i32, op->args[1], + op->args[3], cond); + switch (i ^ inv) { + case 0: + goto do_brcond_const; + case 1: + op->opc = INDEX_op_brcond_i32; + op->args[1] = op->args[2]; + op->args[2] = cond; + op->args[3] = label; + break; + } + break; + + default: + break; + + do_brcond_high: + op->opc = INDEX_op_brcond_i32; + op->args[0] = op->args[1]; + op->args[1] = op->args[3]; + op->args[2] = cond; + op->args[3] = label; + break; + + do_brcond_const: + if (i == 0) { + tcg_op_remove(ctx->tcg, op); + return true; + } + op->opc = INDEX_op_br; + op->args[0] = label; + break; + } + return false; +} + static bool fold_call(OptContext *ctx, TCGOp *op) { TCGContext *s = ctx->tcg; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - case INDEX_op_brcond2_i32: - i = do_constant_folding_cond2(&op->args[0], &op->args[2], - op->args[4]); - if (i == 0) { - do_brcond_false: - tcg_op_remove(s, op); - continue; - } - if (i > 0) { - do_brcond_true: - op->opc = opc = INDEX_op_br; - op->args[0] = op->args[5]; - break; - } - if ((op->args[4] == TCG_COND_LT || op->args[4] == TCG_COND_GE) - && arg_is_const(op->args[2]) - && arg_info(op->args[2])->val == 0 - && arg_is_const(op->args[3]) - && arg_info(op->args[3])->val == 0) { - /* Simplify LT/GE comparisons vs zero to a single compare - vs the high word of the input. */ - do_brcond_high: - op->opc = opc = INDEX_op_brcond_i32; - op->args[0] = op->args[1]; - op->args[1] = op->args[3]; - op->args[2] = op->args[4]; - op->args[3] = op->args[5]; - break; - } - if (op->args[4] == TCG_COND_EQ) { - /* Simplify EQ comparisons where one of the pairs - can be simplified. */ - i = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[0], op->args[2], - TCG_COND_EQ); - if (i == 0) { - goto do_brcond_false; - } else if (i > 0) { - goto do_brcond_high; - } - i = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[1], op->args[3], - TCG_COND_EQ); - if (i == 0) { - goto do_brcond_false; - } else if (i < 0) { - break; - } - do_brcond_low: - memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); - op->opc = INDEX_op_brcond_i32; - op->args[1] = op->args[2]; - op->args[2] = op->args[4]; - op->args[3] = op->args[5]; - break; - } - if (op->args[4] == TCG_COND_NE) { - /* Simplify NE comparisons where one of the pairs - can be simplified. */ - i = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[0], op->args[2], - TCG_COND_NE); - if (i == 0) { - goto do_brcond_high; - } else if (i > 0) { - goto do_brcond_true; - } - i = do_constant_folding_cond(INDEX_op_brcond_i32, - op->args[1], op->args[3], - TCG_COND_NE); - if (i == 0) { - goto do_brcond_low; - } else if (i > 0) { - goto do_brcond_true; - } - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(andc): done = fold_andc(&ctx, op); break; + case INDEX_op_brcond2_i32: + done = fold_brcond2(&ctx, op); + break; CASE_OP_32_64(ctpop): done = fold_ctpop(&ctx, op); break; -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 33 +++++++++++++++++++-------------- 1 file changed, 19 insertions(+), 14 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_andc(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_brcond(OptContext *ctx, TCGOp *op) +{ + TCGCond cond = op->args[2]; + int i = do_constant_folding_cond(op->opc, op->args[0], op->args[1], cond); + + if (i == 0) { + tcg_op_remove(ctx->tcg, op); + return true; + } + if (i > 0) { + op->opc = INDEX_op_br; + op->args[0] = op->args[3]; + } + return false; +} + static bool fold_brcond2(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[4]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(brcond): - i = do_constant_folding_cond(opc, op->args[0], - op->args[1], op->args[2]); - if (i == 0) { - tcg_op_remove(s, op); - continue; - } else if (i > 0) { - memset(&ctx.temps_used, 0, sizeof(ctx.temps_used)); - op->opc = opc = INDEX_op_br; - op->args[0] = op->args[3]; - break; - } - break; - CASE_OP_32_64(movcond): i = do_constant_folding_cond(opc, op->args[1], op->args[2], op->args[5]); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(andc): done = fold_andc(&ctx, op); break; + CASE_OP_32_64(brcond): + done = fold_brcond(&ctx, op); + break; case INDEX_op_brcond2_i32: done = fold_brcond2(&ctx, op); break; -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 23 ++++++++++++++--------- 1 file changed, 14 insertions(+), 9 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_remainder(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_setcond(OptContext *ctx, TCGOp *op) +{ + TCGCond cond = op->args[3]; + int i = do_constant_folding_cond(op->opc, op->args[1], op->args[2], cond); + + if (i >= 0) { + return tcg_opt_gen_movi(ctx, op, op->args[0], i); + } + return false; +} + static bool fold_setcond2(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[5]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(setcond): - i = do_constant_folding_cond(opc, op->args[1], - op->args[2], op->args[3]); - if (i >= 0) { - tcg_opt_gen_movi(&ctx, op, op->args[0], i); - continue; - } - break; - CASE_OP_32_64(movcond): i = do_constant_folding_cond(opc, op->args[1], op->args[2], op->args[5]); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(shr): done = fold_shift(&ctx, op); break; + CASE_OP_32_64(setcond): + done = fold_setcond(&ctx, op); + break; case INDEX_op_setcond2_i32: done = fold_setcond2(&ctx, op); break; -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 37 +++++++++++++++++++++---------------- 1 file changed, 21 insertions(+), 16 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_mul_highpart(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_mulu2_i32(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[2]) && arg_is_const(op->args[3])) { + uint32_t a = arg_info(op->args[2])->val; + uint32_t b = arg_info(op->args[3])->val; + uint64_t r = (uint64_t)a * b; + TCGArg rl, rh; + TCGOp *op2 = tcg_op_insert_before(ctx->tcg, op, INDEX_op_mov_i32); + + rl = op->args[0]; + rh = op->args[1]; + tcg_opt_gen_movi(ctx, op, rl, (int32_t)r); + tcg_opt_gen_movi(ctx, op2, rh, (int32_t)(r >> 32)); + return true; + } + return false; +} + static bool fold_nand(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - case INDEX_op_mulu2_i32: - if (arg_is_const(op->args[2]) && arg_is_const(op->args[3])) { - uint32_t a = arg_info(op->args[2])->val; - uint32_t b = arg_info(op->args[3])->val; - uint64_t r = (uint64_t)a * b; - TCGArg rl, rh; - TCGOp *op2 = tcg_op_insert_before(s, op, INDEX_op_mov_i32); - - rl = op->args[0]; - rh = op->args[1]; - tcg_opt_gen_movi(&ctx, op, rl, (int32_t)r); - tcg_opt_gen_movi(&ctx, op2, rh, (int32_t)(r >> 32)); - continue; - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(muluh): done = fold_mul_highpart(&ctx, op); break; + case INDEX_op_mulu2_i32: + done = fold_mulu2_i32(&ctx, op); + break; CASE_OP_32_64(nand): done = fold_nand(&ctx, op); break; -- 2.25.1 Add two additional helpers, fold_add2_i32 and fold_sub2_i32 which will not be simple wrappers forever. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 70 +++++++++++++++++++++++++++++++------------------- 1 file changed, 44 insertions(+), 26 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_add(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_addsub2_i32(OptContext *ctx, TCGOp *op, bool add) +{ + if (arg_is_const(op->args[2]) && arg_is_const(op->args[3]) && + arg_is_const(op->args[4]) && arg_is_const(op->args[5])) { + uint32_t al = arg_info(op->args[2])->val; + uint32_t ah = arg_info(op->args[3])->val; + uint32_t bl = arg_info(op->args[4])->val; + uint32_t bh = arg_info(op->args[5])->val; + uint64_t a = ((uint64_t)ah << 32) | al; + uint64_t b = ((uint64_t)bh << 32) | bl; + TCGArg rl, rh; + TCGOp *op2 = tcg_op_insert_before(ctx->tcg, op, INDEX_op_mov_i32); + + if (add) { + a += b; + } else { + a -= b; + } + + rl = op->args[0]; + rh = op->args[1]; + tcg_opt_gen_movi(ctx, op, rl, (int32_t)a); + tcg_opt_gen_movi(ctx, op2, rh, (int32_t)(a >> 32)); + return true; + } + return false; +} + +static bool fold_add2_i32(OptContext *ctx, TCGOp *op) +{ + return fold_addsub2_i32(ctx, op, true); +} + static bool fold_and(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ static bool fold_sub(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_sub2_i32(OptContext *ctx, TCGOp *op) +{ + return fold_addsub2_i32(ctx, op, false); +} + static bool fold_xor(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - case INDEX_op_add2_i32: - case INDEX_op_sub2_i32: - if (arg_is_const(op->args[2]) && arg_is_const(op->args[3]) - && arg_is_const(op->args[4]) && arg_is_const(op->args[5])) { - uint32_t al = arg_info(op->args[2])->val; - uint32_t ah = arg_info(op->args[3])->val; - uint32_t bl = arg_info(op->args[4])->val; - uint32_t bh = arg_info(op->args[5])->val; - uint64_t a = ((uint64_t)ah << 32) | al; - uint64_t b = ((uint64_t)bh << 32) | bl; - TCGArg rl, rh; - TCGOp *op2 = tcg_op_insert_before(s, op, INDEX_op_mov_i32); - - if (opc == INDEX_op_add2_i32) { - a += b; - } else { - a -= b; - } - - rl = op->args[0]; - rh = op->args[1]; - tcg_opt_gen_movi(&ctx, op, rl, (int32_t)a); - tcg_opt_gen_movi(&ctx, op2, rh, (int32_t)(a >> 32)); - continue; - } - break; default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(add): done = fold_add(&ctx, op); break; + case INDEX_op_add2_i32: + done = fold_add2_i32(&ctx, op); + break; CASE_OP_32_64_VEC(and): done = fold_and(&ctx, op); break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(sub): done = fold_sub(&ctx, op); break; + case INDEX_op_sub2_i32: + done = fold_sub2_i32(&ctx, op); + break; CASE_OP_32_64_VEC(xor): done = fold_xor(&ctx, op); break; -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 56 ++++++++++++++++++++++++++++---------------------- 1 file changed, 31 insertions(+), 25 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_mb(OptContext *ctx, TCGOp *op) return true; } +static bool fold_movcond(OptContext *ctx, TCGOp *op) +{ + TCGOpcode opc = op->opc; + TCGCond cond = op->args[5]; + int i = do_constant_folding_cond(opc, op->args[1], op->args[2], cond); + + if (i >= 0) { + return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[4 - i]); + } + + if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { + uint64_t tv = arg_info(op->args[3])->val; + uint64_t fv = arg_info(op->args[4])->val; + + opc = (opc == INDEX_op_movcond_i32 + ? INDEX_op_setcond_i32 : INDEX_op_setcond_i64); + + if (tv == 1 && fv == 0) { + op->opc = opc; + op->args[3] = cond; + } else if (fv == 1 && tv == 0) { + op->opc = opc; + op->args[3] = tcg_invert_cond(cond); + } + } + return false; +} + static bool fold_mul(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(movcond): - i = do_constant_folding_cond(opc, op->args[1], - op->args[2], op->args[5]); - if (i >= 0) { - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[4 - i]); - continue; - } - if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { - uint64_t tv = arg_info(op->args[3])->val; - uint64_t fv = arg_info(op->args[4])->val; - TCGCond cond = op->args[5]; - - if (fv == 1 && tv == 0) { - cond = tcg_invert_cond(cond); - } else if (!(tv == 1 && fv == 0)) { - break; - } - op->args[3] = cond; - op->opc = opc = (opc == INDEX_op_movcond_i32 - ? INDEX_op_setcond_i32 - : INDEX_op_setcond_i64); - } - break; - - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_mb: done = fold_mb(&ctx, op); break; + CASE_OP_32_64(movcond): + done = fold_movcond(&ctx, op); + break; CASE_OP_32_64(mul): done = fold_mul(&ctx, op); break; -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 39 ++++++++++++++++++++++----------------- 1 file changed, 22 insertions(+), 17 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_eqv(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_extract2(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { + uint64_t v1 = arg_info(op->args[1])->val; + uint64_t v2 = arg_info(op->args[2])->val; + int shr = op->args[3]; + + if (op->opc == INDEX_op_extract2_i64) { + v1 >>= shr; + v2 <<= 64 - shr; + } else { + v1 = (uint32_t)v1 >> shr; + v2 = (int32_t)v2 << (32 - shr); + } + return tcg_opt_gen_movi(ctx, op, op->args[0], v1 | v2); + } + return false; +} + static bool fold_exts(OptContext *ctx, TCGOp *op) { return fold_const1(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(extract2): - if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - uint64_t v1 = arg_info(op->args[1])->val; - uint64_t v2 = arg_info(op->args[2])->val; - int shr = op->args[3]; - - if (opc == INDEX_op_extract2_i64) { - tmp = (v1 >> shr) | (v2 << (64 - shr)); - } else { - tmp = (int32_t)(((uint32_t)v1 >> shr) | - ((uint32_t)v2 << (32 - shr))); - } - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(eqv): done = fold_eqv(&ctx, op); break; + CASE_OP_32_64(extract2): + done = fold_extract2(&ctx, op); + break; CASE_OP_32_64(ext8s): CASE_OP_32_64(ext16s): case INDEX_op_ext32s_i64: -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 48 ++++++++++++++++++++++++++++++------------------ 1 file changed, 30 insertions(+), 18 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_eqv(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_extract(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1])) { + uint64_t t; + + t = arg_info(op->args[1])->val; + t = extract64(t, op->args[2], op->args[3]); + return tcg_opt_gen_movi(ctx, op, op->args[0], t); + } + return false; +} + static bool fold_extract2(OptContext *ctx, TCGOp *op) { if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { @@ -XXX,XX +XXX,XX @@ static bool fold_setcond2(OptContext *ctx, TCGOp *op) return tcg_opt_gen_movi(ctx, op, op->args[0], i); } +static bool fold_sextract(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1])) { + uint64_t t; + + t = arg_info(op->args[1])->val; + t = sextract64(t, op->args[2], op->args[3]); + return tcg_opt_gen_movi(ctx, op, op->args[0], t); + } + return false; +} + static bool fold_shift(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(extract): - if (arg_is_const(op->args[1])) { - tmp = extract64(arg_info(op->args[1])->val, - op->args[2], op->args[3]); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - - CASE_OP_32_64(sextract): - if (arg_is_const(op->args[1])) { - tmp = sextract64(arg_info(op->args[1])->val, - op->args[2], op->args[3]); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(eqv): done = fold_eqv(&ctx, op); break; + CASE_OP_32_64(extract): + done = fold_extract(&ctx, op); + break; CASE_OP_32_64(extract2): done = fold_extract2(&ctx, op); break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_setcond2_i32: done = fold_setcond2(&ctx, op); break; + CASE_OP_32_64(sextract): + done = fold_sextract(&ctx, op); + break; CASE_OP_32_64_VEC(sub): done = fold_sub(&ctx, op); break; -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 25 +++++++++++++++---------- 1 file changed, 15 insertions(+), 10 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_ctpop(OptContext *ctx, TCGOp *op) return fold_const1(ctx, op); } +static bool fold_deposit(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { + uint64_t t1 = arg_info(op->args[1])->val; + uint64_t t2 = arg_info(op->args[2])->val; + + t1 = deposit64(t1, op->args[3], op->args[4], t2); + return tcg_opt_gen_movi(ctx, op, op->args[0], t1); + } + return false; +} + static bool fold_divide(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(deposit): - if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - tmp = deposit64(arg_info(op->args[1])->val, - op->args[3], op->args[4], - arg_info(op->args[2])->val); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(ctpop): done = fold_ctpop(&ctx, op); break; + CASE_OP_32_64(deposit): + done = fold_deposit(&ctx, op); + break; CASE_OP_32_64(div): CASE_OP_32_64(divu): done = fold_divide(&ctx, op); -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 32 ++++++++++++++++++-------------- 1 file changed, 18 insertions(+), 14 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_call(OptContext *ctx, TCGOp *op) return true; } +static bool fold_count_zeros(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1])) { + uint64_t t = arg_info(op->args[1])->val; + + if (t != 0) { + t = do_constant_folding(op->opc, t, 0); + return tcg_opt_gen_movi(ctx, op, op->args[0], t); + } + return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[2]); + } + return false; +} + static bool fold_ctpop(OptContext *ctx, TCGOp *op) { return fold_const1(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(clz): - CASE_OP_32_64(ctz): - if (arg_is_const(op->args[1])) { - TCGArg v = arg_info(op->args[1])->val; - if (v != 0) { - tmp = do_constant_folding(opc, v, 0); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - } else { - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[2]); - } - continue; - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_brcond2_i32: done = fold_brcond2(&ctx, op); break; + CASE_OP_32_64(clz): + CASE_OP_32_64(ctz): + done = fold_count_zeros(&ctx, op); + break; CASE_OP_32_64(ctpop): done = fold_ctpop(&ctx, op); break; -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 27 ++++++++++++++++----------- 1 file changed, 16 insertions(+), 11 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_brcond2(OptContext *ctx, TCGOp *op) return false; } +static bool fold_bswap(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1])) { + uint64_t t = arg_info(op->args[1])->val; + + t = do_constant_folding(op->opc, t, op->args[2]); + return tcg_opt_gen_movi(ctx, op, op->args[0], t); + } + return false; +} + static bool fold_call(OptContext *ctx, TCGOp *op) { TCGContext *s = ctx->tcg; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } break; - CASE_OP_32_64(bswap16): - CASE_OP_32_64(bswap32): - case INDEX_op_bswap64_i64: - if (arg_is_const(op->args[1])) { - tmp = do_constant_folding(opc, arg_info(op->args[1])->val, - op->args[2]); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_brcond2_i32: done = fold_brcond2(&ctx, op); break; + CASE_OP_32_64(bswap16): + CASE_OP_32_64(bswap32): + case INDEX_op_bswap64_i64: + done = fold_bswap(&ctx, op); + break; CASE_OP_32_64(clz): CASE_OP_32_64(ctz): done = fold_count_zeros(&ctx, op); -- 2.25.1 Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 53 +++++++++++++++++++++++++++++--------------------- 1 file changed, 31 insertions(+), 22 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_divide(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_dup(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1])) { + uint64_t t = arg_info(op->args[1])->val; + t = dup_const(TCGOP_VECE(op), t); + return tcg_opt_gen_movi(ctx, op, op->args[0], t); + } + return false; +} + +static bool fold_dup2(OptContext *ctx, TCGOp *op) +{ + if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { + uint64_t t = deposit64(arg_info(op->args[1])->val, 32, 32, + arg_info(op->args[2])->val); + return tcg_opt_gen_movi(ctx, op, op->args[0], t); + } + + if (args_are_copies(op->args[1], op->args[2])) { + op->opc = INDEX_op_dup_vec; + TCGOP_VECE(op) = MO_32; + } + return false; +} + static bool fold_eqv(OptContext *ctx, TCGOp *op) { return fold_const2(ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) done = tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); break; - case INDEX_op_dup_vec: - if (arg_is_const(op->args[1])) { - tmp = arg_info(op->args[1])->val; - tmp = dup_const(TCGOP_VECE(op), tmp); - tcg_opt_gen_movi(&ctx, op, op->args[0], tmp); - continue; - } - break; - - case INDEX_op_dup2_vec: - assert(TCG_TARGET_REG_BITS == 32); - if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - tcg_opt_gen_movi(&ctx, op, op->args[0], - deposit64(arg_info(op->args[1])->val, 32, 32, - arg_info(op->args[2])->val)); - continue; - } else if (args_are_copies(op->args[1], op->args[2])) { - op->opc = INDEX_op_dup_vec; - TCGOP_VECE(op) = MO_32; - } - break; - default: break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(divu): done = fold_divide(&ctx, op); break; + case INDEX_op_dup_vec: + done = fold_dup(&ctx, op); + break; + case INDEX_op_dup2_vec: + done = fold_dup2(&ctx, op); + break; CASE_OP_32_64(eqv): done = fold_eqv(&ctx, op); break; -- 2.25.1 This is the final entry in the main switch that was in a different form. After this, we have the option to convert the switch into a function dispatch table. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 27 ++++++++++++++------------- 1 file changed, 14 insertions(+), 13 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_mb(OptContext *ctx, TCGOp *op) return true; } +static bool fold_mov(OptContext *ctx, TCGOp *op) +{ + return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[1]); +} + static bool fold_movcond(OptContext *ctx, TCGOp *op) { TCGOpcode opc = op->opc; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - /* Propagate constants through copy operations and do constant - folding. Constants will be substituted to arguments by register - allocator where needed and possible. Also detect copies. */ + /* + * Process each opcode. + * Sorted alphabetically by opcode as much as possible. + */ switch (opc) { - CASE_OP_32_64_VEC(mov): - done = tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); - break; - - default: - break; - - /* ---------------------------------------------------------- */ - /* Sorted alphabetically by opcode as much as possible. */ - CASE_OP_32_64_VEC(add): done = fold_add(&ctx, op); break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_mb: done = fold_mb(&ctx, op); break; + CASE_OP_32_64_VEC(mov): + done = fold_mov(&ctx, op); + break; CASE_OP_32_64(movcond): done = fold_movcond(&ctx, op); break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(xor): done = fold_xor(&ctx, op); break; + default: + break; } if (!done) { -- 2.25.1 Pull the "op r, a, a => movi r, 0" optimization into a function, and use it in the outer opcode fold functions. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 41 ++++++++++++++++++++++++----------------- 1 file changed, 24 insertions(+), 17 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_const2(OptContext *ctx, TCGOp *op) return false; } +/* If the binary operation has both arguments equal, fold to @i. */ +static bool fold_xx_to_i(OptContext *ctx, TCGOp *op, uint64_t i) +{ + if (args_are_copies(op->args[1], op->args[2])) { + return tcg_opt_gen_movi(ctx, op, op->args[0], i); + } + return false; +} + /* * These outermost fold_<op> functions are sorted alphabetically. */ @@ -XXX,XX +XXX,XX @@ static bool fold_and(OptContext *ctx, TCGOp *op) static bool fold_andc(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xx_to_i(ctx, op, 0)) { + return true; + } + return false; } static bool fold_brcond(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_shift(OptContext *ctx, TCGOp *op) static bool fold_sub(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xx_to_i(ctx, op, 0)) { + return true; + } + return false; } static bool fold_sub2_i32(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_sub2_i32(OptContext *ctx, TCGOp *op) static bool fold_xor(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xx_to_i(ctx, op, 0)) { + return true; + } + return false; } /* Propagate constants and copies, fold constant expressions. */ @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - /* Simplify expression for "op r, a, a => movi r, 0" cases */ - switch (opc) { - CASE_OP_32_64_VEC(andc): - CASE_OP_32_64_VEC(sub): - CASE_OP_32_64_VEC(xor): - if (args_are_copies(op->args[1], op->args[2])) { - tcg_opt_gen_movi(&ctx, op, op->args[0], 0); - continue; - } - break; - default: - break; - } - /* * Process each opcode. * Sorted alphabetically by opcode as much as possible. -- 2.25.1 Pull the "op r, a, a => mov r, a" optimization into a function, and use it in the outer opcode fold functions. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 39 ++++++++++++++++++++++++--------------- 1 file changed, 24 insertions(+), 15 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_xx_to_i(OptContext *ctx, TCGOp *op, uint64_t i) return false; } +/* If the binary operation has both arguments equal, fold to identity. */ +static bool fold_xx_to_x(OptContext *ctx, TCGOp *op) +{ + if (args_are_copies(op->args[1], op->args[2])) { + return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[1]); + } + return false; +} + /* * These outermost fold_<op> functions are sorted alphabetically. + * + * The ordering of the transformations should be: + * 1) those that produce a constant + * 2) those that produce a copy + * 3) those that produce information about the result value. */ static bool fold_add(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_add2_i32(OptContext *ctx, TCGOp *op) static bool fold_and(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xx_to_x(ctx, op)) { + return true; + } + return false; } static bool fold_andc(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_not(OptContext *ctx, TCGOp *op) static bool fold_or(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xx_to_x(ctx, op)) { + return true; + } + return false; } static bool fold_orc(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - /* Simplify expression for "op r, a, a => mov r, a" cases */ - switch (opc) { - CASE_OP_32_64_VEC(or): - CASE_OP_32_64_VEC(and): - if (args_are_copies(op->args[1], op->args[2])) { - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); - continue; - } - break; - default: - break; - } - /* * Process each opcode. * Sorted alphabetically by opcode as much as possible. -- 2.25.1 Pull the "op r, a, 0 => movi r, 0" optimization into a function, and use it in the outer opcode fold functions. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 38 ++++++++++++++++++++------------------ 1 file changed, 20 insertions(+), 18 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_const2(OptContext *ctx, TCGOp *op) return false; } +/* If the binary operation has second argument @i, fold to @i. */ +static bool fold_xi_to_i(OptContext *ctx, TCGOp *op, uint64_t i) +{ + if (arg_is_const(op->args[2]) && arg_info(op->args[2])->val == i) { + return tcg_opt_gen_movi(ctx, op, op->args[0], i); + } + return false; +} + /* If the binary operation has both arguments equal, fold to @i. */ static bool fold_xx_to_i(OptContext *ctx, TCGOp *op, uint64_t i) { @@ -XXX,XX +XXX,XX @@ static bool fold_add2_i32(OptContext *ctx, TCGOp *op) static bool fold_and(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || + fold_xi_to_i(ctx, op, 0) || fold_xx_to_x(ctx, op)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_movcond(OptContext *ctx, TCGOp *op) static bool fold_mul(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_i(ctx, op, 0)) { + return true; + } + return false; } static bool fold_mul_highpart(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_i(ctx, op, 0)) { + return true; + } + return false; } static bool fold_mulu2_i32(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) continue; } - /* Simplify expression for "op r, a, 0 => movi r, 0" cases */ - switch (opc) { - CASE_OP_32_64_VEC(and): - CASE_OP_32_64_VEC(mul): - CASE_OP_32_64(muluh): - CASE_OP_32_64(mulsh): - if (arg_is_const(op->args[2]) - && arg_info(op->args[2])->val == 0) { - tcg_opt_gen_movi(&ctx, op, op->args[0], 0); - continue; - } - break; - default: - break; - } - /* * Process each opcode. * Sorted alphabetically by opcode as much as possible. -- 2.25.1 Compute the type of the operation early. There are at least 4 places that used a def->flags ladder to determine the type of the operation being optimized. There were two places that assumed !TCG_OPF_64BIT means TCG_TYPE_I32, and so could potentially compute incorrect results for vector operations. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 149 +++++++++++++++++++++++++++++-------------------- 1 file changed, 89 insertions(+), 60 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct OptContext { /* In flight values from optimization. */ uint64_t z_mask; + TCGType type; } OptContext; static inline TempOptInfo *ts_info(TCGTemp *ts) @@ -XXX,XX +XXX,XX @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) { TCGTemp *dst_ts = arg_temp(dst); TCGTemp *src_ts = arg_temp(src); - const TCGOpDef *def; TempOptInfo *di; TempOptInfo *si; uint64_t z_mask; @@ -XXX,XX +XXX,XX @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) reset_ts(dst_ts); di = ts_info(dst_ts); si = ts_info(src_ts); - def = &tcg_op_defs[op->opc]; - if (def->flags & TCG_OPF_VECTOR) { - new_op = INDEX_op_mov_vec; - } else if (def->flags & TCG_OPF_64BIT) { - new_op = INDEX_op_mov_i64; - } else { + + switch (ctx->type) { + case TCG_TYPE_I32: new_op = INDEX_op_mov_i32; + break; + case TCG_TYPE_I64: + new_op = INDEX_op_mov_i64; + break; + case TCG_TYPE_V64: + case TCG_TYPE_V128: + case TCG_TYPE_V256: + /* TCGOP_VECL and TCGOP_VECE remain unchanged. */ + new_op = INDEX_op_mov_vec; + break; + default: + g_assert_not_reached(); } op->opc = new_op; - /* TCGOP_VECL and TCGOP_VECE remain unchanged. */ op->args[0] = dst; op->args[1] = src; @@ -XXX,XX +XXX,XX @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) static bool tcg_opt_gen_movi(OptContext *ctx, TCGOp *op, TCGArg dst, uint64_t val) { - const TCGOpDef *def = &tcg_op_defs[op->opc]; - TCGType type; - TCGTemp *tv; - - if (def->flags & TCG_OPF_VECTOR) { - type = TCGOP_VECL(op) + TCG_TYPE_V64; - } else if (def->flags & TCG_OPF_64BIT) { - type = TCG_TYPE_I64; - } else { - type = TCG_TYPE_I32; - } - /* Convert movi to mov with constant temp. */ - tv = tcg_constant_internal(type, val); + TCGTemp *tv = tcg_constant_internal(ctx->type, val); + init_ts_info(ctx, tv); return tcg_opt_gen_mov(ctx, op, dst, temp_arg(tv)); } @@ -XXX,XX +XXX,XX @@ static uint64_t do_constant_folding_2(TCGOpcode op, uint64_t x, uint64_t y) } } -static uint64_t do_constant_folding(TCGOpcode op, uint64_t x, uint64_t y) +static uint64_t do_constant_folding(TCGOpcode op, TCGType type, + uint64_t x, uint64_t y) { - const TCGOpDef *def = &tcg_op_defs[op]; uint64_t res = do_constant_folding_2(op, x, y); - if (!(def->flags & TCG_OPF_64BIT)) { + if (type == TCG_TYPE_I32) { res = (int32_t)res; } return res; @@ -XXX,XX +XXX,XX @@ static bool do_constant_folding_cond_eq(TCGCond c) * Return -1 if the condition can't be simplified, * and the result of the condition (0 or 1) if it can. */ -static int do_constant_folding_cond(TCGOpcode op, TCGArg x, +static int do_constant_folding_cond(TCGType type, TCGArg x, TCGArg y, TCGCond c) { uint64_t xv = arg_info(x)->val; uint64_t yv = arg_info(y)->val; if (arg_is_const(x) && arg_is_const(y)) { - const TCGOpDef *def = &tcg_op_defs[op]; - tcg_debug_assert(!(def->flags & TCG_OPF_VECTOR)); - if (def->flags & TCG_OPF_64BIT) { - return do_constant_folding_cond_64(xv, yv, c); - } else { + switch (type) { + case TCG_TYPE_I32: return do_constant_folding_cond_32(xv, yv, c); + case TCG_TYPE_I64: + return do_constant_folding_cond_64(xv, yv, c); + default: + /* Only scalar comparisons are optimizable */ + return -1; } } else if (args_are_copies(x, y)) { return do_constant_folding_cond_eq(c); @@ -XXX,XX +XXX,XX @@ static bool fold_const1(OptContext *ctx, TCGOp *op) uint64_t t; t = arg_info(op->args[1])->val; - t = do_constant_folding(op->opc, t, 0); + t = do_constant_folding(op->opc, ctx->type, t, 0); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } return false; @@ -XXX,XX +XXX,XX @@ static bool fold_const2(OptContext *ctx, TCGOp *op) uint64_t t1 = arg_info(op->args[1])->val; uint64_t t2 = arg_info(op->args[2])->val; - t1 = do_constant_folding(op->opc, t1, t2); + t1 = do_constant_folding(op->opc, ctx->type, t1, t2); return tcg_opt_gen_movi(ctx, op, op->args[0], t1); } return false; @@ -XXX,XX +XXX,XX @@ static bool fold_andc(OptContext *ctx, TCGOp *op) static bool fold_brcond(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[2]; - int i = do_constant_folding_cond(op->opc, op->args[0], op->args[1], cond); + int i = do_constant_folding_cond(ctx->type, op->args[0], op->args[1], cond); if (i == 0) { tcg_op_remove(ctx->tcg, op); @@ -XXX,XX +XXX,XX @@ static bool fold_brcond2(OptContext *ctx, TCGOp *op) * Simplify EQ/NE comparisons where one of the pairs * can be simplified. */ - i = do_constant_folding_cond(INDEX_op_brcond_i32, op->args[0], + i = do_constant_folding_cond(TCG_TYPE_I32, op->args[0], op->args[2], cond); switch (i ^ inv) { case 0: @@ -XXX,XX +XXX,XX @@ static bool fold_brcond2(OptContext *ctx, TCGOp *op) goto do_brcond_high; } - i = do_constant_folding_cond(INDEX_op_brcond_i32, op->args[1], + i = do_constant_folding_cond(TCG_TYPE_I32, op->args[1], op->args[3], cond); switch (i ^ inv) { case 0: @@ -XXX,XX +XXX,XX @@ static bool fold_bswap(OptContext *ctx, TCGOp *op) if (arg_is_const(op->args[1])) { uint64_t t = arg_info(op->args[1])->val; - t = do_constant_folding(op->opc, t, op->args[2]); + t = do_constant_folding(op->opc, ctx->type, t, op->args[2]); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } return false; @@ -XXX,XX +XXX,XX @@ static bool fold_count_zeros(OptContext *ctx, TCGOp *op) uint64_t t = arg_info(op->args[1])->val; if (t != 0) { - t = do_constant_folding(op->opc, t, 0); + t = do_constant_folding(op->opc, ctx->type, t, 0); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[2]); @@ -XXX,XX +XXX,XX @@ static bool fold_mov(OptContext *ctx, TCGOp *op) static bool fold_movcond(OptContext *ctx, TCGOp *op) { - TCGOpcode opc = op->opc; TCGCond cond = op->args[5]; - int i = do_constant_folding_cond(opc, op->args[1], op->args[2], cond); + int i = do_constant_folding_cond(ctx->type, op->args[1], op->args[2], cond); if (i >= 0) { return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[4 - i]); @@ -XXX,XX +XXX,XX @@ static bool fold_movcond(OptContext *ctx, TCGOp *op) if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { uint64_t tv = arg_info(op->args[3])->val; uint64_t fv = arg_info(op->args[4])->val; + TCGOpcode opc; - opc = (opc == INDEX_op_movcond_i32 - ? INDEX_op_setcond_i32 : INDEX_op_setcond_i64); + switch (ctx->type) { + case TCG_TYPE_I32: + opc = INDEX_op_setcond_i32; + break; + case TCG_TYPE_I64: + opc = INDEX_op_setcond_i64; + break; + default: + g_assert_not_reached(); + } if (tv == 1 && fv == 0) { op->opc = opc; @@ -XXX,XX +XXX,XX @@ static bool fold_remainder(OptContext *ctx, TCGOp *op) static bool fold_setcond(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[3]; - int i = do_constant_folding_cond(op->opc, op->args[1], op->args[2], cond); + int i = do_constant_folding_cond(ctx->type, op->args[1], op->args[2], cond); if (i >= 0) { return tcg_opt_gen_movi(ctx, op, op->args[0], i); @@ -XXX,XX +XXX,XX @@ static bool fold_setcond2(OptContext *ctx, TCGOp *op) * Simplify EQ/NE comparisons where one of the pairs * can be simplified. */ - i = do_constant_folding_cond(INDEX_op_setcond_i32, op->args[1], + i = do_constant_folding_cond(TCG_TYPE_I32, op->args[1], op->args[3], cond); switch (i ^ inv) { case 0: @@ -XXX,XX +XXX,XX @@ static bool fold_setcond2(OptContext *ctx, TCGOp *op) goto do_setcond_high; } - i = do_constant_folding_cond(INDEX_op_setcond_i32, op->args[2], + i = do_constant_folding_cond(TCG_TYPE_I32, op->args[2], op->args[4], cond); switch (i ^ inv) { case 0: @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) init_arguments(&ctx, op, def->nb_oargs + def->nb_iargs); copy_propagate(&ctx, op, def->nb_oargs, def->nb_iargs); + /* Pre-compute the type of the operation. */ + if (def->flags & TCG_OPF_VECTOR) { + ctx.type = TCG_TYPE_V64 + TCGOP_VECL(op); + } else if (def->flags & TCG_OPF_64BIT) { + ctx.type = TCG_TYPE_I64; + } else { + ctx.type = TCG_TYPE_I32; + } + /* For commutative operations make constant second argument */ switch (opc) { CASE_OP_32_64_VEC(add): @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) /* Proceed with possible constant folding. */ break; } - if (opc == INDEX_op_sub_i32) { + switch (ctx.type) { + case TCG_TYPE_I32: neg_op = INDEX_op_neg_i32; have_neg = TCG_TARGET_HAS_neg_i32; - } else if (opc == INDEX_op_sub_i64) { + break; + case TCG_TYPE_I64: neg_op = INDEX_op_neg_i64; have_neg = TCG_TARGET_HAS_neg_i64; - } else if (TCG_TARGET_HAS_neg_vec) { - TCGType type = TCGOP_VECL(op) + TCG_TYPE_V64; - unsigned vece = TCGOP_VECE(op); - neg_op = INDEX_op_neg_vec; - have_neg = tcg_can_emit_vec_op(neg_op, type, vece) > 0; - } else { break; + case TCG_TYPE_V64: + case TCG_TYPE_V128: + case TCG_TYPE_V256: + neg_op = INDEX_op_neg_vec; + have_neg = tcg_can_emit_vec_op(neg_op, ctx.type, + TCGOP_VECE(op)) > 0; + break; + default: + g_assert_not_reached(); } if (!have_neg) { break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) TCGOpcode not_op; bool have_not; - if (def->flags & TCG_OPF_VECTOR) { - not_op = INDEX_op_not_vec; - have_not = TCG_TARGET_HAS_not_vec; - } else if (def->flags & TCG_OPF_64BIT) { - not_op = INDEX_op_not_i64; - have_not = TCG_TARGET_HAS_not_i64; - } else { + switch (ctx.type) { + case TCG_TYPE_I32: not_op = INDEX_op_not_i32; have_not = TCG_TARGET_HAS_not_i32; + break; + case TCG_TYPE_I64: + not_op = INDEX_op_not_i64; + have_not = TCG_TARGET_HAS_not_i64; + break; + case TCG_TYPE_V64: + case TCG_TYPE_V128: + case TCG_TYPE_V256: + not_op = INDEX_op_not_vec; + have_not = TCG_TARGET_HAS_not_vec; + break; + default: + g_assert_not_reached(); } if (!have_not) { break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) below, we can ignore high bits, but for further optimizations we need to record that the high bits contain garbage. */ partmask = z_mask; - if (!(def->flags & TCG_OPF_64BIT)) { + if (ctx.type == TCG_TYPE_I32) { z_mask |= ~(tcg_target_ulong)0xffffffffu; partmask &= 0xffffffffu; affected &= 0xffffffffu; -- 2.25.1 Split out the conditional conversion from a more complex logical operation to a simple NOT. Create a couple more helpers to make this easy for the outer-most logical operations. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 158 +++++++++++++++++++++++++++---------------------- 1 file changed, 86 insertions(+), 72 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_const2(OptContext *ctx, TCGOp *op) return false; } +/* + * Convert @op to NOT, if NOT is supported by the host. + * Return true f the conversion is successful, which will still + * indicate that the processing is complete. + */ +static bool fold_not(OptContext *ctx, TCGOp *op); +static bool fold_to_not(OptContext *ctx, TCGOp *op, int idx) +{ + TCGOpcode not_op; + bool have_not; + + switch (ctx->type) { + case TCG_TYPE_I32: + not_op = INDEX_op_not_i32; + have_not = TCG_TARGET_HAS_not_i32; + break; + case TCG_TYPE_I64: + not_op = INDEX_op_not_i64; + have_not = TCG_TARGET_HAS_not_i64; + break; + case TCG_TYPE_V64: + case TCG_TYPE_V128: + case TCG_TYPE_V256: + not_op = INDEX_op_not_vec; + have_not = TCG_TARGET_HAS_not_vec; + break; + default: + g_assert_not_reached(); + } + if (have_not) { + op->opc = not_op; + op->args[1] = op->args[idx]; + return fold_not(ctx, op); + } + return false; +} + +/* If the binary operation has first argument @i, fold to NOT. */ +static bool fold_ix_to_not(OptContext *ctx, TCGOp *op, uint64_t i) +{ + if (arg_is_const(op->args[1]) && arg_info(op->args[1])->val == i) { + return fold_to_not(ctx, op, 2); + } + return false; +} + /* If the binary operation has second argument @i, fold to @i. */ static bool fold_xi_to_i(OptContext *ctx, TCGOp *op, uint64_t i) { @@ -XXX,XX +XXX,XX @@ static bool fold_xi_to_i(OptContext *ctx, TCGOp *op, uint64_t i) return false; } +/* If the binary operation has second argument @i, fold to NOT. */ +static bool fold_xi_to_not(OptContext *ctx, TCGOp *op, uint64_t i) +{ + if (arg_is_const(op->args[2]) && arg_info(op->args[2])->val == i) { + return fold_to_not(ctx, op, 1); + } + return false; +} + /* If the binary operation has both arguments equal, fold to @i. */ static bool fold_xx_to_i(OptContext *ctx, TCGOp *op, uint64_t i) { @@ -XXX,XX +XXX,XX @@ static bool fold_and(OptContext *ctx, TCGOp *op) static bool fold_andc(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || - fold_xx_to_i(ctx, op, 0)) { + fold_xx_to_i(ctx, op, 0) || + fold_ix_to_not(ctx, op, -1)) { return true; } return false; @@ -XXX,XX +XXX,XX @@ static bool fold_dup2(OptContext *ctx, TCGOp *op) static bool fold_eqv(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_not(ctx, op, 0)) { + return true; + } + return false; } static bool fold_extract(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_mulu2_i32(OptContext *ctx, TCGOp *op) static bool fold_nand(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_not(ctx, op, -1)) { + return true; + } + return false; } static bool fold_neg(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_neg(OptContext *ctx, TCGOp *op) static bool fold_nor(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_not(ctx, op, 0)) { + return true; + } + return false; } static bool fold_not(OptContext *ctx, TCGOp *op) { - return fold_const1(ctx, op); + if (fold_const1(ctx, op)) { + return true; + } + + /* Because of fold_to_not, we want to always return true, via finish. */ + finish_folding(ctx, op); + return true; } static bool fold_or(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_or(OptContext *ctx, TCGOp *op) static bool fold_orc(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_ix_to_not(ctx, op, 0)) { + return true; + } + return false; } static bool fold_qemu_ld(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_sub2_i32(OptContext *ctx, TCGOp *op) static bool fold_xor(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || - fold_xx_to_i(ctx, op, 0)) { + fold_xx_to_i(ctx, op, 0) || + fold_xi_to_not(ctx, op, -1)) { return true; } return false; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } } break; - CASE_OP_32_64_VEC(xor): - CASE_OP_32_64(nand): - if (!arg_is_const(op->args[1]) - && arg_is_const(op->args[2]) - && arg_info(op->args[2])->val == -1) { - i = 1; - goto try_not; - } - break; - CASE_OP_32_64(nor): - if (!arg_is_const(op->args[1]) - && arg_is_const(op->args[2]) - && arg_info(op->args[2])->val == 0) { - i = 1; - goto try_not; - } - break; - CASE_OP_32_64_VEC(andc): - if (!arg_is_const(op->args[2]) - && arg_is_const(op->args[1]) - && arg_info(op->args[1])->val == -1) { - i = 2; - goto try_not; - } - break; - CASE_OP_32_64_VEC(orc): - CASE_OP_32_64(eqv): - if (!arg_is_const(op->args[2]) - && arg_is_const(op->args[1]) - && arg_info(op->args[1])->val == 0) { - i = 2; - goto try_not; - } - break; - try_not: - { - TCGOpcode not_op; - bool have_not; - - switch (ctx.type) { - case TCG_TYPE_I32: - not_op = INDEX_op_not_i32; - have_not = TCG_TARGET_HAS_not_i32; - break; - case TCG_TYPE_I64: - not_op = INDEX_op_not_i64; - have_not = TCG_TARGET_HAS_not_i64; - break; - case TCG_TYPE_V64: - case TCG_TYPE_V128: - case TCG_TYPE_V256: - not_op = INDEX_op_not_vec; - have_not = TCG_TARGET_HAS_not_vec; - break; - default: - g_assert_not_reached(); - } - if (!have_not) { - break; - } - op->opc = not_op; - reset_temp(op->args[0]); - op->args[1] = op->args[i]; - continue; - } default: break; } -- 2.25.1 Even though there is only one user, place this more complex conversion into its own helper. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 89 ++++++++++++++++++++++++++------------------------ 1 file changed, 47 insertions(+), 42 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_nand(OptContext *ctx, TCGOp *op) static bool fold_neg(OptContext *ctx, TCGOp *op) { - return fold_const1(ctx, op); + if (fold_const1(ctx, op)) { + return true; + } + /* + * Because of fold_sub_to_neg, we want to always return true, + * via finish_folding. + */ + finish_folding(ctx, op); + return true; } static bool fold_nor(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_shift(OptContext *ctx, TCGOp *op) return fold_const2(ctx, op); } +static bool fold_sub_to_neg(OptContext *ctx, TCGOp *op) +{ + TCGOpcode neg_op; + bool have_neg; + + if (!arg_is_const(op->args[1]) || arg_info(op->args[1])->val != 0) { + return false; + } + + switch (ctx->type) { + case TCG_TYPE_I32: + neg_op = INDEX_op_neg_i32; + have_neg = TCG_TARGET_HAS_neg_i32; + break; + case TCG_TYPE_I64: + neg_op = INDEX_op_neg_i64; + have_neg = TCG_TARGET_HAS_neg_i64; + break; + case TCG_TYPE_V64: + case TCG_TYPE_V128: + case TCG_TYPE_V256: + neg_op = INDEX_op_neg_vec; + have_neg = (TCG_TARGET_HAS_neg_vec && + tcg_can_emit_vec_op(neg_op, ctx->type, TCGOP_VECE(op)) > 0); + break; + default: + g_assert_not_reached(); + } + if (have_neg) { + op->opc = neg_op; + op->args[1] = op->args[2]; + return fold_neg(ctx, op); + } + return false; +} + static bool fold_sub(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || - fold_xx_to_i(ctx, op, 0)) { + fold_xx_to_i(ctx, op, 0) || + fold_sub_to_neg(ctx, op)) { return true; } return false; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) continue; } break; - CASE_OP_32_64_VEC(sub): - { - TCGOpcode neg_op; - bool have_neg; - - if (arg_is_const(op->args[2])) { - /* Proceed with possible constant folding. */ - break; - } - switch (ctx.type) { - case TCG_TYPE_I32: - neg_op = INDEX_op_neg_i32; - have_neg = TCG_TARGET_HAS_neg_i32; - break; - case TCG_TYPE_I64: - neg_op = INDEX_op_neg_i64; - have_neg = TCG_TARGET_HAS_neg_i64; - break; - case TCG_TYPE_V64: - case TCG_TYPE_V128: - case TCG_TYPE_V256: - neg_op = INDEX_op_neg_vec; - have_neg = tcg_can_emit_vec_op(neg_op, ctx.type, - TCGOP_VECE(op)) > 0; - break; - default: - g_assert_not_reached(); - } - if (!have_neg) { - break; - } - if (arg_is_const(op->args[1]) - && arg_info(op->args[1])->val == 0) { - op->opc = neg_op; - reset_temp(op->args[0]); - op->args[1] = op->args[2]; - continue; - } - } - break; default: break; } -- 2.25.1 Pull the "op r, a, i => mov r, a" optimization into a function, and use them in the outer-most logical operations. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 61 +++++++++++++++++++++----------------------------- 1 file changed, 26 insertions(+), 35 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_xi_to_i(OptContext *ctx, TCGOp *op, uint64_t i) return false; } +/* If the binary operation has second argument @i, fold to identity. */ +static bool fold_xi_to_x(OptContext *ctx, TCGOp *op, uint64_t i) +{ + if (arg_is_const(op->args[2]) && arg_info(op->args[2])->val == i) { + return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[1]); + } + return false; +} + /* If the binary operation has second argument @i, fold to NOT. */ static bool fold_xi_to_not(OptContext *ctx, TCGOp *op, uint64_t i) { @@ -XXX,XX +XXX,XX @@ static bool fold_xx_to_x(OptContext *ctx, TCGOp *op) static bool fold_add(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_x(ctx, op, 0)) { + return true; + } + return false; } static bool fold_addsub2_i32(OptContext *ctx, TCGOp *op, bool add) @@ -XXX,XX +XXX,XX @@ static bool fold_and(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || fold_xi_to_i(ctx, op, 0) || + fold_xi_to_x(ctx, op, -1) || fold_xx_to_x(ctx, op)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_andc(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || fold_xx_to_i(ctx, op, 0) || + fold_xi_to_x(ctx, op, 0) || fold_ix_to_not(ctx, op, -1)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_dup2(OptContext *ctx, TCGOp *op) static bool fold_eqv(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || + fold_xi_to_x(ctx, op, -1) || fold_xi_to_not(ctx, op, 0)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_not(OptContext *ctx, TCGOp *op) static bool fold_or(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || + fold_xi_to_x(ctx, op, 0) || fold_xx_to_x(ctx, op)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_or(OptContext *ctx, TCGOp *op) static bool fold_orc(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || + fold_xi_to_x(ctx, op, -1) || fold_ix_to_not(ctx, op, 0)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_sextract(OptContext *ctx, TCGOp *op) static bool fold_shift(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_x(ctx, op, 0)) { + return true; + } + return false; } static bool fold_sub_to_neg(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_sub(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || fold_xx_to_i(ctx, op, 0) || + fold_xi_to_x(ctx, op, 0) || fold_sub_to_neg(ctx, op)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_xor(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || fold_xx_to_i(ctx, op, 0) || + fold_xi_to_x(ctx, op, 0) || fold_xi_to_not(ctx, op, -1)) { return true; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - /* Simplify expression for "op r, a, const => mov r, a" cases */ - switch (opc) { - CASE_OP_32_64_VEC(add): - CASE_OP_32_64_VEC(sub): - CASE_OP_32_64_VEC(or): - CASE_OP_32_64_VEC(xor): - CASE_OP_32_64_VEC(andc): - CASE_OP_32_64(shl): - CASE_OP_32_64(shr): - CASE_OP_32_64(sar): - CASE_OP_32_64(rotl): - CASE_OP_32_64(rotr): - if (!arg_is_const(op->args[1]) - && arg_is_const(op->args[2]) - && arg_info(op->args[2])->val == 0) { - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); - continue; - } - break; - CASE_OP_32_64_VEC(and): - CASE_OP_32_64_VEC(orc): - CASE_OP_32_64(eqv): - if (!arg_is_const(op->args[1]) - && arg_is_const(op->args[2]) - && arg_info(op->args[2])->val == -1) { - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); - continue; - } - break; - default: - break; - } - /* Simplify using known-zero bits. Currently only ops with a single output argument is supported. */ z_mask = -1; -- 2.25.1 Pull the "op r, 0, b => movi r, 0" optimization into a function, and use it in fold_shift. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 28 ++++++++++------------------ 1 file changed, 10 insertions(+), 18 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_to_not(OptContext *ctx, TCGOp *op, int idx) return false; } +/* If the binary operation has first argument @i, fold to @i. */ +static bool fold_ix_to_i(OptContext *ctx, TCGOp *op, uint64_t i) +{ + if (arg_is_const(op->args[1]) && arg_info(op->args[1])->val == i) { + return tcg_opt_gen_movi(ctx, op, op->args[0], i); + } + return false; +} + /* If the binary operation has first argument @i, fold to NOT. */ static bool fold_ix_to_not(OptContext *ctx, TCGOp *op, uint64_t i) { @@ -XXX,XX +XXX,XX @@ static bool fold_sextract(OptContext *ctx, TCGOp *op) static bool fold_shift(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || + fold_ix_to_i(ctx, op, 0) || fold_xi_to_x(ctx, op, 0)) { return true; } @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - /* Simplify expressions for "shift/rot r, 0, a => movi r, 0", - and "sub r, 0, a => neg r, a" case. */ - switch (opc) { - CASE_OP_32_64(shl): - CASE_OP_32_64(shr): - CASE_OP_32_64(sar): - CASE_OP_32_64(rotl): - CASE_OP_32_64(rotr): - if (arg_is_const(op->args[1]) - && arg_info(op->args[1])->val == 0) { - tcg_opt_gen_movi(&ctx, op, op->args[0], 0); - continue; - } - break; - default: - break; - } - /* Simplify using known-zero bits. Currently only ops with a single output argument is supported. */ z_mask = -1; -- 2.25.1 Move all of the known-zero optimizations into the per-opcode functions. Use fold_masks when there is a possibility of the result being determined, and simply set ctx->z_mask otherwise. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 545 ++++++++++++++++++++++++++----------------------- 1 file changed, 294 insertions(+), 251 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct OptContext { TCGTempSet temps_used; /* In flight values from optimization. */ - uint64_t z_mask; + uint64_t a_mask; /* mask bit is 0 iff value identical to first input */ + uint64_t z_mask; /* mask bit is 0 iff value bit is 0 */ TCGType type; } OptContext; @@ -XXX,XX +XXX,XX @@ static bool fold_const2(OptContext *ctx, TCGOp *op) return false; } +static bool fold_masks(OptContext *ctx, TCGOp *op) +{ + uint64_t a_mask = ctx->a_mask; + uint64_t z_mask = ctx->z_mask; + + /* + * 32-bit ops generate 32-bit results. For the result is zero test + * below, we can ignore high bits, but for further optimizations we + * need to record that the high bits contain garbage. + */ + if (ctx->type == TCG_TYPE_I32) { + ctx->z_mask |= MAKE_64BIT_MASK(32, 32); + a_mask &= MAKE_64BIT_MASK(0, 32); + z_mask &= MAKE_64BIT_MASK(0, 32); + } + + if (z_mask == 0) { + return tcg_opt_gen_movi(ctx, op, op->args[0], 0); + } + if (a_mask == 0) { + return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[1]); + } + return false; +} + /* * Convert @op to NOT, if NOT is supported by the host. * Return true f the conversion is successful, which will still @@ -XXX,XX +XXX,XX @@ static bool fold_add2_i32(OptContext *ctx, TCGOp *op) static bool fold_and(OptContext *ctx, TCGOp *op) { + uint64_t z1, z2; + if (fold_const2(ctx, op) || fold_xi_to_i(ctx, op, 0) || fold_xi_to_x(ctx, op, -1) || fold_xx_to_x(ctx, op)) { return true; } - return false; + + z1 = arg_info(op->args[1])->z_mask; + z2 = arg_info(op->args[2])->z_mask; + ctx->z_mask = z1 & z2; + + /* + * Known-zeros does not imply known-ones. Therefore unless + * arg2 is constant, we can't infer affected bits from it. + */ + if (arg_is_const(op->args[2])) { + ctx->a_mask = z1 & ~z2; + } + + return fold_masks(ctx, op); } static bool fold_andc(OptContext *ctx, TCGOp *op) { + uint64_t z1; + if (fold_const2(ctx, op) || fold_xx_to_i(ctx, op, 0) || fold_xi_to_x(ctx, op, 0) || fold_ix_to_not(ctx, op, -1)) { return true; } - return false; + + z1 = arg_info(op->args[1])->z_mask; + + /* + * Known-zeros does not imply known-ones. Therefore unless + * arg2 is constant, we can't infer anything from it. + */ + if (arg_is_const(op->args[2])) { + uint64_t z2 = ~arg_info(op->args[2])->z_mask; + ctx->a_mask = z1 & ~z2; + z1 &= z2; + } + ctx->z_mask = z1; + + return fold_masks(ctx, op); } static bool fold_brcond(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_brcond2(OptContext *ctx, TCGOp *op) static bool fold_bswap(OptContext *ctx, TCGOp *op) { + uint64_t z_mask, sign; + if (arg_is_const(op->args[1])) { uint64_t t = arg_info(op->args[1])->val; t = do_constant_folding(op->opc, ctx->type, t, op->args[2]); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } - return false; + + z_mask = arg_info(op->args[1])->z_mask; + switch (op->opc) { + case INDEX_op_bswap16_i32: + case INDEX_op_bswap16_i64: + z_mask = bswap16(z_mask); + sign = INT16_MIN; + break; + case INDEX_op_bswap32_i32: + case INDEX_op_bswap32_i64: + z_mask = bswap32(z_mask); + sign = INT32_MIN; + break; + case INDEX_op_bswap64_i64: + z_mask = bswap64(z_mask); + sign = INT64_MIN; + break; + default: + g_assert_not_reached(); + } + + switch (op->args[2] & (TCG_BSWAP_OZ | TCG_BSWAP_OS)) { + case TCG_BSWAP_OZ: + break; + case TCG_BSWAP_OS: + /* If the sign bit may be 1, force all the bits above to 1. */ + if (z_mask & sign) { + z_mask |= sign; + } + break; + default: + /* The high bits are undefined: force all bits above the sign to 1. */ + z_mask |= sign << 1; + break; + } + ctx->z_mask = z_mask; + + return fold_masks(ctx, op); } static bool fold_call(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_call(OptContext *ctx, TCGOp *op) static bool fold_count_zeros(OptContext *ctx, TCGOp *op) { + uint64_t z_mask; + if (arg_is_const(op->args[1])) { uint64_t t = arg_info(op->args[1])->val; @@ -XXX,XX +XXX,XX @@ static bool fold_count_zeros(OptContext *ctx, TCGOp *op) } return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[2]); } + + switch (ctx->type) { + case TCG_TYPE_I32: + z_mask = 31; + break; + case TCG_TYPE_I64: + z_mask = 63; + break; + default: + g_assert_not_reached(); + } + ctx->z_mask = arg_info(op->args[2])->z_mask | z_mask; + return false; } static bool fold_ctpop(OptContext *ctx, TCGOp *op) { - return fold_const1(ctx, op); + if (fold_const1(ctx, op)) { + return true; + } + + switch (ctx->type) { + case TCG_TYPE_I32: + ctx->z_mask = 32 | 31; + break; + case TCG_TYPE_I64: + ctx->z_mask = 64 | 63; + break; + default: + g_assert_not_reached(); + } + return false; } static bool fold_deposit(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_deposit(OptContext *ctx, TCGOp *op) t1 = deposit64(t1, op->args[3], op->args[4], t2); return tcg_opt_gen_movi(ctx, op, op->args[0], t1); } + + ctx->z_mask = deposit64(arg_info(op->args[1])->z_mask, + op->args[3], op->args[4], + arg_info(op->args[2])->z_mask); return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_eqv(OptContext *ctx, TCGOp *op) static bool fold_extract(OptContext *ctx, TCGOp *op) { + uint64_t z_mask_old, z_mask; + if (arg_is_const(op->args[1])) { uint64_t t; @@ -XXX,XX +XXX,XX @@ static bool fold_extract(OptContext *ctx, TCGOp *op) t = extract64(t, op->args[2], op->args[3]); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } - return false; + + z_mask_old = arg_info(op->args[1])->z_mask; + z_mask = extract64(z_mask_old, op->args[2], op->args[3]); + if (op->args[2] == 0) { + ctx->a_mask = z_mask_old ^ z_mask; + } + ctx->z_mask = z_mask; + + return fold_masks(ctx, op); } static bool fold_extract2(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_extract2(OptContext *ctx, TCGOp *op) static bool fold_exts(OptContext *ctx, TCGOp *op) { - return fold_const1(ctx, op); + uint64_t z_mask_old, z_mask, sign; + bool type_change = false; + + if (fold_const1(ctx, op)) { + return true; + } + + z_mask_old = z_mask = arg_info(op->args[1])->z_mask; + + switch (op->opc) { + CASE_OP_32_64(ext8s): + sign = INT8_MIN; + z_mask = (uint8_t)z_mask; + break; + CASE_OP_32_64(ext16s): + sign = INT16_MIN; + z_mask = (uint16_t)z_mask; + break; + case INDEX_op_ext_i32_i64: + type_change = true; + QEMU_FALLTHROUGH; + case INDEX_op_ext32s_i64: + sign = INT32_MIN; + z_mask = (uint32_t)z_mask; + break; + default: + g_assert_not_reached(); + } + + if (z_mask & sign) { + z_mask |= sign; + } else if (!type_change) { + ctx->a_mask = z_mask_old ^ z_mask; + } + ctx->z_mask = z_mask; + + return fold_masks(ctx, op); } static bool fold_extu(OptContext *ctx, TCGOp *op) { - return fold_const1(ctx, op); + uint64_t z_mask_old, z_mask; + bool type_change = false; + + if (fold_const1(ctx, op)) { + return true; + } + + z_mask_old = z_mask = arg_info(op->args[1])->z_mask; + + switch (op->opc) { + CASE_OP_32_64(ext8u): + z_mask = (uint8_t)z_mask; + break; + CASE_OP_32_64(ext16u): + z_mask = (uint16_t)z_mask; + break; + case INDEX_op_extrl_i64_i32: + case INDEX_op_extu_i32_i64: + type_change = true; + QEMU_FALLTHROUGH; + case INDEX_op_ext32u_i64: + z_mask = (uint32_t)z_mask; + break; + case INDEX_op_extrh_i64_i32: + type_change = true; + z_mask >>= 32; + break; + default: + g_assert_not_reached(); + } + + ctx->z_mask = z_mask; + if (!type_change) { + ctx->a_mask = z_mask_old ^ z_mask; + } + return fold_masks(ctx, op); } static bool fold_mb(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_movcond(OptContext *ctx, TCGOp *op) return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[4 - i]); } + ctx->z_mask = arg_info(op->args[3])->z_mask + | arg_info(op->args[4])->z_mask; + if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { uint64_t tv = arg_info(op->args[3])->val; uint64_t fv = arg_info(op->args[4])->val; @@ -XXX,XX +XXX,XX @@ static bool fold_nand(OptContext *ctx, TCGOp *op) static bool fold_neg(OptContext *ctx, TCGOp *op) { + uint64_t z_mask; + if (fold_const1(ctx, op)) { return true; } + + /* Set to 1 all bits to the left of the rightmost. */ + z_mask = arg_info(op->args[1])->z_mask; + ctx->z_mask = -(z_mask & -z_mask); + /* * Because of fold_sub_to_neg, we want to always return true, * via finish_folding. @@ -XXX,XX +XXX,XX @@ static bool fold_or(OptContext *ctx, TCGOp *op) fold_xx_to_x(ctx, op)) { return true; } - return false; + + ctx->z_mask = arg_info(op->args[1])->z_mask + | arg_info(op->args[2])->z_mask; + return fold_masks(ctx, op); } static bool fold_orc(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_orc(OptContext *ctx, TCGOp *op) static bool fold_qemu_ld(OptContext *ctx, TCGOp *op) { + const TCGOpDef *def = &tcg_op_defs[op->opc]; + MemOpIdx oi = op->args[def->nb_oargs + def->nb_iargs]; + MemOp mop = get_memop(oi); + int width = 8 * memop_size(mop); + + if (!(mop & MO_SIGN) && width < 64) { + ctx->z_mask = MAKE_64BIT_MASK(0, width); + } + /* Opcodes that touch guest memory stop the mb optimization. */ ctx->prev_mb = NULL; return false; @@ -XXX,XX +XXX,XX @@ static bool fold_setcond(OptContext *ctx, TCGOp *op) if (i >= 0) { return tcg_opt_gen_movi(ctx, op, op->args[0], i); } + + ctx->z_mask = 1; return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_setcond2(OptContext *ctx, TCGOp *op) op->opc = INDEX_op_setcond_i32; break; } + + ctx->z_mask = 1; return false; do_setcond_const: @@ -XXX,XX +XXX,XX @@ static bool fold_setcond2(OptContext *ctx, TCGOp *op) static bool fold_sextract(OptContext *ctx, TCGOp *op) { + int64_t z_mask_old, z_mask; + if (arg_is_const(op->args[1])) { uint64_t t; @@ -XXX,XX +XXX,XX @@ static bool fold_sextract(OptContext *ctx, TCGOp *op) t = sextract64(t, op->args[2], op->args[3]); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } - return false; + + z_mask_old = arg_info(op->args[1])->z_mask; + z_mask = sextract64(z_mask_old, op->args[2], op->args[3]); + if (op->args[2] == 0 && z_mask >= 0) { + ctx->a_mask = z_mask_old ^ z_mask; + } + ctx->z_mask = z_mask; + + return fold_masks(ctx, op); } static bool fold_shift(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_shift(OptContext *ctx, TCGOp *op) fold_xi_to_x(ctx, op, 0)) { return true; } + + if (arg_is_const(op->args[2])) { + ctx->z_mask = do_constant_folding(op->opc, ctx->type, + arg_info(op->args[1])->z_mask, + arg_info(op->args[2])->val); + return fold_masks(ctx, op); + } return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_sub2_i32(OptContext *ctx, TCGOp *op) return fold_addsub2_i32(ctx, op, false); } +static bool fold_tcg_ld(OptContext *ctx, TCGOp *op) +{ + /* We can't do any folding with a load, but we can record bits. */ + switch (op->opc) { + CASE_OP_32_64(ld8u): + ctx->z_mask = MAKE_64BIT_MASK(0, 8); + break; + CASE_OP_32_64(ld16u): + ctx->z_mask = MAKE_64BIT_MASK(0, 16); + break; + case INDEX_op_ld32u_i64: + ctx->z_mask = MAKE_64BIT_MASK(0, 32); + break; + default: + g_assert_not_reached(); + } + return false; +} + static bool fold_xor(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || @@ -XXX,XX +XXX,XX @@ static bool fold_xor(OptContext *ctx, TCGOp *op) fold_xi_to_not(ctx, op, -1)) { return true; } - return false; + + ctx->z_mask = arg_info(op->args[1])->z_mask + | arg_info(op->args[2])->z_mask; + return fold_masks(ctx, op); } /* Propagate constants and copies, fold constant expressions. */ @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) } QTAILQ_FOREACH_SAFE(op, &s->ops, link, op_next) { - uint64_t z_mask, partmask, affected, tmp; TCGOpcode opc = op->opc; const TCGOpDef *def; bool done = false; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } - /* Simplify using known-zero bits. Currently only ops with a single - output argument is supported. */ - z_mask = -1; - affected = -1; - switch (opc) { - CASE_OP_32_64(ext8s): - if ((arg_info(op->args[1])->z_mask & 0x80) != 0) { - break; - } - QEMU_FALLTHROUGH; - CASE_OP_32_64(ext8u): - z_mask = 0xff; - goto and_const; - CASE_OP_32_64(ext16s): - if ((arg_info(op->args[1])->z_mask & 0x8000) != 0) { - break; - } - QEMU_FALLTHROUGH; - CASE_OP_32_64(ext16u): - z_mask = 0xffff; - goto and_const; - case INDEX_op_ext32s_i64: - if ((arg_info(op->args[1])->z_mask & 0x80000000) != 0) { - break; - } - QEMU_FALLTHROUGH; - case INDEX_op_ext32u_i64: - z_mask = 0xffffffffU; - goto and_const; - - CASE_OP_32_64(and): - z_mask = arg_info(op->args[2])->z_mask; - if (arg_is_const(op->args[2])) { - and_const: - affected = arg_info(op->args[1])->z_mask & ~z_mask; - } - z_mask = arg_info(op->args[1])->z_mask & z_mask; - break; - - case INDEX_op_ext_i32_i64: - if ((arg_info(op->args[1])->z_mask & 0x80000000) != 0) { - break; - } - QEMU_FALLTHROUGH; - case INDEX_op_extu_i32_i64: - /* We do not compute affected as it is a size changing op. */ - z_mask = (uint32_t)arg_info(op->args[1])->z_mask; - break; - - CASE_OP_32_64(andc): - /* Known-zeros does not imply known-ones. Therefore unless - op->args[2] is constant, we can't infer anything from it. */ - if (arg_is_const(op->args[2])) { - z_mask = ~arg_info(op->args[2])->z_mask; - goto and_const; - } - /* But we certainly know nothing outside args[1] may be set. */ - z_mask = arg_info(op->args[1])->z_mask; - break; - - case INDEX_op_sar_i32: - if (arg_is_const(op->args[2])) { - tmp = arg_info(op->args[2])->val & 31; - z_mask = (int32_t)arg_info(op->args[1])->z_mask >> tmp; - } - break; - case INDEX_op_sar_i64: - if (arg_is_const(op->args[2])) { - tmp = arg_info(op->args[2])->val & 63; - z_mask = (int64_t)arg_info(op->args[1])->z_mask >> tmp; - } - break; - - case INDEX_op_shr_i32: - if (arg_is_const(op->args[2])) { - tmp = arg_info(op->args[2])->val & 31; - z_mask = (uint32_t)arg_info(op->args[1])->z_mask >> tmp; - } - break; - case INDEX_op_shr_i64: - if (arg_is_const(op->args[2])) { - tmp = arg_info(op->args[2])->val & 63; - z_mask = (uint64_t)arg_info(op->args[1])->z_mask >> tmp; - } - break; - - case INDEX_op_extrl_i64_i32: - z_mask = (uint32_t)arg_info(op->args[1])->z_mask; - break; - case INDEX_op_extrh_i64_i32: - z_mask = (uint64_t)arg_info(op->args[1])->z_mask >> 32; - break; - - CASE_OP_32_64(shl): - if (arg_is_const(op->args[2])) { - tmp = arg_info(op->args[2])->val & (TCG_TARGET_REG_BITS - 1); - z_mask = arg_info(op->args[1])->z_mask << tmp; - } - break; - - CASE_OP_32_64(neg): - /* Set to 1 all bits to the left of the rightmost. */ - z_mask = -(arg_info(op->args[1])->z_mask - & -arg_info(op->args[1])->z_mask); - break; - - CASE_OP_32_64(deposit): - z_mask = deposit64(arg_info(op->args[1])->z_mask, - op->args[3], op->args[4], - arg_info(op->args[2])->z_mask); - break; - - CASE_OP_32_64(extract): - z_mask = extract64(arg_info(op->args[1])->z_mask, - op->args[2], op->args[3]); - if (op->args[2] == 0) { - affected = arg_info(op->args[1])->z_mask & ~z_mask; - } - break; - CASE_OP_32_64(sextract): - z_mask = sextract64(arg_info(op->args[1])->z_mask, - op->args[2], op->args[3]); - if (op->args[2] == 0 && (tcg_target_long)z_mask >= 0) { - affected = arg_info(op->args[1])->z_mask & ~z_mask; - } - break; - - CASE_OP_32_64(or): - CASE_OP_32_64(xor): - z_mask = arg_info(op->args[1])->z_mask - | arg_info(op->args[2])->z_mask; - break; - - case INDEX_op_clz_i32: - case INDEX_op_ctz_i32: - z_mask = arg_info(op->args[2])->z_mask | 31; - break; - - case INDEX_op_clz_i64: - case INDEX_op_ctz_i64: - z_mask = arg_info(op->args[2])->z_mask | 63; - break; - - case INDEX_op_ctpop_i32: - z_mask = 32 | 31; - break; - case INDEX_op_ctpop_i64: - z_mask = 64 | 63; - break; - - CASE_OP_32_64(setcond): - case INDEX_op_setcond2_i32: - z_mask = 1; - break; - - CASE_OP_32_64(movcond): - z_mask = arg_info(op->args[3])->z_mask - | arg_info(op->args[4])->z_mask; - break; - - CASE_OP_32_64(ld8u): - z_mask = 0xff; - break; - CASE_OP_32_64(ld16u): - z_mask = 0xffff; - break; - case INDEX_op_ld32u_i64: - z_mask = 0xffffffffu; - break; - - CASE_OP_32_64(qemu_ld): - { - MemOpIdx oi = op->args[def->nb_oargs + def->nb_iargs]; - MemOp mop = get_memop(oi); - if (!(mop & MO_SIGN)) { - z_mask = (2ULL << ((8 << (mop & MO_SIZE)) - 1)) - 1; - } - } - break; - - CASE_OP_32_64(bswap16): - z_mask = arg_info(op->args[1])->z_mask; - if (z_mask <= 0xffff) { - op->args[2] |= TCG_BSWAP_IZ; - } - z_mask = bswap16(z_mask); - switch (op->args[2] & (TCG_BSWAP_OZ | TCG_BSWAP_OS)) { - case TCG_BSWAP_OZ: - break; - case TCG_BSWAP_OS: - z_mask = (int16_t)z_mask; - break; - default: /* undefined high bits */ - z_mask |= MAKE_64BIT_MASK(16, 48); - break; - } - break; - - case INDEX_op_bswap32_i64: - z_mask = arg_info(op->args[1])->z_mask; - if (z_mask <= 0xffffffffu) { - op->args[2] |= TCG_BSWAP_IZ; - } - z_mask = bswap32(z_mask); - switch (op->args[2] & (TCG_BSWAP_OZ | TCG_BSWAP_OS)) { - case TCG_BSWAP_OZ: - break; - case TCG_BSWAP_OS: - z_mask = (int32_t)z_mask; - break; - default: /* undefined high bits */ - z_mask |= MAKE_64BIT_MASK(32, 32); - break; - } - break; - - default: - break; - } - - /* 32-bit ops generate 32-bit results. For the result is zero test - below, we can ignore high bits, but for further optimizations we - need to record that the high bits contain garbage. */ - partmask = z_mask; - if (ctx.type == TCG_TYPE_I32) { - z_mask |= ~(tcg_target_ulong)0xffffffffu; - partmask &= 0xffffffffu; - affected &= 0xffffffffu; - } - ctx.z_mask = z_mask; - - if (partmask == 0) { - tcg_opt_gen_movi(&ctx, op, op->args[0], 0); - continue; - } - if (affected == 0) { - tcg_opt_gen_mov(&ctx, op, op->args[0], op->args[1]); - continue; - } + /* Assume all bits affected, and no bits known zero. */ + ctx.a_mask = -1; + ctx.z_mask = -1; /* * Process each opcode. @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_extrh_i64_i32: done = fold_extu(&ctx, op); break; + CASE_OP_32_64(ld8u): + CASE_OP_32_64(ld16u): + case INDEX_op_ld32u_i64: + done = fold_tcg_ld(&ctx, op); + break; case INDEX_op_mb: done = fold_mb(&ctx, op); break; -- 2.25.1 Rename to fold_multiply2, and handle muls2_i32, mulu2_i64, and muls2_i64. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 44 +++++++++++++++++++++++++++++++++++--------- 1 file changed, 35 insertions(+), 9 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_mul_highpart(OptContext *ctx, TCGOp *op) return false; } -static bool fold_mulu2_i32(OptContext *ctx, TCGOp *op) +static bool fold_multiply2(OptContext *ctx, TCGOp *op) { if (arg_is_const(op->args[2]) && arg_is_const(op->args[3])) { - uint32_t a = arg_info(op->args[2])->val; - uint32_t b = arg_info(op->args[3])->val; - uint64_t r = (uint64_t)a * b; + uint64_t a = arg_info(op->args[2])->val; + uint64_t b = arg_info(op->args[3])->val; + uint64_t h, l; TCGArg rl, rh; - TCGOp *op2 = tcg_op_insert_before(ctx->tcg, op, INDEX_op_mov_i32); + TCGOp *op2; + + switch (op->opc) { + case INDEX_op_mulu2_i32: + l = (uint64_t)(uint32_t)a * (uint32_t)b; + h = (int32_t)(l >> 32); + l = (int32_t)l; + break; + case INDEX_op_muls2_i32: + l = (int64_t)(int32_t)a * (int32_t)b; + h = l >> 32; + l = (int32_t)l; + break; + case INDEX_op_mulu2_i64: + mulu64(&l, &h, a, b); + break; + case INDEX_op_muls2_i64: + muls64(&l, &h, a, b); + break; + default: + g_assert_not_reached(); + } rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(ctx, op, rl, (int32_t)r); - tcg_opt_gen_movi(ctx, op2, rh, (int32_t)(r >> 32)); + + /* The proper opcode is supplied by tcg_opt_gen_mov. */ + op2 = tcg_op_insert_before(ctx->tcg, op, 0); + + tcg_opt_gen_movi(ctx, op, rl, l); + tcg_opt_gen_movi(ctx, op2, rh, h); return true; } return false; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(muluh): done = fold_mul_highpart(&ctx, op); break; - case INDEX_op_mulu2_i32: - done = fold_mulu2_i32(&ctx, op); + CASE_OP_32_64(muls2): + CASE_OP_32_64(mulu2): + done = fold_multiply2(&ctx, op); break; CASE_OP_32_64(nand): done = fold_nand(&ctx, op); -- 2.25.1 Rename to fold_addsub2. Use Int128 to implement the wider operation. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 65 ++++++++++++++++++++++++++++++++++---------------- 1 file changed, 44 insertions(+), 21 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ */ #include "qemu/osdep.h" +#include "qemu/int128.h" #include "tcg/tcg-op.h" #include "tcg-internal.h" @@ -XXX,XX +XXX,XX @@ static bool fold_add(OptContext *ctx, TCGOp *op) return false; } -static bool fold_addsub2_i32(OptContext *ctx, TCGOp *op, bool add) +static bool fold_addsub2(OptContext *ctx, TCGOp *op, bool add) { if (arg_is_const(op->args[2]) && arg_is_const(op->args[3]) && arg_is_const(op->args[4]) && arg_is_const(op->args[5])) { - uint32_t al = arg_info(op->args[2])->val; - uint32_t ah = arg_info(op->args[3])->val; - uint32_t bl = arg_info(op->args[4])->val; - uint32_t bh = arg_info(op->args[5])->val; - uint64_t a = ((uint64_t)ah << 32) | al; - uint64_t b = ((uint64_t)bh << 32) | bl; + uint64_t al = arg_info(op->args[2])->val; + uint64_t ah = arg_info(op->args[3])->val; + uint64_t bl = arg_info(op->args[4])->val; + uint64_t bh = arg_info(op->args[5])->val; TCGArg rl, rh; - TCGOp *op2 = tcg_op_insert_before(ctx->tcg, op, INDEX_op_mov_i32); + TCGOp *op2; - if (add) { - a += b; + if (ctx->type == TCG_TYPE_I32) { + uint64_t a = deposit64(al, 32, 32, ah); + uint64_t b = deposit64(bl, 32, 32, bh); + + if (add) { + a += b; + } else { + a -= b; + } + + al = sextract64(a, 0, 32); + ah = sextract64(a, 32, 32); } else { - a -= b; + Int128 a = int128_make128(al, ah); + Int128 b = int128_make128(bl, bh); + + if (add) { + a = int128_add(a, b); + } else { + a = int128_sub(a, b); + } + + al = int128_getlo(a); + ah = int128_gethi(a); } rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(ctx, op, rl, (int32_t)a); - tcg_opt_gen_movi(ctx, op2, rh, (int32_t)(a >> 32)); + + /* The proper opcode is supplied by tcg_opt_gen_mov. */ + op2 = tcg_op_insert_before(ctx->tcg, op, 0); + + tcg_opt_gen_movi(ctx, op, rl, al); + tcg_opt_gen_movi(ctx, op2, rh, ah); return true; } return false; } -static bool fold_add2_i32(OptContext *ctx, TCGOp *op) +static bool fold_add2(OptContext *ctx, TCGOp *op) { - return fold_addsub2_i32(ctx, op, true); + return fold_addsub2(ctx, op, true); } static bool fold_and(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static bool fold_sub(OptContext *ctx, TCGOp *op) return false; } -static bool fold_sub2_i32(OptContext *ctx, TCGOp *op) +static bool fold_sub2(OptContext *ctx, TCGOp *op) { - return fold_addsub2_i32(ctx, op, false); + return fold_addsub2(ctx, op, false); } static bool fold_tcg_ld(OptContext *ctx, TCGOp *op) @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(add): done = fold_add(&ctx, op); break; - case INDEX_op_add2_i32: - done = fold_add2_i32(&ctx, op); + CASE_OP_32_64(add2): + done = fold_add2(&ctx, op); break; CASE_OP_32_64_VEC(and): done = fold_and(&ctx, op); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(sub): done = fold_sub(&ctx, op); break; - case INDEX_op_sub2_i32: - done = fold_sub2_i32(&ctx, op); + CASE_OP_32_64(sub2): + done = fold_sub2(&ctx, op); break; CASE_OP_32_64_VEC(xor): done = fold_xor(&ctx, op); -- 2.25.1 Most of these are handled by creating a fold_const2_commutative to handle all of the binary operators. The rest were already handled on a case-by-case basis in the switch, and have their own fold function in which to place the call. We now have only one major switch on TCGOpcode. Introduce NO_DEST and a block comment for swap_commutative in order to make the handling of brcond and movcond opcodes cleaner. Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 142 ++++++++++++++++++++++++------------------------- 1 file changed, 70 insertions(+), 72 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static int do_constant_folding_cond2(TCGArg *p1, TCGArg *p2, TCGCond c) return -1; } +/** + * swap_commutative: + * @dest: TCGArg of the destination argument, or NO_DEST. + * @p1: first paired argument + * @p2: second paired argument + * + * If *@p1 is a constant and *@p2 is not, swap. + * If *@p2 matches @dest, swap. + * Return true if a swap was performed. + */ + +#define NO_DEST temp_arg(NULL) + static bool swap_commutative(TCGArg dest, TCGArg *p1, TCGArg *p2) { TCGArg a1 = *p1, a2 = *p2; @@ -XXX,XX +XXX,XX @@ static bool fold_const2(OptContext *ctx, TCGOp *op) return false; } +static bool fold_const2_commutative(OptContext *ctx, TCGOp *op) +{ + swap_commutative(op->args[0], &op->args[1], &op->args[2]); + return fold_const2(ctx, op); +} + static bool fold_masks(OptContext *ctx, TCGOp *op) { uint64_t a_mask = ctx->a_mask; @@ -XXX,XX +XXX,XX @@ static bool fold_xx_to_x(OptContext *ctx, TCGOp *op) static bool fold_add(OptContext *ctx, TCGOp *op) { - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xi_to_x(ctx, op, 0)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_addsub2(OptContext *ctx, TCGOp *op, bool add) static bool fold_add2(OptContext *ctx, TCGOp *op) { + /* Note that the high and low parts may be independently swapped. */ + swap_commutative(op->args[0], &op->args[2], &op->args[4]); + swap_commutative(op->args[1], &op->args[3], &op->args[5]); + return fold_addsub2(ctx, op, true); } @@ -XXX,XX +XXX,XX @@ static bool fold_and(OptContext *ctx, TCGOp *op) { uint64_t z1, z2; - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xi_to_i(ctx, op, 0) || fold_xi_to_x(ctx, op, -1) || fold_xx_to_x(ctx, op)) { @@ -XXX,XX +XXX,XX @@ static bool fold_andc(OptContext *ctx, TCGOp *op) static bool fold_brcond(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[2]; - int i = do_constant_folding_cond(ctx->type, op->args[0], op->args[1], cond); + int i; + if (swap_commutative(NO_DEST, &op->args[0], &op->args[1])) { + op->args[2] = cond = tcg_swap_cond(cond); + } + + i = do_constant_folding_cond(ctx->type, op->args[0], op->args[1], cond); if (i == 0) { tcg_op_remove(ctx->tcg, op); return true; @@ -XXX,XX +XXX,XX @@ static bool fold_brcond(OptContext *ctx, TCGOp *op) static bool fold_brcond2(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[4]; - int i = do_constant_folding_cond2(&op->args[0], &op->args[2], cond); TCGArg label = op->args[5]; - int inv = 0; + int i, inv = 0; + if (swap_commutative2(&op->args[0], &op->args[2])) { + op->args[4] = cond = tcg_swap_cond(cond); + } + + i = do_constant_folding_cond2(&op->args[0], &op->args[2], cond); if (i >= 0) { goto do_brcond_const; } @@ -XXX,XX +XXX,XX @@ static bool fold_dup2(OptContext *ctx, TCGOp *op) static bool fold_eqv(OptContext *ctx, TCGOp *op) { - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xi_to_x(ctx, op, -1) || fold_xi_to_not(ctx, op, 0)) { return true; @@ -XXX,XX +XXX,XX @@ static bool fold_mov(OptContext *ctx, TCGOp *op) static bool fold_movcond(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[5]; - int i = do_constant_folding_cond(ctx->type, op->args[1], op->args[2], cond); + int i; + if (swap_commutative(NO_DEST, &op->args[1], &op->args[2])) { + op->args[5] = cond = tcg_swap_cond(cond); + } + /* + * Canonicalize the "false" input reg to match the destination reg so + * that the tcg backend can implement a "move if true" operation. + */ + if (swap_commutative(op->args[0], &op->args[4], &op->args[3])) { + op->args[5] = cond = tcg_invert_cond(cond); + } + + i = do_constant_folding_cond(ctx->type, op->args[1], op->args[2], cond); if (i >= 0) { return tcg_opt_gen_mov(ctx, op, op->args[0], op->args[4 - i]); } @@ -XXX,XX +XXX,XX @@ static bool fold_mul(OptContext *ctx, TCGOp *op) static bool fold_mul_highpart(OptContext *ctx, TCGOp *op) { - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xi_to_i(ctx, op, 0)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_mul_highpart(OptContext *ctx, TCGOp *op) static bool fold_multiply2(OptContext *ctx, TCGOp *op) { + swap_commutative(op->args[0], &op->args[2], &op->args[3]); + if (arg_is_const(op->args[2]) && arg_is_const(op->args[3])) { uint64_t a = arg_info(op->args[2])->val; uint64_t b = arg_info(op->args[3])->val; @@ -XXX,XX +XXX,XX @@ static bool fold_multiply2(OptContext *ctx, TCGOp *op) static bool fold_nand(OptContext *ctx, TCGOp *op) { - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xi_to_not(ctx, op, -1)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_neg(OptContext *ctx, TCGOp *op) static bool fold_nor(OptContext *ctx, TCGOp *op) { - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xi_to_not(ctx, op, 0)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool fold_not(OptContext *ctx, TCGOp *op) static bool fold_or(OptContext *ctx, TCGOp *op) { - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xi_to_x(ctx, op, 0) || fold_xx_to_x(ctx, op)) { return true; @@ -XXX,XX +XXX,XX @@ static bool fold_remainder(OptContext *ctx, TCGOp *op) static bool fold_setcond(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[3]; - int i = do_constant_folding_cond(ctx->type, op->args[1], op->args[2], cond); + int i; + if (swap_commutative(op->args[0], &op->args[1], &op->args[2])) { + op->args[3] = cond = tcg_swap_cond(cond); + } + + i = do_constant_folding_cond(ctx->type, op->args[1], op->args[2], cond); if (i >= 0) { return tcg_opt_gen_movi(ctx, op, op->args[0], i); } @@ -XXX,XX +XXX,XX @@ static bool fold_setcond(OptContext *ctx, TCGOp *op) static bool fold_setcond2(OptContext *ctx, TCGOp *op) { TCGCond cond = op->args[5]; - int i = do_constant_folding_cond2(&op->args[1], &op->args[3], cond); - int inv = 0; + int i, inv = 0; + if (swap_commutative2(&op->args[1], &op->args[3])) { + op->args[5] = cond = tcg_swap_cond(cond); + } + + i = do_constant_folding_cond2(&op->args[1], &op->args[3], cond); if (i >= 0) { goto do_setcond_const; } @@ -XXX,XX +XXX,XX @@ static bool fold_tcg_ld(OptContext *ctx, TCGOp *op) static bool fold_xor(OptContext *ctx, TCGOp *op) { - if (fold_const2(ctx, op) || + if (fold_const2_commutative(ctx, op) || fold_xx_to_i(ctx, op, 0) || fold_xi_to_x(ctx, op, 0) || fold_xi_to_not(ctx, op, -1)) { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) ctx.type = TCG_TYPE_I32; } - /* For commutative operations make constant second argument */ - switch (opc) { - CASE_OP_32_64_VEC(add): - CASE_OP_32_64_VEC(mul): - CASE_OP_32_64_VEC(and): - CASE_OP_32_64_VEC(or): - CASE_OP_32_64_VEC(xor): - CASE_OP_32_64(eqv): - CASE_OP_32_64(nand): - CASE_OP_32_64(nor): - CASE_OP_32_64(muluh): - CASE_OP_32_64(mulsh): - swap_commutative(op->args[0], &op->args[1], &op->args[2]); - break; - CASE_OP_32_64(brcond): - if (swap_commutative(-1, &op->args[0], &op->args[1])) { - op->args[2] = tcg_swap_cond(op->args[2]); - } - break; - CASE_OP_32_64(setcond): - if (swap_commutative(op->args[0], &op->args[1], &op->args[2])) { - op->args[3] = tcg_swap_cond(op->args[3]); - } - break; - CASE_OP_32_64(movcond): - if (swap_commutative(-1, &op->args[1], &op->args[2])) { - op->args[5] = tcg_swap_cond(op->args[5]); - } - /* For movcond, we canonicalize the "false" input reg to match - the destination reg so that the tcg backend can implement - a "move if true" operation. */ - if (swap_commutative(op->args[0], &op->args[4], &op->args[3])) { - op->args[5] = tcg_invert_cond(op->args[5]); - } - break; - CASE_OP_32_64(add2): - swap_commutative(op->args[0], &op->args[2], &op->args[4]); - swap_commutative(op->args[1], &op->args[3], &op->args[5]); - break; - CASE_OP_32_64(mulu2): - CASE_OP_32_64(muls2): - swap_commutative(op->args[0], &op->args[2], &op->args[3]); - break; - case INDEX_op_brcond2_i32: - if (swap_commutative2(&op->args[0], &op->args[2])) { - op->args[4] = tcg_swap_cond(op->args[4]); - } - break; - case INDEX_op_setcond2_i32: - if (swap_commutative2(&op->args[1], &op->args[3])) { - op->args[5] = tcg_swap_cond(op->args[5]); - } - break; - default: - break; - } - /* Assume all bits affected, and no bits known zero. */ ctx.a_mask = -1; ctx.z_mask = -1; -- 2.25.1 This "garbage" setting pre-dates the addition of the type changing opcodes INDEX_op_ext_i32_i64, INDEX_op_extu_i32_i64, and INDEX_op_extr{l,h}_i64_i32. So now we have a definitive points at which to adjust z_mask to eliminate such bits from the 32-bit operands. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 35 ++++++++++++++++------------------- 1 file changed, 16 insertions(+), 19 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static void init_ts_info(OptContext *ctx, TCGTemp *ts) ti->is_const = true; ti->val = ts->val; ti->z_mask = ts->val; - if (TCG_TARGET_REG_BITS > 32 && ts->type == TCG_TYPE_I32) { - /* High bits of a 32-bit quantity are garbage. */ - ti->z_mask |= ~0xffffffffull; - } } else { ti->is_const = false; ti->z_mask = -1; @@ -XXX,XX +XXX,XX @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) TCGTemp *src_ts = arg_temp(src); TempOptInfo *di; TempOptInfo *si; - uint64_t z_mask; TCGOpcode new_op; if (ts_are_copies(dst_ts, src_ts)) { @@ -XXX,XX +XXX,XX @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) op->args[0] = dst; op->args[1] = src; - z_mask = si->z_mask; - if (TCG_TARGET_REG_BITS > 32 && new_op == INDEX_op_mov_i32) { - /* High bits of the destination are now garbage. */ - z_mask |= ~0xffffffffull; - } - di->z_mask = z_mask; + di->z_mask = si->z_mask; if (src_ts->type == dst_ts->type) { TempOptInfo *ni = ts_info(si->next_copy); @@ -XXX,XX +XXX,XX @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) static bool tcg_opt_gen_movi(OptContext *ctx, TCGOp *op, TCGArg dst, uint64_t val) { - /* Convert movi to mov with constant temp. */ - TCGTemp *tv = tcg_constant_internal(ctx->type, val); + TCGTemp *tv; + if (ctx->type == TCG_TYPE_I32) { + val = (int32_t)val; + } + + /* Convert movi to mov with constant temp. */ + tv = tcg_constant_internal(ctx->type, val); init_ts_info(ctx, tv); return tcg_opt_gen_mov(ctx, op, dst, temp_arg(tv)); } @@ -XXX,XX +XXX,XX @@ static bool fold_masks(OptContext *ctx, TCGOp *op) uint64_t z_mask = ctx->z_mask; /* - * 32-bit ops generate 32-bit results. For the result is zero test - * below, we can ignore high bits, but for further optimizations we - * need to record that the high bits contain garbage. + * 32-bit ops generate 32-bit results, which for the purpose of + * simplifying tcg are sign-extended. Certainly that's how we + * represent our constants elsewhere. Note that the bits will + * be reset properly for a 64-bit value when encountering the + * type changing opcodes. */ if (ctx->type == TCG_TYPE_I32) { - ctx->z_mask |= MAKE_64BIT_MASK(32, 32); - a_mask &= MAKE_64BIT_MASK(0, 32); - z_mask &= MAKE_64BIT_MASK(0, 32); + a_mask = (int32_t)a_mask; + z_mask = (int32_t)z_mask; + ctx->z_mask = z_mask; } if (z_mask == 0) { -- 2.25.1 Recognize the constant function for or-complement. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_or(OptContext *ctx, TCGOp *op) static bool fold_orc(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || + fold_xx_to_i(ctx, op, -1) || fold_xi_to_x(ctx, op, -1) || fold_ix_to_not(ctx, op, 0)) { return true; -- 2.25.1 Recognize the identity function for low-part multiply. Suggested-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_movcond(OptContext *ctx, TCGOp *op) static bool fold_mul(OptContext *ctx, TCGOp *op) { if (fold_const2(ctx, op) || - fold_xi_to_i(ctx, op, 0)) { + fold_xi_to_i(ctx, op, 0) || + fold_xi_to_x(ctx, op, 1)) { return true; } return false; -- 2.25.1 Recognize the identity function for division. Suggested-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_deposit(OptContext *ctx, TCGOp *op) static bool fold_divide(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xi_to_x(ctx, op, 1)) { + return true; + } + return false; } static bool fold_dup(OptContext *ctx, TCGOp *op) -- 2.25.1 Recognize the constant function for remainder. Suggested-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_qemu_st(OptContext *ctx, TCGOp *op) static bool fold_remainder(OptContext *ctx, TCGOp *op) { - return fold_const2(ctx, op); + if (fold_const2(ctx, op) || + fold_xx_to_i(ctx, op, 0)) { + return true; + } + return false; } static bool fold_setcond(OptContext *ctx, TCGOp *op) -- 2.25.1 Certain targets, like riscv, produce signed 32-bit results. This can lead to lots of redundant extensions as values are manipulated. Begin by tracking only the obvious sign-extensions, and converting them to simple copies when possible. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 123 ++++++++++++++++++++++++++++++++++++++++--------- 1 file changed, 102 insertions(+), 21 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ typedef struct TempOptInfo { TCGTemp *next_copy; uint64_t val; uint64_t z_mask; /* mask bit is 0 if and only if value bit is 0 */ + uint64_t s_mask; /* a left-aligned mask of clrsb(value) bits. */ } TempOptInfo; typedef struct OptContext { @@ -XXX,XX +XXX,XX @@ typedef struct OptContext { /* In flight values from optimization. */ uint64_t a_mask; /* mask bit is 0 iff value identical to first input */ uint64_t z_mask; /* mask bit is 0 iff value bit is 0 */ + uint64_t s_mask; /* mask of clrsb(value) bits */ TCGType type; } OptContext; +/* Calculate the smask for a specific value. */ +static uint64_t smask_from_value(uint64_t value) +{ + int rep = clrsb64(value); + return ~(~0ull >> rep); +} + +/* + * Calculate the smask for a given set of known-zeros. + * If there are lots of zeros on the left, we can consider the remainder + * an unsigned field, and thus the corresponding signed field is one bit + * larger. + */ +static uint64_t smask_from_zmask(uint64_t zmask) +{ + /* + * Only the 0 bits are significant for zmask, thus the msb itself + * must be zero, else we have no sign information. + */ + int rep = clz64(zmask); + if (rep == 0) { + return 0; + } + rep -= 1; + return ~(~0ull >> rep); +} + static inline TempOptInfo *ts_info(TCGTemp *ts) { return ts->state_ptr; @@ -XXX,XX +XXX,XX @@ static void reset_ts(TCGTemp *ts) ti->prev_copy = ts; ti->is_const = false; ti->z_mask = -1; + ti->s_mask = 0; } static void reset_temp(TCGArg arg) @@ -XXX,XX +XXX,XX @@ static void init_ts_info(OptContext *ctx, TCGTemp *ts) ti->is_const = true; ti->val = ts->val; ti->z_mask = ts->val; + ti->s_mask = smask_from_value(ts->val); } else { ti->is_const = false; ti->z_mask = -1; + ti->s_mask = 0; } } @@ -XXX,XX +XXX,XX @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src) op->args[1] = src; di->z_mask = si->z_mask; + di->s_mask = si->s_mask; if (src_ts->type == dst_ts->type) { TempOptInfo *ni = ts_info(si->next_copy); @@ -XXX,XX +XXX,XX @@ static void finish_folding(OptContext *ctx, TCGOp *op) nb_oargs = def->nb_oargs; for (i = 0; i < nb_oargs; i++) { - reset_temp(op->args[i]); + TCGTemp *ts = arg_temp(op->args[i]); + reset_ts(ts); /* - * Save the corresponding known-zero bits mask for the + * Save the corresponding known-zero/sign bits mask for the * first output argument (only one supported so far). */ if (i == 0) { - arg_info(op->args[i])->z_mask = ctx->z_mask; + ts_info(ts)->z_mask = ctx->z_mask; + ts_info(ts)->s_mask = ctx->s_mask; } } } @@ -XXX,XX +XXX,XX @@ static bool fold_masks(OptContext *ctx, TCGOp *op) { uint64_t a_mask = ctx->a_mask; uint64_t z_mask = ctx->z_mask; + uint64_t s_mask = ctx->s_mask; /* * 32-bit ops generate 32-bit results, which for the purpose of @@ -XXX,XX +XXX,XX @@ static bool fold_masks(OptContext *ctx, TCGOp *op) if (ctx->type == TCG_TYPE_I32) { a_mask = (int32_t)a_mask; z_mask = (int32_t)z_mask; + s_mask |= MAKE_64BIT_MASK(32, 32); ctx->z_mask = z_mask; + ctx->s_mask = s_mask; } if (z_mask == 0) { @@ -XXX,XX +XXX,XX @@ static bool fold_brcond2(OptContext *ctx, TCGOp *op) static bool fold_bswap(OptContext *ctx, TCGOp *op) { - uint64_t z_mask, sign; + uint64_t z_mask, s_mask, sign; if (arg_is_const(op->args[1])) { uint64_t t = arg_info(op->args[1])->val; @@ -XXX,XX +XXX,XX @@ static bool fold_bswap(OptContext *ctx, TCGOp *op) } z_mask = arg_info(op->args[1])->z_mask; + switch (op->opc) { case INDEX_op_bswap16_i32: case INDEX_op_bswap16_i64: @@ -XXX,XX +XXX,XX @@ static bool fold_bswap(OptContext *ctx, TCGOp *op) default: g_assert_not_reached(); } + s_mask = smask_from_zmask(z_mask); switch (op->args[2] & (TCG_BSWAP_OZ | TCG_BSWAP_OS)) { case TCG_BSWAP_OZ: @@ -XXX,XX +XXX,XX @@ static bool fold_bswap(OptContext *ctx, TCGOp *op) /* If the sign bit may be 1, force all the bits above to 1. */ if (z_mask & sign) { z_mask |= sign; + s_mask = sign << 1; } break; default: /* The high bits are undefined: force all bits above the sign to 1. */ z_mask |= sign << 1; + s_mask = 0; break; } ctx->z_mask = z_mask; + ctx->s_mask = s_mask; return fold_masks(ctx, op); } @@ -XXX,XX +XXX,XX @@ static bool fold_eqv(OptContext *ctx, TCGOp *op) static bool fold_extract(OptContext *ctx, TCGOp *op) { uint64_t z_mask_old, z_mask; + int pos = op->args[2]; + int len = op->args[3]; if (arg_is_const(op->args[1])) { uint64_t t; t = arg_info(op->args[1])->val; - t = extract64(t, op->args[2], op->args[3]); + t = extract64(t, pos, len); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } z_mask_old = arg_info(op->args[1])->z_mask; - z_mask = extract64(z_mask_old, op->args[2], op->args[3]); - if (op->args[2] == 0) { + z_mask = extract64(z_mask_old, pos, len); + if (pos == 0) { ctx->a_mask = z_mask_old ^ z_mask; } ctx->z_mask = z_mask; + ctx->s_mask = smask_from_zmask(z_mask); return fold_masks(ctx, op); } @@ -XXX,XX +XXX,XX @@ static bool fold_extract2(OptContext *ctx, TCGOp *op) static bool fold_exts(OptContext *ctx, TCGOp *op) { - uint64_t z_mask_old, z_mask, sign; + uint64_t s_mask_old, s_mask, z_mask, sign; bool type_change = false; if (fold_const1(ctx, op)) { return true; } - z_mask_old = z_mask = arg_info(op->args[1])->z_mask; + z_mask = arg_info(op->args[1])->z_mask; + s_mask = arg_info(op->args[1])->s_mask; + s_mask_old = s_mask; switch (op->opc) { CASE_OP_32_64(ext8s): @@ -XXX,XX +XXX,XX @@ static bool fold_exts(OptContext *ctx, TCGOp *op) if (z_mask & sign) { z_mask |= sign; - } else if (!type_change) { - ctx->a_mask = z_mask_old ^ z_mask; } + s_mask |= sign << 1; + ctx->z_mask = z_mask; + ctx->s_mask = s_mask; + if (!type_change) { + ctx->a_mask = s_mask & ~s_mask_old; + } return fold_masks(ctx, op); } @@ -XXX,XX +XXX,XX @@ static bool fold_extu(OptContext *ctx, TCGOp *op) } ctx->z_mask = z_mask; + ctx->s_mask = smask_from_zmask(z_mask); if (!type_change) { ctx->a_mask = z_mask_old ^ z_mask; } @@ -XXX,XX +XXX,XX @@ static bool fold_qemu_ld(OptContext *ctx, TCGOp *op) MemOp mop = get_memop(oi); int width = 8 * memop_size(mop); - if (!(mop & MO_SIGN) && width < 64) { - ctx->z_mask = MAKE_64BIT_MASK(0, width); + if (width < 64) { + ctx->s_mask = MAKE_64BIT_MASK(width, 64 - width); + if (!(mop & MO_SIGN)) { + ctx->z_mask = MAKE_64BIT_MASK(0, width); + ctx->s_mask <<= 1; + } } /* Opcodes that touch guest memory stop the mb optimization. */ @@ -XXX,XX +XXX,XX @@ static bool fold_setcond2(OptContext *ctx, TCGOp *op) static bool fold_sextract(OptContext *ctx, TCGOp *op) { - int64_t z_mask_old, z_mask; + uint64_t z_mask, s_mask, s_mask_old; + int pos = op->args[2]; + int len = op->args[3]; if (arg_is_const(op->args[1])) { uint64_t t; t = arg_info(op->args[1])->val; - t = sextract64(t, op->args[2], op->args[3]); + t = sextract64(t, pos, len); return tcg_opt_gen_movi(ctx, op, op->args[0], t); } - z_mask_old = arg_info(op->args[1])->z_mask; - z_mask = sextract64(z_mask_old, op->args[2], op->args[3]); - if (op->args[2] == 0 && z_mask >= 0) { - ctx->a_mask = z_mask_old ^ z_mask; - } + z_mask = arg_info(op->args[1])->z_mask; + z_mask = sextract64(z_mask, pos, len); ctx->z_mask = z_mask; + s_mask_old = arg_info(op->args[1])->s_mask; + s_mask = sextract64(s_mask_old, pos, len); + s_mask |= MAKE_64BIT_MASK(len, 64 - len); + ctx->s_mask = s_mask; + + if (pos == 0) { + ctx->a_mask = s_mask & ~s_mask_old; + } + return fold_masks(ctx, op); } @@ -XXX,XX +XXX,XX @@ static bool fold_tcg_ld(OptContext *ctx, TCGOp *op) { /* We can't do any folding with a load, but we can record bits. */ switch (op->opc) { + CASE_OP_32_64(ld8s): + ctx->s_mask = MAKE_64BIT_MASK(8, 56); + break; CASE_OP_32_64(ld8u): ctx->z_mask = MAKE_64BIT_MASK(0, 8); + ctx->s_mask = MAKE_64BIT_MASK(9, 55); + break; + CASE_OP_32_64(ld16s): + ctx->s_mask = MAKE_64BIT_MASK(16, 48); break; CASE_OP_32_64(ld16u): ctx->z_mask = MAKE_64BIT_MASK(0, 16); + ctx->s_mask = MAKE_64BIT_MASK(17, 47); + break; + case INDEX_op_ld32s_i64: + ctx->s_mask = MAKE_64BIT_MASK(32, 32); break; case INDEX_op_ld32u_i64: ctx->z_mask = MAKE_64BIT_MASK(0, 32); + ctx->s_mask = MAKE_64BIT_MASK(33, 31); break; default: g_assert_not_reached(); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) ctx.type = TCG_TYPE_I32; } - /* Assume all bits affected, and no bits known zero. */ + /* Assume all bits affected, no bits known zero, no sign reps. */ ctx.a_mask = -1; ctx.z_mask = -1; + ctx.s_mask = 0; /* * Process each opcode. @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) case INDEX_op_extrh_i64_i32: done = fold_extu(&ctx, op); break; + CASE_OP_32_64(ld8s): CASE_OP_32_64(ld8u): + CASE_OP_32_64(ld16s): CASE_OP_32_64(ld16u): + case INDEX_op_ld32s_i64: case INDEX_op_ld32u_i64: done = fold_tcg_ld(&ctx, op); break; -- 2.25.1 Sign repetitions are perforce all identical, whether they are 1 or 0. Bitwise operations preserve the relative quantity of the repetitions. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_and(OptContext *ctx, TCGOp *op) z2 = arg_info(op->args[2])->z_mask; ctx->z_mask = z1 & z2; + /* + * Sign repetitions are perforce all identical, whether they are 1 or 0. + * Bitwise operations preserve the relative quantity of the repetitions. + */ + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; + /* * Known-zeros does not imply known-ones. Therefore unless * arg2 is constant, we can't infer affected bits from it. @@ -XXX,XX +XXX,XX @@ static bool fold_andc(OptContext *ctx, TCGOp *op) } ctx->z_mask = z1; + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; return fold_masks(ctx, op); } @@ -XXX,XX +XXX,XX @@ static bool fold_eqv(OptContext *ctx, TCGOp *op) fold_xi_to_not(ctx, op, 0)) { return true; } + + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_movcond(OptContext *ctx, TCGOp *op) ctx->z_mask = arg_info(op->args[3])->z_mask | arg_info(op->args[4])->z_mask; + ctx->s_mask = arg_info(op->args[3])->s_mask + & arg_info(op->args[4])->s_mask; if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { uint64_t tv = arg_info(op->args[3])->val; @@ -XXX,XX +XXX,XX @@ static bool fold_nand(OptContext *ctx, TCGOp *op) fold_xi_to_not(ctx, op, -1)) { return true; } + + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_nor(OptContext *ctx, TCGOp *op) fold_xi_to_not(ctx, op, 0)) { return true; } + + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_not(OptContext *ctx, TCGOp *op) return true; } + ctx->s_mask = arg_info(op->args[1])->s_mask; + /* Because of fold_to_not, we want to always return true, via finish. */ finish_folding(ctx, op); return true; @@ -XXX,XX +XXX,XX @@ static bool fold_or(OptContext *ctx, TCGOp *op) ctx->z_mask = arg_info(op->args[1])->z_mask | arg_info(op->args[2])->z_mask; + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; return fold_masks(ctx, op); } @@ -XXX,XX +XXX,XX @@ static bool fold_orc(OptContext *ctx, TCGOp *op) fold_ix_to_not(ctx, op, 0)) { return true; } + + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_xor(OptContext *ctx, TCGOp *op) ctx->z_mask = arg_info(op->args[1])->z_mask | arg_info(op->args[2])->z_mask; + ctx->s_mask = arg_info(op->args[1])->s_mask + & arg_info(op->args[2])->s_mask; return fold_masks(ctx, op); } -- 2.25.1 The result is either 0 or 1, which means that we have a 2 bit signed result, and thus 62 bits of sign. For clarity, use the smask_from_zmask function. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_setcond(OptContext *ctx, TCGOp *op) } ctx->z_mask = 1; + ctx->s_mask = smask_from_zmask(1); return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_setcond2(OptContext *ctx, TCGOp *op) } ctx->z_mask = 1; + ctx->s_mask = smask_from_zmask(1); return false; do_setcond_const: -- 2.25.1 The results are generally 6 bit unsigned values, though the count leading and trailing bits may produce any value for a zero input. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static bool fold_count_zeros(OptContext *ctx, TCGOp *op) g_assert_not_reached(); } ctx->z_mask = arg_info(op->args[2])->z_mask | z_mask; - + ctx->s_mask = smask_from_zmask(ctx->z_mask); return false; } @@ -XXX,XX +XXX,XX @@ static bool fold_ctpop(OptContext *ctx, TCGOp *op) default: g_assert_not_reached(); } + ctx->s_mask = smask_from_zmask(ctx->z_mask); return false; } -- 2.25.1 For constant shifts, we can simply shift the s_mask. For variable shifts, we know that sar does not reduce the s_mask, which helps for sequences like ext32s_i64 t, in sar_i64 t, t, v ext32s_i64 out, t allowing the final extend to be eliminated. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Luis Pires <luis.pires@eldorado.org.br> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 47 insertions(+), 3 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ static uint64_t smask_from_zmask(uint64_t zmask) return ~(~0ull >> rep); } +/* + * Recreate a properly left-aligned smask after manipulation. + * Some bit-shuffling, particularly shifts and rotates, may + * retain sign bits on the left, but may scatter disconnected + * sign bits on the right. Retain only what remains to the left. + */ +static uint64_t smask_from_smask(int64_t smask) +{ + /* Only the 1 bits are significant for smask */ + return smask_from_zmask(~smask); +} + static inline TempOptInfo *ts_info(TCGTemp *ts) { return ts->state_ptr; @@ -XXX,XX +XXX,XX @@ static bool fold_sextract(OptContext *ctx, TCGOp *op) static bool fold_shift(OptContext *ctx, TCGOp *op) { + uint64_t s_mask, z_mask, sign; + if (fold_const2(ctx, op) || fold_ix_to_i(ctx, op, 0) || fold_xi_to_x(ctx, op, 0)) { return true; } + s_mask = arg_info(op->args[1])->s_mask; + z_mask = arg_info(op->args[1])->z_mask; + if (arg_is_const(op->args[2])) { - ctx->z_mask = do_constant_folding(op->opc, ctx->type, - arg_info(op->args[1])->z_mask, - arg_info(op->args[2])->val); + int sh = arg_info(op->args[2])->val; + + ctx->z_mask = do_constant_folding(op->opc, ctx->type, z_mask, sh); + + s_mask = do_constant_folding(op->opc, ctx->type, s_mask, sh); + ctx->s_mask = smask_from_smask(s_mask); + return fold_masks(ctx, op); } + + switch (op->opc) { + CASE_OP_32_64(sar): + /* + * Arithmetic right shift will not reduce the number of + * input sign repetitions. + */ + ctx->s_mask = s_mask; + break; + CASE_OP_32_64(shr): + /* + * If the sign bit is known zero, then logical right shift + * will not reduced the number of input sign repetitions. + */ + sign = (s_mask & -s_mask) >> 1; + if (!(z_mask & sign)) { + ctx->s_mask = s_mask; + } + break; + default: + break; + } + return false; } -- 2.25.1 The following changes since commit 8844bb8d896595ee1d25d21c770e6e6f29803097: Merge tag 'or1k-pull-request-20230513' of https://github.com/stffrdhrn/qemu into staging (2023-05-13 11:23:14 +0100) are available in the Git repository at: https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20230516 for you to fetch changes up to ee95d036bf4bfa10be65325a287bf3d0e8b2a0e6: tcg: Split out exec/user/guest-base.h (2023-05-16 08:11:53 -0700) ---------------------------------------------------------------- tcg/i386: Fix tcg_out_addi_ptr for win64 tcg: Implement atomicity for TCGv_i128 tcg: First quarter of cleanups for building tcg once ---------------------------------------------------------------- Richard Henderson (80): tcg/i386: Set P_REXW in tcg_out_addi_ptr include/exec/memop: Add MO_ATOM_* accel/tcg: Honor atomicity of loads accel/tcg: Honor atomicity of stores tcg: Unify helper_{be,le}_{ld,st}* accel/tcg: Implement helper_{ld,st}*_mmu for user-only tcg/tci: Use helper_{ld,st}*_mmu for user-only tcg: Add 128-bit guest memory primitives meson: Detect atomic128 support with optimization tcg/i386: Add have_atomic16 tcg/aarch64: Detect have_lse, have_lse2 for linux tcg/aarch64: Detect have_lse, have_lse2 for darwin tcg/i386: Use full load/store helpers in user-only mode tcg/aarch64: Use full load/store helpers in user-only mode tcg/ppc: Use full load/store helpers in user-only mode tcg/loongarch64: Use full load/store helpers in user-only mode tcg/riscv: Use full load/store helpers in user-only mode tcg/arm: Adjust constraints on qemu_ld/st tcg/arm: Use full load/store helpers in user-only mode tcg/mips: Use full load/store helpers in user-only mode tcg/s390x: Use full load/store helpers in user-only mode tcg/sparc64: Allocate %g2 as a third temporary tcg/sparc64: Rename tcg_out_movi_imm13 to tcg_out_movi_s13 target/sparc64: Remove tcg_out_movi_s13 case from tcg_out_movi_imm32 tcg/sparc64: Rename tcg_out_movi_imm32 to tcg_out_movi_u32 tcg/sparc64: Split out tcg_out_movi_s32 tcg/sparc64: Use standard slow path for softmmu accel/tcg: Remove helper_unaligned_{ld,st} tcg/loongarch64: Check the host supports unaligned accesses tcg/loongarch64: Support softmmu unaligned accesses tcg/riscv: Support softmmu unaligned accesses tcg: Introduce tcg_target_has_memory_bswap tcg: Add INDEX_op_qemu_{ld,st}_i128 tcg: Introduce tcg_out_movext3 tcg: Merge tcg_out_helper_load_regs into caller tcg: Support TCG_TYPE_I128 in tcg_out_{ld,st}_helper_{args,ret} tcg: Introduce atom_and_align_for_opc tcg/i386: Use atom_and_align_for_opc tcg/aarch64: Use atom_and_align_for_opc tcg/arm: Use atom_and_align_for_opc tcg/loongarch64: Use atom_and_align_for_opc tcg/mips: Use atom_and_align_for_opc tcg/ppc: Use atom_and_align_for_opc tcg/riscv: Use atom_and_align_for_opc tcg/s390x: Use atom_and_align_for_opc tcg/sparc64: Use atom_and_align_for_opc tcg/i386: Honor 64-bit atomicity in 32-bit mode tcg/i386: Support 128-bit load/store with have_atomic16 tcg/aarch64: Rename temporaries tcg/aarch64: Support 128-bit load/store tcg/ppc: Support 128-bit load/store tcg/s390x: Support 128-bit load/store tcg: Split out memory ops to tcg-op-ldst.c tcg: Widen gen_insn_data to uint64_t accel/tcg: Widen tcg-ldst.h addresses to uint64_t tcg: Widen helper_{ld,st}_i128 addresses to uint64_t tcg: Widen helper_atomic_* addresses to uint64_t tcg: Widen tcg_gen_code pc_start argument to uint64_t accel/tcg: Merge gen_mem_wrapped with plugin_gen_empty_mem_callback accel/tcg: Merge do_gen_mem_cb into caller tcg: Reduce copies for plugin_gen_mem_callbacks accel/tcg: Widen plugin_gen_empty_mem_callback to i64 tcg: Add addr_type to TCGContext tcg: Remove TCGv from tcg_gen_qemu_{ld,st}_* tcg: Remove TCGv from tcg_gen_atomic_* tcg: Split INDEX_op_qemu_{ld,st}* for guest address size tcg/tci: Elimnate TARGET_LONG_BITS, target_ulong tcg/i386: Always enable TCG_TARGET_HAS_extr[lh]_i64_i32 tcg/i386: Conditionalize tcg_out_extu_i32_i64 tcg/i386: Adjust type of tlb_mask tcg/i386: Remove TARGET_LONG_BITS, TCG_TYPE_TL tcg/arm: Remove TARGET_LONG_BITS tcg/aarch64: Remove USE_GUEST_BASE tcg/aarch64: Remove TARGET_LONG_BITS, TCG_TYPE_TL tcg/loongarch64: Remove TARGET_LONG_BITS, TCG_TYPE_TL tcg/mips: Remove TARGET_LONG_BITS, TCG_TYPE_TL tcg: Remove TARGET_LONG_BITS, TCG_TYPE_TL tcg: Add page_bits and page_mask to TCGContext tcg: Add tlb_dyn_max_bits to TCGContext tcg: Split out exec/user/guest-base.h docs/devel/loads-stores.rst | 36 +- docs/devel/tcg-ops.rst | 11 +- meson.build | 52 +- accel/tcg/tcg-runtime.h | 49 +- include/exec/cpu-all.h | 5 +- include/exec/memop.h | 37 ++ include/exec/plugin-gen.h | 4 +- include/exec/user/guest-base.h | 12 + include/qemu/cpuid.h | 18 + include/tcg/tcg-ldst.h | 72 +-- include/tcg/tcg-op.h | 273 ++++++--- include/tcg/tcg-opc.h | 41 +- include/tcg/tcg.h | 39 +- tcg/aarch64/tcg-target-con-set.h | 2 + tcg/aarch64/tcg-target.h | 15 +- tcg/arm/tcg-target-con-set.h | 16 +- tcg/arm/tcg-target-con-str.h | 5 +- tcg/arm/tcg-target.h | 3 +- tcg/i386/tcg-target.h | 13 +- tcg/loongarch64/tcg-target.h | 3 +- tcg/mips/tcg-target.h | 4 +- tcg/ppc/tcg-target-con-set.h | 2 + tcg/ppc/tcg-target-con-str.h | 1 + tcg/ppc/tcg-target.h | 4 +- tcg/riscv/tcg-target.h | 4 +- tcg/s390x/tcg-target-con-set.h | 2 + tcg/s390x/tcg-target.h | 4 +- tcg/sparc64/tcg-target-con-set.h | 2 - tcg/sparc64/tcg-target-con-str.h | 1 - tcg/sparc64/tcg-target.h | 4 +- tcg/tcg-internal.h | 2 + tcg/tci/tcg-target.h | 4 +- accel/tcg/cputlb.c | 839 ++++++++++++++++--------- accel/tcg/plugin-gen.c | 68 +- accel/tcg/translate-all.c | 35 +- accel/tcg/user-exec.c | 488 ++++++++++----- tcg/optimize.c | 19 +- tcg/tcg-op-ldst.c | 1234 +++++++++++++++++++++++++++++++++++++ tcg/tcg-op.c | 864 -------------------------- tcg/tcg.c | 627 +++++++++++++++---- tcg/tci.c | 243 +++----- accel/tcg/atomic_common.c.inc | 14 +- accel/tcg/ldst_atomicity.c.inc | 1262 ++++++++++++++++++++++++++++++++++++++ tcg/aarch64/tcg-target.c.inc | 438 ++++++++----- tcg/arm/tcg-target.c.inc | 246 +++----- tcg/i386/tcg-target.c.inc | 467 ++++++++++---- tcg/loongarch64/tcg-target.c.inc | 123 ++-- tcg/mips/tcg-target.c.inc | 216 +++---- tcg/ppc/tcg-target.c.inc | 300 +++++---- tcg/riscv/tcg-target.c.inc | 161 ++--- tcg/s390x/tcg-target.c.inc | 207 ++++--- tcg/sparc64/tcg-target.c.inc | 731 ++++++++-------------- tcg/tci/tcg-target.c.inc | 58 +- tcg/meson.build | 1 + 54 files changed, 5988 insertions(+), 3393 deletions(-) create mode 100644 include/exec/user/guest-base.h create mode 100644 tcg/tcg-op-ldst.c create mode 100644 accel/tcg/ldst_atomicity.c.inc The REXW bit must be set to produce a 64-bit pointer result; the bit is disabled in 32-bit mode, so we can do this unconditionally. Fixes: 7d9e1ee424b0 ("tcg/i386: Adjust assert in tcg_out_addi_ptr") Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1592 Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1642 Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_addi_ptr(TCGContext *s, TCGReg rd, TCGReg rs, { /* This function is only used for passing structs by reference. */ tcg_debug_assert(imm == (int32_t)imm); - tcg_out_modrm_offset(s, OPC_LEA, rd, rs, imm); + tcg_out_modrm_offset(s, OPC_LEA | P_REXW, rd, rs, imm); } static inline void tcg_out_pushi(TCGContext *s, tcg_target_long val) -- 2.34.1 This field may be used to describe the precise atomicity requirements of the guest, which may then be used to constrain the methods by which it may be emulated by the host. For instance, the AArch64 LDP (32-bit) instruction changes semantics with ARMv8.4 LSE2, from MO_64 | MO_ATOM_IFALIGN_PAIR (64-bits, single-copy atomic only on 4 byte units, nonatomic if not aligned by 4), to MO_64 | MO_ATOM_WITHIN16 (64-bits, single-copy atomic within a 16 byte block) The former may be implemented with two 4 byte loads, or a single 8 byte load if that happens to be efficient on the host. The latter may not be implemented with two 4 byte loads and may also require a helper when misaligned. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/memop.h | 37 +++++++++++++++++++++++++++++++++++++ tcg/tcg.c | 27 +++++++++++++++++++++------ 2 files changed, 58 insertions(+), 6 deletions(-) diff --git a/include/exec/memop.h b/include/exec/memop.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/memop.h +++ b/include/exec/memop.h @@ -XXX,XX +XXX,XX @@ typedef enum MemOp { MO_ALIGN_64 = 6 << MO_ASHIFT, MO_ALIGN = MO_AMASK, + /* + * MO_ATOM_* describes the atomicity requirements of the operation: + * MO_ATOM_IFALIGN: the operation must be single-copy atomic if it + * is aligned; if unaligned there is no atomicity. + * MO_ATOM_IFALIGN_PAIR: the entire operation may be considered to + * be a pair of half-sized operations which are packed together + * for convenience, with single-copy atomicity on each half if + * the half is aligned. + * This is the atomicity e.g. of Arm pre-FEAT_LSE2 LDP. + * MO_ATOM_WITHIN16: the operation is single-copy atomic, even if it + * is unaligned, so long as it does not cross a 16-byte boundary; + * if it crosses a 16-byte boundary there is no atomicity. + * This is the atomicity e.g. of Arm FEAT_LSE2 LDR. + * MO_ATOM_WITHIN16_PAIR: the entire operation is single-copy atomic, + * if it happens to be within a 16-byte boundary, otherwise it + * devolves to a pair of half-sized MO_ATOM_WITHIN16 operations. + * Depending on alignment, one or both will be single-copy atomic. + * This is the atomicity e.g. of Arm FEAT_LSE2 LDP. + * MO_ATOM_SUBALIGN: the operation is single-copy atomic by parts + * by the alignment. E.g. if the address is 0 mod 4, then each + * 4-byte subobject is single-copy atomic. + * This is the atomicity e.g. of IBM Power. + * MO_ATOM_NONE: the operation has no atomicity requirements. + * + * Note the default (i.e. 0) value is single-copy atomic to the + * size of the operation, if aligned. This retains the behaviour + * from before this field was introduced. + */ + MO_ATOM_SHIFT = 8, + MO_ATOM_IFALIGN = 0 << MO_ATOM_SHIFT, + MO_ATOM_IFALIGN_PAIR = 1 << MO_ATOM_SHIFT, + MO_ATOM_WITHIN16 = 2 << MO_ATOM_SHIFT, + MO_ATOM_WITHIN16_PAIR = 3 << MO_ATOM_SHIFT, + MO_ATOM_SUBALIGN = 4 << MO_ATOM_SHIFT, + MO_ATOM_NONE = 5 << MO_ATOM_SHIFT, + MO_ATOM_MASK = 7 << MO_ATOM_SHIFT, + /* Combinations of the above, for ease of use. */ MO_UB = MO_8, MO_UW = MO_16, diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static const char * const alignment_name[(MO_AMASK >> MO_ASHIFT) + 1] = { [MO_ALIGN_64 >> MO_ASHIFT] = "al64+", }; +static const char * const atom_name[(MO_ATOM_MASK >> MO_ATOM_SHIFT) + 1] = { + [MO_ATOM_IFALIGN >> MO_ATOM_SHIFT] = "", + [MO_ATOM_IFALIGN_PAIR >> MO_ATOM_SHIFT] = "pair+", + [MO_ATOM_WITHIN16 >> MO_ATOM_SHIFT] = "w16+", + [MO_ATOM_WITHIN16_PAIR >> MO_ATOM_SHIFT] = "w16p+", + [MO_ATOM_SUBALIGN >> MO_ATOM_SHIFT] = "sub+", + [MO_ATOM_NONE >> MO_ATOM_SHIFT] = "noat+", +}; + static const char bswap_flag_name[][6] = { [TCG_BSWAP_IZ] = "iz", [TCG_BSWAP_OZ] = "oz", @@ -XXX,XX +XXX,XX @@ static void tcg_dump_ops(TCGContext *s, FILE *f, bool have_prefs) case INDEX_op_qemu_ld_i64: case INDEX_op_qemu_st_i64: { + const char *s_al, *s_op, *s_at; MemOpIdx oi = op->args[k++]; MemOp op = get_memop(oi); unsigned ix = get_mmuidx(oi); - if (op & ~(MO_AMASK | MO_BSWAP | MO_SSIZE)) { - col += ne_fprintf(f, ",$0x%x,%u", op, ix); + s_al = alignment_name[(op & MO_AMASK) >> MO_ASHIFT]; + s_op = ldst_name[op & (MO_BSWAP | MO_SSIZE)]; + s_at = atom_name[(op & MO_ATOM_MASK) >> MO_ATOM_SHIFT]; + op &= ~(MO_AMASK | MO_BSWAP | MO_SSIZE | MO_ATOM_MASK); + + /* If all fields are accounted for, print symbolically. */ + if (!op && s_al && s_op && s_at) { + col += ne_fprintf(f, ",%s%s%s,%u", + s_at, s_al, s_op, ix); } else { - const char *s_al, *s_op; - s_al = alignment_name[(op & MO_AMASK) >> MO_ASHIFT]; - s_op = ldst_name[op & (MO_BSWAP | MO_SSIZE)]; - col += ne_fprintf(f, ",%s%s,%u", s_al, s_op, ix); + op = get_memop(oi); + col += ne_fprintf(f, ",$0x%x,%u", op, ix); } i = 1; } -- 2.34.1 Create ldst_atomicity.c.inc. Not required for user-only code loads, because we've ensured that the page is read-only before beginning to translate code. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/cputlb.c | 175 +++++++--- accel/tcg/user-exec.c | 26 +- accel/tcg/ldst_atomicity.c.inc | 566 +++++++++++++++++++++++++++++++++ 3 files changed, 716 insertions(+), 51 deletions(-) create mode 100644 accel/tcg/ldst_atomicity.c.inc diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -XXX,XX +XXX,XX @@ tb_page_addr_t get_page_addr_code_hostp(CPUArchState *env, target_ulong addr, return qemu_ram_addr_from_host_nofail(p); } +/* Load/store with atomicity primitives. */ +#include "ldst_atomicity.c.inc" + #ifdef CONFIG_PLUGIN /* * Perform a TLB lookup and populate the qemu_plugin_hwaddr structure. @@ -XXX,XX +XXX,XX @@ static void validate_memop(MemOpIdx oi, MemOp expected) * specifically for reading instructions from system memory. It is * called by the translation loop and in some helpers where the code * is disassembled. It shouldn't be called directly by guest code. - */ - -typedef uint64_t FullLoadHelper(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); - -static inline uint64_t QEMU_ALWAYS_INLINE -load_memop(const void *haddr, MemOp op) -{ - switch (op) { - case MO_UB: - return ldub_p(haddr); - case MO_BEUW: - return lduw_be_p(haddr); - case MO_LEUW: - return lduw_le_p(haddr); - case MO_BEUL: - return (uint32_t)ldl_be_p(haddr); - case MO_LEUL: - return (uint32_t)ldl_le_p(haddr); - case MO_BEUQ: - return ldq_be_p(haddr); - case MO_LEUQ: - return ldq_le_p(haddr); - default: - qemu_build_not_reached(); - } -} - -/* + * * For the benefit of TCG generated code, we want to avoid the * complication of ABI-specific return type promotion and always * return a value extended to the register size of the host. This is @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld_bytes_beN(MMULookupPageData *p, uint64_t ret_be) return ret_be; } +/** + * do_ld_parts_beN + * @p: translation parameters + * @ret_be: accumulated data + * + * As do_ld_bytes_beN, but atomically on each aligned part. + */ +static uint64_t do_ld_parts_beN(MMULookupPageData *p, uint64_t ret_be) +{ + void *haddr = p->haddr; + int size = p->size; + + do { + uint64_t x; + int n; + + /* + * Find minimum of alignment and size. + * This is slightly stronger than required by MO_ATOM_SUBALIGN, which + * would have only checked the low bits of addr|size once at the start, + * but is just as easy. + */ + switch (((uintptr_t)haddr | size) & 7) { + case 4: + x = cpu_to_be32(load_atomic4(haddr)); + ret_be = (ret_be << 32) | x; + n = 4; + break; + case 2: + case 6: + x = cpu_to_be16(load_atomic2(haddr)); + ret_be = (ret_be << 16) | x; + n = 2; + break; + default: + x = *(uint8_t *)haddr; + ret_be = (ret_be << 8) | x; + n = 1; + break; + case 0: + g_assert_not_reached(); + } + haddr += n; + size -= n; + } while (size != 0); + return ret_be; +} + +/** + * do_ld_parts_be4 + * @p: translation parameters + * @ret_be: accumulated data + * + * As do_ld_bytes_beN, but with one atomic load. + * Four aligned bytes are guaranteed to cover the load. + */ +static uint64_t do_ld_whole_be4(MMULookupPageData *p, uint64_t ret_be) +{ + int o = p->addr & 3; + uint32_t x = load_atomic4(p->haddr - o); + + x = cpu_to_be32(x); + x <<= o * 8; + x >>= (4 - p->size) * 8; + return (ret_be << (p->size * 8)) | x; +} + +/** + * do_ld_parts_be8 + * @p: translation parameters + * @ret_be: accumulated data + * + * As do_ld_bytes_beN, but with one atomic load. + * Eight aligned bytes are guaranteed to cover the load. + */ +static uint64_t do_ld_whole_be8(CPUArchState *env, uintptr_t ra, + MMULookupPageData *p, uint64_t ret_be) +{ + int o = p->addr & 7; + uint64_t x = load_atomic8_or_exit(env, ra, p->haddr - o); + + x = cpu_to_be64(x); + x <<= o * 8; + x >>= (8 - p->size) * 8; + return (ret_be << (p->size * 8)) | x; +} + /* * Wrapper for the above. */ static uint64_t do_ld_beN(CPUArchState *env, MMULookupPageData *p, - uint64_t ret_be, int mmu_idx, - MMUAccessType type, uintptr_t ra) + uint64_t ret_be, int mmu_idx, MMUAccessType type, + MemOp mop, uintptr_t ra) { + MemOp atom; + unsigned tmp, half_size; + if (unlikely(p->flags & TLB_MMIO)) { return do_ld_mmio_beN(env, p, ret_be, mmu_idx, type, ra); - } else { + } + + /* + * It is a given that we cross a page and therefore there is no + * atomicity for the load as a whole, but subobjects may need attention. + */ + atom = mop & MO_ATOM_MASK; + switch (atom) { + case MO_ATOM_SUBALIGN: + return do_ld_parts_beN(p, ret_be); + + case MO_ATOM_IFALIGN_PAIR: + case MO_ATOM_WITHIN16_PAIR: + tmp = mop & MO_SIZE; + tmp = tmp ? tmp - 1 : 0; + half_size = 1 << tmp; + if (atom == MO_ATOM_IFALIGN_PAIR + ? p->size == half_size + : p->size >= half_size) { + if (!HAVE_al8_fast && p->size < 4) { + return do_ld_whole_be4(p, ret_be); + } else { + return do_ld_whole_be8(env, ra, p, ret_be); + } + } + /* fall through */ + + case MO_ATOM_IFALIGN: + case MO_ATOM_WITHIN16: + case MO_ATOM_NONE: return do_ld_bytes_beN(p, ret_be); + + default: + g_assert_not_reached(); } } @@ -XXX,XX +XXX,XX @@ static uint16_t do_ld_2(CPUArchState *env, MMULookupPageData *p, int mmu_idx, } /* Perform the load host endian, then swap if necessary. */ - ret = load_memop(p->haddr, MO_UW); + ret = load_atom_2(env, ra, p->haddr, memop); if (memop & MO_BSWAP) { ret = bswap16(ret); } @@ -XXX,XX +XXX,XX @@ static uint32_t do_ld_4(CPUArchState *env, MMULookupPageData *p, int mmu_idx, } /* Perform the load host endian. */ - ret = load_memop(p->haddr, MO_UL); + ret = load_atom_4(env, ra, p->haddr, memop); if (memop & MO_BSWAP) { ret = bswap32(ret); } @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld_8(CPUArchState *env, MMULookupPageData *p, int mmu_idx, } /* Perform the load host endian. */ - ret = load_memop(p->haddr, MO_UQ); + ret = load_atom_8(env, ra, p->haddr, memop); if (memop & MO_BSWAP) { ret = bswap64(ret); } @@ -XXX,XX +XXX,XX @@ static uint32_t do_ld4_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return do_ld_4(env, &l.page[0], l.mmu_idx, access_type, l.memop, ra); } - ret = do_ld_beN(env, &l.page[0], 0, l.mmu_idx, access_type, ra); - ret = do_ld_beN(env, &l.page[1], ret, l.mmu_idx, access_type, ra); + ret = do_ld_beN(env, &l.page[0], 0, l.mmu_idx, access_type, l.memop, ra); + ret = do_ld_beN(env, &l.page[1], ret, l.mmu_idx, access_type, l.memop, ra); if ((l.memop & MO_BSWAP) == MO_LE) { ret = bswap32(ret); } @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld8_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return do_ld_8(env, &l.page[0], l.mmu_idx, access_type, l.memop, ra); } - ret = do_ld_beN(env, &l.page[0], 0, l.mmu_idx, access_type, ra); - ret = do_ld_beN(env, &l.page[1], ret, l.mmu_idx, access_type, ra); + ret = do_ld_beN(env, &l.page[0], 0, l.mmu_idx, access_type, l.memop, ra); + ret = do_ld_beN(env, &l.page[1], ret, l.mmu_idx, access_type, l.memop, ra); if ((l.memop & MO_BSWAP) == MO_LE) { ret = bswap64(ret); } diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/user-exec.c +++ b/accel/tcg/user-exec.c @@ -XXX,XX +XXX,XX @@ static void *cpu_mmu_lookup(CPUArchState *env, target_ulong addr, return ret; } +#include "ldst_atomicity.c.inc" + uint8_t cpu_ldb_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { @@ -XXX,XX +XXX,XX @@ uint16_t cpu_ldw_be_mmu(CPUArchState *env, abi_ptr addr, validate_memop(oi, MO_BEUW); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = lduw_be_p(haddr); + ret = load_atom_2(env, ra, haddr, get_memop(oi)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return ret; + return cpu_to_be16(ret); } uint32_t cpu_ldl_be_mmu(CPUArchState *env, abi_ptr addr, @@ -XXX,XX +XXX,XX @@ uint32_t cpu_ldl_be_mmu(CPUArchState *env, abi_ptr addr, validate_memop(oi, MO_BEUL); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = ldl_be_p(haddr); + ret = load_atom_4(env, ra, haddr, get_memop(oi)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return ret; + return cpu_to_be32(ret); } uint64_t cpu_ldq_be_mmu(CPUArchState *env, abi_ptr addr, @@ -XXX,XX +XXX,XX @@ uint64_t cpu_ldq_be_mmu(CPUArchState *env, abi_ptr addr, validate_memop(oi, MO_BEUQ); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = ldq_be_p(haddr); + ret = load_atom_8(env, ra, haddr, get_memop(oi)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return ret; + return cpu_to_be64(ret); } uint16_t cpu_ldw_le_mmu(CPUArchState *env, abi_ptr addr, @@ -XXX,XX +XXX,XX @@ uint16_t cpu_ldw_le_mmu(CPUArchState *env, abi_ptr addr, validate_memop(oi, MO_LEUW); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = lduw_le_p(haddr); + ret = load_atom_2(env, ra, haddr, get_memop(oi)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return ret; + return cpu_to_le16(ret); } uint32_t cpu_ldl_le_mmu(CPUArchState *env, abi_ptr addr, @@ -XXX,XX +XXX,XX @@ uint32_t cpu_ldl_le_mmu(CPUArchState *env, abi_ptr addr, validate_memop(oi, MO_LEUL); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = ldl_le_p(haddr); + ret = load_atom_4(env, ra, haddr, get_memop(oi)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return ret; + return cpu_to_le32(ret); } uint64_t cpu_ldq_le_mmu(CPUArchState *env, abi_ptr addr, @@ -XXX,XX +XXX,XX @@ uint64_t cpu_ldq_le_mmu(CPUArchState *env, abi_ptr addr, validate_memop(oi, MO_LEUQ); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = ldq_le_p(haddr); + ret = load_atom_8(env, ra, haddr, get_memop(oi)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return ret; + return cpu_to_le64(ret); } Int128 cpu_ld16_be_mmu(CPUArchState *env, abi_ptr addr, diff --git a/accel/tcg/ldst_atomicity.c.inc b/accel/tcg/ldst_atomicity.c.inc new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/accel/tcg/ldst_atomicity.c.inc @@ -XXX,XX +XXX,XX @@ +/* + * Routines common to user and system emulation of load/store. + * + * Copyright (c) 2022 Linaro, Ltd. + * + * SPDX-License-Identifier: GPL-2.0-or-later + * + * This work is licensed under the terms of the GNU GPL, version 2 or later. + * See the COPYING file in the top-level directory. + */ + +#ifdef CONFIG_ATOMIC64 +# define HAVE_al8 true +#else +# define HAVE_al8 false +#endif +#define HAVE_al8_fast (ATOMIC_REG_SIZE >= 8) + +#if defined(CONFIG_ATOMIC128) +# define HAVE_al16_fast true +#else +# define HAVE_al16_fast false +#endif + +/** + * required_atomicity: + * + * Return the lg2 bytes of atomicity required by @memop for @p. + * If the operation must be split into two operations to be + * examined separately for atomicity, return -lg2. + */ +static int required_atomicity(CPUArchState *env, uintptr_t p, MemOp memop) +{ + MemOp atom = memop & MO_ATOM_MASK; + MemOp size = memop & MO_SIZE; + MemOp half = size ? size - 1 : 0; + unsigned tmp; + int atmax; + + switch (atom) { + case MO_ATOM_NONE: + atmax = MO_8; + break; + + case MO_ATOM_IFALIGN_PAIR: + size = half; + /* fall through */ + + case MO_ATOM_IFALIGN: + tmp = (1 << size) - 1; + atmax = p & tmp ? MO_8 : size; + break; + + case MO_ATOM_WITHIN16: + tmp = p & 15; + atmax = (tmp + (1 << size) <= 16 ? size : MO_8); + break; + + case MO_ATOM_WITHIN16_PAIR: + tmp = p & 15; + if (tmp + (1 << size) <= 16) { + atmax = size; + } else if (tmp + (1 << half) == 16) { + /* + * The pair exactly straddles the boundary. + * Both halves are naturally aligned and atomic. + */ + atmax = half; + } else { + /* + * One of the pair crosses the boundary, and is non-atomic. + * The other of the pair does not cross, and is atomic. + */ + atmax = -half; + } + break; + + case MO_ATOM_SUBALIGN: + /* + * Examine the alignment of p to determine if there are subobjects + * that must be aligned. Note that we only really need ctz4() -- + * any more sigificant bits are discarded by the immediately + * following comparison. + */ + tmp = ctz32(p); + atmax = MIN(size, tmp); + break; + + default: + g_assert_not_reached(); + } + + /* + * Here we have the architectural atomicity of the operation. + * However, when executing in a serial context, we need no extra + * host atomicity in order to avoid racing. This reduction + * avoids looping with cpu_loop_exit_atomic. + */ + if (cpu_in_serial_context(env_cpu(env))) { + return MO_8; + } + return atmax; +} + +/** + * load_atomic2: + * @pv: host address + * + * Atomically load 2 aligned bytes from @pv. + */ +static inline uint16_t load_atomic2(void *pv) +{ + uint16_t *p = __builtin_assume_aligned(pv, 2); + return qatomic_read(p); +} + +/** + * load_atomic4: + * @pv: host address + * + * Atomically load 4 aligned bytes from @pv. + */ +static inline uint32_t load_atomic4(void *pv) +{ + uint32_t *p = __builtin_assume_aligned(pv, 4); + return qatomic_read(p); +} + +/** + * load_atomic8: + * @pv: host address + * + * Atomically load 8 aligned bytes from @pv. + */ +static inline uint64_t load_atomic8(void *pv) +{ + uint64_t *p = __builtin_assume_aligned(pv, 8); + + qemu_build_assert(HAVE_al8); + return qatomic_read__nocheck(p); +} + +/** + * load_atomic16: + * @pv: host address + * + * Atomically load 16 aligned bytes from @pv. + */ +static inline Int128 load_atomic16(void *pv) +{ +#ifdef CONFIG_ATOMIC128 + __uint128_t *p = __builtin_assume_aligned(pv, 16); + Int128Alias r; + + r.u = qatomic_read__nocheck(p); + return r.s; +#else + qemu_build_not_reached(); +#endif +} + +/** + * load_atomic8_or_exit: + * @env: cpu context + * @ra: host unwind address + * @pv: host address + * + * Atomically load 8 aligned bytes from @pv. + * If this is not possible, longjmp out to restart serially. + */ +static uint64_t load_atomic8_or_exit(CPUArchState *env, uintptr_t ra, void *pv) +{ + if (HAVE_al8) { + return load_atomic8(pv); + } + +#ifdef CONFIG_USER_ONLY + /* + * If the page is not writable, then assume the value is immutable + * and requires no locking. This ignores the case of MAP_SHARED with + * another process, because the fallback start_exclusive solution + * provides no protection across processes. + */ + if (!page_check_range(h2g(pv), 8, PAGE_WRITE)) { + uint64_t *p = __builtin_assume_aligned(pv, 8); + return *p; + } +#endif + + /* Ultimate fallback: re-execute in serial context. */ + cpu_loop_exit_atomic(env_cpu(env), ra); +} + +/** + * load_atomic16_or_exit: + * @env: cpu context + * @ra: host unwind address + * @pv: host address + * + * Atomically load 16 aligned bytes from @pv. + * If this is not possible, longjmp out to restart serially. + */ +static Int128 load_atomic16_or_exit(CPUArchState *env, uintptr_t ra, void *pv) +{ + Int128 *p = __builtin_assume_aligned(pv, 16); + + if (HAVE_al16_fast) { + return load_atomic16(p); + } + +#ifdef CONFIG_USER_ONLY + /* + * We can only use cmpxchg to emulate a load if the page is writable. + * If the page is not writable, then assume the value is immutable + * and requires no locking. This ignores the case of MAP_SHARED with + * another process, because the fallback start_exclusive solution + * provides no protection across processes. + */ + if (!page_check_range(h2g(p), 16, PAGE_WRITE)) { + return *p; + } +#endif + + /* + * In system mode all guest pages are writable, and for user-only + * we have just checked writability. Try cmpxchg. + */ +#if defined(CONFIG_CMPXCHG128) + /* Swap 0 with 0, with the side-effect of returning the old value. */ + { + Int128Alias r; + r.u = __sync_val_compare_and_swap_16((__uint128_t *)p, 0, 0); + return r.s; + } +#endif + + /* Ultimate fallback: re-execute in serial context. */ + cpu_loop_exit_atomic(env_cpu(env), ra); +} + +/** + * load_atom_extract_al4x2: + * @pv: host address + * + * Load 4 bytes from @p, from two sequential atomic 4-byte loads. + */ +static uint32_t load_atom_extract_al4x2(void *pv) +{ + uintptr_t pi = (uintptr_t)pv; + int sh = (pi & 3) * 8; + uint32_t a, b; + + pv = (void *)(pi & ~3); + a = load_atomic4(pv); + b = load_atomic4(pv + 4); + + if (HOST_BIG_ENDIAN) { + return (a << sh) | (b >> (-sh & 31)); + } else { + return (a >> sh) | (b << (-sh & 31)); + } +} + +/** + * load_atom_extract_al8x2: + * @pv: host address + * + * Load 8 bytes from @p, from two sequential atomic 8-byte loads. + */ +static uint64_t load_atom_extract_al8x2(void *pv) +{ + uintptr_t pi = (uintptr_t)pv; + int sh = (pi & 7) * 8; + uint64_t a, b; + + pv = (void *)(pi & ~7); + a = load_atomic8(pv); + b = load_atomic8(pv + 8); + + if (HOST_BIG_ENDIAN) { + return (a << sh) | (b >> (-sh & 63)); + } else { + return (a >> sh) | (b << (-sh & 63)); + } +} + +/** + * load_atom_extract_al8_or_exit: + * @env: cpu context + * @ra: host unwind address + * @pv: host address + * @s: object size in bytes, @s <= 4. + * + * Atomically load @s bytes from @p, when p % s != 0, and [p, p+s-1] does + * not cross an 8-byte boundary. This means that we can perform an atomic + * 8-byte load and extract. + * The value is returned in the low bits of a uint32_t. + */ +static uint32_t load_atom_extract_al8_or_exit(CPUArchState *env, uintptr_t ra, + void *pv, int s) +{ + uintptr_t pi = (uintptr_t)pv; + int o = pi & 7; + int shr = (HOST_BIG_ENDIAN ? 8 - s - o : o) * 8; + + pv = (void *)(pi & ~7); + return load_atomic8_or_exit(env, ra, pv) >> shr; +} + +/** + * load_atom_extract_al16_or_exit: + * @env: cpu context + * @ra: host unwind address + * @p: host address + * @s: object size in bytes, @s <= 8. + * + * Atomically load @s bytes from @p, when p % 16 < 8 + * and p % 16 + s > 8. I.e. does not cross a 16-byte + * boundary, but *does* cross an 8-byte boundary. + * This is the slow version, so we must have eliminated + * any faster load_atom_extract_al8_or_exit case. + * + * If this is not possible, longjmp out to restart serially. + */ +static uint64_t load_atom_extract_al16_or_exit(CPUArchState *env, uintptr_t ra, + void *pv, int s) +{ + uintptr_t pi = (uintptr_t)pv; + int o = pi & 7; + int shr = (HOST_BIG_ENDIAN ? 16 - s - o : o) * 8; + Int128 r; + + /* + * Note constraints above: p & 8 must be clear. + * Provoke SIGBUS if possible otherwise. + */ + pv = (void *)(pi & ~7); + r = load_atomic16_or_exit(env, ra, pv); + + r = int128_urshift(r, shr); + return int128_getlo(r); +} + +/** + * load_atom_extract_al16_or_al8: + * @p: host address + * @s: object size in bytes, @s <= 8. + * + * Load @s bytes from @p, when p % s != 0. If [p, p+s-1] does not + * cross an 16-byte boundary then the access must be 16-byte atomic, + * otherwise the access must be 8-byte atomic. + */ +static inline uint64_t load_atom_extract_al16_or_al8(void *pv, int s) +{ +#if defined(CONFIG_ATOMIC128) + uintptr_t pi = (uintptr_t)pv; + int o = pi & 7; + int shr = (HOST_BIG_ENDIAN ? 16 - s - o : o) * 8; + __uint128_t r; + + pv = (void *)(pi & ~7); + if (pi & 8) { + uint64_t *p8 = __builtin_assume_aligned(pv, 16, 8); + uint64_t a = qatomic_read__nocheck(p8); + uint64_t b = qatomic_read__nocheck(p8 + 1); + + if (HOST_BIG_ENDIAN) { + r = ((__uint128_t)a << 64) | b; + } else { + r = ((__uint128_t)b << 64) | a; + } + } else { + __uint128_t *p16 = __builtin_assume_aligned(pv, 16, 0); + r = qatomic_read__nocheck(p16); + } + return r >> shr; +#else + qemu_build_not_reached(); +#endif +} + +/** + * load_atom_4_by_2: + * @pv: host address + * + * Load 4 bytes from @pv, with two 2-byte atomic loads. + */ +static inline uint32_t load_atom_4_by_2(void *pv) +{ + uint32_t a = load_atomic2(pv); + uint32_t b = load_atomic2(pv + 2); + + if (HOST_BIG_ENDIAN) { + return (a << 16) | b; + } else { + return (b << 16) | a; + } +} + +/** + * load_atom_8_by_2: + * @pv: host address + * + * Load 8 bytes from @pv, with four 2-byte atomic loads. + */ +static inline uint64_t load_atom_8_by_2(void *pv) +{ + uint32_t a = load_atom_4_by_2(pv); + uint32_t b = load_atom_4_by_2(pv + 4); + + if (HOST_BIG_ENDIAN) { + return ((uint64_t)a << 32) | b; + } else { + return ((uint64_t)b << 32) | a; + } +} + +/** + * load_atom_8_by_4: + * @pv: host address + * + * Load 8 bytes from @pv, with two 4-byte atomic loads. + */ +static inline uint64_t load_atom_8_by_4(void *pv) +{ + uint32_t a = load_atomic4(pv); + uint32_t b = load_atomic4(pv + 4); + + if (HOST_BIG_ENDIAN) { + return ((uint64_t)a << 32) | b; + } else { + return ((uint64_t)b << 32) | a; + } +} + +/** + * load_atom_2: + * @p: host address + * @memop: the full memory op + * + * Load 2 bytes from @p, honoring the atomicity of @memop. + */ +static uint16_t load_atom_2(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop) +{ + uintptr_t pi = (uintptr_t)pv; + int atmax; + + if (likely((pi & 1) == 0)) { + return load_atomic2(pv); + } + if (HAVE_al16_fast) { + return load_atom_extract_al16_or_al8(pv, 2); + } + + atmax = required_atomicity(env, pi, memop); + switch (atmax) { + case MO_8: + return lduw_he_p(pv); + case MO_16: + /* The only case remaining is MO_ATOM_WITHIN16. */ + if (!HAVE_al8_fast && (pi & 3) == 1) { + /* Big or little endian, we want the middle two bytes. */ + return load_atomic4(pv - 1) >> 8; + } + if ((pi & 15) != 7) { + return load_atom_extract_al8_or_exit(env, ra, pv, 2); + } + return load_atom_extract_al16_or_exit(env, ra, pv, 2); + default: + g_assert_not_reached(); + } +} + +/** + * load_atom_4: + * @p: host address + * @memop: the full memory op + * + * Load 4 bytes from @p, honoring the atomicity of @memop. + */ +static uint32_t load_atom_4(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop) +{ + uintptr_t pi = (uintptr_t)pv; + int atmax; + + if (likely((pi & 3) == 0)) { + return load_atomic4(pv); + } + if (HAVE_al16_fast) { + return load_atom_extract_al16_or_al8(pv, 4); + } + + atmax = required_atomicity(env, pi, memop); + switch (atmax) { + case MO_8: + case MO_16: + case -MO_16: + /* + * For MO_ATOM_IFALIGN, this is more atomicity than required, + * but it's trivially supported on all hosts, better than 4 + * individual byte loads (when the host requires alignment), + * and overlaps with the MO_ATOM_SUBALIGN case of p % 2 == 0. + */ + return load_atom_extract_al4x2(pv); + case MO_32: + if (!(pi & 4)) { + return load_atom_extract_al8_or_exit(env, ra, pv, 4); + } + return load_atom_extract_al16_or_exit(env, ra, pv, 4); + default: + g_assert_not_reached(); + } +} + +/** + * load_atom_8: + * @p: host address + * @memop: the full memory op + * + * Load 8 bytes from @p, honoring the atomicity of @memop. + */ +static uint64_t load_atom_8(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop) +{ + uintptr_t pi = (uintptr_t)pv; + int atmax; + + /* + * If the host does not support 8-byte atomics, wait until we have + * examined the atomicity parameters below. + */ + if (HAVE_al8 && likely((pi & 7) == 0)) { + return load_atomic8(pv); + } + if (HAVE_al16_fast) { + return load_atom_extract_al16_or_al8(pv, 8); + } + + atmax = required_atomicity(env, pi, memop); + if (atmax == MO_64) { + if (!HAVE_al8 && (pi & 7) == 0) { + load_atomic8_or_exit(env, ra, pv); + } + return load_atom_extract_al16_or_exit(env, ra, pv, 8); + } + if (HAVE_al8_fast) { + return load_atom_extract_al8x2(pv); + } + switch (atmax) { + case MO_8: + return ldq_he_p(pv); + case MO_16: + return load_atom_8_by_2(pv); + case MO_32: + return load_atom_8_by_4(pv); + case -MO_32: + if (HAVE_al8) { + return load_atom_extract_al8x2(pv); + } + cpu_loop_exit_atomic(env_cpu(env), ra); + default: + g_assert_not_reached(); + } +} -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/cputlb.c | 108 ++++---- accel/tcg/user-exec.c | 12 +- accel/tcg/ldst_atomicity.c.inc | 491 +++++++++++++++++++++++++++++++++ 3 files changed, 545 insertions(+), 66 deletions(-) diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -XXX,XX +XXX,XX @@ Int128 cpu_ld16_le_mmu(CPUArchState *env, abi_ptr addr, * Store Helpers */ -static inline void QEMU_ALWAYS_INLINE -store_memop(void *haddr, uint64_t val, MemOp op) -{ - switch (op) { - case MO_UB: - stb_p(haddr, val); - break; - case MO_BEUW: - stw_be_p(haddr, val); - break; - case MO_LEUW: - stw_le_p(haddr, val); - break; - case MO_BEUL: - stl_be_p(haddr, val); - break; - case MO_LEUL: - stl_le_p(haddr, val); - break; - case MO_BEUQ: - stq_be_p(haddr, val); - break; - case MO_LEUQ: - stq_le_p(haddr, val); - break; - default: - qemu_build_not_reached(); - } -} - /** * do_st_mmio_leN: * @env: cpu context @@ -XXX,XX +XXX,XX @@ static uint64_t do_st_mmio_leN(CPUArchState *env, MMULookupPageData *p, return val_le; } -/** - * do_st_bytes_leN: - * @p: translation parameters - * @val_le: data to store - * - * Store @p->size bytes at @p->haddr, which is RAM. - * The bytes to store are extracted in little-endian order from @val_le; - * return the bytes of @val_le beyond @p->size that have not been stored. - */ -static uint64_t do_st_bytes_leN(MMULookupPageData *p, uint64_t val_le) -{ - uint8_t *haddr = p->haddr; - int i, size = p->size; - - for (i = 0; i < size; i++, val_le >>= 8) { - haddr[i] = val_le; - } - return val_le; -} - /* * Wrapper for the above. */ static uint64_t do_st_leN(CPUArchState *env, MMULookupPageData *p, - uint64_t val_le, int mmu_idx, uintptr_t ra) + uint64_t val_le, int mmu_idx, + MemOp mop, uintptr_t ra) { + MemOp atom; + unsigned tmp, half_size; + if (unlikely(p->flags & TLB_MMIO)) { return do_st_mmio_leN(env, p, val_le, mmu_idx, ra); } else if (unlikely(p->flags & TLB_DISCARD_WRITE)) { return val_le >> (p->size * 8); - } else { - return do_st_bytes_leN(p, val_le); + } + + /* + * It is a given that we cross a page and therefore there is no atomicity + * for the store as a whole, but subobjects may need attention. + */ + atom = mop & MO_ATOM_MASK; + switch (atom) { + case MO_ATOM_SUBALIGN: + return store_parts_leN(p->haddr, p->size, val_le); + + case MO_ATOM_IFALIGN_PAIR: + case MO_ATOM_WITHIN16_PAIR: + tmp = mop & MO_SIZE; + tmp = tmp ? tmp - 1 : 0; + half_size = 1 << tmp; + if (atom == MO_ATOM_IFALIGN_PAIR + ? p->size == half_size + : p->size >= half_size) { + if (!HAVE_al8_fast && p->size <= 4) { + return store_whole_le4(p->haddr, p->size, val_le); + } else if (HAVE_al8) { + return store_whole_le8(p->haddr, p->size, val_le); + } else { + cpu_loop_exit_atomic(env_cpu(env), ra); + } + } + /* fall through */ + + case MO_ATOM_IFALIGN: + case MO_ATOM_WITHIN16: + case MO_ATOM_NONE: + return store_bytes_leN(p->haddr, p->size, val_le); + + default: + g_assert_not_reached(); } } @@ -XXX,XX +XXX,XX @@ static void do_st_2(CPUArchState *env, MMULookupPageData *p, uint16_t val, if (memop & MO_BSWAP) { val = bswap16(val); } - store_memop(p->haddr, val, MO_UW); + store_atom_2(env, ra, p->haddr, memop, val); } } @@ -XXX,XX +XXX,XX @@ static void do_st_4(CPUArchState *env, MMULookupPageData *p, uint32_t val, if (memop & MO_BSWAP) { val = bswap32(val); } - store_memop(p->haddr, val, MO_UL); + store_atom_4(env, ra, p->haddr, memop, val); } } @@ -XXX,XX +XXX,XX @@ static void do_st_8(CPUArchState *env, MMULookupPageData *p, uint64_t val, if (memop & MO_BSWAP) { val = bswap64(val); } - store_memop(p->haddr, val, MO_UQ); + store_atom_8(env, ra, p->haddr, memop, val); } } @@ -XXX,XX +XXX,XX @@ static void do_st4_mmu(CPUArchState *env, target_ulong addr, uint32_t val, if ((l.memop & MO_BSWAP) != MO_LE) { val = bswap32(val); } - val = do_st_leN(env, &l.page[0], val, l.mmu_idx, ra); - (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, ra); + val = do_st_leN(env, &l.page[0], val, l.mmu_idx, l.memop, ra); + (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, l.memop, ra); } void helper_le_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, @@ -XXX,XX +XXX,XX @@ static void do_st8_mmu(CPUArchState *env, target_ulong addr, uint64_t val, if ((l.memop & MO_BSWAP) != MO_LE) { val = bswap64(val); } - val = do_st_leN(env, &l.page[0], val, l.mmu_idx, ra); - (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, ra); + val = do_st_leN(env, &l.page[0], val, l.mmu_idx, l.memop, ra); + (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, l.memop, ra); } void helper_le_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/user-exec.c +++ b/accel/tcg/user-exec.c @@ -XXX,XX +XXX,XX @@ void cpu_stw_be_mmu(CPUArchState *env, abi_ptr addr, uint16_t val, validate_memop(oi, MO_BEUW); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - stw_be_p(haddr, val); + store_atom_2(env, ra, haddr, get_memop(oi), be16_to_cpu(val)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } @@ -XXX,XX +XXX,XX @@ void cpu_stl_be_mmu(CPUArchState *env, abi_ptr addr, uint32_t val, validate_memop(oi, MO_BEUL); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - stl_be_p(haddr, val); + store_atom_4(env, ra, haddr, get_memop(oi), be32_to_cpu(val)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } @@ -XXX,XX +XXX,XX @@ void cpu_stq_be_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, validate_memop(oi, MO_BEUQ); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - stq_be_p(haddr, val); + store_atom_8(env, ra, haddr, get_memop(oi), be64_to_cpu(val)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } @@ -XXX,XX +XXX,XX @@ void cpu_stw_le_mmu(CPUArchState *env, abi_ptr addr, uint16_t val, validate_memop(oi, MO_LEUW); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - stw_le_p(haddr, val); + store_atom_2(env, ra, haddr, get_memop(oi), le16_to_cpu(val)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } @@ -XXX,XX +XXX,XX @@ void cpu_stl_le_mmu(CPUArchState *env, abi_ptr addr, uint32_t val, validate_memop(oi, MO_LEUL); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - stl_le_p(haddr, val); + store_atom_4(env, ra, haddr, get_memop(oi), le32_to_cpu(val)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } @@ -XXX,XX +XXX,XX @@ void cpu_stq_le_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, validate_memop(oi, MO_LEUQ); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - stq_le_p(haddr, val); + store_atom_8(env, ra, haddr, get_memop(oi), le64_to_cpu(val)); clear_helper_retaddr(); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } diff --git a/accel/tcg/ldst_atomicity.c.inc b/accel/tcg/ldst_atomicity.c.inc index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/ldst_atomicity.c.inc +++ b/accel/tcg/ldst_atomicity.c.inc @@ -XXX,XX +XXX,XX @@ #else # define HAVE_al16_fast false #endif +#if defined(CONFIG_ATOMIC128) || defined(CONFIG_CMPXCHG128) +# define HAVE_al16 true +#else +# define HAVE_al16 false +#endif + /** * required_atomicity: @@ -XXX,XX +XXX,XX @@ static uint64_t load_atom_8(CPUArchState *env, uintptr_t ra, g_assert_not_reached(); } } + +/** + * store_atomic2: + * @pv: host address + * @val: value to store + * + * Atomically store 2 aligned bytes to @pv. + */ +static inline void store_atomic2(void *pv, uint16_t val) +{ + uint16_t *p = __builtin_assume_aligned(pv, 2); + qatomic_set(p, val); +} + +/** + * store_atomic4: + * @pv: host address + * @val: value to store + * + * Atomically store 4 aligned bytes to @pv. + */ +static inline void store_atomic4(void *pv, uint32_t val) +{ + uint32_t *p = __builtin_assume_aligned(pv, 4); + qatomic_set(p, val); +} + +/** + * store_atomic8: + * @pv: host address + * @val: value to store + * + * Atomically store 8 aligned bytes to @pv. + */ +static inline void store_atomic8(void *pv, uint64_t val) +{ + uint64_t *p = __builtin_assume_aligned(pv, 8); + + qemu_build_assert(HAVE_al8); + qatomic_set__nocheck(p, val); +} + +/** + * store_atom_4x2 + */ +static inline void store_atom_4_by_2(void *pv, uint32_t val) +{ + store_atomic2(pv, val >> (HOST_BIG_ENDIAN ? 16 : 0)); + store_atomic2(pv + 2, val >> (HOST_BIG_ENDIAN ? 0 : 16)); +} + +/** + * store_atom_8_by_2 + */ +static inline void store_atom_8_by_2(void *pv, uint64_t val) +{ + store_atom_4_by_2(pv, val >> (HOST_BIG_ENDIAN ? 32 : 0)); + store_atom_4_by_2(pv + 4, val >> (HOST_BIG_ENDIAN ? 0 : 32)); +} + +/** + * store_atom_8_by_4 + */ +static inline void store_atom_8_by_4(void *pv, uint64_t val) +{ + store_atomic4(pv, val >> (HOST_BIG_ENDIAN ? 32 : 0)); + store_atomic4(pv + 4, val >> (HOST_BIG_ENDIAN ? 0 : 32)); +} + +/** + * store_atom_insert_al4: + * @p: host address + * @val: shifted value to store + * @msk: mask for value to store + * + * Atomically store @val to @p, masked by @msk. + */ +static void store_atom_insert_al4(uint32_t *p, uint32_t val, uint32_t msk) +{ + uint32_t old, new; + + p = __builtin_assume_aligned(p, 4); + old = qatomic_read(p); + do { + new = (old & ~msk) | val; + } while (!__atomic_compare_exchange_n(p, &old, new, true, + __ATOMIC_RELAXED, __ATOMIC_RELAXED)); +} + +/** + * store_atom_insert_al8: + * @p: host address + * @val: shifted value to store + * @msk: mask for value to store + * + * Atomically store @val to @p masked by @msk. + */ +static void store_atom_insert_al8(uint64_t *p, uint64_t val, uint64_t msk) +{ + uint64_t old, new; + + qemu_build_assert(HAVE_al8); + p = __builtin_assume_aligned(p, 8); + old = qatomic_read__nocheck(p); + do { + new = (old & ~msk) | val; + } while (!__atomic_compare_exchange_n(p, &old, new, true, + __ATOMIC_RELAXED, __ATOMIC_RELAXED)); +} + +/** + * store_atom_insert_al16: + * @p: host address + * @val: shifted value to store + * @msk: mask for value to store + * + * Atomically store @val to @p masked by @msk. + */ +static void store_atom_insert_al16(Int128 *ps, Int128Alias val, Int128Alias msk) +{ +#if defined(CONFIG_ATOMIC128) + __uint128_t *pu, old, new; + + /* With CONFIG_ATOMIC128, we can avoid the memory barriers. */ + pu = __builtin_assume_aligned(ps, 16); + old = *pu; + do { + new = (old & ~msk.u) | val.u; + } while (!__atomic_compare_exchange_n(pu, &old, new, true, + __ATOMIC_RELAXED, __ATOMIC_RELAXED)); +#elif defined(CONFIG_CMPXCHG128) + __uint128_t *pu, old, new; + + /* + * Without CONFIG_ATOMIC128, __atomic_compare_exchange_n will always + * defer to libatomic, so we must use __sync_*_compare_and_swap_16 + * and accept the sequential consistency that comes with it. + */ + pu = __builtin_assume_aligned(ps, 16); + do { + old = *pu; + new = (old & ~msk.u) | val.u; + } while (!__sync_bool_compare_and_swap_16(pu, old, new)); +#else + qemu_build_not_reached(); +#endif +} + +/** + * store_bytes_leN: + * @pv: host address + * @size: number of bytes to store + * @val_le: data to store + * + * Store @size bytes at @p. The bytes to store are extracted in little-endian order + * from @val_le; return the bytes of @val_le beyond @size that have not been stored. + */ +static uint64_t store_bytes_leN(void *pv, int size, uint64_t val_le) +{ + uint8_t *p = pv; + for (int i = 0; i < size; i++, val_le >>= 8) { + p[i] = val_le; + } + return val_le; +} + +/** + * store_parts_leN + * @pv: host address + * @size: number of bytes to store + * @val_le: data to store + * + * As store_bytes_leN, but atomically on each aligned part. + */ +G_GNUC_UNUSED +static uint64_t store_parts_leN(void *pv, int size, uint64_t val_le) +{ + do { + int n; + + /* Find minimum of alignment and size */ + switch (((uintptr_t)pv | size) & 7) { + case 4: + store_atomic4(pv, le32_to_cpu(val_le)); + val_le >>= 32; + n = 4; + break; + case 2: + case 6: + store_atomic2(pv, le16_to_cpu(val_le)); + val_le >>= 16; + n = 2; + break; + default: + *(uint8_t *)pv = val_le; + val_le >>= 8; + n = 1; + break; + case 0: + g_assert_not_reached(); + } + pv += n; + size -= n; + } while (size != 0); + + return val_le; +} + +/** + * store_whole_le4 + * @pv: host address + * @size: number of bytes to store + * @val_le: data to store + * + * As store_bytes_leN, but atomically as a whole. + * Four aligned bytes are guaranteed to cover the store. + */ +static uint64_t store_whole_le4(void *pv, int size, uint64_t val_le) +{ + int sz = size * 8; + int o = (uintptr_t)pv & 3; + int sh = o * 8; + uint32_t m = MAKE_64BIT_MASK(0, sz); + uint32_t v; + + if (HOST_BIG_ENDIAN) { + v = bswap32(val_le) >> sh; + m = bswap32(m) >> sh; + } else { + v = val_le << sh; + m <<= sh; + } + store_atom_insert_al4(pv - o, v, m); + return val_le >> sz; +} + +/** + * store_whole_le8 + * @pv: host address + * @size: number of bytes to store + * @val_le: data to store + * + * As store_bytes_leN, but atomically as a whole. + * Eight aligned bytes are guaranteed to cover the store. + */ +static uint64_t store_whole_le8(void *pv, int size, uint64_t val_le) +{ + int sz = size * 8; + int o = (uintptr_t)pv & 7; + int sh = o * 8; + uint64_t m = MAKE_64BIT_MASK(0, sz); + uint64_t v; + + qemu_build_assert(HAVE_al8); + if (HOST_BIG_ENDIAN) { + v = bswap64(val_le) >> sh; + m = bswap64(m) >> sh; + } else { + v = val_le << sh; + m <<= sh; + } + store_atom_insert_al8(pv - o, v, m); + return val_le >> sz; +} + +/** + * store_whole_le16 + * @pv: host address + * @size: number of bytes to store + * @val_le: data to store + * + * As store_bytes_leN, but atomically as a whole. + * 16 aligned bytes are guaranteed to cover the store. + */ +static uint64_t store_whole_le16(void *pv, int size, Int128 val_le) +{ + int sz = size * 8; + int o = (uintptr_t)pv & 15; + int sh = o * 8; + Int128 m, v; + + qemu_build_assert(HAVE_al16); + + /* Like MAKE_64BIT_MASK(0, sz), but larger. */ + if (sz <= 64) { + m = int128_make64(MAKE_64BIT_MASK(0, sz)); + } else { + m = int128_make128(-1, MAKE_64BIT_MASK(0, sz - 64)); + } + + if (HOST_BIG_ENDIAN) { + v = int128_urshift(bswap128(val_le), sh); + m = int128_urshift(bswap128(m), sh); + } else { + v = int128_lshift(val_le, sh); + m = int128_lshift(m, sh); + } + store_atom_insert_al16(pv - o, v, m); + + /* Unused if sz <= 64. */ + return int128_gethi(val_le) >> (sz - 64); +} + +/** + * store_atom_2: + * @p: host address + * @val: the value to store + * @memop: the full memory op + * + * Store 2 bytes to @p, honoring the atomicity of @memop. + */ +static void store_atom_2(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop, uint16_t val) +{ + uintptr_t pi = (uintptr_t)pv; + int atmax; + + if (likely((pi & 1) == 0)) { + store_atomic2(pv, val); + return; + } + + atmax = required_atomicity(env, pi, memop); + if (atmax == MO_8) { + stw_he_p(pv, val); + return; + } + + /* + * The only case remaining is MO_ATOM_WITHIN16. + * Big or little endian, we want the middle two bytes in each test. + */ + if ((pi & 3) == 1) { + store_atom_insert_al4(pv - 1, (uint32_t)val << 8, MAKE_64BIT_MASK(8, 16)); + return; + } else if ((pi & 7) == 3) { + if (HAVE_al8) { + store_atom_insert_al8(pv - 3, (uint64_t)val << 24, MAKE_64BIT_MASK(24, 16)); + return; + } + } else if ((pi & 15) == 7) { + if (HAVE_al16) { + Int128 v = int128_lshift(int128_make64(val), 56); + Int128 m = int128_lshift(int128_make64(0xffff), 56); + store_atom_insert_al16(pv - 7, v, m); + return; + } + } else { + g_assert_not_reached(); + } + + cpu_loop_exit_atomic(env_cpu(env), ra); +} + +/** + * store_atom_4: + * @p: host address + * @val: the value to store + * @memop: the full memory op + * + * Store 4 bytes to @p, honoring the atomicity of @memop. + */ +static void store_atom_4(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop, uint32_t val) +{ + uintptr_t pi = (uintptr_t)pv; + int atmax; + + if (likely((pi & 3) == 0)) { + store_atomic4(pv, val); + return; + } + + atmax = required_atomicity(env, pi, memop); + switch (atmax) { + case MO_8: + stl_he_p(pv, val); + return; + case MO_16: + store_atom_4_by_2(pv, val); + return; + case -MO_16: + { + uint32_t val_le = cpu_to_le32(val); + int s2 = pi & 3; + int s1 = 4 - s2; + + switch (s2) { + case 1: + val_le = store_whole_le4(pv, s1, val_le); + *(uint8_t *)(pv + 3) = val_le; + break; + case 3: + *(uint8_t *)pv = val_le; + store_whole_le4(pv + 1, s2, val_le >> 8); + break; + case 0: /* aligned */ + case 2: /* atmax MO_16 */ + default: + g_assert_not_reached(); + } + } + return; + case MO_32: + if ((pi & 7) < 4) { + if (HAVE_al8) { + store_whole_le8(pv, 4, cpu_to_le32(val)); + return; + } + } else { + if (HAVE_al16) { + store_whole_le16(pv, 4, int128_make64(cpu_to_le32(val))); + return; + } + } + cpu_loop_exit_atomic(env_cpu(env), ra); + default: + g_assert_not_reached(); + } +} + +/** + * store_atom_8: + * @p: host address + * @val: the value to store + * @memop: the full memory op + * + * Store 8 bytes to @p, honoring the atomicity of @memop. + */ +static void store_atom_8(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop, uint64_t val) +{ + uintptr_t pi = (uintptr_t)pv; + int atmax; + + if (HAVE_al8 && likely((pi & 7) == 0)) { + store_atomic8(pv, val); + return; + } + + atmax = required_atomicity(env, pi, memop); + switch (atmax) { + case MO_8: + stq_he_p(pv, val); + return; + case MO_16: + store_atom_8_by_2(pv, val); + return; + case MO_32: + store_atom_8_by_4(pv, val); + return; + case -MO_32: + if (HAVE_al8) { + uint64_t val_le = cpu_to_le64(val); + int s2 = pi & 7; + int s1 = 8 - s2; + + switch (s2) { + case 1 ... 