The following changes since commit 45240eed4f064576d589ea60ebadf3c11d7ab891: Merge remote-tracking branch 'remotes/armbru/tags/pull-yank-2021-01-13' into staging (2021-01-13 14:19:24 +0000) are available in the Git repository at: https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20210113 for you to fetch changes up to 4cacecaaa2bbf8af0967bd3eee43297fada475a9: decodetree: Open files with encoding='utf-8' (2021-01-13 08:39:08 -1000) ---------------------------------------------------------------- Improvements to tcg constant handling. Force utf8 for decodetree. ---------------------------------------------------------------- Philippe Mathieu-Daudé (1): decodetree: Open files with encoding='utf-8' Richard Henderson (23): tcg: Use tcg_out_dupi_vec from temp_load tcg: Increase tcg_out_dupi_vec immediate to int64_t tcg: Consolidate 3 bits into enum TCGTempKind tcg: Add temp_readonly tcg: Expand TCGTemp.val to 64-bits tcg: Rename struct tcg_temp_info to TempOptInfo tcg: Expand TempOptInfo to 64-bits tcg: Introduce TYPE_CONST temporaries tcg/optimize: Improve find_better_copy tcg/optimize: Adjust TempOptInfo allocation tcg/optimize: Use tcg_constant_internal with constant folding tcg: Convert tcg_gen_dupi_vec to TCG_CONST tcg: Use tcg_constant_i32 with icount expander tcg: Use tcg_constant_{i32,i64} with tcg int expanders tcg: Use tcg_constant_{i32,i64} with tcg plugins tcg: Use tcg_constant_{i32,i64,vec} with gvec expanders tcg/tci: Add special tci_movi_{i32,i64} opcodes tcg: Remove movi and dupi opcodes tcg: Add tcg_reg_alloc_dup2 tcg/i386: Use tcg_constant_vec with tcg vec expanders tcg: Remove tcg_gen_dup{8,16,32,64}i_vec tcg/ppc: Use tcg_constant_vec with tcg vec expanders tcg/aarch64: Use tcg_constant_vec with tcg vec expanders include/exec/gen-icount.h | 25 +-- include/tcg/tcg-op.h | 17 +- include/tcg/tcg-opc.h | 11 +- include/tcg/tcg.h | 50 ++++- accel/tcg/plugin-gen.c | 49 ++--- tcg/optimize.c | 249 +++++++++++----------- tcg/tcg-op-gvec.c | 129 +++++------- tcg/tcg-op-vec.c | 52 +---- tcg/tcg-op.c | 227 ++++++++++---------- tcg/tcg.c | 488 +++++++++++++++++++++++++++++++++---------- tcg/tci.c | 4 +- tcg/aarch64/tcg-target.c.inc | 32 +-- tcg/arm/tcg-target.c.inc | 1 - tcg/i386/tcg-target.c.inc | 112 ++++++---- tcg/mips/tcg-target.c.inc | 2 - tcg/ppc/tcg-target.c.inc | 90 ++++---- tcg/riscv/tcg-target.c.inc | 2 - tcg/s390/tcg-target.c.inc | 2 - tcg/sparc/tcg-target.c.inc | 2 - tcg/tci/tcg-target.c.inc | 6 +- scripts/decodetree.py | 9 +- 21 files changed, 890 insertions(+), 669 deletions(-) Having dupi pass though movi is confusing and arguably wrong. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 6 +++- tcg/aarch64/tcg-target.c.inc | 7 ---- tcg/i386/tcg-target.c.inc | 63 ++++++++++++++++++++++++------------ tcg/ppc/tcg-target.c.inc | 6 ---- 4 files changed, 47 insertions(+), 35 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 temp_load(TCGContext *s, TCGTemp *ts, TCGRegSet desired_regs, case TEMP_VAL_CONST: reg = tcg_reg_alloc(s, desired_regs, allocated_regs, preferred_regs, ts->indirect_base); - tcg_out_movi(s, ts->type, reg, ts->val); + if (ts->type <= TCG_TYPE_I64) { + tcg_out_movi(s, ts->type, reg, ts->val); + } else { + tcg_out_dupi_vec(s, ts->type, reg, ts->val); + } ts->mem_coherent = 0; break; case TEMP_VAL_MEM: 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_movi(TCGContext *s, TCGType type, TCGReg rd, case TCG_TYPE_I64: tcg_debug_assert(rd < 32); break; - - case TCG_TYPE_V64: - case TCG_TYPE_V128: - tcg_debug_assert(rd >= 32); - tcg_out_dupi_vec(s, type, rd, value); - return; - default: g_assert_not_reached(); } 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_dupi_vec(TCGContext *s, TCGType type, } } -static void tcg_out_movi(TCGContext *s, TCGType type, - TCGReg ret, tcg_target_long arg) +static void tcg_out_movi_vec(TCGContext *s, TCGType type, + TCGReg ret, tcg_target_long arg) +{ + if (arg == 0) { + tcg_out_vex_modrm(s, OPC_PXOR, ret, ret, ret); + return; + } + if (arg == -1) { + tcg_out_vex_modrm(s, OPC_PCMPEQB, ret, ret, ret); + return; + } + + int rexw = (type == TCG_TYPE_I32 ? 0 : P_REXW); + tcg_out_vex_modrm_pool(s, OPC_MOVD_VyEy + rexw, ret); + if (TCG_TARGET_REG_BITS == 64) { + new_pool_label(s, arg, R_386_PC32, s->code_ptr - 4, -4); + } else { + new_pool_label(s, arg, R_386_32, s->code_ptr - 4, 0); + } +} + +static void tcg_out_movi_int(TCGContext *s, TCGType type, + TCGReg ret, tcg_target_long arg) { tcg_target_long diff; - switch (type) { - case TCG_TYPE_I32: -#if TCG_TARGET_REG_BITS == 64 - case TCG_TYPE_I64: -#endif - if (ret < 16) { - break; - } - /* fallthru */ - case TCG_TYPE_V64: - case TCG_TYPE_V128: - case TCG_TYPE_V256: - tcg_debug_assert(ret >= 16); - tcg_out_dupi_vec(s, type, ret, arg); - return; - default: - g_assert_not_reached(); - } - if (arg == 0) { tgen_arithr(s, ARITH_XOR, ret, ret); return; @@ -XXX,XX +XXX,XX @@ static void tcg_out_movi(TCGContext *s, TCGType type, tcg_out64(s, arg); } +static void tcg_out_movi(TCGContext *s, TCGType type, + TCGReg ret, tcg_target_long arg) +{ + switch (type) { + case TCG_TYPE_I32: +#if TCG_TARGET_REG_BITS == 64 + case TCG_TYPE_I64: +#endif + if (ret < 16) { + tcg_out_movi_int(s, type, ret, arg); + } else { + tcg_out_movi_vec(s, type, ret, arg); + } + break; + default: + g_assert_not_reached(); + } +} + static inline void tcg_out_pushi(TCGContext *s, tcg_target_long val) { if (val == (int8_t)val) { 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_movi(TCGContext *s, TCGType type, TCGReg ret, tcg_out_movi_int(s, type, ret, arg, false); break; - case TCG_TYPE_V64: - case TCG_TYPE_V128: - tcg_debug_assert(ret >= TCG_REG_V0); - tcg_out_dupi_vec(s, type, ret, arg); - break; - default: g_assert_not_reached(); } -- 2.25.1 While we don't store more than tcg_target_long in TCGTemp, we shouldn't be limited to that for code generation. We will be able to use this for INDEX_op_dup2_vec with 2 constants. Also pass along the minimal vece that may be said to apply to the constant. This allows some simplification in the various backends. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 31 +++++++++++++++++++++++++----- tcg/aarch64/tcg-target.c.inc | 12 ++++++------ tcg/i386/tcg-target.c.inc | 22 ++++++++++++--------- tcg/ppc/tcg-target.c.inc | 37 +++++++++++++++++++++++------------- 4 files changed, 69 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 bool tcg_out_dup_vec(TCGContext *s, TCGType type, unsigned vece, TCGReg dst, TCGReg src); static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece, TCGReg dst, TCGReg base, intptr_t offset); -static void tcg_out_dupi_vec(TCGContext *s, TCGType type, - TCGReg dst, tcg_target_long arg); +static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece, + TCGReg dst, int64_t arg); static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, unsigned vecl, unsigned vece, const TCGArg *args, const int *const_args); @@ -XXX,XX +XXX,XX @@ static inline bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece, { g_assert_not_reached(); } -static inline void tcg_out_dupi_vec(TCGContext *s, TCGType type, - TCGReg dst, tcg_target_long arg) +static inline void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece, + TCGReg dst, int64_t arg) { g_assert_not_reached(); } @@ -XXX,XX +XXX,XX @@ static void temp_load(TCGContext *s, TCGTemp *ts, TCGRegSet desired_regs, if (ts->type <= TCG_TYPE_I64) { tcg_out_movi(s, ts->type, reg, ts->val); } else { - tcg_out_dupi_vec(s, ts->type, reg, ts->val); + uint64_t val = ts->val; + MemOp vece = MO_64; + + /* + * Find the minimal vector element that matches the constant. + * The targets will, in general, have to do this search anyway, + * do this generically. + */ + if (TCG_TARGET_REG_BITS == 32) { + val = dup_const(MO_32, val); + vece = MO_32; + } + if (val == dup_const(MO_8, val)) { + vece = MO_8; + } else if (val == dup_const(MO_16, val)) { + vece = MO_16; + } else if (TCG_TARGET_REG_BITS == 64 && + val == dup_const(MO_32, val)) { + vece = MO_32; + } + + tcg_out_dupi_vec(s, ts->type, vece, reg, ts->val); } ts->mem_coherent = 0; break; 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_logicali(TCGContext *s, AArch64Insn insn, TCGType ext, tcg_out_insn_3404(s, insn, ext, rd, rn, ext, r, c); } -static void tcg_out_dupi_vec(TCGContext *s, TCGType type, - TCGReg rd, tcg_target_long v64) +static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece, + TCGReg rd, int64_t v64) { bool q = type == TCG_TYPE_V128; int cmode, imm8, i; /* Test all bytes equal first. */ - if (v64 == dup_const(MO_8, v64)) { + if (vece == MO_8) { imm8 = (uint8_t)v64; tcg_out_insn(s, 3606, MOVI, q, rd, 0, 0xe, imm8); return; @@ -XXX,XX +XXX,XX @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, * cannot find an expansion there's no point checking a larger * width because we already know by replication it cannot match. */ - if (v64 == dup_const(MO_16, v64)) { + if (vece == MO_16) { uint16_t v16 = v64; if (is_shimm16(v16, &cmode, &imm8)) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, tcg_out_insn(s, 3606, MOVI, q, rd, 0, 0x8, v16 & 0xff); tcg_out_insn(s, 3606, ORR, q, rd, 0, 0xa, v16 >> 8); return; - } else if (v64 == dup_const(MO_32, v64)) { + } else if (vece == MO_32) { uint32_t v32 = v64; uint32_t n32 = ~v32; @@ -XXX,XX +XXX,XX @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, tcg_out_insn_3617(s, insn, is_q, vece, a0, a1); break; } - tcg_out_dupi_vec(s, type, TCG_VEC_TMP, 0); + tcg_out_dupi_vec(s, type, MO_8, TCG_VEC_TMP, 0); a2 = TCG_VEC_TMP; } insn = cmp_insn[cond]; 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_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece, return true; } -static void tcg_out_dupi_vec(TCGContext *s, TCGType type, - TCGReg ret, tcg_target_long arg) +static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece, + TCGReg ret, int64_t arg) { int vex_l = (type == TCG_TYPE_V256 ? P_VEXL : 0); @@ -XXX,XX +XXX,XX @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, return; } - if (TCG_TARGET_REG_BITS == 64) { + if (TCG_TARGET_REG_BITS == 32 && vece < MO_64) { + if (have_avx2) { + tcg_out_vex_modrm_pool(s, OPC_VPBROADCASTD + vex_l, ret); + } else { + tcg_out_vex_modrm_pool(s, OPC_VBROADCASTSS, ret); + } + new_pool_label(s, arg, R_386_32, s->code_ptr - 4, 0); + } else { if (type == TCG_TYPE_V64) { tcg_out_vex_modrm_pool(s, OPC_MOVQ_VqWq, ret); } else if (have_avx2) { @@ -XXX,XX +XXX,XX @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, } else { tcg_out_vex_modrm_pool(s, OPC_MOVDDUP, ret); } - new_pool_label(s, arg, R_386_PC32, s->code_ptr - 4, -4); - } else { - if (have_avx2) { - tcg_out_vex_modrm_pool(s, OPC_VPBROADCASTD + vex_l, ret); + if (TCG_TARGET_REG_BITS == 64) { + new_pool_label(s, arg, R_386_PC32, s->code_ptr - 4, -4); } else { - tcg_out_vex_modrm_pool(s, OPC_VBROADCASTSS, ret); + new_pool_l2(s, R_386_32, s->code_ptr - 4, 0, arg, arg >> 32); } - new_pool_label(s, arg, R_386_32, s->code_ptr - 4, 0); } } 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_movi_int(TCGContext *s, TCGType type, TCGReg ret, } } -static void tcg_out_dupi_vec(TCGContext *s, TCGType type, TCGReg ret, - tcg_target_long val) +static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece, + TCGReg ret, int64_t val) { uint32_t load_insn; int rel, low; intptr_t add; - low = (int8_t)val; - if (low >= -16 && low < 16) { - if (val == (tcg_target_long)dup_const(MO_8, low)) { + switch (vece) { + case MO_8: + low = (int8_t)val; + if (low >= -16 && low < 16) { tcg_out32(s, VSPLTISB | VRT(ret) | ((val & 31) << 16)); return; } - if (val == (tcg_target_long)dup_const(MO_16, low)) { + if (have_isa_3_00) { + tcg_out32(s, XXSPLTIB | VRT(ret) | ((val & 0xff) << 11)); + return; + } + break; + + case MO_16: + low = (int16_t)val; + if (low >= -16 && low < 16) { tcg_out32(s, VSPLTISH | VRT(ret) | ((val & 31) << 16)); return; } - if (val == (tcg_target_long)dup_const(MO_32, low)) { + break; + + case MO_32: + low = (int32_t)val; + if (low >= -16 && low < 16) { tcg_out32(s, VSPLTISW | VRT(ret) | ((val & 31) << 16)); return; } - } - if (have_isa_3_00 && val == (tcg_target_long)dup_const(MO_8, val)) { - tcg_out32(s, XXSPLTIB | VRT(ret) | ((val & 0xff) << 11)); - return; + break; } /* @@ -XXX,XX +XXX,XX @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, TCGReg ret, if (TCG_TARGET_REG_BITS == 64) { new_pool_label(s, val, rel, s->code_ptr, add); } else { - new_pool_l2(s, rel, s->code_ptr, add, val, val); + new_pool_l2(s, rel, s->code_ptr, add, val >> 32, val); } } else { load_insn = LVX | VRT(ret) | RB(TCG_REG_TMP1); if (TCG_TARGET_REG_BITS == 64) { new_pool_l2(s, rel, s->code_ptr, add, val, val); } else { - new_pool_l4(s, rel, s->code_ptr, add, val, val, val, val); + new_pool_l4(s, rel, s->code_ptr, add, + val >> 32, val, val >> 32, val); } } -- 2.25.1 The temp_fixed, temp_global, temp_local bits are all related. Combine them into a single enumeration. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 20 +++++--- tcg/optimize.c | 8 +-- tcg/tcg.c | 126 ++++++++++++++++++++++++++++------------------ 3 files changed, 92 insertions(+), 62 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 @@ typedef enum TCGTempVal { TEMP_VAL_CONST, } TCGTempVal; +typedef enum TCGTempKind { + /* Temp is dead at the end of all basic blocks. */ + TEMP_NORMAL, + /* Temp is saved across basic blocks but dead at the end of TBs. */ + TEMP_LOCAL, + /* Temp is saved across both basic blocks and translation blocks. */ + TEMP_GLOBAL, + /* Temp is in a fixed register. */ + TEMP_FIXED, +} TCGTempKind; + typedef struct TCGTemp { TCGReg reg:8; TCGTempVal val_type:8; TCGType base_type:8; TCGType type:8; - unsigned int fixed_reg:1; + TCGTempKind kind:3; unsigned int indirect_reg:1; unsigned int indirect_base:1; unsigned int mem_coherent:1; unsigned int mem_allocated:1; - /* If true, the temp is saved across both basic blocks and - translation blocks. */ - unsigned int temp_global:1; - /* If true, the temp is saved across basic blocks but dead - at the end of translation blocks. If false, the temp is - dead at the end of basic blocks. */ - unsigned int temp_local:1; unsigned int temp_allocated:1; tcg_target_long val; 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 TCGTemp *find_better_copy(TCGContext *s, TCGTemp *ts) TCGTemp *i; /* If this is already a global, we can't do better. */ - if (ts->temp_global) { + if (ts->kind >= TEMP_GLOBAL) { return ts; } /* Search for a global first. */ for (i = ts_info(ts)->next_copy; i != ts; i = ts_info(i)->next_copy) { - if (i->temp_global) { + if (i->kind >= TEMP_GLOBAL) { return i; } } /* If it is a temp, search for a temp local. */ - if (!ts->temp_local) { + if (ts->kind == TEMP_NORMAL) { for (i = ts_info(ts)->next_copy; i != ts; i = ts_info(i)->next_copy) { - if (ts->temp_local) { + if (i->kind >= TEMP_LOCAL) { return i; } } 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 inline TCGTemp *tcg_global_alloc(TCGContext *s) tcg_debug_assert(s->nb_globals == s->nb_temps); s->nb_globals++; ts = tcg_temp_alloc(s); - ts->temp_global = 1; + ts->kind = TEMP_GLOBAL; return ts; } @@ -XXX,XX +XXX,XX @@ static TCGTemp *tcg_global_reg_new_internal(TCGContext *s, TCGType type, ts = tcg_global_alloc(s); ts->base_type = type; ts->type = type; - ts->fixed_reg = 1; + ts->kind = TEMP_FIXED; ts->reg = reg; ts->name = name; tcg_regset_set_reg(s->reserved_regs, reg); @@ -XXX,XX +XXX,XX @@ TCGTemp *tcg_global_mem_new_internal(TCGType type, TCGv_ptr base, bigendian = 1; #endif - if (!base_ts->fixed_reg) { + if (base_ts->kind != TEMP_FIXED) { /* We do not support double-indirect registers. */ tcg_debug_assert(!base_ts->indirect_reg); base_ts->indirect_base = 1; @@ -XXX,XX +XXX,XX @@ TCGTemp *tcg_global_mem_new_internal(TCGType type, TCGv_ptr base, TCGTemp *tcg_temp_new_internal(TCGType type, bool temp_local) { TCGContext *s = tcg_ctx; + TCGTempKind kind = temp_local ? TEMP_LOCAL : TEMP_NORMAL; TCGTemp *ts; int idx, k; @@ -XXX,XX +XXX,XX @@ TCGTemp *tcg_temp_new_internal(TCGType type, bool temp_local) ts = &s->temps[idx]; ts->temp_allocated = 1; tcg_debug_assert(ts->base_type == type); - tcg_debug_assert(ts->temp_local == temp_local); + tcg_debug_assert(ts->kind == kind); } else { ts = tcg_temp_alloc(s); if (TCG_TARGET_REG_BITS == 32 && type == TCG_TYPE_I64) { @@ -XXX,XX +XXX,XX @@ TCGTemp *tcg_temp_new_internal(TCGType type, bool temp_local) ts->base_type = type; ts->type = TCG_TYPE_I32; ts->temp_allocated = 1; - ts->temp_local = temp_local; + ts->kind = kind; tcg_debug_assert(ts2 == ts + 1); ts2->base_type = TCG_TYPE_I64; ts2->type = TCG_TYPE_I32; ts2->temp_allocated = 1; - ts2->temp_local = temp_local; + ts2->kind = kind; } else { ts->base_type = type; ts->type = type; ts->temp_allocated = 1; - ts->temp_local = temp_local; + ts->kind = kind; } } @@ -XXX,XX +XXX,XX @@ void tcg_temp_free_internal(TCGTemp *ts) } #endif - tcg_debug_assert(ts->temp_global == 0); + tcg_debug_assert(ts->kind < TEMP_GLOBAL); tcg_debug_assert(ts->temp_allocated != 0); ts->temp_allocated = 0; idx = temp_idx(ts); - k = ts->base_type + (ts->temp_local ? TCG_TYPE_COUNT : 0); + k = ts->base_type + (ts->kind == TEMP_NORMAL ? 0 : TCG_TYPE_COUNT); set_bit(idx, s->free_temps[k].l); } @@ -XXX,XX +XXX,XX @@ void tcg_gen_callN(void *func, TCGTemp *ret, int nargs, TCGTemp **args) static void tcg_reg_alloc_start(TCGContext *s) { int i, n; - TCGTemp *ts; - for (i = 0, n = s->nb_globals; i < n; i++) { - ts = &s->temps[i]; - ts->val_type = (ts->fixed_reg ? TEMP_VAL_REG : TEMP_VAL_MEM); - } - for (n = s->nb_temps; i < n; i++) { - ts = &s->temps[i]; - ts->val_type = (ts->temp_local ? TEMP_VAL_MEM : TEMP_VAL_DEAD); - ts->mem_allocated = 0; - ts->fixed_reg = 0; + for (i = 0, n = s->nb_temps; i < n; i++) { + TCGTemp *ts = &s->temps[i]; + TCGTempVal val = TEMP_VAL_MEM; + + switch (ts->kind) { + case TEMP_FIXED: + val = TEMP_VAL_REG; + break; + case TEMP_GLOBAL: + break; + case TEMP_NORMAL: + val = TEMP_VAL_DEAD; + /* fall through */ + case TEMP_LOCAL: + ts->mem_allocated = 0; + break; + default: + g_assert_not_reached(); + } + ts->val_type = val; } memset(s->reg_to_temp, 0, sizeof(s->reg_to_temp)); @@ -XXX,XX +XXX,XX @@ static char *tcg_get_arg_str_ptr(TCGContext *s, char *buf, int buf_size, { int idx = temp_idx(ts); - if (ts->temp_global) { + switch (ts->kind) { + case TEMP_FIXED: + case TEMP_GLOBAL: pstrcpy(buf, buf_size, ts->name); - } else if (ts->temp_local) { + break; + case TEMP_LOCAL: snprintf(buf, buf_size, "loc%d", idx - s->nb_globals); - } else { + break; + case TEMP_NORMAL: snprintf(buf, buf_size, "tmp%d", idx - s->nb_globals); + break; } return buf; } @@ -XXX,XX +XXX,XX @@ static void la_bb_end(TCGContext *s, int ng, int nt) { int i; - for (i = 0; i < ng; ++i) { - s->temps[i].state = TS_DEAD | TS_MEM; - la_reset_pref(&s->temps[i]); - } - for (i = ng; i < nt; ++i) { - s->temps[i].state = (s->temps[i].temp_local - ? TS_DEAD | TS_MEM - : TS_DEAD); - la_reset_pref(&s->temps[i]); + for (i = 0; i < nt; ++i) { + TCGTemp *ts = &s->temps[i]; + int state; + + switch (ts->kind) { + case TEMP_FIXED: + case TEMP_GLOBAL: + case TEMP_LOCAL: + state = TS_DEAD | TS_MEM; + break; + case TEMP_NORMAL: + state = TS_DEAD; + break; + default: + g_assert_not_reached(); + } + ts->state = state; + la_reset_pref(ts); } } @@ -XXX,XX +XXX,XX @@ static void la_bb_sync(TCGContext *s, int ng, int nt) la_global_sync(s, ng); for (int i = ng; i < nt; ++i) { - if (s->temps[i].temp_local) { + if (s->temps[i].kind == TEMP_LOCAL) { int state = s->temps[i].state; s->temps[i].state = state | TS_MEM; if (state != TS_DEAD) { @@ -XXX,XX +XXX,XX @@ static void check_regs(TCGContext *s) } for (k = 0; k < s->nb_temps; k++) { ts = &s->temps[k]; - if (ts->val_type == TEMP_VAL_REG && !ts->fixed_reg + if (ts->val_type == TEMP_VAL_REG + && ts->kind != TEMP_FIXED && s->reg_to_temp[ts->reg] != ts) { printf("Inconsistency for temp %s:\n", tcg_get_arg_str_ptr(s, buf, sizeof(buf), ts)); @@ -XXX,XX +XXX,XX @@ static void temp_load(TCGContext *, TCGTemp *, TCGRegSet, TCGRegSet, TCGRegSet); mark it free; otherwise mark it dead. */ static void temp_free_or_dead(TCGContext *s, TCGTemp *ts, int free_or_dead) { - if (ts->fixed_reg) { + if (ts->kind == TEMP_FIXED) { return; } if (ts->val_type == TEMP_VAL_REG) { s->reg_to_temp[ts->reg] = NULL; } ts->val_type = (free_or_dead < 0 - || ts->temp_local - || ts->temp_global + || ts->kind != TEMP_NORMAL ? TEMP_VAL_MEM : TEMP_VAL_DEAD); } @@ -XXX,XX +XXX,XX @@ static inline void temp_dead(TCGContext *s, TCGTemp *ts) static void temp_sync(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs, TCGRegSet preferred_regs, int free_or_dead) { - if (ts->fixed_reg) { + if (ts->kind == TEMP_FIXED) { return; } if (!ts->mem_coherent) { @@ -XXX,XX +XXX,XX @@ static void temp_save(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs) { /* The liveness analysis already ensures that globals are back in memory. Keep an tcg_debug_assert for safety. */ - tcg_debug_assert(ts->val_type == TEMP_VAL_MEM || ts->fixed_reg); + tcg_debug_assert(ts->val_type == TEMP_VAL_MEM + || ts->kind == TEMP_FIXED); } /* save globals to their canonical location and assume they can be @@ -XXX,XX +XXX,XX @@ static void sync_globals(TCGContext *s, TCGRegSet allocated_regs) for (i = 0, n = s->nb_globals; i < n; i++) { TCGTemp *ts = &s->temps[i]; tcg_debug_assert(ts->val_type != TEMP_VAL_REG - || ts->fixed_reg + || ts->kind == TEMP_FIXED || ts->mem_coherent); } } @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_bb_end(TCGContext *s, TCGRegSet allocated_regs) for (i = s->nb_globals; i < s->nb_temps; i++) { TCGTemp *ts = &s->temps[i]; - if (ts->temp_local) { + if (ts->kind == TEMP_LOCAL) { temp_save(s, ts, allocated_regs); } else { /* The liveness analysis already ensures that temps are dead. @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_cbranch(TCGContext *s, TCGRegSet allocated_regs) * The liveness analysis already ensures that temps are dead. * Keep tcg_debug_asserts for safety. */ - if (ts->temp_local) { + if (ts->kind == TEMP_LOCAL) { tcg_debug_assert(ts->val_type != TEMP_VAL_REG || ts->mem_coherent); } else { tcg_debug_assert(ts->val_type == TEMP_VAL_DEAD); @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_do_movi(TCGContext *s, TCGTemp *ots, TCGRegSet preferred_regs) { /* ENV should not be modified. */ - tcg_debug_assert(!ots->fixed_reg); + tcg_debug_assert(ots->kind != TEMP_FIXED); /* The movi is not explicitly generated here. */ if (ots->val_type == TEMP_VAL_REG) { @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOp *op) ts = arg_temp(op->args[1]); /* ENV should not be modified. */ - tcg_debug_assert(!ots->fixed_reg); + tcg_debug_assert(ots->kind != TEMP_FIXED); /* Note that otype != itype for no-op truncation. */ otype = ots->type; @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOp *op) } temp_dead(s, ots); } else { - if (IS_DEAD_ARG(1) && !ts->fixed_reg) { + if (IS_DEAD_ARG(1) && ts->kind != TEMP_FIXED) { /* the mov can be suppressed */ if (ots->val_type == TEMP_VAL_REG) { s->reg_to_temp[ots->reg] = NULL; @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOp *op) * Store the source register into the destination slot * and leave the destination temp as TEMP_VAL_MEM. */ - assert(!ots->fixed_reg); + assert(ots->kind != TEMP_FIXED); if (!ts->mem_allocated) { temp_allocate_frame(s, ots); } @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_dup(TCGContext *s, const TCGOp *op) its = arg_temp(op->args[1]); /* ENV should not be modified. */ - tcg_debug_assert(!ots->fixed_reg); + tcg_debug_assert(ots->kind != TEMP_FIXED); itype = its->type; vece = TCGOP_VECE(op); @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) i_preferred_regs = o_preferred_regs = 0; if (arg_ct->ialias) { o_preferred_regs = op->output_pref[arg_ct->alias_index]; - if (ts->fixed_reg) { + if (ts->kind == TEMP_FIXED) { /* if fixed register, we must allocate a new register if the alias is not the same register */ if (arg != op->args[arg_ct->alias_index]) { @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) ts = arg_temp(arg); /* ENV should not be modified. */ - tcg_debug_assert(!ts->fixed_reg); + tcg_debug_assert(ts->kind != TEMP_FIXED); if (arg_ct->oalias && !const_args[arg_ct->alias_index]) { reg = new_args[arg_ct->alias_index]; @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) ts = arg_temp(op->args[i]); /* ENV should not be modified. */ - tcg_debug_assert(!ts->fixed_reg); + tcg_debug_assert(ts->kind != TEMP_FIXED); if (NEED_SYNC_ARG(i)) { temp_sync(s, ts, o_allocated_regs, 0, IS_DEAD_ARG(i)); @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_call(TCGContext *s, TCGOp *op) ts = arg_temp(arg); /* ENV should not be modified. */ - tcg_debug_assert(!ts->fixed_reg); + tcg_debug_assert(ts->kind != TEMP_FIXED); reg = tcg_target_call_oarg_regs[i]; tcg_debug_assert(s->reg_to_temp[reg] == NULL); -- 2.25.1 In most, but not all, places that we check for TEMP_FIXED, we are really testing that we do not modify the temporary. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 5 +++++ tcg/tcg.c | 21 ++++++++++----------- 2 files changed, 15 insertions(+), 11 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 { target_ulong gen_insn_data[TCG_MAX_INSNS][TARGET_INSN_START_WORDS]; }; +static inline bool temp_readonly(TCGTemp *ts) +{ + return ts->kind == TEMP_FIXED; +} + extern TCGContext tcg_init_ctx; extern __thread TCGContext *tcg_ctx; extern const void *tcg_code_gen_epilogue; 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 temp_load(TCGContext *, TCGTemp *, TCGRegSet, TCGRegSet, TCGRegSet); mark it free; otherwise mark it dead. */ static void temp_free_or_dead(TCGContext *s, TCGTemp *ts, int free_or_dead) { - if (ts->kind == TEMP_FIXED) { + if (temp_readonly(ts)) { return; } if (ts->val_type == TEMP_VAL_REG) { @@ -XXX,XX +XXX,XX @@ static inline void temp_dead(TCGContext *s, TCGTemp *ts) static void temp_sync(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs, TCGRegSet preferred_regs, int free_or_dead) { - if (ts->kind == TEMP_FIXED) { + if (temp_readonly(ts)) { return; } if (!ts->mem_coherent) { @@ -XXX,XX +XXX,XX @@ static void temp_save(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs) { /* The liveness analysis already ensures that globals are back in memory. Keep an tcg_debug_assert for safety. */ - tcg_debug_assert(ts->val_type == TEMP_VAL_MEM - || ts->kind == TEMP_FIXED); + tcg_debug_assert(ts->val_type == TEMP_VAL_MEM || temp_readonly(ts)); } /* save globals to their canonical location and assume they can be @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_do_movi(TCGContext *s, TCGTemp *ots, TCGRegSet preferred_regs) { /* ENV should not be modified. */ - tcg_debug_assert(ots->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ots)); /* The movi is not explicitly generated here. */ if (ots->val_type == TEMP_VAL_REG) { @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOp *op) ts = arg_temp(op->args[1]); /* ENV should not be modified. */ - tcg_debug_assert(ots->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ots)); /* Note that otype != itype for no-op truncation. */ otype = ots->type; @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOp *op) * Store the source register into the destination slot * and leave the destination temp as TEMP_VAL_MEM. */ - assert(ots->kind != TEMP_FIXED); + assert(!temp_readonly(ots)); if (!ts->mem_allocated) { temp_allocate_frame(s, ots); } @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_dup(TCGContext *s, const TCGOp *op) its = arg_temp(op->args[1]); /* ENV should not be modified. */ - tcg_debug_assert(ots->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ots)); itype = its->type; vece = TCGOP_VECE(op); @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) ts = arg_temp(arg); /* ENV should not be modified. */ - tcg_debug_assert(ts->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ts)); if (arg_ct->oalias && !const_args[arg_ct->alias_index]) { reg = new_args[arg_ct->alias_index]; @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) ts = arg_temp(op->args[i]); /* ENV should not be modified. */ - tcg_debug_assert(ts->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ts)); if (NEED_SYNC_ARG(i)) { temp_sync(s, ts, o_allocated_regs, 0, IS_DEAD_ARG(i)); @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_call(TCGContext *s, TCGOp *op) ts = arg_temp(arg); /* ENV should not be modified. */ - tcg_debug_assert(ts->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ts)); reg = tcg_target_call_oarg_regs[i]; tcg_debug_assert(s->reg_to_temp[reg] == NULL); -- 2.25.1 This will reduce the differences between 32-bit and 64-bit hosts, allowing full 64-bit constants to be created with the same interface. 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 @@ typedef struct TCGTemp { unsigned int mem_allocated:1; unsigned int temp_allocated:1; - tcg_target_long val; + int64_t val; struct TCGTemp *mem_base; intptr_t mem_offset; const char *name; 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 dump_regs(TCGContext *s) tcg_target_reg_names[ts->mem_base->reg]); break; case TEMP_VAL_CONST: - printf("$0x%" TCG_PRIlx, ts->val); + printf("$0x%" PRIx64, ts->val); break; case TEMP_VAL_DEAD: printf("D"); -- 2.25.1 Fix this name vs our coding style. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 32 ++++++++++++++++---------------- 1 file changed, 16 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 @@ glue(glue(case INDEX_op_, x), _i64): \ glue(glue(case INDEX_op_, x), _vec) -struct tcg_temp_info { +typedef struct TempOptInfo { bool is_const; TCGTemp *prev_copy; TCGTemp *next_copy; tcg_target_ulong val; tcg_target_ulong mask; -}; +} TempOptInfo; -static inline struct tcg_temp_info *ts_info(TCGTemp *ts) +static inline TempOptInfo *ts_info(TCGTemp *ts) { return ts->state_ptr; } -static inline struct tcg_temp_info *arg_info(TCGArg arg) +static inline TempOptInfo *arg_info(TCGArg arg) { return ts_info(arg_temp(arg)); } @@ -XXX,XX +XXX,XX @@ static inline bool ts_is_copy(TCGTemp *ts) /* Reset TEMP's state, possibly removing the temp for the list of copies. */ static void reset_ts(TCGTemp *ts) { - struct tcg_temp_info *ti = ts_info(ts); - struct tcg_temp_info *pi = ts_info(ti->prev_copy); - struct tcg_temp_info *ni = ts_info(ti->next_copy); + TempOptInfo *ti = ts_info(ts); + TempOptInfo *pi = ts_info(ti->prev_copy); + TempOptInfo *ni = ts_info(ti->next_copy); ni->prev_copy = ti->prev_copy; pi->next_copy = ti->next_copy; @@ -XXX,XX +XXX,XX @@ static void reset_temp(TCGArg arg) } /* Initialize and activate a temporary. */ -static void init_ts_info(struct tcg_temp_info *infos, +static void init_ts_info(TempOptInfo *infos, TCGTempSet *temps_used, TCGTemp *ts) { size_t idx = temp_idx(ts); if (!test_bit(idx, temps_used->l)) { - struct tcg_temp_info *ti = &infos[idx]; + TempOptInfo *ti = &infos[idx]; ts->state_ptr = ti; ti->next_copy = ts; @@ -XXX,XX +XXX,XX @@ static void init_ts_info(struct tcg_temp_info *infos, } } -static void init_arg_info(struct tcg_temp_info *infos, +static void init_arg_info(TempOptInfo *infos, TCGTempSet *temps_used, TCGArg arg) { init_ts_info(infos, temps_used, arg_temp(arg)); @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_movi(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg val) const TCGOpDef *def; TCGOpcode new_op; tcg_target_ulong mask; - struct tcg_temp_info *di = arg_info(dst); + TempOptInfo *di = arg_info(dst); def = &tcg_op_defs[op->opc]; if (def->flags & TCG_OPF_VECTOR) { @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) TCGTemp *dst_ts = arg_temp(dst); TCGTemp *src_ts = arg_temp(src); const TCGOpDef *def; - struct tcg_temp_info *di; - struct tcg_temp_info *si; + TempOptInfo *di; + TempOptInfo *si; tcg_target_ulong mask; TCGOpcode new_op; @@ -XXX,XX +XXX,XX @@ static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) di->mask = mask; if (src_ts->type == dst_ts->type) { - struct tcg_temp_info *ni = ts_info(si->next_copy); + TempOptInfo *ni = ts_info(si->next_copy); di->next_copy = si->next_copy; di->prev_copy = src_ts; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) { int nb_temps, nb_globals; TCGOp *op, *op_next, *prev_mb = NULL; - struct tcg_temp_info *infos; + TempOptInfo *infos; TCGTempSet temps_used; /* Array VALS has an element for each temp. @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) nb_temps = s->nb_temps; nb_globals = s->nb_globals; bitmap_zero(temps_used.l, nb_temps); - infos = tcg_malloc(sizeof(struct tcg_temp_info) * nb_temps); + infos = tcg_malloc(sizeof(TempOptInfo) * nb_temps); QTAILQ_FOREACH_SAFE(op, &s->ops, link, op_next) { tcg_target_ulong mask, partmask, affected; -- 2.25.1 This propagates the extended value of TCGTemp.val that we did before. In addition, it will be required for vector constants. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 40 +++++++++++++++++++++------------------- 1 file changed, 21 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 @@ typedef struct TempOptInfo { bool is_const; TCGTemp *prev_copy; TCGTemp *next_copy; - tcg_target_ulong val; - tcg_target_ulong mask; + uint64_t val; + uint64_t mask; } TempOptInfo; static inline TempOptInfo *ts_info(TCGTemp *ts) @@ -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_movi(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg val) +static void tcg_opt_gen_movi(TCGContext *s, TCGOp *op, TCGArg dst, uint64_t val) { const TCGOpDef *def; TCGOpcode new_op; - tcg_target_ulong mask; + uint64_t mask; TempOptInfo *di = arg_info(dst); def = &tcg_op_defs[op->opc]; @@ -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; - tcg_target_ulong mask; + uint64_t 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) } } -static TCGArg do_constant_folding_2(TCGOpcode op, TCGArg x, TCGArg y) +static uint64_t do_constant_folding_2(TCGOpcode op, uint64_t x, uint64_t y) { uint64_t l64, h64; @@ -XXX,XX +XXX,XX @@ static TCGArg do_constant_folding_2(TCGOpcode op, TCGArg x, TCGArg y) } } -static TCGArg do_constant_folding(TCGOpcode op, TCGArg x, TCGArg y) +static uint64_t do_constant_folding(TCGOpcode op, uint64_t x, uint64_t y) { const TCGOpDef *def = &tcg_op_defs[op]; - TCGArg res = do_constant_folding_2(op, x, y); + uint64_t res = do_constant_folding_2(op, x, y); if (!(def->flags & TCG_OPF_64BIT)) { res = (int32_t)res; } @@ -XXX,XX +XXX,XX @@ static bool do_constant_folding_cond_eq(TCGCond c) static TCGArg do_constant_folding_cond(TCGOpcode op, TCGArg x, TCGArg y, TCGCond c) { - tcg_target_ulong xv = arg_info(x)->val; - tcg_target_ulong yv = arg_info(y)->val; + 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)); @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) infos = tcg_malloc(sizeof(TempOptInfo) * nb_temps); QTAILQ_FOREACH_SAFE(op, &s->ops, link, op_next) { - tcg_target_ulong mask, partmask, affected; + uint64_t mask, partmask, affected, tmp; int nb_oargs, nb_iargs, i; - TCGArg tmp; TCGOpcode opc = op->opc; const TCGOpDef *def = &tcg_op_defs[opc]; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64(extract2): if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - TCGArg v1 = arg_info(op->args[1])->val; - TCGArg v2 = arg_info(op->args[2])->val; + 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 >> op->args[3]) | (v2 << (64 - op->args[3])); + tmp = (v1 >> shr) | (v2 << (64 - shr)); } else { - tmp = (int32_t)(((uint32_t)v1 >> op->args[3]) | - ((uint32_t)v2 << (32 - op->args[3]))); + tmp = (int32_t)(((uint32_t)v1 >> shr) | + ((uint32_t)v2 << (32 - shr))); } tcg_opt_gen_movi(s, op, op->args[0], tmp); break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; } if (arg_is_const(op->args[3]) && arg_is_const(op->args[4])) { - tcg_target_ulong tv = arg_info(op->args[3])->val; - tcg_target_ulong fv = arg_info(op->args[4])->val; + 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)) { -- 2.25.1 These will hold a single constant for the duration of the TB. They are hashed, so that each value has one temp across the TB. Not used yet, this is all infrastructure. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 24 ++++- tcg/optimize.c | 13 ++- tcg/tcg.c | 224 ++++++++++++++++++++++++++++++++++++---------- 3 files changed, 211 insertions(+), 50 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 @@ typedef enum TCGTempKind { TEMP_GLOBAL, /* Temp is in a fixed register. */ TEMP_FIXED, + /* Temp is a fixed constant. */ + TEMP_CONST, } TCGTempKind; typedef struct TCGTemp { @@ -XXX,XX +XXX,XX @@ struct TCGContext { QSIMPLEQ_HEAD(, TCGOp) plugin_ops; #endif + GHashTable *const_table[TCG_TYPE_COUNT]; TCGTempSet free_temps[TCG_TYPE_COUNT * 2]; TCGTemp temps[TCG_MAX_TEMPS]; /* globals first, temps after */ @@ -XXX,XX +XXX,XX @@ struct TCGContext { static inline bool temp_readonly(TCGTemp *ts) { - return ts->kind == TEMP_FIXED; + return ts->kind >= TEMP_FIXED; } extern TCGContext tcg_init_ctx; @@ -XXX,XX +XXX,XX @@ TCGOp *tcg_op_insert_after(TCGContext *s, TCGOp *op, TCGOpcode opc); void tcg_optimize(TCGContext *s); +/* Allocate a new temporary and initialize it with a constant. */ TCGv_i32 tcg_const_i32(int32_t val); TCGv_i64 tcg_const_i64(int64_t val); TCGv_i32 tcg_const_local_i32(int32_t val); @@ -XXX,XX +XXX,XX @@ TCGv_vec tcg_const_ones_vec(TCGType); TCGv_vec tcg_const_zeros_vec_matching(TCGv_vec); TCGv_vec tcg_const_ones_vec_matching(TCGv_vec); +/* + * Locate or create a read-only temporary that is a constant. + * This kind of temporary need not and should not be freed. + */ +TCGTemp *tcg_constant_internal(TCGType type, int64_t val); + +static inline TCGv_i32 tcg_constant_i32(int32_t val) +{ + return temp_tcgv_i32(tcg_constant_internal(TCG_TYPE_I32, val)); +} + +static inline TCGv_i64 tcg_constant_i64(int64_t val) +{ + return temp_tcgv_i64(tcg_constant_internal(TCG_TYPE_I64, val)); +} + +TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val); + #if UINTPTR_MAX == UINT32_MAX # define tcg_const_ptr(x) ((TCGv_ptr)tcg_const_i32((intptr_t)(x))) # define tcg_const_local_ptr(x) ((TCGv_ptr)tcg_const_local_i32((intptr_t)(x))) 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(TempOptInfo *infos, ts->state_ptr = ti; ti->next_copy = ts; ti->prev_copy = ts; - ti->is_const = false; - ti->mask = -1; + if (ts->kind == TEMP_CONST) { + ti->is_const = true; + ti->val = ti->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; + } + } else { + ti->is_const = false; + ti->mask = -1; + } set_bit(idx, temps_used->l); } } 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) /* No temps have been previously allocated for size or locality. */ memset(s->free_temps, 0, sizeof(s->free_temps)); + /* No constant temps have been previously allocated. */ + for (int i = 0; i < TCG_TYPE_COUNT; ++i) { + if (s->const_table[i]) { + g_hash_table_remove_all(s->const_table[i]); + } + } + s->nb_ops = 0; s->nb_labels = 0; s->current_frame_offset = s->frame_start; @@ -XXX,XX +XXX,XX @@ TCGTemp *tcg_global_mem_new_internal(TCGType type, TCGv_ptr base, bigendian = 1; #endif - if (base_ts->kind != TEMP_FIXED) { + switch (base_ts->kind) { + case TEMP_FIXED: + break; + case TEMP_GLOBAL: /* We do not support double-indirect registers. */ tcg_debug_assert(!base_ts->indirect_reg); base_ts->indirect_base = 1; s->nb_indirects += (TCG_TARGET_REG_BITS == 32 && type == TCG_TYPE_I64 ? 2 : 1); indirect_reg = 1; + break; + default: + g_assert_not_reached(); } if (TCG_TARGET_REG_BITS == 32 && type == TCG_TYPE_I64) { @@ -XXX,XX +XXX,XX @@ void tcg_temp_free_internal(TCGTemp *ts) TCGContext *s = tcg_ctx; int k, idx; + /* In order to simplify users of tcg_constant_*, silently ignore free. */ + if (ts->kind == TEMP_CONST) { + return; + } + #if defined(CONFIG_DEBUG_TCG) s->temps_in_use--; if (s->temps_in_use < 0) { @@ -XXX,XX +XXX,XX @@ void tcg_temp_free_internal(TCGTemp *ts) set_bit(idx, s->free_temps[k].l); } +TCGTemp *tcg_constant_internal(TCGType type, int64_t val) +{ + TCGContext *s = tcg_ctx; + GHashTable *h = s->const_table[type]; + TCGTemp *ts; + + if (h == NULL) { + h = g_hash_table_new(g_int64_hash, g_int64_equal); + s->const_table[type] = h; + } + + ts = g_hash_table_lookup(h, &val); + if (ts == NULL) { + ts = tcg_temp_alloc(s); + + if (TCG_TARGET_REG_BITS == 32 && type == TCG_TYPE_I64) { + TCGTemp *ts2 = tcg_temp_alloc(s); + + ts->base_type = TCG_TYPE_I64; + ts->type = TCG_TYPE_I32; + ts->kind = TEMP_CONST; + ts->temp_allocated = 1; + /* + * Retain the full value of the 64-bit constant in the low + * part, so that the hash table works. Actual uses will + * truncate the value to the low part. + */ + ts->val = val; + + tcg_debug_assert(ts2 == ts + 1); + ts2->base_type = TCG_TYPE_I64; + ts2->type = TCG_TYPE_I32; + ts2->kind = TEMP_CONST; + ts2->temp_allocated = 1; + ts2->val = val >> 32; + } else { + ts->base_type = type; + ts->type = type; + ts->kind = TEMP_CONST; + ts->temp_allocated = 1; + ts->val = val; + } + g_hash_table_insert(h, &ts->val, ts); + } + + return ts; +} + +TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val) +{ + val = dup_const(vece, val); + return temp_tcgv_vec(tcg_constant_internal(type, val)); +} + TCGv_i32 tcg_const_i32(int32_t val) { TCGv_i32 t0; @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_start(TCGContext *s) TCGTempVal val = TEMP_VAL_MEM; switch (ts->kind) { + case TEMP_CONST: + val = TEMP_VAL_CONST; + break; case TEMP_FIXED: val = TEMP_VAL_REG; break; @@ -XXX,XX +XXX,XX @@ static char *tcg_get_arg_str_ptr(TCGContext *s, char *buf, int buf_size, case TEMP_NORMAL: snprintf(buf, buf_size, "tmp%d", idx - s->nb_globals); break; + case TEMP_CONST: + switch (ts->type) { + case TCG_TYPE_I32: + snprintf(buf, buf_size, "$0x%x", (int32_t)ts->val); + break; +#if TCG_TARGET_REG_BITS > 32 + case TCG_TYPE_I64: + snprintf(buf, buf_size, "$0x%" PRIx64, ts->val); + break; +#endif + case TCG_TYPE_V64: + case TCG_TYPE_V128: + case TCG_TYPE_V256: + snprintf(buf, buf_size, "v%d$0x%" PRIx64, + 64 << (ts->type - TCG_TYPE_V64), ts->val); + break; + default: + g_assert_not_reached(); + } + break; } return buf; } @@ -XXX,XX +XXX,XX @@ static void la_bb_end(TCGContext *s, int ng, int nt) state = TS_DEAD | TS_MEM; break; case TEMP_NORMAL: + case TEMP_CONST: state = TS_DEAD; break; default: @@ -XXX,XX +XXX,XX @@ static void la_bb_sync(TCGContext *s, int ng, int nt) la_global_sync(s, ng); for (int i = ng; i < nt; ++i) { - if (s->temps[i].kind == TEMP_LOCAL) { - int state = s->temps[i].state; - s->temps[i].state = state | TS_MEM; + TCGTemp *ts = &s->temps[i]; + int state; + + switch (ts->kind) { + case TEMP_LOCAL: + state = ts->state; + ts->state = state | TS_MEM; if (state != TS_DEAD) { continue; } - } else { + break; + case TEMP_NORMAL: s->temps[i].state = TS_DEAD; + break; + case TEMP_CONST: + continue; + default: + g_assert_not_reached(); } la_reset_pref(&s->temps[i]); } @@ -XXX,XX +XXX,XX @@ static void temp_load(TCGContext *, TCGTemp *, TCGRegSet, TCGRegSet, TCGRegSet); mark it free; otherwise mark it dead. */ static void temp_free_or_dead(TCGContext *s, TCGTemp *ts, int free_or_dead) { - if (temp_readonly(ts)) { + TCGTempVal new_type; + + switch (ts->kind) { + case TEMP_FIXED: return; + case TEMP_GLOBAL: + case TEMP_LOCAL: + new_type = TEMP_VAL_MEM; + break; + case TEMP_NORMAL: + new_type = free_or_dead < 0 ? TEMP_VAL_MEM : TEMP_VAL_DEAD; + break; + case TEMP_CONST: + new_type = TEMP_VAL_CONST; + break; + default: + g_assert_not_reached(); } if (ts->val_type == TEMP_VAL_REG) { s->reg_to_temp[ts->reg] = NULL; } - ts->val_type = (free_or_dead < 0 - || ts->kind != TEMP_NORMAL - ? TEMP_VAL_MEM : TEMP_VAL_DEAD); + ts->val_type = new_type; } /* Mark a temporary as dead. */ @@ -XXX,XX +XXX,XX @@ static inline void temp_dead(TCGContext *s, TCGTemp *ts) static void temp_sync(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs, TCGRegSet preferred_regs, int free_or_dead) { - if (temp_readonly(ts)) { - return; - } - if (!ts->mem_coherent) { + if (!temp_readonly(ts) && !ts->mem_coherent) { if (!ts->mem_allocated) { temp_allocate_frame(s, ts); } @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_bb_end(TCGContext *s, TCGRegSet allocated_regs) for (i = s->nb_globals; i < s->nb_temps; i++) { TCGTemp *ts = &s->temps[i]; - if (ts->kind == TEMP_LOCAL) { + + switch (ts->kind) { + case TEMP_LOCAL: temp_save(s, ts, allocated_regs); - } else { + break; + case TEMP_NORMAL: /* The liveness analysis already ensures that temps are dead. Keep an tcg_debug_assert for safety. */ tcg_debug_assert(ts->val_type == TEMP_VAL_DEAD); + break; + case TEMP_CONST: + /* Similarly, we should have freed any allocated register. */ + tcg_debug_assert(ts->val_type == TEMP_VAL_CONST); + break; + default: + g_assert_not_reached(); } } @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_cbranch(TCGContext *s, TCGRegSet allocated_regs) * The liveness analysis already ensures that temps are dead. * Keep tcg_debug_asserts for safety. */ - if (ts->kind == TEMP_LOCAL) { + switch (ts->kind) { + case TEMP_LOCAL: tcg_debug_assert(ts->val_type != TEMP_VAL_REG || ts->mem_coherent); - } else { + break; + case TEMP_NORMAL: tcg_debug_assert(ts->val_type == TEMP_VAL_DEAD); + break; + case TEMP_CONST: + break; + default: + g_assert_not_reached(); } } } @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) i_preferred_regs = o_preferred_regs = 0; if (arg_ct->ialias) { o_preferred_regs = op->output_pref[arg_ct->alias_index]; - if (ts->kind == TEMP_FIXED) { - /* if fixed register, we must allocate a new register - if the alias is not the same register */ - if (arg != op->args[arg_ct->alias_index]) { - goto allocate_in_reg; - } - } else { - /* if the input is aliased to an output and if it is - not dead after the instruction, we must allocate - a new register and move it */ - if (!IS_DEAD_ARG(i)) { - goto allocate_in_reg; - } - /* check if the current register has already been allocated - for another input aliased to an output */ - if (ts->val_type == TEMP_VAL_REG) { - int k2, i2; - reg = ts->reg; - for (k2 = 0 ; k2 < k ; k2++) { - i2 = def->args_ct[nb_oargs + k2].sort_index; - if (def->args_ct[i2].ialias && reg == new_args[i2]) { - goto allocate_in_reg; - } + /* + * If the input is readonly, then it cannot also be an + * output and aliased to itself. If the input is not + * dead after the instruction, we must allocate a new + * register and move it. + */ + if (temp_readonly(ts) || !IS_DEAD_ARG(i)) { + goto allocate_in_reg; + } + + /* + * Check if the current register has already been allocated + * for another input aliased to an output. + */ + if (ts->val_type == TEMP_VAL_REG) { + reg = ts->reg; + for (int k2 = 0; k2 < k; k2++) { + int i2 = def->args_ct[nb_oargs + k2].sort_index; + if (def->args_ct[i2].ialias && reg == new_args[i2]) { + goto allocate_in_reg; } } - i_preferred_regs = o_preferred_regs; } + i_preferred_regs = o_preferred_regs; } temp_load(s, ts, arg_ct->regs, i_allocated_regs, i_preferred_regs); reg = ts->reg; - if (tcg_regset_test_reg(arg_ct->regs, reg)) { - /* nothing to do : the constraint is satisfied */ - } else { - allocate_in_reg: - /* allocate a new register matching the constraint - and move the temporary register into it */ + if (!tcg_regset_test_reg(arg_ct->regs, reg)) { + allocate_in_reg: + /* + * Allocate a new register matching the constraint + * and move the temporary register into it. + */ temp_load(s, ts, tcg_target_available_regs[ts->type], i_allocated_regs, 0); reg = tcg_reg_alloc(s, arg_ct->regs, i_allocated_regs, -- 2.25.1 Prefer TEMP_CONST over anything else. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 27 ++++++++++++--------------- 1 file changed, 12 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 void init_arg_info(TempOptInfo *infos, static TCGTemp *find_better_copy(TCGContext *s, TCGTemp *ts) { - TCGTemp *i; + TCGTemp *i, *g, *l; - /* If this is already a global, we can't do better. */ - if (ts->kind >= TEMP_GLOBAL) { + /* If this is already readonly, we can't do better. */ + if (temp_readonly(ts)) { return ts; } - /* Search for a global first. */ + g = l = NULL; for (i = ts_info(ts)->next_copy; i != ts; i = ts_info(i)->next_copy) { - if (i->kind >= TEMP_GLOBAL) { + if (temp_readonly(i)) { return i; - } - } - - /* If it is a temp, search for a temp local. */ - if (ts->kind == TEMP_NORMAL) { - for (i = ts_info(ts)->next_copy; i != ts; i = ts_info(i)->next_copy) { - if (i->kind >= TEMP_LOCAL) { - return i; + } else if (i->kind > ts->kind) { + if (i->kind == TEMP_GLOBAL) { + g = i; + } else if (i->kind == TEMP_LOCAL) { + l = i; } } } - /* Failure to find a better representation, return the same temp. */ - return ts; + /* If we didn't find a better representation, return the same temp. */ + return g ? g : l ? l : ts; } static bool ts_are_copies(TCGTemp *ts1, TCGTemp *ts2) -- 2.25.1 Do not allocate a large block for indexing. Instead, allocate for each temporary as they are seen. In general, this will use less memory, if we consider that most TBs do not touch every target register. This also allows us to allocate TempOptInfo for new temps created during optimization. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 60 ++++++++++++++++++++++++++++---------------------- 1 file changed, 34 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 void reset_temp(TCGArg arg) } /* Initialize and activate a temporary. */ -static void init_ts_info(TempOptInfo *infos, - TCGTempSet *temps_used, TCGTemp *ts) +static void init_ts_info(TCGTempSet *temps_used, TCGTemp *ts) { size_t idx = temp_idx(ts); - if (!test_bit(idx, temps_used->l)) { - TempOptInfo *ti = &infos[idx]; + TempOptInfo *ti; + if (test_bit(idx, temps_used->l)) { + return; + } + set_bit(idx, temps_used->l); + + ti = ts->state_ptr; + if (ti == NULL) { + ti = tcg_malloc(sizeof(TempOptInfo)); ts->state_ptr = ti; - ti->next_copy = ts; - ti->prev_copy = ts; - if (ts->kind == TEMP_CONST) { - ti->is_const = true; - ti->val = ti->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; - } - } else { - ti->is_const = false; - ti->mask = -1; + } + + ti->next_copy = ts; + ti->prev_copy = ts; + if (ts->kind == TEMP_CONST) { + ti->is_const = true; + ti->val = ts->val; + ti->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; } - set_bit(idx, temps_used->l); + } else { + ti->is_const = false; + ti->mask = -1; } } -static void init_arg_info(TempOptInfo *infos, - TCGTempSet *temps_used, TCGArg arg) +static void init_arg_info(TCGTempSet *temps_used, TCGArg arg) { - init_ts_info(infos, temps_used, arg_temp(arg)); + init_ts_info(temps_used, arg_temp(arg)); } static TCGTemp *find_better_copy(TCGContext *s, TCGTemp *ts) @@ -XXX,XX +XXX,XX @@ static bool swap_commutative2(TCGArg *p1, TCGArg *p2) /* Propagate constants and copies, fold constant expressions. */ void tcg_optimize(TCGContext *s) { - int nb_temps, nb_globals; + int nb_temps, nb_globals, i; TCGOp *op, *op_next, *prev_mb = NULL; - TempOptInfo *infos; TCGTempSet temps_used; /* Array VALS has an element for each temp. @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) nb_temps = s->nb_temps; nb_globals = s->nb_globals; + bitmap_zero(temps_used.l, nb_temps); - infos = tcg_malloc(sizeof(TempOptInfo) * nb_temps); + for (i = 0; i < nb_temps; ++i) { + s->temps[i].state_ptr = NULL; + } QTAILQ_FOREACH_SAFE(op, &s->ops, link, op_next) { uint64_t mask, partmask, affected, tmp; - int nb_oargs, nb_iargs, i; + 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) for (i = 0; i < nb_oargs + nb_iargs; i++) { TCGTemp *ts = arg_temp(op->args[i]); if (ts) { - init_ts_info(infos, &temps_used, ts); + init_ts_info(&temps_used, ts); } } } else { nb_oargs = def->nb_oargs; nb_iargs = def->nb_iargs; for (i = 0; i < nb_oargs + nb_iargs; i++) { - init_arg_info(infos, &temps_used, op->args[i]); + init_arg_info(&temps_used, op->args[i]); } } -- 2.25.1 Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 108 ++++++++++++++++++++++--------------------------- 1 file changed, 49 insertions(+), 59 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_movi(TCGContext *s, TCGOp *op, TCGArg dst, uint64_t val) -{ - const TCGOpDef *def; - TCGOpcode new_op; - uint64_t mask; - TempOptInfo *di = arg_info(dst); - - def = &tcg_op_defs[op->opc]; - if (def->flags & TCG_OPF_VECTOR) { - new_op = INDEX_op_dupi_vec; - } else if (def->flags & TCG_OPF_64BIT) { - new_op = INDEX_op_movi_i64; - } else { - new_op = INDEX_op_movi_i32; - } - op->opc = new_op; - /* TCGOP_VECL and TCGOP_VECE remain unchanged. */ - op->args[0] = dst; - op->args[1] = val; - - reset_temp(dst); - di->is_const = true; - di->val = val; - mask = val; - if (TCG_TARGET_REG_BITS > 32 && new_op == INDEX_op_movi_i32) { - /* High bits of the destination are now garbage. */ - mask |= ~0xffffffffull; - } - di->mask = mask; -} - static void tcg_opt_gen_mov(TCGContext *s, TCGOp *op, TCGArg dst, TCGArg src) { TCGTemp *dst_ts = arg_temp(dst); @@ -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, + 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); + init_ts_info(temps_used, tv); + tcg_opt_gen_mov(s, op, dst, temp_arg(tv)); +} + static uint64_t do_constant_folding_2(TCGOpcode op, uint64_t x, uint64_t y) { uint64_t l64, h64; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) nb_temps = s->nb_temps; nb_globals = s->nb_globals; - bitmap_zero(temps_used.l, nb_temps); + 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) CASE_OP_32_64(rotr): if (arg_is_const(op->args[1]) && arg_info(op->args[1])->val == 0) { - tcg_opt_gen_movi(s, op, op->args[0], 0); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], 0); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], 0); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], 0); + tcg_opt_gen_movi(s, &temps_used, op, op->args[0], 0); continue; } break; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) break; CASE_OP_32_64(movi): case INDEX_op_dupi_vec: - tcg_opt_gen_movi(s, op, op->args[0], op->args[1]); + tcg_opt_gen_movi(s, &temps_used, 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(s, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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 = arg_info(op->args[1])->val; if (tmp == arg_info(op->args[2])->val) { - tcg_opt_gen_movi(s, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); break; } } else if (args_are_copies(op->args[1], op->args[2])) { @@ -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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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) { - bitmap_zero(temps_used.l, nb_temps); + memset(&temps_used, 0, sizeof(temps_used)); op->opc = INDEX_op_br; op->args[0] = op->args[3]; } else { @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) 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_movi_i32); + TCGOp *op2 = tcg_op_insert_before(s, op, INDEX_op_mov_i32); if (opc == INDEX_op_add2_i32) { a += b; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(s, op, rl, (int32_t)a); - tcg_opt_gen_movi(s, op2, rh, (int32_t)(a >> 32)); + 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)); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) 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_movi_i32); + TCGOp *op2 = tcg_op_insert_before(s, op, INDEX_op_mov_i32); rl = op->args[0]; rh = op->args[1]; - tcg_opt_gen_movi(s, op, rl, (int32_t)r); - tcg_opt_gen_movi(s, op2, rh, (int32_t)(r >> 32)); + 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)); break; } goto do_default; @@ -XXX,XX +XXX,XX @@ void tcg_optimize(TCGContext *s) if (tmp != 2) { if (tmp) { do_brcond_true: - bitmap_zero(temps_used.l, nb_temps); + memset(&temps_used, 0, sizeof(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: - bitmap_zero(temps_used.l, nb_temps); + memset(&temps_used, 0, sizeof(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: - bitmap_zero(temps_used.l, nb_temps); + memset(&temps_used, 0, sizeof(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, op, op->args[0], tmp); + tcg_opt_gen_movi(s, &temps_used, 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) block, otherwise we only trash the output args. "mask" is the non-zero bits mask for the first output arg. */ if (def->flags & TCG_OPF_BB_END) { - bitmap_zero(temps_used.l, nb_temps); + memset(&temps_used, 0, sizeof(temps_used)); } else { do_reset_output: for (i = 0; i < nb_oargs; i++) { -- 2.25.1 Because we now store uint64_t in TCGTemp, we can now always store the full 64-bit duplicate immediate. So remove the difference between 32- and 64-bit hosts. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/optimize.c | 9 ++++----- tcg/tcg-op-vec.c | 39 ++++++++++----------------------------- tcg/tcg.c | 7 +------ 3 files changed, 15 insertions(+), 40 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) case INDEX_op_dup2_vec: assert(TCG_TARGET_REG_BITS == 32); if (arg_is_const(op->args[1]) && arg_is_const(op->args[2])) { - tmp = arg_info(op->args[1])->val; - if (tmp == arg_info(op->args[2])->val) { - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], tmp); - break; - } + tcg_opt_gen_movi(s, &temps_used, op, op->args[0], + deposit64(arg_info(op->args[1])->val, 32, 32, + arg_info(op->args[2])->val)); + break; } else if (args_are_copies(op->args[1], op->args[2])) { op->opc = INDEX_op_dup_vec; TCGOP_VECE(op) = MO_32; diff --git a/tcg/tcg-op-vec.c b/tcg/tcg-op-vec.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-vec.c +++ b/tcg/tcg-op-vec.c @@ -XXX,XX +XXX,XX @@ void tcg_gen_mov_vec(TCGv_vec r, TCGv_vec a) } } -#define MO_REG (TCG_TARGET_REG_BITS == 64 ? MO_64 : MO_32) - -static void do_dupi_vec(TCGv_vec r, unsigned vece, TCGArg a) -{ - TCGTemp *rt = tcgv_vec_temp(r); - vec_gen_2(INDEX_op_dupi_vec, rt->base_type, vece, temp_arg(rt), a); -} - TCGv_vec tcg_const_zeros_vec(TCGType type) { TCGv_vec ret = tcg_temp_new_vec(type); - do_dupi_vec(ret, MO_REG, 0); + tcg_gen_dupi_vec(MO_64, ret, 0); return ret; } TCGv_vec tcg_const_ones_vec(TCGType type) { TCGv_vec ret = tcg_temp_new_vec(type); - do_dupi_vec(ret, MO_REG, -1); + tcg_gen_dupi_vec(MO_64, ret, -1); return ret; } @@ -XXX,XX +XXX,XX @@ TCGv_vec tcg_const_ones_vec_matching(TCGv_vec m) void tcg_gen_dup64i_vec(TCGv_vec r, uint64_t a) { - if (TCG_TARGET_REG_BITS == 64) { - do_dupi_vec(r, MO_64, a); - } else if (a == dup_const(MO_32, a)) { - do_dupi_vec(r, MO_32, a); - } else { - TCGv_i64 c = tcg_const_i64(a); - tcg_gen_dup_i64_vec(MO_64, r, c); - tcg_temp_free_i64(c); - } + tcg_gen_dupi_vec(MO_64, r, a); } void tcg_gen_dup32i_vec(TCGv_vec r, uint32_t a) { - do_dupi_vec(r, MO_REG, dup_const(MO_32, a)); + tcg_gen_dupi_vec(MO_32, r, a); } void tcg_gen_dup16i_vec(TCGv_vec r, uint32_t a) { - do_dupi_vec(r, MO_REG, dup_const(MO_16, a)); + tcg_gen_dupi_vec(MO_16, r, a); } void tcg_gen_dup8i_vec(TCGv_vec r, uint32_t a) { - do_dupi_vec(r, MO_REG, dup_const(MO_8, a)); + tcg_gen_dupi_vec(MO_8, r, a); } void tcg_gen_dupi_vec(unsigned vece, TCGv_vec r, uint64_t a) { - if (vece == MO_64) { - tcg_gen_dup64i_vec(r, a); - } else { - do_dupi_vec(r, MO_REG, dup_const(vece, a)); - } + TCGTemp *rt = tcgv_vec_temp(r); + tcg_gen_mov_vec(r, tcg_constant_vec(rt->base_type, vece, a)); } void tcg_gen_dup_i64_vec(unsigned vece, TCGv_vec r, TCGv_i64 a) @@ -XXX,XX +XXX,XX @@ void tcg_gen_abs_vec(unsigned vece, TCGv_vec r, TCGv_vec a) if (tcg_can_emit_vec_op(INDEX_op_sari_vec, type, vece) > 0) { tcg_gen_sari_vec(vece, t, a, (8 << vece) - 1); } else { - do_dupi_vec(t, MO_REG, 0); - tcg_gen_cmp_vec(TCG_COND_LT, vece, t, a, t); + tcg_gen_cmp_vec(TCG_COND_LT, vece, t, a, + tcg_constant_vec(type, vece, 0)); } tcg_gen_xor_vec(vece, r, a, t); tcg_gen_sub_vec(vece, r, r, 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 @@ static void temp_load(TCGContext *s, TCGTemp *ts, TCGRegSet desired_regs, * The targets will, in general, have to do this search anyway, * do this generically. */ - if (TCG_TARGET_REG_BITS == 32) { - val = dup_const(MO_32, val); - vece = MO_32; - } if (val == dup_const(MO_8, val)) { vece = MO_8; } else if (val == dup_const(MO_16, val)) { vece = MO_16; - } else if (TCG_TARGET_REG_BITS == 64 && - val == dup_const(MO_32, val)) { + } else if (val == dup_const(MO_32, val)) { vece = MO_32; } -- 2.25.1 We must do this before we adjust tcg_out_movi_i32, lest the under-the-hood poking that we do for icount be broken. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/gen-icount.h | 25 +++++++++++++------------ 1 file changed, 13 insertions(+), 12 deletions(-) diff --git a/include/exec/gen-icount.h b/include/exec/gen-icount.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/gen-icount.h +++ b/include/exec/gen-icount.h @@ -XXX,XX +XXX,XX @@ static inline void gen_io_end(void) static inline void gen_tb_start(const TranslationBlock *tb) { - TCGv_i32 count, imm; + TCGv_i32 count; tcg_ctx->exitreq_label = gen_new_label(); if (tb_cflags(tb) & CF_USE_ICOUNT) { @@ -XXX,XX +XXX,XX @@ static inline void gen_tb_start(const TranslationBlock *tb) offsetof(ArchCPU, env)); if (tb_cflags(tb) & CF_USE_ICOUNT) { - imm = tcg_temp_new_i32(); - /* We emit a movi with a dummy immediate argument. Keep the insn index - * of the movi so that we later (when we know the actual insn count) - * can update the immediate argument with the actual insn count. */ - tcg_gen_movi_i32(imm, 0xdeadbeef); + /* + * We emit a sub with a dummy immediate argument. Keep the insn index + * of the sub so that we later (when we know the actual insn count) + * can update the argument with the actual insn count. + */ + tcg_gen_sub_i32(count, count, tcg_constant_i32(0)); icount_start_insn = tcg_last_op(); - - tcg_gen_sub_i32(count, count, imm); - tcg_temp_free_i32(imm); } tcg_gen_brcondi_i32(TCG_COND_LT, count, 0, tcg_ctx->exitreq_label); @@ -XXX,XX +XXX,XX @@ static inline void gen_tb_start(const TranslationBlock *tb) static inline void gen_tb_end(const TranslationBlock *tb, int num_insns) { if (tb_cflags(tb) & CF_USE_ICOUNT) { - /* Update the num_insn immediate parameter now that we know - * the actual insn count. */ - tcg_set_insn_param(icount_start_insn, 1, num_insns); + /* + * Update the num_insn immediate parameter now that we know + * the actual insn count. + */ + tcg_set_insn_param(icount_start_insn, 2, + tcgv_i32_arg(tcg_constant_i32(num_insns))); } gen_set_label(tcg_ctx->exitreq_label); -- 2.25.1 Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-op.h | 13 +-- tcg/tcg-op.c | 227 ++++++++++++++++++++----------------------- 2 files changed, 109 insertions(+), 131 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 @@ void tcg_gen_mb(TCGBar); /* 32 bit ops */ +void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg); void tcg_gen_addi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2); void tcg_gen_subfi_i32(TCGv_i32 ret, int32_t arg1, TCGv_i32 arg2); void tcg_gen_subi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2); @@ -XXX,XX +XXX,XX @@ static inline void tcg_gen_mov_i32(TCGv_i32 ret, TCGv_i32 arg) } } -static inline void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg) -{ - tcg_gen_op2i_i32(INDEX_op_movi_i32, ret, arg); -} - static inline void tcg_gen_ld8u_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset) { @@ -XXX,XX +XXX,XX @@ static inline void tcg_gen_not_i32(TCGv_i32 ret, TCGv_i32 arg) /* 64 bit ops */ +void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg); void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2); void tcg_gen_subfi_i64(TCGv_i64 ret, int64_t arg1, TCGv_i64 arg2); void tcg_gen_subi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2); @@ -XXX,XX +XXX,XX @@ static inline void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg) } } -static inline void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg) -{ - tcg_gen_op2i_i64(INDEX_op_movi_i64, ret, arg); -} - static inline void tcg_gen_ld8u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset) { @@ -XXX,XX +XXX,XX @@ static inline void tcg_gen_sub_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2) void tcg_gen_discard_i64(TCGv_i64 arg); void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg); -void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg); void tcg_gen_ld8u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset); void tcg_gen_ld8s_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset); void tcg_gen_ld16u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset); 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_mb(TCGBar mb_type) /* 32 bit ops */ +void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg) +{ + tcg_gen_mov_i32(ret, tcg_constant_i32(arg)); +} + void tcg_gen_addi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) { /* some cases can be optimized here */ if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_add_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_add_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_subfi_i32(TCGv_i32 ret, int32_t arg1, TCGv_i32 arg2) /* Don't recurse with tcg_gen_neg_i32. */ tcg_gen_op2_i32(INDEX_op_neg_i32, ret, arg2); } else { - TCGv_i32 t0 = tcg_const_i32(arg1); - tcg_gen_sub_i32(ret, t0, arg2); - tcg_temp_free_i32(t0); + tcg_gen_sub_i32(ret, tcg_constant_i32(arg1), arg2); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_subi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_sub_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_sub_i32(ret, arg1, tcg_constant_i32(arg2)); } } void tcg_gen_andi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) { - TCGv_i32 t0; /* Some cases can be optimized here. */ switch (arg2) { case 0: @@ -XXX,XX +XXX,XX @@ void tcg_gen_andi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) } break; } - t0 = tcg_const_i32(arg2); - tcg_gen_and_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + + tcg_gen_and_i32(ret, arg1, tcg_constant_i32(arg2)); } void tcg_gen_ori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) @@ -XXX,XX +XXX,XX @@ void tcg_gen_ori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_or_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_or_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_xori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) /* Don't recurse with tcg_gen_not_i32. */ tcg_gen_op2_i32(INDEX_op_not_i32, ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_xor_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_xor_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_shli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_shl_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_shl_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_shri_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_shr_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_shr_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_sari_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_sar_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_sar_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_brcondi_i32(TCGCond cond, TCGv_i32 arg1, int32_t arg2, TCGLabel *l) if (cond == TCG_COND_ALWAYS) { tcg_gen_br(l); } else if (cond != TCG_COND_NEVER) { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_brcond_i32(cond, arg1, t0, l); - tcg_temp_free_i32(t0); + tcg_gen_brcond_i32(cond, arg1, tcg_constant_i32(arg2), l); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_setcond_i32(TCGCond cond, TCGv_i32 ret, void tcg_gen_setcondi_i32(TCGCond cond, TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_setcond_i32(cond, ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_setcond_i32(cond, ret, arg1, tcg_constant_i32(arg2)); } void tcg_gen_muli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) @@ -XXX,XX +XXX,XX @@ void tcg_gen_muli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) } else if (is_power_of_2(arg2)) { tcg_gen_shli_i32(ret, arg1, ctz32(arg2)); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_mul_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_mul_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_clz_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2) void tcg_gen_clzi_i32(TCGv_i32 ret, TCGv_i32 arg1, uint32_t arg2) { - TCGv_i32 t = tcg_const_i32(arg2); - tcg_gen_clz_i32(ret, arg1, t); - tcg_temp_free_i32(t); + tcg_gen_clz_i32(ret, arg1, tcg_constant_i32(arg2)); } void tcg_gen_ctz_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2) @@ -XXX,XX +XXX,XX @@ void tcg_gen_ctz_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2) tcg_gen_clzi_i32(t, t, 32); tcg_gen_xori_i32(t, t, 31); } - z = tcg_const_i32(0); + z = tcg_constant_i32(0); tcg_gen_movcond_i32(TCG_COND_EQ, ret, arg1, z, arg2, t); tcg_temp_free_i32(t); - tcg_temp_free_i32(z); } else { gen_helper_ctz_i32(ret, arg1, arg2); } @@ -XXX,XX +XXX,XX @@ void tcg_gen_ctzi_i32(TCGv_i32 ret, TCGv_i32 arg1, uint32_t arg2) tcg_gen_ctpop_i32(ret, t); tcg_temp_free_i32(t); } else { - TCGv_i32 t = tcg_const_i32(arg2); - tcg_gen_ctz_i32(ret, arg1, t); - tcg_temp_free_i32(t); + tcg_gen_ctz_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_rotli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else if (TCG_TARGET_HAS_rot_i32) { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_rotl_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_rotl_i32(ret, arg1, tcg_constant_i32(arg2)); } else { TCGv_i32 t0, t1; t0 = tcg_temp_new_i32(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_deposit_z_i32(TCGv_i32 ret, TCGv_i32 arg, tcg_gen_andi_i32(ret, arg, (1u << len) - 1); } else if (TCG_TARGET_HAS_deposit_i32 && TCG_TARGET_deposit_i32_valid(ofs, len)) { - TCGv_i32 zero = tcg_const_i32(0); + TCGv_i32 zero = tcg_constant_i32(0); tcg_gen_op5ii_i32(INDEX_op_deposit_i32, ret, zero, arg, ofs, len); - tcg_temp_free_i32(zero); } else { /* To help two-operand hosts we prefer to zero-extend first, which allows ARG to stay live. */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_bswap32_i32(TCGv_i32 ret, TCGv_i32 arg) } else { TCGv_i32 t0 = tcg_temp_new_i32(); TCGv_i32 t1 = tcg_temp_new_i32(); - TCGv_i32 t2 = tcg_const_i32(0x00ff00ff); + TCGv_i32 t2 = tcg_constant_i32(0x00ff00ff); /* arg = abcd */ tcg_gen_shri_i32(t0, arg, 8); /* t0 = .abc */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_bswap32_i32(TCGv_i32 ret, TCGv_i32 arg) tcg_temp_free_i32(t0); tcg_temp_free_i32(t1); - tcg_temp_free_i32(t2); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_discard_i64(TCGv_i64 arg) void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg) { - tcg_gen_mov_i32(TCGV_LOW(ret), TCGV_LOW(arg)); - tcg_gen_mov_i32(TCGV_HIGH(ret), TCGV_HIGH(arg)); + TCGTemp *ts = tcgv_i64_temp(arg); + + /* Canonicalize TCGv_i64 TEMP_CONST into TCGv_i32 TEMP_CONST. */ + if (ts->kind == TEMP_CONST) { + tcg_gen_movi_i64(ret, ts->val); + } else { + tcg_gen_mov_i32(TCGV_LOW(ret), TCGV_LOW(arg)); + tcg_gen_mov_i32(TCGV_HIGH(ret), TCGV_HIGH(arg)); + } } void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg) @@ -XXX,XX +XXX,XX @@ void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2) tcg_temp_free_i64(t0); tcg_temp_free_i32(t1); } + +#else + +void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg) +{ + tcg_gen_mov_i64(ret, tcg_constant_i64(arg)); +} + #endif /* TCG_TARGET_REG_SIZE == 32 */ void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) @@ -XXX,XX +XXX,XX @@ void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) /* some cases can be optimized here */ if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); + } else if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_add_i64(ret, arg1, tcg_constant_i64(arg2)); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_add_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_add2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), tcg_constant_i32(arg2 >> 32)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_subfi_i64(TCGv_i64 ret, int64_t arg1, TCGv_i64 arg2) if (arg1 == 0 && TCG_TARGET_HAS_neg_i64) { /* Don't recurse with tcg_gen_neg_i64. */ tcg_gen_op2_i64(INDEX_op_neg_i64, ret, arg2); + } else if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_sub_i64(ret, tcg_constant_i64(arg1), arg2); } else { - TCGv_i64 t0 = tcg_const_i64(arg1); - tcg_gen_sub_i64(ret, t0, arg2); - tcg_temp_free_i64(t0); + tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), + tcg_constant_i32(arg1), tcg_constant_i32(arg1 >> 32), + TCGV_LOW(arg2), TCGV_HIGH(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_subi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) /* some cases can be optimized here */ if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); + } else if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_sub_i64(ret, arg1, tcg_constant_i64(arg2)); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_sub_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), tcg_constant_i32(arg2 >> 32)); } } void tcg_gen_andi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) { - TCGv_i64 t0; - if (TCG_TARGET_REG_BITS == 32) { tcg_gen_andi_i32(TCGV_LOW(ret), TCGV_LOW(arg1), arg2); tcg_gen_andi_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), arg2 >> 32); @@ -XXX,XX +XXX,XX @@ void tcg_gen_andi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } break; } - t0 = tcg_const_i64(arg2); - tcg_gen_and_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + + tcg_gen_and_i64(ret, arg1, tcg_constant_i64(arg2)); } void tcg_gen_ori_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) @@ -XXX,XX +XXX,XX @@ void tcg_gen_ori_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_or_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_or_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_xori_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) /* Don't recurse with tcg_gen_not_i64. */ tcg_gen_op2_i64(INDEX_op_not_i64, ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_xor_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_xor_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_shli_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_shl_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_shl_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_shri_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_shr_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_shr_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_sari_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_sar_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_sar_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_brcond_i64(TCGCond cond, TCGv_i64 arg1, TCGv_i64 arg2, TCGLabel *l) void tcg_gen_brcondi_i64(TCGCond cond, TCGv_i64 arg1, int64_t arg2, TCGLabel *l) { - if (cond == TCG_COND_ALWAYS) { + if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_brcond_i64(cond, arg1, tcg_constant_i64(arg2), l); + } else if (cond == TCG_COND_ALWAYS) { tcg_gen_br(l); } else