3: + val_le = store_whole_le8(pv, s1, val_le); + store_bytes_leN(pv + s1, s2, val_le); + break; + case 5 ... 7: + val_le = store_bytes_leN(pv, s1, val_le); + store_whole_le8(pv + s1, s2, val_le); + break; + case 0: /* aligned */ + case 4: /* atmax MO_32 */ + default: + g_assert_not_reached(); + } + return; + } + break; + case MO_64: + if (HAVE_al16) { + store_whole_le16(pv, 8, int128_make64(cpu_to_le64(val))); + return; + } + break; + default: + g_assert_not_reached(); + } + cpu_loop_exit_atomic(env_cpu(env), ra); +} -- 2.34.1 With the current structure of cputlb.c, there is no difference between the little-endian and big-endian entry points, aside from the assert. Unify the pairs of functions. Hoist the qemu_{ld,st}_helpers arrays to tcg.c. Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- docs/devel/loads-stores.rst | 36 ++---- include/tcg/tcg-ldst.h | 60 ++++------ accel/tcg/cputlb.c | 190 ++++++++++--------------------- tcg/tcg.c | 21 ++++ tcg/tci.c | 61 ++++------ tcg/aarch64/tcg-target.c.inc | 33 ------ tcg/arm/tcg-target.c.inc | 37 ------ tcg/i386/tcg-target.c.inc | 30 +---- tcg/loongarch64/tcg-target.c.inc | 23 ---- tcg/mips/tcg-target.c.inc | 31 ----- tcg/ppc/tcg-target.c.inc | 30 +---- tcg/riscv/tcg-target.c.inc | 42 ------- tcg/s390x/tcg-target.c.inc | 31 +---- tcg/sparc64/tcg-target.c.inc | 32 +----- 14 files changed, 146 insertions(+), 511 deletions(-) diff --git a/docs/devel/loads-stores.rst b/docs/devel/loads-stores.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/devel/loads-stores.rst +++ b/docs/devel/loads-stores.rst @@ -XXX,XX +XXX,XX @@ swap: ``translator_ld{sign}{size}_swap(env, ptr, swap)`` Regexes for git grep - ``\<translator_ld[us]\?[bwlq]\(_swap\)\?\>`` -``helper_*_{ld,st}*_mmu`` +``helper_{ld,st}*_mmu`` ~~~~~~~~~~~~~~~~~~~~~~~~~ These functions are intended primarily to be called by the code -generated by the TCG backend. They may also be called by target -CPU helper function code. Like the ``cpu_{ld,st}_mmuidx_ra`` functions -they perform accesses by guest virtual address, with a given ``mmuidx``. +generated by the TCG backend. Like the ``cpu_{ld,st}_mmu`` functions +they perform accesses by guest virtual address, with a given ``MemOpIdx``. -These functions specify an ``opindex`` parameter which encodes -(among other things) the mmu index to use for the access. This parameter -should be created by calling ``make_memop_idx()``. +They differ from ``cpu_{ld,st}_mmu`` in that they take the endianness +of the operation only from the MemOpIdx, and loads extend the return +value to the size of a host general register (``tcg_target_ulong``). -The ``retaddr`` parameter should be the result of GETPC() called directly -from the top level HELPER(foo) function (or 0 if no guest CPU state -unwinding is required). +load: ``helper_ld{sign}{size}_mmu(env, addr, opindex, retaddr)`` -**TODO** The names of these functions are a bit odd for historical -reasons because they were originally expected to be called only from -within generated code. We should rename them to bring them more in -line with the other memory access functions. The explicit endianness -is the only feature they have beyond ``*_mmuidx_ra``. - -load: ``helper_{endian}_ld{sign}{size}_mmu(env, addr, opindex, retaddr)`` - -store: ``helper_{endian}_st{size}_mmu(env, addr, val, opindex, retaddr)`` +store: ``helper_{size}_mmu(env, addr, val, opindex, retaddr)`` ``sign`` - (empty) : for 32 or 64 bit sizes @@ -XXX,XX +XXX,XX @@ store: ``helper_{endian}_st{size}_mmu(env, addr, val, opindex, retaddr)`` - ``l`` : 32 bits - ``q`` : 64 bits -``endian`` - - ``le`` : little endian - - ``be`` : big endian - - ``ret`` : target endianness - Regexes for git grep - - ``\<helper_\(le\|be\|ret\)_ld[us]\?[bwlq]_mmu\>`` - - ``\<helper_\(le\|be\|ret\)_st[bwlq]_mmu\>`` + - ``\<helper_ld[us]\?[bwlq]_mmu\>`` + - ``\<helper_st[bwlq]_mmu\>`` ``address_space_*`` ~~~~~~~~~~~~~~~~~~~ diff --git a/include/tcg/tcg-ldst.h b/include/tcg/tcg-ldst.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-ldst.h +++ b/include/tcg/tcg-ldst.h @@ -XXX,XX +XXX,XX @@ #ifdef CONFIG_SOFTMMU /* Value zero-extended to tcg register size. */ -tcg_target_ulong helper_ret_ldub_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_le_lduw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_le_ldul_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -uint64_t helper_le_ldq_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_be_lduw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_be_ldul_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -uint64_t helper_be_ldq_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); +tcg_target_ulong helper_ldub_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); +tcg_target_ulong helper_lduw_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); +tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); +uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); /* Value sign-extended to tcg register size. */ -tcg_target_ulong helper_ret_ldsb_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_le_ldsw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_le_ldsl_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_be_ldsw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_be_ldsl_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr); +tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); +tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); +tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); /* * Value extended to at least uint32_t, so that some ABIs do not require * zero-extension from uint8_t or uint16_t. */ -void helper_ret_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr); -void helper_le_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr); -void helper_le_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr); -void helper_le_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, - MemOpIdx oi, uintptr_t retaddr); -void helper_be_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr); -void helper_be_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr); -void helper_be_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, - MemOpIdx oi, uintptr_t retaddr); +void helper_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t retaddr); +void helper_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t retaddr); +void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t retaddr); +void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, + MemOpIdx oi, uintptr_t retaddr); #else diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -XXX,XX +XXX,XX @@ static void *atomic_mmu_lookup(CPUArchState *env, target_ulong addr, cpu_loop_exit_atomic(env_cpu(env), retaddr); } -/* - * Verify that we have passed the correct MemOp to the correct function. - * - * In the case of the helper_*_mmu functions, we will have done this by - * using the MemOp to look up the helper during code generation. - * - * In the case of the cpu_*_mmu functions, this is up to the caller. - * We could present one function to target code, and dispatch based on - * the MemOp, but so far we have worked hard to avoid an indirect function - * call along the memory path. - */ -static void validate_memop(MemOpIdx oi, MemOp expected) -{ -#ifdef CONFIG_DEBUG_TCG - MemOp have = get_memop(oi) & (MO_SIZE | MO_BSWAP); - assert(have == expected); -#endif -} - /* * Load Helpers * @@ -XXX,XX +XXX,XX @@ static uint8_t do_ld1_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return do_ld_1(env, &l.page[0], l.mmu_idx, access_type, ra); } -tcg_target_ulong helper_ret_ldub_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) +tcg_target_ulong helper_ldub_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr) { - validate_memop(oi, MO_UB); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_8); return do_ld1_mmu(env, addr, oi, retaddr, MMU_DATA_LOAD); } @@ -XXX,XX +XXX,XX @@ static uint16_t do_ld2_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return ret; } -tcg_target_ulong helper_le_lduw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) +tcg_target_ulong helper_lduw_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr) { - validate_memop(oi, MO_LEUW); - return do_ld2_mmu(env, addr, oi, retaddr, MMU_DATA_LOAD); -} - -tcg_target_ulong helper_be_lduw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) -{ - validate_memop(oi, MO_BEUW); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_16); return do_ld2_mmu(env, addr, oi, retaddr, MMU_DATA_LOAD); } @@ -XXX,XX +XXX,XX @@ static uint32_t do_ld4_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return ret; } -tcg_target_ulong helper_le_ldul_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) +tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr) { - validate_memop(oi, MO_LEUL); - return do_ld4_mmu(env, addr, oi, retaddr, MMU_DATA_LOAD); -} - -tcg_target_ulong helper_be_ldul_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) -{ - validate_memop(oi, MO_BEUL); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_32); return do_ld4_mmu(env, addr, oi, retaddr, MMU_DATA_LOAD); } @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld8_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return ret; } -uint64_t helper_le_ldq_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) +uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr) { - validate_memop(oi, MO_LEUQ); - return do_ld8_mmu(env, addr, oi, retaddr, MMU_DATA_LOAD); -} - -uint64_t helper_be_ldq_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) -{ - validate_memop(oi, MO_BEUQ); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_64); return do_ld8_mmu(env, addr, oi, retaddr, MMU_DATA_LOAD); } @@ -XXX,XX +XXX,XX @@ uint64_t helper_be_ldq_mmu(CPUArchState *env, target_ulong addr, * avoid this for 64-bit data, or for 32-bit data on 32-bit host. */ - -tcg_target_ulong helper_ret_ldsb_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) +tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr) { - return (int8_t)helper_ret_ldub_mmu(env, addr, oi, retaddr); + return (int8_t)helper_ldub_mmu(env, addr, oi, retaddr); } -tcg_target_ulong helper_le_ldsw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) +tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr) { - return (int16_t)helper_le_lduw_mmu(env, addr, oi, retaddr); + return (int16_t)helper_lduw_mmu(env, addr, oi, retaddr); } -tcg_target_ulong helper_be_ldsw_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) +tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr) { - return (int16_t)helper_be_lduw_mmu(env, addr, oi, retaddr); -} - -tcg_target_ulong helper_le_ldsl_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) -{ - return (int32_t)helper_le_ldul_mmu(env, addr, oi, retaddr); -} - -tcg_target_ulong helper_be_ldsl_mmu(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t retaddr) -{ - return (int32_t)helper_be_ldul_mmu(env, addr, oi, retaddr); + return (int32_t)helper_ldul_mmu(env, addr, oi, retaddr); } /* @@ -XXX,XX +XXX,XX @@ uint8_t cpu_ldb_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { uint8_t ret; - validate_memop(oi, MO_UB); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_UB); ret = do_ld1_mmu(env, addr, oi, ra, MMU_DATA_LOAD); plugin_load_cb(env, addr, oi); return ret; @@ -XXX,XX +XXX,XX @@ uint16_t cpu_ldw_be_mmu(CPUArchState *env, abi_ptr addr, { uint16_t ret; - validate_memop(oi, MO_BEUW); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_BEUW); ret = do_ld2_mmu(env, addr, oi, ra, MMU_DATA_LOAD); plugin_load_cb(env, addr, oi); return ret; @@ -XXX,XX +XXX,XX @@ uint32_t cpu_ldl_be_mmu(CPUArchState *env, abi_ptr addr, { uint32_t ret; - validate_memop(oi, MO_BEUL); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_BEUL); ret = do_ld4_mmu(env, addr, oi, ra, MMU_DATA_LOAD); plugin_load_cb(env, addr, oi); return ret; @@ -XXX,XX +XXX,XX @@ uint64_t cpu_ldq_be_mmu(CPUArchState *env, abi_ptr addr, { uint64_t ret; - validate_memop(oi, MO_BEUQ); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_BEUQ); ret = do_ld8_mmu(env, addr, oi, ra, MMU_DATA_LOAD); plugin_load_cb(env, addr, oi); return ret; @@ -XXX,XX +XXX,XX @@ uint16_t cpu_ldw_le_mmu(CPUArchState *env, abi_ptr addr, { uint16_t ret; - validate_memop(oi, MO_LEUW); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_LEUW); ret = do_ld2_mmu(env, addr, oi, ra, MMU_DATA_LOAD); plugin_load_cb(env, addr, oi); return ret; @@ -XXX,XX +XXX,XX @@ uint32_t cpu_ldl_le_mmu(CPUArchState *env, abi_ptr addr, { uint32_t ret; - validate_memop(oi, MO_LEUL); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_LEUL); ret = do_ld4_mmu(env, addr, oi, ra, MMU_DATA_LOAD); plugin_load_cb(env, addr, oi); return ret; @@ -XXX,XX +XXX,XX @@ uint64_t cpu_ldq_le_mmu(CPUArchState *env, abi_ptr addr, { uint64_t ret; - validate_memop(oi, MO_LEUQ); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_LEUQ); ret = do_ld8_mmu(env, addr, oi, ra, MMU_DATA_LOAD); plugin_load_cb(env, addr, oi); return ret; @@ -XXX,XX +XXX,XX @@ Int128 cpu_ld16_be_mmu(CPUArchState *env, abi_ptr addr, mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; new_oi = make_memop_idx(mop, mmu_idx); - h = helper_be_ldq_mmu(env, addr, new_oi, ra); - l = helper_be_ldq_mmu(env, addr + 8, new_oi, ra); + h = helper_ldq_mmu(env, addr, new_oi, ra); + l = helper_ldq_mmu(env, addr + 8, new_oi, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); return int128_make128(l, h); @@ -XXX,XX +XXX,XX @@ Int128 cpu_ld16_le_mmu(CPUArchState *env, abi_ptr addr, mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; new_oi = make_memop_idx(mop, mmu_idx); - l = helper_le_ldq_mmu(env, addr, new_oi, ra); - h = helper_le_ldq_mmu(env, addr + 8, new_oi, ra); + l = helper_ldq_mmu(env, addr, new_oi, ra); + h = helper_ldq_mmu(env, addr + 8, new_oi, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); return int128_make128(l, h); @@ -XXX,XX +XXX,XX @@ static void do_st_8(CPUArchState *env, MMULookupPageData *p, uint64_t val, } } -void helper_ret_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t ra) +void helper_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t ra) { MMULookupLocals l; bool crosspage; - validate_memop(oi, MO_UB); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_8); crosspage = mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE, &l); tcg_debug_assert(!crosspage); @@ -XXX,XX +XXX,XX @@ static void do_st2_mmu(CPUArchState *env, target_ulong addr, uint16_t val, do_st_1(env, &l.page[1], b, l.mmu_idx, ra); } -void helper_le_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr) +void helper_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t retaddr) { - validate_memop(oi, MO_LEUW); - do_st2_mmu(env, addr, val, oi, retaddr); -} - -void helper_be_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr) -{ - validate_memop(oi, MO_BEUW); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_16); do_st2_mmu(env, addr, val, oi, retaddr); } @@ -XXX,XX +XXX,XX @@ static void do_st4_mmu(CPUArchState *env, target_ulong addr, uint32_t val, (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, l.memop, ra); } -void helper_le_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr) +void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t retaddr) { - validate_memop(oi, MO_LEUL); - do_st4_mmu(env, addr, val, oi, retaddr); -} - -void helper_be_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, - MemOpIdx oi, uintptr_t retaddr) -{ - validate_memop(oi, MO_BEUL); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_32); do_st4_mmu(env, addr, val, oi, retaddr); } @@ -XXX,XX +XXX,XX @@ static void do_st8_mmu(CPUArchState *env, target_ulong addr, uint64_t val, (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, l.memop, ra); } -void helper_le_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, - MemOpIdx oi, uintptr_t retaddr) +void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, + MemOpIdx oi, uintptr_t retaddr) { - validate_memop(oi, MO_LEUQ); - do_st8_mmu(env, addr, val, oi, retaddr); -} - -void helper_be_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, - MemOpIdx oi, uintptr_t retaddr) -{ - validate_memop(oi, MO_BEUQ); + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_64); do_st8_mmu(env, addr, val, oi, retaddr); } @@ -XXX,XX +XXX,XX @@ static void plugin_store_cb(CPUArchState *env, abi_ptr addr, MemOpIdx oi) void cpu_stb_mmu(CPUArchState *env, target_ulong addr, uint8_t val, MemOpIdx oi, uintptr_t retaddr) { - helper_ret_stb_mmu(env, addr, val, oi, retaddr); + helper_stb_mmu(env, addr, val, oi, retaddr); plugin_store_cb(env, addr, oi); } void cpu_stw_be_mmu(CPUArchState *env, target_ulong addr, uint16_t val, MemOpIdx oi, uintptr_t retaddr) { - helper_be_stw_mmu(env, addr, val, oi, retaddr); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_BEUW); + do_st2_mmu(env, addr, val, oi, retaddr); plugin_store_cb(env, addr, oi); } void cpu_stl_be_mmu(CPUArchState *env, target_ulong addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr) { - helper_be_stl_mmu(env, addr, val, oi, retaddr); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_BEUL); + do_st4_mmu(env, addr, val, oi, retaddr); plugin_store_cb(env, addr, oi); } void cpu_stq_be_mmu(CPUArchState *env, target_ulong addr, uint64_t val, MemOpIdx oi, uintptr_t retaddr) { - helper_be_stq_mmu(env, addr, val, oi, retaddr); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_BEUQ); + do_st8_mmu(env, addr, val, oi, retaddr); plugin_store_cb(env, addr, oi); } void cpu_stw_le_mmu(CPUArchState *env, target_ulong addr, uint16_t val, MemOpIdx oi, uintptr_t retaddr) { - helper_le_stw_mmu(env, addr, val, oi, retaddr); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_LEUW); + do_st2_mmu(env, addr, val, oi, retaddr); plugin_store_cb(env, addr, oi); } void cpu_stl_le_mmu(CPUArchState *env, target_ulong addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr) { - helper_le_stl_mmu(env, addr, val, oi, retaddr); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_LEUL); + do_st4_mmu(env, addr, val, oi, retaddr); plugin_store_cb(env, addr, oi); } void cpu_stq_le_mmu(CPUArchState *env, target_ulong addr, uint64_t val, MemOpIdx oi, uintptr_t retaddr) { - helper_le_stq_mmu(env, addr, val, oi, retaddr); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == MO_LEUQ); + do_st8_mmu(env, addr, val, oi, retaddr); plugin_store_cb(env, addr, oi); } @@ -XXX,XX +XXX,XX @@ void cpu_st16_be_mmu(CPUArchState *env, abi_ptr addr, Int128 val, mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; new_oi = make_memop_idx(mop, mmu_idx); - helper_be_stq_mmu(env, addr, int128_gethi(val), new_oi, ra); - helper_be_stq_mmu(env, addr + 8, int128_getlo(val), new_oi, ra); + helper_stq_mmu(env, addr, int128_gethi(val), new_oi, ra); + helper_stq_mmu(env, addr + 8, int128_getlo(val), new_oi, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } @@ -XXX,XX +XXX,XX @@ void cpu_st16_le_mmu(CPUArchState *env, abi_ptr addr, Int128 val, mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; new_oi = make_memop_idx(mop, mmu_idx); - helper_le_stq_mmu(env, addr, int128_getlo(val), new_oi, ra); - helper_le_stq_mmu(env, addr + 8, int128_gethi(val), new_oi, ra); + helper_stq_mmu(env, addr, int128_getlo(val), new_oi, ra); + helper_stq_mmu(env, addr + 8, int128_gethi(val), new_oi, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *l, const TCGLdstHelperParam *p) __attribute__((unused)); +#ifdef CONFIG_SOFTMMU +static void * const qemu_ld_helpers[MO_SSIZE + 1] = { + [MO_UB] = helper_ldub_mmu, + [MO_SB] = helper_ldsb_mmu, + [MO_UW] = helper_lduw_mmu, + [MO_SW] = helper_ldsw_mmu, + [MO_UL] = helper_ldul_mmu, + [MO_UQ] = helper_ldq_mmu, +#if TCG_TARGET_REG_BITS == 64 + [MO_SL] = helper_ldsl_mmu, +#endif +}; + +static void * const qemu_st_helpers[MO_SIZE + 1] = { + [MO_8] = helper_stb_mmu, + [MO_16] = helper_stw_mmu, + [MO_32] = helper_stl_mmu, + [MO_64] = helper_stq_mmu, +}; +#endif + TCGContext tcg_init_ctx; __thread TCGContext *tcg_ctx; diff --git a/tcg/tci.c b/tcg/tci.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci.c +++ b/tcg/tci.c @@ -XXX,XX +XXX,XX @@ static uint64_t tci_qemu_ld(CPUArchState *env, target_ulong taddr, uintptr_t ra = (uintptr_t)tb_ptr; #ifdef CONFIG_SOFTMMU - switch (mop & (MO_BSWAP | MO_SSIZE)) { + switch (mop & MO_SSIZE) { case MO_UB: - return helper_ret_ldub_mmu(env, taddr, oi, ra); + return helper_ldub_mmu(env, taddr, oi, ra); case MO_SB: - return helper_ret_ldsb_mmu(env, taddr, oi, ra); - case MO_LEUW: - return helper_le_lduw_mmu(env, taddr, oi, ra); - case MO_LESW: - return helper_le_ldsw_mmu(env, taddr, oi, ra); - case MO_LEUL: - return helper_le_ldul_mmu(env, taddr, oi, ra); - case MO_LESL: - return helper_le_ldsl_mmu(env, taddr, oi, ra); - case MO_LEUQ: - return helper_le_ldq_mmu(env, taddr, oi, ra); - case MO_BEUW: - return helper_be_lduw_mmu(env, taddr, oi, ra); - case MO_BESW: - return helper_be_ldsw_mmu(env, taddr, oi, ra); - case MO_BEUL: - return helper_be_ldul_mmu(env, taddr, oi, ra); - case MO_BESL: - return helper_be_ldsl_mmu(env, taddr, oi, ra); - case MO_BEUQ: - return helper_be_ldq_mmu(env, taddr, oi, ra); + return helper_ldsb_mmu(env, taddr, oi, ra); + case MO_UW: + return helper_lduw_mmu(env, taddr, oi, ra); + case MO_SW: + return helper_ldsw_mmu(env, taddr, oi, ra); + case MO_UL: + return helper_ldul_mmu(env, taddr, oi, ra); + case MO_SL: + return helper_ldsl_mmu(env, taddr, oi, ra); + case MO_UQ: + return helper_ldq_mmu(env, taddr, oi, ra); default: g_assert_not_reached(); } @@ -XXX,XX +XXX,XX @@ static void tci_qemu_st(CPUArchState *env, target_ulong taddr, uint64_t val, uintptr_t ra = (uintptr_t)tb_ptr; #ifdef CONFIG_SOFTMMU - switch (mop & (MO_BSWAP | MO_SIZE)) { + switch (mop & MO_SIZE) { case MO_UB: - helper_ret_stb_mmu(env, taddr, val, oi, ra); + helper_stb_mmu(env, taddr, val, oi, ra); break; - case MO_LEUW: - helper_le_stw_mmu(env, taddr, val, oi, ra); + case MO_UW: + helper_stw_mmu(env, taddr, val, oi, ra); break; - case MO_LEUL: - helper_le_stl_mmu(env, taddr, val, oi, ra); + case MO_UL: + helper_stl_mmu(env, taddr, val, oi, ra); break; - case MO_LEUQ: - helper_le_stq_mmu(env, taddr, val, oi, ra); - break; - case MO_BEUW: - helper_be_stw_mmu(env, taddr, val, oi, ra); - break; - case MO_BEUL: - helper_be_stl_mmu(env, taddr, val, oi, ra); - break; - case MO_BEUQ: - helper_be_stq_mmu(env, taddr, val, oi, ra); + case MO_UQ: + helper_stq_mmu(env, taddr, val, oi, ra); break; default: g_assert_not_reached(); diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { } HostAddress; #ifdef CONFIG_SOFTMMU -/* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, - * MemOpIdx oi, uintptr_t ra) - */ -static void * const qemu_ld_helpers[MO_SIZE + 1] = { - [MO_8] = helper_ret_ldub_mmu, -#if HOST_BIG_ENDIAN - [MO_16] = helper_be_lduw_mmu, - [MO_32] = helper_be_ldul_mmu, - [MO_64] = helper_be_ldq_mmu, -#else - [MO_16] = helper_le_lduw_mmu, - [MO_32] = helper_le_ldul_mmu, - [MO_64] = helper_le_ldq_mmu, -#endif -}; - -/* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, - * uintxx_t val, MemOpIdx oi, - * uintptr_t ra) - */ -static void * const qemu_st_helpers[MO_SIZE + 1] = { - [MO_8] = helper_ret_stb_mmu, -#if HOST_BIG_ENDIAN - [MO_16] = helper_be_stw_mmu, - [MO_32] = helper_be_stl_mmu, - [MO_64] = helper_be_stq_mmu, -#else - [MO_16] = helper_le_stw_mmu, - [MO_32] = helper_le_stl_mmu, - [MO_64] = helper_le_stq_mmu, -#endif -}; - static const TCGLdstHelperParam ldst_helper_param = { .ntmp = 1, .tmp = { TCG_REG_TMP } }; diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { } HostAddress; #ifdef CONFIG_SOFTMMU -/* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, - * int mmu_idx, uintptr_t ra) - */ -static void * const qemu_ld_helpers[MO_SSIZE + 1] = { - [MO_UB] = helper_ret_ldub_mmu, - [MO_SB] = helper_ret_ldsb_mmu, -#if HOST_BIG_ENDIAN - [MO_UW] = helper_be_lduw_mmu, - [MO_UL] = helper_be_ldul_mmu, - [MO_UQ] = helper_be_ldq_mmu, - [MO_SW] = helper_be_ldsw_mmu, - [MO_SL] = helper_be_ldul_mmu, -#else - [MO_UW] = helper_le_lduw_mmu, - [MO_UL] = helper_le_ldul_mmu, - [MO_UQ] = helper_le_ldq_mmu, - [MO_SW] = helper_le_ldsw_mmu, - [MO_SL] = helper_le_ldul_mmu, -#endif -}; - -/* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, - * uintxx_t val, int mmu_idx, uintptr_t ra) - */ -static void * const qemu_st_helpers[MO_SIZE + 1] = { - [MO_8] = helper_ret_stb_mmu, -#if HOST_BIG_ENDIAN - [MO_16] = helper_be_stw_mmu, - [MO_32] = helper_be_stl_mmu, - [MO_64] = helper_be_stq_mmu, -#else - [MO_16] = helper_le_stw_mmu, - [MO_32] = helper_le_stl_mmu, - [MO_64] = helper_le_stq_mmu, -#endif -}; - static TCGReg ldst_ra_gen(TCGContext *s, const TCGLabelQemuLdst *l, int arg) { /* We arrive at the slow path via "BLNE", so R14 contains l->raddr. */ diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { } HostAddress; #if defined(CONFIG_SOFTMMU) -/* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, - * int mmu_idx, uintptr_t ra) - */ -static void * const qemu_ld_helpers[(MO_SIZE | MO_BSWAP) + 1] = { - [MO_UB] = helper_ret_ldub_mmu, - [MO_LEUW] = helper_le_lduw_mmu, - [MO_LEUL] = helper_le_ldul_mmu, - [MO_LEUQ] = helper_le_ldq_mmu, - [MO_BEUW] = helper_be_lduw_mmu, - [MO_BEUL] = helper_be_ldul_mmu, - [MO_BEUQ] = helper_be_ldq_mmu, -}; - -/* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, - * uintxx_t val, int mmu_idx, uintptr_t ra) - */ -static void * const qemu_st_helpers[(MO_SIZE | MO_BSWAP) + 1] = { - [MO_UB] = helper_ret_stb_mmu, - [MO_LEUW] = helper_le_stw_mmu, - [MO_LEUL] = helper_le_stl_mmu, - [MO_LEUQ] = helper_le_stq_mmu, - [MO_BEUW] = helper_be_stw_mmu, - [MO_BEUL] = helper_be_stl_mmu, - [MO_BEUQ] = helper_be_stq_mmu, -}; - /* * Because i686 has no register parameters and because x86_64 has xchg * to handle addr/data register overlap, we have placed all input arguments @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) } tcg_out_ld_helper_args(s, l, &ldst_helper_param); - tcg_out_branch(s, 1, qemu_ld_helpers[opc & (MO_BSWAP | MO_SIZE)]); + tcg_out_branch(s, 1, qemu_ld_helpers[opc & MO_SIZE]); tcg_out_ld_helper_ret(s, l, false, &ldst_helper_param); tcg_out_jmp(s, l->raddr); @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) } tcg_out_st_helper_args(s, l, &ldst_helper_param); - tcg_out_branch(s, 1, qemu_st_helpers[opc & (MO_BSWAP | MO_SIZE)]); + tcg_out_branch(s, 1, qemu_st_helpers[opc & MO_SIZE]); tcg_out_jmp(s, l->raddr); return true; diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, */ #if defined(CONFIG_SOFTMMU) -/* - * helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, - * MemOpIdx oi, uintptr_t ra) - */ -static void * const qemu_ld_helpers[4] = { - [MO_8] = helper_ret_ldub_mmu, - [MO_16] = helper_le_lduw_mmu, - [MO_32] = helper_le_ldul_mmu, - [MO_64] = helper_le_ldq_mmu, -}; - -/* - * helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, - * uintxx_t val, MemOpIdx oi, - * uintptr_t ra) - */ -static void * const qemu_st_helpers[4] = { - [MO_8] = helper_ret_stb_mmu, - [MO_16] = helper_le_stw_mmu, - [MO_32] = helper_le_stl_mmu, - [MO_64] = helper_le_stq_mmu, -}; - static bool tcg_out_goto(TCGContext *s, const tcg_insn_unit *target) { tcg_out_opc_b(s, 0); diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_call(TCGContext *s, const tcg_insn_unit *arg, } #if defined(CONFIG_SOFTMMU) -static void * const qemu_ld_helpers[MO_SSIZE + 1] = { - [MO_UB] = helper_ret_ldub_mmu, - [MO_SB] = helper_ret_ldsb_mmu, -#if HOST_BIG_ENDIAN - [MO_UW] = helper_be_lduw_mmu, - [MO_SW] = helper_be_ldsw_mmu, - [MO_UL] = helper_be_ldul_mmu, - [MO_SL] = helper_be_ldsl_mmu, - [MO_UQ] = helper_be_ldq_mmu, -#else - [MO_UW] = helper_le_lduw_mmu, - [MO_SW] = helper_le_ldsw_mmu, - [MO_UL] = helper_le_ldul_mmu, - [MO_UQ] = helper_le_ldq_mmu, - [MO_SL] = helper_le_ldsl_mmu, -#endif -}; - -static void * const qemu_st_helpers[MO_SIZE + 1] = { - [MO_UB] = helper_ret_stb_mmu, -#if HOST_BIG_ENDIAN - [MO_UW] = helper_be_stw_mmu, - [MO_UL] = helper_be_stl_mmu, - [MO_UQ] = helper_be_stq_mmu, -#else - [MO_UW] = helper_le_stw_mmu, - [MO_UL] = helper_le_stl_mmu, - [MO_UQ] = helper_le_stq_mmu, -#endif -}; - /* We have four temps, we might as well expose three of them. */ static const TCGLdstHelperParam ldst_helper_param = { .ntmp = 3, .tmp = { TCG_TMP0, TCG_TMP1, TCG_TMP2 } diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static const uint32_t qemu_stx_opc[(MO_SIZE + MO_BSWAP) + 1] = { }; #if defined (CONFIG_SOFTMMU) -/* helper signature: helper_ld_mmu(CPUState *env, target_ulong addr, - * int mmu_idx, uintptr_t ra) - */ -static void * const qemu_ld_helpers[(MO_SIZE | MO_BSWAP) + 1] = { - [MO_UB] = helper_ret_ldub_mmu, - [MO_LEUW] = helper_le_lduw_mmu, - [MO_LEUL] = helper_le_ldul_mmu, - [MO_LEUQ] = helper_le_ldq_mmu, - [MO_BEUW] = helper_be_lduw_mmu, - [MO_BEUL] = helper_be_ldul_mmu, - [MO_BEUQ] = helper_be_ldq_mmu, -}; - -/* helper signature: helper_st_mmu(CPUState *env, target_ulong addr, - * uintxx_t val, int mmu_idx, uintptr_t ra) - */ -static void * const qemu_st_helpers[(MO_SIZE | MO_BSWAP) + 1] = { - [MO_UB] = helper_ret_stb_mmu, - [MO_LEUW] = helper_le_stw_mmu, - [MO_LEUL] = helper_le_stl_mmu, - [MO_LEUQ] = helper_le_stq_mmu, - [MO_BEUW] = helper_be_stw_mmu, - [MO_BEUL] = helper_be_stl_mmu, - [MO_BEUQ] = helper_be_stq_mmu, -}; - static TCGReg ldst_ra_gen(TCGContext *s, const TCGLabelQemuLdst *l, int arg) { if (arg < 0) { @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) } tcg_out_ld_helper_args(s, lb, &ldst_helper_param); - tcg_out_call_int(s, LK, qemu_ld_helpers[opc & (MO_BSWAP | MO_SIZE)]); + tcg_out_call_int(s, LK, qemu_ld_helpers[opc & MO_SIZE]); tcg_out_ld_helper_ret(s, lb, false, &ldst_helper_param); tcg_out_b(s, 0, lb->raddr); @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) } tcg_out_st_helper_args(s, lb, &ldst_helper_param); - tcg_out_call_int(s, LK, qemu_st_helpers[opc & (MO_BSWAP | MO_SIZE)]); + tcg_out_call_int(s, LK, qemu_st_helpers[opc & MO_SIZE]); tcg_out_b(s, 0, lb->raddr); return true; diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_mb(TCGContext *s, TCGArg a0) */ #if defined(CONFIG_SOFTMMU) -/* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr, - * MemOpIdx oi, uintptr_t ra) - */ -static void * const qemu_ld_helpers[MO_SSIZE + 1] = { - [MO_UB] = helper_ret_ldub_mmu, - [MO_SB] = helper_ret_ldsb_mmu, -#if HOST_BIG_ENDIAN - [MO_UW] = helper_be_lduw_mmu, - [MO_SW] = helper_be_ldsw_mmu, - [MO_UL] = helper_be_ldul_mmu, -#if TCG_TARGET_REG_BITS == 64 - [MO_SL] = helper_be_ldsl_mmu, -#endif - [MO_UQ] = helper_be_ldq_mmu, -#else - [MO_UW] = helper_le_lduw_mmu, - [MO_SW] = helper_le_ldsw_mmu, - [MO_UL] = helper_le_ldul_mmu, -#if TCG_TARGET_REG_BITS == 64 - [MO_SL] = helper_le_ldsl_mmu, -#endif - [MO_UQ] = helper_le_ldq_mmu, -#endif -}; - -/* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr, - * uintxx_t val, MemOpIdx oi, - * uintptr_t ra) - */ -static void * const qemu_st_helpers[MO_SIZE + 1] = { - [MO_8] = helper_ret_stb_mmu, -#if HOST_BIG_ENDIAN - [MO_16] = helper_be_stw_mmu, - [MO_32] = helper_be_stl_mmu, - [MO_64] = helper_be_stq_mmu, -#else - [MO_16] = helper_le_stw_mmu, - [MO_32] = helper_le_stl_mmu, - [MO_64] = helper_le_stq_mmu, -#endif -}; - static void tcg_out_goto(TCGContext *s, const tcg_insn_unit *target) { tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, 0); diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static const uint8_t tcg_cond_to_ltr_cond[] = { [TCG_COND_GEU] = S390_CC_ALWAYS, }; -#ifdef CONFIG_SOFTMMU -static void * const qemu_ld_helpers[(MO_SSIZE | MO_BSWAP) + 1] = { - [MO_UB] = helper_ret_ldub_mmu, - [MO_SB] = helper_ret_ldsb_mmu, - [MO_LEUW] = helper_le_lduw_mmu, - [MO_LESW] = helper_le_ldsw_mmu, - [MO_LEUL] = helper_le_ldul_mmu, - [MO_LESL] = helper_le_ldsl_mmu, - [MO_LEUQ] = helper_le_ldq_mmu, - [MO_BEUW] = helper_be_lduw_mmu, - [MO_BESW] = helper_be_ldsw_mmu, - [MO_BEUL] = helper_be_ldul_mmu, - [MO_BESL] = helper_be_ldsl_mmu, - [MO_BEUQ] = helper_be_ldq_mmu, -}; - -static void * const qemu_st_helpers[(MO_SIZE | MO_BSWAP) + 1] = { - [MO_UB] = helper_ret_stb_mmu, - [MO_LEUW] = helper_le_stw_mmu, - [MO_LEUL] = helper_le_stl_mmu, - [MO_LEUQ] = helper_le_stq_mmu, - [MO_BEUW] = helper_be_stw_mmu, - [MO_BEUL] = helper_be_stl_mmu, - [MO_BEUQ] = helper_be_stq_mmu, -}; -#endif - static const tcg_insn_unit *tb_ret_addr; uint64_t s390_facilities[3]; @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) } tcg_out_ld_helper_args(s, lb, &ldst_helper_param); - tcg_out_call_int(s, qemu_ld_helpers[opc & (MO_BSWAP | MO_SIZE)]); + tcg_out_call_int(s, qemu_ld_helpers[opc & MO_SIZE]); tcg_out_ld_helper_ret(s, lb, false, &ldst_helper_param); tgen_gotoi(s, S390_CC_ALWAYS, lb->raddr); @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) } tcg_out_st_helper_args(s, lb, &ldst_helper_param); - tcg_out_call_int(s, qemu_st_helpers[opc & (MO_BSWAP | MO_SIZE)]); + tcg_out_call_int(s, qemu_st_helpers[opc & MO_SIZE]); tgen_gotoi(s, S390_CC_ALWAYS, lb->raddr); return true; diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_mb(TCGContext *s, TCGArg a0) } #ifdef CONFIG_SOFTMMU -static const tcg_insn_unit *qemu_ld_trampoline[(MO_SSIZE | MO_BSWAP) + 1]; -static const tcg_insn_unit *qemu_st_trampoline[(MO_SIZE | MO_BSWAP) + 1]; +static const tcg_insn_unit *qemu_ld_trampoline[MO_SSIZE + 1]; +static const tcg_insn_unit *qemu_st_trampoline[MO_SIZE + 1]; static void build_trampolines(TCGContext *s) { - static void * const qemu_ld_helpers[] = { - [MO_UB] = helper_ret_ldub_mmu, - [MO_SB] = helper_ret_ldsb_mmu, - [MO_LEUW] = helper_le_lduw_mmu, - [MO_LESW] = helper_le_ldsw_mmu, - [MO_LEUL] = helper_le_ldul_mmu, - [MO_LEUQ] = helper_le_ldq_mmu, - [MO_BEUW] = helper_be_lduw_mmu, - [MO_BESW] = helper_be_ldsw_mmu, - [MO_BEUL] = helper_be_ldul_mmu, - [MO_BEUQ] = helper_be_ldq_mmu, - }; - static void * const qemu_st_helpers[] = { - [MO_UB] = helper_ret_stb_mmu, - [MO_LEUW] = helper_le_stw_mmu, - [MO_LEUL] = helper_le_stl_mmu, - [MO_LEUQ] = helper_le_stq_mmu, - [MO_BEUW] = helper_be_stw_mmu, - [MO_BEUL] = helper_be_stl_mmu, - [MO_BEUQ] = helper_be_stq_mmu, - }; - int i; for (i = 0; i < ARRAY_SIZE(qemu_ld_helpers); ++i) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr, /* We use the helpers to extend SB and SW data, leaving the case of SL needing explicit extending below. */ if ((memop & MO_SSIZE) == MO_SL) { - func = qemu_ld_trampoline[memop & (MO_BSWAP | MO_SIZE)]; + func = qemu_ld_trampoline[MO_UL]; } else { - func = qemu_ld_trampoline[memop & (MO_BSWAP | MO_SSIZE)]; + func = qemu_ld_trampoline[memop & MO_SSIZE]; } tcg_debug_assert(func != NULL); tcg_out_call_nodelay(s, func, false); @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data, TCGReg addr, tcg_out_movext(s, (memop & MO_SIZE) == MO_64 ? TCG_TYPE_I64 : TCG_TYPE_I32, TCG_REG_O2, data_type, memop & MO_SIZE, data); - func = qemu_st_trampoline[memop & (MO_BSWAP | MO_SIZE)]; + func = qemu_st_trampoline[memop & MO_SIZE]; tcg_debug_assert(func != NULL); tcg_out_call_nodelay(s, func, false); /* delay slot */ -- 2.34.1 TCG backends may need to defer to a helper to implement the atomicity required by a given operation. Mirror the interface used in system mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-ldst.h | 6 +- accel/tcg/user-exec.c | 393 ++++++++++++++++++++++++++++------------- tcg/tcg.c | 6 +- 3 files changed, 278 insertions(+), 127 deletions(-) diff --git a/include/tcg/tcg-ldst.h b/include/tcg/tcg-ldst.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-ldst.h +++ b/include/tcg/tcg-ldst.h @@ -XXX,XX +XXX,XX @@ #ifndef TCG_LDST_H #define TCG_LDST_H -#ifdef CONFIG_SOFTMMU - /* Value zero-extended to tcg register size. */ tcg_target_ulong helper_ldub_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, uintptr_t retaddr); @@ -XXX,XX +XXX,XX @@ void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, MemOpIdx oi, uintptr_t retaddr); -#else +#ifdef CONFIG_USER_ONLY G_NORETURN void helper_unaligned_ld(CPUArchState *env, target_ulong addr); G_NORETURN void helper_unaligned_st(CPUArchState *env, target_ulong addr); -#endif /* CONFIG_SOFTMMU */ +#endif /* CONFIG_USER_ONLY */ #endif /* TCG_LDST_H */ diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/user-exec.c +++ b/accel/tcg/user-exec.c @@ -XXX,XX +XXX,XX @@ void page_reset_target_data(target_ulong start, target_ulong last) { } /* The softmmu versions of these helpers are in cputlb.c. */ -/* - * Verify that we have passed the correct MemOp to the correct function. - * - * We could present one function to target code, and dispatch based on - * the MemOp, but so far we have worked hard to avoid an indirect function - * call along the memory path. - */ -static void validate_memop(MemOpIdx oi, MemOp expected) -{ -#ifdef CONFIG_DEBUG_TCG - MemOp have = get_memop(oi) & (MO_SIZE | MO_BSWAP); - assert(have == expected); -#endif -} - void helper_unaligned_ld(CPUArchState *env, target_ulong addr) { cpu_loop_exit_sigbus(env_cpu(env), addr, MMU_DATA_LOAD, GETPC()); @@ -XXX,XX +XXX,XX @@ void helper_unaligned_st(CPUArchState *env, target_ulong addr) cpu_loop_exit_sigbus(env_cpu(env), addr, MMU_DATA_STORE, GETPC()); } -static void *cpu_mmu_lookup(CPUArchState *env, target_ulong addr, - MemOpIdx oi, uintptr_t ra, MMUAccessType type) +static void *cpu_mmu_lookup(CPUArchState *env, abi_ptr addr, + MemOp mop, uintptr_t ra, MMUAccessType type) { - MemOp mop = get_memop(oi); int a_bits = get_alignment_bits(mop); void *ret; @@ -XXX,XX +XXX,XX @@ static void *cpu_mmu_lookup(CPUArchState *env, target_ulong addr, #include "ldst_atomicity.c.inc" -uint8_t cpu_ldb_mmu(CPUArchState *env, abi_ptr addr, - MemOpIdx oi, uintptr_t ra) +static uint8_t do_ld1_mmu(CPUArchState *env, abi_ptr addr, + MemOp mop, uintptr_t ra) { void *haddr; uint8_t ret; - validate_memop(oi, MO_UB); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); + tcg_debug_assert((mop & MO_SIZE) == MO_8); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_LOAD); ret = ldub_p(haddr); clear_helper_retaddr(); + return ret; +} + +tcg_target_ulong helper_ldub_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + return do_ld1_mmu(env, addr, get_memop(oi), ra); +} + +tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + return (int8_t)do_ld1_mmu(env, addr, get_memop(oi), ra); +} + +uint8_t cpu_ldb_mmu(CPUArchState *env, abi_ptr addr, + MemOpIdx oi, uintptr_t ra) +{ + uint8_t ret = do_ld1_mmu(env, addr, get_memop(oi), ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); return ret; } +static uint16_t do_ld2_he_mmu(CPUArchState *env, abi_ptr addr, + MemOp mop, uintptr_t ra) +{ + void *haddr; + uint16_t ret; + + tcg_debug_assert((mop & MO_SIZE) == MO_16); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_LOAD); + ret = load_atom_2(env, ra, haddr, mop); + clear_helper_retaddr(); + return ret; +} + +tcg_target_ulong helper_lduw_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + uint16_t ret = do_ld2_he_mmu(env, addr, mop, ra); + + if (mop & MO_BSWAP) { + ret = bswap16(ret); + } + return ret; +} + +tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + int16_t ret = do_ld2_he_mmu(env, addr, mop, ra); + + if (mop & MO_BSWAP) { + ret = bswap16(ret); + } + return ret; +} + uint16_t cpu_ldw_be_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); uint16_t ret; - validate_memop(oi, MO_BEUW); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = load_atom_2(env, ra, haddr, get_memop(oi)); - clear_helper_retaddr(); + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); + ret = do_ld2_he_mmu(env, addr, mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); return cpu_to_be16(ret); } -uint32_t cpu_ldl_be_mmu(CPUArchState *env, abi_ptr addr, - MemOpIdx oi, uintptr_t ra) -{ - void *haddr; - uint32_t ret; - - validate_memop(oi, MO_BEUL); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = load_atom_4(env, ra, haddr, get_memop(oi)); - clear_helper_retaddr(); - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return cpu_to_be32(ret); -} - -uint64_t cpu_ldq_be_mmu(CPUArchState *env, abi_ptr addr, - MemOpIdx oi, uintptr_t ra) -{ - void *haddr; - uint64_t ret; - - validate_memop(oi, MO_BEUQ); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = load_atom_8(env, ra, haddr, get_memop(oi)); - clear_helper_retaddr(); - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return cpu_to_be64(ret); -} - uint16_t cpu_ldw_le_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); uint16_t ret; - validate_memop(oi, MO_LEUW); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = load_atom_2(env, ra, haddr, get_memop(oi)); - clear_helper_retaddr(); + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); + ret = do_ld2_he_mmu(env, addr, mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); return cpu_to_le16(ret); } +static uint32_t do_ld4_he_mmu(CPUArchState *env, abi_ptr addr, + MemOp mop, uintptr_t ra) +{ + void *haddr; + uint32_t ret; + + tcg_debug_assert((mop & MO_SIZE) == MO_32); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_LOAD); + ret = load_atom_4(env, ra, haddr, mop); + clear_helper_retaddr(); + return ret; +} + +tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + uint32_t ret = do_ld4_he_mmu(env, addr, mop, ra); + + if (mop & MO_BSWAP) { + ret = bswap32(ret); + } + return ret; +} + +tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + int32_t ret = do_ld4_he_mmu(env, addr, mop, ra); + + if (mop & MO_BSWAP) { + ret = bswap32(ret); + } + return ret; +} + +uint32_t cpu_ldl_be_mmu(CPUArchState *env, abi_ptr addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + uint32_t ret; + + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); + ret = do_ld4_he_mmu(env, addr, mop, ra); + qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); + return cpu_to_be32(ret); +} + uint32_t cpu_ldl_le_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); uint32_t ret; - validate_memop(oi, MO_LEUL); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = load_atom_4(env, ra, haddr, get_memop(oi)); - clear_helper_retaddr(); + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); + ret = do_ld4_he_mmu(env, addr, mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); return cpu_to_le32(ret); } +static uint64_t do_ld8_he_mmu(CPUArchState *env, abi_ptr addr, + MemOp mop, uintptr_t ra) +{ + void *haddr; + uint64_t ret; + + tcg_debug_assert((mop & MO_SIZE) == MO_64); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_LOAD); + ret = load_atom_8(env, ra, haddr, mop); + clear_helper_retaddr(); + return ret; +} + +uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + uint64_t ret = do_ld8_he_mmu(env, addr, mop, ra); + + if (mop & MO_BSWAP) { + ret = bswap64(ret); + } + return ret; +} + +uint64_t cpu_ldq_be_mmu(CPUArchState *env, abi_ptr addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + uint64_t ret; + + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); + ret = do_ld8_he_mmu(env, addr, mop, ra); + qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); + return cpu_to_be64(ret); +} + uint64_t cpu_ldq_le_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); uint64_t ret; - validate_memop(oi, MO_LEUQ); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - ret = load_atom_8(env, ra, haddr, get_memop(oi)); - clear_helper_retaddr(); + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); + ret = do_ld8_he_mmu(env, addr, mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); return cpu_to_le64(ret); } @@ -XXX,XX +XXX,XX @@ Int128 cpu_ld16_be_mmu(CPUArchState *env, abi_ptr addr, void *haddr; Int128 ret; - validate_memop(oi, MO_128 | MO_BE); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_BE)); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); memcpy(&ret, haddr, 16); clear_helper_retaddr(); @@ -XXX,XX +XXX,XX @@ Int128 cpu_ld16_le_mmu(CPUArchState *env, abi_ptr addr, void *haddr; Int128 ret; - validate_memop(oi, MO_128 | MO_LE); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_LE)); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); memcpy(&ret, haddr, 16); clear_helper_retaddr(); @@ -XXX,XX +XXX,XX @@ Int128 cpu_ld16_le_mmu(CPUArchState *env, abi_ptr addr, return ret; } -void cpu_stb_mmu(CPUArchState *env, abi_ptr addr, uint8_t val, - MemOpIdx oi, uintptr_t ra) +static void do_st1_mmu(CPUArchState *env, abi_ptr addr, uint8_t val, + MemOp mop, uintptr_t ra) { void *haddr; - validate_memop(oi, MO_UB); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); + tcg_debug_assert((mop & MO_SIZE) == MO_8); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_STORE); stb_p(haddr, val); clear_helper_retaddr(); +} + +void helper_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t ra) +{ + do_st1_mmu(env, addr, val, get_memop(oi), ra); +} + +void cpu_stb_mmu(CPUArchState *env, abi_ptr addr, uint8_t val, + MemOpIdx oi, uintptr_t ra) +{ + do_st1_mmu(env, addr, val, get_memop(oi), ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } +static void do_st2_he_mmu(CPUArchState *env, abi_ptr addr, uint16_t val, + MemOp mop, uintptr_t ra) +{ + void *haddr; + + tcg_debug_assert((mop & MO_SIZE) == MO_16); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_STORE); + store_atom_2(env, ra, haddr, mop, val); + clear_helper_retaddr(); +} + +void helper_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + if (mop & MO_BSWAP) { + val = bswap16(val); + } + do_st2_he_mmu(env, addr, val, mop, ra); +} + void cpu_stw_be_mmu(CPUArchState *env, abi_ptr addr, uint16_t val, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); - validate_memop(oi, MO_BEUW); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - store_atom_2(env, ra, haddr, get_memop(oi), be16_to_cpu(val)); - clear_helper_retaddr(); - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); -} - -void cpu_stl_be_mmu(CPUArchState *env, abi_ptr addr, uint32_t val, - MemOpIdx oi, uintptr_t ra) -{ - void *haddr; - - validate_memop(oi, MO_BEUL); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - store_atom_4(env, ra, haddr, get_memop(oi), be32_to_cpu(val)); - clear_helper_retaddr(); - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); -} - -void cpu_stq_be_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, - MemOpIdx oi, uintptr_t ra) -{ - void *haddr; - - validate_memop(oi, MO_BEUQ); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - store_atom_8(env, ra, haddr, get_memop(oi), be64_to_cpu(val)); - clear_helper_retaddr(); + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); + do_st2_he_mmu(env, addr, be16_to_cpu(val), mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } void cpu_stw_le_mmu(CPUArchState *env, abi_ptr addr, uint16_t val, MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); + do_st2_he_mmu(env, addr, le16_to_cpu(val), mop, ra); + qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); +} + +static void do_st4_he_mmu(CPUArchState *env, abi_ptr addr, uint32_t val, + MemOp mop, uintptr_t ra) { void *haddr; - validate_memop(oi, MO_LEUW); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - store_atom_2(env, ra, haddr, get_memop(oi), le16_to_cpu(val)); + tcg_debug_assert((mop & MO_SIZE) == MO_32); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_STORE); + store_atom_4(env, ra, haddr, mop, val); clear_helper_retaddr(); +} + +void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + if (mop & MO_BSWAP) { + val = bswap32(val); + } + do_st4_he_mmu(env, addr, val, mop, ra); +} + +void cpu_stl_be_mmu(CPUArchState *env, abi_ptr addr, uint32_t val, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); + do_st4_he_mmu(env, addr, be32_to_cpu(val), mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } void cpu_stl_le_mmu(CPUArchState *env, abi_ptr addr, uint32_t val, MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); + do_st4_he_mmu(env, addr, le32_to_cpu(val), mop, ra); + qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); +} + +static void do_st8_he_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, + MemOp mop, uintptr_t ra) { void *haddr; - validate_memop(oi, MO_LEUL); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - store_atom_4(env, ra, haddr, get_memop(oi), le32_to_cpu(val)); + tcg_debug_assert((mop & MO_SIZE) == MO_64); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_STORE); + store_atom_8(env, ra, haddr, mop, val); clear_helper_retaddr(); +} + +void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + if (mop & MO_BSWAP) { + val = bswap64(val); + } + do_st8_he_mmu(env, addr, val, mop, ra); +} + +void cpu_stq_be_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); + do_st8_he_mmu(env, addr, cpu_to_be64(val), mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } void cpu_stq_le_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); - validate_memop(oi, MO_LEUQ); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); - store_atom_8(env, ra, haddr, get_memop(oi), le64_to_cpu(val)); - clear_helper_retaddr(); + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); + do_st8_he_mmu(env, addr, cpu_to_le64(val), mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } @@ -XXX,XX +XXX,XX @@ void cpu_st16_be_mmu(CPUArchState *env, abi_ptr addr, { void *haddr; - validate_memop(oi, MO_128 | MO_BE); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_BE)); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); if (!HOST_BIG_ENDIAN) { val = bswap128(val); @@ -XXX,XX +XXX,XX @@ void cpu_st16_le_mmu(CPUArchState *env, abi_ptr addr, { void *haddr; - validate_memop(oi, MO_128 | MO_LE); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_LE)); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); if (HOST_BIG_ENDIAN) { val = bswap128(val); @@ -XXX,XX +XXX,XX @@ uint64_t cpu_ldq_code_mmu(CPUArchState *env, abi_ptr addr, void *haddr; uint64_t ret; - validate_memop(oi, MO_BEUQ); haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); ret = ldq_p(haddr); clear_helper_retaddr(); diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *l, const TCGLdstHelperParam *p) __attribute__((unused)); -#ifdef CONFIG_SOFTMMU -static void * const qemu_ld_helpers[MO_SSIZE + 1] = { +static void * const qemu_ld_helpers[MO_SSIZE + 1] __attribute__((unused)) = { [MO_UB] = helper_ldub_mmu, [MO_SB] = helper_ldsb_mmu, [MO_UW] = helper_lduw_mmu, @@ -XXX,XX +XXX,XX @@ static void * const qemu_ld_helpers[MO_SSIZE + 1] = { #endif }; -static void * const qemu_st_helpers[MO_SIZE + 1] = { +static void * const qemu_st_helpers[MO_SIZE + 1] __attribute__((unused)) = { [MO_8] = helper_stb_mmu, [MO_16] = helper_stw_mmu, [MO_32] = helper_stl_mmu, [MO_64] = helper_stq_mmu, }; -#endif TCGContext tcg_init_ctx; __thread TCGContext *tcg_ctx; -- 2.34.1 We can now fold these two pieces of code. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tci.c | 89 ------------------------------------------------------- 1 file changed, 89 deletions(-) diff --git a/tcg/tci.c b/tcg/tci.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci.c +++ b/tcg/tci.c @@ -XXX,XX +XXX,XX @@ static uint64_t tci_qemu_ld(CPUArchState *env, target_ulong taddr, MemOp mop = get_memop(oi); uintptr_t ra = (uintptr_t)tb_ptr; -#ifdef CONFIG_SOFTMMU switch (mop & MO_SSIZE) { case MO_UB: return helper_ldub_mmu(env, taddr, oi, ra); @@ -XXX,XX +XXX,XX @@ static uint64_t tci_qemu_ld(CPUArchState *env, target_ulong taddr, default: g_assert_not_reached(); } -#else - void *haddr = g2h(env_cpu(env), taddr); - unsigned a_mask = (1u << get_alignment_bits(mop)) - 1; - uint64_t ret; - - set_helper_retaddr(ra); - if (taddr & a_mask) { - helper_unaligned_ld(env, taddr); - } - switch (mop & (MO_BSWAP | MO_SSIZE)) { - case MO_UB: - ret = ldub_p(haddr); - break; - case MO_SB: - ret = ldsb_p(haddr); - break; - case MO_LEUW: - ret = lduw_le_p(haddr); - break; - case MO_LESW: - ret = ldsw_le_p(haddr); - break; - case MO_LEUL: - ret = (uint32_t)ldl_le_p(haddr); - break; - case MO_LESL: - ret = (int32_t)ldl_le_p(haddr); - break; - case MO_LEUQ: - ret = ldq_le_p(haddr); - break; - case MO_BEUW: - ret = lduw_be_p(haddr); - break; - case MO_BESW: - ret = ldsw_be_p(haddr); - break; - case MO_BEUL: - ret = (uint32_t)ldl_be_p(haddr); - break; - case MO_BESL: - ret = (int32_t)ldl_be_p(haddr); - break; - case MO_BEUQ: - ret = ldq_be_p(haddr); - break; - default: - g_assert_not_reached(); - } - clear_helper_retaddr(); - return ret; -#endif } static void tci_qemu_st(CPUArchState *env, target_ulong taddr, uint64_t val, @@ -XXX,XX +XXX,XX @@ static void tci_qemu_st(CPUArchState *env, target_ulong taddr, uint64_t val, MemOp mop = get_memop(oi); uintptr_t ra = (uintptr_t)tb_ptr; -#ifdef CONFIG_SOFTMMU switch (mop & MO_SIZE) { case MO_UB: helper_stb_mmu(env, taddr, val, oi, ra); @@ -XXX,XX +XXX,XX @@ static void tci_qemu_st(CPUArchState *env, target_ulong taddr, uint64_t val, default: g_assert_not_reached(); } -#else - void *haddr = g2h(env_cpu(env), taddr); - unsigned a_mask = (1u << get_alignment_bits(mop)) - 1; - - set_helper_retaddr(ra); - if (taddr & a_mask) { - helper_unaligned_st(env, taddr); - } - switch (mop & (MO_BSWAP | MO_SIZE)) { - case MO_UB: - stb_p(haddr, val); - break; - case MO_LEUW: - stw_le_p(haddr, val); - break; - case MO_LEUL: - stl_le_p(haddr, val); - break; - case MO_LEUQ: - stq_le_p(haddr, val); - break; - case MO_BEUW: - stw_be_p(haddr, val); - break; - case MO_BEUL: - stl_be_p(haddr, val); - break; - case MO_BEUQ: - stq_be_p(haddr, val); - break; - default: - g_assert_not_reached(); - } - clear_helper_retaddr(); -#endif } #if TCG_TARGET_REG_BITS == 64 -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/tcg-runtime.h | 3 + include/tcg/tcg-ldst.h | 4 + accel/tcg/cputlb.c | 399 +++++++++++++++++++++++++-------- accel/tcg/user-exec.c | 94 ++++++-- tcg/tcg-op.c | 173 +++++++++----- accel/tcg/ldst_atomicity.c.inc | 184 +++++++++++++++ 6 files changed, 679 insertions(+), 178 deletions(-) diff --git a/accel/tcg/tcg-runtime.h b/accel/tcg/tcg-runtime.h index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/tcg-runtime.h +++ b/accel/tcg/tcg-runtime.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_1(exit_atomic, TCG_CALL_NO_WG, noreturn, env) DEF_HELPER_FLAGS_3(memset, TCG_CALL_NO_RWG, ptr, ptr, int, ptr) #endif /* IN_HELPER_PROTO */ +DEF_HELPER_FLAGS_3(ld_i128, TCG_CALL_NO_WG, i128, env, tl, i32) +DEF_HELPER_FLAGS_4(st_i128, TCG_CALL_NO_WG, void, env, tl, i128, i32) + DEF_HELPER_FLAGS_5(atomic_cmpxchgb, TCG_CALL_NO_WG, i32, env, tl, i32, i32, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgw_be, TCG_CALL_NO_WG, diff --git a/include/tcg/tcg-ldst.h b/include/tcg/tcg-ldst.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-ldst.h +++ b/include/tcg/tcg-ldst.h @@ -XXX,XX +XXX,XX @@ tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, uintptr_t retaddr); uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, uintptr_t retaddr); +Int128 helper_ld16_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t retaddr); /* Value sign-extended to tcg register size. */ tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, target_ulong addr, @@ -XXX,XX +XXX,XX @@ void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr); void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, MemOpIdx oi, uintptr_t retaddr); +void helper_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, + MemOpIdx oi, uintptr_t retaddr); #ifdef CONFIG_USER_ONLY diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -XXX,XX +XXX,XX @@ #include "qemu/plugin-memory.h" #endif #include "tcg/tcg-ldst.h" +#include "exec/helper-proto.h" /* DEBUG defines, enable DEBUG_TLB_LOG to log to the CPU_LOG_MMU target */ /* #define DEBUG_TLB */ @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld_whole_be8(CPUArchState *env, uintptr_t ra, return (ret_be << (p->size * 8)) | x; } +/** + * do_ld_parts_be16 + * @p: translation parameters + * @ret_be: accumulated data + * + * As do_ld_bytes_beN, but with one atomic load. + * 16 aligned bytes are guaranteed to cover the load. + */ +static Int128 do_ld_whole_be16(CPUArchState *env, uintptr_t ra, + MMULookupPageData *p, uint64_t ret_be) +{ + int o = p->addr & 15; + Int128 x, y = load_atomic16_or_exit(env, ra, p->haddr - o); + int size = p->size; + + if (!HOST_BIG_ENDIAN) { + y = bswap128(y); + } + y = int128_lshift(y, o * 8); + y = int128_urshift(y, (16 - size) * 8); + x = int128_make64(ret_be); + x = int128_lshift(x, size * 8); + return int128_or(x, y); +} + /* * Wrapper for the above. */ @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld_beN(CPUArchState *env, MMULookupPageData *p, } } +/* + * Wrapper for the above, for 8 < size < 16. + */ +static Int128 do_ld16_beN(CPUArchState *env, MMULookupPageData *p, + uint64_t a, int mmu_idx, MemOp mop, uintptr_t ra) +{ + int size = p->size; + uint64_t b; + MemOp atom; + + if (unlikely(p->flags & TLB_MMIO)) { + p->size = size - 8; + a = do_ld_mmio_beN(env, p, a, mmu_idx, MMU_DATA_LOAD, ra); + p->addr += p->size; + p->size = 8; + b = do_ld_mmio_beN(env, p, 0, mmu_idx, MMU_DATA_LOAD, ra); + return int128_make128(b, a); + } + + /* + * It is a given that we cross a page and therefore there is no + * atomicity for the load as a whole, but subobjects may need attention. + */ + atom = mop & MO_ATOM_MASK; + switch (atom) { + case MO_ATOM_SUBALIGN: + p->size = size - 8; + a = do_ld_parts_beN(p, a); + p->haddr += size - 8; + p->size = 8; + b = do_ld_parts_beN(p, 0); + break; + + case MO_ATOM_WITHIN16_PAIR: + /* Since size > 8, this is the half that must be atomic. */ + return do_ld_whole_be16(env, ra, p, a); + + case MO_ATOM_IFALIGN_PAIR: + /* + * Since size > 8, both halves are misaligned, + * and so neither is atomic. + */ + case MO_ATOM_IFALIGN: + case MO_ATOM_WITHIN16: + case MO_ATOM_NONE: + p->size = size - 8; + a = do_ld_bytes_beN(p, a); + b = ldq_be_p(p->haddr + size - 8); + break; + + default: + g_assert_not_reached(); + } + + return int128_make128(b, a); +} + static uint8_t do_ld_1(CPUArchState *env, MMULookupPageData *p, int mmu_idx, MMUAccessType type, uintptr_t ra) { @@ -XXX,XX +XXX,XX @@ tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, target_ulong addr, return (int32_t)helper_ldul_mmu(env, addr, oi, retaddr); } +static Int128 do_ld16_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + MMULookupLocals l; + bool crosspage; + uint64_t a, b; + Int128 ret; + int first; + + crosspage = mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD, &l); + if (likely(!crosspage)) { + /* Perform the load host endian. */ + if (unlikely(l.page[0].flags & TLB_MMIO)) { + QEMU_IOTHREAD_LOCK_GUARD(); + a = io_readx(env, l.page[0].full, l.mmu_idx, addr, + ra, MMU_DATA_LOAD, MO_64); + b = io_readx(env, l.page[0].full, l.mmu_idx, addr + 8, + ra, MMU_DATA_LOAD, MO_64); + ret = int128_make128(HOST_BIG_ENDIAN ? b : a, + HOST_BIG_ENDIAN ? a : b); + } else { + ret = load_atom_16(env, ra, l.page[0].haddr, l.memop); + } + if (l.memop & MO_BSWAP) { + ret = bswap128(ret); + } + return ret; + } + + first = l.page[0].size; + if (first == 8) { + MemOp mop8 = (l.memop & ~MO_SIZE) | MO_64; + + a = do_ld_8(env, &l.page[0], l.mmu_idx, MMU_DATA_LOAD, mop8, ra); + b = do_ld_8(env, &l.page[1], l.mmu_idx, MMU_DATA_LOAD, mop8, ra); + if ((mop8 & MO_BSWAP) == MO_LE) { + ret = int128_make128(a, b); + } else { + ret = int128_make128(b, a); + } + return ret; + } + + if (first < 8) { + a = do_ld_beN(env, &l.page[0], 0, l.mmu_idx, + MMU_DATA_LOAD, l.memop, ra); + ret = do_ld16_beN(env, &l.page[1], a, l.mmu_idx, l.memop, ra); + } else { + ret = do_ld16_beN(env, &l.page[0], 0, l.mmu_idx, l.memop, ra); + b = int128_getlo(ret); + ret = int128_lshift(ret, l.page[1].size * 8); + a = int128_gethi(ret); + b = do_ld_beN(env, &l.page[1], b, l.mmu_idx, + MMU_DATA_LOAD, l.memop, ra); + ret = int128_make128(b, a); + } + if ((l.memop & MO_BSWAP) == MO_LE) { + ret = bswap128(ret); + } + return ret; +} + +Int128 helper_ld16_mmu(CPUArchState *env, target_ulong addr, + uint32_t oi, uintptr_t retaddr) +{ + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_128); + return do_ld16_mmu(env, addr, oi, retaddr); +} + +Int128 helper_ld_i128(CPUArchState *env, target_ulong addr, uint32_t oi) +{ + return helper_ld16_mmu(env, addr, oi, GETPC()); +} + /* * Load helpers for cpu_ldst.h. */ @@ -XXX,XX +XXX,XX @@ uint64_t cpu_ldq_le_mmu(CPUArchState *env, abi_ptr addr, Int128 cpu_ld16_be_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { - MemOp mop = get_memop(oi); - int mmu_idx = get_mmuidx(oi); - MemOpIdx new_oi; - unsigned a_bits; - uint64_t h, l; + Int128 ret; - tcg_debug_assert((mop & (MO_BSWAP|MO_SSIZE)) == (MO_BE|MO_128)); - a_bits = get_alignment_bits(mop); - - /* Handle CPU specific unaligned behaviour */ - if (addr & ((1 << a_bits) - 1)) { - cpu_unaligned_access(env_cpu(env), addr, MMU_DATA_LOAD, - mmu_idx, ra); - } - - /* Construct an unaligned 64-bit replacement MemOpIdx. */ - mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; - new_oi = make_memop_idx(mop, mmu_idx); - - h = helper_ldq_mmu(env, addr, new_oi, ra); - l = helper_ldq_mmu(env, addr + 8, new_oi, ra); - - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return int128_make128(l, h); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP|MO_SIZE)) == (MO_BE|MO_128)); + ret = do_ld16_mmu(env, addr, oi, ra); + plugin_load_cb(env, addr, oi); + return ret; } Int128 cpu_ld16_le_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { - MemOp mop = get_memop(oi); - int mmu_idx = get_mmuidx(oi); - MemOpIdx new_oi; - unsigned a_bits; - uint64_t h, l; + Int128 ret; - tcg_debug_assert((mop & (MO_BSWAP|MO_SSIZE)) == (MO_LE|MO_128)); - a_bits = get_alignment_bits(mop); - - /* Handle CPU specific unaligned behaviour */ - if (addr & ((1 << a_bits) - 1)) { - cpu_unaligned_access(env_cpu(env), addr, MMU_DATA_LOAD, - mmu_idx, ra); - } - - /* Construct an unaligned 64-bit replacement MemOpIdx. */ - mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; - new_oi = make_memop_idx(mop, mmu_idx); - - l = helper_ldq_mmu(env, addr, new_oi, ra); - h = helper_ldq_mmu(env, addr + 8, new_oi, ra); - - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - return int128_make128(l, h); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP|MO_SIZE)) == (MO_LE|MO_128)); + ret = do_ld16_mmu(env, addr, oi, ra); + plugin_load_cb(env, addr, oi); + return ret; } /* @@ -XXX,XX +XXX,XX @@ static uint64_t do_st_leN(CPUArchState *env, MMULookupPageData *p, } } +/* + * Wrapper for the above, for 8 < size < 16. + */ +static uint64_t do_st16_leN(CPUArchState *env, MMULookupPageData *p, + Int128 val_le, int mmu_idx, + MemOp mop, uintptr_t ra) +{ + int size = p->size; + MemOp atom; + + if (unlikely(p->flags & TLB_MMIO)) { + p->size = 8; + do_st_mmio_leN(env, p, int128_getlo(val_le), mmu_idx, ra); + p->size = size - 8; + p->addr += 8; + return do_st_mmio_leN(env, p, int128_gethi(val_le), mmu_idx, ra); + } else if (unlikely(p->flags & TLB_DISCARD_WRITE)) { + return int128_gethi(val_le) >> ((size - 8) * 8); + } + + /* + * It is a given that we cross a page and therefore there is no atomicity + * for the store as a whole, but subobjects may need attention. + */ + atom = mop & MO_ATOM_MASK; + switch (atom) { + case MO_ATOM_SUBALIGN: + store_parts_leN(p->haddr, 8, int128_getlo(val_le)); + return store_parts_leN(p->haddr + 8, p->size - 8, + int128_gethi(val_le)); + + case MO_ATOM_WITHIN16_PAIR: + /* Since size > 8, this is the half that must be atomic. */ + if (!HAVE_al16) { + cpu_loop_exit_atomic(env_cpu(env), ra); + } + return store_whole_le16(p->haddr, p->size, val_le); + + case MO_ATOM_IFALIGN_PAIR: + /* + * Since size > 8, both halves are misaligned, + * and so neither is atomic. + */ + case MO_ATOM_IFALIGN: + case MO_ATOM_NONE: + stq_le_p(p->haddr, int128_getlo(val_le)); + return store_bytes_leN(p->haddr + 8, p->size - 8, + int128_gethi(val_le)); + + default: + g_assert_not_reached(); + } +} + static void do_st_1(CPUArchState *env, MMULookupPageData *p, uint8_t val, int mmu_idx, uintptr_t ra) { @@ -XXX,XX +XXX,XX @@ void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, do_st8_mmu(env, addr, val, oi, retaddr); } +static void do_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, + MemOpIdx oi, uintptr_t ra) +{ + MMULookupLocals l; + bool crosspage; + uint64_t a, b; + int first; + + crosspage = mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE, &l); + if (likely(!crosspage)) { + /* Swap to host endian if necessary, then store. */ + if (l.memop & MO_BSWAP) { + val = bswap128(val); + } + if (unlikely(l.page[0].flags & TLB_MMIO)) { + QEMU_IOTHREAD_LOCK_GUARD(); + if (HOST_BIG_ENDIAN) { + b = int128_getlo(val), a = int128_gethi(val); + } else { + a = int128_getlo(val), b = int128_gethi(val); + } + io_writex(env, l.page[0].full, l.mmu_idx, a, addr, ra, MO_64); + io_writex(env, l.page[0].full, l.mmu_idx, b, addr + 8, ra, MO_64); + } else if (unlikely(l.page[0].flags & TLB_DISCARD_WRITE)) { + /* nothing */ + } else { + store_atom_16(env, ra, l.page[0].haddr, l.memop, val); + } + return; + } + + first = l.page[0].size; + if (first == 8) { + MemOp mop8 = (l.memop & ~(MO_SIZE | MO_BSWAP)) | MO_64; + + if (l.memop & MO_BSWAP) { + val = bswap128(val); + } + if (HOST_BIG_ENDIAN) { + b = int128_getlo(val), a = int128_gethi(val); + } else { + a = int128_getlo(val), b = int128_gethi(val); + } + do_st_8(env, &l.page[0], a, l.mmu_idx, mop8, ra); + do_st_8(env, &l.page[1], b, l.mmu_idx, mop8, ra); + return; + } + + if ((l.memop & MO_BSWAP) != MO_LE) { + val = bswap128(val); + } + if (first < 8) { + do_st_leN(env, &l.page[0], int128_getlo(val), l.mmu_idx, l.memop, ra); + val = int128_urshift(val, first * 8); + do_st16_leN(env, &l.page[1], val, l.mmu_idx, l.memop, ra); + } else { + b = do_st16_leN(env, &l.page[0], val, l.mmu_idx, l.memop, ra); + do_st_leN(env, &l.page[1], b, l.mmu_idx, l.memop, ra); + } +} + +void helper_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, + MemOpIdx oi, uintptr_t retaddr) +{ + tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_128); + do_st16_mmu(env, addr, val, oi, retaddr); +} + +void helper_st_i128(CPUArchState *env, target_ulong addr, Int128 val, + MemOpIdx oi) +{ + helper_st16_mmu(env, addr, val, oi, GETPC()); +} + /* * Store Helpers for cpu_ldst.h */ @@ -XXX,XX +XXX,XX @@ void cpu_stq_le_mmu(CPUArchState *env, target_ulong addr, uint64_t val, plugin_store_cb(env, addr, oi); } -void cpu_st16_be_mmu(CPUArchState *env, abi_ptr addr, Int128 val, - MemOpIdx oi, uintptr_t ra) +void cpu_st16_be_mmu(CPUArchState *env, target_ulong addr, Int128 val, + MemOpIdx oi, uintptr_t retaddr) { - MemOp mop = get_memop(oi); - int mmu_idx = get_mmuidx(oi); - MemOpIdx new_oi; - unsigned a_bits; - - tcg_debug_assert((mop & (MO_BSWAP|MO_SSIZE)) == (MO_BE|MO_128)); - a_bits = get_alignment_bits(mop); - - /* Handle CPU specific unaligned behaviour */ - if (addr & ((1 << a_bits) - 1)) { - cpu_unaligned_access(env_cpu(env), addr, MMU_DATA_STORE, - mmu_idx, ra); - } - - /* Construct an unaligned 64-bit replacement MemOpIdx. */ - mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; - new_oi = make_memop_idx(mop, mmu_idx); - - helper_stq_mmu(env, addr, int128_gethi(val), new_oi, ra); - helper_stq_mmu(env, addr + 8, int128_getlo(val), new_oi, ra); - - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP|MO_SIZE)) == (MO_BE|MO_128)); + do_st16_mmu(env, addr, val, oi, retaddr); + plugin_store_cb(env, addr, oi); } -void cpu_st16_le_mmu(CPUArchState *env, abi_ptr addr, Int128 val, - MemOpIdx oi, uintptr_t ra) +void cpu_st16_le_mmu(CPUArchState *env, target_ulong addr, Int128 val, + MemOpIdx oi, uintptr_t retaddr) { - MemOp mop = get_memop(oi); - int mmu_idx = get_mmuidx(oi); - MemOpIdx new_oi; - unsigned a_bits; - - tcg_debug_assert((mop & (MO_BSWAP|MO_SSIZE)) == (MO_LE|MO_128)); - a_bits = get_alignment_bits(mop); - - /* Handle CPU specific unaligned behaviour */ - if (addr & ((1 << a_bits) - 1)) { - cpu_unaligned_access(env_cpu(env), addr, MMU_DATA_STORE, - mmu_idx, ra); - } - - /* Construct an unaligned 64-bit replacement MemOpIdx. */ - mop = (mop & ~(MO_SIZE | MO_AMASK)) | MO_64 | MO_UNALN; - new_oi = make_memop_idx(mop, mmu_idx); - - helper_stq_mmu(env, addr, int128_getlo(val), new_oi, ra); - helper_stq_mmu(env, addr + 8, int128_gethi(val), new_oi, ra); - - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); + tcg_debug_assert((get_memop(oi) & (MO_BSWAP|MO_SIZE)) == (MO_LE|MO_128)); + do_st16_mmu(env, addr, val, oi, retaddr); + plugin_store_cb(env, addr, oi); } #include "ldst_common.c.inc" diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/user-exec.c +++ b/accel/tcg/user-exec.c @@ -XXX,XX +XXX,XX @@ uint64_t cpu_ldq_le_mmu(CPUArchState *env, abi_ptr addr, return cpu_to_le64(ret); } -Int128 cpu_ld16_be_mmu(CPUArchState *env, abi_ptr addr, - MemOpIdx oi, uintptr_t ra) +static Int128 do_ld16_he_mmu(CPUArchState *env, abi_ptr addr, + MemOp mop, uintptr_t ra) { void *haddr; Int128 ret; - tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_BE)); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - memcpy(&ret, haddr, 16); + tcg_debug_assert((mop & MO_SIZE) == MO_128); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_LOAD); + ret = load_atom_16(env, ra, haddr, mop); clear_helper_retaddr(); - qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); + return ret; +} +Int128 helper_ld16_mmu(CPUArchState *env, target_ulong addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + Int128 ret = do_ld16_he_mmu(env, addr, mop, ra); + + if (mop & MO_BSWAP) { + ret = bswap128(ret); + } + return ret; +} + +Int128 helper_ld_i128(CPUArchState *env, target_ulong addr, MemOpIdx oi) +{ + return helper_ld16_mmu(env, addr, oi, GETPC()); +} + +Int128 cpu_ld16_be_mmu(CPUArchState *env, abi_ptr addr, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + Int128 ret; + + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); + ret = do_ld16_he_mmu(env, addr, mop, ra); + qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); if (!HOST_BIG_ENDIAN) { ret = bswap128(ret); } @@ -XXX,XX +XXX,XX @@ Int128 cpu_ld16_be_mmu(CPUArchState *env, abi_ptr addr, Int128 cpu_ld16_le_mmu(CPUArchState *env, abi_ptr addr, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); Int128 ret; - tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_LE)); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_LOAD); - memcpy(&ret, haddr, 16); - clear_helper_retaddr(); + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); + ret = do_ld16_he_mmu(env, addr, mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); - if (HOST_BIG_ENDIAN) { ret = bswap128(ret); } @@ -XXX,XX +XXX,XX @@ void cpu_stq_le_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } -void cpu_st16_be_mmu(CPUArchState *env, abi_ptr addr, - Int128 val, MemOpIdx oi, uintptr_t ra) +static void do_st16_he_mmu(CPUArchState *env, abi_ptr addr, Int128 val, + MemOp mop, uintptr_t ra) { void *haddr; - tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_BE)); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); + tcg_debug_assert((mop & MO_SIZE) == MO_128); + haddr = cpu_mmu_lookup(env, addr, mop, ra, MMU_DATA_STORE); + store_atom_16(env, ra, haddr, mop, val); + clear_helper_retaddr(); +} + +void helper_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, + MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + if (mop & MO_BSWAP) { + val = bswap128(val); + } + do_st16_he_mmu(env, addr, val, mop, ra); +} + +void helper_st_i128(CPUArchState *env, target_ulong addr, + Int128 val, MemOpIdx oi) +{ + helper_st16_mmu(env, addr, val, oi, GETPC()); +} + +void cpu_st16_be_mmu(CPUArchState *env, abi_ptr addr, + Int128 val, MemOpIdx oi, uintptr_t ra) +{ + MemOp mop = get_memop(oi); + + tcg_debug_assert((mop & MO_BSWAP) == MO_BE); if (!HOST_BIG_ENDIAN) { val = bswap128(val); } - memcpy(haddr, &val, 16); - clear_helper_retaddr(); + do_st16_he_mmu(env, addr, val, mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } void cpu_st16_le_mmu(CPUArchState *env, abi_ptr addr, Int128 val, MemOpIdx oi, uintptr_t ra) { - void *haddr; + MemOp mop = get_memop(oi); - tcg_debug_assert((get_memop(oi) & (MO_BSWAP | MO_SIZE)) == (MO_128 | MO_LE)); - haddr = cpu_mmu_lookup(env, addr, oi, ra, MMU_DATA_STORE); + tcg_debug_assert((mop & MO_BSWAP) == MO_LE); if (HOST_BIG_ENDIAN) { val = bswap128(val); } - memcpy(haddr, &val, 16); - clear_helper_retaddr(); + do_st16_he_mmu(env, addr, val, mop, ra); qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); } diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op.c +++ b/tcg/tcg-op.c @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) } } +/* + * Return true if @mop, without knowledge of the pointer alignment, + * does not require 16-byte atomicity, and it would be adventagous + * to avoid a call to a helper function. + */ +static bool use_two_i64_for_i128(MemOp mop) +{ +#ifdef CONFIG_SOFTMMU + /* Two softmmu tlb lookups is larger than one function call. */ + return false; +#else + /* + * For user-only, two 64-bit operations may well be smaller than a call. + * Determine if that would be legal for the requested atomicity. + */ + switch (mop & MO_ATOM_MASK) { + case MO_ATOM_NONE: + case MO_ATOM_IFALIGN_PAIR: + return true; + case MO_ATOM_IFALIGN: + case MO_ATOM_SUBALIGN: + case MO_ATOM_WITHIN16: + case MO_ATOM_WITHIN16_PAIR: + /* In a serialized context, no atomicity is required. */ + return !(tcg_ctx->gen_tb->cflags & CF_PARALLEL); + default: + g_assert_not_reached(); + } +#endif +} + static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) { MemOp mop_1 = orig, mop_2; @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) ret[1] = mop_2; } +#if TARGET_LONG_BITS == 64 +#define tcg_temp_ebb_new tcg_temp_ebb_new_i64 +#else +#define tcg_temp_ebb_new tcg_temp_ebb_new_i32 +#endif + void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) { - MemOp mop[2]; - TCGv addr_p8; - TCGv_i64 x, y; + MemOpIdx oi = make_memop_idx(memop, idx); - canonicalize_memop_i128_as_i64(mop, memop); + tcg_debug_assert((memop & MO_SIZE) == MO_128); + tcg_debug_assert((memop & MO_SIGN) == 0); tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); addr = plugin_prep_mem_callbacks(addr); - /* TODO: respect atomicity of the operation. */ /* TODO: allow the tcg backend to see the whole operation. */ - /* - * Since there are no global TCGv_i128, there is no visible state - * changed if the second load faults. Load directly into the two - * subwords. - */ - if ((memop & MO_BSWAP) == MO_LE) { - x = TCGV128_LOW(val); - y = TCGV128_HIGH(val); + if (use_two_i64_for_i128(memop)) { + MemOp mop[2]; + TCGv addr_p8; + TCGv_i64 x, y; + + canonicalize_memop_i128_as_i64(mop, memop); + + /* + * Since there are no global TCGv_i128, there is no visible state + * changed if the second load faults. Load directly into the two + * subwords. + */ + if ((memop & MO_BSWAP) == MO_LE) { + x = TCGV128_LOW(val); + y = TCGV128_HIGH(val); + } else { + x = TCGV128_HIGH(val); + y = TCGV128_LOW(val); + } + + gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, mop[0], idx); + + if ((mop[0] ^ memop) & MO_BSWAP) { + tcg_gen_bswap64_i64(x, x); + } + + addr_p8 = tcg_temp_ebb_new(); + tcg_gen_addi_tl(addr_p8, addr, 8); + gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, mop[1], idx); + tcg_temp_free(addr_p8); + + if ((mop[0] ^ memop) & MO_BSWAP) { + tcg_gen_bswap64_i64(y, y); + } } else { - x = TCGV128_HIGH(val); - y = TCGV128_LOW(val); + gen_helper_ld_i128(val, cpu_env, addr, tcg_constant_i32(oi)); } - gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, mop[0], idx); - - if ((mop[0] ^ memop) & MO_BSWAP) { - tcg_gen_bswap64_i64(x, x); - } - - addr_p8 = tcg_temp_new(); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, mop[1], idx); - tcg_temp_free(addr_p8); - - if ((mop[0] ^ memop) & MO_BSWAP) { - tcg_gen_bswap64_i64(y, y); - } - - plugin_gen_mem_callbacks(addr, make_memop_idx(memop, idx), - QEMU_PLUGIN_MEM_R); + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); } void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) { - MemOp mop[2]; - TCGv addr_p8; - TCGv_i64 x, y; + MemOpIdx oi = make_memop_idx(memop, idx); - canonicalize_memop_i128_as_i64(mop, memop); + tcg_debug_assert((memop & MO_SIZE) == MO_128); + tcg_debug_assert((memop & MO_SIGN) == 0); tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); addr = plugin_prep_mem_callbacks(addr); - /* TODO: respect atomicity of the operation. */ /* TODO: allow the tcg backend to see the whole operation. */ - if ((memop & MO_BSWAP) == MO_LE) { - x = TCGV128_LOW(val); - y = TCGV128_HIGH(val); + if (use_two_i64_for_i128(memop)) { + MemOp mop[2]; + TCGv addr_p8; + TCGv_i64 x, y; + + canonicalize_memop_i128_as_i64(mop, memop); + + if ((memop & MO_BSWAP) == MO_LE) { + x = TCGV128_LOW(val); + y = TCGV128_HIGH(val); + } else { + x = TCGV128_HIGH(val); + y = TCGV128_LOW(val); + } + + addr_p8 = tcg_temp_ebb_new(); + if ((mop[0] ^ memop) & MO_BSWAP) { + TCGv_i64 t = tcg_temp_ebb_new_i64(); + + tcg_gen_bswap64_i64(t, x); + gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr, mop[0], idx); + tcg_gen_bswap64_i64(t, y); + tcg_gen_addi_tl(addr_p8, addr, 8); + gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr_p8, mop[1], idx); + tcg_temp_free_i64(t); + } else { + gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, mop[0], idx); + tcg_gen_addi_tl(addr_p8, addr, 8); + gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, mop[1], idx); + } + tcg_temp_free(addr_p8); } else { - x = TCGV128_HIGH(val); - y = TCGV128_LOW(val); + gen_helper_st_i128(cpu_env, addr, val, tcg_constant_i32(oi)); } - addr_p8 = tcg_temp_new(); - if ((mop[0] ^ memop) & MO_BSWAP) { - TCGv_i64 t = tcg_temp_ebb_new_i64(); - - tcg_gen_bswap64_i64(t, x); - gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr, mop[0], idx); - tcg_gen_bswap64_i64(t, y); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr_p8, mop[1], idx); - tcg_temp_free_i64(t); - } else { - gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, mop[0], idx); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, mop[1], idx); - } - tcg_temp_free(addr_p8); - - plugin_gen_mem_callbacks(addr, make_memop_idx(memop, idx), - QEMU_PLUGIN_MEM_W); + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); } static void tcg_gen_ext_i32(TCGv_i32 ret, TCGv_i32 val, MemOp opc) diff --git a/accel/tcg/ldst_atomicity.c.inc b/accel/tcg/ldst_atomicity.c.inc index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/ldst_atomicity.c.inc +++ b/accel/tcg/ldst_atomicity.c.inc @@ -XXX,XX +XXX,XX @@ static inline uint64_t load_atom_8_by_4(void *pv) } } +/** + * load_atom_8_by_8_or_4: + * @pv: host address + * + * Load 8 bytes from aligned @pv, with at least 4-byte atomicity. + */ +static inline uint64_t load_atom_8_by_8_or_4(void *pv) +{ + if (HAVE_al8_fast) { + return load_atomic8(pv); + } else { + return load_atom_8_by_4(pv); + } +} + /** * load_atom_2: * @p: host address @@ -XXX,XX +XXX,XX @@ static uint64_t load_atom_8(CPUArchState *env, uintptr_t ra, } } +/** + * load_atom_16: + * @p: host address + * @memop: the full memory op + * + * Load 16 bytes from @p, honoring the atomicity of @memop. + */ +static Int128 load_atom_16(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop) +{ + uintptr_t pi = (uintptr_t)pv; + int atmax; + Int128 r; + uint64_t a, b; + + /* + * If the host does not support 16-byte atomics, wait until we have + * examined the atomicity parameters below. + */ + if (HAVE_al16_fast && likely((pi & 15) == 0)) { + return load_atomic16(pv); + } + + atmax = required_atomicity(env, pi, memop); + switch (atmax) { + case MO_8: + memcpy(&r, pv, 16); + return r; + case MO_16: + a = load_atom_8_by_2(pv); + b = load_atom_8_by_2(pv + 8); + break; + case MO_32: + a = load_atom_8_by_4(pv); + b = load_atom_8_by_4(pv + 8); + break; + case MO_64: + if (!HAVE_al8) { + cpu_loop_exit_atomic(env_cpu(env), ra); + } + a = load_atomic8(pv); + b = load_atomic8(pv + 8); + break; + case -MO_64: + if (!HAVE_al8) { + cpu_loop_exit_atomic(env_cpu(env), ra); + } + a = load_atom_extract_al8x2(pv); + b = load_atom_extract_al8x2(pv + 8); + break; + case MO_128: + return load_atomic16_or_exit(env, ra, pv); + default: + g_assert_not_reached(); + } + return int128_make128(HOST_BIG_ENDIAN ? b : a, HOST_BIG_ENDIAN ? a : b); +} + /** * store_atomic2: * @pv: host address @@ -XXX,XX +XXX,XX @@ static inline void store_atomic8(void *pv, uint64_t val) qatomic_set__nocheck(p, val); } +/** + * store_atomic16: + * @pv: host address + * @val: value to store + * + * Atomically store 16 aligned bytes to @pv. + */ +static inline void store_atomic16(void *pv, Int128Alias val) +{ +#if defined(CONFIG_ATOMIC128) + __uint128_t *pu = __builtin_assume_aligned(pv, 16); + qatomic_set__nocheck(pu, val.u); +#elif defined(CONFIG_CMPXCHG128) + __uint128_t *pu = __builtin_assume_aligned(pv, 16); + __uint128_t o; + + /* + * Without CONFIG_ATOMIC128, __atomic_compare_exchange_n will always + * defer to libatomic, so we must use __sync_*_compare_and_swap_16 + * and accept the sequential consistency that comes with it. + */ + do { + o = *pu; + } while (!__sync_bool_compare_and_swap_16(pu, o, val.u)); +#else + qemu_build_not_reached(); +#endif +} + /** * store_atom_4x2 */ @@ -XXX,XX +XXX,XX @@ static void store_atom_8(CPUArchState *env, uintptr_t ra, } cpu_loop_exit_atomic(env_cpu(env), ra); } + +/** + * store_atom_16: + * @p: host address + * @val: the value to store + * @memop: the full memory op + * + * Store 16 bytes to @p, honoring the atomicity of @memop. + */ +static void store_atom_16(CPUArchState *env, uintptr_t ra, + void *pv, MemOp memop, Int128 val) +{ + uintptr_t pi = (uintptr_t)pv; + uint64_t a, b; + int atmax; + + if (HAVE_al16_fast && likely((pi & 15) == 0)) { + store_atomic16(pv, val); + return; + } + + atmax = required_atomicity(env, pi, memop); + + a = HOST_BIG_ENDIAN ? int128_gethi(val) : int128_getlo(val); + b = HOST_BIG_ENDIAN ? int128_getlo(val) : int128_gethi(val); + switch (atmax) { + case MO_8: + memcpy(pv, &val, 16); + return; + case MO_16: + store_atom_8_by_2(pv, a); + store_atom_8_by_2(pv + 8, b); + return; + case MO_32: + store_atom_8_by_4(pv, a); + store_atom_8_by_4(pv + 8, b); + return; + case MO_64: + if (HAVE_al8) { + store_atomic8(pv, a); + store_atomic8(pv + 8, b); + return; + } + break; + case -MO_64: + if (HAVE_al16) { + uint64_t val_le; + int s2 = pi & 15; + int s1 = 16 - s2; + + if (HOST_BIG_ENDIAN) { + val = bswap128(val); + } + switch (s2) { + case 1 ... 7: + val_le = store_whole_le16(pv, s1, val); + store_bytes_leN(pv + s1, s2, val_le); + break; + case 9 ... 15: + store_bytes_leN(pv, s1, int128_getlo(val)); + val = int128_urshift(val, s1 * 8); + store_whole_le16(pv + s1, s2, val); + break; + case 0: /* aligned */ + case 8: /* atmax MO_64 */ + default: + g_assert_not_reached(); + } + return; + } + break; + case MO_128: + if (HAVE_al16) { + store_atomic16(pv, val); + return; + } + break; + default: + g_assert_not_reached(); + } + cpu_loop_exit_atomic(env_cpu(env), ra); +} -- 2.34.1 There is an edge condition prior to gcc13 for which optimization is required to generate 16-byte atomic sequences. Detect this. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- meson.build | 52 ++++++++++++++++++++++------------ accel/tcg/ldst_atomicity.c.inc | 29 ++++++++++++++++--- 2 files changed, 59 insertions(+), 22 deletions(-) diff --git a/meson.build b/meson.build index XXXXXXX..XXXXXXX 100644 --- a/meson.build +++ b/meson.build @@ -XXX,XX +XXX,XX @@ config_host_data.set('HAVE_BROKEN_SIZE_MAX', not cc.compiles(''' return printf("%zu", SIZE_MAX); }''', args: ['-Werror'])) -atomic_test = ''' +# See if 64-bit atomic operations are supported. +# Note that without __atomic builtins, we can only +# assume atomic loads/stores max at pointer size. +config_host_data.set('CONFIG_ATOMIC64', cc.links(''' #include <stdint.h> int main(void) { - @0@ x = 0, y = 0; + uint64_t x = 0, y = 0; y = __atomic_load_n(&x, __ATOMIC_RELAXED); __atomic_store_n(&x, y, __ATOMIC_RELAXED); __atomic_compare_exchange_n(&x, &y, x, 0, __ATOMIC_RELAXED, __ATOMIC_RELAXED); __atomic_exchange_n(&x, y, __ATOMIC_RELAXED); __atomic_fetch_add(&x, y, __ATOMIC_RELAXED); return 0; - }''' - -# See if 64-bit atomic operations are supported. -# Note that without __atomic builtins, we can only -# assume atomic loads/stores max at pointer size. -config_host_data.set('CONFIG_ATOMIC64', cc.links(atomic_test.format('uint64_t'))) + }''')) has_int128 = cc.links(''' __int128_t a; @@ -XXX,XX +XXX,XX @@ if has_int128 # "do we have 128-bit atomics which are handled inline and specifically not # via libatomic". The reason we can't use libatomic is documented in the # comment starting "GCC is a house divided" in include/qemu/atomic128.h. - has_atomic128 = cc.links(atomic_test.format('unsigned __int128')) + # We only care about these operations on 16-byte aligned pointers, so + # force 16-byte alignment of the pointer, which may be greater than + # __alignof(unsigned __int128) for the host. + atomic_test_128 = ''' + int main(int ac, char **av) { + unsigned __int128 *p = __builtin_assume_aligned(av[ac - 1], sizeof(16)); + p[1] = __atomic_load_n(&p[0], __ATOMIC_RELAXED); + __atomic_store_n(&p[2], p[3], __ATOMIC_RELAXED); + __atomic_compare_exchange_n(&p[4], &p[5], p[6], 0, __ATOMIC_RELAXED, __ATOMIC_RELAXED); + return 0; + }''' + has_atomic128 = cc.links(atomic_test_128) config_host_data.set('CONFIG_ATOMIC128', has_atomic128) if not has_atomic128 - has_cmpxchg128 = cc.links(''' - int main(void) - { - unsigned __int128 x = 0, y = 0; - __sync_val_compare_and_swap_16(&x, y, x); - return 0; - } - ''') + # Even with __builtin_assume_aligned, the above test may have failed + # without optimization enabled. Try again with optimizations locally + # enabled for the function. See + # https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107389 + has_atomic128_opt = cc.links('__attribute__((optimize("O1")))' + atomic_test_128) + config_host_data.set('CONFIG_ATOMIC128_OPT', has_atomic128_opt) - config_host_data.set('CONFIG_CMPXCHG128', has_cmpxchg128) + if not has_atomic128_opt + config_host_data.set('CONFIG_CMPXCHG128', cc.links(''' + int main(void) + { + unsigned __int128 x = 0, y = 0; + __sync_val_compare_and_swap_16(&x, y, x); + return 0; + } + ''')) + endif endif endif diff --git a/accel/tcg/ldst_atomicity.c.inc b/accel/tcg/ldst_atomicity.c.inc index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/ldst_atomicity.c.inc +++ b/accel/tcg/ldst_atomicity.c.inc @@ -XXX,XX +XXX,XX @@ #endif #define HAVE_al8_fast (ATOMIC_REG_SIZE >= 8) +/* + * If __alignof(unsigned __int128) < 16, GCC may refuse to inline atomics + * that are supported by the host, e.g. s390x. We can force the pointer to + * have our known alignment with __builtin_assume_aligned, however prior to + * GCC 13 that was only reliable with optimization enabled. See + * https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107389 + */ +#if defined(CONFIG_ATOMIC128_OPT) +# if !defined(__OPTIMIZE__) +# define ATTRIBUTE_ATOMIC128_OPT __attribute__((optimize("O1"))) +# endif +# define CONFIG_ATOMIC128 +#endif +#ifndef ATTRIBUTE_ATOMIC128_OPT +# define ATTRIBUTE_ATOMIC128_OPT +#endif + #if defined(CONFIG_ATOMIC128) # define HAVE_al16_fast true #else @@ -XXX,XX +XXX,XX @@ static inline uint64_t load_atomic8(void *pv) * * Atomically load 16 aligned bytes from @pv. */ -static inline Int128 load_atomic16(void *pv) +static inline Int128 ATTRIBUTE_ATOMIC128_OPT +load_atomic16(void *pv) { #ifdef CONFIG_ATOMIC128 __uint128_t *p = __builtin_assume_aligned(pv, 16); @@ -XXX,XX +XXX,XX @@ static uint64_t load_atom_extract_al16_or_exit(CPUArchState *env, uintptr_t ra, * cross an 16-byte boundary then the access must be 16-byte atomic, * otherwise the access must be 8-byte atomic. */ -static inline uint64_t load_atom_extract_al16_or_al8(void *pv, int s) +static inline uint64_t ATTRIBUTE_ATOMIC128_OPT +load_atom_extract_al16_or_al8(void *pv, int s) { #if defined(CONFIG_ATOMIC128) uintptr_t pi = (uintptr_t)pv; @@ -XXX,XX +XXX,XX @@ static inline void store_atomic8(void *pv, uint64_t val) * * Atomically store 16 aligned bytes to @pv. */ -static inline void store_atomic16(void *pv, Int128Alias val) +static inline void ATTRIBUTE_ATOMIC128_OPT +store_atomic16(void *pv, Int128Alias val) { #if defined(CONFIG_ATOMIC128) __uint128_t *pu = __builtin_assume_aligned(pv, 16); @@ -XXX,XX +XXX,XX @@ static void store_atom_insert_al8(uint64_t *p, uint64_t val, uint64_t msk) * * Atomically store @val to @p masked by @msk. */ -static void store_atom_insert_al16(Int128 *ps, Int128Alias val, Int128Alias msk) +static void ATTRIBUTE_ATOMIC128_OPT +store_atom_insert_al16(Int128 *ps, Int128Alias val, Int128Alias msk) { #if defined(CONFIG_ATOMIC128) __uint128_t *pu, old, new; -- 2.34.1 Notice when Intel or AMD have guaranteed that vmovdqa is atomic. The new variable will also be used in generated code. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/qemu/cpuid.h | 18 ++++++++++++++++++ tcg/i386/tcg-target.h | 1 + tcg/i386/tcg-target.c.inc | 27 +++++++++++++++++++++++++++ 3 files changed, 46 insertions(+) diff --git a/include/qemu/cpuid.h b/include/qemu/cpuid.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/cpuid.h +++ b/include/qemu/cpuid.h @@ -XXX,XX +XXX,XX @@ #define bit_LZCNT (1 << 5) #endif +/* + * Signatures for different CPU implementations as returned from Leaf 0. + */ + +#ifndef signature_INTEL_ecx +/* "Genu" "ineI" "ntel" */ +#define signature_INTEL_ebx 0x756e6547 +#define signature_INTEL_edx 0x49656e69 +#define signature_INTEL_ecx 0x6c65746e +#endif + +#ifndef signature_AMD_ecx +/* "Auth" "enti" "cAMD" */ +#define signature_AMD_ebx 0x68747541 +#define signature_AMD_edx 0x69746e65 +#define signature_AMD_ecx 0x444d4163 +#endif + static inline unsigned xgetbv_low(unsigned c) { unsigned a, d; diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.h +++ b/tcg/i386/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_avx512dq; extern bool have_avx512vbmi2; extern bool have_avx512vl; extern bool have_movbe; +extern bool have_atomic16; /* optional instructions */ #define TCG_TARGET_HAS_div2_i32 1 diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ bool have_avx512dq; bool have_avx512vbmi2; bool have_avx512vl; bool have_movbe; +bool have_atomic16; #ifdef CONFIG_CPUID_H static bool have_bmi2; @@ -XXX,XX +XXX,XX @@ static void tcg_target_init(TCGContext *s) have_avx512dq = (b7 & bit_AVX512DQ) != 0; have_avx512vbmi2 = (c7 & bit_AVX512VBMI2) != 0; } + + /* + * The Intel SDM has added: + * Processors that enumerate support for Intel® AVX + * (by setting the feature flag CPUID.01H:ECX.AVX[bit 28]) + * guarantee that the 16-byte memory operations performed + * by the following instructions will always be carried + * out atomically: + * - MOVAPD, MOVAPS, and MOVDQA. + * - VMOVAPD, VMOVAPS, and VMOVDQA when encoded with VEX.128. + * - VMOVAPD, VMOVAPS, VMOVDQA32, and VMOVDQA64 when encoded + * with EVEX.128 and k0 (masking disabled). + * Note that these instructions require the linear addresses + * of their memory operands to be 16-byte aligned. + * + * AMD has provided an even stronger guarantee that processors + * with AVX provide 16-byte atomicity for all cachable, + * naturally aligned single loads and stores, e.g. MOVDQU. + * + * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104688 + */ + if (have_avx1) { + __cpuid(0, a, b, c, d); + have_atomic16 = (c == signature_INTEL_ecx || + c == signature_AMD_ecx); + } } } } -- 2.34.1 Notice when the host has additional atomic instructions. The new variables will also be used in generated code. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.h | 3 +++ tcg/aarch64/tcg-target.c.inc | 12 ++++++++++++ 2 files changed, 15 insertions(+) diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.h +++ b/tcg/aarch64/tcg-target.h @@ -XXX,XX +XXX,XX @@ typedef enum { #define TCG_TARGET_CALL_ARG_I128 TCG_CALL_ARG_EVEN #define TCG_TARGET_CALL_RET_I128 TCG_CALL_RET_NORMAL +extern bool have_lse; +extern bool have_lse2; + /* optional instructions */ #define TCG_TARGET_HAS_div_i32 1 #define TCG_TARGET_HAS_rem_i32 1 diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ #include "../tcg-ldst.c.inc" #include "../tcg-pool.c.inc" #include "qemu/bitops.h" +#ifdef __linux__ +#include <asm/hwcap.h> +#endif /* We're going to re-use TCGType in setting of the SF bit, which controls the size of the operation performed. If we know the values match, it @@ -XXX,XX +XXX,XX @@ static TCGReg tcg_target_call_oarg_reg(TCGCallReturnKind kind, int slot) return TCG_REG_X0 + slot; } +bool have_lse; +bool have_lse2; + #define TCG_REG_TMP TCG_REG_X30 #define TCG_VEC_TMP TCG_REG_V31 @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) static void tcg_target_init(TCGContext *s) { +#ifdef __linux__ + unsigned long hwcap = qemu_getauxval(AT_HWCAP); + have_lse = hwcap & HWCAP_ATOMICS; + have_lse2 = hwcap & HWCAP_USCAT; +#endif + tcg_target_available_regs[TCG_TYPE_I32] = 0xffffffffu; tcg_target_available_regs[TCG_TYPE_I64] = 0xffffffffu; tcg_target_available_regs[TCG_TYPE_V64] = 0xffffffff00000000ull; -- 2.34.1 These features are present for Apple M1. Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.c.inc | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ #ifdef __linux__ #include <asm/hwcap.h> #endif +#ifdef CONFIG_DARWIN +#include <sys/sysctl.h> +#endif /* We're going to re-use TCGType in setting of the SF bit, which controls the size of the operation performed. If we know the values match, it @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) } } +#ifdef CONFIG_DARWIN +static bool sysctl_for_bool(const char *name) +{ + int val = 0; + size_t len = sizeof(val); + + if (sysctlbyname(name, &val, &len, NULL, 0) == 0) { + return val != 0; + } + + /* + * We might in the future ask for properties not present in older kernels, + * but we're only asking about static properties, all of which should be + * 'int'. So we shouln't see ENOMEM (val too small), or any of the other + * more exotic errors. + */ + assert(errno == ENOENT); + return false; +} +#endif + static void tcg_target_init(TCGContext *s) { #ifdef __linux__ @@ -XXX,XX +XXX,XX @@ static void tcg_target_init(TCGContext *s) have_lse = hwcap & HWCAP_ATOMICS; have_lse2 = hwcap & HWCAP_USCAT; #endif +#ifdef CONFIG_DARWIN + have_lse = sysctl_for_bool("hw.optional.arm.FEAT_LSE"); + have_lse2 = sysctl_for_bool("hw.optional.arm.FEAT_LSE2"); +#endif tcg_target_available_regs[TCG_TYPE_I32] = 0xffffffffu; tcg_target_available_regs[TCG_TYPE_I64] = 0xffffffffu; -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 52 +++------------------------------------ 1 file changed, 4 insertions(+), 48 deletions(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { int seg; } HostAddress; -#if defined(CONFIG_SOFTMMU) /* * Because i686 has no register parameters and because x86_64 has xchg * to handle addr/data register overlap, we have placed all input arguments @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) /* resolve label address */ tcg_patch32(label_ptr[0], s->code_ptr - label_ptr[0] - 4); - if (TARGET_LONG_BITS > TCG_TARGET_REG_BITS) { + if (label_ptr[1]) { tcg_patch32(label_ptr[1], s->code_ptr - label_ptr[1] - 4); } @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) /* resolve label address */ tcg_patch32(label_ptr[0], s->code_ptr - label_ptr[0] - 4); - if (TARGET_LONG_BITS > TCG_TARGET_REG_BITS) { + if (label_ptr[1]) { tcg_patch32(label_ptr[1], s->code_ptr - label_ptr[1] - 4); } @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) tcg_out_jmp(s, l->raddr); return true; } -#else -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - /* resolve label address */ - tcg_patch32(l->label_ptr[0], s->code_ptr - l->label_ptr[0] - 4); - - if (TCG_TARGET_REG_BITS == 32) { - int ofs = 0; - - tcg_out_st(s, TCG_TYPE_PTR, TCG_AREG0, TCG_REG_ESP, ofs); - ofs += 4; - - tcg_out_st(s, TCG_TYPE_I32, l->addrlo_reg, TCG_REG_ESP, ofs); - ofs += 4; - if (TARGET_LONG_BITS == 64) { - tcg_out_st(s, TCG_TYPE_I32, l->addrhi_reg, TCG_REG_ESP, ofs); - ofs += 4; - } - - tcg_out_pushi(s, (uintptr_t)l->raddr); - } else { - tcg_out_mov(s, TCG_TYPE_TL, tcg_target_call_iarg_regs[1], - l->addrlo_reg); - tcg_out_mov(s, TCG_TYPE_PTR, tcg_target_call_iarg_regs[0], TCG_AREG0); - - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_RAX, (uintptr_t)l->raddr); - tcg_out_push(s, TCG_REG_RAX); - } - - /* "Tail call" to the helper, with the return address back inline. */ - tcg_out_jmp(s, (const void *)(l->is_ld ? helper_unaligned_ld - : helper_unaligned_st)); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} +#ifndef CONFIG_SOFTMMU static HostAddress x86_guest_base = { .index = -1 }; @@ -XXX,XX +XXX,XX @@ static inline int setup_guest_base_seg(void) return 0; } #endif /* setup_guest_base_seg */ -#endif /* SOFTMMU */ +#endif /* !SOFTMMU */ /* * For softmmu, perform the TLB load and compare. -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.c.inc | 35 ----------------------------------- 1 file changed, 35 deletions(-) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGType index_ext; } HostAddress; -#ifdef CONFIG_SOFTMMU static const TCGLdstHelperParam ldst_helper_param = { .ntmp = 1, .tmp = { TCG_REG_TMP } }; @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) tcg_out_goto(s, lb->raddr); return true; } -#else -static void tcg_out_adr(TCGContext *s, TCGReg rd, const void *target) -{ - ptrdiff_t offset = tcg_pcrel_diff(s, target); - tcg_debug_assert(offset == sextract64(offset, 0, 21)); - tcg_out_insn(s, 3406, ADR, rd, offset); -} - -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - if (!reloc_pc19(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) { - return false; - } - - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_X1, l->addrlo_reg); - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_X0, TCG_AREG0); - - /* "Tail call" to the helper, with the return address back inline. */ - tcg_out_adr(s, TCG_REG_LR, l->raddr); - tcg_out_goto_long(s, (const void *)(l->is_ld ? helper_unaligned_ld - : helper_unaligned_st)); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} -#endif /* CONFIG_SOFTMMU */ /* * For softmmu, perform the TLB load and compare. -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/ppc/tcg-target.c.inc | 44 ---------------------------------------- 1 file changed, 44 deletions(-) diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static const uint32_t qemu_stx_opc[(MO_SIZE + MO_BSWAP) + 1] = { [MO_BSWAP | MO_UQ] = STDBRX, }; -#if defined (CONFIG_SOFTMMU) static TCGReg ldst_ra_gen(TCGContext *s, const TCGLabelQemuLdst *l, int arg) { if (arg < 0) { @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) tcg_out_b(s, 0, lb->raddr); return true; } -#else -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - if (!reloc_pc14(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) { - return false; - } - - if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) { - TCGReg arg = TCG_REG_R4; - - arg |= (TCG_TARGET_CALL_ARG_I64 == TCG_CALL_ARG_EVEN); - if (l->addrlo_reg != arg) { - tcg_out_mov(s, TCG_TYPE_I32, arg, l->addrhi_reg); - tcg_out_mov(s, TCG_TYPE_I32, arg + 1, l->addrlo_reg); - } else if (l->addrhi_reg != arg + 1) { - tcg_out_mov(s, TCG_TYPE_I32, arg + 1, l->addrlo_reg); - tcg_out_mov(s, TCG_TYPE_I32, arg, l->addrhi_reg); - } else { - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_R0, arg); - tcg_out_mov(s, TCG_TYPE_I32, arg, arg + 1); - tcg_out_mov(s, TCG_TYPE_I32, arg + 1, TCG_REG_R0); - } - } else { - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_R4, l->addrlo_reg); - } - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_R3, TCG_AREG0); - - /* "Tail call" to the helper, with the return address back inline. */ - tcg_out_call_int(s, 0, (const void *)(l->is_ld ? helper_unaligned_ld - : helper_unaligned_st)); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} -#endif /* SOFTMMU */ typedef struct { TCGReg base; -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/loongarch64/tcg-target.c.inc | 30 ------------------------------ 1 file changed, 30 deletions(-) diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, * Load/store helpers for SoftMMU, and qemu_ld/st implementations */ -#if defined(CONFIG_SOFTMMU) static bool tcg_out_goto(TCGContext *s, const tcg_insn_unit *target) { tcg_out_opc_b(s, 0); @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) tcg_out_call_int(s, qemu_st_helpers[opc & MO_SIZE], false); return tcg_out_goto(s, l->raddr); } -#else -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - /* resolve label address */ - if (!reloc_br_sk16(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) { - return false; - } - - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_A1, l->addrlo_reg); - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_A0, TCG_AREG0); - - /* tail call, with the return address back inline. */ - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_RA, (uintptr_t)l->raddr); - tcg_out_call_int(s, (const void *)(l->is_ld ? helper_unaligned_ld - : helper_unaligned_st), true); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -#endif /* CONFIG_SOFTMMU */ typedef struct { TCGReg base; -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/riscv/tcg-target.c.inc | 29 ----------------------------- 1 file changed, 29 deletions(-) diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_mb(TCGContext *s, TCGArg a0) * Load/store and TLB */ -#if defined(CONFIG_SOFTMMU) static void tcg_out_goto(TCGContext *s, const tcg_insn_unit *target) { tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, 0); @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) tcg_out_goto(s, l->raddr); return true; } -#else -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - /* resolve label address */ - if (!reloc_sbimm12(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) { - return false; - } - - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_A1, l->addrlo_reg); - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_A0, TCG_AREG0); - - /* tail call, with the return address back inline. */ - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_RA, (uintptr_t)l->raddr); - tcg_out_call_int(s, (const void *)(l->is_ld ? helper_unaligned_ld - : helper_unaligned_st), true); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} -#endif /* CONFIG_SOFTMMU */ /* * For softmmu, perform the TLB load and compare. -- 2.34.1 Always reserve r3 for tlb softmmu lookup. Fix a bug in user-only ALL_QLDST_REGS, in that r14 is clobbered by the BLNE that leads to the misaligned trap. Remove r0+r1 from user-only ALL_QLDST_REGS; I believe these had been reserved for bswap, which we no longer perform during qemu_st. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/arm/tcg-target-con-set.h | 16 ++++++++-------- tcg/arm/tcg-target-con-str.h | 5 ++--- tcg/arm/tcg-target.c.inc | 23 ++++++++--------------- 3 files changed, 18 insertions(+), 26 deletions(-) diff --git a/tcg/arm/tcg-target-con-set.h b/tcg/arm/tcg-target-con-set.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target-con-set.h +++ b/tcg/arm/tcg-target-con-set.h @@ -XXX,XX +XXX,XX @@ C_O0_I1(r) C_O0_I2(r, r) C_O0_I2(r, rIN) -C_O0_I2(s, s) +C_O0_I2(q, q) C_O0_I2(w, r) -C_O0_I3(s, s, s) -C_O0_I3(S, p, s) +C_O0_I3(q, q, q) +C_O0_I3(Q, p, q) C_O0_I4(r, r, rI, rI) -C_O0_I4(S, p, s, s) -C_O1_I1(r, l) +C_O0_I4(Q, p, q, q) +C_O1_I1(r, q) C_O1_I1(r, r) C_O1_I1(w, r) C_O1_I1(w, w) C_O1_I1(w, wr) C_O1_I2(r, 0, rZ) -C_O1_I2(r, l, l) +C_O1_I2(r, q, q) C_O1_I2(r, r, r) C_O1_I2(r, r, rI) C_O1_I2(r, r, rIK) @@ -XXX,XX +XXX,XX @@ C_O1_I2(w, w, wZ) C_O1_I3(w, w, w, w) C_O1_I4(r, r, r, rI, rI) C_O1_I4(r, r, rIN, rIK, 0) -C_O2_I1(e, p, l) -C_O2_I2(e, p, l, l) +C_O2_I1(e, p, q) +C_O2_I2(e, p, q, q) C_O2_I2(r, r, r, r) C_O2_I4(r, r, r, r, rIN, rIK) C_O2_I4(r, r, rI, rI, rIN, rIK) diff --git a/tcg/arm/tcg-target-con-str.h b/tcg/arm/tcg-target-con-str.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target-con-str.h +++ b/tcg/arm/tcg-target-con-str.h @@ -XXX,XX +XXX,XX @@ */ REGS('e', ALL_GENERAL_REGS & 0x5555) /* even regs */ REGS('r', ALL_GENERAL_REGS) -REGS('l', ALL_QLOAD_REGS) -REGS('s', ALL_QSTORE_REGS) -REGS('S', ALL_QSTORE_REGS & 0x5555) /* even qstore */ +REGS('q', ALL_QLDST_REGS) +REGS('Q', ALL_QLDST_REGS & 0x5555) /* even qldst */ REGS('w', ALL_VECTOR_REGS) /* diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type, #define ALL_VECTOR_REGS 0xffff0000u /* - * r0-r2 will be overwritten when reading the tlb entry (softmmu only) - * and r0-r1 doing the byte swapping, so don't use these. - * r3 is removed for softmmu to avoid clashes with helper arguments. + * r0-r3 will be overwritten when reading the tlb entry (softmmu only); + * r14 will be overwritten by the BLNE branching to the slow path. */ #ifdef CONFIG_SOFTMMU -#define ALL_QLOAD_REGS \ +#define ALL_QLDST_REGS \ (ALL_GENERAL_REGS & ~((1 << TCG_REG_R0) | (1 << TCG_REG_R1) | \ (1 << TCG_REG_R2) | (1 << TCG_REG_R3) | \ (1 << TCG_REG_R14))) -#define ALL_QSTORE_REGS \ - (ALL_GENERAL_REGS & ~((1 << TCG_REG_R0) | (1 << TCG_REG_R1) | \ - (1 << TCG_REG_R2) | (1 << TCG_REG_R14) | \ - ((TARGET_LONG_BITS == 64) << TCG_REG_R3))) #else -#define ALL_QLOAD_REGS ALL_GENERAL_REGS -#define ALL_QSTORE_REGS \ - (ALL_GENERAL_REGS & ~((1 << TCG_REG_R0) | (1 << TCG_REG_R1))) +#define ALL_QLDST_REGS (ALL_GENERAL_REGS & ~(1 << TCG_REG_R14)) #endif /* @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) return C_O1_I4(r, r, r, rI, rI); case INDEX_op_qemu_ld_i32: - return TARGET_LONG_BITS == 32 ? C_O1_I1(r, l) : C_O1_I2(r, l, l); + return TARGET_LONG_BITS == 32 ? C_O1_I1(r, q) : C_O1_I2(r, q, q); case INDEX_op_qemu_ld_i64: - return TARGET_LONG_BITS == 32 ? C_O2_I1(e, p, l) : C_O2_I2(e, p, l, l); + return TARGET_LONG_BITS == 32 ? C_O2_I1(e, p, q) : C_O2_I2(e, p, q, q); case INDEX_op_qemu_st_i32: - return TARGET_LONG_BITS == 32 ? C_O0_I2(s, s) : C_O0_I3(s, s, s); + return TARGET_LONG_BITS == 32 ? C_O0_I2(q, q) : C_O0_I3(q, q, q); case INDEX_op_qemu_st_i64: - return TARGET_LONG_BITS == 32 ? C_O0_I3(S, p, s) : C_O0_I4(S, p, s, s); + return TARGET_LONG_BITS == 32 ? C_O0_I3(Q, p, q) : C_O0_I4(Q, p, q, q); case INDEX_op_st_vec: return C_O0_I2(w, r); -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/arm/tcg-target.c.inc | 45 ---------------------------------------- 1 file changed, 45 deletions(-) diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { bool index_scratch; } HostAddress; -#ifdef CONFIG_SOFTMMU static TCGReg ldst_ra_gen(TCGContext *s, const TCGLabelQemuLdst *l, int arg) { /* We arrive at the slow path via "BLNE", so R14 contains l->raddr. */ @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) tcg_out_goto(s, COND_AL, qemu_st_helpers[opc & MO_SIZE]); return true; } -#else -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - if (!reloc_pc24(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) { - return false; - } - - if (TARGET_LONG_BITS == 64) { - /* 64-bit target address is aligned into R2:R3. */ - TCGMovExtend ext[2] = { - { .dst = TCG_REG_R2, .dst_type = TCG_TYPE_I32, - .src = l->addrlo_reg, - .src_type = TCG_TYPE_I32, .src_ext = MO_UL }, - { .dst = TCG_REG_R3, .dst_type = TCG_TYPE_I32, - .src = l->addrhi_reg, - .src_type = TCG_TYPE_I32, .src_ext = MO_UL }, - }; - tcg_out_movext2(s, &ext[0], &ext[1], TCG_REG_TMP); - } else { - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_R1, l->addrlo_reg); - } - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_R0, TCG_AREG0); - - /* - * Tail call to the helper, with the return address back inline, - * just for the clarity of the debugging traceback -- the helper - * cannot return. We have used BLNE to arrive here, so LR is - * already set. - */ - tcg_out_goto(s, COND_AL, (const void *) - (l->is_ld ? helper_unaligned_ld : helper_unaligned_st)); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} -#endif /* SOFTMMU */ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCGReg addrlo, TCGReg addrhi, -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/mips/tcg-target.c.inc | 57 ++------------------------------------- 1 file changed, 2 insertions(+), 55 deletions(-) diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_call(TCGContext *s, const tcg_insn_unit *arg, tcg_out_nop(s); } -#if defined(CONFIG_SOFTMMU) /* We have four temps, we might as well expose three of them. */ static const TCGLdstHelperParam ldst_helper_param = { .ntmp = 3, .tmp = { TCG_TMP0, TCG_TMP1, TCG_TMP2 } @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) /* resolve label address */ if (!reloc_pc16(l->label_ptr[0], tgt_rx) - || (TCG_TARGET_REG_BITS < TARGET_LONG_BITS - && !reloc_pc16(l->label_ptr[1], tgt_rx))) { + || (l->label_ptr[1] && !reloc_pc16(l->label_ptr[1], tgt_rx))) { return false; } @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) /* resolve label address */ if (!reloc_pc16(l->label_ptr[0], tgt_rx) - || (TCG_TARGET_REG_BITS < TARGET_LONG_BITS - && !reloc_pc16(l->label_ptr[1], tgt_rx))) { + || (l->label_ptr[1] && !reloc_pc16(l->label_ptr[1], tgt_rx))) { return false; } @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) return true; } -#else -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - void *target; - - if (!reloc_pc16(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) { - return false; - } - - if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) { - /* A0 is env, A1 is skipped, A2:A3 is the uint64_t address. */ - TCGReg a2 = MIPS_BE ? l->addrhi_reg : l->addrlo_reg; - TCGReg a3 = MIPS_BE ? l->addrlo_reg : l->addrhi_reg; - - if (a3 != TCG_REG_A2) { - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_A2, a2); - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_A3, a3); - } else if (a2 != TCG_REG_A3) { - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_A3, a3); - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_A2, a2); - } else { - tcg_out_mov(s, TCG_TYPE_I32, TCG_TMP0, TCG_REG_A2); - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_A2, TCG_REG_A3); - tcg_out_mov(s, TCG_TYPE_I32, TCG_REG_A3, TCG_TMP0); - } - } else { - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_A1, l->addrlo_reg); - } - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_A0, TCG_AREG0); - - /* - * Tail call to the helper, with the return address back inline. - * We have arrived here via BNEL, so $31 is already set. - */ - target = (l->is_ld ? helper_unaligned_ld : helper_unaligned_st); - tcg_out_call_int(s, target, true); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} -#endif /* SOFTMMU */ - typedef struct { TCGReg base; MemOp align; -- 2.34.1 Instead of using helper_unaligned_{ld,st}, use the full load/store helpers. This will allow the fast path to increase alignment to implement atomicity while not immediately raising an alignment exception. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/s390x/tcg-target.c.inc | 29 ----------------------------- 1 file changed, 29 deletions(-) diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st_direct(TCGContext *s, MemOp opc, TCGReg data, } } -#if defined(CONFIG_SOFTMMU) static const TCGLdstHelperParam ldst_helper_param = { .ntmp = 1, .tmp = { TCG_TMP0 } }; @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) tgen_gotoi(s, S390_CC_ALWAYS, lb->raddr); return true; } -#else -static bool tcg_out_fail_alignment(TCGContext *s, TCGLabelQemuLdst *l) -{ - if (!patch_reloc(l->label_ptr[0], R_390_PC16DBL, - (intptr_t)tcg_splitwx_to_rx(s->code_ptr), 2)) { - return false; - } - - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_R3, l->addrlo_reg); - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_R2, TCG_AREG0); - - /* "Tail call" to the helper, with the return address back inline. */ - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R14, (uintptr_t)l->raddr); - tgen_gotoi(s, S390_CC_ALWAYS, (const void *)(l->is_ld ? helper_unaligned_ld - : helper_unaligned_st)); - return true; -} - -static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} - -static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) -{ - return tcg_out_fail_alignment(s, l); -} -#endif /* CONFIG_SOFTMMU */ /* * For softmmu, perform the TLB load and compare. -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/sparc64/tcg-target.c.inc | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static const char * const tcg_target_reg_names[TCG_TARGET_NB_REGS] = { #define ALL_GENERAL_REGS MAKE_64BIT_MASK(0, 32) #define ALL_QLDST_REGS (ALL_GENERAL_REGS & ~SOFTMMU_RESERVE_REGS) -/* Define some temporary registers. T2 is used for constant generation. */ +/* Define some temporary registers. T3 is used for constant generation. */ #define TCG_REG_T1 TCG_REG_G1 -#define TCG_REG_T2 TCG_REG_O7 +#define TCG_REG_T2 TCG_REG_G2 +#define TCG_REG_T3 TCG_REG_O7 #ifndef CONFIG_SOFTMMU # define TCG_GUEST_BASE_REG TCG_REG_I5 @@ -XXX,XX +XXX,XX @@ static const int tcg_target_reg_alloc_order[] = { TCG_REG_I4, TCG_REG_I5, - TCG_REG_G2, TCG_REG_G3, TCG_REG_G4, TCG_REG_G5, @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret, static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg ret, tcg_target_long arg) { - tcg_debug_assert(ret != TCG_REG_T2); - tcg_out_movi_int(s, type, ret, arg, false, TCG_REG_T2); + tcg_debug_assert(ret != TCG_REG_T3); + tcg_out_movi_int(s, type, ret, arg, false, TCG_REG_T3); } static void tcg_out_ext8s(TCGContext *s, TCGType type, TCGReg rd, TCGReg rs) @@ -XXX,XX +XXX,XX @@ static void tcg_out_jmpl_const(TCGContext *s, const tcg_insn_unit *dest, { uintptr_t desti = (uintptr_t)dest; - /* Be careful not to clobber %o7 for a tail call. */ tcg_out_movi_int(s, TCG_TYPE_PTR, TCG_REG_T1, - desti & ~0xfff, in_prologue, - tail_call ? TCG_REG_G2 : TCG_REG_O7); + desti & ~0xfff, in_prologue, TCG_REG_T2); tcg_out_arithi(s, tail_call ? TCG_REG_G0 : TCG_REG_O7, TCG_REG_T1, desti & 0xfff, JMPL); } @@ -XXX,XX +XXX,XX @@ static void tcg_target_init(TCGContext *s) tcg_regset_set_reg(s->reserved_regs, TCG_REG_O6); /* stack pointer */ tcg_regset_set_reg(s->reserved_regs, TCG_REG_T1); /* for internal use */ tcg_regset_set_reg(s->reserved_regs, TCG_REG_T2); /* for internal use */ + tcg_regset_set_reg(s->reserved_regs, TCG_REG_T3); /* for internal use */ } #define ELF_HOST_MACHINE EM_SPARCV9 -- 2.34.1 Emphasize that the constant is signed. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/sparc64/tcg-target.c.inc | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_sethi(TCGContext *s, TCGReg ret, uint32_t arg) tcg_out32(s, SETHI | INSN_RD(ret) | ((arg & 0xfffffc00) >> 10)); } -static void tcg_out_movi_imm13(TCGContext *s, TCGReg ret, int32_t arg) +/* A 13-bit constant sign-extended to 64 bits. */ +static void tcg_out_movi_s13(TCGContext *s, TCGReg ret, int32_t arg) { tcg_out_arithi(s, ret, TCG_REG_G0, arg, ARITH_OR); } @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_imm32(TCGContext *s, TCGReg ret, int32_t arg) { if (check_fit_i32(arg, 13)) { /* A 13-bit constant sign-extended to 64-bits. */ - tcg_out_movi_imm13(s, ret, arg); + tcg_out_movi_s13(s, ret, arg); } else { /* A 32-bit constant zero-extended to 64 bits. */ tcg_out_sethi(s, ret, arg); @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret, /* A 13-bit constant sign-extended to 64-bits. */ if (check_fit_tl(arg, 13)) { - tcg_out_movi_imm13(s, ret, arg); + tcg_out_movi_s13(s, ret, arg); return; } @@ -XXX,XX +XXX,XX @@ static void tcg_out_setcond_i32(TCGContext *s, TCGCond cond, TCGReg ret, default: tcg_out_cmp(s, c1, c2, c2const); - tcg_out_movi_imm13(s, ret, 0); + tcg_out_movi_s13(s, ret, 0); tcg_out_movcc(s, cond, MOVCC_ICC, ret, 1, 1); return; } @@ -XXX,XX +XXX,XX @@ static void tcg_out_setcond_i64(TCGContext *s, TCGCond cond, TCGReg ret, /* For 64-bit signed comparisons vs zero, we can avoid the compare if the input does not overlap the output. */ if (c2 == 0 && !is_unsigned_cond(cond) && c1 != ret) { - tcg_out_movi_imm13(s, ret, 0); + tcg_out_movi_s13(s, ret, 0); tcg_out_movr(s, cond, ret, c1, 1, 1); } else { tcg_out_cmp(s, c1, c2, c2const); - tcg_out_movi_imm13(s, ret, 0); + tcg_out_movi_s13(s, ret, 0); tcg_out_movcc(s, cond, MOVCC_XCC, ret, 1, 1); } } @@ -XXX,XX +XXX,XX @@ static void tcg_out_addsub2_i64(TCGContext *s, TCGReg rl, TCGReg rh, if (use_vis3_instructions && !is_sub) { /* Note that ADDXC doesn't accept immediates. */ if (bhconst && bh != 0) { - tcg_out_movi_imm13(s, TCG_REG_T2, bh); + tcg_out_movi_s13(s, TCG_REG_T2, bh); bh = TCG_REG_T2; } tcg_out_arith(s, rh, ah, bh, ARITH_ADDXC); @@ -XXX,XX +XXX,XX @@ static void tcg_out_addsub2_i64(TCGContext *s, TCGReg rl, TCGReg rh, * so the adjustment fits 12 bits. */ if (bhconst) { - tcg_out_movi_imm13(s, TCG_REG_T2, bh + (is_sub ? -1 : 1)); + tcg_out_movi_s13(s, TCG_REG_T2, bh + (is_sub ? -1 : 1)); } else { tcg_out_arithi(s, TCG_REG_T2, bh, 1, is_sub ? ARITH_SUB : ARITH_ADD); @@ -XXX,XX +XXX,XX @@ static void tcg_target_qemu_prologue(TCGContext *s) tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr); tcg_out_arithi(s, TCG_REG_G0, TCG_REG_I7, 8, RETURN); /* delay slot */ - tcg_out_movi_imm13(s, TCG_REG_O0, 0); + tcg_out_movi_s13(s, TCG_REG_O0, 0); build_trampolines(s); } @@ -XXX,XX +XXX,XX @@ static void tcg_out_exit_tb(TCGContext *s, uintptr_t a0) { if (check_fit_ptr(a0, 13)) { tcg_out_arithi(s, TCG_REG_G0, TCG_REG_I7, 8, RETURN); - tcg_out_movi_imm13(s, TCG_REG_O0, a0); + tcg_out_movi_s13(s, TCG_REG_O0, a0); return; } else { intptr_t tb_diff = tcg_tbrel_diff(s, (void *)a0); -- 2.34.1 Shuffle the order in tcg_out_movi_int to check s13 first, and drop this check from tcg_out_movi_imm32. This might make the sequence for in_prologue larger, but not worth worrying about. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/sparc64/tcg-target.c.inc | 25 ++++++++++--------------- 1 file changed, 10 insertions(+), 15 deletions(-) diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_s13(TCGContext *s, TCGReg ret, int32_t arg) static void tcg_out_movi_imm32(TCGContext *s, TCGReg ret, int32_t arg) { - if (check_fit_i32(arg, 13)) { - /* A 13-bit constant sign-extended to 64-bits. */ - tcg_out_movi_s13(s, ret, arg); - } else { - /* A 32-bit constant zero-extended to 64 bits. */ - tcg_out_sethi(s, ret, arg); - if (arg & 0x3ff) { - tcg_out_arithi(s, ret, ret, arg & 0x3ff, ARITH_OR); - } + /* A 32-bit constant zero-extended to 64 bits. */ + tcg_out_sethi(s, ret, arg); + if (arg & 0x3ff) { + tcg_out_arithi(s, ret, ret, arg & 0x3ff, ARITH_OR); } } @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret, tcg_target_long hi, lo = (int32_t)arg; tcg_target_long test, lsb; - /* A 32-bit constant, or 32-bit zero-extended to 64-bits. */ - if (type == TCG_TYPE_I32 || arg == (uint32_t)arg) { - tcg_out_movi_imm32(s, ret, arg); - return; - } - /* A 13-bit constant sign-extended to 64-bits. */ if (check_fit_tl(arg, 13)) { tcg_out_movi_s13(s, ret, arg); return; } + /* A 32-bit constant, or 32-bit zero-extended to 64-bits. */ + if (type == TCG_TYPE_I32 || arg == (uint32_t)arg) { + tcg_out_movi_imm32(s, ret, arg); + return; + } + /* A 13-bit constant relative to the TB. */ if (!in_prologue) { test = tcg_tbrel_diff(s, (void *)arg); -- 2.34.1 Emphasize that the constant is unsigned. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/sparc64/tcg-target.c.inc | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_s13(TCGContext *s, TCGReg ret, int32_t arg) tcg_out_arithi(s, ret, TCG_REG_G0, arg, ARITH_OR); } -static void tcg_out_movi_imm32(TCGContext *s, TCGReg ret, int32_t arg) +/* A 32-bit constant zero-extended to 64 bits. */ +static void tcg_out_movi_u32(TCGContext *s, TCGReg ret, uint32_t arg) { - /* A 32-bit constant zero-extended to 64 bits. */ tcg_out_sethi(s, ret, arg); if (arg & 0x3ff) { tcg_out_arithi(s, ret, ret, arg & 0x3ff, ARITH_OR); @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret, /* A 32-bit constant, or 32-bit zero-extended to 64-bits. */ if (type == TCG_TYPE_I32 || arg == (uint32_t)arg) { - tcg_out_movi_imm32(s, ret, arg); + tcg_out_movi_u32(s, ret, arg); return; } @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret, /* A 64-bit constant decomposed into 2 32-bit pieces. */ if (check_fit_i32(lo, 13)) { hi = (arg - lo) >> 32; - tcg_out_movi_imm32(s, ret, hi); + tcg_out_movi_u32(s, ret, hi); tcg_out_arithi(s, ret, ret, 32, SHIFT_SLLX); tcg_out_arithi(s, ret, ret, lo, ARITH_ADD); } else { hi = arg >> 32; - tcg_out_movi_imm32(s, ret, hi); - tcg_out_movi_imm32(s, scratch, lo); + tcg_out_movi_u32(s, ret, hi); + tcg_out_movi_u32(s, scratch, lo); tcg_out_arithi(s, ret, ret, 32, SHIFT_SLLX); tcg_out_arith(s, ret, ret, scratch, ARITH_OR); } -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/sparc64/tcg-target.c.inc | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_s13(TCGContext *s, TCGReg ret, int32_t arg) tcg_out_arithi(s, ret, TCG_REG_G0, arg, ARITH_OR); } +/* A 32-bit constant sign-extended to 64 bits. */ +static void tcg_out_movi_s32(TCGContext *s, TCGReg ret, int32_t arg) +{ + tcg_out_sethi(s, ret, ~arg); + tcg_out_arithi(s, ret, ret, (arg & 0x3ff) | -0x400, ARITH_XOR); +} + /* A 32-bit constant zero-extended to 64 bits. */ static void tcg_out_movi_u32(TCGContext *s, TCGReg ret, uint32_t arg) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret, /* A 32-bit constant sign-extended to 64-bits. */ if (arg == lo) { - tcg_out_sethi(s, ret, ~arg); - tcg_out_arithi(s, ret, ret, (arg & 0x3ff) | -0x400, ARITH_XOR); + tcg_out_movi_s32(s, ret, arg); return; } -- 2.34.1 Drop the target-specific trampolines for the standard slow path. This lets us use tcg_out_helper_{ld,st}_args, and handles the new atomicity bits within MemOp. At the same time, use the full load/store helpers for user-only mode. Drop inline unaligned access support for user-only mode, as it does not handle atomicity. Use TCG_REG_T[1-3] in the tlb lookup, instead of TCG_REG_O[0-2]. This allows the constraints to be simplified. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/sparc64/tcg-target-con-set.h | 2 - tcg/sparc64/tcg-target-con-str.h | 1 - tcg/sparc64/tcg-target.h | 1 + tcg/sparc64/tcg-target.c.inc | 610 +++++++++---------------------- 4 files changed, 182 insertions(+), 432 deletions(-) diff --git a/tcg/sparc64/tcg-target-con-set.h b/tcg/sparc64/tcg-target-con-set.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target-con-set.h +++ b/tcg/sparc64/tcg-target-con-set.h @@ -XXX,XX +XXX,XX @@ C_O0_I1(r) C_O0_I2(rZ, r) C_O0_I2(rZ, rJ) -C_O0_I2(sZ, s) -C_O1_I1(r, s) C_O1_I1(r, r) C_O1_I2(r, r, r) C_O1_I2(r, rZ, rJ) diff --git a/tcg/sparc64/tcg-target-con-str.h b/tcg/sparc64/tcg-target-con-str.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target-con-str.h +++ b/tcg/sparc64/tcg-target-con-str.h @@ -XXX,XX +XXX,XX @@ * REGS(letter, register_mask) */ REGS('r', ALL_GENERAL_REGS) -REGS('s', ALL_QLDST_REGS) /* * Define constraint letters for constants: diff --git a/tcg/sparc64/tcg-target.h b/tcg/sparc64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.h +++ b/tcg/sparc64/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool use_vis3_instructions; #define TCG_TARGET_DEFAULT_MO (0) #define TCG_TARGET_HAS_MEMORY_BSWAP 1 +#define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS #endif diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ #error "unsupported code generation mode" #endif +#include "../tcg-ldst.c.inc" #include "../tcg-pool.c.inc" #ifdef CONFIG_DEBUG_TCG @@ -XXX,XX +XXX,XX @@ static const char * const tcg_target_reg_names[TCG_TARGET_NB_REGS] = { #define TCG_CT_CONST_S13 0x200 #define TCG_CT_CONST_ZERO 0x400 -/* - * For softmmu, we need to avoid conflicts with the first 3 - * argument registers to perform the tlb lookup, and to call - * the helper function. - */ -#ifdef CONFIG_SOFTMMU -#define SOFTMMU_RESERVE_REGS MAKE_64BIT_MASK(TCG_REG_O0, 3) -#else -#define SOFTMMU_RESERVE_REGS 0 -#endif -#define ALL_GENERAL_REGS MAKE_64BIT_MASK(0, 32) -#define ALL_QLDST_REGS (ALL_GENERAL_REGS & ~SOFTMMU_RESERVE_REGS) +#define ALL_GENERAL_REGS MAKE_64BIT_MASK(0, 32) /* Define some temporary registers. T3 is used for constant generation. */ #define TCG_REG_T1 TCG_REG_G1 @@ -XXX,XX +XXX,XX @@ static void tcg_out_mb(TCGContext *s, TCGArg a0) tcg_out32(s, MEMBAR | (a0 & TCG_MO_ALL)); } -#ifdef CONFIG_SOFTMMU -static const tcg_insn_unit *qemu_ld_trampoline[MO_SSIZE + 1]; -static const tcg_insn_unit *qemu_st_trampoline[MO_SIZE + 1]; - -static void build_trampolines(TCGContext *s) -{ - int i; - - for (i = 0; i < ARRAY_SIZE(qemu_ld_helpers); ++i) { - if (qemu_ld_helpers[i] == NULL) { - continue; - } - - /* May as well align the trampoline. */ - while ((uintptr_t)s->code_ptr & 15) { - tcg_out_nop(s); - } - qemu_ld_trampoline[i] = tcg_splitwx_to_rx(s->code_ptr); - - /* Set the retaddr operand. */ - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_O3, TCG_REG_O7); - /* Tail call. */ - tcg_out_jmpl_const(s, qemu_ld_helpers[i], true, true); - /* delay slot -- set the env argument */ - tcg_out_mov_delay(s, TCG_REG_O0, TCG_AREG0); - } - - for (i = 0; i < ARRAY_SIZE(qemu_st_helpers); ++i) { - if (qemu_st_helpers[i] == NULL) { - continue; - } - - /* May as well align the trampoline. */ - while ((uintptr_t)s->code_ptr & 15) { - tcg_out_nop(s); - } - qemu_st_trampoline[i] = tcg_splitwx_to_rx(s->code_ptr); - - /* Set the retaddr operand. */ - tcg_out_mov(s, TCG_TYPE_PTR, TCG_REG_O4, TCG_REG_O7); - - /* Tail call. */ - tcg_out_jmpl_const(s, qemu_st_helpers[i], true, true); - /* delay slot -- set the env argument */ - tcg_out_mov_delay(s, TCG_REG_O0, TCG_AREG0); - } -} -#else -static const tcg_insn_unit *qemu_unalign_ld_trampoline; -static const tcg_insn_unit *qemu_unalign_st_trampoline; - -static void build_trampolines(TCGContext *s) -{ - for (int ld = 0; ld < 2; ++ld) { - void *helper; - - while ((uintptr_t)s->code_ptr & 15) { - tcg_out_nop(s); - } - - if (ld) { - helper = helper_unaligned_ld; - qemu_unalign_ld_trampoline = tcg_splitwx_to_rx(s->code_ptr); - } else { - helper = helper_unaligned_st; - qemu_unalign_st_trampoline = tcg_splitwx_to_rx(s->code_ptr); - } - - /* Tail call. */ - tcg_out_jmpl_const(s, helper, true, true); - /* delay slot -- set the env argument */ - tcg_out_mov_delay(s, TCG_REG_O0, TCG_AREG0); - } -} -#endif - /* Generate global QEMU prologue and epilogue code */ static void tcg_target_qemu_prologue(TCGContext *s) { @@ -XXX,XX +XXX,XX @@ static void tcg_target_qemu_prologue(TCGContext *s) tcg_out_arithi(s, TCG_REG_G0, TCG_REG_I7, 8, RETURN); /* delay slot */ tcg_out_movi_s13(s, TCG_REG_O0, 0); - - build_trampolines(s); } static void tcg_out_nop_fill(tcg_insn_unit *p, int count) @@ -XXX,XX +XXX,XX @@ static void tcg_out_nop_fill(tcg_insn_unit *p, int count) } } -#if defined(CONFIG_SOFTMMU) +static const TCGLdstHelperParam ldst_helper_param = { + .ntmp = 1, .tmp = { TCG_REG_T1 } +}; -/* We expect to use a 13-bit negative offset from ENV. */ -QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) > 0); -QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) < -(1 << 12)); - -/* Perform the TLB load and compare. - - Inputs: - ADDRLO and ADDRHI contain the possible two parts of the address. - - MEM_INDEX and S_BITS are the memory context and log2 size of the load. - - WHICH is the offset into the CPUTLBEntry structure of the slot to read. - This should be offsetof addr_read or addr_write. - - The result of the TLB comparison is in %[ix]cc. The sanitized address - is in the returned register, maybe %o0. The TLB addend is in %o1. */ - -static TCGReg tcg_out_tlb_load(TCGContext *s, TCGReg addr, int mem_index, - MemOp opc, int which) +static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) { + MemOp opc = get_memop(lb->oi); + MemOp sgn; + + if (!patch_reloc(lb->label_ptr[0], R_SPARC_WDISP19, + (intptr_t)tcg_splitwx_to_rx(s->code_ptr), 0)) { + return false; + } + + /* Use inline tcg_out_ext32s; otherwise let the helper sign-extend. */ + sgn = (opc & MO_SIZE) < MO_32 ? MO_SIGN : 0; + + tcg_out_ld_helper_args(s, lb, &ldst_helper_param); + tcg_out_call(s, qemu_ld_helpers[opc & (MO_SIZE | sgn)], NULL); + tcg_out_ld_helper_ret(s, lb, sgn, &ldst_helper_param); + + tcg_out_bpcc0(s, COND_A, BPCC_A | BPCC_PT, 0); + return patch_reloc(s->code_ptr - 1, R_SPARC_WDISP19, + (intptr_t)lb->raddr, 0); +} + +static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) +{ + MemOp opc = get_memop(lb->oi); + + if (!patch_reloc(lb->label_ptr[0], R_SPARC_WDISP19, + (intptr_t)tcg_splitwx_to_rx(s->code_ptr), 0)) { + return false; + } + + tcg_out_st_helper_args(s, lb, &ldst_helper_param); + tcg_out_call(s, qemu_st_helpers[opc & MO_SIZE], NULL); + + tcg_out_bpcc0(s, COND_A, BPCC_A | BPCC_PT, 0); + return patch_reloc(s->code_ptr - 1, R_SPARC_WDISP19, + (intptr_t)lb->raddr, 0); +} + +typedef struct { + TCGReg base; + TCGReg index; +} HostAddress; + +/* + * For softmmu, perform the TLB load and compare. + * For useronly, perform any required alignment tests. + * In both cases, return a TCGLabelQemuLdst structure if the slow path + * is required and fill in @h with the host address for the fast path. + */ +static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, + TCGReg addr_reg, MemOpIdx oi, + bool is_ld) +{ + TCGLabelQemuLdst *ldst = NULL; + MemOp opc = get_memop(oi); + unsigned a_bits = get_alignment_bits(opc); + unsigned s_bits = opc & MO_SIZE; + unsigned a_mask; + + /* We don't support unaligned accesses. */ + a_bits = MAX(a_bits, s_bits); + a_mask = (1u << a_bits) - 1; + +#ifdef CONFIG_SOFTMMU + int mem_index = get_mmuidx(oi); int fast_off = TLB_MASK_TABLE_OFS(mem_index); int mask_off = fast_off + offsetof(CPUTLBDescFast, mask); int table_off = fast_off + offsetof(CPUTLBDescFast, table); - const TCGReg r0 = TCG_REG_O0; - const TCGReg r1 = TCG_REG_O1; - const TCGReg r2 = TCG_REG_O2; - unsigned s_bits = opc & MO_SIZE; - unsigned a_bits = get_alignment_bits(opc); - tcg_target_long compare_mask; + int cmp_off = is_ld ? offsetof(CPUTLBEntry, addr_read) + : offsetof(CPUTLBEntry, addr_write); + int add_off = offsetof(CPUTLBEntry, addend); + int compare_mask; + int cc; /* Load tlb_mask[mmu_idx] and tlb_table[mmu_idx]. */ - tcg_out_ld(s, TCG_TYPE_PTR, r0, TCG_AREG0, mask_off); - tcg_out_ld(s, TCG_TYPE_PTR, r1, TCG_AREG0, table_off); + QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) > 0); + QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) < -(1 << 12)); + tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_T2, TCG_AREG0, mask_off); + tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_T3, TCG_AREG0, table_off); /* Extract the page index, shifted into place for tlb index. */ - tcg_out_arithi(s, r2, addr, TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS, - SHIFT_SRL); - tcg_out_arith(s, r2, r2, r0, ARITH_AND); + tcg_out_arithi(s, TCG_REG_T1, addr_reg, + TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS, SHIFT_SRL); + tcg_out_arith(s, TCG_REG_T1, TCG_REG_T1, TCG_REG_T2, ARITH_AND); /* Add the tlb_table pointer, creating the CPUTLBEntry address into R2. */ - tcg_out_arith(s, r2, r2, r1, ARITH_ADD); + tcg_out_arith(s, TCG_REG_T1, TCG_REG_T1, TCG_REG_T3, ARITH_ADD); - /* Load the tlb comparator and the addend. */ - tcg_out_ld(s, TCG_TYPE_TL, r0, r2, which); - tcg_out_ld(s, TCG_TYPE_PTR, r1, r2, offsetof(CPUTLBEntry, addend)); + /* Load the tlb comparator and the addend. */ + tcg_out_ld(s, TCG_TYPE_TL, TCG_REG_T2, TCG_REG_T1, cmp_off); + tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_T1, TCG_REG_T1, add_off); + h->base = TCG_REG_T1; - /* Mask out the page offset, except for the required alignment. - We don't support unaligned accesses. */ - if (a_bits < s_bits) { - a_bits = s_bits; - } - compare_mask = (tcg_target_ulong)TARGET_PAGE_MASK | ((1 << a_bits) - 1); + /* Mask out the page offset, except for the required alignment. */ + compare_mask = TARGET_PAGE_MASK | a_mask; if (check_fit_tl(compare_mask, 13)) { - tcg_out_arithi(s, r2, addr, compare_mask, ARITH_AND); + tcg_out_arithi(s, TCG_REG_T3, addr_reg, compare_mask, ARITH_AND); } else { - tcg_out_movi(s, TCG_TYPE_TL, r2, compare_mask); - tcg_out_arith(s, r2, addr, r2, ARITH_AND); + tcg_out_movi_s32(s, TCG_REG_T3, compare_mask); + tcg_out_arith(s, TCG_REG_T3, addr_reg, TCG_REG_T3, ARITH_AND); } - tcg_out_cmp(s, r0, r2, 0); + tcg_out_cmp(s, TCG_REG_T2, TCG_REG_T3, 0); - /* If the guest address must be zero-extended, do so now. */ + ldst = new_ldst_label(s); + ldst->is_ld = is_ld; + ldst->oi = oi; + ldst->addrlo_reg = addr_reg; + ldst->label_ptr[0] = s->code_ptr; + + /* bne,pn %[xi]cc, label0 */ + cc = TARGET_LONG_BITS == 64 ? BPCC_XCC : BPCC_ICC; + tcg_out_bpcc0(s, COND_NE, BPCC_PN | cc, 0); +#else + if (a_bits != s_bits) { + /* + * Test for at least natural alignment, and defer + * everything else to the helper functions. + */ + tcg_debug_assert(check_fit_tl(a_mask, 13)); + tcg_out_arithi(s, TCG_REG_G0, addr_reg, a_mask, ARITH_ANDCC); + + ldst = new_ldst_label(s); + ldst->is_ld = is_ld; + ldst->oi = oi; + ldst->addrlo_reg = addr_reg; + ldst->label_ptr[0] = s->code_ptr; + + /* bne,pn %icc, label0 */ + tcg_out_bpcc0(s, COND_NE, BPCC_PN | BPCC_ICC, 0); + } + h->base = guest_base ? TCG_GUEST_BASE_REG : TCG_REG_G0; +#endif + + /* If the guest address must be zero-extended, do in the delay slot. */ if (TARGET_LONG_BITS == 32) { - tcg_out_ext32u(s, r0, addr); - return r0; + tcg_out_ext32u(s, TCG_REG_T2, addr_reg); + h->index = TCG_REG_T2; + } else { + if (ldst) { + tcg_out_nop(s); + } + h->index = addr_reg; } - return addr; + return ldst; } -#endif /* CONFIG_SOFTMMU */ - -static const int qemu_ld_opc[(MO_SSIZE | MO_BSWAP) + 1] = { - [MO_UB] = LDUB, - [MO_SB] = LDSB, - [MO_UB | MO_LE] = LDUB, - [MO_SB | MO_LE] = LDSB, - - [MO_BEUW] = LDUH, - [MO_BESW] = LDSH, - [MO_BEUL] = LDUW, - [MO_BESL] = LDSW, - [MO_BEUQ] = LDX, - [MO_BESQ] = LDX, - - [MO_LEUW] = LDUH_LE, - [MO_LESW] = LDSH_LE, - [MO_LEUL] = LDUW_LE, - [MO_LESL] = LDSW_LE, - [MO_LEUQ] = LDX_LE, - [MO_LESQ] = LDX_LE, -}; - -static const int qemu_st_opc[(MO_SIZE | MO_BSWAP) + 1] = { - [MO_UB] = STB, - - [MO_BEUW] = STH, - [MO_BEUL] = STW, - [MO_BEUQ] = STX, - - [MO_LEUW] = STH_LE, - [MO_LEUL] = STW_LE, - [MO_LEUQ] = STX_LE, -}; static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr, MemOpIdx oi, TCGType data_type) { - MemOp memop = get_memop(oi); - tcg_insn_unit *label_ptr; + static const int ld_opc[(MO_SSIZE | MO_BSWAP) + 1] = { + [MO_UB] = LDUB, + [MO_SB] = LDSB, + [MO_UB | MO_LE] = LDUB, + [MO_SB | MO_LE] = LDSB, -#ifdef CONFIG_SOFTMMU - unsigned memi = get_mmuidx(oi); - TCGReg addrz; - const tcg_insn_unit *func; + [MO_BEUW] = LDUH, + [MO_BESW] = LDSH, + [MO_BEUL] = LDUW, + [MO_BESL] = LDSW, + [MO_BEUQ] = LDX, + [MO_BESQ] = LDX, - addrz = tcg_out_tlb_load(s, addr, memi, memop, - offsetof(CPUTLBEntry, addr_read)); + [MO_LEUW] = LDUH_LE, + [MO_LESW] = LDSH_LE, + [MO_LEUL] = LDUW_LE, + [MO_LESL] = LDSW_LE, + [MO_LEUQ] = LDX_LE, + [MO_LESQ] = LDX_LE, + }; - /* The fast path is exactly one insn. Thus we can perform the - entire TLB Hit in the (annulled) delay slot of the branch - over the TLB Miss case. */ + TCGLabelQemuLdst *ldst; + HostAddress h; - /* beq,a,pt %[xi]cc, label0 */ - label_ptr = s->code_ptr; - tcg_out_bpcc0(s, COND_E, BPCC_A | BPCC_PT - | (TARGET_LONG_BITS == 64 ? BPCC_XCC : BPCC_ICC), 0); - /* delay slot */ - tcg_out_ldst_rr(s, data, addrz, TCG_REG_O1, - qemu_ld_opc[memop & (MO_BSWAP | MO_SSIZE)]); + ldst = prepare_host_addr(s, &h, addr, oi, true); - /* TLB Miss. */ + tcg_out_ldst_rr(s, data, h.base, h.index, + ld_opc[get_memop(oi) & (MO_BSWAP | MO_SSIZE)]); - tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_O1, addrz); - - /* We use the helpers to extend SB and SW data, leaving the case - of SL needing explicit extending below. */ - if ((memop & MO_SSIZE) == MO_SL) { - func = qemu_ld_trampoline[MO_UL]; - } else { - func = qemu_ld_trampoline[memop & MO_SSIZE]; + if (ldst) { + ldst->type = data_type; + ldst->datalo_reg = data; + ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); } - tcg_debug_assert(func != NULL); - tcg_out_call_nodelay(s, func, false); - /* delay slot */ - tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_O2, oi); - - /* We let the helper sign-extend SB and SW, but leave SL for here. */ - if ((memop & MO_SSIZE) == MO_SL) { - tcg_out_ext32s(s, data, TCG_REG_O0); - } else { - tcg_out_mov(s, TCG_TYPE_REG, data, TCG_REG_O0); - } - - *label_ptr |= INSN_OFF19(tcg_ptr_byte_diff(s->code_ptr, label_ptr)); -#else - TCGReg index = (guest_base ? TCG_GUEST_BASE_REG : TCG_REG_G0); - unsigned a_bits = get_alignment_bits(memop); - unsigned s_bits = memop & MO_SIZE; - unsigned t_bits; - - if (TARGET_LONG_BITS == 32) { - tcg_out_ext32u(s, TCG_REG_T1, addr); - addr = TCG_REG_T1; - } - - /* - * Normal case: alignment equal to access size. - */ - if (a_bits == s_bits) { - tcg_out_ldst_rr(s, data, addr, index, - qemu_ld_opc[memop & (MO_BSWAP | MO_SSIZE)]); - return; - } - - /* - * Test for at least natural alignment, and assume most accesses - * will be aligned -- perform a straight load in the delay slot. - * This is required to preserve atomicity for aligned accesses. - */ - t_bits = MAX(a_bits, s_bits); - tcg_debug_assert(t_bits < 13); - tcg_out_arithi(s, TCG_REG_G0, addr, (1u << t_bits) - 1, ARITH_ANDCC); - - /* beq,a,pt %icc, label */ - label_ptr = s->code_ptr; - tcg_out_bpcc0(s, COND_E, BPCC_A | BPCC_PT | BPCC_ICC, 0); - /* delay slot */ - tcg_out_ldst_rr(s, data, addr, index, - qemu_ld_opc[memop & (MO_BSWAP | MO_SSIZE)]); - - if (a_bits >= s_bits) { - /* - * Overalignment: A successful alignment test will perform the memory - * operation in the delay slot, and failure need only invoke the - * handler for SIGBUS. - */ - tcg_out_call_nodelay(s, qemu_unalign_ld_trampoline, false); - /* delay slot -- move to low part of argument reg */ - tcg_out_mov_delay(s, TCG_REG_O1, addr); - } else { - /* Underalignment: load by pieces of minimum alignment. */ - int ld_opc, a_size, s_size, i; - - /* - * Force full address into T1 early; avoids problems with - * overlap between @addr and @data. - */ - tcg_out_arith(s, TCG_REG_T1, addr, index, ARITH_ADD); - - a_size = 1 << a_bits; - s_size = 1 << s_bits; - if ((memop & MO_BSWAP) == MO_BE) { - ld_opc = qemu_ld_opc[a_bits | MO_BE | (memop & MO_SIGN)]; - tcg_out_ldst(s, data, TCG_REG_T1, 0, ld_opc); - ld_opc = qemu_ld_opc[a_bits | MO_BE]; - for (i = a_size; i < s_size; i += a_size) { - tcg_out_ldst(s, TCG_REG_T2, TCG_REG_T1, i, ld_opc); - tcg_out_arithi(s, data, data, a_size, SHIFT_SLLX); - tcg_out_arith(s, data, data, TCG_REG_T2, ARITH_OR); - } - } else if (a_bits == 0) { - ld_opc = LDUB; - tcg_out_ldst(s, data, TCG_REG_T1, 0, ld_opc); - for (i = a_size; i < s_size; i += a_size) { - if ((memop & MO_SIGN) && i == s_size - a_size) { - ld_opc = LDSB; - } - tcg_out_ldst(s, TCG_REG_T2, TCG_REG_T1, i, ld_opc); - tcg_out_arithi(s, TCG_REG_T2, TCG_REG_T2, i * 8, SHIFT_SLLX); - tcg_out_arith(s, data, data, TCG_REG_T2, ARITH_OR); - } - } else { - ld_opc = qemu_ld_opc[a_bits | MO_LE]; - tcg_out_ldst_rr(s, data, TCG_REG_T1, TCG_REG_G0, ld_opc); - for (i = a_size; i < s_size; i += a_size) { - tcg_out_arithi(s, TCG_REG_T1, TCG_REG_T1, a_size, ARITH_ADD); - if ((memop & MO_SIGN) && i == s_size - a_size) { - ld_opc = qemu_ld_opc[a_bits | MO_LE | MO_SIGN]; - } - tcg_out_ldst_rr(s, TCG_REG_T2, TCG_REG_T1, TCG_REG_G0, ld_opc); - tcg_out_arithi(s, TCG_REG_T2, TCG_REG_T2, i * 8, SHIFT_SLLX); - tcg_out_arith(s, data, data, TCG_REG_T2, ARITH_OR); - } - } - } - - *label_ptr |= INSN_OFF19(tcg_ptr_byte_diff(s->code_ptr, label_ptr)); -#endif /* CONFIG_SOFTMMU */ } static void tcg_out_qemu_st(TCGContext *s, TCGReg data, TCGReg addr, MemOpIdx oi, TCGType data_type) { - MemOp memop = get_memop(oi); - tcg_insn_unit *label_ptr; + static const int st_opc[(MO_SIZE | MO_BSWAP) + 1] = { + [MO_UB] = STB, -#ifdef CONFIG_SOFTMMU - unsigned memi = get_mmuidx(oi); - TCGReg addrz; - const tcg_insn_unit *func; + [MO_BEUW] = STH, + [MO_BEUL] = STW, + [MO_BEUQ] = STX, - addrz = tcg_out_tlb_load(s, addr, memi, memop, - offsetof(CPUTLBEntry, addr_write)); + [MO_LEUW] = STH_LE, + [MO_LEUL] = STW_LE, + [MO_LEUQ] = STX_LE, + }; - /* The fast path is exactly one insn. Thus we can perform the entire - TLB Hit in the (annulled) delay slot of the branch over TLB Miss. */ - /* beq,a,pt %[xi]cc, label0 */ - label_ptr = s->code_ptr; - tcg_out_bpcc0(s, COND_E, BPCC_A | BPCC_PT - | (TARGET_LONG_BITS == 64 ? BPCC_XCC : BPCC_ICC), 0); - /* delay slot */ - tcg_out_ldst_rr(s, data, addrz, TCG_REG_O1, - qemu_st_opc[memop & (MO_BSWAP | MO_SIZE)]); + TCGLabelQemuLdst *ldst; + HostAddress h; - /* TLB Miss. */ + ldst = prepare_host_addr(s, &h, addr, oi, false); - tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_O1, addrz); - tcg_out_movext(s, (memop & MO_SIZE) == MO_64 ? TCG_TYPE_I64 : TCG_TYPE_I32, - TCG_REG_O2, data_type, memop & MO_SIZE, data); + tcg_out_ldst_rr(s, data, h.base, h.index, + st_opc[get_memop(oi) & (MO_BSWAP | MO_SIZE)]); - func = qemu_st_trampoline[memop & MO_SIZE]; - tcg_debug_assert(func != NULL); - tcg_out_call_nodelay(s, func, false); - /* delay slot */ - tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_O3, oi); - - *label_ptr |= INSN_OFF19(tcg_ptr_byte_diff(s->code_ptr, label_ptr)); -#else - TCGReg index = (guest_base ? TCG_GUEST_BASE_REG : TCG_REG_G0); - unsigned a_bits = get_alignment_bits(memop); - unsigned s_bits = memop & MO_SIZE; - unsigned t_bits; - - if (TARGET_LONG_BITS == 32) { - tcg_out_ext32u(s, TCG_REG_T1, addr); - addr = TCG_REG_T1; + if (ldst) { + ldst->type = data_type; + ldst->datalo_reg = data; + ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); } - - /* - * Normal case: alignment equal to access size. - */ - if (a_bits == s_bits) { - tcg_out_ldst_rr(s, data, addr, index, - qemu_st_opc[memop & (MO_BSWAP | MO_SIZE)]); - return; - } - - /* - * Test for at least natural alignment, and assume most accesses - * will be aligned -- perform a straight store in the delay slot. - * This is required to preserve atomicity for aligned accesses. - */ - t_bits = MAX(a_bits, s_bits); - tcg_debug_assert(t_bits < 13); - tcg_out_arithi(s, TCG_REG_G0, addr, (1u << t_bits) - 1, ARITH_ANDCC); - - /* beq,a,pt %icc, label */ - label_ptr = s->code_ptr; - tcg_out_bpcc0(s, COND_E, BPCC_A | BPCC_PT | BPCC_ICC, 0); - /* delay slot */ - tcg_out_ldst_rr(s, data, addr, index, - qemu_st_opc[memop & (MO_BSWAP | MO_SIZE)]); - - if (a_bits >= s_bits) { - /* - * Overalignment: A successful alignment test will perform the memory - * operation in the delay slot, and failure need only invoke the - * handler for SIGBUS. - */ - tcg_out_call_nodelay(s, qemu_unalign_st_trampoline, false); - /* delay slot -- move to low part of argument reg */ - tcg_out_mov_delay(s, TCG_REG_O1, addr); - } else { - /* Underalignment: store by pieces of minimum alignment. */ - int st_opc, a_size, s_size, i; - - /* - * Force full address into T1 early; avoids problems with - * overlap between @addr and @data. - */ - tcg_out_arith(s, TCG_REG_T1, addr, index, ARITH_ADD); - - a_size = 1 << a_bits; - s_size = 1 << s_bits; - if ((memop & MO_BSWAP) == MO_BE) { - st_opc = qemu_st_opc[a_bits | MO_BE]; - for (i = 0; i < s_size; i += a_size) { - TCGReg d = data; - int shift = (s_size - a_size - i) * 8; - if (shift) { - d = TCG_REG_T2; - tcg_out_arithi(s, d, data, shift, SHIFT_SRLX); - } - tcg_out_ldst(s, d, TCG_REG_T1, i, st_opc); - } - } else if (a_bits == 0) { - tcg_out_ldst(s, data, TCG_REG_T1, 0, STB); - for (i = 1; i < s_size; i++) { - tcg_out_arithi(s, TCG_REG_T2, data, i * 8, SHIFT_SRLX); - tcg_out_ldst(s, TCG_REG_T2, TCG_REG_T1, i, STB); - } - } else { - /* Note that ST*A with immediate asi must use indexed address. */ - st_opc = qemu_st_opc[a_bits + MO_LE]; - tcg_out_ldst_rr(s, data, TCG_REG_T1, TCG_REG_G0, st_opc); - for (i = a_size; i < s_size; i += a_size) { - tcg_out_arithi(s, TCG_REG_T2, data, i * 8, SHIFT_SRLX); - tcg_out_arithi(s, TCG_REG_T1, TCG_REG_T1, a_size, ARITH_ADD); - tcg_out_ldst_rr(s, TCG_REG_T2, TCG_REG_T1, TCG_REG_G0, st_opc); - } - } - } - - *label_ptr |= INSN_OFF19(tcg_ptr_byte_diff(s->code_ptr, label_ptr)); -#endif /* CONFIG_SOFTMMU */ } static void tcg_out_exit_tb(TCGContext *s, uintptr_t a0) @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_extu_i32_i64: case INDEX_op_extrl_i64_i32: case INDEX_op_extrh_i64_i32: + case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_i64: return C_O1_I1(r, r); case INDEX_op_st8_i32: @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_st_i32: case INDEX_op_st32_i64: case INDEX_op_st_i64: + case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_i64: return C_O0_I2(rZ, r); case INDEX_op_add_i32: @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_muluh_i64: return C_O1_I2(r, r, r); - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: - return C_O1_I1(r, s); - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: - return C_O0_I2(sZ, s); - default: g_assert_not_reached(); } -- 2.34.1 These functions are now unused. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-ldst.h | 6 ------ accel/tcg/user-exec.c | 10 ---------- 2 files changed, 16 deletions(-) diff --git a/include/tcg/tcg-ldst.h b/include/tcg/tcg-ldst.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-ldst.h +++ b/include/tcg/tcg-ldst.h @@ -XXX,XX +XXX,XX @@ void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, void helper_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, MemOpIdx oi, uintptr_t retaddr); -#ifdef CONFIG_USER_ONLY - -G_NORETURN void helper_unaligned_ld(CPUArchState *env, target_ulong addr); -G_NORETURN void helper_unaligned_st(CPUArchState *env, target_ulong addr); - -#endif /* CONFIG_USER_ONLY */ #endif /* TCG_LDST_H */ diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/user-exec.c +++ b/accel/tcg/user-exec.c @@ -XXX,XX +XXX,XX @@ void page_reset_target_data(target_ulong start, target_ulong last) { } /* The softmmu versions of these helpers are in cputlb.c. */ -void helper_unaligned_ld(CPUArchState *env, target_ulong addr) -{ - cpu_loop_exit_sigbus(env_cpu(env), addr, MMU_DATA_LOAD, GETPC()); -} - -void helper_unaligned_st(CPUArchState *env, target_ulong addr) -{ - cpu_loop_exit_sigbus(env_cpu(env), addr, MMU_DATA_STORE, GETPC()); -} - static void *cpu_mmu_lookup(CPUArchState *env, abi_ptr addr, MemOp mop, uintptr_t ra, MMUAccessType type) { -- 2.34.1 This should be true of all loongarch64 running Linux. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/loongarch64/tcg-target.c.inc | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ */ #include "../tcg-ldst.c.inc" +#include <asm/hwcap.h> #ifdef CONFIG_DEBUG_TCG static const char * const tcg_target_reg_names[TCG_TARGET_NB_REGS] = { @@ -XXX,XX +XXX,XX @@ static void tcg_target_qemu_prologue(TCGContext *s) static void tcg_target_init(TCGContext *s) { + unsigned long hwcap = qemu_getauxval(AT_HWCAP); + + /* Server and desktop class cpus have UAL; embedded cpus do not. */ + if (!(hwcap & HWCAP_LOONGARCH_UAL)) { + error_report("TCG: unaligned access support required; exiting"); + exit(EXIT_FAILURE); + } + tcg_target_available_regs[TCG_TYPE_I32] = ALL_GENERAL_REGS; tcg_target_available_regs[TCG_TYPE_I64] = ALL_GENERAL_REGS; -- 2.34.1 Test the final byte of an unaligned access. Use BSTRINS.D to clear the range of bits, rather than AND. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/loongarch64/tcg-target.c.inc | 19 ++++++++++++------- 1 file changed, 12 insertions(+), 7 deletions(-) diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, int fast_ofs = TLB_MASK_TABLE_OFS(mem_index); int mask_ofs = fast_ofs + offsetof(CPUTLBDescFast, mask); int table_ofs = fast_ofs + offsetof(CPUTLBDescFast, table); - tcg_target_long compare_mask; ldst = new_ldst_label(s); ldst->is_ld = is_ld; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP2, TCG_REG_TMP2, offsetof(CPUTLBEntry, addend)); - /* We don't support unaligned accesses. */ + /* + * For aligned accesses, we check the first byte and include the alignment + * bits within the address. For unaligned access, we check that we don't + * cross pages using the address of the last byte of the access. + */ if (a_bits < s_bits) { - a_bits = s_bits; + unsigned a_mask = (1u << a_bits) - 1; + unsigned s_mask = (1u << s_bits) - 1; + tcg_out_addi(s, TCG_TYPE_TL, TCG_REG_TMP1, addr_reg, s_mask - a_mask); + } else { + tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_TMP1, addr_reg); } - /* Clear the non-page, non-alignment bits from the address. */ - compare_mask = (tcg_target_long)TARGET_PAGE_MASK | ((1 << a_bits) - 1); - tcg_out_movi(s, TCG_TYPE_TL, TCG_REG_TMP1, compare_mask); - tcg_out_opc_and(s, TCG_REG_TMP1, TCG_REG_TMP1, addr_reg); + tcg_out_opc_bstrins_d(s, TCG_REG_TMP1, TCG_REG_ZERO, + a_bits, TARGET_PAGE_BITS - 1); /* Compare masked address with the TLB entry. */ ldst->label_ptr[0] = s->code_ptr; -- 2.34.1 The system is required to emulate unaligned accesses, even if the hardware does not support it. The resulting trap may or may not be more efficient than the qemu slow path. There are linux kernel patches in flight to allow userspace to query hardware support; we can re-evaluate whether to enable this by default after that. In the meantime, softmmu now matches useronly, where we already assumed that unaligned accesses are supported. Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/riscv/tcg-target.c.inc | 48 ++++++++++++++++++++++---------------- 1 file changed, 28 insertions(+), 20 deletions(-) diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, #ifdef CONFIG_SOFTMMU unsigned s_bits = opc & MO_SIZE; + unsigned s_mask = (1u << s_bits) - 1; int mem_index = get_mmuidx(oi); int fast_ofs = TLB_MASK_TABLE_OFS(mem_index); int mask_ofs = fast_ofs + offsetof(CPUTLBDescFast, mask); int table_ofs = fast_ofs + offsetof(CPUTLBDescFast, table); - TCGReg mask_base = TCG_AREG0, table_base = TCG_AREG0; - tcg_target_long compare_mask; + int compare_mask; + TCGReg addr_adj; ldst = new_ldst_label(s); ldst->is_ld = is_ld; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) > 0); QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) < -(1 << 11)); - tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP0, mask_base, mask_ofs); - tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, table_base, table_ofs); + tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP0, TCG_AREG0, mask_ofs); + tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_AREG0, table_ofs); tcg_out_opc_imm(s, OPC_SRLI, TCG_REG_TMP2, addr_reg, TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); tcg_out_opc_reg(s, OPC_AND, TCG_REG_TMP2, TCG_REG_TMP2, TCG_REG_TMP0); tcg_out_opc_reg(s, OPC_ADD, TCG_REG_TMP2, TCG_REG_TMP2, TCG_REG_TMP1); + /* + * For aligned accesses, we check the first byte and include the alignment + * bits within the address. For unaligned access, we check that we don't + * cross pages using the address of the last byte of the access. + */ + addr_adj = addr_reg; + if (a_bits < s_bits) { + addr_adj = TCG_REG_TMP0; + tcg_out_opc_imm(s, TARGET_LONG_BITS == 32 ? OPC_ADDIW : OPC_ADDI, + addr_adj, addr_reg, s_mask - a_mask); + } + compare_mask = TARGET_PAGE_MASK | a_mask; + if (compare_mask == sextreg(compare_mask, 0, 12)) { + tcg_out_opc_imm(s, OPC_ANDI, TCG_REG_TMP1, addr_adj, compare_mask); + } else { + tcg_out_movi(s, TCG_TYPE_TL, TCG_REG_TMP1, compare_mask); + tcg_out_opc_reg(s, OPC_AND, TCG_REG_TMP1, TCG_REG_TMP1, addr_adj); + } + /* Load the tlb comparator and the addend. */ tcg_out_ld(s, TCG_TYPE_TL, TCG_REG_TMP0, TCG_REG_TMP2, is_ld ? offsetof(CPUTLBEntry, addr_read) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP2, TCG_REG_TMP2, offsetof(CPUTLBEntry, addend)); - /* We don't support unaligned accesses. */ - if (a_bits < s_bits) { - a_bits = s_bits; - } - /* Clear the non-page, non-alignment bits from the address. */ - compare_mask = (tcg_target_long)TARGET_PAGE_MASK | a_mask; - if (compare_mask == sextreg(compare_mask, 0, 12)) { - tcg_out_opc_imm(s, OPC_ANDI, TCG_REG_TMP1, addr_reg, compare_mask); - } else { - tcg_out_movi(s, TCG_TYPE_TL, TCG_REG_TMP1, compare_mask); - tcg_out_opc_reg(s, OPC_AND, TCG_REG_TMP1, TCG_REG_TMP1, addr_reg); - } - /* Compare masked address with the TLB entry. */ ldst->label_ptr[0] = s->code_ptr; tcg_out_opc_branch(s, OPC_BNE, TCG_REG_TMP0, TCG_REG_TMP1, 0); /* TLB Hit - translate address using addend. */ + addr_adj = addr_reg; if (TARGET_LONG_BITS == 32) { - tcg_out_ext32u(s, TCG_REG_TMP0, addr_reg); - addr_reg = TCG_REG_TMP0; + addr_adj = TCG_REG_TMP0; + tcg_out_ext32u(s, addr_adj, addr_reg); } - tcg_out_opc_reg(s, OPC_ADD, TCG_REG_TMP0, TCG_REG_TMP2, addr_reg); + tcg_out_opc_reg(s, OPC_ADD, TCG_REG_TMP0, TCG_REG_TMP2, addr_adj); *pbase = TCG_REG_TMP0; #else if (a_mask) { -- 2.34.1 Replace the unparameterized TCG_TARGET_HAS_MEMORY_BSWAP macro with a function with a memop argument. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.h | 1 - tcg/arm/tcg-target.h | 1 - tcg/i386/tcg-target.h | 3 --- tcg/loongarch64/tcg-target.h | 2 -- tcg/mips/tcg-target.h | 2 -- tcg/ppc/tcg-target.h | 1 - tcg/riscv/tcg-target.h | 2 -- tcg/s390x/tcg-target.h | 2 -- tcg/sparc64/tcg-target.h | 1 - tcg/tcg-internal.h | 2 ++ tcg/tci/tcg-target.h | 2 -- tcg/tcg-op.c | 20 +++++++++++--------- tcg/aarch64/tcg-target.c.inc | 5 +++++ tcg/arm/tcg-target.c.inc | 5 +++++ tcg/i386/tcg-target.c.inc | 5 +++++ tcg/loongarch64/tcg-target.c.inc | 5 +++++ tcg/mips/tcg-target.c.inc | 5 +++++ tcg/ppc/tcg-target.c.inc | 5 +++++ tcg/riscv/tcg-target.c.inc | 5 +++++ tcg/s390x/tcg-target.c.inc | 5 +++++ tcg/sparc64/tcg-target.c.inc | 5 +++++ tcg/tci/tcg-target.c.inc | 5 +++++ 22 files changed, 63 insertions(+), 26 deletions(-) diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.h +++ b/tcg/aarch64/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_lse2; #define TCG_TARGET_HAS_cmpsel_vec 0 #define TCG_TARGET_DEFAULT_MO (0) -#define TCG_TARGET_HAS_MEMORY_BSWAP 0 #define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.h +++ b/tcg/arm/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool use_neon_instructions; #define TCG_TARGET_HAS_cmpsel_vec 0 #define TCG_TARGET_DEFAULT_MO (0) -#define TCG_TARGET_HAS_MEMORY_BSWAP 0 #define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.h +++ b/tcg/i386/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_atomic16; #include "tcg/tcg-mo.h" #define TCG_TARGET_DEFAULT_MO (TCG_MO_ALL & ~TCG_MO_ST_LD) - -#define TCG_TARGET_HAS_MEMORY_BSWAP have_movbe - #define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS diff --git a/tcg/loongarch64/tcg-target.h b/tcg/loongarch64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.h +++ b/tcg/loongarch64/tcg-target.h @@ -XXX,XX +XXX,XX @@ typedef enum { #define TCG_TARGET_NEED_LDST_LABELS -#define TCG_TARGET_HAS_MEMORY_BSWAP 0 - #endif /* LOONGARCH_TCG_TARGET_H */ diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.h +++ b/tcg/mips/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool use_mips32r2_instructions; #endif #define TCG_TARGET_DEFAULT_MO 0 -#define TCG_TARGET_HAS_MEMORY_BSWAP 0 - #define TCG_TARGET_NEED_LDST_LABELS #endif diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.h +++ b/tcg/ppc/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_vsx; #define TCG_TARGET_HAS_cmpsel_vec 0 #define TCG_TARGET_DEFAULT_MO (0) -#define TCG_TARGET_HAS_MEMORY_BSWAP 1 #define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.h +++ b/tcg/riscv/tcg-target.h @@ -XXX,XX +XXX,XX @@ typedef enum { #define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS -#define TCG_TARGET_HAS_MEMORY_BSWAP 0 - #endif diff --git a/tcg/s390x/tcg-target.h b/tcg/s390x/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.h +++ b/tcg/s390x/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern uint64_t s390_facilities[3]; #define TCG_TARGET_CALL_ARG_I128 TCG_CALL_ARG_BY_REF #define TCG_TARGET_CALL_RET_I128 TCG_CALL_RET_BY_REF -#define TCG_TARGET_HAS_MEMORY_BSWAP 1 - #define TCG_TARGET_DEFAULT_MO (TCG_MO_ALL & ~TCG_MO_ST_LD) #define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS diff --git a/tcg/sparc64/tcg-target.h b/tcg/sparc64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.h +++ b/tcg/sparc64/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool use_vis3_instructions; #define TCG_AREG0 TCG_REG_I0 #define TCG_TARGET_DEFAULT_MO (0) -#define TCG_TARGET_HAS_MEMORY_BSWAP 1 #define TCG_TARGET_NEED_LDST_LABELS #define TCG_TARGET_NEED_POOL_LABELS diff --git a/tcg/tcg-internal.h b/tcg/tcg-internal.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-internal.h +++ b/tcg/tcg-internal.h @@ -XXX,XX +XXX,XX @@ static inline TCGv_i64 TCGV128_HIGH(TCGv_i128 t) return temp_tcgv_i64(tcgv_i128_temp(t) + o); } +bool tcg_target_has_memory_bswap(MemOp memop); + #endif /* TCG_INTERNAL_H */ diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci/tcg-target.h +++ b/tcg/tci/tcg-target.h @@ -XXX,XX +XXX,XX @@ typedef enum { We prefer consistency across hosts on this. */ #define TCG_TARGET_DEFAULT_MO (0) -#define TCG_TARGET_HAS_MEMORY_BSWAP 1 - #endif /* TCG_TARGET_H */ diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op.c +++ b/tcg/tcg-op.c @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) oi = make_memop_idx(memop, idx); orig_memop = memop; - if (!TCG_TARGET_HAS_MEMORY_BSWAP && (memop & MO_BSWAP)) { + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { memop &= ~MO_BSWAP; /* The bswap primitive benefits from zero-extended input. */ if ((memop & MO_SSIZE) == MO_SW) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) memop = tcg_canonicalize_memop(memop, 0, 1); oi = make_memop_idx(memop, idx); - if (!TCG_TARGET_HAS_MEMORY_BSWAP && (memop & MO_BSWAP)) { + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { swap = tcg_temp_ebb_new_i32(); switch (memop & MO_SIZE) { case MO_16: @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) oi = make_memop_idx(memop, idx); orig_memop = memop; - if (!TCG_TARGET_HAS_MEMORY_BSWAP && (memop & MO_BSWAP)) { + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { memop &= ~MO_BSWAP; /* The bswap primitive benefits from zero-extended input. */ if ((memop & MO_SIGN) && (memop & MO_SIZE) < MO_64) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) memop = tcg_canonicalize_memop(memop, 1, 1); oi = make_memop_idx(memop, idx); - if (!TCG_TARGET_HAS_MEMORY_BSWAP && (memop & MO_BSWAP)) { + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { swap = tcg_temp_ebb_new_i64(); switch (memop & MO_SIZE) { case MO_16: @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) tcg_debug_assert((orig & MO_SIZE) == MO_128); tcg_debug_assert((orig & MO_SIGN) == 0); - /* Use a memory ordering implemented by the host. */ - if (!TCG_TARGET_HAS_MEMORY_BSWAP && (orig & MO_BSWAP)) { - mop_1 &= ~MO_BSWAP; - } - /* Reduce the size to 64-bit. */ mop_1 = (mop_1 & ~MO_SIZE) | MO_64; @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) default: g_assert_not_reached(); } + + /* Use a memory ordering implemented by the host. */ + if ((orig & MO_BSWAP) && !tcg_target_has_memory_bswap(mop_1)) { + mop_1 &= ~MO_BSWAP; + mop_2 &= ~MO_BSWAP; + } + ret[0] = mop_1; ret[1] = mop_2; } diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGType index_ext; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return false; +} + static const TCGLdstHelperParam ldst_helper_param = { .ntmp = 1, .tmp = { TCG_REG_TMP } }; diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { bool index_scratch; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return false; +} + static TCGReg ldst_ra_gen(TCGContext *s, const TCGLabelQemuLdst *l, int arg) { /* We arrive at the slow path via "BLNE", so R14 contains l->raddr. */ diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { int seg; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return have_movbe; +} + /* * Because i686 has no register parameters and because x86_64 has xchg * to handle addr/data register overlap, we have placed all input arguments diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGReg index; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return false; +} + /* * For softmmu, perform the TLB load and compare. * For useronly, perform any required alignment tests. diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { MemOp align; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return false; +} + /* * For softmmu, perform the TLB load and compare. * For useronly, perform any required alignment tests. diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGReg index; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return true; +} + /* * For softmmu, perform the TLB load and compare. * For useronly, perform any required alignment tests. diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_goto(TCGContext *s, const tcg_insn_unit *target) tcg_debug_assert(ok); } +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return false; +} + /* We have three temps, we might as well expose them. */ static const TCGLdstHelperParam ldst_helper_param = { .ntmp = 3, .tmp = { TCG_REG_TMP0, TCG_REG_TMP1, TCG_REG_TMP2 } diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { int disp; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return true; +} + static void tcg_out_qemu_ld_direct(TCGContext *s, MemOp opc, TCGReg data, HostAddress h) { diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGReg index; } HostAddress; +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return true; +} + /* * For softmmu, perform the TLB load and compare. * For useronly, perform any required alignment tests. diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci/tcg-target.c.inc +++ b/tcg/tci/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_target_init(TCGContext *s) static inline void tcg_target_qemu_prologue(TCGContext *s) { } + +bool tcg_target_has_memory_bswap(MemOp memop) +{ + return true; +} -- 2.34.1 Add opcodes for backend support for 128-bit memory operations. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- docs/devel/tcg-ops.rst | 11 +++--- include/tcg/tcg-opc.h | 8 +++++ tcg/aarch64/tcg-target.h | 2 ++ tcg/arm/tcg-target.h | 2 ++ tcg/i386/tcg-target.h | 2 ++ tcg/loongarch64/tcg-target.h | 1 + tcg/mips/tcg-target.h | 2 ++ tcg/ppc/tcg-target.h | 2 ++ tcg/riscv/tcg-target.h | 2 ++ tcg/s390x/tcg-target.h | 2 ++ tcg/sparc64/tcg-target.h | 2 ++ tcg/tci/tcg-target.h | 2 ++ tcg/tcg-op.c | 69 ++++++++++++++++++++++++++++++++---- tcg/tcg.c | 6 ++++ 14 files changed, 103 insertions(+), 10 deletions(-) diff --git a/docs/devel/tcg-ops.rst b/docs/devel/tcg-ops.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/devel/tcg-ops.rst +++ b/docs/devel/tcg-ops.rst @@ -XXX,XX +XXX,XX @@ QEMU specific operations | This operation is optional. If the TCG backend does not implement the goto_ptr opcode, emitting this op is equivalent to emitting exit_tb(0). - * - qemu_ld_i32/i64 *t0*, *t1*, *flags*, *memidx* + * - qemu_ld_i32/i64/i128 *t0*, *t1*, *flags*, *memidx* - qemu_st_i32/i64 *t0*, *t1*, *flags*, *memidx* + qemu_st_i32/i64/i128 *t0*, *t1*, *flags*, *memidx* qemu_st8_i32 *t0*, *t1*, *flags*, *memidx* - | Load data at the guest address *t1* into *t0*, or store data in *t0* at guest - address *t1*. The _i32/_i64 size applies to the size of the input/output + address *t1*. The _i32/_i64/_i128 size applies to the size of the input/output register *t0* only. The address *t1* is always sized according to the guest, and the width of the memory operation is controlled by *flags*. | | Both *t0* and *t1* may be split into little-endian ordered pairs of registers - if dealing with 64-bit quantities on a 32-bit host. + if dealing with 64-bit quantities on a 32-bit host, or 128-bit quantities on + a 64-bit host. | | The *memidx* selects the qemu tlb index to use (e.g. user or kernel access). The flags are the MemOp bits, selecting the sign, width, and endianness @@ -XXX,XX +XXX,XX @@ QEMU specific operations | For a 32-bit host, qemu_ld/st_i64 is guaranteed to only be used with a 64-bit memory access specified in *flags*. | + | For qemu_ld/st_i128, these are only supported for a 64-bit host. + | | For i386, qemu_st8_i32 is exactly like qemu_st_i32, except the size of the memory operation is known to be 8-bit. This allows the backend to provide a different set of register constraints. diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-opc.h +++ b/include/tcg/tcg-opc.h @@ -XXX,XX +XXX,XX @@ DEF(qemu_st8_i32, 0, TLADDR_ARGS + 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | IMPL(TCG_TARGET_HAS_qemu_st8_i32)) +/* Only for 64-bit hosts at the moment. */ +DEF(qemu_ld_i128, 2, 1, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT | + IMPL(TCG_TARGET_HAS_qemu_ldst_i128)) +DEF(qemu_st_i128, 0, 3, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT | + IMPL(TCG_TARGET_HAS_qemu_ldst_i128)) + /* Host vector support. */ #define IMPLVEC TCG_OPF_VECTOR | IMPL(TCG_TARGET_MAYBE_vec) diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.h +++ b/tcg/aarch64/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_lse2; #define TCG_TARGET_HAS_muluh_i64 1 #define TCG_TARGET_HAS_mulsh_i64 1 +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + #define TCG_TARGET_HAS_v64 1 #define TCG_TARGET_HAS_v128 1 #define TCG_TARGET_HAS_v256 0 diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.h +++ b/tcg/arm/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool use_neon_instructions; #define TCG_TARGET_HAS_rem_i32 0 #define TCG_TARGET_HAS_qemu_st8_i32 0 +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + #define TCG_TARGET_HAS_v64 use_neon_instructions #define TCG_TARGET_HAS_v128 use_neon_instructions #define TCG_TARGET_HAS_v256 0 diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.h +++ b/tcg/i386/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_atomic16; #define TCG_TARGET_HAS_qemu_st8_i32 1 #endif +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + /* We do not support older SSE systems, only beginning with AVX1. */ #define TCG_TARGET_HAS_v64 have_avx1 #define TCG_TARGET_HAS_v128 have_avx1 diff --git a/tcg/loongarch64/tcg-target.h b/tcg/loongarch64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.h +++ b/tcg/loongarch64/tcg-target.h @@ -XXX,XX +XXX,XX @@ typedef enum { #define TCG_TARGET_HAS_muls2_i64 0 #define TCG_TARGET_HAS_muluh_i64 1 #define TCG_TARGET_HAS_mulsh_i64 1 +#define TCG_TARGET_HAS_qemu_ldst_i128 0 #define TCG_TARGET_DEFAULT_MO (0) diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.h +++ b/tcg/mips/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool use_mips32r2_instructions; #define TCG_TARGET_HAS_ext16u_i64 0 /* andi rt, rs, 0xffff */ #endif +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + #define TCG_TARGET_DEFAULT_MO 0 #define TCG_TARGET_NEED_LDST_LABELS diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.h +++ b/tcg/ppc/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_vsx; #define TCG_TARGET_HAS_mulsh_i64 1 #endif +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + /* * While technically Altivec could support V64, it has no 64-bit store * instruction and substituting two 32-bit stores makes the generated diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.h +++ b/tcg/riscv/tcg-target.h @@ -XXX,XX +XXX,XX @@ typedef enum { #define TCG_TARGET_HAS_muluh_i64 1 #define TCG_TARGET_HAS_mulsh_i64 1 +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + #define TCG_TARGET_DEFAULT_MO (0) #define TCG_TARGET_NEED_LDST_LABELS diff --git a/tcg/s390x/tcg-target.h b/tcg/s390x/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.h +++ b/tcg/s390x/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern uint64_t s390_facilities[3]; #define TCG_TARGET_HAS_muluh_i64 0 #define TCG_TARGET_HAS_mulsh_i64 0 +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + #define TCG_TARGET_HAS_v64 HAVE_FACILITY(VECTOR) #define TCG_TARGET_HAS_v128 HAVE_FACILITY(VECTOR) #define TCG_TARGET_HAS_v256 0 diff --git a/tcg/sparc64/tcg-target.h b/tcg/sparc64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.h +++ b/tcg/sparc64/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool use_vis3_instructions; #define TCG_TARGET_HAS_muluh_i64 use_vis3_instructions #define TCG_TARGET_HAS_mulsh_i64 0 +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + #define TCG_AREG0 TCG_REG_I0 #define TCG_TARGET_DEFAULT_MO (0) diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci/tcg-target.h +++ b/tcg/tci/tcg-target.h @@ -XXX,XX +XXX,XX @@ #define TCG_TARGET_HAS_mulu2_i32 1 #endif /* TCG_TARGET_REG_BITS == 64 */ +#define TCG_TARGET_HAS_qemu_ldst_i128 0 + /* Number of registers available. */ #define TCG_TARGET_NB_REGS 16 diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op.c +++ b/tcg/tcg-op.c @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) { - MemOpIdx oi = make_memop_idx(memop, idx); + const MemOpIdx oi = make_memop_idx(memop, idx); tcg_debug_assert((memop & MO_SIZE) == MO_128); tcg_debug_assert((memop & MO_SIGN) == 0); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); addr = plugin_prep_mem_callbacks(addr); - /* TODO: allow the tcg backend to see the whole operation. */ + /* TODO: For now, force 32-bit hosts to use the helper. */ + if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { + TCGv_i64 lo, hi; + TCGArg addr_arg; + MemOpIdx adj_oi; + bool need_bswap = false; - if (use_two_i64_for_i128(memop)) { + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + lo = TCGV128_HIGH(val); + hi = TCGV128_LOW(val); + adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); + need_bswap = true; + } else { + lo = TCGV128_LOW(val); + hi = TCGV128_HIGH(val); + adj_oi = oi; + } + +#if TARGET_LONG_BITS == 32 + addr_arg = tcgv_i32_arg(addr); +#else + addr_arg = tcgv_i64_arg(addr); +#endif + tcg_gen_op4ii_i64(INDEX_op_qemu_ld_i128, lo, hi, addr_arg, adj_oi); + + if (need_bswap) { + tcg_gen_bswap64_i64(lo, lo); + tcg_gen_bswap64_i64(hi, hi); + } + } else if (use_two_i64_for_i128(memop)) { MemOp mop[2]; TCGv addr_p8; TCGv_i64 x, y; @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) { - MemOpIdx oi = make_memop_idx(memop, idx); + const MemOpIdx oi = make_memop_idx(memop, idx); tcg_debug_assert((memop & MO_SIZE) == MO_128); tcg_debug_assert((memop & MO_SIGN) == 0); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); addr = plugin_prep_mem_callbacks(addr); - /* TODO: allow the tcg backend to see the whole operation. */ + /* TODO: For now, force 32-bit hosts to use the helper. */ - if (use_two_i64_for_i128(memop)) { + if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { + TCGv_i64 lo, hi; + TCGArg addr_arg; + MemOpIdx adj_oi; + bool need_bswap = false; + + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + lo = tcg_temp_new_i64(); + hi = tcg_temp_new_i64(); + tcg_gen_bswap64_i64(lo, TCGV128_HIGH(val)); + tcg_gen_bswap64_i64(hi, TCGV128_LOW(val)); + adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); + need_bswap = true; + } else { + lo = TCGV128_LOW(val); + hi = TCGV128_HIGH(val); + adj_oi = oi; + } + +#if TARGET_LONG_BITS == 32 + addr_arg = tcgv_i32_arg(addr); +#else + addr_arg = tcgv_i64_arg(addr); +#endif + tcg_gen_op4ii_i64(INDEX_op_qemu_st_i128, lo, hi, addr_arg, adj_oi); + + if (need_bswap) { + tcg_temp_free_i64(lo); + tcg_temp_free_i64(hi); + } + } else if (use_two_i64_for_i128(memop)) { MemOp mop[2]; TCGv addr_p8; TCGv_i64 x, y; diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ bool tcg_op_supported(TCGOpcode op) case INDEX_op_qemu_st8_i32: return TCG_TARGET_HAS_qemu_st8_i32; + case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_st_i128: + return TCG_TARGET_HAS_qemu_ldst_i128; + case INDEX_op_mov_i32: case INDEX_op_setcond_i32: case INDEX_op_brcond_i32: @@ -XXX,XX +XXX,XX @@ static void tcg_dump_ops(TCGContext *s, FILE *f, bool have_prefs) case INDEX_op_qemu_st8_i32: case INDEX_op_qemu_ld_i64: case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_st_i128: { const char *s_al, *s_op, *s_at; MemOpIdx oi = op->args[k++]; -- 2.34.1 With x86_64 as host, we do not have any temporaries with which to resolve cycles, but we do have xchg. As a side bonus, the set of graphs that can be made with 3 nodes and all nodes conflicting is small: two. We can solve the cycle with a single temp. This is required for x86_64 to handle stores of i128: 1 address register and 2 data registers. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 138 ++++++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 108 insertions(+), 30 deletions(-) diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static void tcg_out_movext2(TCGContext *s, const TCGMovExtend *i1, tcg_out_movext1_new_src(s, i1, src1); } +/** + * tcg_out_movext3 -- move and extend three pair + * @s: tcg context + * @i1: first move description + * @i2: second move description + * @i3: third move description + * @scratch: temporary register, or -1 for none + * + * As tcg_out_movext, for all of @i1, @i2 and @i3, caring for overlap + * between the sources and destinations. + */ + +static void tcg_out_movext3(TCGContext *s, const TCGMovExtend *i1, + const TCGMovExtend *i2, const TCGMovExtend *i3, + int scratch) +{ + TCGReg src1 = i1->src; + TCGReg src2 = i2->src; + TCGReg src3 = i3->src; + + if (i1->dst != src2 && i1->dst != src3) { + tcg_out_movext1(s, i1); + tcg_out_movext2(s, i2, i3, scratch); + return; + } + if (i2->dst != src1 && i2->dst != src3) { + tcg_out_movext1(s, i2); + tcg_out_movext2(s, i1, i3, scratch); + return; + } + if (i3->dst != src1 && i3->dst != src2) { + tcg_out_movext1(s, i3); + tcg_out_movext2(s, i1, i2, scratch); + return; + } + + /* + * There is a cycle. Since there are only 3 nodes, the cycle is + * either "clockwise" or "anti-clockwise", and can be solved with + * a single scratch or two xchg. + */ + if (i1->dst == src2 && i2->dst == src3 && i3->dst == src1) { + /* "Clockwise" */ + if (tcg_out_xchg(s, MAX(i1->src_type, i2->src_type), src1, src2)) { + tcg_out_xchg(s, MAX(i2->src_type, i3->src_type), src2, src3); + /* The data is now in the correct registers, now extend. */ + tcg_out_movext1_new_src(s, i1, i1->dst); + tcg_out_movext1_new_src(s, i2, i2->dst); + tcg_out_movext1_new_src(s, i3, i3->dst); + } else { + tcg_debug_assert(scratch >= 0); + tcg_out_mov(s, i1->src_type, scratch, src1); + tcg_out_movext1(s, i3); + tcg_out_movext1(s, i2); + tcg_out_movext1_new_src(s, i1, scratch); + } + } else if (i1->dst == src3 && i2->dst == src1 && i3->dst == src2) { + /* "Anti-clockwise" */ + if (tcg_out_xchg(s, MAX(i2->src_type, i3->src_type), src2, src3)) { + tcg_out_xchg(s, MAX(i1->src_type, i2->src_type), src1, src2); + /* The data is now in the correct registers, now extend. */ + tcg_out_movext1_new_src(s, i1, i1->dst); + tcg_out_movext1_new_src(s, i2, i2->dst); + tcg_out_movext1_new_src(s, i3, i3->dst); + } else { + tcg_debug_assert(scratch >= 0); + tcg_out_mov(s, i1->src_type, scratch, src1); + tcg_out_movext1(s, i2); + tcg_out_movext1(s, i3); + tcg_out_movext1_new_src(s, i1, scratch); + } + } else { + g_assert_not_reached(); + } +} + #define C_PFX1(P, A) P##A #define C_PFX2(P, A, B) P##A##_##B #define C_PFX3(P, A, B, C) P##A##_##B##_##C @@ -XXX,XX +XXX,XX @@ static int tcg_out_helper_stk_ofs(TCGType type, unsigned slot) static void tcg_out_helper_load_regs(TCGContext *s, unsigned nmov, TCGMovExtend *mov, - unsigned ntmp, const int *tmp) + const TCGLdstHelperParam *parm) { + TCGReg dst3; + switch (nmov) { - default: + case 4: /* The backend must have provided enough temps for the worst case. */ - tcg_debug_assert(ntmp + 1 >= nmov); + tcg_debug_assert(parm->ntmp >= 2); - for (unsigned i = nmov - 1; i >= 2; --i) { - TCGReg dst = mov[i].dst; + dst3 = mov[3].dst; + for (unsigned j = 0; j < 3; ++j) { + if (dst3 == mov[j].src) { + /* + * Conflict. Copy the source to a temporary, perform the + * remaining moves, then the extension from our scratch + * on the way out. + */ + TCGReg scratch = parm->tmp[1]; - for (unsigned j = 0; j < i; ++j) { - if (dst == mov[j].src) { - /* - * Conflict. - * Copy the source to a temporary, recurse for the - * remaining moves, perform the extension from our - * scratch on the way out. - */ - TCGReg scratch = tmp[--ntmp]; - tcg_out_mov(s, mov[i].src_type, scratch, mov[i].src); - mov[i].src = scratch; - - tcg_out_helper_load_regs(s, i, mov, ntmp, tmp); - tcg_out_movext1(s, &mov[i]); - return; - } + tcg_out_mov(s, mov[3].src_type, scratch, mov[3].src); + tcg_out_movext3(s, mov, mov + 1, mov + 2, parm->tmp[0]); + tcg_out_movext1_new_src(s, &mov[3], scratch); + break; } - - /* No conflicts: perform this move and continue. */ - tcg_out_movext1(s, &mov[i]); } - /* fall through for the final two moves */ + /* No conflicts: perform this move and continue. */ + tcg_out_movext1(s, &mov[3]); + /* fall through */ + + case 3: + tcg_out_movext3(s, mov, mov + 1, mov + 2, + parm->ntmp ? parm->tmp[0] : -1); + break; case 2: - tcg_out_movext2(s, mov, mov + 1, ntmp ? tmp[0] : -1); - return; + tcg_out_movext2(s, mov, mov + 1, + parm->ntmp ? parm->tmp[0] : -1); + break; case 1: tcg_out_movext1(s, mov); - return; - case 0: + break; + default: g_assert_not_reached(); } } @@ -XXX,XX +XXX,XX @@ static void tcg_out_helper_load_slots(TCGContext *s, for (i = 0; i < nmov; ++i) { mov[i].dst = tcg_target_call_iarg_regs[mov[i].dst]; } - tcg_out_helper_load_regs(s, nmov, mov, parm->ntmp, parm->tmp); + tcg_out_helper_load_regs(s, nmov, mov, parm); } static void tcg_out_helper_load_imm(TCGContext *s, unsigned slot, -- 2.34.1 Now that tcg_out_helper_load_regs is not recursive, we can merge it into its only caller, tcg_out_helper_load_slots. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 89 +++++++++++++++++++++++++------------------------------ 1 file changed, 41 insertions(+), 48 deletions(-) diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static int tcg_out_helper_stk_ofs(TCGType type, unsigned slot) return ofs; } -static void tcg_out_helper_load_regs(TCGContext *s, - unsigned nmov, TCGMovExtend *mov, - const TCGLdstHelperParam *parm) +static void tcg_out_helper_load_slots(TCGContext *s, + unsigned nmov, TCGMovExtend *mov, + const TCGLdstHelperParam *parm) { + unsigned i; TCGReg dst3; + /* + * Start from the end, storing to the stack first. + * This frees those registers, so we need not consider overlap. + */ + for (i = nmov; i-- > 0; ) { + unsigned slot = mov[i].dst; + + if (arg_slot_reg_p(slot)) { + goto found_reg; + } + + TCGReg src = mov[i].src; + TCGType dst_type = mov[i].dst_type; + MemOp dst_mo = dst_type == TCG_TYPE_I32 ? MO_32 : MO_64; + + /* The argument is going onto the stack; extend into scratch. */ + if ((mov[i].src_ext & MO_SIZE) != dst_mo) { + tcg_debug_assert(parm->ntmp != 0); + mov[i].dst = src = parm->tmp[0]; + tcg_out_movext1(s, &mov[i]); + } + + tcg_out_st(s, dst_type, src, TCG_REG_CALL_STACK, + tcg_out_helper_stk_ofs(dst_type, slot)); + } + return; + + found_reg: + /* + * The remaining arguments are in registers. + * Convert slot numbers to argument registers. + */ + nmov = i + 1; + for (i = 0; i < nmov; ++i) { + mov[i].dst = tcg_target_call_iarg_regs[mov[i].dst]; + } + switch (nmov) { case 4: /* The backend must have provided enough temps for the worst case. */ @@ -XXX,XX +XXX,XX @@ static void tcg_out_helper_load_regs(TCGContext *s, } } -static void tcg_out_helper_load_slots(TCGContext *s, - unsigned nmov, TCGMovExtend *mov, - const TCGLdstHelperParam *parm) -{ - unsigned i; - - /* - * Start from the end, storing to the stack first. - * This frees those registers, so we need not consider overlap. - */ - for (i = nmov; i-- > 0; ) { - unsigned slot = mov[i].dst; - - if (arg_slot_reg_p(slot)) { - goto found_reg; - } - - TCGReg src = mov[i].src; - TCGType dst_type = mov[i].dst_type; - MemOp dst_mo = dst_type == TCG_TYPE_I32 ? MO_32 : MO_64; - - /* The argument is going onto the stack; extend into scratch. */ - if ((mov[i].src_ext & MO_SIZE) != dst_mo) { - tcg_debug_assert(parm->ntmp != 0); - mov[i].dst = src = parm->tmp[0]; - tcg_out_movext1(s, &mov[i]); - } - - tcg_out_st(s, dst_type, src, TCG_REG_CALL_STACK, - tcg_out_helper_stk_ofs(dst_type, slot)); - } - return; - - found_reg: - /* - * The remaining arguments are in registers. - * Convert slot numbers to argument registers. - */ - nmov = i + 1; - for (i = 0; i < nmov; ++i) { - mov[i].dst = tcg_target_call_iarg_regs[mov[i].dst]; - } - tcg_out_helper_load_regs(s, nmov, mov, parm); -} - static void tcg_out_helper_load_imm(TCGContext *s, unsigned slot, TCGType type, tcg_target_long imm, const TCGLdstHelperParam *parm) -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 196 +++++++++++++++++++++++++++++++++++++++++++++--------- 1 file changed, 163 insertions(+), 33 deletions(-) diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static void * const qemu_ld_helpers[MO_SSIZE + 1] __attribute__((unused)) = { [MO_UQ] = helper_ldq_mmu, #if TCG_TARGET_REG_BITS == 64 [MO_SL] = helper_ldsl_mmu, + [MO_128] = helper_ld16_mmu, #endif }; @@ -XXX,XX +XXX,XX @@ static void * const qemu_st_helpers[MO_SIZE + 1] __attribute__((unused)) = { [MO_16] = helper_stw_mmu, [MO_32] = helper_stl_mmu, [MO_64] = helper_stq_mmu, +#if TCG_TARGET_REG_BITS == 64 + [MO_128] = helper_st16_mmu, +#endif }; TCGContext tcg_init_ctx; @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_ld64_mmu = { | dh_typemask(ptr, 4) /* uintptr_t ra */ }; +static TCGHelperInfo info_helper_ld128_mmu = { + .flags = TCG_CALL_NO_WG, + .typemask = dh_typemask(i128, 0) /* return Int128 */ + | dh_typemask(env, 1) + | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i32, 3) /* unsigned oi */ + | dh_typemask(ptr, 4) /* uintptr_t ra */ +}; + static TCGHelperInfo info_helper_st32_mmu = { .flags = TCG_CALL_NO_WG, .typemask = dh_typemask(void, 0) @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_st64_mmu = { | dh_typemask(ptr, 5) /* uintptr_t ra */ }; +static TCGHelperInfo info_helper_st128_mmu = { + .flags = TCG_CALL_NO_WG, + .typemask = dh_typemask(void, 0) + | dh_typemask(env, 1) + | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i128, 3) /* Int128 data */ + | dh_typemask(i32, 4) /* unsigned oi */ + | dh_typemask(ptr, 5) /* uintptr_t ra */ +}; + #ifdef CONFIG_TCG_INTERPRETER static ffi_type *typecode_to_ffi(int argmask) { @@ -XXX,XX +XXX,XX @@ static void tcg_context_init(unsigned max_cpus) init_call_layout(&info_helper_ld32_mmu); init_call_layout(&info_helper_ld64_mmu); + init_call_layout(&info_helper_ld128_mmu); init_call_layout(&info_helper_st32_mmu); init_call_layout(&info_helper_st64_mmu); + init_call_layout(&info_helper_st128_mmu); #ifdef CONFIG_TCG_INTERPRETER init_ffi_layouts(); @@ -XXX,XX +XXX,XX @@ static unsigned tcg_out_helper_add_mov(TCGMovExtend *mov, TCGType dst_type, TCGType src_type, TCGReg lo, TCGReg hi) { + MemOp reg_mo; + if (dst_type <= TCG_TYPE_REG) { MemOp src_ext; @@ -XXX,XX +XXX,XX @@ static unsigned tcg_out_helper_add_mov(TCGMovExtend *mov, return 1; } - assert(TCG_TARGET_REG_BITS == 32); + if (TCG_TARGET_REG_BITS == 32) { + assert(dst_type == TCG_TYPE_I64); + reg_mo = MO_32; + } else { + assert(dst_type == TCG_TYPE_I128); + reg_mo = MO_64; + } mov[0].dst = loc[HOST_BIG_ENDIAN].arg_slot; mov[0].src = lo; - mov[0].dst_type = TCG_TYPE_I32; - mov[0].src_type = TCG_TYPE_I32; - mov[0].src_ext = MO_32; + mov[0].dst_type = TCG_TYPE_REG; + mov[0].src_type = TCG_TYPE_REG; + mov[0].src_ext = reg_mo; mov[1].dst = loc[!HOST_BIG_ENDIAN].arg_slot; mov[1].src = hi; - mov[1].dst_type = TCG_TYPE_I32; - mov[1].src_type = TCG_TYPE_I32; - mov[1].src_ext = MO_32; + mov[1].dst_type = TCG_TYPE_REG; + mov[1].src_type = TCG_TYPE_REG; + mov[1].src_ext = reg_mo; return 2; } @@ -XXX,XX +XXX,XX @@ static void tcg_out_ld_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, case MO_64: info = &info_helper_ld64_mmu; break; + case MO_128: + info = &info_helper_ld128_mmu; + break; default: g_assert_not_reached(); } @@ -XXX,XX +XXX,XX @@ static void tcg_out_ld_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, tcg_out_helper_load_slots(s, nmov, mov, parm); - /* No special attention for 32 and 64-bit return values. */ - tcg_debug_assert(info->out_kind == TCG_CALL_RET_NORMAL); + switch (info->out_kind) { + case TCG_CALL_RET_NORMAL: + case TCG_CALL_RET_BY_VEC: + break; + case TCG_CALL_RET_BY_REF: + /* + * The return reference is in the first argument slot. + * We need memory in which to return: re-use the top of stack. + */ + { + int ofs_slot0 = TCG_TARGET_CALL_STACK_OFFSET; + + if (arg_slot_reg_p(0)) { + tcg_out_addi_ptr(s, tcg_target_call_iarg_regs[0], + TCG_REG_CALL_STACK, ofs_slot0); + } else { + tcg_debug_assert(parm->ntmp != 0); + tcg_out_addi_ptr(s, parm->tmp[0], + TCG_REG_CALL_STACK, ofs_slot0); + tcg_out_st(s, TCG_TYPE_PTR, parm->tmp[0], + TCG_REG_CALL_STACK, ofs_slot0); + } + } + break; + default: + g_assert_not_reached(); + } tcg_out_helper_load_common_args(s, ldst, parm, info, next_arg); } @@ -XXX,XX +XXX,XX @@ static void tcg_out_ld_helper_ret(TCGContext *s, const TCGLabelQemuLdst *ldst, bool load_sign, const TCGLdstHelperParam *parm) { + MemOp mop = get_memop(ldst->oi); TCGMovExtend mov[2]; + int ofs_slot0; - if (ldst->type <= TCG_TYPE_REG) { - MemOp mop = get_memop(ldst->oi); + switch (ldst->type) { + case TCG_TYPE_I64: + if (TCG_TARGET_REG_BITS == 32) { + break; + } + /* fall through */ + case TCG_TYPE_I32: mov[0].dst = ldst->datalo_reg; mov[0].src = tcg_target_call_oarg_reg(TCG_CALL_RET_NORMAL, 0); mov[0].dst_type = ldst->type; @@ -XXX,XX +XXX,XX @@ static void tcg_out_ld_helper_ret(TCGContext *s, const TCGLabelQemuLdst *ldst, mov[0].src_ext = mop & MO_SSIZE; } tcg_out_movext1(s, mov); - } else { - assert(TCG_TARGET_REG_BITS == 32); + return; - mov[0].dst = ldst->datalo_reg; - mov[0].src = - tcg_target_call_oarg_reg(TCG_CALL_RET_NORMAL, HOST_BIG_ENDIAN); - mov[0].dst_type = TCG_TYPE_I32; - mov[0].src_type = TCG_TYPE_I32; - mov[0].src_ext = MO_32; + case TCG_TYPE_I128: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + ofs_slot0 = TCG_TARGET_CALL_STACK_OFFSET; + switch (TCG_TARGET_CALL_RET_I128) { + case TCG_CALL_RET_NORMAL: + break; + case TCG_CALL_RET_BY_VEC: + tcg_out_st(s, TCG_TYPE_V128, + tcg_target_call_oarg_reg(TCG_CALL_RET_BY_VEC, 0), + TCG_REG_CALL_STACK, ofs_slot0); + /* fall through */ + case TCG_CALL_RET_BY_REF: + tcg_out_ld(s, TCG_TYPE_I64, ldst->datalo_reg, + TCG_REG_CALL_STACK, ofs_slot0 + 8 * HOST_BIG_ENDIAN); + tcg_out_ld(s, TCG_TYPE_I64, ldst->datahi_reg, + TCG_REG_CALL_STACK, ofs_slot0 + 8 * !HOST_BIG_ENDIAN); + return; + default: + g_assert_not_reached(); + } + break; - mov[1].dst = ldst->datahi_reg; - mov[1].src = - tcg_target_call_oarg_reg(TCG_CALL_RET_NORMAL, !HOST_BIG_ENDIAN); - mov[1].dst_type = TCG_TYPE_REG; - mov[1].src_type = TCG_TYPE_REG; - mov[1].src_ext = MO_32; - - tcg_out_movext2(s, mov, mov + 1, parm->ntmp ? parm->tmp[0] : -1); + default: + g_assert_not_reached(); } + + mov[0].dst = ldst->datalo_reg; + mov[0].src = + tcg_target_call_oarg_reg(TCG_CALL_RET_NORMAL, HOST_BIG_ENDIAN); + mov[0].dst_type = TCG_TYPE_I32; + mov[0].src_type = TCG_TYPE_I32; + mov[0].src_ext = TCG_TARGET_REG_BITS == 32 ? MO_32 : MO_64; + + mov[1].dst = ldst->datahi_reg; + mov[1].src = + tcg_target_call_oarg_reg(TCG_CALL_RET_NORMAL, !HOST_BIG_ENDIAN); + mov[1].dst_type = TCG_TYPE_REG; + mov[1].src_type = TCG_TYPE_REG; + mov[1].src_ext = TCG_TARGET_REG_BITS == 32 ? MO_32 : MO_64; + + tcg_out_movext2(s, mov, mov + 1, parm->ntmp ? parm->tmp[0] : -1); } static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, info = &info_helper_st64_mmu; data_type = TCG_TYPE_I64; break; + case MO_128: + info = &info_helper_st128_mmu; + data_type = TCG_TYPE_I128; + break; default: g_assert_not_reached(); } @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, /* Handle data argument. */ loc = &info->in[next_arg]; - n = tcg_out_helper_add_mov(mov + nmov, loc, data_type, ldst->type, - ldst->datalo_reg, ldst->datahi_reg); - next_arg += n; - nmov += n; - tcg_debug_assert(nmov <= ARRAY_SIZE(mov)); + switch (loc->kind) { + case TCG_CALL_ARG_NORMAL: + case TCG_CALL_ARG_EXTEND_U: + case TCG_CALL_ARG_EXTEND_S: + n = tcg_out_helper_add_mov(mov + nmov, loc, data_type, ldst->type, + ldst->datalo_reg, ldst->datahi_reg); + next_arg += n; + nmov += n; + tcg_out_helper_load_slots(s, nmov, mov, parm); + break; + + case TCG_CALL_ARG_BY_REF: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + tcg_debug_assert(data_type == TCG_TYPE_I128); + tcg_out_st(s, TCG_TYPE_I64, + HOST_BIG_ENDIAN ? ldst->datahi_reg : ldst->datalo_reg, + TCG_REG_CALL_STACK, arg_slot_stk_ofs(loc[0].ref_slot)); + tcg_out_st(s, TCG_TYPE_I64, + HOST_BIG_ENDIAN ? ldst->datalo_reg : ldst->datahi_reg, + TCG_REG_CALL_STACK, arg_slot_stk_ofs(loc[1].ref_slot)); + + tcg_out_helper_load_slots(s, nmov, mov, parm); + + if (arg_slot_reg_p(loc->arg_slot)) { + tcg_out_addi_ptr(s, tcg_target_call_iarg_regs[loc->arg_slot], + TCG_REG_CALL_STACK, + arg_slot_stk_ofs(loc->ref_slot)); + } else { + tcg_debug_assert(parm->ntmp != 0); + tcg_out_addi_ptr(s, parm->tmp[0], TCG_REG_CALL_STACK, + arg_slot_stk_ofs(loc->ref_slot)); + tcg_out_st(s, TCG_TYPE_PTR, parm->tmp[0], + TCG_REG_CALL_STACK, arg_slot_stk_ofs(loc->arg_slot)); + } + next_arg += 2; + break; + + default: + g_assert_not_reached(); + } - tcg_out_helper_load_slots(s, nmov, mov, parm); tcg_out_helper_load_common_args(s, ldst, parm, info, next_arg); } -- 2.34.1 Examine MemOp for atomicity and alignment, adjusting alignment as required to implement atomicity on the host. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 95 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 95 insertions(+) diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static void * const qemu_st_helpers[MO_SIZE + 1] __attribute__((unused)) = { #endif }; +typedef struct { + MemOp atom; /* lg2 bits of atomicity required */ + MemOp align; /* lg2 bits of alignment to use */ +} TCGAtomAlign; + +static TCGAtomAlign atom_and_align_for_opc(TCGContext *s, MemOp opc, + MemOp host_atom, bool allow_two_ops) + __attribute__((unused)); + TCGContext tcg_init_ctx; __thread TCGContext *tcg_ctx; @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_call(TCGContext *s, TCGOp *op) } } +/** + * atom_and_align_for_opc: + * @s: tcg context + * @opc: memory operation code + * @host_atom: MO_ATOM_{IFALIGN,WITHIN16,SUBALIGN} for host operations + * @allow_two_ops: true if we are prepared to issue two operations + * + * Return the alignment and atomicity to use for the inline fast path + * for the given memory operation. The alignment may be larger than + * that specified in @opc, and the correct alignment will be diagnosed + * by the slow path helper. + * + * If @allow_two_ops, the host is prepared to test for 2x alignment, + * and issue two loads or stores for subalignment. + */ +static TCGAtomAlign atom_and_align_for_opc(TCGContext *s, MemOp opc, + MemOp host_atom, bool allow_two_ops) +{ + MemOp align = get_alignment_bits(opc); + MemOp size = opc & MO_SIZE; + MemOp half = size ? size - 1 : 0; + MemOp atmax; + MemOp atom; + + /* When serialized, no further atomicity required. */ + if (s->gen_tb->cflags & CF_PARALLEL) { + atom = opc & MO_ATOM_MASK; + } else { + atom = MO_ATOM_NONE; + } + + switch (atom) { + case MO_ATOM_NONE: + /* The operation requires no specific atomicity. */ + atmax = MO_8; + break; + + case MO_ATOM_IFALIGN: + atmax = size; + break; + + case MO_ATOM_IFALIGN_PAIR: + atmax = half; + break; + + case MO_ATOM_WITHIN16: + atmax = size; + if (size == MO_128) { + /* Misalignment implies !within16, and therefore no atomicity. */ + } else if (host_atom != MO_ATOM_WITHIN16) { + /* The host does not implement within16, so require alignment. */ + align = MAX(align, size); + } + break; + + case MO_ATOM_WITHIN16_PAIR: + atmax = size; + /* + * Misalignment implies !within16, and therefore half atomicity. + * Any host prepared for two operations can implement this with + * half alignment. + */ + if (host_atom != MO_ATOM_WITHIN16 && allow_two_ops) { + align = MAX(align, half); + } + break; + + case MO_ATOM_SUBALIGN: + atmax = size; + if (host_atom != MO_ATOM_SUBALIGN) { + /* If unaligned but not odd, there are subobjects up to half. */ + if (allow_two_ops) { + align = MAX(align, half); + } else { + align = MAX(align, size); + } + } + break; + + default: + g_assert_not_reached(); + } + + return (TCGAtomAlign){ .atom = atmax, .align = align }; +} + /* * Similarly for qemu_ld/st slow path helpers. * We must re-implement tcg_gen_callN and tcg_reg_alloc_call simultaneously, -- 2.34.1 No change to the ultimate load/store routines yet, so some atomicity conditions not yet honored, but plumbs the change to alignment through the relevant functions. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { int index; int ofs; int seg; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); - unsigned a_mask = (1 << a_bits) - 1; + unsigned a_mask; + +#ifdef CONFIG_SOFTMMU + h->index = TCG_REG_L0; + h->ofs = 0; + h->seg = 0; +#else + *h = x86_guest_base; +#endif + h->base = addrlo; + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + a_mask = (1 << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU int cmp_ofs = is_ld ? offsetof(CPUTLBEntry, addr_read) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * copy the address and mask. For lesser alignments, check that we don't * cross pages for the complete access. */ - if (a_bits >= s_bits) { + if (a_mask >= s_mask) { tcg_out_mov(s, ttype, TCG_REG_L1, addrlo); } else { tcg_out_modrm_offset(s, OPC_LEA + trexw, TCG_REG_L1, @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* TLB Hit. */ tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_L0, TCG_REG_L0, offsetof(CPUTLBEntry, addend)); - - *h = (HostAddress) { - .base = addrlo, - .index = TCG_REG_L0, - }; #else - if (a_bits) { + if (a_mask) { ldst = new_ldst_label(s); ldst->is_ld = is_ld; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->label_ptr[0] = s->code_ptr; s->code_ptr += 4; } - - *h = x86_guest_base; - h->base = addrlo; #endif return ldst; -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.c.inc | 36 ++++++++++++++++++------------------ 1 file changed, 18 insertions(+), 18 deletions(-) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGReg base; TCGReg index; TCGType index_ext; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCGType addr_type = TARGET_LONG_BITS == 64 ? TCG_TYPE_I64 : TCG_TYPE_I32; TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); - unsigned a_mask = (1u << a_bits) - 1; + unsigned a_mask; + + h->aa = atom_and_align_for_opc(s, opc, + have_lse2 ? MO_ATOM_WITHIN16 + : MO_ATOM_IFALIGN, + false); + a_mask = (1 << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU unsigned s_bits = opc & MO_SIZE; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * bits within the address. For unaligned access, we check that we don't * cross pages using the address of the last byte of the access. */ - if (a_bits >= s_bits) { + if (a_mask >= s_mask) { x3 = addr_reg; } else { tcg_out_insn(s, 3401, ADDI, TARGET_LONG_BITS == 64, @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->label_ptr[0] = s->code_ptr; tcg_out_insn(s, 3202, B_C, TCG_COND_NE, 0); - *h = (HostAddress){ - .base = TCG_REG_X1, - .index = addr_reg, - .index_ext = addr_type - }; + h->base = TCG_REG_X1, + h->index = addr_reg; + h->index_ext = addr_type; #else if (a_mask) { ldst = new_ldst_label(s); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, } if (USE_GUEST_BASE) { - *h = (HostAddress){ - .base = TCG_REG_GUEST_BASE, - .index = addr_reg, - .index_ext = addr_type - }; + h->base = TCG_REG_GUEST_BASE; + h->index = addr_reg; + h->index_ext = addr_type; } else { - *h = (HostAddress){ - .base = addr_reg, - .index = TCG_REG_XZR, - .index_ext = TCG_TYPE_I64 - }; + h->base = addr_reg; + h->index = TCG_REG_XZR; + h->index_ext = TCG_TYPE_I64; } #endif -- 2.34.1 No change to the ultimate load/store routines yet, so some atomicity conditions not yet honored, but plumbs the change to alignment through the relevant functions. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/arm/tcg-target.c.inc | 39 ++++++++++++++++++++++----------------- 1 file changed, 22 insertions(+), 17 deletions(-) diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGReg base; int index; bool index_scratch; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - MemOp a_bits = get_alignment_bits(opc); - unsigned a_mask = (1 << a_bits) - 1; + unsigned a_mask; + +#ifdef CONFIG_SOFTMMU + *h = (HostAddress){ + .cond = COND_AL, + .base = addrlo, + .index = TCG_REG_R1, + .index_scratch = true, + }; +#else + *h = (HostAddress){ + .cond = COND_AL, + .base = addrlo, + .index = guest_base ? TCG_REG_GUEST_BASE : -1, + .index_scratch = false, + }; +#endif + + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + a_mask = (1 << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU int mem_index = get_mmuidx(oi); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, if (TARGET_LONG_BITS == 64) { tcg_out_dat_reg(s, COND_EQ, ARITH_CMP, 0, TCG_REG_R3, addrhi, 0); } - - *h = (HostAddress){ - .cond = COND_AL, - .base = addrlo, - .index = TCG_REG_R1, - .index_scratch = true, - }; #else if (a_mask) { ldst = new_ldst_label(s); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->addrlo_reg = addrlo; ldst->addrhi_reg = addrhi; - /* We are expecting a_bits to max out at 7 */ + /* We are expecting alignment to max out at 7 */ tcg_debug_assert(a_mask <= 0xff); /* tst addr, #mask */ tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addrlo, a_mask); } - - *h = (HostAddress){ - .cond = COND_AL, - .base = addrlo, - .index = guest_base ? TCG_REG_GUEST_BASE : -1, - .index_scratch = false, - }; #endif return ldst; -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/loongarch64/tcg-target.c.inc | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) typedef struct { TCGReg base; TCGReg index; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); + MemOp a_bits; + + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + a_bits = h->aa.align; #ifdef CONFIG_SOFTMMU unsigned s_bits = opc & MO_SIZE; -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/mips/tcg-target.c.inc | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l) typedef struct { TCGReg base; - MemOp align; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); + MemOp a_bits; unsigned s_bits = opc & MO_SIZE; - unsigned a_mask = (1 << a_bits) - 1; + unsigned a_mask; TCGReg base; + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + a_bits = h->aa.align; + a_mask = (1 << a_bits) - 1; + #ifdef CONFIG_SOFTMMU unsigned s_mask = (1 << s_bits) - 1; int mem_index = get_mmuidx(oi); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, #endif h->base = base; - h->align = a_bits; return ldst; } @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg datalo, TCGReg datahi, ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, true); - if (use_mips32r6_instructions || h.align >= (opc & MO_SIZE)) { + if (use_mips32r6_instructions || h.aa.align >= (opc & MO_SIZE)) { tcg_out_qemu_ld_direct(s, datalo, datahi, h.base, opc, data_type); } else { tcg_out_qemu_ld_unalign(s, datalo, datahi, h.base, opc, data_type); @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg datalo, TCGReg datahi, ldst = prepare_host_addr(s, &h, addrlo, addrhi, oi, false); - if (use_mips32r6_instructions || h.align >= (opc & MO_SIZE)) { + if (use_mips32r6_instructions || h.aa.align >= (opc & MO_SIZE)) { tcg_out_qemu_st_direct(s, datalo, datahi, h.base, opc); } else { tcg_out_qemu_st_unalign(s, datalo, datahi, h.base, opc); -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/ppc/tcg-target.c.inc | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-) diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) typedef struct { TCGReg base; TCGReg index; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); + MemOp a_bits; + + /* + * Book II, Section 1.4, Single-Copy Atomicity, specifies: + * + * Before 3.0, "An access that is not atomic is performed as a set of + * smaller disjoint atomic accesses. In general, the number and alignment + * of these accesses are implementation-dependent." Thus MO_ATOM_IFALIGN. + * + * As of 3.0, "the non-atomic access is performed as described in + * the corresponding list", which matches MO_ATOM_SUBALIGN. + */ + h->aa = atom_and_align_for_opc(s, opc, + have_isa_3_00 ? MO_ATOM_SUBALIGN + : MO_ATOM_IFALIGN, + false); + a_bits = h->aa.align; #ifdef CONFIG_SOFTMMU int mem_index = get_mmuidx(oi); -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/riscv/tcg-target.c.inc | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); - unsigned a_mask = (1u << a_bits) - 1; + TCGAtomAlign aa; + unsigned a_mask; + + aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + a_mask = (1u << aa.align) - 1; #ifdef CONFIG_SOFTMMU unsigned s_bits = opc & MO_SIZE; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, * cross pages using the address of the last byte of the access. */ addr_adj = addr_reg; - if (a_bits < s_bits) { + if (a_mask < s_mask) { addr_adj = TCG_REG_TMP0; tcg_out_opc_imm(s, TARGET_LONG_BITS == 32 ? OPC_ADDIW : OPC_ADDI, addr_adj, addr_reg, s_mask - a_mask); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, ldst->oi = oi; ldst->addrlo_reg = addr_reg; - /* We are expecting a_bits max 7, so we can always use andi. */ - tcg_debug_assert(a_bits < 12); + /* We are expecting alignment max 7, so we can always use andi. */ + tcg_debug_assert(a_mask == sextreg(a_mask, 0, 12)); tcg_out_opc_imm(s, OPC_ANDI, TCG_REG_TMP1, addr_reg, a_mask); ldst->label_ptr[0] = s->code_ptr; -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/s390x/tcg-target.c.inc | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef struct { TCGReg base; TCGReg index; int disp; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); - unsigned a_mask = (1u << a_bits) - 1; + unsigned a_mask; + + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + a_mask = (1 << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU unsigned s_bits = opc & MO_SIZE; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * bits within the address. For unaligned access, we check that we don't * cross pages using the address of the last byte of the access. */ - a_off = (a_bits >= s_bits ? 0 : s_mask - a_mask); + a_off = (a_mask >= s_mask ? 0 : s_mask - a_mask); tlb_mask = (uint64_t)TARGET_PAGE_MASK | a_mask; if (a_off == 0) { tgen_andi_risbg(s, TCG_REG_R0, addr_reg, tlb_mask); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->addrlo_reg = addr_reg; /* We are expecting a_bits to max out at 7, much lower than TMLL. */ - tcg_debug_assert(a_bits < 16); + tcg_debug_assert(a_mask <= 0xffff); tcg_out_insn(s, RI, TMLL, addr_reg, a_mask); tcg_out16(s, RI_BRC | (7 << 4)); /* CC in {1,2,3} */ -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/sparc64/tcg-target.c.inc | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-) diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) typedef struct { TCGReg base; TCGReg index; + TCGAtomAlign aa; } HostAddress; bool tcg_target_has_memory_bswap(MemOp memop) @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - unsigned a_bits = get_alignment_bits(opc); - unsigned s_bits = opc & MO_SIZE; + MemOp s_bits = opc & MO_SIZE; unsigned a_mask; /* We don't support unaligned accesses. */ - a_bits = MAX(a_bits, s_bits); - a_mask = (1u << a_bits) - 1; + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + h->aa.align = MAX(h->aa.align, s_bits); + a_mask = (1u << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU int mem_index = get_mmuidx(oi); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, cc = TARGET_LONG_BITS == 64 ? BPCC_XCC : BPCC_ICC; tcg_out_bpcc0(s, COND_NE, BPCC_PN | cc, 0); #else - if (a_bits != s_bits) { - /* - * Test for at least natural alignment, and defer - * everything else to the helper functions. - */ + /* + * If the size equals the required alignment, we can skip the test + * and allow host SIGBUS to deliver SIGBUS to the guest. + * Otherwise, test for at least natural alignment and defer + * everything else to the helper functions. + */ + if (s_bits != get_alignment_bits(opc)) { tcg_debug_assert(check_fit_tl(a_mask, 13)); tcg_out_arithi(s, TCG_REG_G0, addr_reg, a_mask, ARITH_ANDCC); -- 2.34.1 Use the fpu to perform 64-bit loads and stores. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 44 +++++++++++++++++++++++++++++++++------ 1 file changed, 38 insertions(+), 6 deletions(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_target_const_match(int64_t val, TCGType type, int ct) #define OPC_GRP5 (0xff) #define OPC_GRP14 (0x73 | P_EXT | P_DATA16) +#define OPC_ESCDF (0xdf) +#define ESCDF_FILD_m64 5 +#define ESCDF_FISTP_m64 7 + /* Group 1 opcode extensions for 0x80-0x83. These are also used as modifiers for OPC_ARITH. */ #define ARITH_ADD 0 @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_ld_direct(TCGContext *s, TCGReg datalo, TCGReg datahi, datalo = datahi; datahi = t; } - if (h.base == datalo || h.index == datalo) { + if (h.aa.atom == MO_64) { + /* + * Atomicity requires that we use use a single 8-byte load. + * For simplicity and code size, always use the FPU for this. + * Similar insns using SSE/AVX are merely larger. + * Load from memory in one go, then store back to the stack, + * from whence we can load into the correct integer regs. + */ + tcg_out_modrm_sib_offset(s, OPC_ESCDF + h.seg, ESCDF_FILD_m64, + h.base, h.index, 0, h.ofs); + tcg_out_modrm_offset(s, OPC_ESCDF, ESCDF_FISTP_m64, TCG_REG_ESP, 0); + tcg_out_modrm_offset(s, movop, datalo, TCG_REG_ESP, 0); + tcg_out_modrm_offset(s, movop, datahi, TCG_REG_ESP, 4); + } else if (h.base == datalo || h.index == datalo) { tcg_out_modrm_sib_offset(s, OPC_LEA, datahi, h.base, h.index, 0, h.ofs); tcg_out_modrm_offset(s, movop + h.seg, datalo, datahi, 0); @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st_direct(TCGContext *s, TCGReg datalo, TCGReg datahi, if (TCG_TARGET_REG_BITS == 64) { tcg_out_modrm_sib_offset(s, movop + P_REXW + h.seg, datalo, h.base, h.index, 0, h.ofs); + break; + } + if (use_movbe) { + TCGReg t = datalo; + datalo = datahi; + datahi = t; + } + if (h.aa.atom == MO_64) { + /* + * Atomicity requires that we use use one 8-byte store. + * For simplicity, and code size, always use the FPU for this. + * Similar insns using SSE/AVX are merely larger. + * Assemble the 8-byte quantity in required endianness + * on the stack, load to coproc unit, and store. + */ + tcg_out_modrm_offset(s, movop, datalo, TCG_REG_ESP, 0); + tcg_out_modrm_offset(s, movop, datahi, TCG_REG_ESP, 4); + tcg_out_modrm_offset(s, OPC_ESCDF, ESCDF_FILD_m64, TCG_REG_ESP, 0); + tcg_out_modrm_sib_offset(s, OPC_ESCDF + h.seg, ESCDF_FISTP_m64, + h.base, h.index, 0, h.ofs); } else { - if (use_movbe) { - TCGReg t = datalo; - datalo = datahi; - datahi = t; - } tcg_out_modrm_sib_offset(s, movop + h.seg, datalo, h.base, h.index, 0, h.ofs); tcg_out_modrm_sib_offset(s, movop + h.seg, datahi, -- 2.34.1 Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.h | 3 +- tcg/i386/tcg-target.c.inc | 181 +++++++++++++++++++++++++++++++++++++- 2 files changed, 180 insertions(+), 4 deletions(-) diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.h +++ b/tcg/i386/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_atomic16; #define TCG_TARGET_HAS_qemu_st8_i32 1 #endif -#define TCG_TARGET_HAS_qemu_ldst_i128 0 +#define TCG_TARGET_HAS_qemu_ldst_i128 \ + (TCG_TARGET_REG_BITS == 64 && have_atomic16) /* We do not support older SSE systems, only beginning with AVX1. */ #define TCG_TARGET_HAS_v64 have_avx1 diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static const int tcg_target_reg_alloc_order[] = { #endif }; +#define TCG_TMP_VEC TCG_REG_XMM5 + static const int tcg_target_call_iarg_regs[] = { #if TCG_TARGET_REG_BITS == 64 #if defined(_WIN64) @@ -XXX,XX +XXX,XX @@ static bool tcg_target_const_match(int64_t val, TCGType type, int ct) #define OPC_PCMPGTW (0x65 | P_EXT | P_DATA16) #define OPC_PCMPGTD (0x66 | P_EXT | P_DATA16) #define OPC_PCMPGTQ (0x37 | P_EXT38 | P_DATA16) +#define OPC_PEXTRD (0x16 | P_EXT3A | P_DATA16) +#define OPC_PINSRD (0x22 | P_EXT3A | P_DATA16) #define OPC_PMAXSB (0x3c | P_EXT38 | P_DATA16) #define OPC_PMAXSW (0xee | P_EXT | P_DATA16) #define OPC_PMAXSD (0x3d | P_EXT38 | P_DATA16) @@ -XXX,XX +XXX,XX @@ typedef struct { bool tcg_target_has_memory_bswap(MemOp memop) { - return have_movbe; + TCGAtomAlign aa; + + if (!have_movbe) { + return false; + } + if ((memop & MO_SIZE) <= MO_64) { + return true; + } + + /* + * Reject 16-byte memop with 16-byte atomicity, i.e. VMOVDQA, + * but do allow a pair of 64-bit operations, i.e. MOVBEQ. + */ + aa = atom_and_align_for_opc(tcg_ctx, memop, MO_ATOM_IFALIGN, true); + return aa.atom <= MO_64; } /* @@ -XXX,XX +XXX,XX @@ static const TCGLdstHelperParam ldst_helper_param = { static const TCGLdstHelperParam ldst_helper_param = { }; #endif +static void tcg_out_vec_to_pair(TCGContext *s, TCGType type, + TCGReg l, TCGReg h, TCGReg v) +{ + int rexw = type == TCG_TYPE_I32 ? 0 : P_REXW; + + /* vpmov{d,q} %v, %l */ + tcg_out_vex_modrm(s, OPC_MOVD_EyVy + rexw, v, 0, l); + /* vpextr{d,q} $1, %v, %h */ + tcg_out_vex_modrm(s, OPC_PEXTRD + rexw, v, 0, h); + tcg_out8(s, 1); +} + +static void tcg_out_pair_to_vec(TCGContext *s, TCGType type, + TCGReg v, TCGReg l, TCGReg h) +{ + int rexw = type == TCG_TYPE_I32 ? 0 : P_REXW; + + /* vmov{d,q} %l, %v */ + tcg_out_vex_modrm(s, OPC_MOVD_VyEy + rexw, v, 0, l); + /* vpinsr{d,q} $1, %h, %v, %v */ + tcg_out_vex_modrm(s, OPC_PINSRD + rexw, v, v, h); + tcg_out8(s, 1); +} + /* * Generate code for the slow path for a load at the end of block */ @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); + MemOp s_bits = opc & MO_SIZE; unsigned a_mask; #ifdef CONFIG_SOFTMMU @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, *h = x86_guest_base; #endif h->base = addrlo; - h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, s_bits == MO_128); a_mask = (1 << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCGType tlbtype = TCG_TYPE_I32; int trexw = 0, hrexw = 0, tlbrexw = 0; unsigned mem_index = get_mmuidx(oi); - unsigned s_bits = opc & MO_SIZE; unsigned s_mask = (1 << s_bits) - 1; target_ulong tlb_mask; @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_ld_direct(TCGContext *s, TCGReg datalo, TCGReg datahi, h.base, h.index, 0, h.ofs + 4); } break; + + case MO_128: + { + TCGLabel *l1 = NULL, *l2 = NULL; + bool use_pair = h.aa.atom < MO_128; + + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + + if (!use_pair) { + tcg_debug_assert(!use_movbe); + /* + * Atomicity requires that we use use VMOVDQA. + * If we've already checked for 16-byte alignment, that's all + * we need. If we arrive here with lesser alignment, then we + * have determined that less than 16-byte alignment can be + * satisfied with two 8-byte loads. + */ + if (h.aa.align < MO_128) { + use_pair = true; + l1 = gen_new_label(); + l2 = gen_new_label(); + + tcg_out_testi(s, h.base, 15); + tcg_out_jxx(s, JCC_JNE, l2, true); + } + + tcg_out_vex_modrm_sib_offset(s, OPC_MOVDQA_VxWx + h.seg, + TCG_TMP_VEC, 0, + h.base, h.index, 0, h.ofs); + tcg_out_vec_to_pair(s, TCG_TYPE_I64, datalo, + datahi, TCG_TMP_VEC); + + if (use_pair) { + tcg_out_jxx(s, JCC_JMP, l1, true); + tcg_out_label(s, l2); + } + } + if (use_pair) { + if (use_movbe) { + TCGReg t = datalo; + datalo = datahi; + datahi = t; + } + if (h.base == datalo || h.index == datalo) { + tcg_out_modrm_sib_offset(s, OPC_LEA + P_REXW, datahi, + h.base, h.index, 0, h.ofs); + tcg_out_modrm_offset(s, movop + P_REXW + h.seg, + datalo, datahi, 0); + tcg_out_modrm_offset(s, movop + P_REXW + h.seg, + datahi, datahi, 8); + } else { + tcg_out_modrm_sib_offset(s, movop + P_REXW + h.seg, datalo, + h.base, h.index, 0, h.ofs); + tcg_out_modrm_sib_offset(s, movop + P_REXW + h.seg, datahi, + h.base, h.index, 0, h.ofs + 8); + } + } + if (l1) { + tcg_out_label(s, l1); + } + } + break; + default: g_assert_not_reached(); } @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st_direct(TCGContext *s, TCGReg datalo, TCGReg datahi, h.base, h.index, 0, h.ofs + 4); } break; + + case MO_128: + { + TCGLabel *l1 = NULL, *l2 = NULL; + bool use_pair = h.aa.atom < MO_128; + + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + + if (!use_pair) { + tcg_debug_assert(!use_movbe); + /* + * Atomicity requires that we use use VMOVDQA. + * If we've already checked for 16-byte alignment, that's all + * we need. If we arrive here with lesser alignment, then we + * have determined that less that 16-byte alignment can be + * satisfied with two 8-byte loads. + */ + if (h.aa.align < MO_128) { + use_pair = true; + l1 = gen_new_label(); + l2 = gen_new_label(); + + tcg_out_testi(s, h.base, 15); + tcg_out_jxx(s, JCC_JNE, l2, true); + } + + tcg_out_pair_to_vec(s, TCG_TYPE_I64, TCG_TMP_VEC, + datalo, datahi); + tcg_out_vex_modrm_sib_offset(s, OPC_MOVDQA_WxVx + h.seg, + TCG_TMP_VEC, 0, + h.base, h.index, 0, h.ofs); + + if (use_pair) { + tcg_out_jxx(s, JCC_JMP, l1, true); + tcg_out_label(s, l2); + } + } + if (use_pair) { + if (use_movbe) { + TCGReg t = datalo; + datalo = datahi; + datahi = t; + } + tcg_out_modrm_sib_offset(s, movop + P_REXW + h.seg, datalo, + h.base, h.index, 0, h.ofs); + tcg_out_modrm_sib_offset(s, movop + P_REXW + h.seg, datahi, + h.base, h.index, 0, h.ofs + 8); + } + if (l1) { + tcg_out_label(s, l1); + } + } + break; + default: g_assert_not_reached(); } @@ -XXX,XX +XXX,XX @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_qemu_ld(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); } break; + case INDEX_op_qemu_ld_i128: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + tcg_out_qemu_ld(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); + break; case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st8_i32: if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { @@ -XXX,XX +XXX,XX @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_qemu_st(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); } break; + case INDEX_op_qemu_st_i128: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + tcg_out_qemu_st(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); + break; OP_32_64(mulu2): tcg_out_modrm(s, OPC_GRP3_Ev + rexw, EXT3_MUL, args[3]); @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) : TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? C_O0_I3(L, L, L) : C_O0_I4(L, L, L, L)); + case INDEX_op_qemu_ld_i128: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + return C_O2_I1(r, r, L); + case INDEX_op_qemu_st_i128: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + return C_O0_I3(L, L, L); + case INDEX_op_brcond2_i32: return C_O0_I4(r, r, ri, ri); @@ -XXX,XX +XXX,XX @@ static void tcg_target_init(TCGContext *s) s->reserved_regs = 0; tcg_regset_set_reg(s->reserved_regs, TCG_REG_CALL_STACK); + tcg_regset_set_reg(s->reserved_regs, TCG_TMP_VEC); #ifdef _WIN64 /* These are call saved, and we don't save them, so don't use them. */ tcg_regset_set_reg(s->reserved_regs, TCG_REG_XMM6); -- 2.34.1 We will need to allocate a second general-purpose temporary. Rename the existing temps to add a distinguishing number. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.c.inc | 50 ++++++++++++++++++------------------ 1 file changed, 25 insertions(+), 25 deletions(-) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGReg tcg_target_call_oarg_reg(TCGCallReturnKind kind, int slot) bool have_lse; bool have_lse2; -#define TCG_REG_TMP TCG_REG_X30 -#define TCG_VEC_TMP TCG_REG_V31 +#define TCG_REG_TMP0 TCG_REG_X30 +#define TCG_VEC_TMP0 TCG_REG_V31 #ifndef CONFIG_SOFTMMU /* Note that XZR cannot be encoded in the address base register slot, @@ -XXX,XX +XXX,XX @@ static bool tcg_out_dup_vec(TCGContext *s, TCGType type, unsigned vece, static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece, TCGReg r, TCGReg base, intptr_t offset) { - TCGReg temp = TCG_REG_TMP; + TCGReg temp = TCG_REG_TMP0; if (offset < -0xffffff || offset > 0xffffff) { tcg_out_movi(s, TCG_TYPE_PTR, temp, offset); @@ -XXX,XX +XXX,XX @@ static void tcg_out_ldst(TCGContext *s, AArch64Insn insn, TCGReg rd, } /* Worst-case scenario, move offset to temp register, use reg offset. */ - tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP, offset); - tcg_out_ldst_r(s, insn, rd, rn, TCG_TYPE_I64, TCG_REG_TMP); + tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP0, offset); + tcg_out_ldst_r(s, insn, rd, rn, TCG_TYPE_I64, TCG_REG_TMP0); } static bool tcg_out_mov(TCGContext *s, TCGType type, TCGReg ret, TCGReg arg) @@ -XXX,XX +XXX,XX @@ static void tcg_out_call_int(TCGContext *s, const tcg_insn_unit *target) if (offset == sextract64(offset, 0, 26)) { tcg_out_insn(s, 3206, BL, offset); } else { - tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP, (intptr_t)target); - tcg_out_insn(s, 3207, BLR, TCG_REG_TMP); + tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP0, (intptr_t)target); + tcg_out_insn(s, 3207, BLR, TCG_REG_TMP0); } } @@ -XXX,XX +XXX,XX @@ static void tcg_out_addsub2(TCGContext *s, TCGType ext, TCGReg rl, AArch64Insn insn; if (rl == ah || (!const_bh && rl == bh)) { - rl = TCG_REG_TMP; + rl = TCG_REG_TMP0; } if (const_bl) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_addsub2(TCGContext *s, TCGType ext, TCGReg rl, possibility of adding 0+const in the low part, and the immediate add instructions encode XSP not XZR. Don't try anything more elaborate here than loading another zero. */ - al = TCG_REG_TMP; + al = TCG_REG_TMP0; tcg_out_movi(s, ext, al, 0); } tcg_out_insn_3401(s, insn, ext, rl, al, bl); @@ -XXX,XX +XXX,XX @@ static void tcg_out_cltz(TCGContext *s, TCGType ext, TCGReg d, { TCGReg a1 = a0; if (is_ctz) { - a1 = TCG_REG_TMP; + a1 = TCG_REG_TMP0; tcg_out_insn(s, 3507, RBIT, ext, a1, a0); } if (const_b && b == (ext ? 64 : 32)) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_cltz(TCGContext *s, TCGType ext, TCGReg d, AArch64Insn sel = I3506_CSEL; tcg_out_cmp(s, ext, a0, 0, 1); - tcg_out_insn(s, 3507, CLZ, ext, TCG_REG_TMP, a1); + tcg_out_insn(s, 3507, CLZ, ext, TCG_REG_TMP0, a1); if (const_b) { if (b == -1) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_cltz(TCGContext *s, TCGType ext, TCGReg d, b = d; } } - tcg_out_insn_3506(s, sel, ext, d, TCG_REG_TMP, b, TCG_COND_NE); + tcg_out_insn_3506(s, sel, ext, d, TCG_REG_TMP0, b, TCG_COND_NE); } } @@ -XXX,XX +XXX,XX @@ bool tcg_target_has_memory_bswap(MemOp memop) } static const TCGLdstHelperParam ldst_helper_param = { - .ntmp = 1, .tmp = { TCG_REG_TMP } + .ntmp = 1, .tmp = { TCG_REG_TMP0 } }; static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) @@ -XXX,XX +XXX,XX @@ static void tcg_out_goto_tb(TCGContext *s, int which) set_jmp_insn_offset(s, which); tcg_out32(s, I3206_B); - tcg_out_insn(s, 3207, BR, TCG_REG_TMP); + tcg_out_insn(s, 3207, BR, TCG_REG_TMP0); set_jmp_reset_offset(s, which); } @@ -XXX,XX +XXX,XX @@ void tb_target_set_jmp_target(const TranslationBlock *tb, int n, ptrdiff_t i_offset = i_addr - jmp_rx; /* Note that we asserted this in range in tcg_out_goto_tb. */ - insn = deposit32(I3305_LDR | TCG_REG_TMP, 5, 19, i_offset >> 2); + insn = deposit32(I3305_LDR | TCG_REG_TMP0, 5, 19, i_offset >> 2); } qatomic_set((uint32_t *)jmp_rw, insn); flush_idcache_range(jmp_rx, jmp_rw, 4); @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_rem_i64: case INDEX_op_rem_i32: - tcg_out_insn(s, 3508, SDIV, ext, TCG_REG_TMP, a1, a2); - tcg_out_insn(s, 3509, MSUB, ext, a0, TCG_REG_TMP, a2, a1); + tcg_out_insn(s, 3508, SDIV, ext, TCG_REG_TMP0, a1, a2); + tcg_out_insn(s, 3509, MSUB, ext, a0, TCG_REG_TMP0, a2, a1); break; case INDEX_op_remu_i64: case INDEX_op_remu_i32: - tcg_out_insn(s, 3508, UDIV, ext, TCG_REG_TMP, a1, a2); - tcg_out_insn(s, 3509, MSUB, ext, a0, TCG_REG_TMP, a2, a1); + tcg_out_insn(s, 3508, UDIV, ext, TCG_REG_TMP0, a1, a2); + tcg_out_insn(s, 3509, MSUB, ext, a0, TCG_REG_TMP0, a2, a1); break; case INDEX_op_shl_i64: @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, if (c2) { tcg_out_rotl(s, ext, a0, a1, a2); } else { - tcg_out_insn(s, 3502, SUB, 0, TCG_REG_TMP, TCG_REG_XZR, a2); - tcg_out_insn(s, 3508, RORV, ext, a0, a1, TCG_REG_TMP); + tcg_out_insn(s, 3502, SUB, 0, TCG_REG_TMP0, TCG_REG_XZR, a2); + tcg_out_insn(s, 3508, RORV, ext, a0, a1, TCG_REG_TMP0); } break; @@ -XXX,XX +XXX,XX @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, break; } } - tcg_out_dupi_vec(s, type, MO_8, TCG_VEC_TMP, 0); - a2 = TCG_VEC_TMP; + tcg_out_dupi_vec(s, type, MO_8, TCG_VEC_TMP0, 0); + a2 = TCG_VEC_TMP0; } if (is_scalar) { insn = cmp_scalar_insn[cond]; @@ -XXX,XX +XXX,XX @@ static void tcg_target_init(TCGContext *s) s->reserved_regs = 0; tcg_regset_set_reg(s->reserved_regs, TCG_REG_SP); tcg_regset_set_reg(s->reserved_regs, TCG_REG_FP); - tcg_regset_set_reg(s->reserved_regs, TCG_REG_TMP); tcg_regset_set_reg(s->reserved_regs, TCG_REG_X18); /* platform register */ - tcg_regset_set_reg(s->reserved_regs, TCG_VEC_TMP); + tcg_regset_set_reg(s->reserved_regs, TCG_REG_TMP0); + tcg_regset_set_reg(s->reserved_regs, TCG_VEC_TMP0); } /* Saving pairs: (X19, X20) .. (X27, X28), (X29(fp), X30(lr)). */ -- 2.34.1 Use LDXP+STXP when LSE2 is not present and 16-byte atomicity is required, and LDP/STP otherwise. This requires allocating a second general-purpose temporary, as Rs cannot overlap Rn in STXP. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target-con-set.h | 2 + tcg/aarch64/tcg-target.h | 11 +- tcg/aarch64/tcg-target.c.inc | 179 ++++++++++++++++++++++++++++++- 3 files changed, 189 insertions(+), 3 deletions(-) diff --git a/tcg/aarch64/tcg-target-con-set.h b/tcg/aarch64/tcg-target-con-set.