if (cond != TCG_COND_NEVER) { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_brcond_i64(cond, arg1, t0, l); - tcg_temp_free_i64(t0); + l->refs++; + tcg_gen_op6ii_i32(INDEX_op_brcond2_i32, + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), + tcg_constant_i32(arg2 >> 32), + cond, label_arg(l)); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_setcond_i64(TCGCond cond, TCGv_i64 ret, void tcg_gen_setcondi_i64(TCGCond cond, TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_setcond_i64(cond, ret, arg1, t0); - tcg_temp_free_i64(t0); + if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_setcond_i64(cond, ret, arg1, tcg_constant_i64(arg2)); + } else if (cond == TCG_COND_ALWAYS) { + tcg_gen_movi_i64(ret, 1); + } else if (cond == TCG_COND_NEVER) { + tcg_gen_movi_i64(ret, 0); + } else { + tcg_gen_op6i_i32(INDEX_op_setcond2_i32, TCGV_LOW(ret), + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), + tcg_constant_i32(arg2 >> 32), cond); + tcg_gen_movi_i32(TCGV_HIGH(ret), 0); + } } void tcg_gen_muli_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) @@ -XXX,XX +XXX,XX @@ void tcg_gen_bswap32_i64(TCGv_i64 ret, TCGv_i64 arg) } else { TCGv_i64 t0 = tcg_temp_new_i64(); TCGv_i64 t1 = tcg_temp_new_i64(); - TCGv_i64 t2 = tcg_const_i64(0x00ff00ff); + TCGv_i64 t2 = tcg_constant_i64(0x00ff00ff); /* arg = ....abcd */ tcg_gen_shri_i64(t0, arg, 8); /* t0 = .....abc */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_bswap32_i64(TCGv_i64 ret, TCGv_i64 arg) tcg_temp_free_i64(t0); tcg_temp_free_i64(t1); - tcg_temp_free_i64(t2); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_clzi_i64(TCGv_i64 ret, TCGv_i64 arg1, uint64_t arg2) if (TCG_TARGET_REG_BITS == 32 && TCG_TARGET_HAS_clz_i32 && arg2 <= 0xffffffffu) { - TCGv_i32 t = tcg_const_i32((uint32_t)arg2 - 32); - tcg_gen_clz_i32(t, TCGV_LOW(arg1), t); + TCGv_i32 t = tcg_temp_new_i32(); + tcg_gen_clzi_i32(t, TCGV_LOW(arg1), arg2 - 32); tcg_gen_addi_i32(t, t, 32); tcg_gen_clz_i32(TCGV_LOW(ret), TCGV_HIGH(arg1), t); tcg_gen_movi_i32(TCGV_HIGH(ret), 0); tcg_temp_free_i32(t); } else { - TCGv_i64 t = tcg_const_i64(arg2); - tcg_gen_clz_i64(ret, arg1, t); - tcg_temp_free_i64(t); + TCGv_i64 t0 = tcg_const_i64(arg2); + tcg_gen_clz_i64(ret, arg1, t0); + tcg_temp_free_i64(t0); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_ctz_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2) tcg_gen_clzi_i64(t, t, 64); tcg_gen_xori_i64(t, t, 63); } - z = tcg_const_i64(0); + z = tcg_constant_i64(0); tcg_gen_movcond_i64(TCG_COND_EQ, ret, arg1, z, arg2, t); tcg_temp_free_i64(t); tcg_temp_free_i64(z); @@ -XXX,XX +XXX,XX @@ void tcg_gen_ctzi_i64(TCGv_i64 ret, TCGv_i64 arg1, uint64_t arg2) if (TCG_TARGET_REG_BITS == 32 && TCG_TARGET_HAS_ctz_i32 && arg2 <= 0xffffffffu) { - TCGv_i32 t32 = tcg_const_i32((uint32_t)arg2 - 32); - tcg_gen_ctz_i32(t32, TCGV_HIGH(arg1), t32); + TCGv_i32 t32 = tcg_temp_new_i32(); + tcg_gen_ctzi_i32(t32, TCGV_HIGH(arg1), arg2 - 32); tcg_gen_addi_i32(t32, t32, 32); tcg_gen_ctz_i32(TCGV_LOW(ret), TCGV_LOW(arg1), t32); tcg_gen_movi_i32(TCGV_HIGH(ret), 0); @@ -XXX,XX +XXX,XX @@ void tcg_gen_ctzi_i64(TCGv_i64 ret, TCGv_i64 arg1, uint64_t arg2) tcg_gen_ctpop_i64(ret, t); tcg_temp_free_i64(t); } else { - TCGv_i64 t64 = tcg_const_i64(arg2); - tcg_gen_ctz_i64(ret, arg1, t64); - tcg_temp_free_i64(t64); + TCGv_i64 t0 = tcg_const_i64(arg2); + tcg_gen_ctz_i64(ret, arg1, t0); + tcg_temp_free_i64(t0); } } @@ -XXX,XX +XXX,XX @@ void tcg_gen_rotli_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else if (TCG_TARGET_HAS_rot_i64) { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_rotl_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_rotl_i64(ret, arg1, tcg_constant_i64(arg2)); } else { TCGv_i64 t0, t1; t0 = tcg_temp_new_i64(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_deposit_z_i64(TCGv_i64 ret, TCGv_i64 arg, tcg_gen_andi_i64(ret, arg, (1ull << len) - 1); } else if (TCG_TARGET_HAS_deposit_i64 && TCG_TARGET_deposit_i64_valid(ofs, len)) { - TCGv_i64 zero = tcg_const_i64(0); + TCGv_i64 zero = tcg_constant_i64(0); tcg_gen_op5ii_i64(INDEX_op_deposit_i64, ret, zero, arg, ofs, len); - tcg_temp_free_i64(zero); } else { if (TCG_TARGET_REG_BITS == 32) { if (ofs >= 32) { @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop & ~MO_SIGN, idx)); - gen(retv, cpu_env, addr, cmpv, newv, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); } #else gen(retv, cpu_env, addr, cmpv, newv); @@ -XXX,XX +XXX,XX @@ void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop, idx)); - gen(retv, cpu_env, addr, cmpv, newv, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop, idx); + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); } #else gen(retv, cpu_env, addr, cmpv, newv); @@ -XXX,XX +XXX,XX @@ static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop & ~MO_SIGN, idx)); - gen(ret, cpu_env, addr, val, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); } #else gen(ret, cpu_env, addr, val); @@ -XXX,XX +XXX,XX @@ static void do_atomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop & ~MO_SIGN, idx)); - gen(ret, cpu_env, addr, val, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); } #else gen(ret, cpu_env, addr, val); -- 2.25.1 Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/plugin-gen.c | 49 +++++++++++++++++++----------------------- 1 file changed, 22 insertions(+), 27 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 TCGOp *copy_extu_i32_i64(TCGOp **begin_op, TCGOp *op) if (TCG_TARGET_REG_BITS == 32) { /* mov_i32 */ op = copy_op(begin_op, op, INDEX_op_mov_i32); - /* movi_i32 */ - op = copy_op(begin_op, op, INDEX_op_movi_i32); + /* mov_i32 w/ $0 */ + op = copy_op(begin_op, op, INDEX_op_mov_i32); } else { /* extu_i32_i64 */ op = copy_op(begin_op, op, INDEX_op_extu_i32_i64); @@ -XXX,XX +XXX,XX @@ static TCGOp *copy_mov_i64(TCGOp **begin_op, TCGOp *op) return op; } -static TCGOp *copy_movi_i64(TCGOp **begin_op, TCGOp *op, uint64_t v) -{ - if (TCG_TARGET_REG_BITS == 32) { - /* 2x movi_i32 */ - op = copy_op(begin_op, op, INDEX_op_movi_i32); - op->args[1] = v; - - op = copy_op(begin_op, op, INDEX_op_movi_i32); - op->args[1] = v >> 32; - } else { - /* movi_i64 */ - op = copy_op(begin_op, op, INDEX_op_movi_i64); - op->args[1] = v; - } - return op; -} - static TCGOp *copy_const_ptr(TCGOp **begin_op, TCGOp *op, void *ptr) { if (UINTPTR_MAX == UINT32_MAX) { - /* movi_i32 */ - op = copy_op(begin_op, op, INDEX_op_movi_i32); - op->args[1] = (uintptr_t)ptr; + /* mov_i32 */ + op = copy_op(begin_op, op, INDEX_op_mov_i32); + op->args[1] = tcgv_i32_arg(tcg_constant_i32((uintptr_t)ptr)); } else { - /* movi_i64 */ - op = copy_movi_i64(begin_op, op, (uint64_t)(uintptr_t)ptr); + /* mov_i64 */ + op = copy_op(begin_op, op, INDEX_op_mov_i64); + op->args[1] = tcgv_i64_arg(tcg_constant_i64((uintptr_t)ptr)); } return op; } static TCGOp *copy_const_i64(TCGOp **begin_op, TCGOp *op, uint64_t v) { - return copy_movi_i64(begin_op, op, v); + if (TCG_TARGET_REG_BITS == 32) { + /* 2x mov_i32 */ + op = copy_op(begin_op, op, INDEX_op_mov_i32); + op->args[1] = tcgv_i32_arg(tcg_constant_i32(v)); + op = copy_op(begin_op, op, INDEX_op_mov_i32); + op->args[1] = tcgv_i32_arg(tcg_constant_i32(v >> 32)); + } else { + /* mov_i64 */ + op = copy_op(begin_op, op, INDEX_op_mov_i64); + op->args[1] = tcgv_i64_arg(tcg_constant_i64(v)); + } + return op; } static TCGOp *copy_extu_tl_i64(TCGOp **begin_op, TCGOp *op) @@ -XXX,XX +XXX,XX @@ static TCGOp *append_mem_cb(const struct qemu_plugin_dyn_cb *cb, tcg_debug_assert(type == PLUGIN_GEN_CB_MEM); - /* const_i32 == movi_i32 ("info", so it remains as is) */ - op = copy_op(&begin_op, op, INDEX_op_movi_i32); + /* const_i32 == mov_i32 ("info", so it remains as is) */ + op = copy_op(&begin_op, op, INDEX_op_mov_i32); /* const_ptr */ op = copy_const_ptr(&begin_op, op, cb->userp); -- 2.25.1 Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg.h | 1 + tcg/tcg-op-gvec.c | 129 ++++++++++++++++++---------------------------- tcg/tcg.c | 8 +++ 3 files changed, 60 insertions(+), 78 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 @@ static inline TCGv_i64 tcg_constant_i64(int64_t val) } TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val); +TCGv_vec tcg_constant_vec_matching(TCGv_vec match, unsigned vece, int64_t val); #if UINTPTR_MAX == UINT32_MAX # define tcg_const_ptr(x) ((TCGv_ptr)tcg_const_i32((intptr_t)(x))) diff --git a/tcg/tcg-op-gvec.c b/tcg/tcg-op-gvec.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-gvec.c +++ b/tcg/tcg-op-gvec.c @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_2_ool(uint32_t dofs, uint32_t aofs, gen_helper_gvec_2 *fn) { TCGv_ptr a0, a1; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_2_ool(uint32_t dofs, uint32_t aofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with two vector operands @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_2i_ool(uint32_t dofs, uint32_t aofs, TCGv_i64 c, gen_helper_gvec_2i *fn) { TCGv_ptr a0, a1; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_2i_ool(uint32_t dofs, uint32_t aofs, TCGv_i64 c, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with three vector operands. */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_3_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, gen_helper_gvec_3 *fn) { TCGv_ptr a0, a1, a2; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_3_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with four vector operands. */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_4_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, int32_t data, gen_helper_gvec_4 *fn) { TCGv_ptr a0, a1, a2, a3; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_4_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with five vector operands. */ @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_5_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, uint32_t maxsz, int32_t data, gen_helper_gvec_5 *fn) { TCGv_ptr a0, a1, a2, a3, a4; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_5_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); tcg_temp_free_ptr(a4); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with three vector operands @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_2_ptr(uint32_t dofs, uint32_t aofs, int32_t data, gen_helper_gvec_2_ptr *fn) { TCGv_ptr a0, a1; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_2_ptr(uint32_t dofs, uint32_t aofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with three vector operands @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_3_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, int32_t data, gen_helper_gvec_3_ptr *fn) { TCGv_ptr a0, a1, a2; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_3_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with four vector operands @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_4_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, gen_helper_gvec_4_ptr *fn) { TCGv_ptr a0, a1, a2, a3; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_4_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with five vector operands @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_5_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, gen_helper_gvec_5_ptr *fn) { TCGv_ptr a0, a1, a2, a3, a4; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_5_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); tcg_temp_free_ptr(a4); - tcg_temp_free_i32(desc); } /* Return true if we want to implement something of OPRSZ bytes @@ -XXX,XX +XXX,XX @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, || (TCG_TARGET_REG_BITS == 64 && (in_c == 0 || in_c == -1 || !check_size_impl(oprsz, 4)))) { - t_64 = tcg_const_i64(in_c); + t_64 = tcg_constant_i64(in_c); } else { - t_32 = tcg_const_i32(in_c); + t_32 = tcg_constant_i32(in_c); } } @@ -XXX,XX +XXX,XX @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, t_val = tcg_temp_new_i32(); tcg_gen_extrl_i64_i32(t_val, in_64); } else { - t_val = tcg_const_i32(in_c); + t_val = tcg_constant_i32(in_c); } gen_helper_memset(t_ptr, t_ptr, t_val, t_size); - if (!in_32) { + if (in_64) { tcg_temp_free_i32(t_val); } tcg_temp_free_ptr(t_size); @@ -XXX,XX +XXX,XX @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, return; } - t_desc = tcg_const_i32(simd_desc(oprsz, maxsz, 0)); + t_desc = tcg_constant_i32(simd_desc(oprsz, maxsz, 0)); if (vece == MO_64) { if (in_64) { gen_helper_gvec_dup64(t_ptr, t_desc, in_64); } else { - t_64 = tcg_const_i64(in_c); + t_64 = tcg_constant_i64(in_c); gen_helper_gvec_dup64(t_ptr, t_desc, t_64); - tcg_temp_free_i64(t_64); } } else { typedef void dup_fn(TCGv_ptr, TCGv_i32, TCGv_i32); @@ -XXX,XX +XXX,XX @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, if (in_32) { fns[vece](t_ptr, t_desc, in_32); - } else { + } else if (in_64) { t_32 = tcg_temp_new_i32(); - if (in_64) { - tcg_gen_extrl_i64_i32(t_32, in_64); - } else if (vece == MO_8) { - tcg_gen_movi_i32(t_32, in_c & 0xff); - } else if (vece == MO_16) { - tcg_gen_movi_i32(t_32, in_c & 0xffff); - } else { - tcg_gen_movi_i32(t_32, in_c); - } + tcg_gen_extrl_i64_i32(t_32, in_64); fns[vece](t_ptr, t_desc, t_32); tcg_temp_free_i32(t_32); + } else { + if (vece == MO_8) { + in_c &= 0xff; + } else if (vece == MO_16) { + in_c &= 0xffff; + } + t_32 = tcg_constant_i32(in_c); + fns[vece](t_ptr, t_desc, t_32); } } tcg_temp_free_ptr(t_ptr); - tcg_temp_free_i32(t_desc); return; done: @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_2i(uint32_t dofs, uint32_t aofs, uint32_t oprsz, if (g->fno) { tcg_gen_gvec_2_ool(dofs, aofs, oprsz, maxsz, c, g->fno); } else { - TCGv_i64 tcg_c = tcg_const_i64(c); + TCGv_i64 tcg_c = tcg_constant_i64(c); tcg_gen_gvec_2i_ool(dofs, aofs, tcg_c, oprsz, maxsz, c, g->fnoi); - tcg_temp_free_i64(tcg_c); } oprsz = maxsz; } @@ -XXX,XX +XXX,XX @@ static void gen_addv_mask(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b, TCGv_i64 m) void tcg_gen_vec_add8_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80)); gen_addv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_add16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000)); gen_addv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_add32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_adds(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_addi(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(c); + TCGv_i64 tmp = tcg_constant_i64(c); tcg_gen_gvec_adds(vece, dofs, aofs, tmp, oprsz, maxsz); - tcg_temp_free_i64(tmp); } static const TCGOpcode vecop_list_sub[] = { INDEX_op_sub_vec, 0 }; @@ -XXX,XX +XXX,XX @@ static void gen_subv_mask(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b, TCGv_i64 m) void tcg_gen_vec_sub8_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80)); gen_subv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_sub16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000)); gen_subv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_sub32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_muls(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_muli(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(c); + TCGv_i64 tmp = tcg_constant_i64(c); tcg_gen_gvec_muls(vece, dofs, aofs, tmp, oprsz, maxsz); - tcg_temp_free_i64(tmp); } void tcg_gen_gvec_ssadd(unsigned vece, uint32_t dofs, uint32_t aofs, @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_sssub(unsigned vece, uint32_t dofs, uint32_t aofs, static void tcg_gen_usadd_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b) { - TCGv_i32 max = tcg_const_i32(-1); + TCGv_i32 max = tcg_constant_i32(-1); tcg_gen_add_i32(d, a, b); tcg_gen_movcond_i32(TCG_COND_LTU, d, d, a, max, d); - tcg_temp_free_i32(max); } static void tcg_gen_usadd_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 max = tcg_const_i64(-1); + TCGv_i64 max = tcg_constant_i64(-1); tcg_gen_add_i64(d, a, b); tcg_gen_movcond_i64(TCG_COND_LTU, d, d, a, max, d); - tcg_temp_free_i64(max); } void tcg_gen_gvec_usadd(unsigned vece, uint32_t dofs, uint32_t aofs, @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_usadd(unsigned vece, uint32_t dofs, uint32_t aofs, static void tcg_gen_ussub_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b) { - TCGv_i32 min = tcg_const_i32(0); + TCGv_i32 min = tcg_constant_i32(0); tcg_gen_sub_i32(d, a, b); tcg_gen_movcond_i32(TCG_COND_LTU, d, a, b, min, d); - tcg_temp_free_i32(min); } static void tcg_gen_ussub_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 min = tcg_const_i64(0); + TCGv_i64 min = tcg_constant_i64(0); tcg_gen_sub_i64(d, a, b); tcg_gen_movcond_i64(TCG_COND_LTU, d, a, b, min, d); - tcg_temp_free_i64(min); } void tcg_gen_gvec_ussub(unsigned vece, uint32_t dofs, uint32_t aofs, @@ -XXX,XX +XXX,XX @@ static void gen_negv_mask(TCGv_i64 d, TCGv_i64 b, TCGv_i64 m) void tcg_gen_vec_neg8_i64(TCGv_i64 d, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80)); gen_negv_mask(d, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_neg16_i64(TCGv_i64 d, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000)); gen_negv_mask(d, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_neg32_i64(TCGv_i64 d, TCGv_i64 b) @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_ands(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_andi(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c)); + TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c)); tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_ands); - tcg_temp_free_i64(tmp); } static const GVecGen2s gop_xors = { @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_xors(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_xori(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c)); + TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c)); tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_xors); - tcg_temp_free_i64(tmp); } static const GVecGen2s gop_ors = { @@ -XXX,XX +XXX,XX @@ void tcg_gen_gvec_ors(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_ori(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c)); + TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c)); tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_ors); - tcg_temp_free_i64(tmp); } void tcg_gen_vec_shl8i_i64(TCGv_i64 d, TCGv_i64 a, int64_t c) @@ -XXX,XX +XXX,XX @@ static void tcg_gen_shlv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_shlv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -XXX,XX +XXX,XX @@ static void tcg_gen_shrv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_shrv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -XXX,XX +XXX,XX @@ static void tcg_gen_sarv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_sarv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -XXX,XX +XXX,XX @@ static void tcg_gen_rotlv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_rotlv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -XXX,XX +XXX,XX @@ static void tcg_gen_rotrv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_rotrv_vec(vece, d, a, t); tcg_temp_free_vec(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 @@ TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val) return temp_tcgv_vec(tcg_constant_internal(type, val)); } +TCGv_vec tcg_constant_vec_matching(TCGv_vec match, unsigned vece, int64_t val) +{ + TCGTemp *t = tcgv_vec_temp(match); + + tcg_debug_assert(t->temp_allocated != 0); + return tcg_constant_vec(t->base_type, vece, val); +} + TCGv_i32 tcg_const_i32(int32_t val) { TCGv_i32 t0; -- 2.25.1 The normal movi opcodes are going away. We need something for TCI to use internally. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-opc.h | 8 ++++++++ tcg/tci.c | 4 ++-- tcg/tci/tcg-target.c.inc | 4 ++-- 3 files changed, 12 insertions(+), 4 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(last_generic, 0, 0, 0, TCG_OPF_NOT_PRESENT) #include "tcg-target.opc.h" #endif +#ifdef TCG_TARGET_INTERPRETER +/* These opcodes are only for use between the tci generator and interpreter. */ +DEF(tci_movi_i32, 1, 0, 1, TCG_OPF_NOT_PRESENT) +#if TCG_TARGET_REG_BITS == 64 +DEF(tci_movi_i64, 1, 0, 1, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) +#endif +#endif + #undef TLADDR_ARGS #undef DATA64_ARGS #undef IMPL 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, t1 = tci_read_r32(regs, &tb_ptr); tci_write_reg32(regs, t0, t1); break; - case INDEX_op_movi_i32: + case INDEX_op_tci_movi_i32: t0 = *tb_ptr++; t1 = tci_read_i32(&tb_ptr); tci_write_reg32(regs, t0, t1); @@ -XXX,XX +XXX,XX @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, t1 = tci_read_r64(regs, &tb_ptr); tci_write_reg64(regs, t0, t1); break; - case INDEX_op_movi_i64: + case INDEX_op_tci_movi_i64: t0 = *tb_ptr++; t1 = tci_read_i64(&tb_ptr); tci_write_reg64(regs, t0, t1); 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_out_movi(TCGContext *s, TCGType type, uint8_t *old_code_ptr = s->code_ptr; uint32_t arg32 = arg; if (type == TCG_TYPE_I32 || arg == arg32) { - tcg_out_op_t(s, INDEX_op_movi_i32); + tcg_out_op_t(s, INDEX_op_tci_movi_i32); tcg_out_r(s, t0); tcg_out32(s, arg32); } else { tcg_debug_assert(type == TCG_TYPE_I64); #if TCG_TARGET_REG_BITS == 64 - tcg_out_op_t(s, INDEX_op_movi_i64); + tcg_out_op_t(s, INDEX_op_tci_movi_i64); tcg_out_r(s, t0); tcg_out64(s, arg); #else -- 2.25.1 These are now completely covered by mov from a TYPE_CONST temporary. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Aleksandar Markovic <aleksandar.qemu.devel@gmail.com> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-opc.h | 3 --- tcg/optimize.c | 4 ---- tcg/tcg-op-vec.c | 1 - tcg/tcg.c | 18 +----------------- tcg/aarch64/tcg-target.c.inc | 3 --- tcg/arm/tcg-target.c.inc | 1 - tcg/i386/tcg-target.c.inc | 3 --- tcg/mips/tcg-target.c.inc | 2 -- tcg/ppc/tcg-target.c.inc | 3 --- tcg/riscv/tcg-target.c.inc | 2 -- tcg/s390/tcg-target.c.inc | 2 -- tcg/sparc/tcg-target.c.inc | 2 -- tcg/tci/tcg-target.c.inc | 2 -- 13 files changed, 1 insertion(+), 45 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(br, 0, 0, 1, TCG_OPF_BB_END) DEF(mb, 0, 0, 1, 0) DEF(mov_i32, 1, 1, 0, TCG_OPF_NOT_PRESENT) -DEF(movi_i32, 1, 0, 1, TCG_OPF_NOT_PRESENT) DEF(setcond_i32, 1, 2, 1, 0) DEF(movcond_i32, 1, 4, 1, IMPL(TCG_TARGET_HAS_movcond_i32)) /* load/store */ @@ -XXX,XX +XXX,XX @@ DEF(ctz_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_ctz_i32)) DEF(ctpop_i32, 1, 1, 0, IMPL(TCG_TARGET_HAS_ctpop_i32)) DEF(mov_i64, 1, 1, 0, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) -DEF(movi_i64, 1, 0, 1, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) DEF(setcond_i64, 1, 2, 1, IMPL64) DEF(movcond_i64, 1, 4, 1, IMPL64 | IMPL(TCG_TARGET_HAS_movcond_i64)) /* load/store */ @@ -XXX,XX +XXX,XX @@ DEF(qemu_st8_i32, 0, TLADDR_ARGS + 1, 1, #define IMPLVEC TCG_OPF_VECTOR | IMPL(TCG_TARGET_MAYBE_vec) DEF(mov_vec, 1, 1, 0, TCG_OPF_VECTOR | TCG_OPF_NOT_PRESENT) -DEF(dupi_vec, 1, 0, 1, TCG_OPF_VECTOR | TCG_OPF_NOT_PRESENT) DEF(dup_vec, 1, 1, 0, IMPLVEC) DEF(dup2_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_REG_BITS == 32)) 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(mov): tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); break; - CASE_OP_32_64(movi): - case INDEX_op_dupi_vec: - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], op->args[1]); - break; case INDEX_op_dup_vec: if (arg_is_const(op->args[1])) { diff --git a/tcg/tcg-op-vec.c b/tcg/tcg-op-vec.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-vec.c +++ b/tcg/tcg-op-vec.c @@ -XXX,XX +XXX,XX @@ bool tcg_can_emit_vecop_list(const TCGOpcode *list, case INDEX_op_xor_vec: case INDEX_op_mov_vec: case INDEX_op_dup_vec: - case INDEX_op_dupi_vec: case INDEX_op_dup2_vec: case INDEX_op_ld_vec: case INDEX_op_st_vec: 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) return TCG_TARGET_HAS_goto_ptr; case INDEX_op_mov_i32: - case INDEX_op_movi_i32: case INDEX_op_setcond_i32: case INDEX_op_brcond_i32: case INDEX_op_ld8u_i32: @@ -XXX,XX +XXX,XX @@ bool tcg_op_supported(TCGOpcode op) return TCG_TARGET_REG_BITS == 32; case INDEX_op_mov_i64: - case INDEX_op_movi_i64: case INDEX_op_setcond_i64: case INDEX_op_brcond_i64: case INDEX_op_ld8u_i64: @@ -XXX,XX +XXX,XX @@ bool tcg_op_supported(TCGOpcode op) case INDEX_op_mov_vec: case INDEX_op_dup_vec: - case INDEX_op_dupi_vec: case INDEX_op_dupm_vec: case INDEX_op_ld_vec: case INDEX_op_st_vec: @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_cbranch(TCGContext *s, TCGRegSet allocated_regs) } /* - * Specialized code generation for INDEX_op_movi_*. + * Specialized code generation for INDEX_op_mov_* with a constant. */ static void tcg_reg_alloc_do_movi(TCGContext *s, TCGTemp *ots, tcg_target_ulong val, TCGLifeData arg_life, @@ -XXX,XX +XXX,XX @@ static void tcg_reg_alloc_do_movi(TCGContext *s, TCGTemp *ots, } } -static void tcg_reg_alloc_movi(TCGContext *s, const TCGOp *op) -{ - TCGTemp *ots = arg_temp(op->args[0]); - tcg_target_ulong val = op->args[1]; - - tcg_reg_alloc_do_movi(s, ots, val, op->life, op->output_pref[0]); -} - /* * Specialized code generation for INDEX_op_mov_*. */ @@ -XXX,XX +XXX,XX @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb) case INDEX_op_mov_vec: tcg_reg_alloc_mov(s, op); break; - case INDEX_op_movi_i32: - case INDEX_op_movi_i64: - case INDEX_op_dupi_vec: - tcg_reg_alloc_movi(s, op); - break; case INDEX_op_dup_vec: tcg_reg_alloc_dup(s, op); break; 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, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: g_assert_not_reached(); @@ -XXX,XX +XXX,XX @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ - case INDEX_op_dupi_vec: /* Always emitted via tcg_out_movi. */ case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ default: g_assert_not_reached(); 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 inline void tcg_out_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); 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, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); @@ -XXX,XX +XXX,XX @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ - case INDEX_op_dupi_vec: /* Always emitted via tcg_out_movi. */ case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ default: g_assert_not_reached(); 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 inline void tcg_out_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); 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, const TCGArg *args, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); @@ -XXX,XX +XXX,XX @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, return; case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ - case INDEX_op_dupi_vec: /* Always emitted via tcg_out_movi. */ case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ default: g_assert_not_reached(); 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, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: g_assert_not_reached(); diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/s390/tcg-target.c.inc +++ b/tcg/s390/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc index XXXXXXX..XXXXXXX 100644 --- a/tcg/sparc/tcg-target.c.inc +++ b/tcg/sparc/tcg-target.c.inc @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); 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_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); -- 2.25.1 There are several ways we can expand a vector dup of a 64-bit element on a 32-bit host. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/tcg.c | 97 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 97 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 tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) } } +static bool tcg_reg_alloc_dup2(TCGContext *s, const TCGOp *op) +{ + const TCGLifeData arg_life = op->life; + TCGTemp *ots, *itsl, *itsh; + TCGType vtype = TCGOP_VECL(op) + TCG_TYPE_V64; + + /* This opcode is only valid for 32-bit hosts, for 64-bit elements. */ + tcg_debug_assert(TCG_TARGET_REG_BITS == 32); + tcg_debug_assert(TCGOP_VECE(op) == MO_64); + + ots = arg_temp(op->args[0]); + itsl = arg_temp(op->args[1]); + itsh = arg_temp(op->args[2]); + + /* ENV should not be modified. */ + tcg_debug_assert(!temp_readonly(ots)); + + /* Allocate the output register now. */ + if (ots->val_type != TEMP_VAL_REG) { + TCGRegSet allocated_regs = s->reserved_regs; + TCGRegSet dup_out_regs = + tcg_op_defs[INDEX_op_dup_vec].args_ct[0].regs; + + /* Make sure to not spill the input registers. */ + if (!IS_DEAD_ARG(1) && itsl->val_type == TEMP_VAL_REG) { + tcg_regset_set_reg(allocated_regs, itsl->reg); + } + if (!IS_DEAD_ARG(2) && itsh->val_type == TEMP_VAL_REG) { + tcg_regset_set_reg(allocated_regs, itsh->reg); + } + + ots->reg = tcg_reg_alloc(s, dup_out_regs, allocated_regs, + op->output_pref[0], ots->indirect_base); + ots->val_type = TEMP_VAL_REG; + ots->mem_coherent = 0; + s->reg_to_temp[ots->reg] = ots; + } + + /* Promote dup2 of immediates to dupi_vec. */ + if (itsl->val_type == TEMP_VAL_CONST && itsh->val_type == TEMP_VAL_CONST) { + uint64_t val = deposit64(itsl->val, 32, 32, itsh->val); + MemOp vece = MO_64; + + if (val == dup_const(MO_8, val)) { + vece = MO_8; + } else if (val == dup_const(MO_16, val)) { + vece = MO_16; + } else if (val == dup_const(MO_32, val)) { + vece = MO_32; + } + + tcg_out_dupi_vec(s, vtype, vece, ots->reg, val); + goto done; + } + + /* If the two inputs form one 64-bit value, try dupm_vec. */ + if (itsl + 1 == itsh && itsl->base_type == TCG_TYPE_I64) { + if (!itsl->mem_coherent) { + temp_sync(s, itsl, s->reserved_regs, 0, 0); + } + if (!itsh->mem_coherent) { + temp_sync(s, itsh, s->reserved_regs, 0, 0); + } +#ifdef HOST_WORDS_BIGENDIAN + TCGTemp *its = itsh; +#else + TCGTemp *its = itsl; +#endif + if (tcg_out_dupm_vec(s, vtype, MO_64, ots->reg, + its->mem_base->reg, its->mem_offset)) { + goto done; + } + } + + /* Fall back to generic expansion. */ + return false; + + done: + if (IS_DEAD_ARG(1)) { + temp_dead(s, itsl); + } + if (IS_DEAD_ARG(2)) { + temp_dead(s, itsh); + } + if (NEED_SYNC_ARG(0)) { + temp_sync(s, ots, s->reserved_regs, 0, IS_DEAD_ARG(0)); + } else if (IS_DEAD_ARG(0)) { + temp_dead(s, ots); + } + return true; +} + #ifdef TCG_TARGET_STACK_GROWSUP #define STACK_DIR(x) (-(x)) #else @@ -XXX,XX +XXX,XX @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb) case INDEX_op_call: tcg_reg_alloc_call(s, op); break; + case INDEX_op_dup2_vec: + if (tcg_reg_alloc_dup2(s, op)) { + break; + } + /* fall through */ default: /* Sanity check that we've not introduced any unhandled opcodes. */ tcg_debug_assert(tcg_op_supported(opc)); -- 2.25.1 Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/i386/tcg-target.c.inc | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 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 void expand_vec_rotv(TCGType type, unsigned vece, TCGv_vec v0, static void expand_vec_mul(TCGType type, unsigned vece, TCGv_vec v0, TCGv_vec v1, TCGv_vec v2) { - TCGv_vec t1, t2, t3, t4; + TCGv_vec t1, t2, t3, t4, zero; tcg_debug_assert(vece == MO_8); @@ -XXX,XX +XXX,XX @@ static void expand_vec_mul(TCGType type, unsigned vece, case TCG_TYPE_V64: t1 = tcg_temp_new_vec(TCG_TYPE_V128); t2 = tcg_temp_new_vec(TCG_TYPE_V128); - tcg_gen_dup16i_vec(t2, 0); + zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0); vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8, - tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t2)); + tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero)); vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8, - tcgv_vec_arg(t2), tcgv_vec_arg(t2), tcgv_vec_arg(v2)); + tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2)); tcg_gen_mul_vec(MO_16, t1, t1, t2); tcg_gen_shri_vec(MO_16, t1, t1, 8); vec_gen_3(INDEX_op_x86_packus_vec, TCG_TYPE_V128, MO_8, @@ -XXX,XX +XXX,XX @@ static void expand_vec_mul(TCGType type, unsigned vece, t2 = tcg_temp_new_vec(type); t3 = tcg_temp_new_vec(type); t4 = tcg_temp_new_vec(type); - tcg_gen_dup16i_vec(t4, 0); + zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0); vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8, - tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t4)); + tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero)); vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8, - tcgv_vec_arg(t2), tcgv_vec_arg(t4), tcgv_vec_arg(v2)); + tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2)); vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8, - tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(t4)); + tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(zero)); vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8, - tcgv_vec_arg(t4), tcgv_vec_arg(t4), tcgv_vec_arg(v2)); + tcgv_vec_arg(t4), tcgv_vec_arg(zero), tcgv_vec_arg(v2)); tcg_gen_mul_vec(MO_16, t1, t1, t2); tcg_gen_mul_vec(MO_16, t3, t3, t4); tcg_gen_shri_vec(MO_16, t1, t1, 8); @@ -XXX,XX +XXX,XX @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0, NEED_UMIN = 8, NEED_UMAX = 16, }; - TCGv_vec t1, t2; + TCGv_vec t1, t2, t3; uint8_t fixup; switch (cond) { @@ -XXX,XX +XXX,XX @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0, } else if (fixup & NEED_BIAS) { t1 = tcg_temp_new_vec(type); t2 = tcg_temp_new_vec(type); - tcg_gen_dupi_vec(vece, t2, 1ull << ((8 << vece) - 1)); - tcg_gen_sub_vec(vece, t1, v1, t2); - tcg_gen_sub_vec(vece, t2, v2, t2); + t3 = tcg_constant_vec(type, vece, 1ull << ((8 << vece) - 1)); + tcg_gen_sub_vec(vece, t1, v1, t3); + tcg_gen_sub_vec(vece, t2, v2, t3); v1 = t1; v2 = t2; cond = tcg_signed_cond(cond); -- 2.25.1 These interfaces have been replaced by tcg_gen_dupi_vec and tcg_constant_vec. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/tcg/tcg-op.h | 4 ---- tcg/tcg-op-vec.c | 20 -------------------- 2 files changed, 24 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 @@ void tcg_gen_mov_vec(TCGv_vec, TCGv_vec); void tcg_gen_dup_i32_vec(unsigned vece, TCGv_vec, TCGv_i32); void tcg_gen_dup_i64_vec(unsigned vece, TCGv_vec, TCGv_i64); void tcg_gen_dup_mem_vec(unsigned vece, TCGv_vec, TCGv_ptr, tcg_target_long); -void tcg_gen_dup8i_vec(TCGv_vec, uint32_t); -void tcg_gen_dup16i_vec(TCGv_vec, uint32_t); -void tcg_gen_dup32i_vec(TCGv_vec, uint32_t); -void tcg_gen_dup64i_vec(TCGv_vec, uint64_t); void tcg_gen_dupi_vec(unsigned vece, TCGv_vec, uint64_t); void tcg_gen_add_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b); void tcg_gen_sub_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b); diff --git a/tcg/tcg-op-vec.c b/tcg/tcg-op-vec.c index XXXXXXX..XXXXXXX 100644 --- a/tcg/tcg-op-vec.c +++ b/tcg/tcg-op-vec.c @@ -XXX,XX +XXX,XX @@ TCGv_vec tcg_const_ones_vec_matching(TCGv_vec m) return tcg_const_ones_vec(t->base_type); } -void tcg_gen_dup64i_vec(TCGv_vec r, uint64_t a) -{ - tcg_gen_dupi_vec(MO_64, r, a); -} - -void tcg_gen_dup32i_vec(TCGv_vec r, uint32_t a) -{ - tcg_gen_dupi_vec(MO_32, r, a); -} - -void tcg_gen_dup16i_vec(TCGv_vec r, uint32_t a) -{ - tcg_gen_dupi_vec(MO_16, r, a); -} - -void tcg_gen_dup8i_vec(TCGv_vec r, uint32_t a) -{ - tcg_gen_dupi_vec(MO_8, r, a); -} - void tcg_gen_dupi_vec(unsigned vece, TCGv_vec r, uint64_t a) { TCGTemp *rt = tcgv_vec_temp(r); -- 2.25.1 Improve expand_vec_shi to use sign-extraction for MO_32. This allows a single VSPLTISB instruction to load all of the valid shift constants. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/ppc/tcg-target.c.inc | 44 ++++++++++++++++++++++++---------------- 1 file changed, 27 insertions(+), 17 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 void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, static void expand_vec_shi(TCGType type, unsigned vece, TCGv_vec v0, TCGv_vec v1, TCGArg imm, TCGOpcode opci) { - TCGv_vec t1 = tcg_temp_new_vec(type); + TCGv_vec t1; - /* Splat w/bytes for xxspltib. */ - tcg_gen_dupi_vec(MO_8, t1, imm & ((8 << vece) - 1)); + if (vece == MO_32) { + /* + * Only 5 bits are significant, and VSPLTISB can represent -16..15. + * So using negative numbers gets us the 4th bit easily. + */ + imm = sextract32(imm, 0, 5); + } else { + imm &= (8 << vece) - 1; + } + + /* Splat w/bytes for xxspltib when 2.07 allows MO_64. */ + t1 = tcg_constant_vec(type, MO_8, imm); vec_gen_3(opci, type, vece, tcgv_vec_arg(v0), tcgv_vec_arg(v1), tcgv_vec_arg(t1)); - tcg_temp_free_vec(t1); } static void expand_vec_cmp(TCGType type, unsigned vece, TCGv_vec v0, @@ -XXX,XX +XXX,XX @@ static void expand_vec_mul(TCGType type, unsigned vece, TCGv_vec v0, { TCGv_vec t1 = tcg_temp_new_vec(type); TCGv_vec t2 = tcg_temp_new_vec(type); - TCGv_vec t3, t4; + TCGv_vec c0, c16; switch (vece) { case MO_8: @@ -XXX,XX +XXX,XX @@ static void expand_vec_mul(TCGType type, unsigned vece, TCGv_vec v0, case MO_32: tcg_debug_assert(!have_isa_2_07); - t3 = tcg_temp_new_vec(type); - t4 = tcg_temp_new_vec(type); - tcg_gen_dupi_vec(MO_8, t4, -16); + /* + * Only 5 bits are significant, and VSPLTISB can represent -16..15. + * So using -16 is a quick way to represent 16. + */ + c16 = tcg_constant_vec(type, MO_8, -16); + c0 = tcg_constant_vec(type, MO_8, 0); + vec_gen_3(INDEX_op_rotlv_vec, type, MO_32, tcgv_vec_arg(t1), - tcgv_vec_arg(v2), tcgv_vec_arg(t4)); + tcgv_vec_arg(v2), tcgv_vec_arg(c16)); vec_gen_3(INDEX_op_ppc_mulou_vec, type, MO_16, tcgv_vec_arg(t2), tcgv_vec_arg(v1), tcgv_vec_arg(v2)); - tcg_gen_dupi_vec(MO_8, t3, 0); - vec_gen_4(INDEX_op_ppc_msum_vec, type, MO_16, tcgv_vec_arg(t3), - tcgv_vec_arg(v1), tcgv_vec_arg(t1), tcgv_vec_arg(t3)); - vec_gen_3(INDEX_op_shlv_vec, type, MO_32, tcgv_vec_arg(t3), - tcgv_vec_arg(t3), tcgv_vec_arg(t4)); - tcg_gen_add_vec(MO_32, v0, t2, t3); - tcg_temp_free_vec(t3); - tcg_temp_free_vec(t4); + vec_gen_4(INDEX_op_ppc_msum_vec, type, MO_16, tcgv_vec_arg(t1), + tcgv_vec_arg(v1), tcgv_vec_arg(t1), tcgv_vec_arg(c0)); + vec_gen_3(INDEX_op_shlv_vec, type, MO_32, tcgv_vec_arg(t1), + tcgv_vec_arg(t1), tcgv_vec_arg(c16)); + tcg_gen_add_vec(MO_32, v0, t1, t2); break; default: -- 2.25.1 Improve rotrv_vec to reduce "t1 = -v2, t2 = t1 + c" to "t1 = -v2, t2 = c - v2". This avoids a serial dependency between t1 and t2. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/aarch64/tcg-target.c.inc | 10 +++++----- 1 file changed, 5 insertions(+), 5 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 @@ void tcg_expand_vec_op(TCGOpcode opc, TCGType type, unsigned vece, TCGArg a0, ...) { va_list va; - TCGv_vec v0, v1, v2, t1, t2; + TCGv_vec v0, v1, v2, t1, t2, c1; TCGArg a2; va_start(va, a0); @@ -XXX,XX +XXX,XX @@ void tcg_expand_vec_op(TCGOpcode opc, TCGType type, unsigned vece, case INDEX_op_rotlv_vec: t1 = tcg_temp_new_vec(type); - tcg_gen_dupi_vec(vece, t1, 8 << vece); - tcg_gen_sub_vec(vece, t1, v2, t1); + c1 = tcg_constant_vec(type, vece, 8 << vece); + tcg_gen_sub_vec(vece, t1, v2, c1); /* Right shifts are negative left shifts for AArch64. */ vec_gen_3(INDEX_op_shlv_vec, type, vece, tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t1)); @@ -XXX,XX +XXX,XX @@ void tcg_expand_vec_op(TCGOpcode opc, TCGType type, unsigned vece, case INDEX_op_rotrv_vec: t1 = tcg_temp_new_vec(type); t2 = tcg_temp_new_vec(type); + c1 = tcg_constant_vec(type, vece, 8 << vece); tcg_gen_neg_vec(vece, t1, v2); - tcg_gen_dupi_vec(vece, t2, 8 << vece); - tcg_gen_add_vec(vece, t2, t1, t2); + tcg_gen_sub_vec(vece, t2, c1, v2); /* Right shifts are negative left shifts for AArch64. */ vec_gen_3(INDEX_op_shlv_vec, type, vece, tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t1)); -- 2.25.1 From: Philippe Mathieu-Daudé <f4bug@amsat.org> When decodetree.py was added in commit 568ae7efae7, QEMU was using Python 2 which happily reads UTF-8 files in text mode. Python 3 requires either UTF-8 locale or an explicit encoding passed to open(). Now that Python 3 is required, explicit UTF-8 encoding for decodetree source files. To avoid further problems with the user locale, also explicit UTF-8 encoding for the generated C files. Explicit both input/output are plain text by using the 't' mode. This fixes: $ /usr/bin/python3 scripts/decodetree.py test.decode Traceback (most recent call last): File "scripts/decodetree.py", line 1397, in <module> main() File "scripts/decodetree.py", line 1308, in main parse_file(f, toppat) File "scripts/decodetree.py", line 994, in parse_file for line in f: File "/usr/lib/python3.6/encodings/ascii.py", line 26, in decode return codecs.ascii_decode(input, self.errors)[0] UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 80: ordinal not in range(128) Reported-by: Peter Maydell <peter.maydell@linaro.org> Suggested-by: Yonggang Luo <luoyonggang@gmail.com> Reviewed-by: Eduardo Habkost <ehabkost@redhat.com> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Message-Id: <20210110000240.761122-1-f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- scripts/decodetree.py | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/scripts/decodetree.py b/scripts/decodetree.py index XXXXXXX..XXXXXXX 100644 --- a/scripts/decodetree.py +++ b/scripts/decodetree.py @@ -XXX,XX +XXX,XX @@ # See the syntax and semantics in docs/devel/decodetree.rst. # +import io import os import re import sys @@ -XXX,XX +XXX,XX @@ def main(): for filename in args: input_file = filename - f = open(filename, 'r') + f = open(filename, 'rt', encoding='utf-8') parse_file(f, toppat) f.close() @@ -XXX,XX +XXX,XX @@ def main(): prop_size(stree) if output_file: - output_fd = open(output_file, 'w') + output_fd = open(output_file, 'wt', encoding='utf-8') else: - output_fd = sys.stdout + output_fd = io.TextIOWrapper(sys.stdout.buffer, + encoding=sys.stdout.encoding, + errors="ignore") output_autogen() for n in sorted(arguments.keys()): -- 2.25.1 TCG patch queue, plus one target/sh4 patch that Yoshinori Sato asked me to process. r~ The following changes since commit efbf38d73e5dcc4d5f8b98c6e7a12be1f3b91745: Merge tag 'for-upstream' of git://repo.or.cz/qemu/kevin into staging (2022-10-03 15:06:07 -0400) are available in the Git repository at: https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20221004 for you to fetch changes up to ab419fd8a035a65942de4e63effcd55ccbf1a9fe: target/sh4: Fix TB_FLAG_UNALIGN (2022-10-04 12:33:05 -0700) ---------------------------------------------------------------- Cache CPUClass for use in hot code paths. Add CPUTLBEntryFull, probe_access_full, tlb_set_page_full. Add generic support for TARGET_TB_PCREL. tcg/ppc: Optimize 26-bit jumps using STQ for POWER 2.07 target/sh4: Fix TB_FLAG_UNALIGN ---------------------------------------------------------------- Alex Bennée (3): cpu: cache CPUClass in CPUState for hot code paths hw/core/cpu-sysemu: used cached class in cpu_asidx_from_attrs cputlb: used cached CPUClass in our hot-paths Leandro Lupori (1): tcg/ppc: Optimize 26-bit jumps Richard Henderson (16): accel/tcg: Rename CPUIOTLBEntry to CPUTLBEntryFull accel/tcg: Drop addr member from SavedIOTLB accel/tcg: Suppress auto-invalidate in probe_access_internal accel/tcg: Introduce probe_access_full accel/tcg: Introduce tlb_set_page_full include/exec: Introduce TARGET_PAGE_ENTRY_EXTRA accel/tcg: Remove PageDesc code_bitmap accel/tcg: Use bool for page_find_alloc accel/tcg: Use DisasContextBase in plugin_gen_tb_start accel/tcg: Do not align tb->page_addr[0] accel/tcg: Inline tb_flush_jmp_cache include/hw/core: Create struct CPUJumpCache hw/core: Add CPUClass.get_pc accel/tcg: Introduce tb_pc and log_pc accel/tcg: Introduce TARGET_TB_PCREL target/sh4: Fix TB_FLAG_UNALIGN accel/tcg/internal.h | 10 ++ accel/tcg/tb-hash.h | 1 + accel/tcg/tb-jmp-cache.h | 65 ++++++++ include/exec/cpu-common.h | 1 + include/exec/cpu-defs.h | 48 ++++-- include/exec/exec-all.h | 75 ++++++++- include/exec/plugin-gen.h | 7 +- include/hw/core/cpu.h | 28 ++-- include/qemu/typedefs.h | 2 + include/tcg/tcg.h | 2 +- target/sh4/cpu.h | 56 ++++--- accel/stubs/tcg-stub.c | 4 + accel/tcg/cpu-exec.c | 80 +++++----- accel/tcg/cputlb.c | 259 ++++++++++++++++++-------------- accel/tcg/plugin-gen.c | 22 +-- accel/tcg/translate-all.c | 214 ++++++++++++-------------- accel/tcg/translator.c | 2 +- cpu.c | 9 +- hw/core/cpu-common.c | 3 +- hw/core/cpu-sysemu.c | 5 +- linux-user/sh4/signal.c | 6 +- plugins/core.c | 2 +- target/alpha/cpu.c | 9 ++ target/arm/cpu.c | 17 ++- target/arm/mte_helper.c | 14 +- target/arm/sve_helper.c | 4 +- target/arm/translate-a64.c | 2 +- target/avr/cpu.c | 10 +- target/cris/cpu.c | 8 + target/hexagon/cpu.c | 10 +- target/hppa/cpu.c | 12 +- target/i386/cpu.c | 9 ++ target/i386/tcg/tcg-cpu.c | 2 +- target/loongarch/cpu.c | 11 +- target/m68k/cpu.c | 8 + target/microblaze/cpu.c | 10 +- target/mips/cpu.c | 8 + target/mips/tcg/exception.c | 2 +- target/mips/tcg/sysemu/special_helper.c | 2 +- target/nios2/cpu.c | 9 ++ target/openrisc/cpu.c | 10 +- target/ppc/cpu_init.c | 8 + target/riscv/cpu.c | 17 ++- target/rx/cpu.c | 10 +- target/s390x/cpu.c | 8 + target/s390x/tcg/mem_helper.c | 4 - target/sh4/cpu.c | 18 ++- target/sh4/helper.c | 6 +- target/sh4/translate.c | 90 +++++------ target/sparc/cpu.c | 10 +- target/tricore/cpu.c | 11 +- target/xtensa/cpu.c | 8 + tcg/tcg.c | 8 +- trace/control-target.c | 2 +- tcg/ppc/tcg-target.c.inc | 119 +++++++++++---- 55 files changed, 915 insertions(+), 462 deletions(-) create mode 100644 accel/tcg/tb-jmp-cache.h From: Alex Bennée <alex.bennee@linaro.org> The class cast checkers are quite expensive and always on (unlike the dynamic case who's checks are gated by CONFIG_QOM_CAST_DEBUG). To avoid the overhead of repeatedly checking something which should never change we cache the CPUClass reference for use in the hot code paths. Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20220811151413.3350684-3-alex.bennee@linaro.org> Signed-off-by: Cédric Le Goater <clg@kaod.org> Message-Id: <20220923084803.498337-3-clg@kaod.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/hw/core/cpu.h | 9 +++++++++ cpu.c | 9 ++++----- 2 files changed, 13 insertions(+), 5 deletions(-) diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/core/cpu.h +++ b/include/hw/core/cpu.h @@ -XXX,XX +XXX,XX @@ typedef int (*WriteCoreDumpFunction)(const void *buf, size_t size, */ #define CPU(obj) ((CPUState *)(obj)) +/* + * The class checkers bring in CPU_GET_CLASS() which is potentially + * expensive given the eventual call to + * object_class_dynamic_cast_assert(). Because of this the CPUState + * has a cached value for the class in cs->cc which is set up in + * cpu_exec_realizefn() for use in hot code paths. + */ typedef struct CPUClass CPUClass; DECLARE_CLASS_CHECKERS(CPUClass, CPU, TYPE_CPU) @@ -XXX,XX +XXX,XX @@ struct qemu_work_item; struct CPUState { /*< private >*/ DeviceState parent_obj; + /* cache to avoid expensive CPU_GET_CLASS */ + CPUClass *cc; /*< public >*/ int nr_cores; diff --git a/cpu.c b/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/cpu.c +++ b/cpu.c @@ -XXX,XX +XXX,XX @@ const VMStateDescription vmstate_cpu_common = { void cpu_exec_realizefn(CPUState *cpu, Error **errp) { -#ifndef CONFIG_USER_ONLY - CPUClass *cc = CPU_GET_CLASS(cpu); -#endif + /* cache the cpu class for the hotpath */ + cpu->cc = CPU_GET_CLASS(cpu); cpu_list_add(cpu); if (!accel_cpu_realizefn(cpu, errp)) { @@ -XXX,XX +XXX,XX @@ void cpu_exec_realizefn(CPUState *cpu, Error **errp) if (qdev_get_vmsd(DEVICE(cpu)) == NULL) { vmstate_register(NULL, cpu->cpu_index, &vmstate_cpu_common, cpu); } - if (cc->sysemu_ops->legacy_vmsd != NULL) { - vmstate_register(NULL, cpu->cpu_index, cc->sysemu_ops->legacy_vmsd, cpu); + if (cpu->cc->sysemu_ops->legacy_vmsd != NULL) { + vmstate_register(NULL, cpu->cpu_index, cpu->cc->sysemu_ops->legacy_vmsd, cpu); } #endif /* CONFIG_USER_ONLY */ } -- 2.34.1 From: Alex Bennée <alex.bennee@linaro.org> This is a heavily used function so lets avoid the cost of CPU_GET_CLASS. On the romulus-bmc run it has a modest effect: Before: 36.812 s ± 0.506 s After: 35.912 s ± 0.168 s Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20220811151413.3350684-4-alex.bennee@linaro.org> Signed-off-by: Cédric Le Goater <clg@kaod.org> Message-Id: <20220923084803.498337-4-clg@kaod.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- hw/core/cpu-sysemu.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/hw/core/cpu-sysemu.c b/hw/core/cpu-sysemu.c index XXXXXXX..XXXXXXX 100644 --- a/hw/core/cpu-sysemu.c +++ b/hw/core/cpu-sysemu.c @@ -XXX,XX +XXX,XX @@ hwaddr cpu_get_phys_page_debug(CPUState *cpu, vaddr addr) int cpu_asidx_from_attrs(CPUState *cpu, MemTxAttrs attrs) { - CPUClass *cc = CPU_GET_CLASS(cpu); int ret = 0; - if (cc->sysemu_ops->asidx_from_attrs) { - ret = cc->sysemu_ops->asidx_from_attrs(cpu, attrs); + if (cpu->cc->sysemu_ops->asidx_from_attrs) { + ret = cpu->cc->sysemu_ops->asidx_from_attrs(cpu, attrs); assert(ret < cpu->num_ases && ret >= 0); } return ret; -- 2.34.1 From: Alex Bennée <alex.bennee@linaro.org> Before: 35.912 s ± 0.168 s After: 35.565 s ± 0.087 s Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20220811151413.3350684-5-alex.bennee@linaro.org> Signed-off-by: Cédric Le Goater <clg@kaod.org> Message-Id: <20220923084803.498337-5-clg@kaod.