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target-con-set.h +++ b/tcg/aarch64/tcg-target-con-set.h @@ -XXX,XX +XXX,XX @@ C_O0_I2(lZ, l) C_O0_I2(r, rA) C_O0_I2(rZ, r) C_O0_I2(w, r) +C_O0_I3(lZ, lZ, l) C_O1_I1(r, l) C_O1_I1(r, r) C_O1_I1(w, r) @@ -XXX,XX +XXX,XX @@ C_O1_I2(w, w, wO) C_O1_I2(w, w, wZ) C_O1_I3(w, w, w, w) C_O1_I4(r, r, rA, rZ, rZ) +C_O2_I1(r, r, l) C_O2_I4(r, r, rZ, rZ, rA, rMZ) diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.h +++ b/tcg/aarch64/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_lse2; #define TCG_TARGET_HAS_muluh_i64 1 #define TCG_TARGET_HAS_mulsh_i64 1 -#define TCG_TARGET_HAS_qemu_ldst_i128 0 +/* + * Without FEAT_LSE2, we must use LDXP+STXP to implement atomic 128-bit load, + * which requires writable pages. We must defer to the helper for user-only, + * but in system mode all ram is writable for the host. + */ +#ifdef CONFIG_USER_ONLY +#define TCG_TARGET_HAS_qemu_ldst_i128 have_lse2 +#else +#define TCG_TARGET_HAS_qemu_ldst_i128 1 +#endif #define TCG_TARGET_HAS_v64 1 #define TCG_TARGET_HAS_v128 1 diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ bool have_lse; bool have_lse2; #define TCG_REG_TMP0 TCG_REG_X30 +#define TCG_REG_TMP1 TCG_REG_X17 #define TCG_VEC_TMP0 TCG_REG_V31 #ifndef CONFIG_SOFTMMU @@ -XXX,XX +XXX,XX @@ typedef enum { I3305_LDR_v64 = 0x5c000000, I3305_LDR_v128 = 0x9c000000, + /* Load/store exclusive. */ + I3306_LDXP = 0xc8600000, + I3306_STXP = 0xc8200000, + /* Load/store register. Described here as 3.3.12, but the helper that emits them can transform to 3.3.10 or 3.3.13. */ I3312_STRB = 0x38000000 | LDST_ST << 22 | MO_8 << 30, @@ -XXX,XX +XXX,XX @@ typedef enum { I3406_ADR = 0x10000000, I3406_ADRP = 0x90000000, + /* Add/subtract extended register instructions. */ + I3501_ADD = 0x0b200000, + /* Add/subtract shifted register instructions (without a shift). */ I3502_ADD = 0x0b000000, I3502_ADDS = 0x2b000000, @@ -XXX,XX +XXX,XX @@ static void tcg_out_insn_3305(TCGContext *s, AArch64Insn insn, tcg_out32(s, insn | (imm19 & 0x7ffff) << 5 | rt); } +static void tcg_out_insn_3306(TCGContext *s, AArch64Insn insn, TCGReg rs, + TCGReg rt, TCGReg rt2, TCGReg rn) +{ + tcg_out32(s, insn | rs << 16 | rt2 << 10 | rn << 5 | rt); +} + static void tcg_out_insn_3201(TCGContext *s, AArch64Insn insn, TCGType ext, TCGReg rt, int imm19) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_insn_3406(TCGContext *s, AArch64Insn insn, tcg_out32(s, insn | (disp & 3) << 29 | (disp & 0x1ffffc) << (5 - 2) | rd); } +static inline void tcg_out_insn_3501(TCGContext *s, AArch64Insn insn, + TCGType sf, TCGReg rd, TCGReg rn, + TCGReg rm, int opt, int imm3) +{ + tcg_out32(s, insn | sf << 31 | rm << 16 | opt << 13 | + imm3 << 10 | rn << 5 | rd); +} + /* This function is for both 3.5.2 (Add/Subtract shifted register), for the rare occasion when we actually want to supply a shift amount. */ static inline void tcg_out_insn_3502S(TCGContext *s, AArch64Insn insn, @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCGType addr_type = TARGET_LONG_BITS == 64 ? TCG_TYPE_I64 : TCG_TYPE_I32; TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); + MemOp s_bits = opc & MO_SIZE; unsigned a_mask; h->aa = atom_and_align_for_opc(s, opc, have_lse2 ? MO_ATOM_WITHIN16 : MO_ATOM_IFALIGN, - false); + s_bits == MO_128); a_mask = (1 << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU - unsigned s_bits = opc & MO_SIZE; unsigned s_mask = (1u << s_bits) - 1; unsigned mem_index = get_mmuidx(oi); TCGReg x3; @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data_reg, TCGReg addr_reg, } } +static TCGLabelQemuLdst * +prepare_host_addr_base_only(TCGContext *s, HostAddress *h, TCGReg addr_reg, + MemOpIdx oi, bool is_ld) +{ + TCGLabelQemuLdst *ldst; + + ldst = prepare_host_addr(s, h, addr_reg, oi, true); + + /* Compose the final address, as LDP/STP have no indexing. */ + if (h->index != TCG_REG_XZR) { + tcg_out_insn(s, 3501, ADD, TCG_TYPE_I64, TCG_REG_TMP0, + h->base, h->index, + h->index_ext == TCG_TYPE_I32 ? MO_32 : MO_64, 0); + h->base = TCG_REG_TMP0; + h->index = TCG_REG_XZR; + h->index_ext = TCG_TYPE_I64; + } + + return ldst; +} + +static void tcg_out_qemu_ld128(TCGContext *s, TCGReg datalo, TCGReg datahi, + TCGReg addr_reg, MemOpIdx oi) +{ + TCGLabelQemuLdst *ldst; + HostAddress h; + + ldst = prepare_host_addr_base_only(s, &h, addr_reg, oi, true); + + if (h.aa.atom < MO_128 || have_lse2) { + tcg_out_insn(s, 3314, LDP, datalo, datahi, h.base, 0, 0, 0); + } else { + TCGLabel *l0, *l1 = NULL; + + /* + * 16-byte atomicity without LSE2 requires LDXP+STXP loop: + * 1: ldxp lo,hi,[addr] + * stxp tmp1,lo,hi,[addr] + * cbnz tmp1, 1b + * + * If we have already checked for 16-byte alignment, that's all + * we need. Otherwise we have determined that misaligned atomicity + * may be handled with two 8-byte loads. + */ + if (h.aa.align < MO_128) { + /* + * TODO: align should be MO_64, so we only need test bit 3, + * which means we could use TBNZ instead of AND+CBNE. + */ + l1 = gen_new_label(); + tcg_out_logicali(s, I3404_ANDI, 0, TCG_REG_TMP1, addr_reg, 15); + tcg_out_brcond(s, TCG_TYPE_I32, TCG_COND_NE, + TCG_REG_TMP1, 0, 1, l1); + } + + l0 = gen_new_label(); + tcg_out_label(s, l0); + + tcg_out_insn(s, 3306, LDXP, TCG_REG_XZR, datalo, datahi, h.base); + tcg_out_insn(s, 3306, STXP, TCG_REG_TMP1, datalo, datahi, h.base); + tcg_out_brcond(s, TCG_TYPE_I32, TCG_COND_NE, TCG_REG_TMP1, 0, 1, l0); + + if (l1) { + TCGLabel *l2 = gen_new_label(); + tcg_out_goto_label(s, l2); + + tcg_out_label(s, l1); + tcg_out_insn(s, 3314, LDP, datalo, datahi, h.base, 0, 0, 0); + + tcg_out_label(s, l2); + } + } + + if (ldst) { + ldst->type = TCG_TYPE_I128; + ldst->datalo_reg = datalo; + ldst->datahi_reg = datahi; + ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); + } +} + +static void tcg_out_qemu_st128(TCGContext *s, TCGReg datalo, TCGReg datahi, + TCGReg addr_reg, MemOpIdx oi) +{ + TCGLabelQemuLdst *ldst; + HostAddress h; + + ldst = prepare_host_addr_base_only(s, &h, addr_reg, oi, false); + + if (h.aa.atom < MO_128 || have_lse2) { + tcg_out_insn(s, 3314, STP, datalo, datahi, h.base, 0, 0, 0); + } else { + TCGLabel *l0, *l1 = NULL; + + /* + * 16-byte atomicity without LSE2 requires LDXP+STXP loop: + * 1: ldxp xzr,tmp1,[addr] + * stxp tmp1,lo,hi,[addr] + * cbnz tmp1, 1b + * + * If we have already checked for 16-byte alignment, that's all + * we need. Otherwise we have determined that misaligned atomicity + * may be handled with two 8-byte stores. + */ + if (h.aa.align < MO_128) { + /* + * TODO: align should be MO_64, so we only need test bit 3, + * which means we could use TBNZ instead of AND+CBNE. + */ + l1 = gen_new_label(); + tcg_out_logicali(s, I3404_ANDI, 0, TCG_REG_TMP1, addr_reg, 15); + tcg_out_brcond(s, TCG_TYPE_I32, TCG_COND_NE, + TCG_REG_TMP1, 0, 1, l1); + } + + l0 = gen_new_label(); + tcg_out_label(s, l0); + + tcg_out_insn(s, 3306, LDXP, TCG_REG_XZR, + TCG_REG_XZR, TCG_REG_TMP1, h.base); + tcg_out_insn(s, 3306, STXP, TCG_REG_TMP1, datalo, datahi, h.base); + tcg_out_brcond(s, TCG_TYPE_I32, TCG_COND_NE, TCG_REG_TMP1, 0, 1, l0); + + if (l1) { + TCGLabel *l2 = gen_new_label(); + tcg_out_goto_label(s, l2); + + tcg_out_label(s, l1); + tcg_out_insn(s, 3314, STP, datalo, datahi, h.base, 0, 0, 0); + + tcg_out_label(s, l2); + } + } + + if (ldst) { + ldst->type = TCG_TYPE_I128; + ldst->datalo_reg = datalo; + ldst->datahi_reg = datahi; + ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); + } +} + static const tcg_insn_unit *tb_ret_addr; static void tcg_out_exit_tb(TCGContext *s, uintptr_t a0) @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, REG0(0), a1, a2, ext); break; + case INDEX_op_qemu_ld_i128: + tcg_out_qemu_ld128(s, a0, a1, a2, args[3]); + break; + case INDEX_op_qemu_st_i128: + tcg_out_qemu_st128(s, REG0(0), REG0(1), a2, args[3]); + break; case INDEX_op_bswap64_i64: tcg_out_rev(s, TCG_TYPE_I64, MO_64, a0, a1); @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_qemu_ld_i32: case INDEX_op_qemu_ld_i64: return C_O1_I1(r, l); + case INDEX_op_qemu_ld_i128: + return C_O2_I1(r, r, l); case INDEX_op_qemu_st_i32: case INDEX_op_qemu_st_i64: return C_O0_I2(lZ, l); + case INDEX_op_qemu_st_i128: + return C_O0_I3(lZ, lZ, l); case INDEX_op_deposit_i32: case INDEX_op_deposit_i64: @@ -XXX,XX +XXX,XX @@ static void tcg_target_init(TCGContext *s) tcg_regset_set_reg(s->reserved_regs, TCG_REG_FP); tcg_regset_set_reg(s->reserved_regs, TCG_REG_X18); /* platform register */ tcg_regset_set_reg(s->reserved_regs, TCG_REG_TMP0); + tcg_regset_set_reg(s->reserved_regs, TCG_REG_TMP1); tcg_regset_set_reg(s->reserved_regs, TCG_VEC_TMP0); } -- 2.34.1 Use LQ/STQ with ISA v2.07, and 16-byte atomicity is required. Note that these instructions do not require 16-byte alignment. Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/ppc/tcg-target-con-set.h | 2 + tcg/ppc/tcg-target-con-str.h | 1 + tcg/ppc/tcg-target.h | 3 +- tcg/ppc/tcg-target.c.inc | 115 +++++++++++++++++++++++++++++++---- 4 files changed, 108 insertions(+), 13 deletions(-) diff --git a/tcg/ppc/tcg-target-con-set.h b/tcg/ppc/tcg-target-con-set.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target-con-set.h +++ b/tcg/ppc/tcg-target-con-set.h @@ -XXX,XX +XXX,XX @@ C_O0_I2(r, r) C_O0_I2(r, ri) C_O0_I2(v, r) C_O0_I3(r, r, r) +C_O0_I3(o, m, r) C_O0_I4(r, r, ri, ri) C_O0_I4(r, r, r, r) C_O1_I1(r, r) @@ -XXX,XX +XXX,XX @@ C_O1_I3(v, v, v, v) C_O1_I4(r, r, ri, rZ, rZ) C_O1_I4(r, r, r, ri, ri) C_O2_I1(r, r, r) +C_O2_I1(o, m, r) C_O2_I2(r, r, r, r) C_O2_I4(r, r, rI, rZM, r, r) C_O2_I4(r, r, r, r, rI, rZM) diff --git a/tcg/ppc/tcg-target-con-str.h b/tcg/ppc/tcg-target-con-str.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target-con-str.h +++ b/tcg/ppc/tcg-target-con-str.h @@ -XXX,XX +XXX,XX @@ * REGS(letter, register_mask) */ REGS('r', ALL_GENERAL_REGS) +REGS('o', ALL_GENERAL_REGS & 0xAAAAAAAAu) /* odd registers */ REGS('v', ALL_VECTOR_REGS) /* diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.h +++ b/tcg/ppc/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_vsx; #define TCG_TARGET_HAS_mulsh_i64 1 #endif -#define TCG_TARGET_HAS_qemu_ldst_i128 0 +#define TCG_TARGET_HAS_qemu_ldst_i128 \ + (TCG_TARGET_REG_BITS == 64 && have_isa_2_07) /* * While technically Altivec could support V64, it has no 64-bit store diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool tcg_target_const_match(int64_t val, TCGType type, int ct) #define B OPCD( 18) #define BC OPCD( 16) + #define LBZ OPCD( 34) #define LHZ OPCD( 40) #define LHA OPCD( 42) #define LWZ OPCD( 32) #define LWZUX XO31( 55) -#define STB OPCD( 38) -#define STH OPCD( 44) -#define STW OPCD( 36) - -#define STD XO62( 0) -#define STDU XO62( 1) -#define STDX XO31(149) - #define LD XO58( 0) #define LDX XO31( 21) #define LDU XO58( 1) #define LDUX XO31( 53) #define LWA XO58( 2) #define LWAX XO31(341) +#define LQ OPCD( 56) + +#define STB OPCD( 38) +#define STH OPCD( 44) +#define STW OPCD( 36) +#define STD XO62( 0) +#define STDU XO62( 1) +#define STDX XO31(149) +#define STQ XO62( 2) #define ADDIC OPCD( 12) #define ADDI OPCD( 14) @@ -XXX,XX +XXX,XX @@ typedef struct { bool tcg_target_has_memory_bswap(MemOp memop) { - return true; + TCGAtomAlign aa; + + if ((memop & MO_SIZE) <= MO_64) { + return true; + } + + /* + * Reject 16-byte memop with 16-byte atomicity, + * but do allow a pair of 64-bit operations. + */ + aa = atom_and_align_for_opc(tcg_ctx, memop, MO_ATOM_IFALIGN, true); + return aa.atom <= MO_64; } /* @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); - MemOp a_bits; + MemOp a_bits, s_bits; /* * Book II, Section 1.4, Single-Copy Atomicity, specifies: @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * As of 3.0, "the non-atomic access is performed as described in * the corresponding list", which matches MO_ATOM_SUBALIGN. */ + s_bits = opc & MO_SIZE; h->aa = atom_and_align_for_opc(s, opc, have_isa_3_00 ? MO_ATOM_SUBALIGN : MO_ATOM_IFALIGN, - false); + s_bits == MO_128); a_bits = h->aa.align; #ifdef CONFIG_SOFTMMU @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, int fast_off = TLB_MASK_TABLE_OFS(mem_index); int mask_off = fast_off + offsetof(CPUTLBDescFast, mask); int table_off = fast_off + offsetof(CPUTLBDescFast, table); - unsigned s_bits = opc & MO_SIZE; ldst = new_ldst_label(s); ldst->is_ld = is_ld; @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg datalo, TCGReg datahi, } } +static TCGLabelQemuLdst * +prepare_host_addr_index_only(TCGContext *s, HostAddress *h, TCGReg addr_reg, + MemOpIdx oi, bool is_ld) +{ + TCGLabelQemuLdst *ldst; + + ldst = prepare_host_addr(s, h, addr_reg, -1, oi, true); + + /* Compose the final address, as LQ/STQ have no indexing. */ + if (h->base != 0) { + tcg_out32(s, ADD | TAB(TCG_REG_TMP1, h->base, h->index)); + h->index = TCG_REG_TMP1; + h->base = 0; + } + + return ldst; +} + +static void tcg_out_qemu_ldst_i128(TCGContext *s, TCGReg datalo, TCGReg datahi, + TCGReg addr_reg, MemOpIdx oi, bool is_ld) +{ + TCGLabelQemuLdst *ldst; + HostAddress h; + bool need_bswap; + uint32_t insn; + + ldst = prepare_host_addr_index_only(s, &h, addr_reg, oi, is_ld); + need_bswap = get_memop(oi) & MO_BSWAP; + + if (h.aa.atom == MO_128) { + tcg_debug_assert(!need_bswap); + tcg_debug_assert(datalo & 1); + tcg_debug_assert(datahi == datalo - 1); + insn = is_ld ? LQ : STQ; + tcg_out32(s, insn | TAI(datahi, h.index, 0)); + } else { + TCGReg d1, d2; + + if (HOST_BIG_ENDIAN ^ need_bswap) { + d1 = datahi, d2 = datalo; + } else { + d1 = datalo, d2 = datahi; + } + + if (need_bswap) { + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R0, 8); + insn = is_ld ? LDBRX : STDBRX; + tcg_out32(s, insn | TAB(d1, 0, h.index)); + tcg_out32(s, insn | TAB(d2, h.index, TCG_REG_R0)); + } else { + insn = is_ld ? LD : STD; + tcg_out32(s, insn | TAI(d1, h.index, 0)); + tcg_out32(s, insn | TAI(d2, h.index, 8)); + } + } + + if (ldst) { + ldst->type = TCG_TYPE_I128; + ldst->datalo_reg = datalo; + ldst->datahi_reg = datahi; + ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); + } +} + static void tcg_out_nop_fill(tcg_insn_unit *p, int count) { int i; @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, args[4], TCG_TYPE_I64); } break; + case INDEX_op_qemu_ld_i128: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], true); + break; + case INDEX_op_qemu_st_i32: if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { tcg_out_qemu_st(s, args[0], -1, args[1], -1, @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, args[4], TCG_TYPE_I64); } break; + case INDEX_op_qemu_st_i128: + tcg_debug_assert(TCG_TARGET_REG_BITS == 64); + tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], false); + break; case INDEX_op_setcond_i32: tcg_out_setcond(s, TCG_TYPE_I32, args[3], args[0], args[1], args[2], @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) : TARGET_LONG_BITS == 32 ? C_O0_I3(r, r, r) : C_O0_I4(r, r, r, r)); + case INDEX_op_qemu_ld_i128: + return C_O2_I1(o, m, r); + case INDEX_op_qemu_st_i128: + return C_O0_I3(o, m, r); + case INDEX_op_add_vec: case INDEX_op_sub_vec: case INDEX_op_mul_vec: -- 2.34.1 Use LPQ/STPQ when 16-byte atomicity is required. Note that these instructions require 16-byte alignment. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/s390x/tcg-target-con-set.h | 2 + tcg/s390x/tcg-target.h | 2 +- tcg/s390x/tcg-target.c.inc | 103 ++++++++++++++++++++++++++++++++- 3 files changed, 103 insertions(+), 4 deletions(-) diff --git a/tcg/s390x/tcg-target-con-set.h b/tcg/s390x/tcg-target-con-set.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target-con-set.h +++ b/tcg/s390x/tcg-target-con-set.h @@ -XXX,XX +XXX,XX @@ C_O0_I2(r, r) C_O0_I2(r, ri) C_O0_I2(r, rA) C_O0_I2(v, r) +C_O0_I3(o, m, r) C_O1_I1(r, r) C_O1_I1(v, r) C_O1_I1(v, v) @@ -XXX,XX +XXX,XX @@ C_O1_I2(v, v, v) C_O1_I3(v, v, v, v) C_O1_I4(r, r, ri, rI, r) C_O1_I4(r, r, rA, rI, r) +C_O2_I1(o, m, r) C_O2_I2(o, m, 0, r) C_O2_I2(o, m, r, r) C_O2_I3(o, m, 0, 1, r) diff --git a/tcg/s390x/tcg-target.h b/tcg/s390x/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.h +++ b/tcg/s390x/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern uint64_t s390_facilities[3]; #define TCG_TARGET_HAS_muluh_i64 0 #define TCG_TARGET_HAS_mulsh_i64 0 -#define TCG_TARGET_HAS_qemu_ldst_i128 0 +#define TCG_TARGET_HAS_qemu_ldst_i128 1 #define TCG_TARGET_HAS_v64 HAVE_FACILITY(VECTOR) #define TCG_TARGET_HAS_v128 HAVE_FACILITY(VECTOR) diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef enum S390Opcode { RXY_LLGF = 0xe316, RXY_LLGH = 0xe391, RXY_LMG = 0xeb04, + RXY_LPQ = 0xe38f, RXY_LRV = 0xe31e, RXY_LRVG = 0xe30f, RXY_LRVH = 0xe31f, @@ -XXX,XX +XXX,XX @@ typedef enum S390Opcode { RXY_STG = 0xe324, RXY_STHY = 0xe370, RXY_STMG = 0xeb24, + RXY_STPQ = 0xe38e, RXY_STRV = 0xe33e, RXY_STRVG = 0xe32f, RXY_STRVH = 0xe33f, @@ -XXX,XX +XXX,XX @@ typedef struct { bool tcg_target_has_memory_bswap(MemOp memop) { - return true; + TCGAtomAlign aa; + + if ((memop & MO_SIZE) <= MO_64) { + return true; + } + + /* + * Reject 16-byte memop with 16-byte atomicity, + * but do allow a pair of 64-bit operations. + */ + aa = atom_and_align_for_opc(tcg_ctx, memop, MO_ATOM_IFALIGN, true); + return aa.atom <= MO_64; } static void tcg_out_qemu_ld_direct(TCGContext *s, MemOp opc, TCGReg data, @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, { TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); + MemOp s_bits = opc & MO_SIZE; unsigned a_mask; - h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, false); + h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, s_bits == MO_128); a_mask = (1 << h->aa.align) - 1; #ifdef CONFIG_SOFTMMU - unsigned s_bits = opc & MO_SIZE; unsigned s_mask = (1 << s_bits) - 1; int mem_index = get_mmuidx(oi); int fast_off = TLB_MASK_TABLE_OFS(mem_index); @@ -XXX,XX +XXX,XX @@ static void tcg_out_qemu_st(TCGContext* s, TCGReg data_reg, TCGReg addr_reg, } } +static void tcg_out_qemu_ldst_i128(TCGContext *s, TCGReg datalo, TCGReg datahi, + TCGReg addr_reg, MemOpIdx oi, bool is_ld) +{ + TCGLabel *l1 = NULL, *l2 = NULL; + TCGLabelQemuLdst *ldst; + HostAddress h; + bool need_bswap; + bool use_pair; + S390Opcode insn; + + ldst = prepare_host_addr(s, &h, addr_reg, oi, is_ld); + + use_pair = h.aa.atom < MO_128; + need_bswap = get_memop(oi) & MO_BSWAP; + + if (!use_pair) { + /* + * Atomicity requires we use LPQ. If we've already checked for + * 16-byte alignment, that's all we need. If we arrive with + * lesser alignment, we have determined that less than 16-byte + * alignment can be satisfied with two 8-byte loads. + */ + if (h.aa.align < MO_128) { + use_pair = true; + l1 = gen_new_label(); + l2 = gen_new_label(); + + tcg_out_insn(s, RI, TMLL, addr_reg, 15); + tgen_branch(s, 7, l1); /* CC in {1,2,3} */ + } + + tcg_debug_assert(!need_bswap); + tcg_debug_assert(datalo & 1); + tcg_debug_assert(datahi == datalo - 1); + insn = is_ld ? RXY_LPQ : RXY_STPQ; + tcg_out_insn_RXY(s, insn, datahi, h.base, h.index, h.disp); + + if (use_pair) { + tgen_branch(s, S390_CC_ALWAYS, l2); + tcg_out_label(s, l1); + } + } + if (use_pair) { + TCGReg d1, d2; + + if (need_bswap) { + d1 = datalo, d2 = datahi; + insn = is_ld ? RXY_LRVG : RXY_STRVG; + } else { + d1 = datahi, d2 = datalo; + insn = is_ld ? RXY_LG : RXY_STG; + } + + if (h.base == d1 || h.index == d1) { + tcg_out_insn(s, RXY, LAY, TCG_TMP0, h.base, h.index, h.disp); + h.base = TCG_TMP0; + h.index = TCG_REG_NONE; + h.disp = 0; + } + tcg_out_insn_RXY(s, insn, d1, h.base, h.index, h.disp); + tcg_out_insn_RXY(s, insn, d2, h.base, h.index, h.disp + 8); + } + if (l2) { + tcg_out_label(s, l2); + } + + if (ldst) { + ldst->type = TCG_TYPE_I128; + ldst->datalo_reg = datalo; + ldst->datahi_reg = datahi; + ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); + } +} + static void tcg_out_exit_tb(TCGContext *s, uintptr_t a0) { /* Reuse the zeroing that exists for goto_ptr. */ @@ -XXX,XX +XXX,XX @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_qemu_st_i64: tcg_out_qemu_st(s, args[0], args[1], args[2], TCG_TYPE_I64); break; + case INDEX_op_qemu_ld_i128: + tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], true); + break; + case INDEX_op_qemu_st_i128: + tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], false); + break; case INDEX_op_ld16s_i64: tcg_out_mem(s, 0, RXY_LGH, args[0], args[1], TCG_REG_NONE, args[2]); @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_qemu_st_i64: case INDEX_op_qemu_st_i32: return C_O0_I2(r, r); + case INDEX_op_qemu_ld_i128: + return C_O2_I1(o, m, r); + case INDEX_op_qemu_st_i128: + return C_O0_I3(o, m, r); case INDEX_op_deposit_i32: case INDEX_op_deposit_i64: -- 2.34.1 Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg-op-ldst.c | 1006 +++++++++++++++++++++++++++++++++++++++++++++ tcg/tcg-op.c | 974 ------------------------------------------- tcg/meson.build | 1 + 3 files changed, 1007 insertions(+), 974 deletions(-) create mode 100644 tcg/tcg-op-ldst.c diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ +/* + * Tiny Code Generator for QEMU + * + * Copyright (c) 2008 Fabrice Bellard + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +#include "qemu/osdep.h" +#include "exec/exec-all.h" +#include "tcg/tcg.h" +#include "tcg/tcg-temp-internal.h" +#include "tcg/tcg-op.h" +#include "tcg/tcg-mo.h" +#include "exec/plugin-gen.h" +#include "tcg-internal.h" + + +static inline MemOp tcg_canonicalize_memop(MemOp op, bool is64, bool st) +{ + /* Trigger the asserts within as early as possible. */ + unsigned a_bits = get_alignment_bits(op); + + /* Prefer MO_ALIGN+MO_XX over MO_ALIGN_XX+MO_XX */ + if (a_bits == (op & MO_SIZE)) { + op = (op & ~MO_AMASK) | MO_ALIGN; + } + + switch (op & MO_SIZE) { + case MO_8: + op &= ~MO_BSWAP; + break; + case MO_16: + break; + case MO_32: + if (!is64) { + op &= ~MO_SIGN; + } + break; + case MO_64: + if (is64) { + op &= ~MO_SIGN; + break; + } + /* fall through */ + default: + g_assert_not_reached(); + } + if (st) { + op &= ~MO_SIGN; + } + return op; +} + +static void gen_ldst_i32(TCGOpcode opc, TCGv_i32 val, TCGv addr, + MemOp memop, TCGArg idx) +{ + MemOpIdx oi = make_memop_idx(memop, idx); +#if TARGET_LONG_BITS == 32 + tcg_gen_op3i_i32(opc, val, addr, oi); +#else + if (TCG_TARGET_REG_BITS == 32) { + tcg_gen_op4i_i32(opc, val, TCGV_LOW(addr), TCGV_HIGH(addr), oi); + } else { + tcg_gen_op3(opc, tcgv_i32_arg(val), tcgv_i64_arg(addr), oi); + } +#endif +} + +static void gen_ldst_i64(TCGOpcode opc, TCGv_i64 val, TCGv addr, + MemOp memop, TCGArg idx) +{ + MemOpIdx oi = make_memop_idx(memop, idx); +#if TARGET_LONG_BITS == 32 + if (TCG_TARGET_REG_BITS == 32) { + tcg_gen_op4i_i32(opc, TCGV_LOW(val), TCGV_HIGH(val), addr, oi); + } else { + tcg_gen_op3(opc, tcgv_i64_arg(val), tcgv_i32_arg(addr), oi); + } +#else + if (TCG_TARGET_REG_BITS == 32) { + tcg_gen_op5i_i32(opc, TCGV_LOW(val), TCGV_HIGH(val), + TCGV_LOW(addr), TCGV_HIGH(addr), oi); + } else { + tcg_gen_op3i_i64(opc, val, addr, oi); + } +#endif +} + +static void tcg_gen_req_mo(TCGBar type) +{ +#ifdef TCG_GUEST_DEFAULT_MO + type &= TCG_GUEST_DEFAULT_MO; +#endif + type &= ~TCG_TARGET_DEFAULT_MO; + if (type) { + tcg_gen_mb(type | TCG_BAR_SC); + } +} + +static inline TCGv plugin_prep_mem_callbacks(TCGv vaddr) +{ +#ifdef CONFIG_PLUGIN + if (tcg_ctx->plugin_insn != NULL) { + /* Save a copy of the vaddr for use after a load. */ + TCGv temp = tcg_temp_new(); + tcg_gen_mov_tl(temp, vaddr); + return temp; + } +#endif + return vaddr; +} + +static void plugin_gen_mem_callbacks(TCGv vaddr, MemOpIdx oi, + enum qemu_plugin_mem_rw rw) +{ +#ifdef CONFIG_PLUGIN + if (tcg_ctx->plugin_insn != NULL) { + qemu_plugin_meminfo_t info = make_plugin_meminfo(oi, rw); + plugin_gen_empty_mem_callback(vaddr, info); + tcg_temp_free(vaddr); + } +#endif +} + +void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) +{ + MemOp orig_memop; + MemOpIdx oi; + + tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); + memop = tcg_canonicalize_memop(memop, 0, 0); + oi = make_memop_idx(memop, idx); + + orig_memop = memop; + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + memop &= ~MO_BSWAP; + /* The bswap primitive benefits from zero-extended input. */ + if ((memop & MO_SSIZE) == MO_SW) { + memop &= ~MO_SIGN; + } + } + + addr = plugin_prep_mem_callbacks(addr); + gen_ldst_i32(INDEX_op_qemu_ld_i32, val, addr, memop, idx); + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); + + if ((orig_memop ^ memop) & MO_BSWAP) { + switch (orig_memop & MO_SIZE) { + case MO_16: + tcg_gen_bswap16_i32(val, val, (orig_memop & MO_SIGN + ? TCG_BSWAP_IZ | TCG_BSWAP_OS + : TCG_BSWAP_IZ | TCG_BSWAP_OZ)); + break; + case MO_32: + tcg_gen_bswap32_i32(val, val); + break; + default: + g_assert_not_reached(); + } + } +} + +void tcg_gen_qemu_st_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) +{ + TCGv_i32 swap = NULL; + MemOpIdx oi; + + tcg_gen_req_mo(TCG_MO_LD_ST | TCG_MO_ST_ST); + memop = tcg_canonicalize_memop(memop, 0, 1); + oi = make_memop_idx(memop, idx); + + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + swap = tcg_temp_ebb_new_i32(); + switch (memop & MO_SIZE) { + case MO_16: + tcg_gen_bswap16_i32(swap, val, 0); + break; + case MO_32: + tcg_gen_bswap32_i32(swap, val); + break; + default: + g_assert_not_reached(); + } + val = swap; + memop &= ~MO_BSWAP; + } + + addr = plugin_prep_mem_callbacks(addr); + if (TCG_TARGET_HAS_qemu_st8_i32 && (memop & MO_SIZE) == MO_8) { + gen_ldst_i32(INDEX_op_qemu_st8_i32, val, addr, memop, idx); + } else { + gen_ldst_i32(INDEX_op_qemu_st_i32, val, addr, memop, idx); + } + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); + + if (swap) { + tcg_temp_free_i32(swap); + } +} + +void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) +{ + MemOp orig_memop; + MemOpIdx oi; + + if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { + tcg_gen_qemu_ld_i32(TCGV_LOW(val), addr, idx, memop); + if (memop & MO_SIGN) { + tcg_gen_sari_i32(TCGV_HIGH(val), TCGV_LOW(val), 31); + } else { + tcg_gen_movi_i32(TCGV_HIGH(val), 0); + } + return; + } + + tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); + memop = tcg_canonicalize_memop(memop, 1, 0); + oi = make_memop_idx(memop, idx); + + orig_memop = memop; + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + memop &= ~MO_BSWAP; + /* The bswap primitive benefits from zero-extended input. */ + if ((memop & MO_SIGN) && (memop & MO_SIZE) < MO_64) { + memop &= ~MO_SIGN; + } + } + + addr = plugin_prep_mem_callbacks(addr); + gen_ldst_i64(INDEX_op_qemu_ld_i64, val, addr, memop, idx); + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); + + if ((orig_memop ^ memop) & MO_BSWAP) { + int flags = (orig_memop & MO_SIGN + ? TCG_BSWAP_IZ | TCG_BSWAP_OS + : TCG_BSWAP_IZ | TCG_BSWAP_OZ); + switch (orig_memop & MO_SIZE) { + case MO_16: + tcg_gen_bswap16_i64(val, val, flags); + break; + case MO_32: + tcg_gen_bswap32_i64(val, val, flags); + break; + case MO_64: + tcg_gen_bswap64_i64(val, val); + break; + default: + g_assert_not_reached(); + } + } +} + +void tcg_gen_qemu_st_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) +{ + TCGv_i64 swap = NULL; + MemOpIdx oi; + + if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { + tcg_gen_qemu_st_i32(TCGV_LOW(val), addr, idx, memop); + return; + } + + tcg_gen_req_mo(TCG_MO_LD_ST | TCG_MO_ST_ST); + memop = tcg_canonicalize_memop(memop, 1, 1); + oi = make_memop_idx(memop, idx); + + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + swap = tcg_temp_ebb_new_i64(); + switch (memop & MO_SIZE) { + case MO_16: + tcg_gen_bswap16_i64(swap, val, 0); + break; + case MO_32: + tcg_gen_bswap32_i64(swap, val, 0); + break; + case MO_64: + tcg_gen_bswap64_i64(swap, val); + break; + default: + g_assert_not_reached(); + } + val = swap; + memop &= ~MO_BSWAP; + } + + addr = plugin_prep_mem_callbacks(addr); + gen_ldst_i64(INDEX_op_qemu_st_i64, val, addr, memop, idx); + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); + + if (swap) { + tcg_temp_free_i64(swap); + } +} + +/* + * Return true if @mop, without knowledge of the pointer alignment, + * does not require 16-byte atomicity, and it would be adventagous + * to avoid a call to a helper function. + */ +static bool use_two_i64_for_i128(MemOp mop) +{ +#ifdef CONFIG_SOFTMMU + /* Two softmmu tlb lookups is larger than one function call. */ + return false; +#else + /* + * For user-only, two 64-bit operations may well be smaller than a call. + * Determine if that would be legal for the requested atomicity. + */ + switch (mop & MO_ATOM_MASK) { + case MO_ATOM_NONE: + case MO_ATOM_IFALIGN_PAIR: + return true; + case MO_ATOM_IFALIGN: + case MO_ATOM_SUBALIGN: + case MO_ATOM_WITHIN16: + case MO_ATOM_WITHIN16_PAIR: + /* In a serialized context, no atomicity is required. */ + return !(tcg_ctx->gen_tb->cflags & CF_PARALLEL); + default: + g_assert_not_reached(); + } +#endif +} + +static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) +{ + MemOp mop_1 = orig, mop_2; + + tcg_debug_assert((orig & MO_SIZE) == MO_128); + tcg_debug_assert((orig & MO_SIGN) == 0); + + /* Reduce the size to 64-bit. */ + mop_1 = (mop_1 & ~MO_SIZE) | MO_64; + + /* Retain the alignment constraints of the original. */ + switch (orig & MO_AMASK) { + case MO_UNALN: + case MO_ALIGN_2: + case MO_ALIGN_4: + mop_2 = mop_1; + break; + case MO_ALIGN_8: + /* Prefer MO_ALIGN+MO_64 to MO_ALIGN_8+MO_64. */ + mop_1 = (mop_1 & ~MO_AMASK) | MO_ALIGN; + mop_2 = mop_1; + break; + case MO_ALIGN: + /* Second has 8-byte alignment; first has 16-byte alignment. */ + mop_2 = mop_1; + mop_1 = (mop_1 & ~MO_AMASK) | MO_ALIGN_16; + break; + case MO_ALIGN_16: + case MO_ALIGN_32: + case MO_ALIGN_64: + /* Second has 8-byte alignment; first retains original. */ + mop_2 = (mop_1 & ~MO_AMASK) | MO_ALIGN; + break; + default: + g_assert_not_reached(); + } + + /* Use a memory ordering implemented by the host. */ + if ((orig & MO_BSWAP) && !tcg_target_has_memory_bswap(mop_1)) { + mop_1 &= ~MO_BSWAP; + mop_2 &= ~MO_BSWAP; + } + + ret[0] = mop_1; + ret[1] = mop_2; +} + +#if TARGET_LONG_BITS == 64 +#define tcg_temp_ebb_new tcg_temp_ebb_new_i64 +#else +#define tcg_temp_ebb_new tcg_temp_ebb_new_i32 +#endif + +void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) +{ + const MemOpIdx oi = make_memop_idx(memop, idx); + + tcg_debug_assert((memop & MO_SIZE) == MO_128); + tcg_debug_assert((memop & MO_SIGN) == 0); + + tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); + addr = plugin_prep_mem_callbacks(addr); + + /* TODO: For now, force 32-bit hosts to use the helper. */ + if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { + TCGv_i64 lo, hi; + TCGArg addr_arg; + MemOpIdx adj_oi; + bool need_bswap = false; + + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + lo = TCGV128_HIGH(val); + hi = TCGV128_LOW(val); + adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); + need_bswap = true; + } else { + lo = TCGV128_LOW(val); + hi = TCGV128_HIGH(val); + adj_oi = oi; + } + +#if TARGET_LONG_BITS == 32 + addr_arg = tcgv_i32_arg(addr); +#else + addr_arg = tcgv_i64_arg(addr); +#endif + tcg_gen_op4ii_i64(INDEX_op_qemu_ld_i128, lo, hi, addr_arg, adj_oi); + + if (need_bswap) { + tcg_gen_bswap64_i64(lo, lo); + tcg_gen_bswap64_i64(hi, hi); + } + } else if (use_two_i64_for_i128(memop)) { + MemOp mop[2]; + TCGv addr_p8; + TCGv_i64 x, y; + + canonicalize_memop_i128_as_i64(mop, memop); + + /* + * Since there are no global TCGv_i128, there is no visible state + * changed if the second load faults. Load directly into the two + * subwords. + */ + if ((memop & MO_BSWAP) == MO_LE) { + x = TCGV128_LOW(val); + y = TCGV128_HIGH(val); + } else { + x = TCGV128_HIGH(val); + y = TCGV128_LOW(val); + } + + gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, mop[0], idx); + + if ((mop[0] ^ memop) & MO_BSWAP) { + tcg_gen_bswap64_i64(x, x); + } + + addr_p8 = tcg_temp_ebb_new(); + tcg_gen_addi_tl(addr_p8, addr, 8); + gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, mop[1], idx); + tcg_temp_free(addr_p8); + + if ((mop[0] ^ memop) & MO_BSWAP) { + tcg_gen_bswap64_i64(y, y); + } + } else { + gen_helper_ld_i128(val, cpu_env, addr, tcg_constant_i32(oi)); + } + + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); +} + +void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) +{ + const MemOpIdx oi = make_memop_idx(memop, idx); + + tcg_debug_assert((memop & MO_SIZE) == MO_128); + tcg_debug_assert((memop & MO_SIGN) == 0); + + tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); + addr = plugin_prep_mem_callbacks(addr); + + /* TODO: For now, force 32-bit hosts to use the helper. */ + + if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { + TCGv_i64 lo, hi; + TCGArg addr_arg; + MemOpIdx adj_oi; + bool need_bswap = false; + + if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { + lo = tcg_temp_new_i64(); + hi = tcg_temp_new_i64(); + tcg_gen_bswap64_i64(lo, TCGV128_HIGH(val)); + tcg_gen_bswap64_i64(hi, TCGV128_LOW(val)); + adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); + need_bswap = true; + } else { + lo = TCGV128_LOW(val); + hi = TCGV128_HIGH(val); + adj_oi = oi; + } + +#if TARGET_LONG_BITS == 32 + addr_arg = tcgv_i32_arg(addr); +#else + addr_arg = tcgv_i64_arg(addr); +#endif + tcg_gen_op4ii_i64(INDEX_op_qemu_st_i128, lo, hi, addr_arg, adj_oi); + + if (need_bswap) { + tcg_temp_free_i64(lo); + tcg_temp_free_i64(hi); + } + } else if (use_two_i64_for_i128(memop)) { + MemOp mop[2]; + TCGv addr_p8; + TCGv_i64 x, y; + + canonicalize_memop_i128_as_i64(mop, memop); + + if ((memop & MO_BSWAP) == MO_LE) { + x = TCGV128_LOW(val); + y = TCGV128_HIGH(val); + } else { + x = TCGV128_HIGH(val); + y = TCGV128_LOW(val); + } + + addr_p8 = tcg_temp_ebb_new(); + if ((mop[0] ^ memop) & MO_BSWAP) { + TCGv_i64 t = tcg_temp_ebb_new_i64(); + + tcg_gen_bswap64_i64(t, x); + gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr, mop[0], idx); + tcg_gen_bswap64_i64(t, y); + tcg_gen_addi_tl(addr_p8, addr, 8); + gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr_p8, mop[1], idx); + tcg_temp_free_i64(t); + } else { + gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, mop[0], idx); + tcg_gen_addi_tl(addr_p8, addr, 8); + gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, mop[1], idx); + } + tcg_temp_free(addr_p8); + } else { + gen_helper_st_i128(cpu_env, addr, val, tcg_constant_i32(oi)); + } + + plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); +} + +static void tcg_gen_ext_i32(TCGv_i32 ret, TCGv_i32 val, MemOp opc) +{ + switch (opc & MO_SSIZE) { + case MO_SB: + tcg_gen_ext8s_i32(ret, val); + break; + case MO_UB: + tcg_gen_ext8u_i32(ret, val); + break; + case MO_SW: + tcg_gen_ext16s_i32(ret, val); + break; + case MO_UW: + tcg_gen_ext16u_i32(ret, val); + break; + default: + tcg_gen_mov_i32(ret, val); + break; + } +} + +static void tcg_gen_ext_i64(TCGv_i64 ret, TCGv_i64 val, MemOp opc) +{ + switch (opc & MO_SSIZE) { + case MO_SB: + tcg_gen_ext8s_i64(ret, val); + break; + case MO_UB: + tcg_gen_ext8u_i64(ret, val); + break; + case MO_SW: + tcg_gen_ext16s_i64(ret, val); + break; + case MO_UW: + tcg_gen_ext16u_i64(ret, val); + break; + case MO_SL: + tcg_gen_ext32s_i64(ret, val); + break; + case MO_UL: + tcg_gen_ext32u_i64(ret, val); + break; + default: + tcg_gen_mov_i64(ret, val); + break; + } +} + +typedef void (*gen_atomic_cx_i32)(TCGv_i32, TCGv_env, TCGv, + TCGv_i32, TCGv_i32, TCGv_i32); +typedef void (*gen_atomic_cx_i64)(TCGv_i64, TCGv_env, TCGv, + TCGv_i64, TCGv_i64, TCGv_i32); +typedef void (*gen_atomic_cx_i128)(TCGv_i128, TCGv_env, TCGv, + TCGv_i128, TCGv_i128, TCGv_i32); +typedef void (*gen_atomic_op_i32)(TCGv_i32, TCGv_env, TCGv, + TCGv_i32, TCGv_i32); +typedef void (*gen_atomic_op_i64)(TCGv_i64, TCGv_env, TCGv, + TCGv_i64, TCGv_i32); + +#ifdef CONFIG_ATOMIC64 +# define WITH_ATOMIC64(X) X, +#else +# define WITH_ATOMIC64(X) +#endif +#ifdef CONFIG_CMPXCHG128 +# define WITH_ATOMIC128(X) X, +#else +# define WITH_ATOMIC128(X) +#endif + +static void * const table_cmpxchg[(MO_SIZE | MO_BSWAP) + 1] = { + [MO_8] = gen_helper_atomic_cmpxchgb, + [MO_16 | MO_LE] = gen_helper_atomic_cmpxchgw_le, + [MO_16 | MO_BE] = gen_helper_atomic_cmpxchgw_be, + [MO_32 | MO_LE] = gen_helper_atomic_cmpxchgl_le, + [MO_32 | MO_BE] = gen_helper_atomic_cmpxchgl_be, + WITH_ATOMIC64([MO_64 | MO_LE] = gen_helper_atomic_cmpxchgq_le) + WITH_ATOMIC64([MO_64 | MO_BE] = gen_helper_atomic_cmpxchgq_be) + WITH_ATOMIC128([MO_128 | MO_LE] = gen_helper_atomic_cmpxchgo_le) + WITH_ATOMIC128([MO_128 | MO_BE] = gen_helper_atomic_cmpxchgo_be) +}; + +void tcg_gen_nonatomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, + TCGv_i32 newv, TCGArg idx, MemOp memop) +{ + TCGv_i32 t1 = tcg_temp_ebb_new_i32(); + TCGv_i32 t2 = tcg_temp_ebb_new_i32(); + + tcg_gen_ext_i32(t2, cmpv, memop & MO_SIZE); + + tcg_gen_qemu_ld_i32(t1, addr, idx, memop & ~MO_SIGN); + tcg_gen_movcond_i32(TCG_COND_EQ, t2, t1, t2, newv, t1); + tcg_gen_qemu_st_i32(t2, addr, idx, memop); + tcg_temp_free_i32(t2); + + if (memop & MO_SIGN) { + tcg_gen_ext_i32(retv, t1, memop); + } else { + tcg_gen_mov_i32(retv, t1); + } + tcg_temp_free_i32(t1); +} + +void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, + TCGv_i32 newv, TCGArg idx, MemOp memop) +{ + gen_atomic_cx_i32 gen; + MemOpIdx oi; + + if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { + tcg_gen_nonatomic_cmpxchg_i32(retv, addr, cmpv, newv, idx, memop); + return; + } + + memop = tcg_canonicalize_memop(memop, 0, 0); + gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; + tcg_debug_assert(gen != NULL); + + oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + + if (memop & MO_SIGN) { + tcg_gen_ext_i32(retv, retv, memop); + } +} + +void tcg_gen_nonatomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, + TCGv_i64 newv, TCGArg idx, MemOp memop) +{ + TCGv_i64 t1, t2; + + if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { + tcg_gen_nonatomic_cmpxchg_i32(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), + TCGV_LOW(newv), idx, memop); + if (memop & MO_SIGN) { + tcg_gen_sari_i32(TCGV_HIGH(retv), TCGV_LOW(retv), 31); + } else { + tcg_gen_movi_i32(TCGV_HIGH(retv), 0); + } + return; + } + + t1 = tcg_temp_ebb_new_i64(); + t2 = tcg_temp_ebb_new_i64(); + + tcg_gen_ext_i64(t2, cmpv, memop & MO_SIZE); + + tcg_gen_qemu_ld_i64(t1, addr, idx, memop & ~MO_SIGN); + tcg_gen_movcond_i64(TCG_COND_EQ, t2, t1, t2, newv, t1); + tcg_gen_qemu_st_i64(t2, addr, idx, memop); + tcg_temp_free_i64(t2); + + if (memop & MO_SIGN) { + tcg_gen_ext_i64(retv, t1, memop); + } else { + tcg_gen_mov_i64(retv, t1); + } + tcg_temp_free_i64(t1); +} + +void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, + TCGv_i64 newv, TCGArg idx, MemOp memop) +{ + if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { + tcg_gen_nonatomic_cmpxchg_i64(retv, addr, cmpv, newv, idx, memop); + return; + } + + if ((memop & MO_SIZE) == MO_64) { + gen_atomic_cx_i64 gen; + + memop = tcg_canonicalize_memop(memop, 1, 0); + gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; + if (gen) { + MemOpIdx oi = make_memop_idx(memop, idx); + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + return; + } + + gen_helper_exit_atomic(cpu_env); + + /* + * Produce a result for a well-formed opcode stream. This satisfies + * liveness for set before used, which happens before this dead code + * is removed. + */ + tcg_gen_movi_i64(retv, 0); + return; + } + + if (TCG_TARGET_REG_BITS == 32) { + tcg_gen_atomic_cmpxchg_i32(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), + TCGV_LOW(newv), idx, memop); + if (memop & MO_SIGN) { + tcg_gen_sari_i32(TCGV_HIGH(retv), TCGV_LOW(retv), 31); + } else { + tcg_gen_movi_i32(TCGV_HIGH(retv), 0); + } + } else { + TCGv_i32 c32 = tcg_temp_ebb_new_i32(); + TCGv_i32 n32 = tcg_temp_ebb_new_i32(); + TCGv_i32 r32 = tcg_temp_ebb_new_i32(); + + tcg_gen_extrl_i64_i32(c32, cmpv); + tcg_gen_extrl_i64_i32(n32, newv); + tcg_gen_atomic_cmpxchg_i32(r32, addr, c32, n32, idx, memop & ~MO_SIGN); + tcg_temp_free_i32(c32); + tcg_temp_free_i32(n32); + + tcg_gen_extu_i32_i64(retv, r32); + tcg_temp_free_i32(r32); + + if (memop & MO_SIGN) { + tcg_gen_ext_i64(retv, retv, memop); + } + } +} + +void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, + TCGv_i128 newv, TCGArg idx, MemOp memop) +{ + if (TCG_TARGET_REG_BITS == 32) { + /* Inline expansion below is simply too large for 32-bit hosts. */ + gen_atomic_cx_i128 gen = ((memop & MO_BSWAP) == MO_LE + ? gen_helper_nonatomic_cmpxchgo_le + : gen_helper_nonatomic_cmpxchgo_be); + MemOpIdx oi = make_memop_idx(memop, idx); + + tcg_debug_assert((memop & MO_SIZE) == MO_128); + tcg_debug_assert((memop & MO_SIGN) == 0); + + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + } else { + TCGv_i128 oldv = tcg_temp_ebb_new_i128(); + TCGv_i128 tmpv = tcg_temp_ebb_new_i128(); + TCGv_i64 t0 = tcg_temp_ebb_new_i64(); + TCGv_i64 t1 = tcg_temp_ebb_new_i64(); + TCGv_i64 z = tcg_constant_i64(0); + + tcg_gen_qemu_ld_i128(oldv, addr, idx, memop); + + /* Compare i128 */ + tcg_gen_xor_i64(t0, TCGV128_LOW(oldv), TCGV128_LOW(cmpv)); + tcg_gen_xor_i64(t1, TCGV128_HIGH(oldv), TCGV128_HIGH(cmpv)); + tcg_gen_or_i64(t0, t0, t1); + + /* tmpv = equal ? newv : oldv */ + tcg_gen_movcond_i64(TCG_COND_EQ, TCGV128_LOW(tmpv), t0, z, + TCGV128_LOW(newv), TCGV128_LOW(oldv)); + tcg_gen_movcond_i64(TCG_COND_EQ, TCGV128_HIGH(tmpv), t0, z, + TCGV128_HIGH(newv), TCGV128_HIGH(oldv)); + + /* Unconditional writeback. */ + tcg_gen_qemu_st_i128(tmpv, addr, idx, memop); + tcg_gen_mov_i128(retv, oldv); + + tcg_temp_free_i64(t0); + tcg_temp_free_i64(t1); + tcg_temp_free_i128(tmpv); + tcg_temp_free_i128(oldv); + } +} + +void tcg_gen_atomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, + TCGv_i128 newv, TCGArg idx, MemOp memop) +{ + gen_atomic_cx_i128 gen; + + if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { + tcg_gen_nonatomic_cmpxchg_i128(retv, addr, cmpv, newv, idx, memop); + return; + } + + tcg_debug_assert((memop & MO_SIZE) == MO_128); + tcg_debug_assert((memop & MO_SIGN) == 0); + gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; + + if (gen) { + MemOpIdx oi = make_memop_idx(memop, idx); + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + return; + } + + gen_helper_exit_atomic(cpu_env); + + /* + * Produce a result for a well-formed opcode stream. This satisfies + * liveness for set before used, which happens before this dead code + * is removed. + */ + tcg_gen_movi_i64(TCGV128_LOW(retv), 0); + tcg_gen_movi_i64(TCGV128_HIGH(retv), 0); +} + +static void do_nonatomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, + TCGArg idx, MemOp memop, bool new_val, + void (*gen)(TCGv_i32, TCGv_i32, TCGv_i32)) +{ + TCGv_i32 t1 = tcg_temp_ebb_new_i32(); + TCGv_i32 t2 = tcg_temp_ebb_new_i32(); + + memop = tcg_canonicalize_memop(memop, 0, 0); + + tcg_gen_qemu_ld_i32(t1, addr, idx, memop); + tcg_gen_ext_i32(t2, val, memop); + gen(t2, t1, t2); + tcg_gen_qemu_st_i32(t2, addr, idx, memop); + + tcg_gen_ext_i32(ret, (new_val ? t2 : t1), memop); + tcg_temp_free_i32(t1); + tcg_temp_free_i32(t2); +} + +static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, + TCGArg idx, MemOp memop, void * const table[]) +{ + gen_atomic_op_i32 gen; + MemOpIdx oi; + + memop = tcg_canonicalize_memop(memop, 0, 0); + + gen = table[memop & (MO_SIZE | MO_BSWAP)]; + tcg_debug_assert(gen != NULL); + + oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); + + if (memop & MO_SIGN) { + tcg_gen_ext_i32(ret, ret, memop); + } +} + +static void do_nonatomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, + TCGArg idx, MemOp memop, bool new_val, + void (*gen)(TCGv_i64, TCGv_i64, TCGv_i64)) +{ + TCGv_i64 t1 = tcg_temp_ebb_new_i64(); + TCGv_i64 t2 = tcg_temp_ebb_new_i64(); + + memop = tcg_canonicalize_memop(memop, 1, 0); + + tcg_gen_qemu_ld_i64(t1, addr, idx, memop); + tcg_gen_ext_i64(t2, val, memop); + gen(t2, t1, t2); + tcg_gen_qemu_st_i64(t2, addr, idx, memop); + + tcg_gen_ext_i64(ret, (new_val ? t2 : t1), memop); + tcg_temp_free_i64(t1); + tcg_temp_free_i64(t2); +} + +static void do_atomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, + TCGArg idx, MemOp memop, void * const table[]) +{ + memop = tcg_canonicalize_memop(memop, 1, 0); + + if ((memop & MO_SIZE) == MO_64) { +#ifdef CONFIG_ATOMIC64 + gen_atomic_op_i64 gen; + MemOpIdx oi; + + gen = table[memop & (MO_SIZE | MO_BSWAP)]; + tcg_debug_assert(gen != NULL); + + oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); +#else + gen_helper_exit_atomic(cpu_env); + /* Produce a result, so that we have a well-formed opcode stream + with respect to uses of the result in the (dead) code following. */ + tcg_gen_movi_i64(ret, 0); +#endif /* CONFIG_ATOMIC64 */ + } else { + TCGv_i32 v32 = tcg_temp_ebb_new_i32(); + TCGv_i32 r32 = tcg_temp_ebb_new_i32(); + + tcg_gen_extrl_i64_i32(v32, val); + do_atomic_op_i32(r32, addr, v32, idx, memop & ~MO_SIGN, table); + tcg_temp_free_i32(v32); + + tcg_gen_extu_i32_i64(ret, r32); + tcg_temp_free_i32(r32); + + if (memop & MO_SIGN) { + tcg_gen_ext_i64(ret, ret, memop); + } + } +} + +#define GEN_ATOMIC_HELPER(NAME, OP, NEW) \ +static void * const table_##NAME[(MO_SIZE | MO_BSWAP) + 1] = { \ + [MO_8] = gen_helper_atomic_##NAME##b, \ + [MO_16 | MO_LE] = gen_helper_atomic_##NAME##w_le, \ + [MO_16 | MO_BE] = gen_helper_atomic_##NAME##w_be, \ + [MO_32 | MO_LE] = gen_helper_atomic_##NAME##l_le, \ + [MO_32 | MO_BE] = gen_helper_atomic_##NAME##l_be, \ + WITH_ATOMIC64([MO_64 | MO_LE] = gen_helper_atomic_##NAME##q_le) \ + WITH_ATOMIC64([MO_64 | MO_BE] = gen_helper_atomic_##NAME##q_be) \ +}; \ +void tcg_gen_atomic_##NAME##_i32 \ + (TCGv_i32 ret, TCGv addr, TCGv_i32 val, TCGArg idx, MemOp memop) \ +{ \ + if (tcg_ctx->gen_tb->cflags & CF_PARALLEL) { \ + do_atomic_op_i32(ret, addr, val, idx, memop, table_##NAME); \ + } else { \ + do_nonatomic_op_i32(ret, addr, val, idx, memop, NEW, \ + tcg_gen_##OP##_i32); \ + } \ +} \ +void tcg_gen_atomic_##NAME##_i64 \ + (TCGv_i64 ret, TCGv addr, TCGv_i64 val, TCGArg idx, MemOp memop) \ +{ \ + if (tcg_ctx->gen_tb->cflags & CF_PARALLEL) { \ + do_atomic_op_i64(ret, addr, val, idx, memop, table_##NAME); \ + } else { \ + do_nonatomic_op_i64(ret, addr, val, idx, memop, NEW, \ + tcg_gen_##OP##_i64); \ + } \ +} + +GEN_ATOMIC_HELPER(fetch_add, add, 0) +GEN_ATOMIC_HELPER(fetch_and, and, 0) +GEN_ATOMIC_HELPER(fetch_or, or, 0) +GEN_ATOMIC_HELPER(fetch_xor, xor, 0) +GEN_ATOMIC_HELPER(fetch_smin, smin, 0) +GEN_ATOMIC_HELPER(fetch_umin, umin, 0) +GEN_ATOMIC_HELPER(fetch_smax, smax, 0) +GEN_ATOMIC_HELPER(fetch_umax, umax, 0) + +GEN_ATOMIC_HELPER(add_fetch, add, 1) +GEN_ATOMIC_HELPER(and_fetch, and, 1) +GEN_ATOMIC_HELPER(or_fetch, or, 1) +GEN_ATOMIC_HELPER(xor_fetch, xor, 1) +GEN_ATOMIC_HELPER(smin_fetch, smin, 1) +GEN_ATOMIC_HELPER(umin_fetch, umin, 1) +GEN_ATOMIC_HELPER(smax_fetch, smax, 1) +GEN_ATOMIC_HELPER(umax_fetch, umax, 1) + +static void tcg_gen_mov2_i32(TCGv_i32 r, TCGv_i32 a, TCGv_i32 b) +{ + tcg_gen_mov_i32(r, b); +} + +static void tcg_gen_mov2_i64(TCGv_i64 r, TCGv_i64 a, TCGv_i64 b) +{ + tcg_gen_mov_i64(r, b); +} + +GEN_ATOMIC_HELPER(xchg, mov2, 0) + +#undef GEN_ATOMIC_HELPER diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op.c +++ b/tcg/tcg-op.c @@ -XXX,XX +XXX,XX @@ #include "tcg/tcg.h" #include "tcg/tcg-temp-internal.h" #include "tcg/tcg-op.h" -#include "tcg/tcg-mo.h" #include "exec/plugin-gen.h" #include "tcg-internal.h" @@ -XXX,XX +XXX,XX @@ void tcg_gen_lookup_and_goto_ptr(void) tcg_gen_op1i(INDEX_op_goto_ptr, tcgv_ptr_arg(ptr)); tcg_temp_free_ptr(ptr); } - -static inline MemOp tcg_canonicalize_memop(MemOp op, bool is64, bool st) -{ - /* Trigger the asserts within as early as possible. */ - unsigned a_bits = get_alignment_bits(op); - - /* Prefer MO_ALIGN+MO_XX over MO_ALIGN_XX+MO_XX */ - if (a_bits == (op & MO_SIZE)) { - op = (op & ~MO_AMASK) | MO_ALIGN; - } - - switch (op & MO_SIZE) { - case MO_8: - op &= ~MO_BSWAP; - break; - case MO_16: - break; - case MO_32: - if (!is64) { - op &= ~MO_SIGN; - } - break; - case MO_64: - if (is64) { - op &= ~MO_SIGN; - break; - } - /* fall through */ - default: - g_assert_not_reached(); - } - if (st) { - op &= ~MO_SIGN; - } - return op; -} - -static void gen_ldst_i32(TCGOpcode opc, TCGv_i32 val, TCGv addr, - MemOp memop, TCGArg idx) -{ - MemOpIdx oi = make_memop_idx(memop, idx); -#if TARGET_LONG_BITS == 32 - tcg_gen_op3i_i32(opc, val, addr, oi); -#else - if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_op4i_i32(opc, val, TCGV_LOW(addr), TCGV_HIGH(addr), oi); - } else { - tcg_gen_op3(opc, tcgv_i32_arg(val), tcgv_i64_arg(addr), oi); - } -#endif -} - -static void gen_ldst_i64(TCGOpcode opc, TCGv_i64 val, TCGv addr, - MemOp memop, TCGArg idx) -{ - MemOpIdx oi = make_memop_idx(memop, idx); -#if TARGET_LONG_BITS == 32 - if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_op4i_i32(opc, TCGV_LOW(val), TCGV_HIGH(val), addr, oi); - } else { - tcg_gen_op3(opc, tcgv_i64_arg(val), tcgv_i32_arg(addr), oi); - } -#else - if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_op5i_i32(opc, TCGV_LOW(val), TCGV_HIGH(val), - TCGV_LOW(addr), TCGV_HIGH(addr), oi); - } else { - tcg_gen_op3i_i64(opc, val, addr, oi); - } -#endif -} - -static void tcg_gen_req_mo(TCGBar type) -{ -#ifdef TCG_GUEST_DEFAULT_MO - type &= TCG_GUEST_DEFAULT_MO; -#endif - type &= ~TCG_TARGET_DEFAULT_MO; - if (type) { - tcg_gen_mb(type | TCG_BAR_SC); - } -} - -static inline TCGv plugin_prep_mem_callbacks(TCGv vaddr) -{ -#ifdef CONFIG_PLUGIN - if (tcg_ctx->plugin_insn != NULL) { - /* Save a copy of the vaddr for use after a load. */ - TCGv temp = tcg_temp_new(); - tcg_gen_mov_tl(temp, vaddr); - return temp; - } -#endif - return vaddr; -} - -static void plugin_gen_mem_callbacks(TCGv vaddr, MemOpIdx oi, - enum qemu_plugin_mem_rw rw) -{ -#ifdef CONFIG_PLUGIN - if (tcg_ctx->plugin_insn != NULL) { - qemu_plugin_meminfo_t info = make_plugin_meminfo(oi, rw); - plugin_gen_empty_mem_callback(vaddr, info); - tcg_temp_free(vaddr); - } -#endif -} - -void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) -{ - MemOp orig_memop; - MemOpIdx oi; - - tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); - memop = tcg_canonicalize_memop(memop, 0, 0); - oi = make_memop_idx(memop, idx); - - orig_memop = memop; - if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { - memop &= ~MO_BSWAP; - /* The bswap primitive benefits from zero-extended input. */ - if ((memop & MO_SSIZE) == MO_SW) { - memop &= ~MO_SIGN; - } - } - - addr = plugin_prep_mem_callbacks(addr); - gen_ldst_i32(INDEX_op_qemu_ld_i32, val, addr, memop, idx); - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); - - if ((orig_memop ^ memop) & MO_BSWAP) { - switch (orig_memop & MO_SIZE) { - case MO_16: - tcg_gen_bswap16_i32(val, val, (orig_memop & MO_SIGN - ? TCG_BSWAP_IZ | TCG_BSWAP_OS - : TCG_BSWAP_IZ | TCG_BSWAP_OZ)); - break; - case MO_32: - tcg_gen_bswap32_i32(val, val); - break; - default: - g_assert_not_reached(); - } - } -} - -void tcg_gen_qemu_st_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) -{ - TCGv_i32 swap = NULL; - MemOpIdx oi; - - tcg_gen_req_mo(TCG_MO_LD_ST | TCG_MO_ST_ST); - memop = tcg_canonicalize_memop(memop, 0, 1); - oi = make_memop_idx(memop, idx); - - if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { - swap = tcg_temp_ebb_new_i32(); - switch (memop & MO_SIZE) { - case MO_16: - tcg_gen_bswap16_i32(swap, val, 0); - break; - case MO_32: - tcg_gen_bswap32_i32(swap, val); - break; - default: - g_assert_not_reached(); - } - val = swap; - memop &= ~MO_BSWAP; - } - - addr = plugin_prep_mem_callbacks(addr); - if (TCG_TARGET_HAS_qemu_st8_i32 && (memop & MO_SIZE) == MO_8) { - gen_ldst_i32(INDEX_op_qemu_st8_i32, val, addr, memop, idx); - } else { - gen_ldst_i32(INDEX_op_qemu_st_i32, val, addr, memop, idx); - } - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); - - if (swap) { - tcg_temp_free_i32(swap); - } -} - -void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) -{ - MemOp orig_memop; - MemOpIdx oi; - - if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { - tcg_gen_qemu_ld_i32(TCGV_LOW(val), addr, idx, memop); - if (memop & MO_SIGN) { - tcg_gen_sari_i32(TCGV_HIGH(val), TCGV_LOW(val), 31); - } else { - tcg_gen_movi_i32(TCGV_HIGH(val), 0); - } - return; - } - - tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); - memop = tcg_canonicalize_memop(memop, 1, 0); - oi = make_memop_idx(memop, idx); - - orig_memop = memop; - if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { - memop &= ~MO_BSWAP; - /* The bswap primitive benefits from zero-extended input. */ - if ((memop & MO_SIGN) && (memop & MO_SIZE) < MO_64) { - memop &= ~MO_SIGN; - } - } - - addr = plugin_prep_mem_callbacks(addr); - gen_ldst_i64(INDEX_op_qemu_ld_i64, val, addr, memop, idx); - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); - - if ((orig_memop ^ memop) & MO_BSWAP) { - int flags = (orig_memop & MO_SIGN - ? TCG_BSWAP_IZ | TCG_BSWAP_OS - : TCG_BSWAP_IZ | TCG_BSWAP_OZ); - switch (orig_memop & MO_SIZE) { - case MO_16: - tcg_gen_bswap16_i64(val, val, flags); - break; - case MO_32: - tcg_gen_bswap32_i64(val, val, flags); - break; - case MO_64: - tcg_gen_bswap64_i64(val, val); - break; - default: - g_assert_not_reached(); - } - } -} - -void tcg_gen_qemu_st_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) -{ - TCGv_i64 swap = NULL; - MemOpIdx oi; - - if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { - tcg_gen_qemu_st_i32(TCGV_LOW(val), addr, idx, memop); - return; - } - - tcg_gen_req_mo(TCG_MO_LD_ST | TCG_MO_ST_ST); - memop = tcg_canonicalize_memop(memop, 1, 1); - oi = make_memop_idx(memop, idx); - - if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { - swap = tcg_temp_ebb_new_i64(); - switch (memop & MO_SIZE) { - case MO_16: - tcg_gen_bswap16_i64(swap, val, 0); - break; - case MO_32: - tcg_gen_bswap32_i64(swap, val, 0); - break; - case MO_64: - tcg_gen_bswap64_i64(swap, val); - break; - default: - g_assert_not_reached(); - } - val = swap; - memop &= ~MO_BSWAP; - } - - addr = plugin_prep_mem_callbacks(addr); - gen_ldst_i64(INDEX_op_qemu_st_i64, val, addr, memop, idx); - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); - - if (swap) { - tcg_temp_free_i64(swap); - } -} - -/* - * Return true if @mop, without knowledge of the pointer alignment, - * does not require 16-byte atomicity, and it would be adventagous - * to avoid a call to a helper function. - */ -static bool use_two_i64_for_i128(MemOp mop) -{ -#ifdef CONFIG_SOFTMMU - /* Two softmmu tlb lookups is larger than one function call. */ - return false; -#else - /* - * For user-only, two 64-bit operations may well be smaller than a call. - * Determine if that would be legal for the requested atomicity. - */ - switch (mop & MO_ATOM_MASK) { - case MO_ATOM_NONE: - case MO_ATOM_IFALIGN_PAIR: - return true; - case MO_ATOM_IFALIGN: - case MO_ATOM_SUBALIGN: - case MO_ATOM_WITHIN16: - case MO_ATOM_WITHIN16_PAIR: - /* In a serialized context, no atomicity is required. */ - return !(tcg_ctx->gen_tb->cflags & CF_PARALLEL); - default: - g_assert_not_reached(); - } -#endif -} - -static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) -{ - MemOp mop_1 = orig, mop_2; - - tcg_debug_assert((orig & MO_SIZE) == MO_128); - tcg_debug_assert((orig & MO_SIGN) == 0); - - /* Reduce the size to 64-bit. */ - mop_1 = (mop_1 & ~MO_SIZE) | MO_64; - - /* Retain the alignment constraints of the original. */ - switch (orig & MO_AMASK) { - case MO_UNALN: - case MO_ALIGN_2: - case MO_ALIGN_4: - mop_2 = mop_1; - break; - case MO_ALIGN_8: - /* Prefer MO_ALIGN+MO_64 to MO_ALIGN_8+MO_64. */ - mop_1 = (mop_1 & ~MO_AMASK) | MO_ALIGN; - mop_2 = mop_1; - break; - case MO_ALIGN: - /* Second has 8-byte alignment; first has 16-byte alignment. */ - mop_2 = mop_1; - mop_1 = (mop_1 & ~MO_AMASK) | MO_ALIGN_16; - break; - case MO_ALIGN_16: - case MO_ALIGN_32: - case MO_ALIGN_64: - /* Second has 8-byte alignment; first retains original. */ - mop_2 = (mop_1 & ~MO_AMASK) | MO_ALIGN; - break; - default: - g_assert_not_reached(); - } - - /* Use a memory ordering implemented by the host. */ - if ((orig & MO_BSWAP) && !tcg_target_has_memory_bswap(mop_1)) { - mop_1 &= ~MO_BSWAP; - mop_2 &= ~MO_BSWAP; - } - - ret[0] = mop_1; - ret[1] = mop_2; -} - -#if TARGET_LONG_BITS == 64 -#define tcg_temp_ebb_new tcg_temp_ebb_new_i64 -#else -#define tcg_temp_ebb_new tcg_temp_ebb_new_i32 -#endif - -void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) -{ - const MemOpIdx oi = make_memop_idx(memop, idx); - - tcg_debug_assert((memop & MO_SIZE) == MO_128); - tcg_debug_assert((memop & MO_SIGN) == 0); - - tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); - addr = plugin_prep_mem_callbacks(addr); - - /* TODO: For now, force 32-bit hosts to use the helper. */ - if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { - TCGv_i64 lo, hi; - TCGArg addr_arg; - MemOpIdx adj_oi; - bool need_bswap = false; - - if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { - lo = TCGV128_HIGH(val); - hi = TCGV128_LOW(val); - adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); - need_bswap = true; - } else { - lo = TCGV128_LOW(val); - hi = TCGV128_HIGH(val); - adj_oi = oi; - } - -#if TARGET_LONG_BITS == 32 - addr_arg = tcgv_i32_arg(addr); -#else - addr_arg = tcgv_i64_arg(addr); -#endif - tcg_gen_op4ii_i64(INDEX_op_qemu_ld_i128, lo, hi, addr_arg, adj_oi); - - if (need_bswap) { - tcg_gen_bswap64_i64(lo, lo); - tcg_gen_bswap64_i64(hi, hi); - } - } else if (use_two_i64_for_i128(memop)) { - MemOp mop[2]; - TCGv addr_p8; - TCGv_i64 x, y; - - canonicalize_memop_i128_as_i64(mop, memop); - - /* - * Since there are no global TCGv_i128, there is no visible state - * changed if the second load faults. Load directly into the two - * subwords. - */ - if ((memop & MO_BSWAP) == MO_LE) { - x = TCGV128_LOW(val); - y = TCGV128_HIGH(val); - } else { - x = TCGV128_HIGH(val); - y = TCGV128_LOW(val); - } - - gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, mop[0], idx); - - if ((mop[0] ^ memop) & MO_BSWAP) { - tcg_gen_bswap64_i64(x, x); - } - - addr_p8 = tcg_temp_ebb_new(); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, mop[1], idx); - tcg_temp_free(addr_p8); - - if ((mop[0] ^ memop) & MO_BSWAP) { - tcg_gen_bswap64_i64(y, y); - } - } else { - gen_helper_ld_i128(val, cpu_env, addr, tcg_constant_i32(oi)); - } - - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); -} - -void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) -{ - const MemOpIdx oi = make_memop_idx(memop, idx); - - tcg_debug_assert((memop & MO_SIZE) == MO_128); - tcg_debug_assert((memop & MO_SIGN) == 0); - - tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); - addr = plugin_prep_mem_callbacks(addr); - - /* TODO: For now, force 32-bit hosts to use the helper. */ - - if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { - TCGv_i64 lo, hi; - TCGArg addr_arg; - MemOpIdx adj_oi; - bool need_bswap = false; - - if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { - lo = tcg_temp_new_i64(); - hi = tcg_temp_new_i64(); - tcg_gen_bswap64_i64(lo, TCGV128_HIGH(val)); - tcg_gen_bswap64_i64(hi, TCGV128_LOW(val)); - adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); - need_bswap = true; - } else { - lo = TCGV128_LOW(val); - hi = TCGV128_HIGH(val); - adj_oi = oi; - } - -#if TARGET_LONG_BITS == 32 - addr_arg = tcgv_i32_arg(addr); -#else - addr_arg = tcgv_i64_arg(addr); -#endif - tcg_gen_op4ii_i64(INDEX_op_qemu_st_i128, lo, hi, addr_arg, adj_oi); - - if (need_bswap) { - tcg_temp_free_i64(lo); - tcg_temp_free_i64(hi); - } - } else if (use_two_i64_for_i128(memop)) { - MemOp mop[2]; - TCGv addr_p8; - TCGv_i64 x, y; - - canonicalize_memop_i128_as_i64(mop, memop); - - if ((memop & MO_BSWAP) == MO_LE) { - x = TCGV128_LOW(val); - y = TCGV128_HIGH(val); - } else { - x = TCGV128_HIGH(val); - y = TCGV128_LOW(val); - } - - addr_p8 = tcg_temp_ebb_new(); - if ((mop[0] ^ memop) & MO_BSWAP) { - TCGv_i64 t = tcg_temp_ebb_new_i64(); - - tcg_gen_bswap64_i64(t, x); - gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr, mop[0], idx); - tcg_gen_bswap64_i64(t, y); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr_p8, mop[1], idx); - tcg_temp_free_i64(t); - } else { - gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, mop[0], idx); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, mop[1], idx); - } - tcg_temp_free(addr_p8); - } else { - gen_helper_st_i128(cpu_env, addr, val, tcg_constant_i32(oi)); - } - - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); -} - -static void tcg_gen_ext_i32(TCGv_i32 ret, TCGv_i32 val, MemOp opc) -{ - switch (opc & MO_SSIZE) { - case MO_SB: - tcg_gen_ext8s_i32(ret, val); - break; - case MO_UB: - tcg_gen_ext8u_i32(ret, val); - break; - case MO_SW: - tcg_gen_ext16s_i32(ret, val); - break; - case MO_UW: - tcg_gen_ext16u_i32(ret, val); - break; - default: - tcg_gen_mov_i32(ret, val); - break; - } -} - -static void tcg_gen_ext_i64(TCGv_i64 ret, TCGv_i64 val, MemOp opc) -{ - switch (opc & MO_SSIZE) { - case MO_SB: - tcg_gen_ext8s_i64(ret, val); - break; - case MO_UB: - tcg_gen_ext8u_i64(ret, val); - break; - case MO_SW: - tcg_gen_ext16s_i64(ret, val); - break; - case MO_UW: - tcg_gen_ext16u_i64(ret, val); - break; - case MO_SL: - tcg_gen_ext32s_i64(ret, val); - break; - case MO_UL: - tcg_gen_ext32u_i64(ret, val); - break; - default: - tcg_gen_mov_i64(ret, val); - break; - } -} - -typedef void (*gen_atomic_cx_i32)(TCGv_i32, TCGv_env, TCGv, - TCGv_i32, TCGv_i32, TCGv_i32); -typedef void (*gen_atomic_cx_i64)(TCGv_i64, TCGv_env, TCGv, - TCGv_i64, TCGv_i64, TCGv_i32); -typedef void (*gen_atomic_cx_i128)(TCGv_i128, TCGv_env, TCGv, - TCGv_i128, TCGv_i128, TCGv_i32); -typedef void (*gen_atomic_op_i32)(TCGv_i32, TCGv_env, TCGv, - TCGv_i32, TCGv_i32); -typedef void (*gen_atomic_op_i64)(TCGv_i64, TCGv_env, TCGv, - TCGv_i64, TCGv_i32); - -#ifdef CONFIG_ATOMIC64 -# define WITH_ATOMIC64(X) X, -#else -# define WITH_ATOMIC64(X) -#endif -#ifdef CONFIG_CMPXCHG128 -# define WITH_ATOMIC128(X) X, -#else -# define WITH_ATOMIC128(X) -#endif - -static void * const table_cmpxchg[(MO_SIZE | MO_BSWAP) + 1] = { - [MO_8] = gen_helper_atomic_cmpxchgb, - [MO_16 | MO_LE] = gen_helper_atomic_cmpxchgw_le, - [MO_16 | MO_BE] = gen_helper_atomic_cmpxchgw_be, - [MO_32 | MO_LE] = gen_helper_atomic_cmpxchgl_le, - [MO_32 | MO_BE] = gen_helper_atomic_cmpxchgl_be, - WITH_ATOMIC64([MO_64 | MO_LE] = gen_helper_atomic_cmpxchgq_le) - WITH_ATOMIC64([MO_64 | MO_BE] = gen_helper_atomic_cmpxchgq_be) - WITH_ATOMIC128([MO_128 | MO_LE] = gen_helper_atomic_cmpxchgo_le) - WITH_ATOMIC128([MO_128 | MO_BE] = gen_helper_atomic_cmpxchgo_be) -}; - -void tcg_gen_nonatomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, - TCGv_i32 newv, TCGArg idx, MemOp memop) -{ - TCGv_i32 t1 = tcg_temp_ebb_new_i32(); - TCGv_i32 t2 = tcg_temp_ebb_new_i32(); - - tcg_gen_ext_i32(t2, cmpv, memop & MO_SIZE); - - tcg_gen_qemu_ld_i32(t1, addr, idx, memop & ~MO_SIGN); - tcg_gen_movcond_i32(TCG_COND_EQ, t2, t1, t2, newv, t1); - tcg_gen_qemu_st_i32(t2, addr, idx, memop); - tcg_temp_free_i32(t2); - - if (memop & MO_SIGN) { - tcg_gen_ext_i32(retv, t1, memop); - } else { - tcg_gen_mov_i32(retv, t1); - } - tcg_temp_free_i32(t1); -} - -void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, - TCGv_i32 newv, TCGArg idx, MemOp memop) -{ - gen_atomic_cx_i32 gen; - MemOpIdx oi; - - if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { - tcg_gen_nonatomic_cmpxchg_i32(retv, addr, cmpv, newv, idx, memop); - return; - } - - memop = tcg_canonicalize_memop(memop, 0, 0); - gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; - tcg_debug_assert(gen != NULL); - - oi = make_memop_idx(memop & ~MO_SIGN, idx); - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); - - if (memop & MO_SIGN) { - tcg_gen_ext_i32(retv, retv, memop); - } -} - -void tcg_gen_nonatomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, - TCGv_i64 newv, TCGArg idx, MemOp memop) -{ - TCGv_i64 t1, t2; - - if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { - tcg_gen_nonatomic_cmpxchg_i32(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), - TCGV_LOW(newv), idx, memop); - if (memop & MO_SIGN) { - tcg_gen_sari_i32(TCGV_HIGH(retv), TCGV_LOW(retv), 31); - } else { - tcg_gen_movi_i32(TCGV_HIGH(retv), 0); - } - return; - } - - t1 = tcg_temp_ebb_new_i64(); - t2 = tcg_temp_ebb_new_i64(); - - tcg_gen_ext_i64(t2, cmpv, memop & MO_SIZE); - - tcg_gen_qemu_ld_i64(t1, addr, idx, memop & ~MO_SIGN); - tcg_gen_movcond_i64(TCG_COND_EQ, t2, t1, t2, newv, t1); - tcg_gen_qemu_st_i64(t2, addr, idx, memop); - tcg_temp_free_i64(t2); - - if (memop & MO_SIGN) { - tcg_gen_ext_i64(retv, t1, memop); - } else { - tcg_gen_mov_i64(retv, t1); - } - tcg_temp_free_i64(t1); -} - -void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, - TCGv_i64 newv, TCGArg idx, MemOp memop) -{ - if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { - tcg_gen_nonatomic_cmpxchg_i64(retv, addr, cmpv, newv, idx, memop); - return; - } - - if ((memop & MO_SIZE) == MO_64) { - gen_atomic_cx_i64 gen; - - memop = tcg_canonicalize_memop(memop, 1, 0); - gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; - if (gen) { - MemOpIdx oi = make_memop_idx(memop, idx); - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); - return; - } - - gen_helper_exit_atomic(cpu_env); - - /* - * Produce a result for a well-formed opcode stream. This satisfies - * liveness for set before used, which happens before this dead code - * is removed. - */ - tcg_gen_movi_i64(retv, 0); - return; - } - - if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_atomic_cmpxchg_i32(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), - TCGV_LOW(newv), idx, memop); - if (memop & MO_SIGN) { - tcg_gen_sari_i32(TCGV_HIGH(retv), TCGV_LOW(retv), 31); - } else { - tcg_gen_movi_i32(TCGV_HIGH(retv), 0); - } - } else { - TCGv_i32 c32 = tcg_temp_ebb_new_i32(); - TCGv_i32 n32 = tcg_temp_ebb_new_i32(); - TCGv_i32 r32 = tcg_temp_ebb_new_i32(); - - tcg_gen_extrl_i64_i32(c32, cmpv); - tcg_gen_extrl_i64_i32(n32, newv); - tcg_gen_atomic_cmpxchg_i32(r32, addr, c32, n32, idx, memop & ~MO_SIGN); - tcg_temp_free_i32(c32); - tcg_temp_free_i32(n32); - - tcg_gen_extu_i32_i64(retv, r32); - tcg_temp_free_i32(r32); - - if (memop & MO_SIGN) { - tcg_gen_ext_i64(retv, retv, memop); - } - } -} - -void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, - TCGv_i128 newv, TCGArg idx, MemOp memop) -{ - if (TCG_TARGET_REG_BITS == 32) { - /* Inline expansion below is simply too large for 32-bit hosts. */ - gen_atomic_cx_i128 gen = ((memop & MO_BSWAP) == MO_LE - ? gen_helper_nonatomic_cmpxchgo_le - : gen_helper_nonatomic_cmpxchgo_be); - MemOpIdx oi = make_memop_idx(memop, idx); - - tcg_debug_assert((memop & MO_SIZE) == MO_128); - tcg_debug_assert((memop & MO_SIGN) == 0); - - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); - } else { - TCGv_i128 oldv = tcg_temp_ebb_new_i128(); - TCGv_i128 tmpv = tcg_temp_ebb_new_i128(); - TCGv_i64 t0 = tcg_temp_ebb_new_i64(); - TCGv_i64 t1 = tcg_temp_ebb_new_i64(); - TCGv_i64 z = tcg_constant_i64(0); - - tcg_gen_qemu_ld_i128(oldv, addr, idx, memop); - - /* Compare i128 */ - tcg_gen_xor_i64(t0, TCGV128_LOW(oldv), TCGV128_LOW(cmpv)); - tcg_gen_xor_i64(t1, TCGV128_HIGH(oldv), TCGV128_HIGH(cmpv)); - tcg_gen_or_i64(t0, t0, t1); - - /* tmpv = equal ? newv : oldv */ - tcg_gen_movcond_i64(TCG_COND_EQ, TCGV128_LOW(tmpv), t0, z, - TCGV128_LOW(newv), TCGV128_LOW(oldv)); - tcg_gen_movcond_i64(TCG_COND_EQ, TCGV128_HIGH(tmpv), t0, z, - TCGV128_HIGH(newv), TCGV128_HIGH(oldv)); - - /* Unconditional writeback. */ - tcg_gen_qemu_st_i128(tmpv, addr, idx, memop); - tcg_gen_mov_i128(retv, oldv); - - tcg_temp_free_i64(t0); - tcg_temp_free_i64(t1); - tcg_temp_free_i128(tmpv); - tcg_temp_free_i128(oldv); - } -} - -void tcg_gen_atomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, - TCGv_i128 newv, TCGArg idx, MemOp memop) -{ - gen_atomic_cx_i128 gen; - - if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { - tcg_gen_nonatomic_cmpxchg_i128(retv, addr, cmpv, newv, idx, memop); - return; - } - - tcg_debug_assert((memop & MO_SIZE) == MO_128); - tcg_debug_assert((memop & MO_SIGN) == 0); - gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; - - if (gen) { - MemOpIdx oi = make_memop_idx(memop, idx); - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); - return; - } - - gen_helper_exit_atomic(cpu_env); - - /* - * Produce a result for a well-formed opcode stream. This satisfies - * liveness for set before used, which happens before this dead code - * is removed. - */ - tcg_gen_movi_i64(TCGV128_LOW(retv), 0); - tcg_gen_movi_i64(TCGV128_HIGH(retv), 0); -} - -static void do_nonatomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, - TCGArg idx, MemOp memop, bool new_val, - void (*gen)(TCGv_i32, TCGv_i32, TCGv_i32)) -{ - TCGv_i32 t1 = tcg_temp_ebb_new_i32(); - TCGv_i32 t2 = tcg_temp_ebb_new_i32(); - - memop = tcg_canonicalize_memop(memop, 0, 0); - - tcg_gen_qemu_ld_i32(t1, addr, idx, memop); - tcg_gen_ext_i32(t2, val, memop); - gen(t2, t1, t2); - tcg_gen_qemu_st_i32(t2, addr, idx, memop); - - tcg_gen_ext_i32(ret, (new_val ? t2 : t1), memop); - tcg_temp_free_i32(t1); - tcg_temp_free_i32(t2); -} - -static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, - TCGArg idx, MemOp memop, void * const table[]) -{ - gen_atomic_op_i32 gen; - MemOpIdx oi; - - memop = tcg_canonicalize_memop(memop, 0, 0); - - gen = table[memop & (MO_SIZE | MO_BSWAP)]; - tcg_debug_assert(gen != NULL); - - oi = make_memop_idx(memop & ~MO_SIGN, idx); - gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); - - if (memop & MO_SIGN) { - tcg_gen_ext_i32(ret, ret, memop); - } -} - -static void do_nonatomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, - TCGArg idx, MemOp memop, bool new_val, - void (*gen)(TCGv_i64, TCGv_i64, TCGv_i64)) -{ - TCGv_i64 t1 = tcg_temp_ebb_new_i64(); - TCGv_i64 t2 = tcg_temp_ebb_new_i64(); - - memop = tcg_canonicalize_memop(memop, 1, 0); - - tcg_gen_qemu_ld_i64(t1, addr, idx, memop); - tcg_gen_ext_i64(t2, val, memop); - gen(t2, t1, t2); - tcg_gen_qemu_st_i64(t2, addr, idx, memop); - - tcg_gen_ext_i64(ret, (new_val ? t2 : t1), memop); - tcg_temp_free_i64(t1); - tcg_temp_free_i64(t2); -} - -static void do_atomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, - TCGArg idx, MemOp memop, void * const table[]) -{ - memop = tcg_canonicalize_memop(memop, 1, 0); - - if ((memop & MO_SIZE) == MO_64) { -#ifdef CONFIG_ATOMIC64 - gen_atomic_op_i64 gen; - MemOpIdx oi; - - gen = table[memop & (MO_SIZE | MO_BSWAP)]; - tcg_debug_assert(gen != NULL); - - oi = make_memop_idx(memop & ~MO_SIGN, idx); - gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); -#else - gen_helper_exit_atomic(cpu_env); - /* Produce a result, so that we have a well-formed opcode stream - with respect to uses of the result in the (dead) code following. */ - tcg_gen_movi_i64(ret, 0); -#endif /* CONFIG_ATOMIC64 */ - } else { - TCGv_i32 v32 = tcg_temp_ebb_new_i32(); - TCGv_i32 r32 = tcg_temp_ebb_new_i32(); - - tcg_gen_extrl_i64_i32(v32, val); - do_atomic_op_i32(r32, addr, v32, idx, memop & ~MO_SIGN, table); - tcg_temp_free_i32(v32); - - tcg_gen_extu_i32_i64(ret, r32); - tcg_temp_free_i32(r32); - - if (memop & MO_SIGN) { - tcg_gen_ext_i64(ret, ret, memop); - } - } -} - -#define GEN_ATOMIC_HELPER(NAME, OP, NEW) \ -static void * const table_##NAME[(MO_SIZE | MO_BSWAP) + 1] = { \ - [MO_8] = gen_helper_atomic_##NAME##b, \ - [MO_16 | MO_LE] = gen_helper_atomic_##NAME##w_le, \ - [MO_16 | MO_BE] = gen_helper_atomic_##NAME##w_be, \ - [MO_32 | MO_LE] = gen_helper_atomic_##NAME##l_le, \ - [MO_32 | MO_BE] = gen_helper_atomic_##NAME##l_be, \ - WITH_ATOMIC64([MO_64 | MO_LE] = gen_helper_atomic_##NAME##q_le) \ - WITH_ATOMIC64([MO_64 | MO_BE] = gen_helper_atomic_##NAME##q_be) \ -}; \ -void tcg_gen_atomic_##NAME##_i32 \ - (TCGv_i32 ret, TCGv addr, TCGv_i32 val, TCGArg idx, MemOp memop) \ -{ \ - if (tcg_ctx->gen_tb->cflags & CF_PARALLEL) { \ - do_atomic_op_i32(ret, addr, val, idx, memop, table_##NAME); \ - } else { \ - do_nonatomic_op_i32(ret, addr, val, idx, memop, NEW, \ - tcg_gen_##OP##_i32); \ - } \ -} \ -void tcg_gen_atomic_##NAME##_i64 \ - (TCGv_i64 ret, TCGv addr, TCGv_i64 val, TCGArg idx, MemOp memop) \ -{ \ - if (tcg_ctx->gen_tb->cflags & CF_PARALLEL) { \ - do_atomic_op_i64(ret, addr, val, idx, memop, table_##NAME); \ - } else { \ - do_nonatomic_op_i64(ret, addr, val, idx, memop, NEW, \ - tcg_gen_##OP##_i64); \ - } \ -} - -GEN_ATOMIC_HELPER(fetch_add, add, 0) -GEN_ATOMIC_HELPER(fetch_and, and, 0) -GEN_ATOMIC_HELPER(fetch_or, or, 0) -GEN_ATOMIC_HELPER(fetch_xor, xor, 0) -GEN_ATOMIC_HELPER(fetch_smin, smin, 0) -GEN_ATOMIC_HELPER(fetch_umin, umin, 0) -GEN_ATOMIC_HELPER(fetch_smax, smax, 0) -GEN_ATOMIC_HELPER(fetch_umax, umax, 0) - -GEN_ATOMIC_HELPER(add_fetch, add, 1) -GEN_ATOMIC_HELPER(and_fetch, and, 1) -GEN_ATOMIC_HELPER(or_fetch, or, 1) -GEN_ATOMIC_HELPER(xor_fetch, xor, 1) -GEN_ATOMIC_HELPER(smin_fetch, smin, 1) -GEN_ATOMIC_HELPER(umin_fetch, umin, 1) -GEN_ATOMIC_HELPER(smax_fetch, smax, 1) -GEN_ATOMIC_HELPER(umax_fetch, umax, 1) - -static void tcg_gen_mov2_i32(TCGv_i32 r, TCGv_i32 a, TCGv_i32 b) -{ - tcg_gen_mov_i32(r, b); -} - -static void tcg_gen_mov2_i64(TCGv_i64 r, TCGv_i64 a, TCGv_i64 b) -{ - tcg_gen_mov_i64(r, b); -} - -GEN_ATOMIC_HELPER(xchg, mov2, 0) - -#undef GEN_ATOMIC_HELPER diff --git a/tcg/meson.build b/tcg/meson.build index XXXXXXX..XXXXXXX 100644 --- a/tcg/meson.build +++ b/tcg/meson.build @@ -XXX,XX +XXX,XX @@ tcg_ss.add(files( 'tcg.c', 'tcg-common.c', 'tcg-op.c', + 'tcg-op-ldst.c', 'tcg-op-gvec.c', 'tcg-op-vec.c', )) -- 2.34.1 We already pass uint64_t to restore_state_to_opc; this changes all of the other uses from insn_start through the encoding to decoding. Reviewed-by: Anton Johansson <anjo@rev.ng> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-op.h | 39 +++++++++------------------------------ include/tcg/tcg-opc.h | 2 +- include/tcg/tcg.h | 30 +++++++++++++++--------------- accel/tcg/translate-all.c | 28 ++++++++++++++++------------ tcg/tcg.c | 18 ++++-------------- 5 files changed, 45 insertions(+), 72 deletions(-) diff --git a/include/tcg/tcg-op.h b/include/tcg/tcg-op.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-op.h +++ b/include/tcg/tcg-op.h @@ -XXX,XX +XXX,XX @@ static inline void tcg_gen_concat32_i64(TCGv_i64 ret, TCGv_i64 lo, TCGv_i64 hi) #endif #if TARGET_INSN_START_WORDS == 1 -# if TARGET_LONG_BITS <= TCG_TARGET_REG_BITS static inline void tcg_gen_insn_start(target_ulong pc) { - tcg_gen_op1(INDEX_op_insn_start, pc); + TCGOp *op = tcg_emit_op(INDEX_op_insn_start, 64 / TCG_TARGET_REG_BITS); + tcg_set_insn_start_param(op, 0, pc); } -# else -static inline void tcg_gen_insn_start(target_ulong pc) -{ - tcg_gen_op2(INDEX_op_insn_start, (uint32_t)pc, (uint32_t)(pc >> 32)); -} -# endif #elif TARGET_INSN_START_WORDS == 2 -# if TARGET_LONG_BITS <= TCG_TARGET_REG_BITS static inline void tcg_gen_insn_start(target_ulong pc, target_ulong a1) { - tcg_gen_op2(INDEX_op_insn_start, pc, a1); + TCGOp *op = tcg_emit_op(INDEX_op_insn_start, 2 * 64 / TCG_TARGET_REG_BITS); + tcg_set_insn_start_param(op, 0, pc); + tcg_set_insn_start_param(op, 1, a1); } -# else -static inline void tcg_gen_insn_start(target_ulong pc, target_ulong a1) -{ - tcg_gen_op4(INDEX_op_insn_start, - (uint32_t)pc, (uint32_t)(pc >> 32), - (uint32_t)a1, (uint32_t)(a1 >> 32)); -} -# endif #elif TARGET_INSN_START_WORDS == 3 -# if TARGET_LONG_BITS <= TCG_TARGET_REG_BITS static inline void tcg_gen_insn_start(target_ulong pc, target_ulong a1, target_ulong a2) { - tcg_gen_op3(INDEX_op_insn_start, pc, a1, a2); + TCGOp *op = tcg_emit_op(INDEX_op_insn_start, 3 * 64 / TCG_TARGET_REG_BITS); + tcg_set_insn_start_param(op, 0, pc); + tcg_set_insn_start_param(op, 1, a1); + tcg_set_insn_start_param(op, 2, a2); } -# else -static inline void tcg_gen_insn_start(target_ulong pc, target_ulong a1, - target_ulong a2) -{ - tcg_gen_op6(INDEX_op_insn_start, - (uint32_t)pc, (uint32_t)(pc >> 32), - (uint32_t)a1, (uint32_t)(a1 >> 32), - (uint32_t)a2, (uint32_t)(a2 >> 32)); -} -# endif #else # error "Unhandled number of operands to insn_start" #endif diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-opc.h +++ b/include/tcg/tcg-opc.h @@ -XXX,XX +XXX,XX @@ DEF(mulsh_i64, 1, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_mulsh_i64)) #define DATA64_ARGS (TCG_TARGET_REG_BITS == 64 ? 1 : 2) /* QEMU specific */ -DEF(insn_start, 0, 0, TLADDR_ARGS * TARGET_INSN_START_WORDS, +DEF(insn_start, 0, 0, DATA64_ARGS * TARGET_INSN_START_WORDS, TCG_OPF_NOT_PRESENT) DEF(exit_tb, 0, 0, 1, TCG_OPF_BB_EXIT | TCG_OPF_BB_END) DEF(goto_tb, 0, 0, 1, TCG_OPF_BB_EXIT | TCG_OPF_BB_END) diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -XXX,XX +XXX,XX @@ struct TCGContext { TCGTemp *reg_to_temp[TCG_TARGET_NB_REGS]; uint16_t gen_insn_end_off[TCG_MAX_INSNS]; - target_ulong gen_insn_data[TCG_MAX_INSNS][TARGET_INSN_START_WORDS]; + uint64_t gen_insn_data[TCG_MAX_INSNS][TARGET_INSN_START_WORDS]; /* Exit to translator on overflow. */ sigjmp_buf jmp_trans; @@ -XXX,XX +XXX,XX @@ static inline void tcg_set_insn_param(TCGOp *op, int arg, TCGArg v) op->args[arg] = v; } -static inline target_ulong tcg_get_insn_start_param(TCGOp *op, int arg) +static inline uint64_t tcg_get_insn_start_param(TCGOp *op, int arg) { -#if TARGET_LONG_BITS <= TCG_TARGET_REG_BITS - return tcg_get_insn_param(op, arg); -#else - return tcg_get_insn_param(op, arg * 2) | - ((uint64_t)tcg_get_insn_param(op, arg * 2 + 1) << 32); -#endif + if (TCG_TARGET_REG_BITS == 64) { + return tcg_get_insn_param(op, arg); + } else { + return deposit64(tcg_get_insn_param(op, arg * 2), 32, 32, + tcg_get_insn_param(op, arg * 2 + 1)); + } } -static inline void tcg_set_insn_start_param(TCGOp *op, int arg, target_ulong v) +static inline void tcg_set_insn_start_param(TCGOp *op, int arg, uint64_t v) { -#if TARGET_LONG_BITS <= TCG_TARGET_REG_BITS - tcg_set_insn_param(op, arg, v); -#else - tcg_set_insn_param(op, arg * 2, v); - tcg_set_insn_param(op, arg * 2 + 1, v >> 32); -#endif + if (TCG_TARGET_REG_BITS == 64) { + tcg_set_insn_param(op, arg, v); + } else { + tcg_set_insn_param(op, arg * 2, v); + tcg_set_insn_param(op, arg * 2 + 1, v >> 32); + } } /* The last op that was emitted. */ diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/translate-all.c +++ b/accel/tcg/translate-all.c @@ -XXX,XX +XXX,XX @@ QEMU_BUILD_BUG_ON(CPU_TRACE_DSTATE_MAX_EVENTS > TBContext tb_ctx; -/* Encode VAL as a signed leb128 sequence at P. - Return P incremented past the encoded value. */ -static uint8_t *encode_sleb128(uint8_t *p, target_long val) +/* + * Encode VAL as a signed leb128 sequence at P. + * Return P incremented past the encoded value. + */ +static uint8_t *encode_sleb128(uint8_t *p, int64_t val) { int more, byte; @@ -XXX,XX +XXX,XX @@ static uint8_t *encode_sleb128(uint8_t *p, target_long val) return p; } -/* Decode a signed leb128 sequence at *PP; increment *PP past the - decoded value. Return the decoded value. */ -static target_long decode_sleb128(const uint8_t **pp) +/* + * Decode a signed leb128 sequence at *PP; increment *PP past the + * decoded value. Return the decoded value. + */ +static int64_t decode_sleb128(const uint8_t **pp) { const uint8_t *p = *pp; - target_long val = 0; + int64_t val = 0; int byte, shift = 0; do { byte = *p++; - val |= (target_ulong)(byte & 0x7f) << shift; + val |= (int64_t)(byte & 0x7f) << shift; shift += 7; } while (byte & 0x80); if (shift < TARGET_LONG_BITS && (byte & 0x40)) { - val |= -(target_ulong)1 << shift; + val |= -(int64_t)1 << shift; } *pp = p; @@ -XXX,XX +XXX,XX @@ static int encode_search(TranslationBlock *tb, uint8_t *block) int i, j, n; for (i = 0, n = tb->icount; i < n; ++i) { - target_ulong prev; + uint64_t prev; for (j = 0; j < TARGET_INSN_START_WORDS; ++j) { if (i == 0) { @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, /* Dump header and the first instruction */ fprintf(logfile, "OUT: [size=%d]\n", gen_code_size); fprintf(logfile, - " -- guest addr 0x" TARGET_FMT_lx " + tb prologue\n", + " -- guest addr 0x%016" PRIx64 " + tb prologue\n", tcg_ctx->gen_insn_data[insn][0]); chunk_start = tcg_ctx->gen_insn_end_off[insn]; disas(logfile, tb->tc.ptr, chunk_start); @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, while (insn < tb->icount) { size_t chunk_end = tcg_ctx->gen_insn_end_off[insn]; if (chunk_end > chunk_start) { - fprintf(logfile, " -- guest addr 0x" TARGET_FMT_lx "\n", + fprintf(logfile, " -- guest addr 0x%016" PRIx64 "\n", tcg_ctx->gen_insn_data[insn][0]); disas(logfile, tb->tc.ptr + chunk_start, chunk_end - chunk_start); diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static void tcg_dump_ops(TCGContext *s, FILE *f, bool have_prefs) col += ne_fprintf(f, "\n ----"); for (i = 0; i < TARGET_INSN_START_WORDS; ++i) { - target_ulong a; -#if TARGET_LONG_BITS > TCG_TARGET_REG_BITS - a = deposit64(op->args[i * 2], 32, 32, op->args[i * 2 + 1]); -#else - a = op->args[i]; -#endif - col += ne_fprintf(f, " " TARGET_FMT_lx, a); + col += ne_fprintf(f, " %016" PRIx64, + tcg_get_insn_start_param(op, i)); } } else if (c == INDEX_op_call) { const TCGHelperInfo *info = tcg_call_info(op); @@ -XXX,XX +XXX,XX @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb, target_ulong pc_start) } num_insns++; for (i = 0; i < TARGET_INSN_START_WORDS; ++i) { - target_ulong a; -#if TARGET_LONG_BITS > TCG_TARGET_REG_BITS - a = deposit64(op->args[i * 2], 32, 32, op->args[i * 2 + 1]); -#else - a = op->args[i]; -#endif - s->gen_insn_data[num_insns][i] = a; + s->gen_insn_data[num_insns][i] = + tcg_get_insn_start_param(op, i); } break; case INDEX_op_discard: -- 2.34.1 Always pass the target address as uint64_t. Adjust tcg_out_{ld,st}_helper_args to match. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-ldst.h | 26 +++++++++--------- accel/tcg/cputlb.c | 26 +++++++++--------- accel/tcg/user-exec.c | 26 +++++++++--------- tcg/tcg.c | 62 ++++++++++++++++++++++++++++++++---------- 4 files changed, 87 insertions(+), 53 deletions(-) diff --git a/include/tcg/tcg-ldst.h b/include/tcg/tcg-ldst.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-ldst.h +++ b/include/tcg/tcg-ldst.h @@ -XXX,XX +XXX,XX @@ #define TCG_LDST_H /* Value zero-extended to tcg register size. */ -tcg_target_ulong helper_ldub_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldub_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_lduw_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_lduw_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldul_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); -uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, +uint64_t helper_ldq_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); -Int128 helper_ld16_mmu(CPUArchState *env, target_ulong addr, +Int128 helper_ld16_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); /* Value sign-extended to tcg register size. */ -tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); -tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr); /* * Value extended to at least uint32_t, so that some ABIs do not require * zero-extension from uint8_t or uint16_t. */ -void helper_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stb_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr); -void helper_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stw_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr); -void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stl_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr); -void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, +void helper_stq_mmu(CPUArchState *env, uint64_t addr, uint64_t val, MemOpIdx oi, uintptr_t retaddr); -void helper_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, +void helper_st16_mmu(CPUArchState *env, uint64_t addr, Int128 val, MemOpIdx oi, uintptr_t retaddr); #endif /* TCG_LDST_H */ diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -XXX,XX +XXX,XX @@ static uint8_t do_ld1_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return do_ld_1(env, &l.page[0], l.mmu_idx, access_type, ra); } -tcg_target_ulong helper_ldub_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldub_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_8); @@ -XXX,XX +XXX,XX @@ static uint16_t do_ld2_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return ret; } -tcg_target_ulong helper_lduw_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_lduw_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_16); @@ -XXX,XX +XXX,XX @@ static uint32_t do_ld4_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return ret; } -tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldul_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_32); @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld8_mmu(CPUArchState *env, target_ulong addr, MemOpIdx oi, return ret; } -uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, +uint64_t helper_ldq_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_64); @@ -XXX,XX +XXX,XX @@ uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, * avoid this for 64-bit data, or for 32-bit data on 32-bit host. */ -tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr) { return (int8_t)helper_ldub_mmu(env, addr, oi, retaddr); } -tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr) { return (int16_t)helper_lduw_mmu(env, addr, oi, retaddr); } -tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t retaddr) { return (int32_t)helper_ldul_mmu(env, addr, oi, retaddr); @@ -XXX,XX +XXX,XX @@ static Int128 do_ld16_mmu(CPUArchState *env, target_ulong addr, return ret; } -Int128 helper_ld16_mmu(CPUArchState *env, target_ulong addr, +Int128 helper_ld16_mmu(CPUArchState *env, uint64_t addr, uint32_t oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_128); @@ -XXX,XX +XXX,XX @@ static void do_st_8(CPUArchState *env, MMULookupPageData *p, uint64_t val, } } -void helper_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stb_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t ra) { MMULookupLocals l; @@ -XXX,XX +XXX,XX @@ static void do_st2_mmu(CPUArchState *env, target_ulong addr, uint16_t val, do_st_1(env, &l.page[1], b, l.mmu_idx, ra); } -void helper_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stw_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_16); @@ -XXX,XX +XXX,XX @@ static void do_st4_mmu(CPUArchState *env, target_ulong addr, uint32_t val, (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, l.memop, ra); } -void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stl_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_32); @@ -XXX,XX +XXX,XX @@ static void do_st8_mmu(CPUArchState *env, target_ulong addr, uint64_t val, (void) do_st_leN(env, &l.page[1], val, l.mmu_idx, l.memop, ra); } -void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, +void helper_stq_mmu(CPUArchState *env, uint64_t addr, uint64_t val, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_64); @@ -XXX,XX +XXX,XX @@ static void do_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, } } -void helper_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, +void helper_st16_mmu(CPUArchState *env, uint64_t addr, Int128 val, MemOpIdx oi, uintptr_t retaddr) { tcg_debug_assert((get_memop(oi) & MO_SIZE) == MO_128); diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/user-exec.c +++ b/accel/tcg/user-exec.c @@ -XXX,XX +XXX,XX @@ static uint8_t do_ld1_mmu(CPUArchState *env, abi_ptr addr, return ret; } -tcg_target_ulong helper_ldub_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldub_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { return do_ld1_mmu(env, addr, get_memop(oi), ra); } -tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsb_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { return (int8_t)do_ld1_mmu(env, addr, get_memop(oi), ra); @@ -XXX,XX +XXX,XX @@ static uint16_t do_ld2_he_mmu(CPUArchState *env, abi_ptr addr, return ret; } -tcg_target_ulong helper_lduw_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_lduw_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ tcg_target_ulong helper_lduw_mmu(CPUArchState *env, target_ulong addr, return ret; } -tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsw_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static uint32_t do_ld4_he_mmu(CPUArchState *env, abi_ptr addr, return ret; } -tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldul_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ tcg_target_ulong helper_ldul_mmu(CPUArchState *env, target_ulong addr, return ret; } -tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, target_ulong addr, +tcg_target_ulong helper_ldsl_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static uint64_t do_ld8_he_mmu(CPUArchState *env, abi_ptr addr, return ret; } -uint64_t helper_ldq_mmu(CPUArchState *env, target_ulong addr, +uint64_t helper_ldq_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static Int128 do_ld16_he_mmu(CPUArchState *env, abi_ptr addr, return ret; } -Int128 helper_ld16_mmu(CPUArchState *env, target_ulong addr, +Int128 helper_ld16_mmu(CPUArchState *env, uint64_t addr, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static void do_st1_mmu(CPUArchState *env, abi_ptr addr, uint8_t val, clear_helper_retaddr(); } -void helper_stb_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stb_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t ra) { do_st1_mmu(env, addr, val, get_memop(oi), ra); @@ -XXX,XX +XXX,XX @@ static void do_st2_he_mmu(CPUArchState *env, abi_ptr addr, uint16_t val, clear_helper_retaddr(); } -void helper_stw_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stw_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static void do_st4_he_mmu(CPUArchState *env, abi_ptr addr, uint32_t val, clear_helper_retaddr(); } -void helper_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val, +void helper_stl_mmu(CPUArchState *env, uint64_t addr, uint32_t val, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static void do_st8_he_mmu(CPUArchState *env, abi_ptr addr, uint64_t val, clear_helper_retaddr(); } -void helper_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val, +void helper_stq_mmu(CPUArchState *env, uint64_t addr, uint64_t val, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static void do_st16_he_mmu(CPUArchState *env, abi_ptr addr, Int128 val, clear_helper_retaddr(); } -void helper_st16_mmu(CPUArchState *env, target_ulong addr, Int128 val, +void helper_st16_mmu(CPUArchState *env, uint64_t addr, Int128 val, MemOpIdx oi, uintptr_t ra) { MemOp mop = get_memop(oi); diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_ld32_mmu = { .flags = TCG_CALL_NO_WG, .typemask = dh_typemask(ttl, 0) /* return tcg_target_ulong */ | dh_typemask(env, 1) - | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i64, 2) /* uint64_t addr */ | dh_typemask(i32, 3) /* unsigned oi */ | dh_typemask(ptr, 4) /* uintptr_t ra */ }; @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_ld64_mmu = { .flags = TCG_CALL_NO_WG, .typemask = dh_typemask(i64, 0) /* return uint64_t */ | dh_typemask(env, 1) - | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i64, 2) /* uint64_t addr */ | dh_typemask(i32, 3) /* unsigned oi */ | dh_typemask(ptr, 4) /* uintptr_t ra */ }; @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_ld128_mmu = { .flags = TCG_CALL_NO_WG, .typemask = dh_typemask(i128, 0) /* return Int128 */ | dh_typemask(env, 1) - | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i64, 2) /* uint64_t addr */ | dh_typemask(i32, 3) /* unsigned oi */ | dh_typemask(ptr, 4) /* uintptr_t ra */ }; @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_st32_mmu = { .flags = TCG_CALL_NO_WG, .typemask = dh_typemask(void, 0) | dh_typemask(env, 1) - | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i64, 2) /* uint64_t addr */ | dh_typemask(i32, 3) /* uint32_t data */ | dh_typemask(i32, 4) /* unsigned oi */ | dh_typemask(ptr, 5) /* uintptr_t ra */ @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_st64_mmu = { .flags = TCG_CALL_NO_WG, .typemask = dh_typemask(void, 0) | dh_typemask(env, 1) - | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i64, 2) /* uint64_t addr */ | dh_typemask(i64, 3) /* uint64_t data */ | dh_typemask(i32, 4) /* unsigned oi */ | dh_typemask(ptr, 5) /* uintptr_t ra */ @@ -XXX,XX +XXX,XX @@ static TCGHelperInfo info_helper_st128_mmu = { .flags = TCG_CALL_NO_WG, .typemask = dh_typemask(void, 0) | dh_typemask(env, 1) - | dh_typemask(tl, 2) /* target_ulong addr */ + | dh_typemask(i64, 2) /* uint64_t addr */ | dh_typemask(i128, 3) /* Int128 data */ | dh_typemask(i32, 4) /* unsigned oi */ | dh_typemask(ptr, 5) /* uintptr_t ra */ @@ -XXX,XX +XXX,XX @@ static void tcg_out_ld_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, next_arg = 1; loc = &info->in[next_arg]; - nmov = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_TL, TCG_TYPE_TL, - ldst->addrlo_reg, ldst->addrhi_reg); - next_arg += nmov; + if (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 64) { + nmov = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, TCG_TYPE_TL, + ldst->addrlo_reg, ldst->addrhi_reg); + tcg_out_helper_load_slots(s, nmov, mov, parm); + next_arg += nmov; + } else { + /* + * 32-bit host with 32-bit guest: zero-extend the guest address + * to 64-bits for the helper by storing the low part, then + * load a zero for the high part. + */ + tcg_out_helper_add_mov(mov, loc + HOST_BIG_ENDIAN, + TCG_TYPE_I32, TCG_TYPE_I32, + ldst->addrlo_reg, -1); + tcg_out_helper_load_slots(s, 1, mov, parm); - tcg_out_helper_load_slots(s, nmov, mov, parm); + tcg_out_helper_load_imm(s, loc[!HOST_BIG_ENDIAN].arg_slot, + TCG_TYPE_I32, 0, parm); + next_arg += 2; + } switch (info->out_kind) { case TCG_CALL_RET_NORMAL: @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, /* Handle addr argument. */ loc = &info->in[next_arg]; - n = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_TL, TCG_TYPE_TL, - ldst->addrlo_reg, ldst->addrhi_reg); - next_arg += n; - nmov += n; + if (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 64) { + n = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, TCG_TYPE_TL, + ldst->addrlo_reg, ldst->addrhi_reg); + next_arg += n; + nmov += n; + } else { + /* + * 32-bit host with 32-bit guest: zero-extend the guest address + * to 64-bits for the helper by storing the low part. Later, + * after we have processed the register inputs, we will load a + * zero for the high part. + */ + tcg_out_helper_add_mov(mov, loc + HOST_BIG_ENDIAN, + TCG_TYPE_I32, TCG_TYPE_I32, + ldst->addrlo_reg, -1); + next_arg += 2; + nmov += 1; + } /* Handle data argument. */ loc = &info->in[next_arg]; @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, g_assert_not_reached(); } + if (TCG_TARGET_REG_BITS == 32 && TARGET_LONG_BITS == 32) { + loc = &info->in[1 + !HOST_BIG_ENDIAN]; + tcg_out_helper_load_imm(s, loc->arg_slot, TCG_TYPE_I32, 0, parm); + } + tcg_out_helper_load_common_args(s, ldst, parm, info, next_arg); } -- 2.34.1 Always pass the target address as uint64_t. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/tcg-runtime.h | 4 ++-- accel/tcg/cputlb.c | 5 ++--- accel/tcg/user-exec.c | 5 ++--- tcg/tcg-op-ldst.c | 26 ++++++++++++++++++++++++-- 4 files changed, 30 insertions(+), 10 deletions(-) diff --git a/accel/tcg/tcg-runtime.h b/accel/tcg/tcg-runtime.h index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/tcg-runtime.h +++ b/accel/tcg/tcg-runtime.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_1(exit_atomic, TCG_CALL_NO_WG, noreturn, env) DEF_HELPER_FLAGS_3(memset, TCG_CALL_NO_RWG, ptr, ptr, int, ptr) #endif /* IN_HELPER_PROTO */ -DEF_HELPER_FLAGS_3(ld_i128, TCG_CALL_NO_WG, i128, env, tl, i32) -DEF_HELPER_FLAGS_4(st_i128, TCG_CALL_NO_WG, void, env, tl, i128, i32) +DEF_HELPER_FLAGS_3(ld_i128, TCG_CALL_NO_WG, i128, env, i64, i32) +DEF_HELPER_FLAGS_4(st_i128, TCG_CALL_NO_WG, void, env, i64, i128, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgb, TCG_CALL_NO_WG, i32, env, tl, i32, i32, i32) diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -XXX,XX +XXX,XX @@ Int128 helper_ld16_mmu(CPUArchState *env, uint64_t addr, return do_ld16_mmu(env, addr, oi, retaddr); } -Int128 helper_ld_i128(CPUArchState *env, target_ulong addr, uint32_t oi) +Int128 helper_ld_i128(CPUArchState *env, uint64_t addr, uint32_t oi) { return helper_ld16_mmu(env, addr, oi, GETPC()); } @@ -XXX,XX +XXX,XX @@ void helper_st16_mmu(CPUArchState *env, uint64_t addr, Int128 val, do_st16_mmu(env, addr, val, oi, retaddr); } -void helper_st_i128(CPUArchState *env, target_ulong addr, Int128 val, - MemOpIdx oi) +void helper_st_i128(CPUArchState *env, uint64_t addr, Int128 val, MemOpIdx oi) { helper_st16_mmu(env, addr, val, oi, GETPC()); } diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/user-exec.c +++ b/accel/tcg/user-exec.c @@ -XXX,XX +XXX,XX @@ Int128 helper_ld16_mmu(CPUArchState *env, uint64_t addr, return ret; } -Int128 helper_ld_i128(CPUArchState *env, target_ulong addr, MemOpIdx oi) +Int128 helper_ld_i128(CPUArchState *env, uint64_t addr, MemOpIdx oi) { return helper_ld16_mmu(env, addr, oi, GETPC()); } @@ -XXX,XX +XXX,XX @@ void helper_st16_mmu(CPUArchState *env, uint64_t addr, Int128 val, do_st16_he_mmu(env, addr, val, mop, ra); } -void helper_st_i128(CPUArchState *env, target_ulong addr, - Int128 val, MemOpIdx oi) +void helper_st_i128(CPUArchState *env, uint64_t addr, Int128 val, MemOpIdx oi) { helper_st16_mmu(env, addr, val, oi, GETPC()); } diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) #define tcg_temp_ebb_new tcg_temp_ebb_new_i32 #endif +static TCGv_i64 maybe_extend_addr64(TCGv addr) +{ +#if TARGET_LONG_BITS == 32 + TCGv_i64 a64 = tcg_temp_ebb_new_i64(); + tcg_gen_extu_i32_i64(a64, addr); + return a64; +#else + return addr; +#endif +} + +static void maybe_free_addr64(TCGv_i64 a64) +{ +#if TARGET_LONG_BITS == 32 + tcg_temp_free_i64(a64); +#endif +} + void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) { const MemOpIdx oi = make_memop_idx(memop, idx); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) tcg_gen_bswap64_i64(y, y); } } else { - gen_helper_ld_i128(val, cpu_env, addr, tcg_constant_i32(oi)); + TCGv_i64 a64 = maybe_extend_addr64(addr); + gen_helper_ld_i128(val, cpu_env, a64, tcg_constant_i32(oi)); + maybe_free_addr64(a64); } plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) } tcg_temp_free(addr_p8); } else { - gen_helper_st_i128(cpu_env, addr, val, tcg_constant_i32(oi)); + TCGv_i64 a64 = maybe_extend_addr64(addr); + gen_helper_st_i128(cpu_env, a64, val, tcg_constant_i32(oi)); + maybe_free_addr64(a64); } plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); -- 2.34.1 Always pass the target address as uint64_t. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/tcg-runtime.h | 46 +++++++++++++++++------------------ tcg/tcg-op-ldst.c | 38 ++++++++++++++++++++--------- accel/tcg/atomic_common.c.inc | 14 +++++------ 3 files changed, 57 insertions(+), 41 deletions(-) diff --git a/accel/tcg/tcg-runtime.h b/accel/tcg/tcg-runtime.h index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/tcg-runtime.h +++ b/accel/tcg/tcg-runtime.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_3(ld_i128, TCG_CALL_NO_WG, i128, env, i64, i32) DEF_HELPER_FLAGS_4(st_i128, TCG_CALL_NO_WG, void, env, i64, i128, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgb, TCG_CALL_NO_WG, - i32, env, tl, i32, i32, i32) + i32, env, i64, i32, i32, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgw_be, TCG_CALL_NO_WG, - i32, env, tl, i32, i32, i32) + i32, env, i64, i32, i32, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgw_le, TCG_CALL_NO_WG, - i32, env, tl, i32, i32, i32) + i32, env, i64, i32, i32, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgl_be, TCG_CALL_NO_WG, - i32, env, tl, i32, i32, i32) + i32, env, i64, i32, i32, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgl_le, TCG_CALL_NO_WG, - i32, env, tl, i32, i32, i32) + i32, env, i64, i32, i32, i32) #ifdef CONFIG_ATOMIC64 DEF_HELPER_FLAGS_5(atomic_cmpxchgq_be, TCG_CALL_NO_WG, - i64, env, tl, i64, i64, i32) + i64, env, i64, i64, i64, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgq_le, TCG_CALL_NO_WG, - i64, env, tl, i64, i64, i32) + i64, env, i64, i64, i64, i32) #endif #ifdef CONFIG_CMPXCHG128 DEF_HELPER_FLAGS_5(atomic_cmpxchgo_be, TCG_CALL_NO_WG, - i128, env, tl, i128, i128, i32) + i128, env, i64, i128, i128, i32) DEF_HELPER_FLAGS_5(atomic_cmpxchgo_le, TCG_CALL_NO_WG, - i128, env, tl, i128, i128, i32) + i128, env, i64, i128, i128, i32) #endif DEF_HELPER_FLAGS_5(nonatomic_cmpxchgo_be, TCG_CALL_NO_WG, - i128, env, tl, i128, i128, i32) + i128, env, i64, i128, i128, i32) DEF_HELPER_FLAGS_5(nonatomic_cmpxchgo_le, TCG_CALL_NO_WG, - i128, env, tl, i128, i128, i32) + i128, env, i64, i128, i128, i32) #ifdef CONFIG_ATOMIC64 #define GEN_ATOMIC_HELPERS(NAME) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), b), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), w_le), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), w_be), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), l_le), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), l_be), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), q_le), \ - TCG_CALL_NO_WG, i64, env, tl, i64, i32) \ + TCG_CALL_NO_WG, i64, env, i64, i64, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), q_be), \ - TCG_CALL_NO_WG, i64, env, tl, i64, i32) + TCG_CALL_NO_WG, i64, env, i64, i64, i32) #else #define GEN_ATOMIC_HELPERS(NAME) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), b), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), w_le), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), w_be), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), l_le), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) \ + TCG_CALL_NO_WG, i32, env, i64, i32, i32) \ DEF_HELPER_FLAGS_4(glue(glue(atomic_, NAME), l_be), \ - TCG_CALL_NO_WG, i32, env, tl, i32, i32) + TCG_CALL_NO_WG, i32, env, i64, i32, i32) #endif /* CONFIG_ATOMIC64 */ GEN_ATOMIC_HELPERS(fetch_add) diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ static void tcg_gen_ext_i64(TCGv_i64 ret, TCGv_i64 val, MemOp opc) } } -typedef void (*gen_atomic_cx_i32)(TCGv_i32, TCGv_env, TCGv, +typedef void (*gen_atomic_cx_i32)(TCGv_i32, TCGv_env, TCGv_i64, TCGv_i32, TCGv_i32, TCGv_i32); -typedef void (*gen_atomic_cx_i64)(TCGv_i64, TCGv_env, TCGv, +typedef void (*gen_atomic_cx_i64)(TCGv_i64, TCGv_env, TCGv_i64, TCGv_i64, TCGv_i64, TCGv_i32); -typedef void (*gen_atomic_cx_i128)(TCGv_i128, TCGv_env, TCGv, +typedef void (*gen_atomic_cx_i128)(TCGv_i128, TCGv_env, TCGv_i64, TCGv_i128, TCGv_i128, TCGv_i32); -typedef void (*gen_atomic_op_i32)(TCGv_i32, TCGv_env, TCGv, +typedef void (*gen_atomic_op_i32)(TCGv_i32, TCGv_env, TCGv_i64, TCGv_i32, TCGv_i32); -typedef void (*gen_atomic_op_i64)(TCGv_i64, TCGv_env, TCGv, +typedef void (*gen_atomic_op_i64)(TCGv_i64, TCGv_env, TCGv_i64, TCGv_i64, TCGv_i32); #ifdef CONFIG_ATOMIC64 @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, TCGv_i32 newv, TCGArg idx, MemOp memop) { gen_atomic_cx_i32 gen; + TCGv_i64 a64; MemOpIdx oi; if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, tcg_debug_assert(gen != NULL); oi = make_memop_idx(memop & ~MO_SIGN, idx); - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + a64 = maybe_extend_addr64(addr); + gen(retv, cpu_env, a64, cmpv, newv, tcg_constant_i32(oi)); + maybe_free_addr64(a64); if (memop & MO_SIGN) { tcg_gen_ext_i32(retv, retv, memop); @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; if (gen) { MemOpIdx oi = make_memop_idx(memop, idx); - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + TCGv_i64 a64 = maybe_extend_addr64(addr); + gen(retv, cpu_env, a64, cmpv, newv, tcg_constant_i32(oi)); + maybe_free_addr64(a64); return; } @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, ? gen_helper_nonatomic_cmpxchgo_le : gen_helper_nonatomic_cmpxchgo_be); MemOpIdx oi = make_memop_idx(memop, idx); + TCGv_i64 a64; tcg_debug_assert((memop & MO_SIZE) == MO_128); tcg_debug_assert((memop & MO_SIGN) == 0); - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + a64 = maybe_extend_addr64(addr); + gen(retv, cpu_env, a64, cmpv, newv, tcg_constant_i32(oi)); + maybe_free_addr64(a64); } else { TCGv_i128 oldv = tcg_temp_ebb_new_i128(); TCGv_i128 tmpv = tcg_temp_ebb_new_i128(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, if (gen) { MemOpIdx oi = make_memop_idx(memop, idx); - gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); + TCGv_i64 a64 = maybe_extend_addr64(addr); + gen(retv, cpu_env, a64, cmpv, newv, tcg_constant_i32(oi)); + maybe_free_addr64(a64); return; } @@ -XXX,XX +XXX,XX @@ static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, TCGArg idx, MemOp memop, void * const table[]) { gen_atomic_op_i32 gen; + TCGv_i64 a64; MemOpIdx oi; memop = tcg_canonicalize_memop(memop, 0, 0); @@ -XXX,XX +XXX,XX @@ static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, tcg_debug_assert(gen != NULL); oi = make_memop_idx(memop & ~MO_SIGN, idx); - gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); + a64 = maybe_extend_addr64(addr); + gen(ret, cpu_env, a64, val, tcg_constant_i32(oi)); + maybe_free_addr64(a64); if (memop & MO_SIGN) { tcg_gen_ext_i32(ret, ret, memop); @@ -XXX,XX +XXX,XX @@ static void do_atomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, if ((memop & MO_SIZE) == MO_64) { #ifdef CONFIG_ATOMIC64 gen_atomic_op_i64 gen; + TCGv_i64 a64; MemOpIdx oi; gen = table[memop & (MO_SIZE | MO_BSWAP)]; tcg_debug_assert(gen != NULL); oi = make_memop_idx(memop & ~MO_SIGN, idx); - gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); + a64 = maybe_extend_addr64(addr); + gen(ret, cpu_env, a64, val, tcg_constant_i32(oi)); + maybe_free_addr64(a64); #else gen_helper_exit_atomic(cpu_env); /* Produce a result, so that we have a well-formed opcode stream diff --git a/accel/tcg/atomic_common.c.inc b/accel/tcg/atomic_common.c.inc index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/atomic_common.c.inc +++ b/accel/tcg/atomic_common.c.inc @@ -XXX,XX +XXX,XX @@ * See the COPYING file in the top-level directory. */ -static void atomic_trace_rmw_post(CPUArchState *env, target_ulong addr, +static void atomic_trace_rmw_post(CPUArchState *env, uint64_t addr, MemOpIdx oi) { qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_RW); } #if HAVE_ATOMIC128 -static void atomic_trace_ld_post(CPUArchState *env, target_ulong addr, +static void atomic_trace_ld_post(CPUArchState *env, uint64_t addr, MemOpIdx oi) { qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_R); } -static void atomic_trace_st_post(CPUArchState *env, target_ulong addr, +static void atomic_trace_st_post(CPUArchState *env, uint64_t addr, MemOpIdx oi) { qemu_plugin_vcpu_mem_cb(env_cpu(env), addr, oi, QEMU_PLUGIN_MEM_W); @@ -XXX,XX +XXX,XX @@ static void atomic_trace_st_post(CPUArchState *env, target_ulong addr, */ #define CMPXCHG_HELPER(OP, TYPE) \ - TYPE HELPER(atomic_##OP)(CPUArchState *env, target_ulong addr, \ + TYPE HELPER(atomic_##OP)(CPUArchState *env, uint64_t addr, \ TYPE oldv, TYPE newv, uint32_t oi) \ { return cpu_atomic_##OP##_mmu(env, addr, oldv, newv, oi, GETPC()); } @@ -XXX,XX +XXX,XX @@ CMPXCHG_HELPER(cmpxchgo_le, Int128) #undef CMPXCHG_HELPER -Int128 HELPER(nonatomic_cmpxchgo_be)(CPUArchState *env, target_ulong addr, +Int128 HELPER(nonatomic_cmpxchgo_be)(CPUArchState *env, uint64_t addr, Int128 cmpv, Int128 newv, uint32_t oi) { #if TCG_TARGET_REG_BITS == 32 @@ -XXX,XX +XXX,XX @@ Int128 HELPER(nonatomic_cmpxchgo_be)(CPUArchState *env, target_ulong addr, #endif } -Int128 HELPER(nonatomic_cmpxchgo_le)(CPUArchState *env, target_ulong addr, +Int128 HELPER(nonatomic_cmpxchgo_le)(CPUArchState *env, uint64_t addr, Int128 cmpv, Int128 newv, uint32_t oi) { #if TCG_TARGET_REG_BITS == 32 @@ -XXX,XX +XXX,XX @@ Int128 HELPER(nonatomic_cmpxchgo_le)(CPUArchState *env, target_ulong addr, } #define ATOMIC_HELPER(OP, TYPE) \ - TYPE HELPER(glue(atomic_,OP))(CPUArchState *env, target_ulong addr, \ + TYPE HELPER(glue(atomic_,OP))(CPUArchState *env, uint64_t addr, \ TYPE val, uint32_t oi) \ { return glue(glue(cpu_atomic_,OP),_mmu)(env, addr, val, oi, GETPC()); } -- 2.34.1 Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 2 +- tcg/tcg.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -XXX,XX +XXX,XX @@ void tcg_register_thread(void); void tcg_prologue_init(TCGContext *s); void tcg_func_start(TCGContext *s); -int tcg_gen_code(TCGContext *s, TranslationBlock *tb, target_ulong pc_start); +int tcg_gen_code(TCGContext *s, TranslationBlock *tb, uint64_t pc_start); void tb_target_set_jmp_target(const TranslationBlock *, int, uintptr_t, uintptr_t); diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ int64_t tcg_cpu_exec_time(void) #endif -int tcg_gen_code(TCGContext *s, TranslationBlock *tb, target_ulong pc_start) +int tcg_gen_code(TCGContext *s, TranslationBlock *tb, uint64_t pc_start) { #ifdef CONFIG_PROFILER TCGProfile *prof = &s->prof; -- 2.34.1 As gen_mem_wrapped is only used in plugin_gen_empty_mem_callback, we can avoid the curiosity of union mem_gen_fn by inlining it. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/plugin-gen.c | 30 ++++++------------------------ 1 file changed, 6 insertions(+), 24 deletions(-) diff --git a/accel/tcg/plugin-gen.c b/accel/tcg/plugin-gen.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/plugin-gen.c +++ b/accel/tcg/plugin-gen.c @@ -XXX,XX +XXX,XX @@ static void plugin_gen_empty_callback(enum plugin_gen_from from) } } -union mem_gen_fn { - void (*mem_fn)(TCGv, uint32_t); - void (*inline_fn)(void); -}; - -static void gen_mem_wrapped(enum plugin_gen_cb type, - const union mem_gen_fn *f, TCGv addr, - uint32_t info, bool is_mem) +void plugin_gen_empty_mem_callback(TCGv addr, uint32_t info) { enum qemu_plugin_mem_rw rw = get_plugin_meminfo_rw(info); - gen_plugin_cb_start(PLUGIN_GEN_FROM_MEM, type, rw); - if (is_mem) { - f->mem_fn(addr, info); - } else { - f->inline_fn(); - } + gen_plugin_cb_start(PLUGIN_GEN_FROM_MEM, PLUGIN_GEN_CB_MEM, rw); + gen_empty_mem_cb(addr, info); tcg_gen_plugin_cb_end(); -} -void plugin_gen_empty_mem_callback(TCGv addr, uint32_t info) -{ - union mem_gen_fn fn; - - fn.mem_fn = gen_empty_mem_cb; - gen_mem_wrapped(PLUGIN_GEN_CB_MEM, &fn, addr, info, true); - - fn.inline_fn = gen_empty_inline_cb; - gen_mem_wrapped(PLUGIN_GEN_CB_INLINE, &fn, 0, info, false); + gen_plugin_cb_start(PLUGIN_GEN_FROM_MEM, PLUGIN_GEN_CB_INLINE, rw); + gen_empty_inline_cb(); + tcg_gen_plugin_cb_end(); } static TCGOp *find_op(TCGOp *op, TCGOpcode opc) -- 2.34.1 As do_gen_mem_cb is called once, merge it into gen_empty_mem_cb. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/plugin-gen.c | 39 +++++++++++++++++---------------------- 1 file changed, 17 insertions(+), 22 deletions(-) diff --git a/accel/tcg/plugin-gen.c b/accel/tcg/plugin-gen.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/plugin-gen.c +++ b/accel/tcg/plugin-gen.c @@ -XXX,XX +XXX,XX @@ void HELPER(plugin_vcpu_mem_cb)(unsigned int vcpu_index, void *userdata) { } -static void do_gen_mem_cb(TCGv vaddr, uint32_t info) -{ - TCGv_i32 cpu_index = tcg_temp_ebb_new_i32(); - TCGv_i32 meminfo = tcg_temp_ebb_new_i32(); - TCGv_i64 vaddr64 = tcg_temp_ebb_new_i64(); - TCGv_ptr udata = tcg_temp_ebb_new_ptr(); - - tcg_gen_movi_i32(meminfo, info); - tcg_gen_movi_ptr(udata, 0); - tcg_gen_ld_i32(cpu_index, cpu_env, - -offsetof(ArchCPU, env) + offsetof(CPUState, cpu_index)); - tcg_gen_extu_tl_i64(vaddr64, vaddr); - - gen_helper_plugin_vcpu_mem_cb(cpu_index, meminfo, vaddr64, udata); - - tcg_temp_free_ptr(udata); - tcg_temp_free_i64(vaddr64); - tcg_temp_free_i32(meminfo); - tcg_temp_free_i32(cpu_index); -} - static void gen_empty_udata_cb(void) { TCGv_i32 cpu_index = tcg_temp_ebb_new_i32(); @@ -XXX,XX +XXX,XX @@ static void gen_empty_inline_cb(void) static void gen_empty_mem_cb(TCGv addr, uint32_t info) { - do_gen_mem_cb(addr, info); + TCGv_i32 cpu_index = tcg_temp_ebb_new_i32(); + TCGv_i32 meminfo = tcg_temp_ebb_new_i32(); + TCGv_i64 addr64 = tcg_temp_ebb_new_i64(); + TCGv_ptr udata = tcg_temp_ebb_new_ptr(); + + tcg_gen_movi_i32(meminfo, info); + tcg_gen_movi_ptr(udata, 0); + tcg_gen_ld_i32(cpu_index, cpu_env, + -offsetof(ArchCPU, env) + offsetof(CPUState, cpu_index)); + tcg_gen_extu_tl_i64(addr64, addr); + + gen_helper_plugin_vcpu_mem_cb(cpu_index, meminfo, addr64, udata); + + tcg_temp_free_ptr(udata); + tcg_temp_free_i64(addr64); + tcg_temp_free_i32(meminfo); + tcg_temp_free_i32(cpu_index); } /* -- 2.34.1 We only need to make copies for loads, when the destination overlaps the address. For now, only eliminate the copy for stores and 128-bit loads. Rename plugin_prep_mem_callbacks to plugin_maybe_preserve_addr, returning NULL if no copy is made. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg-op-ldst.c | 38 ++++++++++++++++++++------------------ 1 file changed, 20 insertions(+), 18 deletions(-) diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ static void tcg_gen_req_mo(TCGBar type) } } -static inline TCGv plugin_prep_mem_callbacks(TCGv vaddr) +/* Only required for loads, where value might overlap addr. */ +static TCGv plugin_maybe_preserve_addr(TCGv vaddr) { #ifdef CONFIG_PLUGIN if (tcg_ctx->plugin_insn != NULL) { @@ -XXX,XX +XXX,XX @@ static inline TCGv plugin_prep_mem_callbacks(TCGv vaddr) return temp; } #endif - return vaddr; + return NULL; } -static void plugin_gen_mem_callbacks(TCGv vaddr, MemOpIdx oi, - enum qemu_plugin_mem_rw rw) +static void +plugin_gen_mem_callbacks(TCGv copy_addr, TCGv orig_addr, MemOpIdx oi, + enum qemu_plugin_mem_rw rw) { #ifdef CONFIG_PLUGIN if (tcg_ctx->plugin_insn != NULL) { qemu_plugin_meminfo_t info = make_plugin_meminfo(oi, rw); - plugin_gen_empty_mem_callback(vaddr, info); - tcg_temp_free(vaddr); + plugin_gen_empty_mem_callback(copy_addr ? : orig_addr, info); + if (copy_addr) { + tcg_temp_free(copy_addr); + } } #endif } @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) { MemOp orig_memop; MemOpIdx oi; + TCGv copy_addr; tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); memop = tcg_canonicalize_memop(memop, 0, 0); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) } } - addr = plugin_prep_mem_callbacks(addr); + copy_addr = plugin_maybe_preserve_addr(addr); gen_ldst_i32(INDEX_op_qemu_ld_i32, val, addr, memop, idx); - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); + plugin_gen_mem_callbacks(copy_addr, addr, oi, QEMU_PLUGIN_MEM_R); if ((orig_memop ^ memop) & MO_BSWAP) { switch (orig_memop & MO_SIZE) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) memop &= ~MO_BSWAP; } - addr = plugin_prep_mem_callbacks(addr); if (TCG_TARGET_HAS_qemu_st8_i32 && (memop & MO_SIZE) == MO_8) { gen_ldst_i32(INDEX_op_qemu_st8_i32, val, addr, memop, idx); } else { gen_ldst_i32(INDEX_op_qemu_st_i32, val, addr, memop, idx); } - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); + plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_W); if (swap) { tcg_temp_free_i32(swap); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) { MemOp orig_memop; MemOpIdx oi; + TCGv copy_addr; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { tcg_gen_qemu_ld_i32(TCGV_LOW(val), addr, idx, memop); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) } } - addr = plugin_prep_mem_callbacks(addr); + copy_addr = plugin_maybe_preserve_addr(addr); gen_ldst_i64(INDEX_op_qemu_ld_i64, val, addr, memop, idx); - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); + plugin_gen_mem_callbacks(copy_addr, addr, oi, QEMU_PLUGIN_MEM_R); if ((orig_memop ^ memop) & MO_BSWAP) { int flags = (orig_memop & MO_SIGN @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) memop &= ~MO_BSWAP; } - addr = plugin_prep_mem_callbacks(addr); gen_ldst_i64(INDEX_op_qemu_st_i64, val, addr, memop, idx); - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); + plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_W); if (swap) { tcg_temp_free_i64(swap); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) tcg_debug_assert((memop & MO_SIGN) == 0); tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); - addr = plugin_prep_mem_callbacks(addr); /* TODO: For now, force 32-bit hosts to use the helper. */ if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) maybe_free_addr64(a64); } - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_R); + plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_R); } void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) tcg_debug_assert((memop & MO_SIGN) == 0); tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); - addr = plugin_prep_mem_callbacks(addr); /* TODO: For now, force 32-bit hosts to use the helper. */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) maybe_free_addr64(a64); } - plugin_gen_mem_callbacks(addr, oi, QEMU_PLUGIN_MEM_W); + plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_W); } static void tcg_gen_ext_i32(TCGv_i32 ret, TCGv_i32 val, MemOp opc) -- 2.34.1 Since we do this inside gen_empty_mem_cb anyway, let's do this earlier inside tcg expansion. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/plugin-gen.h | 4 ++-- accel/tcg/plugin-gen.c | 9 +++------ tcg/tcg-op-ldst.c | 28 ++++++++++++++++++++-------- 3 files changed, 25 insertions(+), 16 deletions(-) diff --git a/include/exec/plugin-gen.h b/include/exec/plugin-gen.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/plugin-gen.h +++ b/include/exec/plugin-gen.h @@ -XXX,XX +XXX,XX @@ void plugin_gen_insn_start(CPUState *cpu, const struct DisasContextBase *db); void plugin_gen_insn_end(void); void plugin_gen_disable_mem_helpers(void); -void plugin_gen_empty_mem_callback(TCGv addr, uint32_t info); +void plugin_gen_empty_mem_callback(TCGv_i64 addr, uint32_t info); static inline void plugin_insn_append(abi_ptr pc, const void *from, size_t size) { @@ -XXX,XX +XXX,XX @@ static inline void plugin_gen_tb_end(CPUState *cpu) static inline void plugin_gen_disable_mem_helpers(void) { } -static inline void plugin_gen_empty_mem_callback(TCGv addr, uint32_t info) +static inline void plugin_gen_empty_mem_callback(TCGv_i64 addr, uint32_t info) { } static inline void plugin_insn_append(abi_ptr pc, const void *from, size_t size) diff --git a/accel/tcg/plugin-gen.c b/accel/tcg/plugin-gen.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/plugin-gen.c +++ b/accel/tcg/plugin-gen.c @@ -XXX,XX +XXX,XX @@ static void gen_empty_inline_cb(void) tcg_temp_free_i64(val); } -static void gen_empty_mem_cb(TCGv addr, uint32_t info) +static void gen_empty_mem_cb(TCGv_i64 addr, uint32_t info) { TCGv_i32 cpu_index = tcg_temp_ebb_new_i32(); TCGv_i32 meminfo = tcg_temp_ebb_new_i32(); - TCGv_i64 addr64 = tcg_temp_ebb_new_i64(); TCGv_ptr udata = tcg_temp_ebb_new_ptr(); tcg_gen_movi_i32(meminfo, info); tcg_gen_movi_ptr(udata, 0); tcg_gen_ld_i32(cpu_index, cpu_env, -offsetof(ArchCPU, env) + offsetof(CPUState, cpu_index)); - tcg_gen_extu_tl_i64(addr64, addr); - gen_helper_plugin_vcpu_mem_cb(cpu_index, meminfo, addr64, udata); + gen_helper_plugin_vcpu_mem_cb(cpu_index, meminfo, addr, udata); tcg_temp_free_ptr(udata); - tcg_temp_free_i64(addr64); tcg_temp_free_i32(meminfo); tcg_temp_free_i32(cpu_index); } @@ -XXX,XX +XXX,XX @@ static void plugin_gen_empty_callback(enum plugin_gen_from from) } } -void plugin_gen_empty_mem_callback(TCGv addr, uint32_t info) +void plugin_gen_empty_mem_callback(TCGv_i64 addr, uint32_t info) { enum qemu_plugin_mem_rw rw = get_plugin_meminfo_rw(info); diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ static void tcg_gen_req_mo(TCGBar type) } /* Only required for loads, where value might overlap addr. */ -static TCGv plugin_maybe_preserve_addr(TCGv vaddr) +static TCGv_i64 plugin_maybe_preserve_addr(TCGv vaddr) { #ifdef CONFIG_PLUGIN if (tcg_ctx->plugin_insn != NULL) { /* Save a copy of the vaddr for use after a load. */ - TCGv temp = tcg_temp_new(); - tcg_gen_mov_tl(temp, vaddr); + TCGv_i64 temp = tcg_temp_ebb_new_i64(); + tcg_gen_extu_tl_i64(temp, vaddr); return temp; } #endif @@ -XXX,XX +XXX,XX @@ static TCGv plugin_maybe_preserve_addr(TCGv vaddr) } static void -plugin_gen_mem_callbacks(TCGv copy_addr, TCGv orig_addr, MemOpIdx oi, +plugin_gen_mem_callbacks(TCGv_i64 copy_addr, TCGv orig_addr, MemOpIdx oi, enum qemu_plugin_mem_rw rw) { #ifdef CONFIG_PLUGIN if (tcg_ctx->plugin_insn != NULL) { qemu_plugin_meminfo_t info = make_plugin_meminfo(oi, rw); - plugin_gen_empty_mem_callback(copy_addr ? : orig_addr, info); + +#if TARGET_LONG_BITS == 64 if (copy_addr) { - tcg_temp_free(copy_addr); + plugin_gen_empty_mem_callback(copy_addr, info); + tcg_temp_free_i64(copy_addr); + } else { + plugin_gen_empty_mem_callback(orig_addr, info); } +#else + if (!copy_addr) { + copy_addr = tcg_temp_ebb_new_i64(); + tcg_gen_extu_tl_i64(copy_addr, orig_addr); + } + plugin_gen_empty_mem_callback(copy_addr, info); + tcg_temp_free_i64(copy_addr); +#endif } #endif } @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) { MemOp orig_memop; MemOpIdx oi; - TCGv copy_addr; + TCGv_i64 copy_addr; tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); memop = tcg_canonicalize_memop(memop, 0, 0); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) { MemOp orig_memop; MemOpIdx oi; - TCGv copy_addr; + TCGv_i64 copy_addr; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { tcg_gen_qemu_ld_i32(TCGV_LOW(val), addr, idx, memop); -- 2.34.1 This will enable replacement of TARGET_LONG_BITS within tcg/. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 1 + accel/tcg/translate-all.c | 2 ++ tcg/tcg.c | 3 +++ 3 files changed, 6 insertions(+) diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -XXX,XX +XXX,XX @@ struct TCGContext { int nb_temps; int nb_indirects; int nb_ops; + TCGType addr_type; /* TCG_TYPE_I32 or TCG_TYPE_I64 */ TCGRegSet reserved_regs; intptr_t current_frame_offset; diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/translate-all.c +++ b/accel/tcg/translate-all.c @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, tb_set_page_addr0(tb, phys_pc); tb_set_page_addr1(tb, -1); tcg_ctx->gen_tb = tb; + tcg_ctx->addr_type = TCG_TYPE_TL; + tb_overflow: #ifdef CONFIG_PROFILER diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ void tcg_func_start(TCGContext *s) QTAILQ_INIT(&s->ops); QTAILQ_INIT(&s->free_ops); QSIMPLEQ_INIT(&s->labels); + + tcg_debug_assert(s->addr_type == TCG_TYPE_I32 || + s->addr_type == TCG_TYPE_I64); } static TCGTemp *tcg_temp_alloc(TCGContext *s) -- 2.34.1 Expand from TCGv to TCGTemp inline in the translators, and validate that the size matches tcg_ctx->addr_type. These inlines will eventually be seen only by target-specific code. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-op.h | 50 ++++++- tcg/tcg-op-ldst.c | 343 ++++++++++++++++++++++++++----------------- 2 files changed, 251 insertions(+), 142 deletions(-) diff --git a/include/tcg/tcg-op.h b/include/tcg/tcg-op.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-op.h +++ b/include/tcg/tcg-op.h @@ -XXX,XX +XXX,XX @@ static inline void tcg_gen_plugin_cb_end(void) #define tcg_temp_new() tcg_temp_new_i32() #define tcg_global_mem_new tcg_global_mem_new_i32 #define tcg_temp_free tcg_temp_free_i32 +#define tcgv_tl_temp tcgv_i32_temp #define tcg_gen_qemu_ld_tl tcg_gen_qemu_ld_i32 #define tcg_gen_qemu_st_tl tcg_gen_qemu_st_i32 #else #define tcg_temp_new() tcg_temp_new_i64() #define tcg_global_mem_new tcg_global_mem_new_i64 #define tcg_temp_free tcg_temp_free_i64 +#define tcgv_tl_temp tcgv_i64_temp #define tcg_gen_qemu_ld_tl tcg_gen_qemu_ld_i64 #define tcg_gen_qemu_st_tl tcg_gen_qemu_st_i64 #endif -void tcg_gen_qemu_ld_i32(TCGv_i32, TCGv, TCGArg, MemOp); -void tcg_gen_qemu_st_i32(TCGv_i32, TCGv, TCGArg, MemOp); -void tcg_gen_qemu_ld_i64(TCGv_i64, TCGv, TCGArg, MemOp); -void tcg_gen_qemu_st_i64(TCGv_i64, TCGv, TCGArg, MemOp); -void tcg_gen_qemu_ld_i128(TCGv_i128, TCGv, TCGArg, MemOp); -void tcg_gen_qemu_st_i128(TCGv_i128, TCGv, TCGArg, MemOp); +void tcg_gen_qemu_ld_i32_chk(TCGv_i32, TCGTemp *, TCGArg, MemOp, TCGType); +void tcg_gen_qemu_st_i32_chk(TCGv_i32, TCGTemp *, TCGArg, MemOp, TCGType); +void tcg_gen_qemu_ld_i64_chk(TCGv_i64, TCGTemp *, TCGArg, MemOp, TCGType); +void tcg_gen_qemu_st_i64_chk(TCGv_i64, TCGTemp *, TCGArg, MemOp, TCGType); +void tcg_gen_qemu_ld_i128_chk(TCGv_i128, TCGTemp *, TCGArg, MemOp, TCGType); +void tcg_gen_qemu_st_i128_chk(TCGv_i128, TCGTemp *, TCGArg, MemOp, TCGType); + +static inline void +tcg_gen_qemu_ld_i32(TCGv_i32 v, TCGv a, TCGArg i, MemOp m) +{ + tcg_gen_qemu_ld_i32_chk(v, tcgv_tl_temp(a), i, m, TCG_TYPE_TL); +} + +static inline void +tcg_gen_qemu_st_i32(TCGv_i32 v, TCGv a, TCGArg i, MemOp m) +{ + tcg_gen_qemu_st_i32_chk(v, tcgv_tl_temp(a), i, m, TCG_TYPE_TL); +} + +static inline void +tcg_gen_qemu_ld_i64(TCGv_i64 v, TCGv a, TCGArg i, MemOp m) +{ + tcg_gen_qemu_ld_i64_chk(v, tcgv_tl_temp(a), i, m, TCG_TYPE_TL); +} + +static inline void +tcg_gen_qemu_st_i64(TCGv_i64 v, TCGv a, TCGArg i, MemOp m) +{ + tcg_gen_qemu_st_i64_chk(v, tcgv_tl_temp(a), i, m, TCG_TYPE_TL); +} + +static inline void +tcg_gen_qemu_ld_i128(TCGv_i128 v, TCGv a, TCGArg i, MemOp m) +{ + tcg_gen_qemu_ld_i128_chk(v, tcgv_tl_temp(a), i, m, TCG_TYPE_TL); +} + +static inline void +tcg_gen_qemu_st_i128(TCGv_i128 v, TCGv a, TCGArg i, MemOp m) +{ + tcg_gen_qemu_st_i128_chk(v, tcgv_tl_temp(a), i, m, TCG_TYPE_TL); +} void tcg_gen_atomic_cmpxchg_i32(TCGv_i32, TCGv, TCGv_i32, TCGv_i32, TCGArg, MemOp); diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ static inline MemOp tcg_canonicalize_memop(MemOp op, bool is64, bool st) return op; } -static void gen_ldst_i32(TCGOpcode opc, TCGv_i32 val, TCGv addr, - MemOp memop, TCGArg idx) +static void gen_ldst(TCGOpcode opc, TCGTemp *vl, TCGTemp *vh, + TCGTemp *addr, MemOpIdx oi) { - MemOpIdx oi = make_memop_idx(memop, idx); -#if TARGET_LONG_BITS == 32 - tcg_gen_op3i_i32(opc, val, addr, oi); -#else - if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_op4i_i32(opc, val, TCGV_LOW(addr), TCGV_HIGH(addr), oi); + if (TCG_TARGET_REG_BITS == 64 || tcg_ctx->addr_type == TCG_TYPE_I32) { + if (vh) { + tcg_gen_op4(opc, temp_arg(vl), temp_arg(vh), temp_arg(addr), oi); + } else { + tcg_gen_op3(opc, temp_arg(vl), temp_arg(addr), oi); + } } else { - tcg_gen_op3(opc, tcgv_i32_arg(val), tcgv_i64_arg(addr), oi); + /* See TCGV_LOW/HIGH. */ + TCGTemp *al = addr + HOST_BIG_ENDIAN; + TCGTemp *ah = addr + !HOST_BIG_ENDIAN; + + if (vh) { + tcg_gen_op5(opc, temp_arg(vl), temp_arg(vh), + temp_arg(al), temp_arg(ah), oi); + } else { + tcg_gen_op4(opc, temp_arg(vl), temp_arg(al), temp_arg(ah), oi); + } } -#endif } -static void gen_ldst_i64(TCGOpcode opc, TCGv_i64 val, TCGv addr, - MemOp memop, TCGArg idx) +static void gen_ldst_i64(TCGOpcode opc, TCGv_i64 v, TCGTemp *addr, MemOpIdx oi) { - MemOpIdx oi = make_memop_idx(memop, idx); -#if TARGET_LONG_BITS == 32 if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_op4i_i32(opc, TCGV_LOW(val), TCGV_HIGH(val), addr, oi); + TCGTemp *vl = tcgv_i32_temp(TCGV_LOW(v)); + TCGTemp *vh = tcgv_i32_temp(TCGV_HIGH(v)); + gen_ldst(opc, vl, vh, addr, oi); } else { - tcg_gen_op3(opc, tcgv_i64_arg(val), tcgv_i32_arg(addr), oi); + gen_ldst(opc, tcgv_i64_temp(v), NULL, addr, oi); } -#else - if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_op5i_i32(opc, TCGV_LOW(val), TCGV_HIGH(val), - TCGV_LOW(addr), TCGV_HIGH(addr), oi); - } else { - tcg_gen_op3i_i64(opc, val, addr, oi); - } -#endif } static void tcg_gen_req_mo(TCGBar type) @@ -XXX,XX +XXX,XX @@ static void tcg_gen_req_mo(TCGBar type) } /* Only required for loads, where value might overlap addr. */ -static TCGv_i64 plugin_maybe_preserve_addr(TCGv vaddr) +static TCGv_i64 plugin_maybe_preserve_addr(TCGTemp *addr) { #ifdef CONFIG_PLUGIN if (tcg_ctx->plugin_insn != NULL) { /* Save a copy of the vaddr for use after a load. */ TCGv_i64 temp = tcg_temp_ebb_new_i64(); - tcg_gen_extu_tl_i64(temp, vaddr); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + tcg_gen_extu_i32_i64(temp, temp_tcgv_i32(addr)); + } else { + tcg_gen_mov_i64(temp, temp_tcgv_i64(addr)); + } return temp; } #endif @@ -XXX,XX +XXX,XX @@ static TCGv_i64 plugin_maybe_preserve_addr(TCGv vaddr) } static void -plugin_gen_mem_callbacks(TCGv_i64 copy_addr, TCGv orig_addr, MemOpIdx oi, +plugin_gen_mem_callbacks(TCGv_i64 copy_addr, TCGTemp *orig_addr, MemOpIdx oi, enum qemu_plugin_mem_rw rw) { #ifdef CONFIG_PLUGIN if (tcg_ctx->plugin_insn != NULL) { qemu_plugin_meminfo_t info = make_plugin_meminfo(oi, rw); -#if TARGET_LONG_BITS == 64 - if (copy_addr) { + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + if (!copy_addr) { + copy_addr = tcg_temp_ebb_new_i64(); + tcg_gen_extu_i32_i64(copy_addr, temp_tcgv_i32(orig_addr)); + } plugin_gen_empty_mem_callback(copy_addr, info); tcg_temp_free_i64(copy_addr); } else { - plugin_gen_empty_mem_callback(orig_addr, info); + if (copy_addr) { + plugin_gen_empty_mem_callback(copy_addr, info); + tcg_temp_free_i64(copy_addr); + } else { + plugin_gen_empty_mem_callback(temp_tcgv_i64(orig_addr), info); + } } -#else - if (!copy_addr) { - copy_addr = tcg_temp_ebb_new_i64(); - tcg_gen_extu_tl_i64(copy_addr, orig_addr); - } - plugin_gen_empty_mem_callback(copy_addr, info); - tcg_temp_free_i64(copy_addr); -#endif } #endif } -void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) +static void tcg_gen_qemu_ld_i32_int(TCGv_i32 val, TCGTemp *addr, + TCGArg idx, MemOp memop) { MemOp orig_memop; - MemOpIdx oi; + MemOpIdx orig_oi, oi; TCGv_i64 copy_addr; tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); - memop = tcg_canonicalize_memop(memop, 0, 0); - oi = make_memop_idx(memop, idx); + orig_memop = memop = tcg_canonicalize_memop(memop, 0, 0); + orig_oi = oi = make_memop_idx(memop, idx); - orig_memop = memop; if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { memop &= ~MO_BSWAP; /* The bswap primitive benefits from zero-extended input. */ if ((memop & MO_SSIZE) == MO_SW) { memop &= ~MO_SIGN; } + oi = make_memop_idx(memop, idx); } copy_addr = plugin_maybe_preserve_addr(addr); - gen_ldst_i32(INDEX_op_qemu_ld_i32, val, addr, memop, idx); - plugin_gen_mem_callbacks(copy_addr, addr, oi, QEMU_PLUGIN_MEM_R); + gen_ldst(INDEX_op_qemu_ld_i32, tcgv_i32_temp(val), NULL, addr, oi); + plugin_gen_mem_callbacks(copy_addr, addr, orig_oi, QEMU_PLUGIN_MEM_R); if ((orig_memop ^ memop) & MO_BSWAP) { switch (orig_memop & MO_SIZE) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) } } -void tcg_gen_qemu_st_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) +void tcg_gen_qemu_ld_i32_chk(TCGv_i32 val, TCGTemp *addr, TCGArg idx, + MemOp memop, TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_32); + tcg_gen_qemu_ld_i32_int(val, addr, idx, memop); +} + +static void tcg_gen_qemu_st_i32_int(TCGv_i32 val, TCGTemp *addr, + TCGArg idx, MemOp memop) { TCGv_i32 swap = NULL; - MemOpIdx oi; + MemOpIdx orig_oi, oi; + TCGOpcode opc; tcg_gen_req_mo(TCG_MO_LD_ST | TCG_MO_ST_ST); memop = tcg_canonicalize_memop(memop, 0, 1); - oi = make_memop_idx(memop, idx); + orig_oi = oi = make_memop_idx(memop, idx); if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { swap = tcg_temp_ebb_new_i32(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i32(TCGv_i32 val, TCGv addr, TCGArg idx, MemOp memop) } val = swap; memop &= ~MO_BSWAP; + oi = make_memop_idx(memop, idx); } if (TCG_TARGET_HAS_qemu_st8_i32 && (memop & MO_SIZE) == MO_8) { - gen_ldst_i32(INDEX_op_qemu_st8_i32, val, addr, memop, idx); + opc = INDEX_op_qemu_st8_i32; } else { - gen_ldst_i32(INDEX_op_qemu_st_i32, val, addr, memop, idx); + opc = INDEX_op_qemu_st_i32; } - plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_W); + gen_ldst(opc, tcgv_i32_temp(val), NULL, addr, oi); + plugin_gen_mem_callbacks(NULL, addr, orig_oi, QEMU_PLUGIN_MEM_W); if (swap) { tcg_temp_free_i32(swap); } } -void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) +void tcg_gen_qemu_st_i32_chk(TCGv_i32 val, TCGTemp *addr, TCGArg idx, + MemOp memop, TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_32); + tcg_gen_qemu_st_i32_int(val, addr, idx, memop); +} + +static void tcg_gen_qemu_ld_i64_int(TCGv_i64 val, TCGTemp *addr, + TCGArg idx, MemOp memop) { MemOp orig_memop; - MemOpIdx oi; + MemOpIdx orig_oi, oi; TCGv_i64 copy_addr; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { - tcg_gen_qemu_ld_i32(TCGV_LOW(val), addr, idx, memop); + tcg_gen_qemu_ld_i32_int(TCGV_LOW(val), addr, idx, memop); if (memop & MO_SIGN) { tcg_gen_sari_i32(TCGV_HIGH(val), TCGV_LOW(val), 31); } else { @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) } tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); - memop = tcg_canonicalize_memop(memop, 1, 0); - oi = make_memop_idx(memop, idx); + orig_memop = memop = tcg_canonicalize_memop(memop, 1, 0); + orig_oi = oi = make_memop_idx(memop, idx); - orig_memop = memop; if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { memop &= ~MO_BSWAP; /* The bswap primitive benefits from zero-extended input. */ if ((memop & MO_SIGN) && (memop & MO_SIZE) < MO_64) { memop &= ~MO_SIGN; } + oi = make_memop_idx(memop, idx); } copy_addr = plugin_maybe_preserve_addr(addr); - gen_ldst_i64(INDEX_op_qemu_ld_i64, val, addr, memop, idx); - plugin_gen_mem_callbacks(copy_addr, addr, oi, QEMU_PLUGIN_MEM_R); + gen_ldst_i64(INDEX_op_qemu_ld_i64, val, addr, oi); + plugin_gen_mem_callbacks(copy_addr, addr, orig_oi, QEMU_PLUGIN_MEM_R); if ((orig_memop ^ memop) & MO_BSWAP) { int flags = (orig_memop & MO_SIGN @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) } } -void tcg_gen_qemu_st_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) +void tcg_gen_qemu_ld_i64_chk(TCGv_i64 val, TCGTemp *addr, TCGArg idx, + MemOp memop, TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_64); + tcg_gen_qemu_ld_i64_int(val, addr, idx, memop); +} + +static void tcg_gen_qemu_st_i64_int(TCGv_i64 val, TCGTemp *addr, + TCGArg idx, MemOp memop) { TCGv_i64 swap = NULL; - MemOpIdx oi; + MemOpIdx orig_oi, oi; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { - tcg_gen_qemu_st_i32(TCGV_LOW(val), addr, idx, memop); + tcg_gen_qemu_st_i32_int(TCGV_LOW(val), addr, idx, memop); return; } tcg_gen_req_mo(TCG_MO_LD_ST | TCG_MO_ST_ST); memop = tcg_canonicalize_memop(memop, 1, 1); - oi = make_memop_idx(memop, idx); + orig_oi = oi = make_memop_idx(memop, idx); if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { swap = tcg_temp_ebb_new_i64(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i64(TCGv_i64 val, TCGv addr, TCGArg idx, MemOp memop) } val = swap; memop &= ~MO_BSWAP; + oi = make_memop_idx(memop, idx); } - gen_ldst_i64(INDEX_op_qemu_st_i64, val, addr, memop, idx); - plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_W); + gen_ldst_i64(INDEX_op_qemu_st_i64, val, addr, oi); + plugin_gen_mem_callbacks(NULL, addr, orig_oi, QEMU_PLUGIN_MEM_W); if (swap) { tcg_temp_free_i64(swap); } } +void tcg_gen_qemu_st_i64_chk(TCGv_i64 val, TCGTemp *addr, TCGArg idx, + MemOp memop, TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_64); + tcg_gen_qemu_st_i64_int(val, addr, idx, memop); +} + /* * Return true if @mop, without knowledge of the pointer alignment, * does not require 16-byte atomicity, and it would be adventagous @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) { MemOp mop_1 = orig, mop_2; - tcg_debug_assert((orig & MO_SIZE) == MO_128); - tcg_debug_assert((orig & MO_SIGN) == 0); - /* Reduce the size to 64-bit. */ mop_1 = (mop_1 & ~MO_SIZE) | MO_64; @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) ret[1] = mop_2; } -#if TARGET_LONG_BITS == 64 -#define tcg_temp_ebb_new tcg_temp_ebb_new_i64 -#else -#define tcg_temp_ebb_new tcg_temp_ebb_new_i32 -#endif - static TCGv_i64 maybe_extend_addr64(TCGv addr) { #if TARGET_LONG_BITS == 32 @@ -XXX,XX +XXX,XX @@ static void maybe_free_addr64(TCGv_i64 a64) #endif } -void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) +static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, + TCGArg idx, MemOp memop) { - const MemOpIdx oi = make_memop_idx(memop, idx); - - tcg_debug_assert((memop & MO_SIZE) == MO_128); - tcg_debug_assert((memop & MO_SIGN) == 0); + const MemOpIdx orig_oi = make_memop_idx(memop, idx); + TCGv_i64 ext_addr = NULL; tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); /* TODO: For now, force 32-bit hosts to use the helper. */ if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { TCGv_i64 lo, hi; - TCGArg addr_arg; - MemOpIdx adj_oi; bool need_bswap = false; + MemOpIdx oi = orig_oi; if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { lo = TCGV128_HIGH(val); hi = TCGV128_LOW(val); - adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); + oi = make_memop_idx(memop & ~MO_BSWAP, idx); need_bswap = true; } else { lo = TCGV128_LOW(val); hi = TCGV128_HIGH(val); - adj_oi = oi; } -#if TARGET_LONG_BITS == 32 - addr_arg = tcgv_i32_arg(addr); -#else - addr_arg = tcgv_i64_arg(addr); -#endif - tcg_gen_op4ii_i64(INDEX_op_qemu_ld_i128, lo, hi, addr_arg, adj_oi); + gen_ldst(INDEX_op_qemu_ld_i128, tcgv_i64_temp(lo), + tcgv_i64_temp(hi), addr, oi); if (need_bswap) { tcg_gen_bswap64_i64(lo, lo); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) } } else if (use_two_i64_for_i128(memop)) { MemOp mop[2]; - TCGv addr_p8; + TCGTemp *addr_p8; TCGv_i64 x, y; + MemOpIdx oi; + bool need_bswap; canonicalize_memop_i128_as_i64(mop, memop); + need_bswap = (mop[0] ^ memop) & MO_BSWAP; /* * Since there are no global TCGv_i128, there is no visible state @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_ld_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) y = TCGV128_LOW(val); } - gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, mop[0], idx); + oi = make_memop_idx(mop[0], idx); + gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, oi); - if ((mop[0] ^ memop) & MO_BSWAP) { + if (need_bswap) { tcg_gen_bswap64_i64(x, x); } - addr_p8 = tcg_temp_ebb_new(); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, mop[1], idx); - tcg_temp_free(addr_p8); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + TCGv_i32 t = tcg_temp_ebb_new_i32(); + tcg_gen_addi_i32(t, temp_tcgv_i32(addr), 8); + addr_p8 = tcgv_i32_temp(t); + } else { + TCGv_i64 t = tcg_temp_ebb_new_i64(); + tcg_gen_addi_i64(t, temp_tcgv_i64(addr), 8); + addr_p8 = tcgv_i64_temp(t); + } - if ((mop[0] ^ memop) & MO_BSWAP) { + gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, oi); + tcg_temp_free_internal(addr_p8); + + if (need_bswap) { tcg_gen_bswap64_i64(y, y); } } else { - TCGv_i64 a64 = maybe_extend_addr64(addr); - gen_helper_ld_i128(val, cpu_env, a64, tcg_constant_i32(oi)); - maybe_free_addr64(a64); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + ext_addr = tcg_temp_ebb_new_i64(); + tcg_gen_extu_i32_i64(ext_addr, temp_tcgv_i32(addr)); + addr = tcgv_i64_temp(ext_addr); + } + gen_helper_ld_i128(val, cpu_env, temp_tcgv_i64(addr), + tcg_constant_i32(orig_oi)); } - plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_R); + plugin_gen_mem_callbacks(ext_addr, addr, orig_oi, QEMU_PLUGIN_MEM_R); } -void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) +void tcg_gen_qemu_ld_i128_chk(TCGv_i128 val, TCGTemp *addr, TCGArg idx, + MemOp memop, TCGType addr_type) { - const MemOpIdx oi = make_memop_idx(memop, idx); - + tcg_debug_assert(addr_type == tcg_ctx->addr_type); tcg_debug_assert((memop & MO_SIZE) == MO_128); tcg_debug_assert((memop & MO_SIGN) == 0); + tcg_gen_qemu_ld_i128_int(val, addr, idx, memop); +} + +static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, + TCGArg idx, MemOp memop) +{ + const MemOpIdx orig_oi = make_memop_idx(memop, idx); + TCGv_i64 ext_addr = NULL; tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) if (TCG_TARGET_HAS_qemu_ldst_i128 && TCG_TARGET_REG_BITS == 64) { TCGv_i64 lo, hi; - TCGArg addr_arg; - MemOpIdx adj_oi; + MemOpIdx oi = orig_oi; bool need_bswap = false; if ((memop & MO_BSWAP) && !tcg_target_has_memory_bswap(memop)) { - lo = tcg_temp_new_i64(); - hi = tcg_temp_new_i64(); + lo = tcg_temp_ebb_new_i64(); + hi = tcg_temp_ebb_new_i64(); tcg_gen_bswap64_i64(lo, TCGV128_HIGH(val)); tcg_gen_bswap64_i64(hi, TCGV128_LOW(val)); - adj_oi = make_memop_idx(memop & ~MO_BSWAP, idx); + oi = make_memop_idx(memop & ~MO_BSWAP, idx); need_bswap = true; } else { lo = TCGV128_LOW(val); hi = TCGV128_HIGH(val); - adj_oi = oi; } -#if TARGET_LONG_BITS == 32 - addr_arg = tcgv_i32_arg(addr); -#else - addr_arg = tcgv_i64_arg(addr); -#endif - tcg_gen_op4ii_i64(INDEX_op_qemu_st_i128, lo, hi, addr_arg, adj_oi); + gen_ldst(INDEX_op_qemu_st_i128, tcgv_i64_temp(lo), + tcgv_i64_temp(hi), addr, oi); if (need_bswap) { tcg_temp_free_i64(lo); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) } } else if (use_two_i64_for_i128(memop)) { MemOp mop[2]; - TCGv addr_p8; - TCGv_i64 x, y; + TCGTemp *addr_p8; + TCGv_i64 x, y, b = NULL; canonicalize_memop_i128_as_i64(mop, memop); @@ -XXX,XX +XXX,XX @@ void tcg_gen_qemu_st_i128(TCGv_i128 val, TCGv addr, TCGArg idx, MemOp memop) y = TCGV128_LOW(val); } - addr_p8 = tcg_temp_ebb_new(); if ((mop[0] ^ memop) & MO_BSWAP) { - TCGv_i64 t = tcg_temp_ebb_new_i64(); - - tcg_gen_bswap64_i64(t, x); - gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr, mop[0], idx); - tcg_gen_bswap64_i64(t, y); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_st_i64, t, addr_p8, mop[1], idx); - tcg_temp_free_i64(t); - } else { - gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, mop[0], idx); - tcg_gen_addi_tl(addr_p8, addr, 8); - gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, mop[1], idx); + b = tcg_temp_ebb_new_i64(); + tcg_gen_bswap64_i64(b, x); + x = b; } - tcg_temp_free(addr_p8); + gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, + make_memop_idx(mop[0], idx)); + + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + TCGv_i32 t = tcg_temp_ebb_new_i32(); + tcg_gen_addi_i32(t, temp_tcgv_i32(addr), 8); + addr_p8 = tcgv_i32_temp(t); + } else { + TCGv_i64 t = tcg_temp_ebb_new_i64(); + tcg_gen_addi_i64(t, temp_tcgv_i64(addr), 8); + addr_p8 = tcgv_i64_temp(t); + } + + if (b) { + tcg_gen_bswap64_i64(b, y); + y = b; + } + gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, + make_memop_idx(mop[1], idx)); + + if (b) { + tcg_temp_free_i64(b); + } + tcg_temp_free_internal(addr_p8); } else { - TCGv_i64 a64 = maybe_extend_addr64(addr); - gen_helper_st_i128(cpu_env, a64, val, tcg_constant_i32(oi)); - maybe_free_addr64(a64); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + ext_addr = tcg_temp_ebb_new_i64(); + tcg_gen_extu_i32_i64(ext_addr, temp_tcgv_i32(addr)); + addr = tcgv_i64_temp(ext_addr); + } + gen_helper_st_i128(cpu_env, temp_tcgv_i64(addr), val, + tcg_constant_i32(orig_oi)); } - plugin_gen_mem_callbacks(NULL, addr, oi, QEMU_PLUGIN_MEM_W); + plugin_gen_mem_callbacks(ext_addr, addr, orig_oi, QEMU_PLUGIN_MEM_W); +} + +void tcg_gen_qemu_st_i128_chk(TCGv_i128 val, TCGTemp *addr, TCGArg idx, + MemOp memop, TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) == MO_128); + tcg_debug_assert((memop & MO_SIGN) == 0); + tcg_gen_qemu_st_i128_int(val, addr, idx, memop); } static void tcg_gen_ext_i32(TCGv_i32 ret, TCGv_i32 val, MemOp opc) -- 2.34.1 Expand from TCGv to TCGTemp inline in the translators, and validate that the size matches tcg_ctx->addr_type. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-op.h | 184 ++++++++++++++++++++++++++++++---------- tcg/tcg-op-ldst.c | 198 ++++++++++++++++++++++++++++--------------- 2 files changed, 267 insertions(+), 115 deletions(-) diff --git a/include/tcg/tcg-op.h b/include/tcg/tcg-op.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-op.h +++ b/include/tcg/tcg-op.h @@ -XXX,XX +XXX,XX @@ tcg_gen_qemu_st_i128(TCGv_i128 v, TCGv a, TCGArg i, MemOp m) tcg_gen_qemu_st_i128_chk(v, tcgv_tl_temp(a), i, m, TCG_TYPE_TL); } -void tcg_gen_atomic_cmpxchg_i32(TCGv_i32, TCGv, TCGv_i32, TCGv_i32, - TCGArg, MemOp); -void tcg_gen_atomic_cmpxchg_i64(TCGv_i64, TCGv, TCGv_i64, TCGv_i64, - TCGArg, MemOp); -void tcg_gen_atomic_cmpxchg_i128(TCGv_i128, TCGv, TCGv_i128, TCGv_i128, - TCGArg, MemOp); +void tcg_gen_atomic_cmpxchg_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_cmpxchg_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_cmpxchg_i128_chk(TCGv_i128, TCGTemp *, TCGv_i128, + TCGv_i128, TCGArg, MemOp, TCGType); -void tcg_gen_nonatomic_cmpxchg_i32(TCGv_i32, TCGv, TCGv_i32, TCGv_i32, - TCGArg, MemOp); -void tcg_gen_nonatomic_cmpxchg_i64(TCGv_i64, TCGv, TCGv_i64, TCGv_i64, - TCGArg, MemOp); -void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128, TCGv, TCGv_i128, TCGv_i128, - TCGArg, MemOp); +void tcg_gen_nonatomic_cmpxchg_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_nonatomic_cmpxchg_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_nonatomic_cmpxchg_i128_chk(TCGv_i128, TCGTemp *, TCGv_i128, + TCGv_i128, TCGArg, MemOp, TCGType); -void tcg_gen_atomic_xchg_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_xchg_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); +void tcg_gen_atomic_xchg_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_xchg_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); -void tcg_gen_atomic_fetch_add_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_add_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_fetch_and_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_and_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_fetch_or_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_or_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_fetch_xor_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_xor_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_fetch_smin_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_smin_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_fetch_umin_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_umin_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_fetch_smax_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_smax_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_fetch_umax_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_fetch_umax_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); +void tcg_gen_atomic_fetch_add_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_add_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_and_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_and_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_or_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_or_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_xor_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_xor_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_smin_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_smin_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_umin_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_umin_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_smax_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_smax_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_umax_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_fetch_umax_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); -void tcg_gen_atomic_add_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_add_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_and_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_and_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_or_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_or_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_xor_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_xor_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_smin_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_smin_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_umin_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_umin_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_smax_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_smax_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); -void tcg_gen_atomic_umax_fetch_i32(TCGv_i32, TCGv, TCGv_i32, TCGArg, MemOp); -void tcg_gen_atomic_umax_fetch_i64(TCGv_i64, TCGv, TCGv_i64, TCGArg, MemOp); +void tcg_gen_atomic_add_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_add_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_and_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_and_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_or_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_or_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_xor_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_xor_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_smin_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_smin_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_umin_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_umin_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_smax_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_smax_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_umax_fetch_i32_chk(TCGv_i32, TCGTemp *, TCGv_i32, + TCGArg, MemOp, TCGType); +void tcg_gen_atomic_umax_fetch_i64_chk(TCGv_i64, TCGTemp *, TCGv_i64, + TCGArg, MemOp, TCGType); + +#define DEF_ATOMIC2(N, S) \ + static inline void N##_##S(TCGv_##S r, TCGv a, TCGv_##S v, \ + TCGArg i, MemOp m) \ + { N##_##S##_chk(r, tcgv_tl_temp(a), v, i, m, TCG_TYPE_TL); } + +#define DEF_ATOMIC3(N, S) \ + static inline void N##_##S(TCGv_##S r, TCGv a, TCGv_##S o, \ + TCGv_##S n, TCGArg i, MemOp m) \ + { N##_##S##_chk(r, tcgv_tl_temp(a), o, n, i, m, TCG_TYPE_TL); } + +DEF_ATOMIC3(tcg_gen_atomic_cmpxchg, i32) +DEF_ATOMIC3(tcg_gen_atomic_cmpxchg, i64) +DEF_ATOMIC3(tcg_gen_atomic_cmpxchg, i128) + +DEF_ATOMIC3(tcg_gen_nonatomic_cmpxchg, i32) +DEF_ATOMIC3(tcg_gen_nonatomic_cmpxchg, i64) +DEF_ATOMIC3(tcg_gen_nonatomic_cmpxchg, i128) + +DEF_ATOMIC2(tcg_gen_atomic_xchg, i32) +DEF_ATOMIC2(tcg_gen_atomic_xchg, i64) + +DEF_ATOMIC2(tcg_gen_atomic_fetch_add, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_add, i64) +DEF_ATOMIC2(tcg_gen_atomic_fetch_and, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_and, i64) +DEF_ATOMIC2(tcg_gen_atomic_fetch_or, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_or, i64) +DEF_ATOMIC2(tcg_gen_atomic_fetch_xor, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_xor, i64) +DEF_ATOMIC2(tcg_gen_atomic_fetch_smin, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_smin, i64) +DEF_ATOMIC2(tcg_gen_atomic_fetch_umin, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_umin, i64) +DEF_ATOMIC2(tcg_gen_atomic_fetch_smax, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_smax, i64) +DEF_ATOMIC2(tcg_gen_atomic_fetch_umax, i32) +DEF_ATOMIC2(tcg_gen_atomic_fetch_umax, i64) + +DEF_ATOMIC2(tcg_gen_atomic_add_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_add_fetch, i64) +DEF_ATOMIC2(tcg_gen_atomic_and_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_and_fetch, i64) +DEF_ATOMIC2(tcg_gen_atomic_or_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_or_fetch, i64) +DEF_ATOMIC2(tcg_gen_atomic_xor_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_xor_fetch, i64) +DEF_ATOMIC2(tcg_gen_atomic_smin_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_smin_fetch, i64) +DEF_ATOMIC2(tcg_gen_atomic_umin_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_umin_fetch, i64) +DEF_ATOMIC2(tcg_gen_atomic_smax_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_smax_fetch, i64) +DEF_ATOMIC2(tcg_gen_atomic_umax_fetch, i32) +DEF_ATOMIC2(tcg_gen_atomic_umax_fetch, i64) + +#undef DEF_ATOMIC2 +#undef DEF_ATOMIC3 void tcg_gen_mov_vec(TCGv_vec, TCGv_vec); void tcg_gen_dup_i32_vec(unsigned vece, TCGv_vec, TCGv_i32); diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ static void canonicalize_memop_i128_as_i64(MemOp ret[2], MemOp orig) ret[1] = mop_2; } -static TCGv_i64 maybe_extend_addr64(TCGv addr) +static TCGv_i64 maybe_extend_addr64(TCGTemp *addr) { -#if TARGET_LONG_BITS == 32 - TCGv_i64 a64 = tcg_temp_ebb_new_i64(); - tcg_gen_extu_i32_i64(a64, addr); - return a64; -#else - return addr; -#endif + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + TCGv_i64 a64 = tcg_temp_ebb_new_i64(); + tcg_gen_extu_i32_i64(a64, temp_tcgv_i32(addr)); + return a64; + } + return temp_tcgv_i64(addr); } static void maybe_free_addr64(TCGv_i64 a64) { -#if TARGET_LONG_BITS == 32 - tcg_temp_free_i64(a64); -#endif + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + tcg_temp_free_i64(a64); + } } static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, @@ -XXX,XX +XXX,XX @@ static void * const table_cmpxchg[(MO_SIZE | MO_BSWAP) + 1] = { WITH_ATOMIC128([MO_128 | MO_BE] = gen_helper_atomic_cmpxchgo_be) }; -void tcg_gen_nonatomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, - TCGv_i32 newv, TCGArg idx, MemOp memop) +static void tcg_gen_nonatomic_cmpxchg_i32_int(TCGv_i32 retv, TCGTemp *addr, + TCGv_i32 cmpv, TCGv_i32 newv, + TCGArg idx, MemOp memop) { TCGv_i32 t1 = tcg_temp_ebb_new_i32(); TCGv_i32 t2 = tcg_temp_ebb_new_i32(); tcg_gen_ext_i32(t2, cmpv, memop & MO_SIZE); - tcg_gen_qemu_ld_i32(t1, addr, idx, memop & ~MO_SIGN); + tcg_gen_qemu_ld_i32_int(t1, addr, idx, memop & ~MO_SIGN); tcg_gen_movcond_i32(TCG_COND_EQ, t2, t1, t2, newv, t1); - tcg_gen_qemu_st_i32(t2, addr, idx, memop); + tcg_gen_qemu_st_i32_int(t2, addr, idx, memop); tcg_temp_free_i32(t2); if (memop & MO_SIGN) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, tcg_temp_free_i32(t1); } -void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, - TCGv_i32 newv, TCGArg idx, MemOp memop) +void tcg_gen_nonatomic_cmpxchg_i32_chk(TCGv_i32 retv, TCGTemp *addr, + TCGv_i32 cmpv, TCGv_i32 newv, + TCGArg idx, MemOp memop, + TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_32); + tcg_gen_nonatomic_cmpxchg_i32_int(retv, addr, cmpv, newv, idx, memop); +} + +static void tcg_gen_atomic_cmpxchg_i32_int(TCGv_i32 retv, TCGTemp *addr, + TCGv_i32 cmpv, TCGv_i32 newv, + TCGArg idx, MemOp memop) { gen_atomic_cx_i32 gen; TCGv_i64 a64; MemOpIdx oi; if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { - tcg_gen_nonatomic_cmpxchg_i32(retv, addr, cmpv, newv, idx, memop); + tcg_gen_nonatomic_cmpxchg_i32_int(retv, addr, cmpv, newv, idx, memop); return; } @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, } } -void tcg_gen_nonatomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, - TCGv_i64 newv, TCGArg idx, MemOp memop) +void tcg_gen_atomic_cmpxchg_i32_chk(TCGv_i32 retv, TCGTemp *addr, + TCGv_i32 cmpv, TCGv_i32 newv, + TCGArg idx, MemOp memop, + TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_32); + tcg_gen_atomic_cmpxchg_i32_int(retv, addr, cmpv, newv, idx, memop); +} + +static void tcg_gen_nonatomic_cmpxchg_i64_int(TCGv_i64 retv, TCGTemp *addr, + TCGv_i64 cmpv, TCGv_i64 newv, + TCGArg idx, MemOp memop) { TCGv_i64 t1, t2; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { - tcg_gen_nonatomic_cmpxchg_i32(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), - TCGV_LOW(newv), idx, memop); + tcg_gen_nonatomic_cmpxchg_i32_int(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), + TCGV_LOW(newv), idx, memop); if (memop & MO_SIGN) { tcg_gen_sari_i32(TCGV_HIGH(retv), TCGV_LOW(retv), 31); } else { @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, tcg_gen_ext_i64(t2, cmpv, memop & MO_SIZE); - tcg_gen_qemu_ld_i64(t1, addr, idx, memop & ~MO_SIGN); + tcg_gen_qemu_ld_i64_int(t1, addr, idx, memop & ~MO_SIGN); tcg_gen_movcond_i64(TCG_COND_EQ, t2, t1, t2, newv, t1); - tcg_gen_qemu_st_i64(t2, addr, idx, memop); + tcg_gen_qemu_st_i64_int(t2, addr, idx, memop); tcg_temp_free_i64(t2); if (memop & MO_SIGN) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, tcg_temp_free_i64(t1); } -void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, - TCGv_i64 newv, TCGArg idx, MemOp memop) +void tcg_gen_nonatomic_cmpxchg_i64_chk(TCGv_i64 retv, TCGTemp *addr, + TCGv_i64 cmpv, TCGv_i64 newv, + TCGArg idx, MemOp memop, + TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_64); + tcg_gen_nonatomic_cmpxchg_i64_int(retv, addr, cmpv, newv, idx, memop); +} + +static void tcg_gen_atomic_cmpxchg_i64_int(TCGv_i64 retv, TCGTemp *addr, + TCGv_i64 cmpv, TCGv_i64 newv, + TCGArg idx, MemOp memop) { if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { - tcg_gen_nonatomic_cmpxchg_i64(retv, addr, cmpv, newv, idx, memop); + tcg_gen_nonatomic_cmpxchg_i64_int(retv, addr, cmpv, newv, idx, memop); return; } @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, } if (TCG_TARGET_REG_BITS == 32) { - tcg_gen_atomic_cmpxchg_i32(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), - TCGV_LOW(newv), idx, memop); + tcg_gen_atomic_cmpxchg_i32_int(TCGV_LOW(retv), addr, TCGV_LOW(cmpv), + TCGV_LOW(newv), idx, memop); if (memop & MO_SIGN) { tcg_gen_sari_i32(TCGV_HIGH(retv), TCGV_LOW(retv), 31); } else { @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, tcg_gen_extrl_i64_i32(c32, cmpv); tcg_gen_extrl_i64_i32(n32, newv); - tcg_gen_atomic_cmpxchg_i32(r32, addr, c32, n32, idx, memop & ~MO_SIGN); + tcg_gen_atomic_cmpxchg_i32_int(r32, addr, c32, n32, + idx, memop & ~MO_SIGN); tcg_temp_free_i32(c32); tcg_temp_free_i32(n32); @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, } } -void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, - TCGv_i128 newv, TCGArg idx, MemOp memop) +void tcg_gen_atomic_cmpxchg_i64_chk(TCGv_i64 retv, TCGTemp *addr, + TCGv_i64 cmpv, TCGv_i64 newv, + TCGArg idx, MemOp memop, TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & MO_SIZE) <= MO_64); + tcg_gen_atomic_cmpxchg_i64_int(retv, addr, cmpv, newv, idx, memop); +} + +static void tcg_gen_nonatomic_cmpxchg_i128_int(TCGv_i128 retv, TCGTemp *addr, + TCGv_i128 cmpv, TCGv_i128 newv, + TCGArg idx, MemOp memop) { if (TCG_TARGET_REG_BITS == 32) { /* Inline expansion below is simply too large for 32-bit hosts. */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, ? gen_helper_nonatomic_cmpxchgo_le : gen_helper_nonatomic_cmpxchgo_be); MemOpIdx oi = make_memop_idx(memop, idx); - TCGv_i64 a64; + TCGv_i64 a64 = maybe_extend_addr64(addr); - tcg_debug_assert((memop & MO_SIZE) == MO_128); - tcg_debug_assert((memop & MO_SIGN) == 0); - - a64 = maybe_extend_addr64(addr); gen(retv, cpu_env, a64, cmpv, newv, tcg_constant_i32(oi)); maybe_free_addr64(a64); } else { @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, TCGv_i64 t1 = tcg_temp_ebb_new_i64(); TCGv_i64 z = tcg_constant_i64(0); - tcg_gen_qemu_ld_i128(oldv, addr, idx, memop); + tcg_gen_qemu_ld_i128_int(oldv, addr, idx, memop); /* Compare i128 */ tcg_gen_xor_i64(t0, TCGV128_LOW(oldv), TCGV128_LOW(cmpv)); @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, TCGV128_HIGH(newv), TCGV128_HIGH(oldv)); /* Unconditional writeback. */ - tcg_gen_qemu_st_i128(tmpv, addr, idx, memop); + tcg_gen_qemu_st_i128_int(tmpv, addr, idx, memop); tcg_gen_mov_i128(retv, oldv); tcg_temp_free_i64(t0); @@ -XXX,XX +XXX,XX @@ void tcg_gen_nonatomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, } } -void tcg_gen_atomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, - TCGv_i128 newv, TCGArg idx, MemOp memop) +void tcg_gen_nonatomic_cmpxchg_i128_chk(TCGv_i128 retv, TCGTemp *addr, + TCGv_i128 cmpv, TCGv_i128 newv, + TCGArg idx, MemOp memop, + TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & (MO_SIZE | MO_SIGN)) == MO_128); + tcg_gen_nonatomic_cmpxchg_i128_int(retv, addr, cmpv, newv, idx, memop); +} + +static void tcg_gen_atomic_cmpxchg_i128_int(TCGv_i128 retv, TCGTemp *addr, + TCGv_i128 cmpv, TCGv_i128 newv, + TCGArg idx, MemOp memop) { gen_atomic_cx_i128 gen; if (!