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/cputlb.c | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 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 @@ void tlb_set_page(CPUState *cpu, target_ulong vaddr, static void tlb_fill(CPUState *cpu, target_ulong addr, int size, MMUAccessType access_type, int mmu_idx, uintptr_t retaddr) { - CPUClass *cc = CPU_GET_CLASS(cpu); bool ok; /* * This is not a probe, so only valid return is success; failure * should result in exception + longjmp to the cpu loop. */ - ok = cc->tcg_ops->tlb_fill(cpu, addr, size, - access_type, mmu_idx, false, retaddr); + ok = cpu->cc->tcg_ops->tlb_fill(cpu, addr, size, + access_type, mmu_idx, false, retaddr); assert(ok); } @@ -XXX,XX +XXX,XX @@ static inline void cpu_unaligned_access(CPUState *cpu, vaddr addr, MMUAccessType access_type, int mmu_idx, uintptr_t retaddr) { - CPUClass *cc = CPU_GET_CLASS(cpu); - - cc->tcg_ops->do_unaligned_access(cpu, addr, access_type, mmu_idx, retaddr); + cpu->cc->tcg_ops->do_unaligned_access(cpu, addr, access_type, + mmu_idx, retaddr); } static inline void cpu_transaction_failed(CPUState *cpu, hwaddr physaddr, @@ -XXX,XX +XXX,XX @@ static int probe_access_internal(CPUArchState *env, target_ulong addr, if (!tlb_hit_page(tlb_addr, page_addr)) { if (!victim_tlb_hit(env, mmu_idx, index, elt_ofs, page_addr)) { CPUState *cs = env_cpu(env); - CPUClass *cc = CPU_GET_CLASS(cs); - if (!cc->tcg_ops->tlb_fill(cs, addr, fault_size, access_type, - mmu_idx, nonfault, retaddr)) { + if (!cs->cc->tcg_ops->tlb_fill(cs, addr, fault_size, access_type, + mmu_idx, nonfault, retaddr)) { /* Non-faulting page table read failed. */ *phost = NULL; return TLB_INVALID_MASK; -- 2.34.1 This structure will shortly contain more than just data for accessing MMIO. Rename the 'addr' member to 'xlat_section' to more clearly indicate its purpose. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/cpu-defs.h | 22 ++++---- accel/tcg/cputlb.c | 102 +++++++++++++++++++------------------ target/arm/mte_helper.c | 14 ++--- target/arm/sve_helper.c | 4 +- target/arm/translate-a64.c | 2 +- 5 files changed, 73 insertions(+), 71 deletions(-) diff --git a/include/exec/cpu-defs.h b/include/exec/cpu-defs.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/cpu-defs.h +++ b/include/exec/cpu-defs.h @@ -XXX,XX +XXX,XX @@ typedef uint64_t target_ulong; # endif # endif +/* Minimalized TLB entry for use by TCG fast path. */ typedef struct CPUTLBEntry { /* bit TARGET_LONG_BITS to TARGET_PAGE_BITS : virtual address bit TARGET_PAGE_BITS-1..4 : Nonzero for accesses that should not @@ -XXX,XX +XXX,XX @@ typedef struct CPUTLBEntry { QEMU_BUILD_BUG_ON(sizeof(CPUTLBEntry) != (1 << CPU_TLB_ENTRY_BITS)); -/* The IOTLB is not accessed directly inline by generated TCG code, - * so the CPUIOTLBEntry layout is not as critical as that of the - * CPUTLBEntry. (This is also why we don't want to combine the two - * structs into one.) +/* + * The full TLB entry, which is not accessed by generated TCG code, + * so the layout is not as critical as that of CPUTLBEntry. This is + * also why we don't want to combine the two structs. */ -typedef struct CPUIOTLBEntry { +typedef struct CPUTLBEntryFull { /* - * @addr contains: + * @xlat_section contains: * - in the lower TARGET_PAGE_BITS, a physical section number * - with the lower TARGET_PAGE_BITS masked off, an offset which * must be added to the virtual address to obtain: @@ -XXX,XX +XXX,XX @@ typedef struct CPUIOTLBEntry { * number is PHYS_SECTION_NOTDIRTY or PHYS_SECTION_ROM) * + the offset within the target MemoryRegion (otherwise) */ - hwaddr addr; + hwaddr xlat_section; MemTxAttrs attrs; -} CPUIOTLBEntry; +} CPUTLBEntryFull; /* * Data elements that are per MMU mode, minus the bits accessed by @@ -XXX,XX +XXX,XX @@ typedef struct CPUTLBDesc { size_t vindex; /* The tlb victim table, in two parts. */ CPUTLBEntry vtable[CPU_VTLB_SIZE]; - CPUIOTLBEntry viotlb[CPU_VTLB_SIZE]; - /* The iotlb. */ - CPUIOTLBEntry *iotlb; + CPUTLBEntryFull vfulltlb[CPU_VTLB_SIZE]; + CPUTLBEntryFull *fulltlb; } CPUTLBDesc; /* 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 tlb_mmu_resize_locked(CPUTLBDesc *desc, CPUTLBDescFast *fast, } g_free(fast->table); - g_free(desc->iotlb); + g_free(desc->fulltlb); tlb_window_reset(desc, now, 0); /* desc->n_used_entries is cleared by the caller */ fast->mask = (new_size - 1) << CPU_TLB_ENTRY_BITS; fast->table = g_try_new(CPUTLBEntry, new_size); - desc->iotlb = g_try_new(CPUIOTLBEntry, new_size); + desc->fulltlb = g_try_new(CPUTLBEntryFull, new_size); /* * If the allocations fail, try smaller sizes. We just freed some @@ -XXX,XX +XXX,XX @@ static void tlb_mmu_resize_locked(CPUTLBDesc *desc, CPUTLBDescFast *fast, * allocations to fail though, so we progressively reduce the allocation * size, aborting if we cannot even allocate the smallest TLB we support. */ - while (fast->table == NULL || desc->iotlb == NULL) { + while (fast->table == NULL || desc->fulltlb == NULL) { if (new_size == (1 << CPU_TLB_DYN_MIN_BITS)) { error_report("%s: %s", __func__, strerror(errno)); abort(); @@ -XXX,XX +XXX,XX @@ static void tlb_mmu_resize_locked(CPUTLBDesc *desc, CPUTLBDescFast *fast, fast->mask = (new_size - 1) << CPU_TLB_ENTRY_BITS; g_free(fast->table); - g_free(desc->iotlb); + g_free(desc->fulltlb); fast->table = g_try_new(CPUTLBEntry, new_size); - desc->iotlb = g_try_new(CPUIOTLBEntry, new_size); + desc->fulltlb = g_try_new(CPUTLBEntryFull, new_size); } } @@ -XXX,XX +XXX,XX @@ static void tlb_mmu_init(CPUTLBDesc *desc, CPUTLBDescFast *fast, int64_t now) desc->n_used_entries = 0; fast->mask = (n_entries - 1) << CPU_TLB_ENTRY_BITS; fast->table = g_new(CPUTLBEntry, n_entries); - desc->iotlb = g_new(CPUIOTLBEntry, n_entries); + desc->fulltlb = g_new(CPUTLBEntryFull, n_entries); tlb_mmu_flush_locked(desc, fast); } @@ -XXX,XX +XXX,XX @@ void tlb_destroy(CPUState *cpu) CPUTLBDescFast *fast = &env_tlb(env)->f[i]; g_free(fast->table); - g_free(desc->iotlb); + g_free(desc->fulltlb); } } @@ -XXX,XX +XXX,XX @@ void tlb_set_page_with_attrs(CPUState *cpu, target_ulong vaddr, /* Evict the old entry into the victim tlb. */ copy_tlb_helper_locked(tv, te); - desc->viotlb[vidx] = desc->iotlb[index]; + desc->vfulltlb[vidx] = desc->fulltlb[index]; tlb_n_used_entries_dec(env, mmu_idx); } @@ -XXX,XX +XXX,XX @@ void tlb_set_page_with_attrs(CPUState *cpu, target_ulong vaddr, * subtract here is that of the page base, and not the same as the * vaddr we add back in io_readx()/io_writex()/get_page_addr_code(). */ - desc->iotlb[index].addr = iotlb - vaddr_page; - desc->iotlb[index].attrs = attrs; + desc->fulltlb[index].xlat_section = iotlb - vaddr_page; + desc->fulltlb[index].attrs = attrs; /* Now calculate the new entry */ tn.addend = addend - vaddr_page; @@ -XXX,XX +XXX,XX @@ static inline void cpu_transaction_failed(CPUState *cpu, hwaddr physaddr, } } -static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry, +static uint64_t io_readx(CPUArchState *env, CPUTLBEntryFull *full, int mmu_idx, target_ulong addr, uintptr_t retaddr, MMUAccessType access_type, MemOp op) { @@ -XXX,XX +XXX,XX @@ static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry, bool locked = false; MemTxResult r; - section = iotlb_to_section(cpu, iotlbentry->addr, iotlbentry->attrs); + section = iotlb_to_section(cpu, full->xlat_section, full->attrs); mr = section->mr; - mr_offset = (iotlbentry->addr & TARGET_PAGE_MASK) + addr; + mr_offset = (full->xlat_section & TARGET_PAGE_MASK) + addr; cpu->mem_io_pc = retaddr; if (!cpu->can_do_io) { cpu_io_recompile(cpu, retaddr); @@ -XXX,XX +XXX,XX @@ static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry, qemu_mutex_lock_iothread(); locked = true; } - r = memory_region_dispatch_read(mr, mr_offset, &val, op, iotlbentry->attrs); + r = memory_region_dispatch_read(mr, mr_offset, &val, op, full->attrs); if (r != MEMTX_OK) { hwaddr physaddr = mr_offset + section->offset_within_address_space - section->offset_within_region; cpu_transaction_failed(cpu, physaddr, addr, memop_size(op), access_type, - mmu_idx, iotlbentry->attrs, r, retaddr); + mmu_idx, full->attrs, r, retaddr); } if (locked) { qemu_mutex_unlock_iothread(); @@ -XXX,XX +XXX,XX @@ static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry, } /* - * Save a potentially trashed IOTLB entry for later lookup by plugin. - * This is read by tlb_plugin_lookup if the iotlb entry doesn't match + * Save a potentially trashed CPUTLBEntryFull for later lookup by plugin. + * This is read by tlb_plugin_lookup if the fulltlb entry doesn't match * because of the side effect of io_writex changing memory layout. */ static void save_iotlb_data(CPUState *cs, hwaddr addr, @@ -XXX,XX +XXX,XX @@ static void save_iotlb_data(CPUState *cs, hwaddr addr, #endif } -static void io_writex(CPUArchState *env, CPUIOTLBEntry *iotlbentry, +static void io_writex(CPUArchState *env, CPUTLBEntryFull *full, int mmu_idx, uint64_t val, target_ulong addr, uintptr_t retaddr, MemOp op) { @@ -XXX,XX +XXX,XX @@ static void io_writex(CPUArchState *env, CPUIOTLBEntry *iotlbentry, bool locked = false; MemTxResult r; - section = iotlb_to_section(cpu, iotlbentry->addr, iotlbentry->attrs); + section = iotlb_to_section(cpu, full->xlat_section, full->attrs); mr = section->mr; - mr_offset = (iotlbentry->addr & TARGET_PAGE_MASK) + addr; + mr_offset = (full->xlat_section & TARGET_PAGE_MASK) + addr; if (!cpu->can_do_io) { cpu_io_recompile(cpu, retaddr); } @@ -XXX,XX +XXX,XX @@ static void io_writex(CPUArchState *env, CPUIOTLBEntry *iotlbentry, * The memory_region_dispatch may trigger a flush/resize * so for plugins we save the iotlb_data just in case. */ - save_iotlb_data(cpu, iotlbentry->addr, section, mr_offset); + save_iotlb_data(cpu, full->xlat_section, section, mr_offset); if (!qemu_mutex_iothread_locked()) { qemu_mutex_lock_iothread(); locked = true; } - r = memory_region_dispatch_write(mr, mr_offset, val, op, iotlbentry->attrs); + r = memory_region_dispatch_write(mr, mr_offset, val, op, full->attrs); if (r != MEMTX_OK) { hwaddr physaddr = mr_offset + section->offset_within_address_space - section->offset_within_region; cpu_transaction_failed(cpu, physaddr, addr, memop_size(op), - MMU_DATA_STORE, mmu_idx, iotlbentry->attrs, r, + MMU_DATA_STORE, mmu_idx, full->attrs, r, retaddr); } if (locked) { @@ -XXX,XX +XXX,XX @@ static bool victim_tlb_hit(CPUArchState *env, size_t mmu_idx, size_t index, copy_tlb_helper_locked(vtlb, &tmptlb); qemu_spin_unlock(&env_tlb(env)->c.lock); - CPUIOTLBEntry tmpio, *io = &env_tlb(env)->d[mmu_idx].iotlb[index]; - CPUIOTLBEntry *vio = &env_tlb(env)->d[mmu_idx].viotlb[vidx]; - tmpio = *io; *io = *vio; *vio = tmpio; + CPUTLBEntryFull *f1 = &env_tlb(env)->d[mmu_idx].fulltlb[index]; + CPUTLBEntryFull *f2 = &env_tlb(env)->d[mmu_idx].vfulltlb[vidx]; + CPUTLBEntryFull tmpf; + tmpf = *f1; *f1 = *f2; *f2 = tmpf; return true; } } @@ -XXX,XX +XXX,XX @@ static bool victim_tlb_hit(CPUArchState *env, size_t mmu_idx, size_t index, (ADDR) & TARGET_PAGE_MASK) static void notdirty_write(CPUState *cpu, vaddr mem_vaddr, unsigned size, - CPUIOTLBEntry *iotlbentry, uintptr_t retaddr) + CPUTLBEntryFull *full, uintptr_t retaddr) { - ram_addr_t ram_addr = mem_vaddr + iotlbentry->addr; + ram_addr_t ram_addr = mem_vaddr + full->xlat_section; trace_memory_notdirty_write_access(mem_vaddr, ram_addr, size); @@ -XXX,XX +XXX,XX @@ int probe_access_flags(CPUArchState *env, target_ulong addr, /* Handle clean RAM pages. */ if (unlikely(flags & TLB_NOTDIRTY)) { uintptr_t index = tlb_index(env, mmu_idx, addr); - CPUIOTLBEntry *iotlbentry = &env_tlb(env)->d[mmu_idx].iotlb[index]; + CPUTLBEntryFull *full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; - notdirty_write(env_cpu(env), addr, 1, iotlbentry, retaddr); + notdirty_write(env_cpu(env), addr, 1, full, retaddr); flags &= ~TLB_NOTDIRTY; } @@ -XXX,XX +XXX,XX @@ void *probe_access(CPUArchState *env, target_ulong addr, int size, if (unlikely(flags & (TLB_NOTDIRTY | TLB_WATCHPOINT))) { uintptr_t index = tlb_index(env, mmu_idx, addr); - CPUIOTLBEntry *iotlbentry = &env_tlb(env)->d[mmu_idx].iotlb[index]; + CPUTLBEntryFull *full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; /* Handle watchpoints. */ if (flags & TLB_WATCHPOINT) { int wp_access = (access_type == MMU_DATA_STORE ? BP_MEM_WRITE : BP_MEM_READ); cpu_check_watchpoint(env_cpu(env), addr, size, - iotlbentry->attrs, wp_access, retaddr); + full->attrs, wp_access, retaddr); } /* Handle clean RAM pages. */ if (flags & TLB_NOTDIRTY) { - notdirty_write(env_cpu(env), addr, 1, iotlbentry, retaddr); + notdirty_write(env_cpu(env), addr, 1, full, retaddr); } } @@ -XXX,XX +XXX,XX @@ tb_page_addr_t get_page_addr_code_hostp(CPUArchState *env, target_ulong addr, * should have just filled the TLB. The one corner case is io_writex * which can cause TLB flushes and potential resizing of the TLBs * losing the information we need. In those cases we need to recover - * data from a copy of the iotlbentry. As long as this always occurs + * data from a copy of the CPUTLBEntryFull. As long as this always occurs * from the same thread (which a mem callback will be) this is safe. */ @@ -XXX,XX +XXX,XX @@ bool tlb_plugin_lookup(CPUState *cpu, target_ulong addr, int mmu_idx, if (likely(tlb_hit(tlb_addr, addr))) { /* We must have an iotlb entry for MMIO */ if (tlb_addr & TLB_MMIO) { - CPUIOTLBEntry *iotlbentry; - iotlbentry = &env_tlb(env)->d[mmu_idx].iotlb[index]; + CPUTLBEntryFull *full; + full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; data->is_io = true; - data->v.io.section = iotlb_to_section(cpu, iotlbentry->addr, iotlbentry->attrs); - data->v.io.offset = (iotlbentry->addr & TARGET_PAGE_MASK) + addr; + data->v.io.section = + iotlb_to_section(cpu, full->xlat_section, full->attrs); + data->v.io.offset = (full->xlat_section & TARGET_PAGE_MASK) + addr; } else { data->is_io = false; data->v.ram.hostaddr = (void *)((uintptr_t)addr + tlbe->addend); @@ -XXX,XX +XXX,XX @@ static void *atomic_mmu_lookup(CPUArchState *env, target_ulong addr, if (unlikely(tlb_addr & TLB_NOTDIRTY)) { notdirty_write(env_cpu(env), addr, size, - &env_tlb(env)->d[mmu_idx].iotlb[index], retaddr); + &env_tlb(env)->d[mmu_idx].fulltlb[index], retaddr); } return hostaddr; @@ -XXX,XX +XXX,XX @@ load_helper(CPUArchState *env, target_ulong addr, MemOpIdx oi, /* Handle anything that isn't just a straight memory access. */ if (unlikely(tlb_addr & ~TARGET_PAGE_MASK)) { - CPUIOTLBEntry *iotlbentry; + CPUTLBEntryFull *full; bool need_swap; /* For anything that is unaligned, recurse through full_load. */ @@ -XXX,XX +XXX,XX @@ load_helper(CPUArchState *env, target_ulong addr, MemOpIdx oi, goto do_unaligned_access; } - iotlbentry = &env_tlb(env)->d[mmu_idx].iotlb[index]; + full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; /* Handle watchpoints. */ if (unlikely(tlb_addr & TLB_WATCHPOINT)) { /* On watchpoint hit, this will longjmp out. */ cpu_check_watchpoint(env_cpu(env), addr, size, - iotlbentry->attrs, BP_MEM_READ, retaddr); + full->attrs, BP_MEM_READ, retaddr); } need_swap = size > 1 && (tlb_addr & TLB_BSWAP); /* Handle I/O access. */ if (likely(tlb_addr & TLB_MMIO)) { - return io_readx(env, iotlbentry, mmu_idx, addr, retaddr, + return io_readx(env, full, mmu_idx, addr, retaddr, access_type, op ^ (need_swap * MO_BSWAP)); } @@ -XXX,XX +XXX,XX @@ store_helper_unaligned(CPUArchState *env, target_ulong addr, uint64_t val, */ if (unlikely(tlb_addr & TLB_WATCHPOINT)) { cpu_check_watchpoint(env_cpu(env), addr, size - size2, - env_tlb(env)->d[mmu_idx].iotlb[index].attrs, + env_tlb(env)->d[mmu_idx].fulltlb[index].attrs, BP_MEM_WRITE, retaddr); } if (unlikely(tlb_addr2 & TLB_WATCHPOINT)) { cpu_check_watchpoint(env_cpu(env), page2, size2, - env_tlb(env)->d[mmu_idx].iotlb[index2].attrs, + env_tlb(env)->d[mmu_idx].fulltlb[index2].attrs, BP_MEM_WRITE, retaddr); } @@ -XXX,XX +XXX,XX @@ store_helper(CPUArchState *env, target_ulong addr, uint64_t val, /* Handle anything that isn't just a straight memory access. */ if (unlikely(tlb_addr & ~TARGET_PAGE_MASK)) { - CPUIOTLBEntry *iotlbentry; + CPUTLBEntryFull *full; bool need_swap; /* For anything that is unaligned, recurse through byte stores. */ @@ -XXX,XX +XXX,XX @@ store_helper(CPUArchState *env, target_ulong addr, uint64_t val, goto do_unaligned_access; } - iotlbentry = &env_tlb(env)->d[mmu_idx].iotlb[index]; + full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; /* Handle watchpoints. */ if (unlikely(tlb_addr & TLB_WATCHPOINT)) { /* On watchpoint hit, this will longjmp out. */ cpu_check_watchpoint(env_cpu(env), addr, size, - iotlbentry->attrs, BP_MEM_WRITE, retaddr); + full->attrs, BP_MEM_WRITE, retaddr); } need_swap = size > 1 && (tlb_addr & TLB_BSWAP); /* Handle I/O access. */ if (tlb_addr & TLB_MMIO) { - io_writex(env, iotlbentry, mmu_idx, val, addr, retaddr, + io_writex(env, full, mmu_idx, val, addr, retaddr, op ^ (need_swap * MO_BSWAP)); return; } @@ -XXX,XX +XXX,XX @@ store_helper(CPUArchState *env, target_ulong addr, uint64_t val, /* Handle clean RAM pages. */ if (tlb_addr & TLB_NOTDIRTY) { - notdirty_write(env_cpu(env), addr, size, iotlbentry, retaddr); + notdirty_write(env_cpu(env), addr, size, full, retaddr); } haddr = (void *)((uintptr_t)addr + entry->addend); diff --git a/target/arm/mte_helper.c b/target/arm/mte_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/mte_helper.c +++ b/target/arm/mte_helper.c @@ -XXX,XX +XXX,XX @@ static uint8_t *allocation_tag_mem(CPUARMState *env, int ptr_mmu_idx, return tags + index; #else uintptr_t index; - CPUIOTLBEntry *iotlbentry; + CPUTLBEntryFull *full; int in_page, flags; ram_addr_t ptr_ra; hwaddr ptr_paddr, tag_paddr, xlat; @@ -XXX,XX +XXX,XX @@ static uint8_t *allocation_tag_mem(CPUARMState *env, int ptr_mmu_idx, assert(!(flags & TLB_INVALID_MASK)); /* - * Find the iotlbentry for ptr. This *must* be present in the TLB + * Find the CPUTLBEntryFull for ptr. This *must* be present in the TLB * because we just found the mapping. * TODO: Perhaps there should be a cputlb helper that returns a * matching tlb entry + iotlb entry. @@ -XXX,XX +XXX,XX @@ static uint8_t *allocation_tag_mem(CPUARMState *env, int ptr_mmu_idx, g_assert(tlb_hit(comparator, ptr)); } # endif - iotlbentry = &env_tlb(env)->d[ptr_mmu_idx].iotlb[index]; + full = &env_tlb(env)->d[ptr_mmu_idx].fulltlb[index]; /* If the virtual page MemAttr != Tagged, access unchecked. */ - if (!arm_tlb_mte_tagged(&iotlbentry->attrs)) { + if (!arm_tlb_mte_tagged(&full->attrs)) { return NULL; } @@ -XXX,XX +XXX,XX @@ static uint8_t *allocation_tag_mem(CPUARMState *env, int ptr_mmu_idx, int wp = ptr_access == MMU_DATA_LOAD ? BP_MEM_READ : BP_MEM_WRITE; assert(ra != 0); cpu_check_watchpoint(env_cpu(env), ptr, ptr_size, - iotlbentry->attrs, wp, ra); + full->attrs, wp, ra); } /* @@ -XXX,XX +XXX,XX @@ static uint8_t *allocation_tag_mem(CPUARMState *env, int ptr_mmu_idx, tag_paddr = ptr_paddr >> (LOG2_TAG_GRANULE + 1); /* Look up the address in tag space. */ - tag_asi = iotlbentry->attrs.secure ? ARMASIdx_TagS : ARMASIdx_TagNS; + tag_asi = full->attrs.secure ? ARMASIdx_TagS : ARMASIdx_TagNS; tag_as = cpu_get_address_space(env_cpu(env), tag_asi); mr = address_space_translate(tag_as, tag_paddr, &xlat, NULL, tag_access == MMU_DATA_STORE, - iotlbentry->attrs); + full->attrs); /* * Note that @mr will never be NULL. If there is nothing in the address diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve_helper.c +++ b/target/arm/sve_helper.c @@ -XXX,XX +XXX,XX @@ bool sve_probe_page(SVEHostPage *info, bool nofault, CPUARMState *env, g_assert(tlb_hit(comparator, addr)); # endif - CPUIOTLBEntry *iotlbentry = &env_tlb(env)->d[mmu_idx].iotlb[index]; - info->attrs = iotlbentry->attrs; + CPUTLBEntryFull *full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; + info->attrs = full->attrs; } #endif diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool is_guarded_page(CPUARMState *env, DisasContext *s) * table entry even for that case. */ return (tlb_hit(entry->addr_code, addr) && - arm_tlb_bti_gp(&env_tlb(env)->d[mmu_idx].iotlb[index].attrs)); + arm_tlb_bti_gp(&env_tlb(env)->d[mmu_idx].fulltlb[index].attrs)); #endif } -- 2.34.1 This field is only written, not read; remove it. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/hw/core/cpu.h | 1 - accel/tcg/cputlb.c | 7 +++---- 2 files changed, 3 insertions(+), 5 deletions(-) diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/core/cpu.h +++ b/include/hw/core/cpu.h @@ -XXX,XX +XXX,XX @@ struct CPUWatchpoint { * the memory regions get moved around by io_writex. */ typedef struct SavedIOTLB { - hwaddr addr; MemoryRegionSection *section; hwaddr mr_offset; } SavedIOTLB; 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 uint64_t io_readx(CPUArchState *env, CPUTLBEntryFull *full, * This is read by tlb_plugin_lookup if the fulltlb entry doesn't match * because of the side effect of io_writex changing memory layout. */ -static void save_iotlb_data(CPUState *cs, hwaddr addr, - MemoryRegionSection *section, hwaddr mr_offset) +static void save_iotlb_data(CPUState *cs, MemoryRegionSection *section, + hwaddr mr_offset) { #ifdef CONFIG_PLUGIN SavedIOTLB *saved = &cs->saved_iotlb; - saved->addr = addr; saved->section = section; saved->mr_offset = mr_offset; #endif @@ -XXX,XX +XXX,XX @@ static void io_writex(CPUArchState *env, CPUTLBEntryFull *full, * The memory_region_dispatch may trigger a flush/resize * so for plugins we save the iotlb_data just in case. */ - save_iotlb_data(cpu, full->xlat_section, section, mr_offset); + save_iotlb_data(cpu, section, mr_offset); if (!qemu_mutex_iothread_locked()) { qemu_mutex_lock_iothread(); -- 2.34.1 When PAGE_WRITE_INV is set when calling tlb_set_page, we immediately set TLB_INVALID_MASK in order to force tlb_fill to be called on the next lookup. Here in probe_access_internal, we have just called tlb_fill and eliminated true misses, thus the lookup must be valid. This allows us to remove a warning comment from s390x. There doesn't seem to be a reason to change the code though. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: David Hildenbrand <david@redhat.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/cputlb.c | 10 +++++++++- target/s390x/tcg/mem_helper.c | 4 ---- 2 files changed, 9 insertions(+), 5 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 @@ static int probe_access_internal(CPUArchState *env, target_ulong addr, } tlb_addr = tlb_read_ofs(entry, elt_ofs); + flags = TLB_FLAGS_MASK; page_addr = addr & TARGET_PAGE_MASK; if (!tlb_hit_page(tlb_addr, page_addr)) { if (!victim_tlb_hit(env, mmu_idx, index, elt_ofs, page_addr)) { @@ -XXX,XX +XXX,XX @@ static int probe_access_internal(CPUArchState *env, target_ulong addr, /* TLB resize via tlb_fill may have moved the entry. */ entry = tlb_entry(env, mmu_idx, addr); + + /* + * With PAGE_WRITE_INV, we set TLB_INVALID_MASK immediately, + * to force the next access through tlb_fill. We've just + * called tlb_fill, so we know that this entry *is* valid. + */ + flags &= ~TLB_INVALID_MASK; } tlb_addr = tlb_read_ofs(entry, elt_ofs); } - flags = tlb_addr & TLB_FLAGS_MASK; + flags &= tlb_addr; /* Fold all "mmio-like" bits into TLB_MMIO. This is not RAM. */ if (unlikely(flags & ~(TLB_WATCHPOINT | TLB_NOTDIRTY))) { diff --git a/target/s390x/tcg/mem_helper.c b/target/s390x/tcg/mem_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/s390x/tcg/mem_helper.c +++ b/target/s390x/tcg/mem_helper.c @@ -XXX,XX +XXX,XX @@ static int s390_probe_access(CPUArchState *env, target_ulong addr, int size, #else int flags; - /* - * For !CONFIG_USER_ONLY, we cannot rely on TLB_INVALID_MASK or haddr==NULL - * to detect if there was an exception during tlb_fill(). - */ env->tlb_fill_exc = 0; flags = probe_access_flags(env, addr, access_type, mmu_idx, nonfault, phost, ra); -- 2.34.1 Add an interface to return the CPUTLBEntryFull struct that goes with the lookup. The result is not intended to be valid across multiple lookups, so the user must use the results immediately. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/exec-all.h | 15 +++++++++++++ include/qemu/typedefs.h | 1 + accel/tcg/cputlb.c | 47 +++++++++++++++++++++++++---------------- 3 files changed, 45 insertions(+), 18 deletions(-) diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/exec-all.h +++ b/include/exec/exec-all.h @@ -XXX,XX +XXX,XX @@ int probe_access_flags(CPUArchState *env, target_ulong addr, MMUAccessType access_type, int mmu_idx, bool nonfault, void **phost, uintptr_t retaddr); +#ifndef CONFIG_USER_ONLY +/** + * probe_access_full: + * Like probe_access_flags, except also return into @pfull. + * + * The CPUTLBEntryFull structure returned via @pfull is transient + * and must be consumed or copied immediately, before any further + * access or changes to TLB @mmu_idx. + */ +int probe_access_full(CPUArchState *env, target_ulong addr, + MMUAccessType access_type, int mmu_idx, + bool nonfault, void **phost, + CPUTLBEntryFull **pfull, uintptr_t retaddr); +#endif + #define CODE_GEN_ALIGN 16 /* must be >= of the size of a icache line */ /* Estimated block size for TB allocation. */ diff --git a/include/qemu/typedefs.h b/include/qemu/typedefs.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/typedefs.h +++ b/include/qemu/typedefs.h @@ -XXX,XX +XXX,XX @@ typedef struct ConfidentialGuestSupport ConfidentialGuestSupport; typedef struct CPUAddressSpace CPUAddressSpace; typedef struct CPUArchState CPUArchState; typedef struct CPUState CPUState; +typedef struct CPUTLBEntryFull CPUTLBEntryFull; typedef struct DeviceListener DeviceListener; typedef struct DeviceState DeviceState; typedef struct DirtyBitmapSnapshot DirtyBitmapSnapshot; 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 notdirty_write(CPUState *cpu, vaddr mem_vaddr, unsigned size, static int probe_access_internal(CPUArchState *env, target_ulong addr, int fault_size, MMUAccessType access_type, int mmu_idx, bool nonfault, - void **phost, uintptr_t retaddr) + void **phost, CPUTLBEntryFull **pfull, + uintptr_t retaddr) { uintptr_t index = tlb_index(env, mmu_idx, addr); CPUTLBEntry *entry = tlb_entry(env, mmu_idx, addr); @@ -XXX,XX +XXX,XX @@ static int probe_access_internal(CPUArchState *env, target_ulong addr, mmu_idx, nonfault, retaddr)) { /* Non-faulting page table read failed. */ *phost = NULL; + *pfull = NULL; return TLB_INVALID_MASK; } /* TLB resize via tlb_fill may have moved the entry. */ + index = tlb_index(env, mmu_idx, addr); entry = tlb_entry(env, mmu_idx, addr); /* @@ -XXX,XX +XXX,XX @@ static int probe_access_internal(CPUArchState *env, target_ulong addr, } flags &= tlb_addr; + *pfull = &env_tlb(env)->d[mmu_idx].fulltlb[index]; + /* Fold all "mmio-like" bits into TLB_MMIO. This is not RAM. */ if (unlikely(flags & ~(TLB_WATCHPOINT | TLB_NOTDIRTY))) { *phost = NULL; @@ -XXX,XX +XXX,XX @@ static int probe_access_internal(CPUArchState *env, target_ulong addr, return flags; } -int probe_access_flags(CPUArchState *env, target_ulong addr, - MMUAccessType access_type, int mmu_idx, - bool nonfault, void **phost, uintptr_t retaddr) +int probe_access_full(CPUArchState *env, target_ulong addr, + MMUAccessType access_type, int mmu_idx, + bool nonfault, void **phost, CPUTLBEntryFull **pfull, + uintptr_t retaddr) { - int flags; - - flags = probe_access_internal(env, addr, 0, access_type, mmu_idx, - nonfault, phost, retaddr); + int flags = probe_access_internal(env, addr, 0, access_type, mmu_idx, + nonfault, phost, pfull, retaddr); /* Handle clean RAM pages. */ if (unlikely(flags & TLB_NOTDIRTY)) { - uintptr_t index = tlb_index(env, mmu_idx, addr); - CPUTLBEntryFull *full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; - - notdirty_write(env_cpu(env), addr, 1, full, retaddr); + notdirty_write(env_cpu(env), addr, 1, *pfull, retaddr); flags &= ~TLB_NOTDIRTY; } return flags; } +int probe_access_flags(CPUArchState *env, target_ulong addr, + MMUAccessType access_type, int mmu_idx, + bool nonfault, void **phost, uintptr_t retaddr) +{ + CPUTLBEntryFull *full; + + return probe_access_full(env, addr, access_type, mmu_idx, + nonfault, phost, &full, retaddr); +} + void *probe_access(CPUArchState *env, target_ulong addr, int size, MMUAccessType access_type, int mmu_idx, uintptr_t retaddr) { + CPUTLBEntryFull *full; void *host; int flags; g_assert(-(addr | TARGET_PAGE_MASK) >= size); flags = probe_access_internal(env, addr, size, access_type, mmu_idx, - false, &host, retaddr); + false, &host, &full, retaddr); /* Per the interface, size == 0 merely faults the access. */ if (size == 0) { @@ -XXX,XX +XXX,XX @@ void *probe_access(CPUArchState *env, target_ulong addr, int size, } if (unlikely(flags & (TLB_NOTDIRTY | TLB_WATCHPOINT))) { - uintptr_t index = tlb_index(env, mmu_idx, addr); - CPUTLBEntryFull *full = &env_tlb(env)->d[mmu_idx].fulltlb[index]; - /* Handle watchpoints. */ if (flags & TLB_WATCHPOINT) { int wp_access = (access_type == MMU_DATA_STORE @@ -XXX,XX +XXX,XX @@ void *probe_access(CPUArchState *env, target_ulong addr, int size, void *tlb_vaddr_to_host(CPUArchState *env, abi_ptr addr, MMUAccessType access_type, int mmu_idx) { + CPUTLBEntryFull *full; void *host; int flags; flags = probe_access_internal(env, addr, 0, access_type, - mmu_idx, true, &host, 0); + mmu_idx, true, &host, &full, 0); /* No combination of flags are expected by the caller. */ return flags ? NULL : host; @@ -XXX,XX +XXX,XX @@ void *tlb_vaddr_to_host(CPUArchState *env, abi_ptr addr, tb_page_addr_t get_page_addr_code_hostp(CPUArchState *env, target_ulong addr, void **hostp) { + CPUTLBEntryFull *full; void *p; (void)probe_access_internal(env, addr, 1, MMU_INST_FETCH, - cpu_mmu_index(env, true), false, &p, 0); + cpu_mmu_index(env, true), false, &p, &full, 0); if (p == NULL) { return -1; } -- 2.34.1 Now that we have collected all of the page data into CPUTLBEntryFull, provide an interface to record that all in one go, instead of using 4 arguments. This interface allows CPUTLBEntryFull to be extended without having to change the number of arguments. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/cpu-defs.h | 14 +++++++++++ include/exec/exec-all.h | 22 ++++++++++++++++++ accel/tcg/cputlb.c | 51 ++++++++++++++++++++++++++--------------- 3 files changed, 69 insertions(+), 18 deletions(-) diff --git a/include/exec/cpu-defs.h b/include/exec/cpu-defs.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/cpu-defs.h +++ b/include/exec/cpu-defs.h @@ -XXX,XX +XXX,XX @@ typedef struct CPUTLBEntryFull { * + the offset within the target MemoryRegion (otherwise) */ hwaddr xlat_section; + + /* + * @phys_addr contains the physical address in the address space + * given by cpu_asidx_from_attrs(cpu, @attrs). + */ + hwaddr phys_addr; + + /* @attrs contains the memory transaction attributes for the page. */ MemTxAttrs attrs; + + /* @prot contains the complete protections for the page. */ + uint8_t prot; + + /* @lg_page_size contains the log2 of the page size. */ + uint8_t lg_page_size; } CPUTLBEntryFull; /* diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/exec-all.h +++ b/include/exec/exec-all.h @@ -XXX,XX +XXX,XX @@ void tlb_flush_range_by_mmuidx_all_cpus_synced(CPUState *cpu, uint16_t idxmap, unsigned bits); +/** + * tlb_set_page_full: + * @cpu: CPU context + * @mmu_idx: mmu index of the tlb to modify + * @vaddr: virtual address of the entry to add + * @full: the details of the tlb entry + * + * Add an entry to @cpu tlb index @mmu_idx. All of the fields of + * @full must be filled, except for xlat_section, and constitute + * the complete description of the translated page. + * + * This is generally called by the target tlb_fill function after + * having performed a successful page table walk to find the physical + * address and attributes for the translation. + * + * At most one entry for a given virtual address is permitted. Only a + * single TARGET_PAGE_SIZE region is mapped; @full->lg_page_size is only + * used by tlb_flush_page. + */ +void tlb_set_page_full(CPUState *cpu, int mmu_idx, target_ulong vaddr, + CPUTLBEntryFull *full); + /** * tlb_set_page_with_attrs: * @cpu: CPU to add this TLB entry for 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 tlb_add_large_page(CPUArchState *env, int mmu_idx, env_tlb(env)->d[mmu_idx].large_page_mask = lp_mask; } -/* Add a new TLB entry. At most one entry for a given virtual address +/* + * Add a new TLB entry. At most one entry for a given virtual address * is permitted. Only a single TARGET_PAGE_SIZE region is mapped, the * supplied size is only used by tlb_flush_page. * * Called from TCG-generated code, which is under an RCU read-side * critical section. */ -void tlb_set_page_with_attrs(CPUState *cpu, target_ulong vaddr, - hwaddr paddr, MemTxAttrs attrs, int prot, - int mmu_idx, target_ulong size) +void tlb_set_page_full(CPUState *cpu, int mmu_idx, + target_ulong vaddr, CPUTLBEntryFull *full) { CPUArchState *env = cpu->env_ptr; CPUTLB *tlb = env_tlb(env); @@ -XXX,XX +XXX,XX @@ void tlb_set_page_with_attrs(CPUState *cpu, target_ulong vaddr, CPUTLBEntry *te, tn; hwaddr iotlb, xlat, sz, paddr_page; target_ulong vaddr_page; - int asidx = cpu_asidx_from_attrs(cpu, attrs); - int wp_flags; + int asidx, wp_flags, prot; bool is_ram, is_romd; assert_cpu_is_self(cpu); - if (size <= TARGET_PAGE_SIZE) { + if (full->lg_page_size <= TARGET_PAGE_BITS) { sz = TARGET_PAGE_SIZE; } else { - tlb_add_large_page(env, mmu_idx, vaddr, size); - sz = size; + sz = (hwaddr)1 << full->lg_page_size; + tlb_add_large_page(env, mmu_idx, vaddr, sz); } vaddr_page = vaddr & TARGET_PAGE_MASK; - paddr_page = paddr & TARGET_PAGE_MASK; + paddr_page = full->phys_addr & TARGET_PAGE_MASK; + prot = full->prot; + asidx = cpu_asidx_from_attrs(cpu, full->attrs); section = address_space_translate_for_iotlb(cpu, asidx, paddr_page, - &xlat, &sz, attrs, &prot); + &xlat, &sz, full->attrs, &prot); assert(sz >= TARGET_PAGE_SIZE); tlb_debug("vaddr=" TARGET_FMT_lx " paddr=0x" TARGET_FMT_plx " prot=%x idx=%d\n", - vaddr, paddr, prot, mmu_idx); + vaddr, full->phys_addr, prot, mmu_idx); address = vaddr_page; - if (size < TARGET_PAGE_SIZE) { + if (full->lg_page_size < TARGET_PAGE_BITS) { /* Repeat the MMU check and TLB fill on every access. */ address |= TLB_INVALID_MASK; } - if (attrs.byte_swap) { + if (full->attrs.byte_swap) { address |= TLB_BSWAP; } @@ -XXX,XX +XXX,XX @@ void tlb_set_page_with_attrs(CPUState *cpu, target_ulong vaddr, * subtract here is that of the page base, and not the same as the * vaddr we add back in io_readx()/io_writex()/get_page_addr_code(). */ + desc->fulltlb[index] = *full; desc->fulltlb[index].xlat_section = iotlb - vaddr_page; - desc->fulltlb[index].attrs = attrs; + desc->fulltlb[index].phys_addr = paddr_page; + desc->fulltlb[index].prot = prot; /* Now calculate the new entry */ tn.addend = addend - vaddr_page; @@ -XXX,XX +XXX,XX @@ void tlb_set_page_with_attrs(CPUState *cpu, target_ulong vaddr, qemu_spin_unlock(&tlb->c.lock); } -/* Add a new TLB entry, but without specifying the memory - * transaction attributes to be used. - */ +void tlb_set_page_with_attrs(CPUState *cpu, target_ulong vaddr, + hwaddr paddr, MemTxAttrs attrs, int prot, + int mmu_idx, target_ulong size) +{ + CPUTLBEntryFull full = { + .phys_addr = paddr, + .attrs = attrs, + .prot = prot, + .lg_page_size = ctz64(size) + }; + + assert(is_power_of_2(size)); + tlb_set_page_full(cpu, mmu_idx, vaddr, &full); +} + void tlb_set_page(CPUState *cpu, target_ulong vaddr, hwaddr paddr, int prot, int mmu_idx, target_ulong size) -- 2.34.1 Allow the target to cache items from the guest page tables. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/cpu-defs.h | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/include/exec/cpu-defs.h b/include/exec/cpu-defs.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/cpu-defs.h +++ b/include/exec/cpu-defs.h @@ -XXX,XX +XXX,XX @@ typedef struct CPUTLBEntryFull { /* @lg_page_size contains the log2 of the page size. */ uint8_t lg_page_size; + + /* + * Allow target-specific additions to this structure. + * This may be used to cache items from the guest cpu + * page tables for later use by the implementation. + */ +#ifdef TARGET_PAGE_ENTRY_EXTRA + TARGET_PAGE_ENTRY_EXTRA +#endif } CPUTLBEntryFull; /* -- 2.34.1 This bitmap is created and discarded immediately. We gain nothing by its existence. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-Id: <20220822232338.1727934-2-richard.henderson@linaro.org> --- accel/tcg/translate-all.c | 78 ++------------------------------------- 1 file changed, 4 insertions(+), 74 deletions(-) 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 @@ #define assert_memory_lock() tcg_debug_assert(have_mmap_lock()) #endif -#define SMC_BITMAP_USE_THRESHOLD 10 - typedef struct PageDesc { /* list of TBs intersecting this ram page */ uintptr_t first_tb; -#ifdef CONFIG_SOFTMMU - /* in order to optimize self modifying code, we count the number - of lookups we do to a given page to use a bitmap */ - unsigned long *code_bitmap; - unsigned int code_write_count; -#else +#ifdef CONFIG_USER_ONLY unsigned long flags; void *target_data; #endif -#ifndef CONFIG_USER_ONLY +#ifdef CONFIG_SOFTMMU QemuSpin lock; #endif } PageDesc; @@ -XXX,XX +XXX,XX @@ void tb_htable_init(void) qht_init(&tb_ctx.htable, tb_cmp, CODE_GEN_HTABLE_SIZE, mode); } -/* call with @p->lock held */ -static inline void invalidate_page_bitmap(PageDesc *p) -{ - assert_page_locked(p); -#ifdef CONFIG_SOFTMMU - g_free(p->code_bitmap); - p->code_bitmap = NULL; - p->code_write_count = 0; -#endif -} - /* Set to NULL all the 'first_tb' fields in all PageDescs. */ static void page_flush_tb_1(int level, void **lp) { @@ -XXX,XX +XXX,XX @@ static void page_flush_tb_1(int level, void **lp) for (i = 0; i < V_L2_SIZE; ++i) { page_lock(&pd[i]); pd[i].first_tb = (uintptr_t)NULL; - invalidate_page_bitmap(pd + i); page_unlock(&pd[i]); } } else { @@ -XXX,XX +XXX,XX @@ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list) if (rm_from_page_list) { p = page_find(tb->page_addr[0] >> TARGET_PAGE_BITS); tb_page_remove(p, tb); - invalidate_page_bitmap(p); if (tb->page_addr[1] != -1) { p = page_find(tb->page_addr[1] >> TARGET_PAGE_BITS); tb_page_remove(p, tb); - invalidate_page_bitmap(p); } } @@ -XXX,XX +XXX,XX @@ void tb_phys_invalidate(TranslationBlock *tb, tb_page_addr_t page_addr) } } -#ifdef CONFIG_SOFTMMU -/* call with @p->lock held */ -static void build_page_bitmap(PageDesc *p) -{ - int n, tb_start, tb_end; - TranslationBlock *tb; - - assert_page_locked(p); - p->code_bitmap = bitmap_new(TARGET_PAGE_SIZE); - - PAGE_FOR_EACH_TB(p, tb, n) { - /* NOTE: this is subtle as a TB may span two physical pages */ - if (n == 0) { - /* NOTE: tb_end may be after the end of the page, but - it is not a problem */ - tb_start = tb->pc & ~TARGET_PAGE_MASK; - tb_end = tb_start + tb->size; - if (tb_end > TARGET_PAGE_SIZE) { - tb_end = TARGET_PAGE_SIZE; - } - } else { - tb_start = 0; - tb_end = ((tb->pc + tb->size) & ~TARGET_PAGE_MASK); - } - bitmap_set(p->code_bitmap, tb_start, tb_end - tb_start); - } -} -#endif - /* add the tb in the target page and protect it if necessary * * Called with mmap_lock held for user-mode emulation. @@ -XXX,XX +XXX,XX @@ static inline void tb_page_add(PageDesc *p, TranslationBlock *tb, page_already_protected = p->first_tb != (uintptr_t)NULL; #endif p->first_tb = (uintptr_t)tb | n; - invalidate_page_bitmap(p); #if defined(CONFIG_USER_ONLY) /* translator_loop() must have made all TB pages non-writable */ @@ -XXX,XX +XXX,XX @@ tb_link_page(TranslationBlock *tb, tb_page_addr_t phys_pc, /* remove TB from the page(s) if we couldn't insert it */ if (unlikely(existing_tb)) { tb_page_remove(p, tb); - invalidate_page_bitmap(p); if (p2) { tb_page_remove(p2, tb); - invalidate_page_bitmap(p2); } tb = existing_tb; } @@ -XXX,XX +XXX,XX @@ tb_invalidate_phys_page_range__locked(struct page_collection *pages, #if !defined(CONFIG_USER_ONLY) /* if no code remaining, no need to continue to use slow writes */ if (!p->first_tb) { - invalidate_page_bitmap(p); tlb_unprotect_code(start); } #endif @@ -XXX,XX +XXX,XX @@ void tb_invalidate_phys_page_fast(struct page_collection *pages, } assert_page_locked(p); - if (!p->code_bitmap && - ++p->code_write_count >= SMC_BITMAP_USE_THRESHOLD) { - build_page_bitmap(p); - } - if (p->code_bitmap) { - unsigned int nr; - unsigned long b; - - nr = start & ~TARGET_PAGE_MASK; - b = p->code_bitmap[BIT_WORD(nr)] >> (nr & (BITS_PER_LONG - 1)); - if (b & ((1 << len) - 1)) { - goto do_invalidate; - } - } else { - do_invalidate: - tb_invalidate_phys_page_range__locked(pages, p, start, start + len, - retaddr); - } + tb_invalidate_phys_page_range__locked(pages, p, start, start + len, + retaddr); } #else /* Called with mmap_lock held. If pc is not 0 then it indicates the -- 2.34.1 Bool is more appropriate type for the alloc parameter. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/translate-all.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) 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 @@ void page_init(void) #endif } -static PageDesc *page_find_alloc(tb_page_addr_t index, int alloc) +static PageDesc *page_find_alloc(tb_page_addr_t index, bool alloc) { PageDesc *pd; void **lp; @@ -XXX,XX +XXX,XX @@ static PageDesc *page_find_alloc(tb_page_addr_t index, int alloc) static inline PageDesc *page_find(tb_page_addr_t index) { - return page_find_alloc(index, 0); + return page_find_alloc(index, false); } static void page_lock_pair(PageDesc **ret_p1, tb_page_addr_t phys1, - PageDesc **ret_p2, tb_page_addr_t phys2, int alloc); + PageDesc **ret_p2, tb_page_addr_t phys2, bool alloc); /* In user-mode page locks aren't used; mmap_lock is enough */ #ifdef CONFIG_USER_ONLY @@ -XXX,XX +XXX,XX @@ static inline void page_unlock(PageDesc *pd) /* lock the page(s) of a TB in the correct acquisition order */ static inline void page_lock_tb(const TranslationBlock *tb) { - page_lock_pair(NULL, tb->page_addr[0], NULL, tb->page_addr[1], 0); + page_lock_pair(NULL, tb->page_addr[0], NULL, tb->page_addr[1], false); } static inline void page_unlock_tb(const TranslationBlock *tb) @@ -XXX,XX +XXX,XX @@ void page_collection_unlock(struct page_collection *set) #endif /* !CONFIG_USER_ONLY */ static void page_lock_pair(PageDesc **ret_p1, tb_page_addr_t phys1, - PageDesc **ret_p2, tb_page_addr_t phys2, int alloc) + PageDesc **ret_p2, tb_page_addr_t phys2, bool alloc) { PageDesc *p1, *p2; tb_page_addr_t page1; @@ -XXX,XX +XXX,XX @@ tb_link_page(TranslationBlock *tb, tb_page_addr_t phys_pc, * Note that inserting into the hash table first isn't an option, since * we can only insert TBs that are fully initialized. */ - page_lock_pair(&p, phys_pc, &p2, phys_page2, 1); + page_lock_pair(&p, phys_pc, &p2, phys_page2, true); tb_page_add(p, tb, 0, phys_pc & TARGET_PAGE_MASK); if (p2) { tb_page_add(p2, tb, 1, phys_page2); @@ -XXX,XX +XXX,XX @@ void page_set_flags(target_ulong start, target_ulong end, int flags) for (addr = start, len = end - start; len != 0; len -= TARGET_PAGE_SIZE, addr += TARGET_PAGE_SIZE) { - PageDesc *p = page_find_alloc(addr >> TARGET_PAGE_BITS, 1); + PageDesc *p = page_find_alloc(addr >> TARGET_PAGE_BITS, true); /* If the write protection bit is set, then we invalidate the code inside. */ -- 2.34.1 Use the pc coming from db->pc_first rather than the TB. Use the cached host_addr rather than re-computing for the first page. We still need a separate lookup for the second page because it won't be computed for DisasContextBase until the translator actually performs a read from the page. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- include/exec/plugin-gen.h | 7 ++++--- accel/tcg/plugin-gen.c | 22 +++++++++++----------- accel/tcg/translator.c | 2 +- 3 files changed, 16 insertions(+), 15 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 @@ struct DisasContextBase; #ifdef CONFIG_PLUGIN -bool plugin_gen_tb_start(CPUState *cpu, const TranslationBlock *tb, bool supress); +bool plugin_gen_tb_start(CPUState *cpu, const struct DisasContextBase *db, + bool supress); void plugin_gen_tb_end(CPUState *cpu); void plugin_gen_insn_start(CPUState *cpu, const struct DisasContextBase *db); void plugin_gen_insn_end(void); @@ -XXX,XX +XXX,XX @@ static inline void plugin_insn_append(abi_ptr pc, const void *from, size_t size) #else /* !CONFIG_PLUGIN */ -static inline -bool plugin_gen_tb_start(CPUState *cpu, const TranslationBlock *tb, bool supress) +static inline bool +plugin_gen_tb_start(CPUState *cpu, const struct DisasContextBase *db, bool sup) { return false; } 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_inject(const struct qemu_plugin_tb *plugin_tb) pr_ops(); } -bool plugin_gen_tb_start(CPUState *cpu, const TranslationBlock *tb, bool mem_only) +bool plugin_gen_tb_start(CPUState *cpu, const DisasContextBase *db, + bool mem_only) { bool ret = false; @@ -XXX,XX +XXX,XX @@ bool plugin_gen_tb_start(CPUState *cpu, const TranslationBlock *tb, bool mem_onl ret = true; - ptb->vaddr = tb->pc; + ptb->vaddr = db->pc_first; ptb->vaddr2 = -1; - get_page_addr_code_hostp(cpu->env_ptr, tb->pc, &ptb->haddr1); + ptb->haddr1 = db->host_addr[0]; ptb->haddr2 = NULL; ptb->mem_only = mem_only; @@ -XXX,XX +XXX,XX @@ void plugin_gen_insn_start(CPUState *cpu, const DisasContextBase *db) * Note that we skip this when haddr1 == NULL, e.g. when we're * fetching instructions from a region not backed by RAM. */ - if (likely(ptb->haddr1 != NULL && ptb->vaddr2 == -1) && - unlikely((db->pc_next & TARGET_PAGE_MASK) != - (db->pc_first & TARGET_PAGE_MASK))) { - get_page_addr_code_hostp(cpu->env_ptr, db->pc_next, - &ptb->haddr2); - ptb->vaddr2 = db->pc_next; - } - if (likely(ptb->vaddr2 == -1)) { + if (ptb->haddr1 == NULL) { + pinsn->haddr = NULL; + } else if (is_same_page(db, db->pc_next)) { pinsn->haddr = ptb->haddr1 + pinsn->vaddr - ptb->vaddr; } else { + if (ptb->vaddr2 == -1) { + ptb->vaddr2 = TARGET_PAGE_ALIGN(db->pc_first); + get_page_addr_code_hostp(cpu->env_ptr, ptb->vaddr2, &ptb->haddr2); + } pinsn->haddr = ptb->haddr2 + pinsn->vaddr - ptb->vaddr2; } } diff --git a/accel/tcg/translator.c b/accel/tcg/translator.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/translator.c +++ b/accel/tcg/translator.c @@ -XXX,XX +XXX,XX @@ void translator_loop(CPUState *cpu, TranslationBlock *tb, int max_insns, ops->tb_start(db, cpu); tcg_debug_assert(db->is_jmp == DISAS_NEXT); /* no early exit */ - plugin_enabled = plugin_gen_tb_start(cpu, tb, cflags & CF_MEMI_ONLY); + plugin_enabled = plugin_gen_tb_start(cpu, db, cflags & CF_MEMI_ONLY); while (true) { db->num_insns++; -- 2.34.1 Let tb->page_addr[0] contain the address of the first byte of the translated block, rather than the address of the page containing the start of the translated block. We need to recover this value anyway at various points, and it is easier to discard a page offset when it is not needed, which happens naturally via the existing find_page shift. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/cpu-exec.c | 16 ++++++++-------- accel/tcg/cputlb.c | 3 ++- accel/tcg/translate-all.c | 9 +++++---- 3 files changed, 15 insertions(+), 13 deletions(-) diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cpu-exec.c +++ b/accel/tcg/cpu-exec.c @@ -XXX,XX +XXX,XX @@ struct tb_desc { target_ulong pc; target_ulong cs_base; CPUArchState *env; - tb_page_addr_t phys_page1; + tb_page_addr_t page_addr0; uint32_t flags; uint32_t cflags; uint32_t trace_vcpu_dstate; @@ -XXX,XX +XXX,XX @@ static bool tb_lookup_cmp(const void *p, const void *d) const struct tb_desc *desc = d; if (tb->pc == desc->pc && - tb->page_addr[0] == desc->phys_page1 && + tb->page_addr[0] == desc->page_addr0 && tb->cs_base == desc->cs_base && tb->flags == desc->flags && tb->trace_vcpu_dstate == desc->trace_vcpu_dstate && @@ -XXX,XX +XXX,XX @@ static bool tb_lookup_cmp(const void *p, const void *d) if (tb->page_addr[1] == -1) { return true; } else { - tb_page_addr_t phys_page2; - target_ulong virt_page2; + tb_page_addr_t phys_page1; + target_ulong virt_page1; /* * We know that the first page matched, and an otherwise valid TB @@ -XXX,XX +XXX,XX @@ static bool tb_lookup_cmp(const void *p, const void *d) * is different for the new TB. Therefore any exception raised * here by the faulting lookup is not premature. */ - virt_page2 = TARGET_PAGE_ALIGN(desc->pc); - phys_page2 = get_page_addr_code(desc->env, virt_page2); - if (tb->page_addr[1] == phys_page2) { + virt_page1 = TARGET_PAGE_ALIGN(desc->pc); + phys_page1 = get_page_addr_code(desc->env, virt_page1); + if (tb->page_addr[1] == phys_page1) { return true; } } @@ -XXX,XX +XXX,XX @@ static TranslationBlock *tb_htable_lookup(CPUState *cpu, target_ulong pc, if (phys_pc == -1) { return NULL; } - desc.phys_page1 = phys_pc & TARGET_PAGE_MASK; + desc.page_addr0 = phys_pc; h = tb_hash_func(phys_pc, pc, flags, cflags, *cpu->trace_dstate); return qht_lookup_custom(&tb_ctx.htable, &desc, h, tb_lookup_cmp); } 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 @@ void tlb_flush_page_bits_by_mmuidx_all_cpus_synced(CPUState *src_cpu, can be detected */ void tlb_protect_code(ram_addr_t ram_addr) { - cpu_physical_memory_test_and_clear_dirty(ram_addr, TARGET_PAGE_SIZE, + cpu_physical_memory_test_and_clear_dirty(ram_addr & TARGET_PAGE_MASK, + TARGET_PAGE_SIZE, DIRTY_MEMORY_CODE); } 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 @@ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list) qemu_spin_unlock(&tb->jmp_lock); /* remove the TB from the hash list */ - phys_pc = tb->page_addr[0] + (tb->pc & ~TARGET_PAGE_MASK); + phys_pc = tb->page_addr[0]; h = tb_hash_func(phys_pc, tb->pc, tb->flags, orig_cflags, tb->trace_vcpu_dstate); if (!qht_remove(&tb_ctx.htable, tb, h)) { @@ -XXX,XX +XXX,XX @@ tb_link_page(TranslationBlock *tb, tb_page_addr_t phys_pc, * we can only insert TBs that are fully initialized. */ page_lock_pair(&p, phys_pc, &p2, phys_page2, true); - tb_page_add(p, tb, 0, phys_pc & TARGET_PAGE_MASK); + tb_page_add(p, tb, 0, phys_pc); if (p2) { tb_page_add(p2, tb, 1, phys_page2); } else { @@ -XXX,XX +XXX,XX @@ tb_invalidate_phys_page_range__locked(struct page_collection *pages, if (n == 0) { /* NOTE: tb_end may be after the end of the page, but it is not a problem */ - tb_start = tb->page_addr[0] + (tb->pc & ~TARGET_PAGE_MASK); + tb_start = tb->page_addr[0]; tb_end = tb_start + tb->size; } else { tb_start = tb->page_addr[1]; - tb_end = tb_start + ((tb->pc + tb->size) & ~TARGET_PAGE_MASK); + tb_end = tb_start + ((tb->page_addr[0] + tb->size) + & ~TARGET_PAGE_MASK); } if (!