(tcg_ctx->gen_tb->cflags & CF_PARALLEL)) { - tcg_gen_nonatomic_cmpxchg_i128(retv, addr, cmpv, newv, idx, memop); + tcg_gen_nonatomic_cmpxchg_i128_int(retv, addr, cmpv, newv, idx, memop); return; } - tcg_debug_assert((memop & MO_SIZE) == MO_128); - tcg_debug_assert((memop & MO_SIGN) == 0); gen = table_cmpxchg[memop & (MO_SIZE | MO_BSWAP)]; - if (gen) { MemOpIdx oi = make_memop_idx(memop, idx); TCGv_i64 a64 = maybe_extend_addr64(addr); @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i128(TCGv_i128 retv, TCGv addr, TCGv_i128 cmpv, tcg_gen_movi_i64(TCGV128_HIGH(retv), 0); } -static void do_nonatomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, +void tcg_gen_atomic_cmpxchg_i128_chk(TCGv_i128 retv, TCGTemp *addr, + TCGv_i128 cmpv, TCGv_i128 newv, + TCGArg idx, MemOp memop, + TCGType addr_type) +{ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); + tcg_debug_assert((memop & (MO_SIZE | MO_SIGN)) == MO_128); + tcg_gen_atomic_cmpxchg_i128_int(retv, addr, cmpv, newv, idx, memop); +} + +static void do_nonatomic_op_i32(TCGv_i32 ret, TCGTemp *addr, TCGv_i32 val, TCGArg idx, MemOp memop, bool new_val, void (*gen)(TCGv_i32, TCGv_i32, TCGv_i32)) { @@ -XXX,XX +XXX,XX @@ static void do_nonatomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, memop = tcg_canonicalize_memop(memop, 0, 0); - tcg_gen_qemu_ld_i32(t1, addr, idx, memop); + tcg_gen_qemu_ld_i32_int(t1, addr, idx, memop); tcg_gen_ext_i32(t2, val, memop); gen(t2, t1, t2); - tcg_gen_qemu_st_i32(t2, addr, idx, memop); + tcg_gen_qemu_st_i32_int(t2, addr, idx, memop); tcg_gen_ext_i32(ret, (new_val ? t2 : t1), memop); tcg_temp_free_i32(t1); tcg_temp_free_i32(t2); } -static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, +static void do_atomic_op_i32(TCGv_i32 ret, TCGTemp *addr, TCGv_i32 val, TCGArg idx, MemOp memop, void * const table[]) { gen_atomic_op_i32 gen; @@ -XXX,XX +XXX,XX @@ static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, } } -static void do_nonatomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, +static void do_nonatomic_op_i64(TCGv_i64 ret, TCGTemp *addr, TCGv_i64 val, TCGArg idx, MemOp memop, bool new_val, void (*gen)(TCGv_i64, TCGv_i64, TCGv_i64)) { @@ -XXX,XX +XXX,XX @@ static void do_nonatomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, memop = tcg_canonicalize_memop(memop, 1, 0); - tcg_gen_qemu_ld_i64(t1, addr, idx, memop); + tcg_gen_qemu_ld_i64_int(t1, addr, idx, memop); tcg_gen_ext_i64(t2, val, memop); gen(t2, t1, t2); - tcg_gen_qemu_st_i64(t2, addr, idx, memop); + tcg_gen_qemu_st_i64_int(t2, addr, idx, memop); tcg_gen_ext_i64(ret, (new_val ? t2 : t1), memop); tcg_temp_free_i64(t1); tcg_temp_free_i64(t2); } -static void do_atomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, +static void do_atomic_op_i64(TCGv_i64 ret, TCGTemp *addr, TCGv_i64 val, TCGArg idx, MemOp memop, void * const table[]) { memop = tcg_canonicalize_memop(memop, 1, 0); if ((memop & MO_SIZE) == MO_64) { -#ifdef CONFIG_ATOMIC64 - gen_atomic_op_i64 gen; - TCGv_i64 a64; - MemOpIdx oi; + gen_atomic_op_i64 gen = table[memop & (MO_SIZE | MO_BSWAP)]; - gen = table[memop & (MO_SIZE | MO_BSWAP)]; - tcg_debug_assert(gen != NULL); + if (gen) { + MemOpIdx oi = make_memop_idx(memop & ~MO_SIGN, idx); + TCGv_i64 a64 = maybe_extend_addr64(addr); + gen(ret, cpu_env, a64, val, tcg_constant_i32(oi)); + maybe_free_addr64(a64); + return; + } - oi = make_memop_idx(memop & ~MO_SIGN, idx); - a64 = maybe_extend_addr64(addr); - gen(ret, cpu_env, a64, val, tcg_constant_i32(oi)); - maybe_free_addr64(a64); -#else gen_helper_exit_atomic(cpu_env); /* Produce a result, so that we have a well-formed opcode stream with respect to uses of the result in the (dead) code following. */ tcg_gen_movi_i64(ret, 0); -#endif /* CONFIG_ATOMIC64 */ } else { TCGv_i32 v32 = tcg_temp_ebb_new_i32(); TCGv_i32 r32 = tcg_temp_ebb_new_i32(); @@ -XXX,XX +XXX,XX @@ static void * const table_##NAME[(MO_SIZE | MO_BSWAP) + 1] = { \ WITH_ATOMIC64([MO_64 | MO_LE] = gen_helper_atomic_##NAME##q_le) \ WITH_ATOMIC64([MO_64 | MO_BE] = gen_helper_atomic_##NAME##q_be) \ }; \ -void tcg_gen_atomic_##NAME##_i32 \ - (TCGv_i32 ret, TCGv addr, TCGv_i32 val, TCGArg idx, MemOp memop) \ +void tcg_gen_atomic_##NAME##_i32_chk(TCGv_i32 ret, TCGTemp *addr, \ + TCGv_i32 val, TCGArg idx, \ + MemOp memop, TCGType addr_type) \ { \ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); \ + tcg_debug_assert((memop & MO_SIZE) <= MO_32); \ if (tcg_ctx->gen_tb->cflags & CF_PARALLEL) { \ do_atomic_op_i32(ret, addr, val, idx, memop, table_##NAME); \ } else { \ @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_##NAME##_i32 \ tcg_gen_##OP##_i32); \ } \ } \ -void tcg_gen_atomic_##NAME##_i64 \ - (TCGv_i64 ret, TCGv addr, TCGv_i64 val, TCGArg idx, MemOp memop) \ +void tcg_gen_atomic_##NAME##_i64_chk(TCGv_i64 ret, TCGTemp *addr, \ + TCGv_i64 val, TCGArg idx, \ + MemOp memop, TCGType addr_type) \ { \ + tcg_debug_assert(addr_type == tcg_ctx->addr_type); \ + tcg_debug_assert((memop & MO_SIZE) <= MO_64); \ if (tcg_ctx->gen_tb->cflags & CF_PARALLEL) { \ do_atomic_op_i64(ret, addr, val, idx, memop, table_##NAME); \ } else { \ -- 2.34.1 For 32-bit hosts, we cannot simply rely on TCGContext.addr_bits, as we need one or two host registers to represent the guest address. Create the new opcodes and update all users. Since we have not yet eliminated TARGET_LONG_BITS, only one of the two opcodes will ever be used, so we can get away with treating them the same in the backends. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-opc.h | 35 ++++++++---- tcg/optimize.c | 19 +++++-- tcg/tcg-op-ldst.c | 83 ++++++++++++++++++++++------- tcg/tcg.c | 42 ++++++++++----- tcg/tci.c | 32 +++++++---- tcg/aarch64/tcg-target.c.inc | 36 ++++++++----- tcg/arm/tcg-target.c.inc | 83 +++++++++++++++-------------- tcg/i386/tcg-target.c.inc | 91 ++++++++++++++++++++------------ tcg/loongarch64/tcg-target.c.inc | 24 ++++++--- tcg/mips/tcg-target.c.inc | 66 ++++++++++++++--------- tcg/ppc/tcg-target.c.inc | 91 +++++++++++++++++++------------- tcg/riscv/tcg-target.c.inc | 24 ++++++--- tcg/s390x/tcg-target.c.inc | 36 ++++++++----- tcg/sparc64/tcg-target.c.inc | 24 ++++++--- tcg/tci/tcg-target.c.inc | 44 ++++++++------- 15 files changed, 468 insertions(+), 262 deletions(-) diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg-opc.h +++ b/include/tcg/tcg-opc.h @@ -XXX,XX +XXX,XX @@ DEF(muls2_i64, 2, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_muls2_i64)) DEF(muluh_i64, 1, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_muluh_i64)) DEF(mulsh_i64, 1, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_mulsh_i64)) -#define TLADDR_ARGS (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? 1 : 2) #define DATA64_ARGS (TCG_TARGET_REG_BITS == 64 ? 1 : 2) /* QEMU specific */ @@ -XXX,XX +XXX,XX @@ DEF(goto_ptr, 0, 1, 0, TCG_OPF_BB_EXIT | TCG_OPF_BB_END) DEF(plugin_cb_start, 0, 0, 3, TCG_OPF_NOT_PRESENT) DEF(plugin_cb_end, 0, 0, 0, TCG_OPF_NOT_PRESENT) -DEF(qemu_ld_i32, 1, TLADDR_ARGS, 1, +/* Replicate ld/st ops for 32 and 64-bit guest addresses. */ +DEF(qemu_ld_a32_i32, 1, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_st_i32, 0, TLADDR_ARGS + 1, 1, +DEF(qemu_st_a32_i32, 0, 1 + 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) -DEF(qemu_ld_i64, DATA64_ARGS, TLADDR_ARGS, 1, +DEF(qemu_ld_a32_i64, DATA64_ARGS, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT) -DEF(qemu_st_i64, 0, TLADDR_ARGS + DATA64_ARGS, 1, +DEF(qemu_st_a32_i64, 0, DATA64_ARGS + 1, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT) + +DEF(qemu_ld_a64_i32, 1, DATA64_ARGS, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) +DEF(qemu_st_a64_i32, 0, 1 + DATA64_ARGS, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS) +DEF(qemu_ld_a64_i64, DATA64_ARGS, DATA64_ARGS, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT) +DEF(qemu_st_a64_i64, 0, DATA64_ARGS + DATA64_ARGS, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT) /* Only used by i386 to cope with stupid register constraints. */ -DEF(qemu_st8_i32, 0, TLADDR_ARGS + 1, 1, +DEF(qemu_st8_a32_i32, 0, 1 + 1, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | + IMPL(TCG_TARGET_HAS_qemu_st8_i32)) +DEF(qemu_st8_a64_i32, 0, 1 + DATA64_ARGS, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | IMPL(TCG_TARGET_HAS_qemu_st8_i32)) /* Only for 64-bit hosts at the moment. */ -DEF(qemu_ld_i128, 2, 1, 1, +DEF(qemu_ld_a32_i128, 2, 1, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT | IMPL(TCG_TARGET_HAS_qemu_ldst_i128)) -DEF(qemu_st_i128, 0, 3, 1, +DEF(qemu_ld_a64_i128, 2, 1, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT | + IMPL(TCG_TARGET_HAS_qemu_ldst_i128)) +DEF(qemu_st_a32_i128, 0, 3, 1, + TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT | + IMPL(TCG_TARGET_HAS_qemu_ldst_i128)) +DEF(qemu_st_a64_i128, 0, 3, 1, TCG_OPF_CALL_CLOBBER | TCG_OPF_SIDE_EFFECTS | TCG_OPF_64BIT | IMPL(TCG_TARGET_HAS_qemu_ldst_i128)) @@ -XXX,XX +XXX,XX @@ DEF(tci_movi, 1, 0, 1, TCG_OPF_NOT_PRESENT) DEF(tci_movl, 1, 0, 1, TCG_OPF_NOT_PRESENT) #endif -#undef TLADDR_ARGS #undef DATA64_ARGS #undef IMPL #undef IMPL64 diff --git a/tcg/optimize.c b/tcg/optimize.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(orc): done = fold_orc(&ctx, op); break; - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: done = fold_qemu_ld(&ctx, op); break; - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st8_i32: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st8_a32_i32: + case INDEX_op_qemu_st8_a64_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: done = fold_qemu_st(&ctx, op); break; CASE_OP_32_64(rem): diff --git a/tcg/tcg-op-ldst.c b/tcg/tcg-op-ldst.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-ldst.c +++ b/tcg/tcg-op-ldst.c @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i32_int(TCGv_i32 val, TCGTemp *addr, MemOp orig_memop; MemOpIdx orig_oi, oi; TCGv_i64 copy_addr; + TCGOpcode opc; tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); orig_memop = memop = tcg_canonicalize_memop(memop, 0, 0); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i32_int(TCGv_i32 val, TCGTemp *addr, } copy_addr = plugin_maybe_preserve_addr(addr); - gen_ldst(INDEX_op_qemu_ld_i32, tcgv_i32_temp(val), NULL, addr, oi); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_ld_a32_i32; + } else { + opc = INDEX_op_qemu_ld_a64_i32; + } + gen_ldst(opc, tcgv_i32_temp(val), NULL, addr, oi); plugin_gen_mem_callbacks(copy_addr, addr, orig_oi, QEMU_PLUGIN_MEM_R); if ((orig_memop ^ memop) & MO_BSWAP) { @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i32_int(TCGv_i32 val, TCGTemp *addr, } if (TCG_TARGET_HAS_qemu_st8_i32 && (memop & MO_SIZE) == MO_8) { - opc = INDEX_op_qemu_st8_i32; + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_st8_a32_i32; + } else { + opc = INDEX_op_qemu_st8_a64_i32; + } } else { - opc = INDEX_op_qemu_st_i32; + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_st_a32_i32; + } else { + opc = INDEX_op_qemu_st_a64_i32; + } } gen_ldst(opc, tcgv_i32_temp(val), NULL, addr, oi); plugin_gen_mem_callbacks(NULL, addr, orig_oi, QEMU_PLUGIN_MEM_W); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i64_int(TCGv_i64 val, TCGTemp *addr, MemOp orig_memop; MemOpIdx orig_oi, oi; TCGv_i64 copy_addr; + TCGOpcode opc; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { tcg_gen_qemu_ld_i32_int(TCGV_LOW(val), addr, idx, memop); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i64_int(TCGv_i64 val, TCGTemp *addr, } copy_addr = plugin_maybe_preserve_addr(addr); - gen_ldst_i64(INDEX_op_qemu_ld_i64, val, addr, oi); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_ld_a32_i64; + } else { + opc = INDEX_op_qemu_ld_a64_i64; + } + gen_ldst_i64(opc, val, addr, oi); plugin_gen_mem_callbacks(copy_addr, addr, orig_oi, QEMU_PLUGIN_MEM_R); if ((orig_memop ^ memop) & MO_BSWAP) { @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i64_int(TCGv_i64 val, TCGTemp *addr, { TCGv_i64 swap = NULL; MemOpIdx orig_oi, oi; + TCGOpcode opc; if (TCG_TARGET_REG_BITS == 32 && (memop & MO_SIZE) < MO_64) { tcg_gen_qemu_st_i32_int(TCGV_LOW(val), addr, idx, memop); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i64_int(TCGv_i64 val, TCGTemp *addr, oi = make_memop_idx(memop, idx); } - gen_ldst_i64(INDEX_op_qemu_st_i64, val, addr, oi); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_st_a32_i64; + } else { + opc = INDEX_op_qemu_st_a64_i64; + } + gen_ldst_i64(opc, val, addr, oi); plugin_gen_mem_callbacks(NULL, addr, orig_oi, QEMU_PLUGIN_MEM_W); if (swap) { @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, { const MemOpIdx orig_oi = make_memop_idx(memop, idx); TCGv_i64 ext_addr = NULL; + TCGOpcode opc; tcg_gen_req_mo(TCG_MO_LD_LD | TCG_MO_ST_LD); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, hi = TCGV128_HIGH(val); } - gen_ldst(INDEX_op_qemu_ld_i128, tcgv_i64_temp(lo), - tcgv_i64_temp(hi), addr, oi); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_ld_a32_i128; + } else { + opc = INDEX_op_qemu_ld_a64_i128; + } + gen_ldst(opc, tcgv_i64_temp(lo), tcgv_i64_temp(hi), addr, oi); if (need_bswap) { tcg_gen_bswap64_i64(lo, lo); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, canonicalize_memop_i128_as_i64(mop, memop); need_bswap = (mop[0] ^ memop) & MO_BSWAP; + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_ld_a32_i64; + } else { + opc = INDEX_op_qemu_ld_a64_i64; + } + /* * Since there are no global TCGv_i128, there is no visible state * changed if the second load faults. Load directly into the two @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, } oi = make_memop_idx(mop[0], idx); - gen_ldst_i64(INDEX_op_qemu_ld_i64, x, addr, oi); + gen_ldst_i64(opc, x, addr, oi); if (need_bswap) { tcg_gen_bswap64_i64(x, x); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_ld_i128_int(TCGv_i128 val, TCGTemp *addr, addr_p8 = tcgv_i64_temp(t); } - gen_ldst_i64(INDEX_op_qemu_ld_i64, y, addr_p8, oi); + gen_ldst_i64(opc, y, addr_p8, oi); tcg_temp_free_internal(addr_p8); if (need_bswap) { @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, { const MemOpIdx orig_oi = make_memop_idx(memop, idx); TCGv_i64 ext_addr = NULL; + TCGOpcode opc; tcg_gen_req_mo(TCG_MO_ST_LD | TCG_MO_ST_ST); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, hi = TCGV128_HIGH(val); } - gen_ldst(INDEX_op_qemu_st_i128, tcgv_i64_temp(lo), - tcgv_i64_temp(hi), addr, oi); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_st_a32_i128; + } else { + opc = INDEX_op_qemu_st_a64_i128; + } + gen_ldst(opc, tcgv_i64_temp(lo), tcgv_i64_temp(hi), addr, oi); if (need_bswap) { tcg_temp_free_i64(lo); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, canonicalize_memop_i128_as_i64(mop, memop); + if (tcg_ctx->addr_type == TCG_TYPE_I32) { + opc = INDEX_op_qemu_st_a32_i64; + } else { + opc = INDEX_op_qemu_st_a64_i64; + } + if ((memop & MO_BSWAP) == MO_LE) { x = TCGV128_LOW(val); y = TCGV128_HIGH(val); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, tcg_gen_bswap64_i64(b, x); x = b; } - gen_ldst_i64(INDEX_op_qemu_st_i64, x, addr, - make_memop_idx(mop[0], idx)); + + gen_ldst_i64(opc, x, addr, make_memop_idx(mop[0], idx)); if (tcg_ctx->addr_type == TCG_TYPE_I32) { TCGv_i32 t = tcg_temp_ebb_new_i32(); @@ -XXX,XX +XXX,XX @@ static void tcg_gen_qemu_st_i128_int(TCGv_i128 val, TCGTemp *addr, if (b) { tcg_gen_bswap64_i64(b, y); - y = b; - } - gen_ldst_i64(INDEX_op_qemu_st_i64, y, addr_p8, - make_memop_idx(mop[1], idx)); - - if (b) { + gen_ldst_i64(opc, b, addr_p8, make_memop_idx(mop[1], idx)); tcg_temp_free_i64(b); + } else { + gen_ldst_i64(opc, y, addr_p8, make_memop_idx(mop[1], idx)); } tcg_temp_free_internal(addr_p8); } else { diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ bool tcg_op_supported(TCGOpcode op) case INDEX_op_exit_tb: case INDEX_op_goto_tb: case INDEX_op_goto_ptr: - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_ld_i64: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: return true; - case INDEX_op_qemu_st8_i32: + case INDEX_op_qemu_st8_a32_i32: + case INDEX_op_qemu_st8_a64_i32: return TCG_TARGET_HAS_qemu_st8_i32; - case INDEX_op_qemu_ld_i128: - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: return TCG_TARGET_HAS_qemu_ldst_i128; case INDEX_op_mov_i32: @@ -XXX,XX +XXX,XX @@ static void tcg_dump_ops(TCGContext *s, FILE *f, bool have_prefs) } i = 1; break; - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st8_i32: - case INDEX_op_qemu_ld_i64: - case INDEX_op_qemu_st_i64: - case INDEX_op_qemu_ld_i128: - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st8_a32_i32: + case INDEX_op_qemu_st8_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: { const char *s_al, *s_op, *s_at; MemOpIdx oi = op->args[k++]; diff --git a/tcg/tci.c b/tcg/tci.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci.c +++ b/tcg/tci.c @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, tb_ptr = ptr; break; - case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, regs[r0] = tmp32; break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, } break; - case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, tci_qemu_st(env, taddr, tmp32, oi, tb_ptr); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; @@ -XXX,XX +XXX,XX @@ int print_insn_tci(bfd_vma addr, disassemble_info *info) str_r(r3), str_r(r4), str_r(r5)); break; - case INDEX_op_qemu_ld_i64: - case INDEX_op_qemu_st_i64: - len = DIV_ROUND_UP(64, TCG_TARGET_REG_BITS); + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_st_a32_i32: + len = 1 + 1; + goto do_qemu_ldst; + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_st_a64_i32: + len = 1 + DIV_ROUND_UP(64, TCG_TARGET_REG_BITS); + goto do_qemu_ldst; + case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_st_a64_i64: + len = 2 * DIV_ROUND_UP(64, TCG_TARGET_REG_BITS); goto do_qemu_ldst; - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_st_i32: - len = 1; do_qemu_ldst: - len += DIV_ROUND_UP(TARGET_LONG_BITS, TCG_TARGET_REG_BITS); switch (len) { case 2: tci_args_rrm(insn, &r0, &r1, &oi); diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_insn(s, 3506, CSEL, ext, a0, REG0(3), REG0(4), args[5]); break; - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: tcg_out_qemu_ld(s, a0, a1, a2, ext); break; - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: tcg_out_qemu_st(s, REG0(0), a1, a2, ext); break; - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: tcg_out_qemu_ld128(s, a0, a1, a2, args[3]); break; - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: tcg_out_qemu_st128(s, REG0(0), REG0(1), a2, args[3]); break; @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_movcond_i64: return C_O1_I4(r, r, rA, rZ, rZ); - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: return C_O1_I1(r, l); - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: return C_O2_I1(r, r, l); - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: return C_O0_I2(lZ, l); - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: return C_O0_I3(lZ, lZ, l); case INDEX_op_deposit_i32: diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, ARITH_MOV, args[0], 0, 0); break; - case INDEX_op_qemu_ld_i32: - if (TARGET_LONG_BITS == 32) { - tcg_out_qemu_ld(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I32); - } else { - tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], - args[3], TCG_TYPE_I32); - } + case INDEX_op_qemu_ld_a32_i32: + tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: - if (TARGET_LONG_BITS == 32) { - tcg_out_qemu_ld(s, args[0], args[1], args[2], -1, - args[3], TCG_TYPE_I64); - } else { - tcg_out_qemu_ld(s, args[0], args[1], args[2], args[3], - args[4], TCG_TYPE_I64); - } + case INDEX_op_qemu_ld_a64_i32: + tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], + args[3], TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i32: - if (TARGET_LONG_BITS == 32) { - tcg_out_qemu_st(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I32); - } else { - tcg_out_qemu_st(s, args[0], -1, args[1], args[2], - args[3], TCG_TYPE_I32); - } + case INDEX_op_qemu_ld_a32_i64: + tcg_out_qemu_ld(s, args[0], args[1], args[2], -1, + args[3], TCG_TYPE_I64); break; - case INDEX_op_qemu_st_i64: - if (TARGET_LONG_BITS == 32) { - tcg_out_qemu_st(s, args[0], args[1], args[2], -1, - args[3], TCG_TYPE_I64); - } else { - tcg_out_qemu_st(s, args[0], args[1], args[2], args[3], - args[4], TCG_TYPE_I64); - } + case INDEX_op_qemu_ld_a64_i64: + tcg_out_qemu_ld(s, args[0], args[1], args[2], args[3], + args[4], TCG_TYPE_I64); + break; + + case INDEX_op_qemu_st_a32_i32: + tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); + break; + case INDEX_op_qemu_st_a64_i32: + tcg_out_qemu_st(s, args[0], -1, args[1], args[2], + args[3], TCG_TYPE_I32); + break; + case INDEX_op_qemu_st_a32_i64: + tcg_out_qemu_st(s, args[0], args[1], args[2], -1, + args[3], TCG_TYPE_I64); + break; + case INDEX_op_qemu_st_a64_i64: + tcg_out_qemu_st(s, args[0], args[1], args[2], args[3], + args[4], TCG_TYPE_I64); break; case INDEX_op_bswap16_i32: @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_setcond2_i32: return C_O1_I4(r, r, r, rI, rI); - case INDEX_op_qemu_ld_i32: - return TARGET_LONG_BITS == 32 ? C_O1_I1(r, q) : C_O1_I2(r, q, q); - case INDEX_op_qemu_ld_i64: - return TARGET_LONG_BITS == 32 ? C_O2_I1(e, p, q) : C_O2_I2(e, p, q, q); - case INDEX_op_qemu_st_i32: - return TARGET_LONG_BITS == 32 ? C_O0_I2(q, q) : C_O0_I3(q, q, q); - case INDEX_op_qemu_st_i64: - return TARGET_LONG_BITS == 32 ? C_O0_I3(Q, p, q) : C_O0_I4(Q, p, q, q); + case INDEX_op_qemu_ld_a32_i32: + return C_O1_I1(r, q); + case INDEX_op_qemu_ld_a64_i32: + return C_O1_I2(r, q, q); + case INDEX_op_qemu_ld_a32_i64: + return C_O2_I1(e, p, q); + case INDEX_op_qemu_ld_a64_i64: + return C_O2_I2(e, p, q, q); + case INDEX_op_qemu_st_a32_i32: + return C_O0_I2(q, q); + case INDEX_op_qemu_st_a64_i32: + return C_O0_I3(q, q, q); + case INDEX_op_qemu_st_a32_i64: + return C_O0_I3(Q, p, q); + case INDEX_op_qemu_st_a64_i64: + return C_O0_I4(Q, p, q, q); case INDEX_op_st_vec: return C_O0_I2(w, r); diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_modrm(s, OPC_GRP3_Ev + rexw, EXT3_NOT, a0); break; - case INDEX_op_qemu_ld_i32: - if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { - tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); - } else { + case INDEX_op_qemu_ld_a64_i32: + if (TCG_TARGET_REG_BITS == 32) { tcg_out_qemu_ld(s, a0, -1, a1, a2, args[3], TCG_TYPE_I32); + break; } + /* fall through */ + case INDEX_op_qemu_ld_a32_i32: + tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); - } else if (TARGET_LONG_BITS == 32) { + } else { tcg_out_qemu_ld(s, a0, a1, a2, -1, args[3], TCG_TYPE_I64); + } + break; + case INDEX_op_qemu_ld_a64_i64: + if (TCG_TARGET_REG_BITS == 64) { + tcg_out_qemu_ld(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); } else { tcg_out_qemu_ld(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); } break; - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_ld(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); break; - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st8_i32: - if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { - tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); - } else { + + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st8_a64_i32: + if (TCG_TARGET_REG_BITS == 32) { tcg_out_qemu_st(s, a0, -1, a1, a2, args[3], TCG_TYPE_I32); + break; } + /* fall through */ + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st8_a32_i32: + tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); - } else if (TARGET_LONG_BITS == 32) { + } else { tcg_out_qemu_st(s, a0, a1, a2, -1, args[3], TCG_TYPE_I64); + } + break; + case INDEX_op_qemu_st_a64_i64: + if (TCG_TARGET_REG_BITS == 64) { + tcg_out_qemu_st(s, a0, -1, a1, -1, a2, TCG_TYPE_I64); } else { tcg_out_qemu_st(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); } break; - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_st(s, a0, a1, a2, -1, args[3], TCG_TYPE_I128); break; @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_clz_i64: return have_lzcnt ? C_N1_I2(r, r, rW) : C_N1_I2(r, r, r); - case INDEX_op_qemu_ld_i32: - return (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS - ? C_O1_I1(r, L) : C_O1_I2(r, L, L)); + case INDEX_op_qemu_ld_a32_i32: + return C_O1_I1(r, L); + case INDEX_op_qemu_ld_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, L) : C_O1_I2(r, L, L); - case INDEX_op_qemu_st_i32: - return (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS - ? C_O0_I2(L, L) : C_O0_I3(L, L, L)); - case INDEX_op_qemu_st8_i32: - return (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS - ? C_O0_I2(s, L) : C_O0_I3(s, L, L)); + case INDEX_op_qemu_st_a32_i32: + return C_O0_I2(L, L); + case INDEX_op_qemu_st_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(L, L) : C_O0_I3(L, L, L); + case INDEX_op_qemu_st8_a32_i32: + return C_O0_I2(s, L); + case INDEX_op_qemu_st8_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(s, L) : C_O0_I3(s, L, L); - case INDEX_op_qemu_ld_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, L) - : TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? C_O2_I1(r, r, L) - : C_O2_I2(r, r, L, L)); + case INDEX_op_qemu_ld_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, L) : C_O2_I1(r, r, L); + case INDEX_op_qemu_ld_a64_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, L) : C_O2_I2(r, r, L, L); - case INDEX_op_qemu_st_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O0_I2(L, L) - : TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? C_O0_I3(L, L, L) - : C_O0_I4(L, L, L, L)); + case INDEX_op_qemu_st_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(L, L) : C_O0_I3(L, L, L); + case INDEX_op_qemu_st_a64_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(L, L) : C_O0_I4(L, L, L, L); - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); return C_O2_I1(r, r, L); - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); return C_O0_I3(L, L, L); diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_ldst(s, OPC_ST_D, a0, a1, a2); break; - case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I64); break; - case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I64); break; @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_st32_i64: case INDEX_op_st_i32: case INDEX_op_st_i64: - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: return C_O0_I2(rZ, r); case INDEX_op_brcond_i32: @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_ld32u_i64: case INDEX_op_ld_i32: case INDEX_op_ld_i64: - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: return C_O1_I1(r, r); case INDEX_op_andc_i32: diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_setcond2(s, args[5], a0, a1, a2, args[3], args[4]); break; - case INDEX_op_qemu_ld_i32: - if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { - tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); - } else { + case INDEX_op_qemu_ld_a64_i32: + if (TCG_TARGET_REG_BITS == 32) { tcg_out_qemu_ld(s, a0, 0, a1, a2, args[3], TCG_TYPE_I32); + break; } + /* fall through */ + case INDEX_op_qemu_ld_a32_i32: + tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); - } else if (TARGET_LONG_BITS == 32) { + } else { tcg_out_qemu_ld(s, a0, a1, a2, 0, args[3], TCG_TYPE_I64); + } + break; + case INDEX_op_qemu_ld_a64_i64: + if (TCG_TARGET_REG_BITS == 64) { + tcg_out_qemu_ld(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); } else { tcg_out_qemu_ld(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); } break; - case INDEX_op_qemu_st_i32: - if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { - tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); - } else { + + case INDEX_op_qemu_st_a64_i32: + if (TCG_TARGET_REG_BITS == 32) { tcg_out_qemu_st(s, a0, 0, a1, a2, args[3], TCG_TYPE_I32); + break; } + /* fall through */ + case INDEX_op_qemu_st_a32_i32: + tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); - } else if (TARGET_LONG_BITS == 32) { + } else { tcg_out_qemu_st(s, a0, a1, a2, 0, args[3], TCG_TYPE_I64); + } + break; + case INDEX_op_qemu_st_a64_i64: + if (TCG_TARGET_REG_BITS == 64) { + tcg_out_qemu_st(s, a0, 0, a1, 0, a2, TCG_TYPE_I64); } else { tcg_out_qemu_st(s, a0, a1, a2, args[3], args[4], TCG_TYPE_I64); } @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_brcond2_i32: return C_O0_I4(rZ, rZ, rZ, rZ); - case INDEX_op_qemu_ld_i32: - return (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 32 - ? C_O1_I1(r, r) : C_O1_I2(r, r, r)); - case INDEX_op_qemu_st_i32: - return (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 32 - ? C_O0_I2(rZ, r) : C_O0_I3(rZ, r, r)); - case INDEX_op_qemu_ld_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) - : TARGET_LONG_BITS == 32 ? C_O2_I1(r, r, r) - : C_O2_I2(r, r, r, r)); - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_ld_a32_i32: + return C_O1_I1(r, r); + case INDEX_op_qemu_ld_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O1_I2(r, r, r); + case INDEX_op_qemu_st_a32_i32: + return C_O0_I2(rZ, r); + case INDEX_op_qemu_st_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rZ, r) : C_O0_I3(rZ, r, r); + case INDEX_op_qemu_ld_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I1(r, r, r); + case INDEX_op_qemu_ld_a64_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I2(r, r, r, r); + case INDEX_op_qemu_st_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rZ, r) : C_O0_I3(rZ, rZ, r); + case INDEX_op_qemu_st_a64_i64: return (TCG_TARGET_REG_BITS == 64 ? C_O0_I2(rZ, r) - : TARGET_LONG_BITS == 32 ? C_O0_I3(rZ, rZ, r) : C_O0_I4(rZ, rZ, r, r)); default: diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out32(s, MODUD | TAB(args[0], args[1], args[2])); break; - case INDEX_op_qemu_ld_i32: - if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { - tcg_out_qemu_ld(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I32); - } else { + case INDEX_op_qemu_ld_a64_i32: + if (TCG_TARGET_REG_BITS == 32) { tcg_out_qemu_ld(s, args[0], -1, args[1], args[2], args[3], TCG_TYPE_I32); + break; } + /* fall through */ + case INDEX_op_qemu_ld_a32_i32: + tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_ld(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I64); - } else if (TARGET_LONG_BITS == 32) { + } else { tcg_out_qemu_ld(s, args[0], args[1], args[2], -1, args[3], TCG_TYPE_I64); + } + break; + case INDEX_op_qemu_ld_a64_i64: + if (TCG_TARGET_REG_BITS == 64) { + tcg_out_qemu_ld(s, args[0], -1, args[1], -1, + args[2], TCG_TYPE_I64); } else { tcg_out_qemu_ld(s, args[0], args[1], args[2], args[3], args[4], TCG_TYPE_I64); } break; - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], true); break; - case INDEX_op_qemu_st_i32: - if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { - tcg_out_qemu_st(s, args[0], -1, args[1], -1, - args[2], TCG_TYPE_I32); - } else { + case INDEX_op_qemu_st_a64_i32: + if (TCG_TARGET_REG_BITS == 32) { tcg_out_qemu_st(s, args[0], -1, args[1], args[2], args[3], TCG_TYPE_I32); + break; } + /* fall through */ + case INDEX_op_qemu_st_a32_i32: + tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_qemu_st(s, args[0], -1, args[1], -1, args[2], TCG_TYPE_I64); - } else if (TARGET_LONG_BITS == 32) { + } else { tcg_out_qemu_st(s, args[0], args[1], args[2], -1, args[3], TCG_TYPE_I64); + } + break; + case INDEX_op_qemu_st_a64_i64: + if (TCG_TARGET_REG_BITS == 64) { + tcg_out_qemu_st(s, args[0], -1, args[1], -1, + args[2], TCG_TYPE_I64); } else { tcg_out_qemu_st(s, args[0], args[1], args[2], args[3], args[4], TCG_TYPE_I64); } break; - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: tcg_debug_assert(TCG_TARGET_REG_BITS == 64); tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], false); break; @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_sub2_i32: return C_O2_I4(r, r, rI, rZM, r, r); - case INDEX_op_qemu_ld_i32: - return (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 32 - ? C_O1_I1(r, r) - : C_O1_I2(r, r, r)); - - case INDEX_op_qemu_st_i32: - return (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 32 - ? C_O0_I2(r, r) - : C_O0_I3(r, r, r)); - - case INDEX_op_qemu_ld_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) - : TARGET_LONG_BITS == 32 ? C_O2_I1(r, r, r) - : C_O2_I2(r, r, r, r)); - - case INDEX_op_qemu_st_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) - : TARGET_LONG_BITS == 32 ? C_O0_I3(r, r, r) - : C_O0_I4(r, r, r, r)); - - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i32: + return C_O1_I1(r, r); + case INDEX_op_qemu_ld_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O1_I2(r, r, r); + case INDEX_op_qemu_ld_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I1(r, r, r); + case INDEX_op_qemu_ld_a64_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I2(r, r, r, r); + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: return C_O2_I1(o, m, r); - case INDEX_op_qemu_st_i128: + + case INDEX_op_qemu_st_a32_i32: + return C_O0_I2(r, r); + case INDEX_op_qemu_st_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); + case INDEX_op_qemu_st_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); + case INDEX_op_qemu_st_a64_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I4(r, r, r, r); + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: return C_O0_I3(o, m, r); case INDEX_op_add_vec: diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_setcond(s, args[3], a0, a1, a2); break; - case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I64); break; - case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I64); break; @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_sub2_i64: return C_O2_I4(r, r, rZ, rZ, rM, rM); - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: return C_O1_I1(r, r); - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: return C_O0_I2(rZ, r); default: diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, args[2], const_args[2], args[3], const_args[3], args[4]); break; - case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: tcg_out_qemu_ld(s, args[0], args[1], args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: tcg_out_qemu_ld(s, args[0], args[1], args[2], TCG_TYPE_I64); break; - case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: tcg_out_qemu_st(s, args[0], args[1], args[2], TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: tcg_out_qemu_st(s, args[0], args[1], args[2], TCG_TYPE_I64); break; - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], true); break; - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], false); break; @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_ctpop_i64: return C_O1_I1(r, r); - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: return C_O1_I1(r, r); - case INDEX_op_qemu_st_i64: - case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: return C_O0_I2(r, r); - case INDEX_op_qemu_ld_i128: + case INDEX_op_qemu_ld_a32_i128: + case INDEX_op_qemu_ld_a64_i128: return C_O2_I1(o, m, r); - case INDEX_op_qemu_st_i128: + case INDEX_op_qemu_st_a32_i128: + case INDEX_op_qemu_st_a64_i128: return C_O0_I3(o, m, r); case INDEX_op_deposit_i32: diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_arithi(s, a1, a0, 32, SHIFT_SRLX); break; - case INDEX_op_qemu_ld_i32: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: tcg_out_qemu_ld(s, a0, a1, a2, TCG_TYPE_I64); break; - case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I32); break; - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: tcg_out_qemu_st(s, a0, a1, a2, TCG_TYPE_I64); break; @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_extu_i32_i64: case INDEX_op_extrl_i64_i32: case INDEX_op_extrh_i64_i32: - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_ld_i64: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: return C_O1_I1(r, r); case INDEX_op_st8_i32: @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_st_i32: case INDEX_op_st32_i64: case INDEX_op_st_i64: - case INDEX_op_qemu_st_i32: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: return C_O0_I2(rZ, r); case INDEX_op_add_i32: diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci/tcg-target.c.inc +++ b/tcg/tci/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) case INDEX_op_setcond2_i32: return C_O1_I4(r, r, r, r, r); - case INDEX_op_qemu_ld_i32: - return (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS - ? C_O1_I1(r, r) - : C_O1_I2(r, r, r)); - case INDEX_op_qemu_ld_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) - : TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? C_O2_I1(r, r, r) - : C_O2_I2(r, r, r, r)); - case INDEX_op_qemu_st_i32: - return (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS - ? C_O0_I2(r, r) - : C_O0_I3(r, r, r)); - case INDEX_op_qemu_st_i64: - return (TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) - : TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? C_O0_I3(r, r, r) - : C_O0_I4(r, r, r, r)); + case INDEX_op_qemu_ld_a32_i32: + return C_O1_I1(r, r); + case INDEX_op_qemu_ld_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O1_I2(r, r, r); + case INDEX_op_qemu_ld_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I1(r, r, r); + case INDEX_op_qemu_ld_a64_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O1_I1(r, r) : C_O2_I2(r, r, r, r); + case INDEX_op_qemu_st_a32_i32: + return C_O0_I2(r, r); + case INDEX_op_qemu_st_a64_i32: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); + case INDEX_op_qemu_st_a32_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I3(r, r, r); + case INDEX_op_qemu_st_a64_i64: + return TCG_TARGET_REG_BITS == 64 ? C_O0_I2(r, r) : C_O0_I4(r, r, r, r); default: g_assert_not_reached(); @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, tcg_out_op_rrrr(s, opc, args[0], args[1], args[2], args[3]); break; - case INDEX_op_qemu_ld_i32: - case INDEX_op_qemu_st_i32: + case INDEX_op_qemu_ld_a32_i32: + case INDEX_op_qemu_ld_a64_i32: + case INDEX_op_qemu_st_a32_i32: + case INDEX_op_qemu_st_a64_i32: if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); } else { @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, } break; - case INDEX_op_qemu_ld_i64: - case INDEX_op_qemu_st_i64: + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_ld_a64_i64: + case INDEX_op_qemu_st_a32_i64: + case INDEX_op_qemu_st_a64_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); } else if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { -- 2.34.1 We now have the address size as part of the opcode, so we no longer need to test TARGET_LONG_BITS. We can use uint64_t for target_ulong, as passed into load/store helpers. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tci.c | 61 +++++++++++++++++++++++++--------------- tcg/tci/tcg-target.c.inc | 15 +++++----- 2 files changed, 46 insertions(+), 30 deletions(-) diff --git a/tcg/tci.c b/tcg/tci.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci.c +++ b/tcg/tci.c @@ -XXX,XX +XXX,XX @@ static bool tci_compare64(uint64_t u0, uint64_t u1, TCGCond condition) return result; } -static uint64_t tci_qemu_ld(CPUArchState *env, target_ulong taddr, +static uint64_t tci_qemu_ld(CPUArchState *env, uint64_t taddr, MemOpIdx oi, const void *tb_ptr) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ static uint64_t tci_qemu_ld(CPUArchState *env, target_ulong taddr, } } -static void tci_qemu_st(CPUArchState *env, target_ulong taddr, uint64_t val, +static void tci_qemu_st(CPUArchState *env, uint64_t taddr, uint64_t val, MemOpIdx oi, const void *tb_ptr) { MemOp mop = get_memop(oi); @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, TCGReg r0, r1, r2, r3, r4, r5; tcg_target_ulong t1; TCGCond condition; - target_ulong taddr; uint8_t pos, len; uint32_t tmp32; - uint64_t tmp64; + uint64_t tmp64, taddr; uint64_t T1, T2; MemOpIdx oi; int32_t ofs; @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, break; case INDEX_op_qemu_ld_a32_i32: + tci_args_rrm(insn, &r0, &r1, &oi); + taddr = (uint32_t)regs[r1]; + goto do_ld_i32; case INDEX_op_qemu_ld_a64_i32: - if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { + if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; } else { tci_args_rrrm(insn, &r0, &r1, &r2, &oi); taddr = tci_uint64(regs[r2], regs[r1]); } - tmp32 = tci_qemu_ld(env, taddr, oi, tb_ptr); - regs[r0] = tmp32; + do_ld_i32: + regs[r0] = tci_qemu_ld(env, taddr, oi, tb_ptr); break; case INDEX_op_qemu_ld_a32_i64: + if (TCG_TARGET_REG_BITS == 64) { + tci_args_rrm(insn, &r0, &r1, &oi); + taddr = (uint32_t)regs[r1]; + } else { + tci_args_rrrm(insn, &r0, &r1, &r2, &oi); + taddr = (uint32_t)regs[r2]; + } + goto do_ld_i64; case INDEX_op_qemu_ld_a64_i64: if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; - } else if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { - tci_args_rrrm(insn, &r0, &r1, &r2, &oi); - taddr = regs[r2]; } else { tci_args_rrrrr(insn, &r0, &r1, &r2, &r3, &r4); taddr = tci_uint64(regs[r3], regs[r2]); oi = regs[r4]; } + do_ld_i64: tmp64 = tci_qemu_ld(env, taddr, oi, tb_ptr); if (TCG_TARGET_REG_BITS == 32) { tci_write_reg64(regs, r1, r0, tmp64); @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, break; case INDEX_op_qemu_st_a32_i32: + tci_args_rrm(insn, &r0, &r1, &oi); + taddr = (uint32_t)regs[r1]; + goto do_st_i32; case INDEX_op_qemu_st_a64_i32: - if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { + if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); taddr = regs[r1]; } else { tci_args_rrrm(insn, &r0, &r1, &r2, &oi); taddr = tci_uint64(regs[r2], regs[r1]); } - tmp32 = regs[r0]; - tci_qemu_st(env, taddr, tmp32, oi, tb_ptr); + do_st_i32: + tci_qemu_st(env, taddr, regs[r0], oi, tb_ptr); break; case INDEX_op_qemu_st_a32_i64: + if (TCG_TARGET_REG_BITS == 64) { + tci_args_rrm(insn, &r0, &r1, &oi); + tmp64 = regs[r0]; + taddr = (uint32_t)regs[r1]; + } else { + tci_args_rrrm(insn, &r0, &r1, &r2, &oi); + tmp64 = tci_uint64(regs[r1], regs[r0]); + taddr = (uint32_t)regs[r2]; + } + goto do_st_i64; case INDEX_op_qemu_st_a64_i64: if (TCG_TARGET_REG_BITS == 64) { tci_args_rrm(insn, &r0, &r1, &oi); - taddr = regs[r1]; tmp64 = regs[r0]; + taddr = regs[r1]; } else { - if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { - tci_args_rrrm(insn, &r0, &r1, &r2, &oi); - taddr = regs[r2]; - } else { - tci_args_rrrrr(insn, &r0, &r1, &r2, &r3, &r4); - taddr = tci_uint64(regs[r3], regs[r2]); - oi = regs[r4]; - } + tci_args_rrrrr(insn, &r0, &r1, &r2, &r3, &r4); tmp64 = tci_uint64(regs[r1], regs[r0]); + taddr = tci_uint64(regs[r3], regs[r2]); + oi = regs[r4]; } + do_st_i64: tci_qemu_st(env, taddr, tmp64, oi, tb_ptr); break; diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/tci/tcg-target.c.inc +++ b/tcg/tci/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type, return false; } -static void stack_bounds_check(TCGReg base, target_long offset) +static void stack_bounds_check(TCGReg base, intptr_t offset) { if (base == TCG_REG_CALL_STACK) { tcg_debug_assert(offset >= 0); @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_qemu_ld_a32_i32: - case INDEX_op_qemu_ld_a64_i32: case INDEX_op_qemu_st_a32_i32: + tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); + break; + case INDEX_op_qemu_ld_a64_i32: case INDEX_op_qemu_st_a64_i32: - if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { + case INDEX_op_qemu_ld_a32_i64: + case INDEX_op_qemu_st_a32_i64: + if (TCG_TARGET_REG_BITS == 64) { tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); } else { tcg_out_op_rrrm(s, opc, args[0], args[1], args[2], args[3]); } break; - - case INDEX_op_qemu_ld_a32_i64: case INDEX_op_qemu_ld_a64_i64: - case INDEX_op_qemu_st_a32_i64: case INDEX_op_qemu_st_a64_i64: if (TCG_TARGET_REG_BITS == 64) { tcg_out_op_rrm(s, opc, args[0], args[1], args[2]); - } else if (TARGET_LONG_BITS <= TCG_TARGET_REG_BITS) { - tcg_out_op_rrrm(s, opc, args[0], args[1], args[2], args[3]); } else { tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_TMP, args[4]); tcg_out_op_rrrrr(s, opc, args[0], args[1], -- 2.34.1 Keep all 32-bit values zero extended in the register, not solely when addresses are 32 bits. This eliminates a dependency on TARGET_LONG_BITS. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.h +++ b/tcg/i386/tcg-target.h @@ -XXX,XX +XXX,XX @@ extern bool have_atomic16; #define TCG_TARGET_HAS_mulsh_i32 0 #if TCG_TARGET_REG_BITS == 64 -/* Keep target addresses zero-extended in a register. */ -#define TCG_TARGET_HAS_extrl_i64_i32 (TARGET_LONG_BITS == 32) -#define TCG_TARGET_HAS_extrh_i64_i32 (TARGET_LONG_BITS == 32) +/* Keep 32-bit values zero-extended in a register. */ +#define TCG_TARGET_HAS_extrl_i64_i32 1 +#define TCG_TARGET_HAS_extrh_i64_i32 1 #define TCG_TARGET_HAS_div2_i64 1 #define TCG_TARGET_HAS_rot_i64 1 #define TCG_TARGET_HAS_ext8s_i64 1 -- 2.34.1 Since TCG_TYPE_I32 values are kept zero-extended in registers, via omission of the REXW bit, we need not extend if the register matches. This is already relied upon by qemu_{ld,st}. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_exts_i32_i64(TCGContext *s, TCGReg dest, TCGReg src) static void tcg_out_extu_i32_i64(TCGContext *s, TCGReg dest, TCGReg src) { - tcg_out_ext32u(s, dest, src); + if (dest != src) { + tcg_out_ext32u(s, dest, src); + } } static void tcg_out_extrl_i64_i32(TCGContext *s, TCGReg dest, TCGReg src) -- 2.34.1 Because of its use on tgen_arithi, this value must be a signed 32-bit quantity, as that is what may be encoded in the insn. The truncation of the value to unsigned for 32-bit guests is done via the REX bit via 'trexw'. Removes the only uses of target_ulong from this tcg backend. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, int trexw = 0, hrexw = 0, tlbrexw = 0; unsigned mem_index = get_mmuidx(oi); unsigned s_mask = (1 << s_bits) - 1; - target_ulong tlb_mask; + int tlb_mask; ldst = new_ldst_label(s); ldst->is_ld = is_ld; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_modrm_offset(s, OPC_LEA + trexw, TCG_REG_L1, addrlo, s_mask - a_mask); } - tlb_mask = (target_ulong)TARGET_PAGE_MASK | a_mask; + tlb_mask = TARGET_PAGE_MASK | a_mask; tgen_arithi(s, ARITH_AND + trexw, TCG_REG_L1, tlb_mask, 0); /* cmp 0(TCG_REG_L0), TCG_REG_L1 */ -- 2.34.1 All uses can be infered from the INDEX_op_qemu_*_a{32,64}_* opcode being used. Add a field into TCGLabelQemuLdst to record the usage. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->addrhi_reg = addrhi; if (TCG_TARGET_REG_BITS == 64) { - if (TARGET_LONG_BITS == 64) { - ttype = TCG_TYPE_I64; - trexw = P_REXW; - } + ttype = s->addr_type; + trexw = (ttype == TCG_TYPE_I32 ? 0 : P_REXW); if (TCG_TYPE_PTR == TCG_TYPE_I64) { hrexw = P_REXW; if (TARGET_PAGE_BITS + CPU_TLB_DYN_MAX_BITS > 32) { @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->label_ptr[0] = s->code_ptr; s->code_ptr += 4; - if (TARGET_LONG_BITS > TCG_TARGET_REG_BITS) { + if (TCG_TARGET_REG_BITS == 32 && s->addr_type == TCG_TYPE_I64) { /* cmp 4(TCG_REG_L0), addrhi */ tcg_out_modrm_offset(s, OPC_CMP_GvEv, addrhi, TCG_REG_L0, cmp_ofs + 4); -- 2.34.1 All uses can be infered from the INDEX_op_qemu_*_a{32,64}_* opcode being used. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/arm/tcg-target.c.inc | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * Load the tlb comparator into R2/R3 and the fast path addend into R1. */ if (cmp_off == 0) { - if (TARGET_LONG_BITS == 64) { - tcg_out_ldrd_rwb(s, COND_AL, TCG_REG_R2, TCG_REG_R1, TCG_REG_R0); - } else { + if (s->addr_type == TCG_TYPE_I32) { tcg_out_ld32_rwb(s, COND_AL, TCG_REG_R2, TCG_REG_R1, TCG_REG_R0); + } else { + tcg_out_ldrd_rwb(s, COND_AL, TCG_REG_R2, TCG_REG_R1, TCG_REG_R0); } } else { tcg_out_dat_reg(s, COND_AL, ARITH_ADD, TCG_REG_R1, TCG_REG_R1, TCG_REG_R0, 0); - if (TARGET_LONG_BITS == 64) { - tcg_out_ldrd_8(s, COND_AL, TCG_REG_R2, TCG_REG_R1, cmp_off); - } else { + if (s->addr_type == TCG_TYPE_I32) { tcg_out_ld32_12(s, COND_AL, TCG_REG_R2, TCG_REG_R1, cmp_off); + } else { + tcg_out_ldrd_8(s, COND_AL, TCG_REG_R2, TCG_REG_R1, cmp_off); } } @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, SHIFT_IMM_LSL(TARGET_PAGE_BITS)); } - if (TARGET_LONG_BITS == 64) { + if (s->addr_type != TCG_TYPE_I32) { tcg_out_dat_reg(s, COND_EQ, ARITH_CMP, 0, TCG_REG_R3, addrhi, 0); } #else -- 2.34.1 Eliminate the test vs TARGET_LONG_BITS by considering this predicate to be always true, and simplify accordingly. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.c.inc | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ bool have_lse2; #define TCG_VEC_TMP0 TCG_REG_V31 #ifndef CONFIG_SOFTMMU -/* Note that XZR cannot be encoded in the address base register slot, - as that actaully encodes SP. So if we need to zero-extend the guest - address, via the address index register slot, we need to load even - a zero guest base into a register. */ -#define USE_GUEST_BASE (guest_base != 0 || TARGET_LONG_BITS == 32) #define TCG_REG_GUEST_BASE TCG_REG_X28 #endif @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_insn(s, 3202, B_C, TCG_COND_NE, 0); } - if (USE_GUEST_BASE) { + if (guest_base || addr_type == TCG_TYPE_I32) { h->base = TCG_REG_GUEST_BASE; h->index = addr_reg; h->index_ext = addr_type; @@ -XXX,XX +XXX,XX @@ static void tcg_target_qemu_prologue(TCGContext *s) CPU_TEMP_BUF_NLONGS * sizeof(long)); #if !defined(CONFIG_SOFTMMU) - if (USE_GUEST_BASE) { - tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_GUEST_BASE, guest_base); - tcg_regset_set_reg(s->reserved_regs, TCG_REG_GUEST_BASE); - } + /* + * Note that XZR cannot be encoded in the address base register slot, + * as that actaully encodes SP. Depending on the guest, we may need + * to zero-extend the guest address via the address index register slot, + * therefore we need to load even a zero guest base into a register. + */ + tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_GUEST_BASE, guest_base); + tcg_regset_set_reg(s->reserved_regs, TCG_REG_GUEST_BASE); #endif tcg_out_mov(s, TCG_TYPE_PTR, TCG_AREG0, tcg_target_call_iarg_regs[0]); -- 2.34.1 All uses replaced with TCGContext.addr_type. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.c.inc | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCGReg addr_reg, MemOpIdx oi, bool is_ld) { - TCGType addr_type = TARGET_LONG_BITS == 64 ? TCG_TYPE_I64 : TCG_TYPE_I32; + TCGType addr_type = s->addr_type; TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); MemOp s_bits = opc & MO_SIZE; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_insn(s, 3502, ADD, 1, TCG_REG_X1, TCG_REG_X1, TCG_REG_X0); /* Load the tlb comparator into X0, and the fast path addend into X1. */ - tcg_out_ld(s, TCG_TYPE_TL, TCG_REG_X0, TCG_REG_X1, + tcg_out_ld(s, addr_type, TCG_REG_X0, TCG_REG_X1, is_ld ? offsetof(CPUTLBEntry, addr_read) : offsetof(CPUTLBEntry, addr_write)); tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_X1, TCG_REG_X1, @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, if (a_mask >= s_mask) { x3 = addr_reg; } else { - tcg_out_insn(s, 3401, ADDI, TARGET_LONG_BITS == 64, + tcg_out_insn(s, 3401, ADDI, addr_type, TCG_REG_X3, addr_reg, s_mask - a_mask); x3 = TCG_REG_X3; } compare_mask = (uint64_t)TARGET_PAGE_MASK | a_mask; /* Store the page mask part of the address into X3. */ - tcg_out_logicali(s, I3404_ANDI, TARGET_LONG_BITS == 64, - TCG_REG_X3, x3, compare_mask); + tcg_out_logicali(s, I3404_ANDI, addr_type, TCG_REG_X3, x3, compare_mask); /* Perform the address comparison. */ - tcg_out_cmp(s, TARGET_LONG_BITS == 64, TCG_REG_X0, TCG_REG_X3, 0); + tcg_out_cmp(s, addr_type, TCG_REG_X0, TCG_REG_X3, 0); /* If not equal, we jump to the slow path. */ ldst->label_ptr[0] = s->code_ptr; -- 2.34.1 All uses replaced with TCGContext.addr_type. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/loongarch64/tcg-target.c.inc | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCGReg addr_reg, MemOpIdx oi, bool is_ld) { + TCGType addr_type = s->addr_type; TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); MemOp a_bits; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_opc_add_d(s, TCG_REG_TMP2, TCG_REG_TMP2, TCG_REG_TMP1); /* Load the tlb comparator and the addend. */ - tcg_out_ld(s, TCG_TYPE_TL, TCG_REG_TMP0, TCG_REG_TMP2, + tcg_out_ld(s, addr_type, TCG_REG_TMP0, TCG_REG_TMP2, is_ld ? offsetof(CPUTLBEntry, addr_read) : offsetof(CPUTLBEntry, addr_write)); tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP2, TCG_REG_TMP2, @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, if (a_bits < s_bits) { unsigned a_mask = (1u << a_bits) - 1; unsigned s_mask = (1u << s_bits) - 1; - tcg_out_addi(s, TCG_TYPE_TL, TCG_REG_TMP1, addr_reg, s_mask - a_mask); + tcg_out_addi(s, addr_type, TCG_REG_TMP1, addr_reg, s_mask - a_mask); } else { - tcg_out_mov(s, TCG_TYPE_TL, TCG_REG_TMP1, addr_reg); + tcg_out_mov(s, addr_type, TCG_REG_TMP1, addr_reg); } tcg_out_opc_bstrins_d(s, TCG_REG_TMP1, TCG_REG_ZERO, a_bits, TARGET_PAGE_BITS - 1); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, h->index = USE_GUEST_BASE ? TCG_GUEST_BASE_REG : TCG_REG_ZERO; #endif - if (TARGET_LONG_BITS == 32) { + if (addr_type == TCG_TYPE_I32) { h->base = TCG_REG_TMP0; tcg_out_ext32u(s, h->base, addr_reg); } else { -- 2.34.1 All uses replaced with TCGContext.addr_type. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/mips/tcg-target.c.inc | 42 +++++++++++++++++++++------------------ 1 file changed, 23 insertions(+), 19 deletions(-) diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ typedef enum { /* Aliases for convenience. */ ALIAS_PADD = sizeof(void *) == 4 ? OPC_ADDU : OPC_DADDU, ALIAS_PADDI = sizeof(void *) == 4 ? OPC_ADDIU : OPC_DADDIU, - ALIAS_TSRL = TARGET_LONG_BITS == 32 || TCG_TARGET_REG_BITS == 32 - ? OPC_SRL : OPC_DSRL, - ALIAS_TADDI = TARGET_LONG_BITS == 32 || TCG_TARGET_REG_BITS == 32 - ? OPC_ADDIU : OPC_DADDIU, } MIPSInsn; /* @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCGReg addrlo, TCGReg addrhi, MemOpIdx oi, bool is_ld) { + TCGType addr_type = s->addr_type; TCGLabelQemuLdst *ldst = NULL; MemOp opc = get_memop(oi); MemOp a_bits; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_ld(s, TCG_TYPE_PTR, TCG_TMP1, TCG_AREG0, table_off); /* Extract the TLB index from the address into TMP3. */ - tcg_out_opc_sa(s, ALIAS_TSRL, TCG_TMP3, addrlo, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + if (TCG_TARGET_REG_BITS == 32 || addr_type == TCG_TYPE_I32) { + tcg_out_opc_sa(s, OPC_SRL, TCG_TMP3, addrlo, + TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + } else { + tcg_out_dsrl(s, TCG_TMP3, addrlo, + TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + } tcg_out_opc_reg(s, OPC_AND, TCG_TMP3, TCG_TMP3, TCG_TMP0); /* Add the tlb_table pointer, creating the CPUTLBEntry address in TMP3. */ tcg_out_opc_reg(s, ALIAS_PADD, TCG_TMP3, TCG_TMP3, TCG_TMP1); - /* Load the (low-half) tlb comparator. */ - if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) { - tcg_out_ldst(s, OPC_LW, TCG_TMP0, TCG_TMP3, cmp_off + LO_OFF); - } else { - tcg_out_ld(s, TCG_TYPE_TL, TCG_TMP0, TCG_TMP3, cmp_off); - } - - if (TCG_TARGET_REG_BITS >= TARGET_LONG_BITS) { + if (TCG_TARGET_REG_BITS == 64 || addr_type == TCG_TYPE_I32) { + /* Load the tlb comparator. */ + tcg_out_ld(s, addr_type, TCG_TMP0, TCG_TMP3, cmp_off); /* Load the tlb addend for the fast path. */ tcg_out_ld(s, TCG_TYPE_PTR, TCG_TMP3, TCG_TMP3, add_off); + } else { + /* Load the low half of the tlb comparator. */ + tcg_out_ldst(s, OPC_LW, TCG_TMP0, TCG_TMP3, cmp_off + LO_OFF); } /* @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * For unaligned accesses, compare against the end of the access to * verify that it does not cross a page boundary. */ - tcg_out_movi(s, TCG_TYPE_TL, TCG_TMP1, TARGET_PAGE_MASK | a_mask); + tcg_out_movi(s, addr_type, TCG_TMP1, TARGET_PAGE_MASK | a_mask); if (a_mask < s_mask) { - tcg_out_opc_imm(s, ALIAS_TADDI, TCG_TMP2, addrlo, s_mask - a_mask); + if (TCG_TARGET_REG_BITS == 32 || addr_type == TCG_TYPE_I32) { + tcg_out_opc_imm(s, OPC_ADDIU, TCG_TMP2, addrlo, s_mask - a_mask); + } else { + tcg_out_opc_imm(s, OPC_DADDIU, TCG_TMP2, addrlo, s_mask - a_mask); + } tcg_out_opc_reg(s, OPC_AND, TCG_TMP1, TCG_TMP1, TCG_TMP2); } else { tcg_out_opc_reg(s, OPC_AND, TCG_TMP1, TCG_TMP1, addrlo); } /* Zero extend a 32-bit guest address for a 64-bit host. */ - if (TCG_TARGET_REG_BITS > TARGET_LONG_BITS) { + if (TCG_TARGET_REG_BITS == 64 && addr_type == TCG_TYPE_I32) { tcg_out_ext32u(s, TCG_TMP2, addrlo); addrlo = TCG_TMP2; } @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_opc_br(s, OPC_BNE, TCG_TMP1, TCG_TMP0); /* Load and test the high half tlb comparator. */ - if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) { + if (TCG_TARGET_REG_BITS == 32 && addr_type != TCG_TYPE_I32) { /* delay slot */ tcg_out_ldst(s, OPC_LW, TCG_TMP0, TCG_TMP3, cmp_off + HI_OFF); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, } base = addrlo; - if (TCG_TARGET_REG_BITS > TARGET_LONG_BITS) { + if (TCG_TARGET_REG_BITS == 64 && addr_type == TCG_TYPE_I32) { tcg_out_ext32u(s, TCG_REG_A0, base); base = TCG_REG_A0; } -- 2.34.1 All uses replaced with TCGContext.addr_type. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 27 ++++++++++++++------------- 1 file changed, 14 insertions(+), 13 deletions(-) diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ static void tcg_out_ld_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, next_arg = 1; loc = &info->in[next_arg]; - if (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 64) { - nmov = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, TCG_TYPE_TL, - ldst->addrlo_reg, ldst->addrhi_reg); - tcg_out_helper_load_slots(s, nmov, mov, parm); - next_arg += nmov; - } else { + if (TCG_TARGET_REG_BITS == 32 && s->addr_type == TCG_TYPE_I32) { /* * 32-bit host with 32-bit guest: zero-extend the guest address * to 64-bits for the helper by storing the low part, then @@ -XXX,XX +XXX,XX @@ static void tcg_out_ld_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, tcg_out_helper_load_imm(s, loc[!HOST_BIG_ENDIAN].arg_slot, TCG_TYPE_I32, 0, parm); next_arg += 2; + } else { + nmov = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, s->addr_type, + ldst->addrlo_reg, ldst->addrhi_reg); + tcg_out_helper_load_slots(s, nmov, mov, parm); + next_arg += nmov; } switch (info->out_kind) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, /* Handle addr argument. */ loc = &info->in[next_arg]; - if (TCG_TARGET_REG_BITS == 64 || TARGET_LONG_BITS == 64) { - n = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, TCG_TYPE_TL, - ldst->addrlo_reg, ldst->addrhi_reg); - next_arg += n; - nmov += n; - } else { + if (TCG_TARGET_REG_BITS == 32 && s->addr_type == TCG_TYPE_I32) { /* * 32-bit host with 32-bit guest: zero-extend the guest address * to 64-bits for the helper by storing the low part. Later, @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, ldst->addrlo_reg, -1); next_arg += 2; nmov += 1; + } else { + n = tcg_out_helper_add_mov(mov, loc, TCG_TYPE_I64, s->addr_type, + ldst->addrlo_reg, ldst->addrhi_reg); + next_arg += n; + nmov += n; } /* Handle data argument. */ @@ -XXX,XX +XXX,XX @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst, g_assert_not_reached(); } - if (TCG_TARGET_REG_BITS == 32 && TARGET_LONG_BITS == 32) { + if (TCG_TARGET_REG_BITS == 32 && s->addr_type == TCG_TYPE_I32) { + /* Zero extend the address by loading a zero for the high part. */ loc = &info->in[1 + !HOST_BIG_ENDIAN]; tcg_out_helper_load_imm(s, loc->arg_slot, TCG_TYPE_I32, 0, parm); } -- 2.34.1 Disconnect guest page size from TCG compilation. While this could be done via exec/target_page.h, we want to cache the value across multiple memory access operations, so we might as well initialize this early. The changes within tcg/ are entirely mechanical: sed -i s/TARGET_PAGE_BITS/s->page_bits/g sed -i s/TARGET_PAGE_MASK/s->page_mask/g Reviewed-by: Anton Johansson <anjo@rev.ng> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 5 +++++ accel/tcg/translate-all.c | 4 ++++ tcg/aarch64/tcg-target.c.inc | 6 +++--- tcg/arm/tcg-target.c.inc | 10 +++++----- tcg/i386/tcg-target.c.inc | 6 +++--- tcg/loongarch64/tcg-target.c.inc | 4 ++-- tcg/mips/tcg-target.c.inc | 6 +++--- tcg/ppc/tcg-target.c.inc | 14 +++++++------- tcg/riscv/tcg-target.c.inc | 4 ++-- tcg/s390x/tcg-target.c.inc | 4 ++-- tcg/sparc64/tcg-target.c.inc | 4 ++-- 11 files changed, 38 insertions(+), 29 deletions(-) diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -XXX,XX +XXX,XX @@ struct TCGContext { int nb_ops; TCGType addr_type; /* TCG_TYPE_I32 or TCG_TYPE_I64 */ +#ifdef CONFIG_SOFTMMU + int page_mask; + uint8_t page_bits; +#endif + TCGRegSet reserved_regs; intptr_t current_frame_offset; intptr_t frame_start; diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/translate-all.c +++ b/accel/tcg/translate-all.c @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, tb_set_page_addr1(tb, -1); tcg_ctx->gen_tb = tb; tcg_ctx->addr_type = TCG_TYPE_TL; +#ifdef CONFIG_SOFTMMU + tcg_ctx->page_bits = TARGET_PAGE_BITS; + tcg_ctx->page_mask = TARGET_PAGE_MASK; +#endif tb_overflow: diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->oi = oi; ldst->addrlo_reg = addr_reg; - mask_type = (TARGET_PAGE_BITS + CPU_TLB_DYN_MAX_BITS > 32 + mask_type = (s->page_bits + CPU_TLB_DYN_MAX_BITS > 32 ? TCG_TYPE_I64 : TCG_TYPE_I32); /* Load env_tlb(env)->f[mmu_idx].{mask,table} into {x0,x1}. */ @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Extract the TLB index from the address into X0. */ tcg_out_insn(s, 3502S, AND_LSR, mask_type == TCG_TYPE_I64, TCG_REG_X0, TCG_REG_X0, addr_reg, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); /* Add the tlb_table pointer, creating the CPUTLBEntry address into X1. */ tcg_out_insn(s, 3502, ADD, 1, TCG_REG_X1, TCG_REG_X1, TCG_REG_X0); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, TCG_REG_X3, addr_reg, s_mask - a_mask); x3 = TCG_REG_X3; } - compare_mask = (uint64_t)TARGET_PAGE_MASK | a_mask; + compare_mask = (uint64_t)s->page_mask | a_mask; /* Store the page mask part of the address into X3. */ tcg_out_logicali(s, I3404_ANDI, addr_type, TCG_REG_X3, x3, compare_mask); diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Extract the tlb index from the address into R0. */ tcg_out_dat_reg(s, COND_AL, ARITH_AND, TCG_REG_R0, TCG_REG_R0, addrlo, - SHIFT_IMM_LSR(TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS)); + SHIFT_IMM_LSR(s->page_bits - CPU_TLB_ENTRY_BITS)); /* * Add the tlb_table pointer, creating the CPUTLBEntry address in R1. @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_dat_imm(s, COND_AL, ARITH_ADD, t_addr, addrlo, s_mask - a_mask); } - if (use_armv7_instructions && TARGET_PAGE_BITS <= 16) { - tcg_out_movi32(s, COND_AL, TCG_REG_TMP, ~(TARGET_PAGE_MASK | a_mask)); + if (use_armv7_instructions && s->page_bits <= 16) { + tcg_out_movi32(s, COND_AL, TCG_REG_TMP, ~(s->page_mask | a_mask)); tcg_out_dat_reg(s, COND_AL, ARITH_BIC, TCG_REG_TMP, t_addr, TCG_REG_TMP, 0); tcg_out_dat_reg(s, COND_AL, ARITH_CMP, 0, TCG_REG_R2, TCG_REG_TMP, 0); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addrlo, a_mask); } tcg_out_dat_reg(s, COND_AL, ARITH_MOV, TCG_REG_TMP, 0, t_addr, - SHIFT_IMM_LSR(TARGET_PAGE_BITS)); + SHIFT_IMM_LSR(s->page_bits)); tcg_out_dat_reg(s, (a_mask ? COND_EQ : COND_AL), ARITH_CMP, 0, TCG_REG_R2, TCG_REG_TMP, - SHIFT_IMM_LSL(TARGET_PAGE_BITS)); + SHIFT_IMM_LSL(s->page_bits)); } if (s->addr_type != TCG_TYPE_I32) { diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, trexw = (ttype == TCG_TYPE_I32 ? 0 : P_REXW); if (TCG_TYPE_PTR == TCG_TYPE_I64) { hrexw = P_REXW; - if (TARGET_PAGE_BITS + CPU_TLB_DYN_MAX_BITS > 32) { + if (s->page_bits + CPU_TLB_DYN_MAX_BITS > 32) { tlbtype = TCG_TYPE_I64; tlbrexw = P_REXW; } @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_mov(s, tlbtype, TCG_REG_L0, addrlo); tcg_out_shifti(s, SHIFT_SHR + tlbrexw, TCG_REG_L0, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); tcg_out_modrm_offset(s, OPC_AND_GvEv + trexw, TCG_REG_L0, TCG_AREG0, TLB_MASK_TABLE_OFS(mem_index) + @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_modrm_offset(s, OPC_LEA + trexw, TCG_REG_L1, addrlo, s_mask - a_mask); } - tlb_mask = TARGET_PAGE_MASK | a_mask; + tlb_mask = s->page_mask | a_mask; tgen_arithi(s, ARITH_AND + trexw, TCG_REG_L1, tlb_mask, 0); /* cmp 0(TCG_REG_L0), TCG_REG_L1 */ diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/loongarch64/tcg-target.c.inc +++ b/tcg/loongarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_AREG0, table_ofs); tcg_out_opc_srli_d(s, TCG_REG_TMP2, addr_reg, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); tcg_out_opc_and(s, TCG_REG_TMP2, TCG_REG_TMP2, TCG_REG_TMP0); tcg_out_opc_add_d(s, TCG_REG_TMP2, TCG_REG_TMP2, TCG_REG_TMP1); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, tcg_out_mov(s, addr_type, TCG_REG_TMP1, addr_reg); } tcg_out_opc_bstrins_d(s, TCG_REG_TMP1, TCG_REG_ZERO, - a_bits, TARGET_PAGE_BITS - 1); + a_bits, s->page_bits - 1); /* Compare masked address with the TLB entry. */ ldst->label_ptr[0] = s->code_ptr; diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Extract the TLB index from the address into TMP3. */ if (TCG_TARGET_REG_BITS == 32 || addr_type == TCG_TYPE_I32) { tcg_out_opc_sa(s, OPC_SRL, TCG_TMP3, addrlo, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); } else { tcg_out_dsrl(s, TCG_TMP3, addrlo, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); } tcg_out_opc_reg(s, OPC_AND, TCG_TMP3, TCG_TMP3, TCG_TMP0); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * For unaligned accesses, compare against the end of the access to * verify that it does not cross a page boundary. */ - tcg_out_movi(s, addr_type, TCG_TMP1, TARGET_PAGE_MASK | a_mask); + tcg_out_movi(s, addr_type, TCG_TMP1, s->page_mask | a_mask); if (a_mask < s_mask) { if (TCG_TARGET_REG_BITS == 32 || addr_type == TCG_TYPE_I32) { tcg_out_opc_imm(s, OPC_ADDIU, TCG_TMP2, addrlo, s_mask - a_mask); diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Extract the page index, shifted into place for tlb index. */ if (TCG_TARGET_REG_BITS == 32) { tcg_out_shri32(s, TCG_REG_R0, addrlo, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); } else { tcg_out_shri64(s, TCG_REG_R0, addrlo, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); } tcg_out32(s, AND | SAB(TCG_REG_TMP1, TCG_REG_TMP1, TCG_REG_R0)); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, a_bits = s_bits; } tcg_out_rlw(s, RLWINM, TCG_REG_R0, addrlo, 0, - (32 - a_bits) & 31, 31 - TARGET_PAGE_BITS); + (32 - a_bits) & 31, 31 - s->page_bits); } else { TCGReg t = addrlo; @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Mask the address for the requested alignment. */ if (TARGET_LONG_BITS == 32) { tcg_out_rlw(s, RLWINM, TCG_REG_R0, t, 0, - (32 - a_bits) & 31, 31 - TARGET_PAGE_BITS); + (32 - a_bits) & 31, 31 - s->page_bits); } else if (a_bits == 0) { - tcg_out_rld(s, RLDICR, TCG_REG_R0, t, 0, 63 - TARGET_PAGE_BITS); + tcg_out_rld(s, RLDICR, TCG_REG_R0, t, 0, 63 - s->page_bits); } else { tcg_out_rld(s, RLDICL, TCG_REG_R0, t, - 64 - TARGET_PAGE_BITS, TARGET_PAGE_BITS - a_bits); - tcg_out_rld(s, RLDICL, TCG_REG_R0, TCG_REG_R0, TARGET_PAGE_BITS, 0); + 64 - s->page_bits, s->page_bits - a_bits); + tcg_out_rld(s, RLDICL, TCG_REG_R0, TCG_REG_R0, s->page_bits, 0); } } diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP1, TCG_AREG0, table_ofs); tcg_out_opc_imm(s, OPC_SRLI, TCG_REG_TMP2, addr_reg, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); tcg_out_opc_reg(s, OPC_AND, TCG_REG_TMP2, TCG_REG_TMP2, TCG_REG_TMP0); tcg_out_opc_reg(s, OPC_ADD, TCG_REG_TMP2, TCG_REG_TMP2, TCG_REG_TMP1); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, TCGReg *pbase, tcg_out_opc_imm(s, TARGET_LONG_BITS == 32 ? OPC_ADDIW : OPC_ADDI, addr_adj, addr_reg, s_mask - a_mask); } - compare_mask = TARGET_PAGE_MASK | a_mask; + compare_mask = s->page_mask | a_mask; if (compare_mask == sextreg(compare_mask, 0, 12)) { tcg_out_opc_imm(s, OPC_ANDI, TCG_REG_TMP1, addr_adj, compare_mask); } else { diff --git a/tcg/s390x/tcg-target.c.inc b/tcg/s390x/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390x/tcg-target.c.inc +++ b/tcg/s390x/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->addrlo_reg = addr_reg; tcg_out_sh64(s, RSY_SRLG, TCG_TMP0, addr_reg, TCG_REG_NONE, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS); + s->page_bits - CPU_TLB_ENTRY_BITS); QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) > 0); QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) < -(1 << 19)); @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, * cross pages using the address of the last byte of the access. */ a_off = (a_mask >= s_mask ? 0 : s_mask - a_mask); - tlb_mask = (uint64_t)TARGET_PAGE_MASK | a_mask; + tlb_mask = (uint64_t)s->page_mask | a_mask; if (a_off == 0) { tgen_andi_risbg(s, TCG_REG_R0, addr_reg, tlb_mask); } else { diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc64/tcg-target.c.inc +++ b/tcg/sparc64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, /* Extract the page index, shifted into place for tlb index. */ tcg_out_arithi(s, TCG_REG_T1, addr_reg, - TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS, SHIFT_SRL); + s->page_bits - CPU_TLB_ENTRY_BITS, SHIFT_SRL); tcg_out_arith(s, TCG_REG_T1, TCG_REG_T1, TCG_REG_T2, ARITH_AND); /* Add the tlb_table pointer, creating the CPUTLBEntry address into R2. */ @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, h->base = TCG_REG_T1; /* Mask out the page offset, except for the required alignment. */ - compare_mask = TARGET_PAGE_MASK | a_mask; + compare_mask = s->page_mask | a_mask; if (check_fit_tl(compare_mask, 13)) { tcg_out_arithi(s, TCG_REG_T3, addr_reg, compare_mask, ARITH_AND); } else { -- 2.34.1 Disconnect guest tlb parameters from TCG compilation. Reviewed-by: Anton Johansson <anjo@rev.ng> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 1 + accel/tcg/translate-all.c | 1 + tcg/aarch64/tcg-target.c.inc | 2 +- tcg/i386/tcg-target.c.inc | 2 +- 4 files changed, 4 insertions(+), 2 deletions(-) diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index XXXXXXX..XXXXXXX 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -XXX,XX +XXX,XX @@ struct TCGContext { #ifdef CONFIG_SOFTMMU int page_mask; uint8_t page_bits; + uint8_t tlb_dyn_max_bits; #endif TCGRegSet reserved_regs; diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/translate-all.c +++ b/accel/tcg/translate-all.c @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, #ifdef CONFIG_SOFTMMU tcg_ctx->page_bits = TARGET_PAGE_BITS; tcg_ctx->page_mask = TARGET_PAGE_MASK; + tcg_ctx->tlb_dyn_max_bits = CPU_TLB_DYN_MAX_BITS; #endif tb_overflow: diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, ldst->oi = oi; ldst->addrlo_reg = addr_reg; - mask_type = (s->page_bits + CPU_TLB_DYN_MAX_BITS > 32 + mask_type = (s->page_bits + s->tlb_dyn_max_bits > 32 ? TCG_TYPE_I64 : TCG_TYPE_I32); /* Load env_tlb(env)->f[mmu_idx].{mask,table} into {x0,x1}. */ diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, trexw = (ttype == TCG_TYPE_I32 ? 0 : P_REXW); if (TCG_TYPE_PTR == TCG_TYPE_I64) { hrexw = P_REXW; - if (s->page_bits + CPU_TLB_DYN_MAX_BITS > 32) { + if (s->page_bits + s->tlb_dyn_max_bits > 32) { tlbtype = TCG_TYPE_I64; tlbrexw = P_REXW; } -- 2.34.1 TCG will need this declaration, without all of the other bits that come with cpu-all.h. Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/cpu-all.h | 5 +---- include/exec/user/guest-base.h | 12 ++++++++++++ tcg/tcg.c | 3 +++ 3 files changed, 16 insertions(+), 4 deletions(-) create mode 100644 include/exec/user/guest-base.h diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/cpu-all.h +++ b/include/exec/cpu-all.h @@ -XXX,XX +XXX,XX @@ #if defined(CONFIG_USER_ONLY) #include "exec/user/abitypes.h" +#include "exec/user/guest-base.h" -/* On some host systems the guest address space is reserved on the host. - * This allows the guest address space to be offset to a convenient location. - */ -extern uintptr_t guest_base; extern bool have_guest_base; /* diff --git a/include/exec/user/guest-base.h b/include/exec/user/guest-base.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/include/exec/user/guest-base.h @@ -XXX,XX +XXX,XX @@ +/* SPDX-License-Identifier: LGPL-2.1-or-later */ +/* + * Declaration of guest_base. + * Copyright (c) 2003 Fabrice Bellard + */ + +#ifndef EXEC_USER_GUEST_BASE_H +#define EXEC_USER_GUEST_BASE_H + +extern uintptr_t guest_base; + +#endif diff --git a/tcg/tcg.c b/tcg/tcg.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -XXX,XX +XXX,XX @@ #include "tcg/tcg-temp-internal.h" #include "tcg-internal.h" #include "accel/tcg/perf.h" +#ifdef CONFIG_USER_ONLY +#include "exec/user/guest-base.h" +#endif /* Forward declarations for functions declared in tcg-target.c.inc and used here. */ -- 2.34.1