(tb_end <= start || tb_start >= end)) { #ifdef TARGET_HAS_PRECISE_SMC -- 2.34.1 This function has two users, who use it incompatibly. In tlb_flush_page_by_mmuidx_async_0, when flushing a single page, we need to flush exactly two pages. In tlb_flush_range_by_mmuidx_async_0, when flushing a range of pages, we need to flush N+1 pages. This avoids double-flushing of jmp cache pages in a range. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/cputlb.c | 25 ++++++++++++++----------- 1 file changed, 14 insertions(+), 11 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 @@ static void tb_jmp_cache_clear_page(CPUState *cpu, target_ulong page_addr) } } -static void tb_flush_jmp_cache(CPUState *cpu, target_ulong addr) -{ - /* Discard jump cache entries for any tb which might potentially - overlap the flushed page. */ - tb_jmp_cache_clear_page(cpu, addr - TARGET_PAGE_SIZE); - tb_jmp_cache_clear_page(cpu, addr); -} - /** * tlb_mmu_resize_locked() - perform TLB resize bookkeeping; resize if necessary * @desc: The CPUTLBDesc portion of the TLB @@ -XXX,XX +XXX,XX @@ static void tlb_flush_page_by_mmuidx_async_0(CPUState *cpu, } qemu_spin_unlock(&env_tlb(env)->c.lock); - tb_flush_jmp_cache(cpu, addr); + /* + * Discard jump cache entries for any tb which might potentially + * overlap the flushed page, which includes the previous. + */ + tb_jmp_cache_clear_page(cpu, addr - TARGET_PAGE_SIZE); + tb_jmp_cache_clear_page(cpu, addr); } /** @@ -XXX,XX +XXX,XX @@ static void tlb_flush_range_by_mmuidx_async_0(CPUState *cpu, return; } - for (target_ulong i = 0; i < d.len; i += TARGET_PAGE_SIZE) { - tb_flush_jmp_cache(cpu, d.addr + i); + /* + * Discard jump cache entries for any tb which might potentially + * overlap the flushed pages, which includes the previous. + */ + d.addr -= TARGET_PAGE_SIZE; + for (target_ulong i = 0, n = d.len / TARGET_PAGE_SIZE + 1; i < n; i++) { + tb_jmp_cache_clear_page(cpu, d.addr); + d.addr += TARGET_PAGE_SIZE; } } -- 2.34.1 Wrap the bare TranslationBlock pointer into a structure. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/tb-hash.h | 1 + accel/tcg/tb-jmp-cache.h | 24 ++++++++++++++++++++++++ include/exec/cpu-common.h | 1 + include/hw/core/cpu.h | 15 +-------------- include/qemu/typedefs.h | 1 + accel/stubs/tcg-stub.c | 4 ++++ accel/tcg/cpu-exec.c | 10 +++++++--- accel/tcg/cputlb.c | 9 +++++---- accel/tcg/translate-all.c | 28 +++++++++++++++++++++++++--- hw/core/cpu-common.c | 3 +-- plugins/core.c | 2 +- trace/control-target.c | 2 +- 12 files changed, 72 insertions(+), 28 deletions(-) create mode 100644 accel/tcg/tb-jmp-cache.h diff --git a/accel/tcg/tb-hash.h b/accel/tcg/tb-hash.h index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/tb-hash.h +++ b/accel/tcg/tb-hash.h @@ -XXX,XX +XXX,XX @@ #include "exec/cpu-defs.h" #include "exec/exec-all.h" #include "qemu/xxhash.h" +#include "tb-jmp-cache.h" #ifdef CONFIG_SOFTMMU diff --git a/accel/tcg/tb-jmp-cache.h b/accel/tcg/tb-jmp-cache.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/accel/tcg/tb-jmp-cache.h @@ -XXX,XX +XXX,XX @@ +/* + * The per-CPU TranslationBlock jump cache. + * + * Copyright (c) 2003 Fabrice Bellard + * + * SPDX-License-Identifier: GPL-2.0-or-later + */ + +#ifndef ACCEL_TCG_TB_JMP_CACHE_H +#define ACCEL_TCG_TB_JMP_CACHE_H + +#define TB_JMP_CACHE_BITS 12 +#define TB_JMP_CACHE_SIZE (1 << TB_JMP_CACHE_BITS) + +/* + * Accessed in parallel; all accesses to 'tb' must be atomic. + */ +struct CPUJumpCache { + struct { + TranslationBlock *tb; + } array[TB_JMP_CACHE_SIZE]; +}; + +#endif /* ACCEL_TCG_TB_JMP_CACHE_H */ diff --git a/include/exec/cpu-common.h b/include/exec/cpu-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/cpu-common.h +++ b/include/exec/cpu-common.h @@ -XXX,XX +XXX,XX @@ void cpu_list_unlock(void); unsigned int cpu_list_generation_id_get(void); void tcg_flush_softmmu_tlb(CPUState *cs); +void tcg_flush_jmp_cache(CPUState *cs); void tcg_iommu_init_notifier_list(CPUState *cpu); void tcg_iommu_free_notifier_list(CPUState *cpu); diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/core/cpu.h +++ b/include/hw/core/cpu.h @@ -XXX,XX +XXX,XX @@ struct kvm_run; struct hax_vcpu_state; struct hvf_vcpu_state; -#define TB_JMP_CACHE_BITS 12 -#define TB_JMP_CACHE_SIZE (1 << TB_JMP_CACHE_BITS) - /* work queue */ /* The union type allows passing of 64 bit target pointers on 32 bit @@ -XXX,XX +XXX,XX @@ struct CPUState { CPUArchState *env_ptr; IcountDecr *icount_decr_ptr; - /* Accessed in parallel; all accesses must be atomic */ - TranslationBlock *tb_jmp_cache[TB_JMP_CACHE_SIZE]; + CPUJumpCache *tb_jmp_cache; struct GDBRegisterState *gdb_regs; int gdb_num_regs; @@ -XXX,XX +XXX,XX @@ extern CPUTailQ cpus; extern __thread CPUState *current_cpu; -static inline void cpu_tb_jmp_cache_clear(CPUState *cpu) -{ - unsigned int i; - - for (i = 0; i < TB_JMP_CACHE_SIZE; i++) { - qatomic_set(&cpu->tb_jmp_cache[i], NULL); - } -} - /** * qemu_tcg_mttcg_enabled: * Check whether we are running MultiThread TCG or not. diff --git a/include/qemu/typedefs.h b/include/qemu/typedefs.h index XXXXXXX..XXXXXXX 100644 --- a/include/qemu/typedefs.h +++ b/include/qemu/typedefs.h @@ -XXX,XX +XXX,XX @@ typedef struct CoMutex CoMutex; typedef struct ConfidentialGuestSupport ConfidentialGuestSupport; typedef struct CPUAddressSpace CPUAddressSpace; typedef struct CPUArchState CPUArchState; +typedef struct CPUJumpCache CPUJumpCache; typedef struct CPUState CPUState; typedef struct CPUTLBEntryFull CPUTLBEntryFull; typedef struct DeviceListener DeviceListener; diff --git a/accel/stubs/tcg-stub.c b/accel/stubs/tcg-stub.c index XXXXXXX..XXXXXXX 100644 --- a/accel/stubs/tcg-stub.c +++ b/accel/stubs/tcg-stub.c @@ -XXX,XX +XXX,XX @@ void tlb_set_dirty(CPUState *cpu, target_ulong vaddr) { } +void tcg_flush_jmp_cache(CPUState *cpu) +{ +} + int probe_access_flags(CPUArchState *env, target_ulong addr, MMUAccessType access_type, int mmu_idx, bool nonfault, void **phost, uintptr_t retaddr) diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cpu-exec.c +++ b/accel/tcg/cpu-exec.c @@ -XXX,XX +XXX,XX @@ #include "sysemu/replay.h" #include "sysemu/tcg.h" #include "exec/helper-proto.h" +#include "tb-jmp-cache.h" #include "tb-hash.h" #include "tb-context.h" #include "internal.h" @@ -XXX,XX +XXX,XX @@ static inline TranslationBlock *tb_lookup(CPUState *cpu, target_ulong pc, tcg_debug_assert(!(cflags & CF_INVALID)); hash = tb_jmp_cache_hash_func(pc); - tb = qatomic_rcu_read(&cpu->tb_jmp_cache[hash]); + tb = qatomic_rcu_read(&cpu->tb_jmp_cache->array[hash].tb); if (likely(tb && tb->pc == pc && @@ -XXX,XX +XXX,XX @@ static inline TranslationBlock *tb_lookup(CPUState *cpu, target_ulong pc, if (tb == NULL) { return NULL; } - qatomic_set(&cpu->tb_jmp_cache[hash], tb); + qatomic_set(&cpu->tb_jmp_cache->array[hash].tb, tb); return tb; } @@ -XXX,XX +XXX,XX @@ int cpu_exec(CPUState *cpu) tb = tb_lookup(cpu, pc, cs_base, flags, cflags); if (tb == NULL) { + uint32_t h; + mmap_lock(); tb = tb_gen_code(cpu, pc, cs_base, flags, cflags); mmap_unlock(); @@ -XXX,XX +XXX,XX @@ int cpu_exec(CPUState *cpu) * We add the TB in the virtual pc hash table * for the fast lookup */ - qatomic_set(&cpu->tb_jmp_cache[tb_jmp_cache_hash_func(pc)], tb); + h = tb_jmp_cache_hash_func(pc); + qatomic_set(&cpu->tb_jmp_cache->array[h].tb, tb); } #ifndef 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 @@ static void tlb_window_reset(CPUTLBDesc *desc, int64_t ns, static void tb_jmp_cache_clear_page(CPUState *cpu, target_ulong page_addr) { - unsigned int i, i0 = tb_jmp_cache_hash_page(page_addr); + int i, i0 = tb_jmp_cache_hash_page(page_addr); + CPUJumpCache *jc = cpu->tb_jmp_cache; for (i = 0; i < TB_JMP_PAGE_SIZE; i++) { - qatomic_set(&cpu->tb_jmp_cache[i0 + i], NULL); + qatomic_set(&jc->array[i0 + i].tb, NULL); } } @@ -XXX,XX +XXX,XX @@ static void tlb_flush_by_mmuidx_async_work(CPUState *cpu, run_on_cpu_data data) qemu_spin_unlock(&env_tlb(env)->c.lock); - cpu_tb_jmp_cache_clear(cpu); + tcg_flush_jmp_cache(cpu); if (to_clean == ALL_MMUIDX_BITS) { qatomic_set(&env_tlb(env)->c.full_flush_count, @@ -XXX,XX +XXX,XX @@ static void tlb_flush_range_by_mmuidx_async_0(CPUState *cpu, * longer to clear each entry individually than it will to clear it all. */ if (d.len >= (TARGET_PAGE_SIZE * TB_JMP_CACHE_SIZE)) { - cpu_tb_jmp_cache_clear(cpu); + tcg_flush_jmp_cache(cpu); return; } 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 @@ #include "sysemu/tcg.h" #include "qapi/error.h" #include "hw/core/tcg-cpu-ops.h" +#include "tb-jmp-cache.h" #include "tb-hash.h" #include "tb-context.h" #include "internal.h" @@ -XXX,XX +XXX,XX @@ static void do_tb_flush(CPUState *cpu, run_on_cpu_data tb_flush_count) } CPU_FOREACH(cpu) { - cpu_tb_jmp_cache_clear(cpu); + tcg_flush_jmp_cache(cpu); } qht_reset_size(&tb_ctx.htable, CODE_GEN_HTABLE_SIZE); @@ -XXX,XX +XXX,XX @@ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list) /* remove the TB from the hash list */ h = tb_jmp_cache_hash_func(tb->pc); CPU_FOREACH(cpu) { - if (qatomic_read(&cpu->tb_jmp_cache[h]) == tb) { - qatomic_set(&cpu->tb_jmp_cache[h], NULL); + CPUJumpCache *jc = cpu->tb_jmp_cache; + if (qatomic_read(&jc->array[h].tb) == tb) { + qatomic_set(&jc->array[h].tb, NULL); } } @@ -XXX,XX +XXX,XX @@ int page_unprotect(target_ulong address, uintptr_t pc) } #endif /* CONFIG_USER_ONLY */ +/* + * Called by generic code at e.g. cpu reset after cpu creation, + * therefore we must be prepared to allocate the jump cache. + */ +void tcg_flush_jmp_cache(CPUState *cpu) +{ + CPUJumpCache *jc = cpu->tb_jmp_cache; + + if (likely(jc)) { + for (int i = 0; i < TB_JMP_CACHE_SIZE; i++) { + qatomic_set(&jc->array[i].tb, NULL); + } + } else { + /* This should happen once during realize, and thus never race. */ + jc = g_new0(CPUJumpCache, 1); + jc = qatomic_xchg(&cpu->tb_jmp_cache, jc); + assert(jc == NULL); + } +} + /* This is a wrapper for common code that can not use CONFIG_SOFTMMU */ void tcg_flush_softmmu_tlb(CPUState *cs) { diff --git a/hw/core/cpu-common.c b/hw/core/cpu-common.c index XXXXXXX..XXXXXXX 100644 --- a/hw/core/cpu-common.c +++ b/hw/core/cpu-common.c @@ -XXX,XX +XXX,XX @@ static void cpu_common_reset(DeviceState *dev) cpu->cflags_next_tb = -1; if (tcg_enabled()) { - cpu_tb_jmp_cache_clear(cpu); - + tcg_flush_jmp_cache(cpu); tcg_flush_softmmu_tlb(cpu); } } diff --git a/plugins/core.c b/plugins/core.c index XXXXXXX..XXXXXXX 100644 --- a/plugins/core.c +++ b/plugins/core.c @@ -XXX,XX +XXX,XX @@ struct qemu_plugin_ctx *plugin_id_to_ctx_locked(qemu_plugin_id_t id) static void plugin_cpu_update__async(CPUState *cpu, run_on_cpu_data data) { bitmap_copy(cpu->plugin_mask, &data.host_ulong, QEMU_PLUGIN_EV_MAX); - cpu_tb_jmp_cache_clear(cpu); + tcg_flush_jmp_cache(cpu); } static void plugin_cpu_update__locked(gpointer k, gpointer v, gpointer udata) diff --git a/trace/control-target.c b/trace/control-target.c index XXXXXXX..XXXXXXX 100644 --- a/trace/control-target.c +++ b/trace/control-target.c @@ -XXX,XX +XXX,XX @@ static void trace_event_synchronize_vcpu_state_dynamic( { bitmap_copy(vcpu->trace_dstate, vcpu->trace_dstate_delayed, CPU_TRACE_DSTATE_MAX_EVENTS); - cpu_tb_jmp_cache_clear(vcpu); + tcg_flush_jmp_cache(vcpu); } void trace_event_set_vcpu_state_dynamic(CPUState *vcpu, -- 2.34.1 Populate this new method for all targets. Always match the result that would be given by cpu_get_tb_cpu_state, as we will want these values to correspond in the logs. Reviewed-by: Taylor Simpson <tsimpson@quicinc.com> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> (target/sparc) Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- Cc: Eduardo Habkost <eduardo@habkost.net> (supporter:Machine core) Cc: Marcel Apfelbaum <marcel.apfelbaum@gmail.com> (supporter:Machine core) Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org> (reviewer:Machine core) Cc: Yanan Wang <wangyanan55@huawei.com> (reviewer:Machine core) Cc: Michael Rolnik <mrolnik@gmail.com> (maintainer:AVR TCG CPUs) Cc: "Edgar E. Iglesias" <edgar.iglesias@gmail.com> (maintainer:CRIS TCG CPUs) Cc: Taylor Simpson <tsimpson@quicinc.com> (supporter:Hexagon TCG CPUs) Cc: Song Gao <gaosong@loongson.cn> (maintainer:LoongArch TCG CPUs) Cc: Xiaojuan Yang <yangxiaojuan@loongson.cn> (maintainer:LoongArch TCG CPUs) Cc: Laurent Vivier <laurent@vivier.eu> (maintainer:M68K TCG CPUs) Cc: Jiaxun Yang <jiaxun.yang@flygoat.com> (reviewer:MIPS TCG CPUs) Cc: Aleksandar Rikalo <aleksandar.rikalo@syrmia.com> (reviewer:MIPS TCG CPUs) Cc: Chris Wulff <crwulff@gmail.com> (maintainer:NiosII TCG CPUs) Cc: Marek Vasut <marex@denx.de> (maintainer:NiosII TCG CPUs) Cc: Stafford Horne <shorne@gmail.com> (odd fixer:OpenRISC TCG CPUs) Cc: Yoshinori Sato <ysato@users.sourceforge.jp> (reviewer:RENESAS RX CPUs) Cc: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> (maintainer:SPARC TCG CPUs) Cc: Bastian Koppelmann <kbastian@mail.uni-paderborn.de> (maintainer:TriCore TCG CPUs) Cc: Max Filippov <jcmvbkbc@gmail.com> (maintainer:Xtensa TCG CPUs) Cc: qemu-arm@nongnu.org (open list:ARM TCG CPUs) Cc: qemu-ppc@nongnu.org (open list:PowerPC TCG CPUs) Cc: qemu-riscv@nongnu.org (open list:RISC-V TCG CPUs) Cc: qemu-s390x@nongnu.org (open list:S390 TCG CPUs) --- include/hw/core/cpu.h | 3 +++ target/alpha/cpu.c | 9 +++++++++ target/arm/cpu.c | 13 +++++++++++++ target/avr/cpu.c | 8 ++++++++ target/cris/cpu.c | 8 ++++++++ target/hexagon/cpu.c | 8 ++++++++ target/hppa/cpu.c | 8 ++++++++ target/i386/cpu.c | 9 +++++++++ target/loongarch/cpu.c | 9 +++++++++ target/m68k/cpu.c | 8 ++++++++ target/microblaze/cpu.c | 8 ++++++++ target/mips/cpu.c | 8 ++++++++ target/nios2/cpu.c | 9 +++++++++ target/openrisc/cpu.c | 8 ++++++++ target/ppc/cpu_init.c | 8 ++++++++ target/riscv/cpu.c | 13 +++++++++++++ target/rx/cpu.c | 8 ++++++++ target/s390x/cpu.c | 8 ++++++++ target/sh4/cpu.c | 8 ++++++++ target/sparc/cpu.c | 8 ++++++++ target/tricore/cpu.c | 9 +++++++++ target/xtensa/cpu.c | 8 ++++++++ 22 files changed, 186 insertions(+) diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/core/cpu.h +++ b/include/hw/core/cpu.h @@ -XXX,XX +XXX,XX @@ struct SysemuCPUOps; * If the target behaviour here is anything other than "set * the PC register to the value passed in" then the target must * also implement the synchronize_from_tb hook. + * @get_pc: Callback for getting the Program Counter register. + * As above, with the semantics of the target architecture. * @gdb_read_register: Callback for letting GDB read a register. * @gdb_write_register: Callback for letting GDB write a register. * @gdb_adjust_breakpoint: Callback for adjusting the address of a @@ -XXX,XX +XXX,XX @@ struct CPUClass { void (*dump_state)(CPUState *cpu, FILE *, int flags); int64_t (*get_arch_id)(CPUState *cpu); void (*set_pc)(CPUState *cpu, vaddr value); + vaddr (*get_pc)(CPUState *cpu); int (*gdb_read_register)(CPUState *cpu, GByteArray *buf, int reg); int (*gdb_write_register)(CPUState *cpu, uint8_t *buf, int reg); vaddr (*gdb_adjust_breakpoint)(CPUState *cpu, vaddr addr); diff --git a/target/alpha/cpu.c b/target/alpha/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/alpha/cpu.c +++ b/target/alpha/cpu.c @@ -XXX,XX +XXX,XX @@ static void alpha_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.pc = value; } +static vaddr alpha_cpu_get_pc(CPUState *cs) +{ + AlphaCPU *cpu = ALPHA_CPU(cs); + + return cpu->env.pc; +} + + static bool alpha_cpu_has_work(CPUState *cs) { /* Here we are checking to see if the CPU should wake up from HALT. @@ -XXX,XX +XXX,XX @@ static void alpha_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = alpha_cpu_has_work; cc->dump_state = alpha_cpu_dump_state; cc->set_pc = alpha_cpu_set_pc; + cc->get_pc = alpha_cpu_get_pc; cc->gdb_read_register = alpha_cpu_gdb_read_register; cc->gdb_write_register = alpha_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_set_pc(CPUState *cs, vaddr value) } } +static vaddr arm_cpu_get_pc(CPUState *cs) +{ + ARMCPU *cpu = ARM_CPU(cs); + CPUARMState *env = &cpu->env; + + if (is_a64(env)) { + return env->pc; + } else { + return env->regs[15]; + } +} + #ifdef CONFIG_TCG void arm_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) @@ -XXX,XX +XXX,XX @@ static void arm_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = arm_cpu_has_work; cc->dump_state = arm_cpu_dump_state; cc->set_pc = arm_cpu_set_pc; + cc->get_pc = arm_cpu_get_pc; cc->gdb_read_register = arm_cpu_gdb_read_register; cc->gdb_write_register = arm_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/avr/cpu.c b/target/avr/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/avr/cpu.c +++ b/target/avr/cpu.c @@ -XXX,XX +XXX,XX @@ static void avr_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.pc_w = value / 2; /* internally PC points to words */ } +static vaddr avr_cpu_get_pc(CPUState *cs) +{ + AVRCPU *cpu = AVR_CPU(cs); + + return cpu->env.pc_w * 2; +} + static bool avr_cpu_has_work(CPUState *cs) { AVRCPU *cpu = AVR_CPU(cs); @@ -XXX,XX +XXX,XX @@ static void avr_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = avr_cpu_has_work; cc->dump_state = avr_cpu_dump_state; cc->set_pc = avr_cpu_set_pc; + cc->get_pc = avr_cpu_get_pc; dc->vmsd = &vms_avr_cpu; cc->sysemu_ops = &avr_sysemu_ops; cc->disas_set_info = avr_cpu_disas_set_info; diff --git a/target/cris/cpu.c b/target/cris/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/cris/cpu.c +++ b/target/cris/cpu.c @@ -XXX,XX +XXX,XX @@ static void cris_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.pc = value; } +static vaddr cris_cpu_get_pc(CPUState *cs) +{ + CRISCPU *cpu = CRIS_CPU(cs); + + return cpu->env.pc; +} + static bool cris_cpu_has_work(CPUState *cs) { return cs->interrupt_request & (CPU_INTERRUPT_HARD | CPU_INTERRUPT_NMI); @@ -XXX,XX +XXX,XX @@ static void cris_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = cris_cpu_has_work; cc->dump_state = cris_cpu_dump_state; cc->set_pc = cris_cpu_set_pc; + cc->get_pc = cris_cpu_get_pc; cc->gdb_read_register = cris_cpu_gdb_read_register; cc->gdb_write_register = cris_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/hexagon/cpu.c b/target/hexagon/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/hexagon/cpu.c +++ b/target/hexagon/cpu.c @@ -XXX,XX +XXX,XX @@ static void hexagon_cpu_set_pc(CPUState *cs, vaddr value) env->gpr[HEX_REG_PC] = value; } +static vaddr hexagon_cpu_get_pc(CPUState *cs) +{ + HexagonCPU *cpu = HEXAGON_CPU(cs); + CPUHexagonState *env = &cpu->env; + return env->gpr[HEX_REG_PC]; +} + static void hexagon_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void hexagon_cpu_class_init(ObjectClass *c, void *data) cc->has_work = hexagon_cpu_has_work; cc->dump_state = hexagon_dump_state; cc->set_pc = hexagon_cpu_set_pc; + cc->get_pc = hexagon_cpu_get_pc; cc->gdb_read_register = hexagon_gdb_read_register; cc->gdb_write_register = hexagon_gdb_write_register; cc->gdb_num_core_regs = TOTAL_PER_THREAD_REGS + NUM_VREGS + NUM_QREGS; diff --git a/target/hppa/cpu.c b/target/hppa/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/hppa/cpu.c +++ b/target/hppa/cpu.c @@ -XXX,XX +XXX,XX @@ static void hppa_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.iaoq_b = value + 4; } +static vaddr hppa_cpu_get_pc(CPUState *cs) +{ + HPPACPU *cpu = HPPA_CPU(cs); + + return cpu->env.iaoq_f; +} + static void hppa_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void hppa_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = hppa_cpu_has_work; cc->dump_state = hppa_cpu_dump_state; cc->set_pc = hppa_cpu_set_pc; + cc->get_pc = hppa_cpu_get_pc; cc->gdb_read_register = hppa_cpu_gdb_read_register; cc->gdb_write_register = hppa_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/i386/cpu.c b/target/i386/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/i386/cpu.c +++ b/target/i386/cpu.c @@ -XXX,XX +XXX,XX @@ static void x86_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.eip = value; } +static vaddr x86_cpu_get_pc(CPUState *cs) +{ + X86CPU *cpu = X86_CPU(cs); + + /* Match cpu_get_tb_cpu_state. */ + return cpu->env.eip + cpu->env.segs[R_CS].base; +} + int x86_cpu_pending_interrupt(CPUState *cs, int interrupt_request) { X86CPU *cpu = X86_CPU(cs); @@ -XXX,XX +XXX,XX @@ static void x86_cpu_common_class_init(ObjectClass *oc, void *data) cc->has_work = x86_cpu_has_work; cc->dump_state = x86_cpu_dump_state; cc->set_pc = x86_cpu_set_pc; + cc->get_pc = x86_cpu_get_pc; cc->gdb_read_register = x86_cpu_gdb_read_register; cc->gdb_write_register = x86_cpu_gdb_write_register; cc->get_arch_id = x86_cpu_get_arch_id; diff --git a/target/loongarch/cpu.c b/target/loongarch/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/loongarch/cpu.c +++ b/target/loongarch/cpu.c @@ -XXX,XX +XXX,XX @@ static void loongarch_cpu_set_pc(CPUState *cs, vaddr value) env->pc = value; } +static vaddr loongarch_cpu_get_pc(CPUState *cs) +{ + LoongArchCPU *cpu = LOONGARCH_CPU(cs); + CPULoongArchState *env = &cpu->env; + + return env->pc; +} + #ifndef CONFIG_USER_ONLY #include "hw/loongarch/virt.h" @@ -XXX,XX +XXX,XX @@ static void loongarch_cpu_class_init(ObjectClass *c, void *data) cc->has_work = loongarch_cpu_has_work; cc->dump_state = loongarch_cpu_dump_state; cc->set_pc = loongarch_cpu_set_pc; + cc->get_pc = loongarch_cpu_get_pc; #ifndef CONFIG_USER_ONLY dc->vmsd = &vmstate_loongarch_cpu; cc->sysemu_ops = &loongarch_sysemu_ops; diff --git a/target/m68k/cpu.c b/target/m68k/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/m68k/cpu.c +++ b/target/m68k/cpu.c @@ -XXX,XX +XXX,XX @@ static void m68k_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.pc = value; } +static vaddr m68k_cpu_get_pc(CPUState *cs) +{ + M68kCPU *cpu = M68K_CPU(cs); + + return cpu->env.pc; +} + static bool m68k_cpu_has_work(CPUState *cs) { return cs->interrupt_request & CPU_INTERRUPT_HARD; @@ -XXX,XX +XXX,XX @@ static void m68k_cpu_class_init(ObjectClass *c, void *data) cc->has_work = m68k_cpu_has_work; cc->dump_state = m68k_cpu_dump_state; cc->set_pc = m68k_cpu_set_pc; + cc->get_pc = m68k_cpu_get_pc; cc->gdb_read_register = m68k_cpu_gdb_read_register; cc->gdb_write_register = m68k_cpu_gdb_write_register; #if defined(CONFIG_SOFTMMU) diff --git a/target/microblaze/cpu.c b/target/microblaze/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/microblaze/cpu.c +++ b/target/microblaze/cpu.c @@ -XXX,XX +XXX,XX @@ static void mb_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.iflags = 0; } +static vaddr mb_cpu_get_pc(CPUState *cs) +{ + MicroBlazeCPU *cpu = MICROBLAZE_CPU(cs); + + return cpu->env.pc; +} + static void mb_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void mb_cpu_class_init(ObjectClass *oc, void *data) cc->dump_state = mb_cpu_dump_state; cc->set_pc = mb_cpu_set_pc; + cc->get_pc = mb_cpu_get_pc; cc->gdb_read_register = mb_cpu_gdb_read_register; cc->gdb_write_register = mb_cpu_gdb_write_register; diff --git a/target/mips/cpu.c b/target/mips/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/mips/cpu.c +++ b/target/mips/cpu.c @@ -XXX,XX +XXX,XX @@ static void mips_cpu_set_pc(CPUState *cs, vaddr value) mips_env_set_pc(&cpu->env, value); } +static vaddr mips_cpu_get_pc(CPUState *cs) +{ + MIPSCPU *cpu = MIPS_CPU(cs); + + return cpu->env.active_tc.PC; +} + static bool mips_cpu_has_work(CPUState *cs) { MIPSCPU *cpu = MIPS_CPU(cs); @@ -XXX,XX +XXX,XX @@ static void mips_cpu_class_init(ObjectClass *c, void *data) cc->has_work = mips_cpu_has_work; cc->dump_state = mips_cpu_dump_state; cc->set_pc = mips_cpu_set_pc; + cc->get_pc = mips_cpu_get_pc; cc->gdb_read_register = mips_cpu_gdb_read_register; cc->gdb_write_register = mips_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/nios2/cpu.c b/target/nios2/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/nios2/cpu.c +++ b/target/nios2/cpu.c @@ -XXX,XX +XXX,XX @@ static void nios2_cpu_set_pc(CPUState *cs, vaddr value) env->pc = value; } +static vaddr nios2_cpu_get_pc(CPUState *cs) +{ + Nios2CPU *cpu = NIOS2_CPU(cs); + CPUNios2State *env = &cpu->env; + + return env->pc; +} + static bool nios2_cpu_has_work(CPUState *cs) { return cs->interrupt_request & CPU_INTERRUPT_HARD; @@ -XXX,XX +XXX,XX @@ static void nios2_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = nios2_cpu_has_work; cc->dump_state = nios2_cpu_dump_state; cc->set_pc = nios2_cpu_set_pc; + cc->get_pc = nios2_cpu_get_pc; cc->disas_set_info = nios2_cpu_disas_set_info; #ifndef CONFIG_USER_ONLY cc->sysemu_ops = &nios2_sysemu_ops; diff --git a/target/openrisc/cpu.c b/target/openrisc/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/openrisc/cpu.c +++ b/target/openrisc/cpu.c @@ -XXX,XX +XXX,XX @@ static void openrisc_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.dflag = 0; } +static vaddr openrisc_cpu_get_pc(CPUState *cs) +{ + OpenRISCCPU *cpu = OPENRISC_CPU(cs); + + return cpu->env.pc; +} + static void openrisc_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void openrisc_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = openrisc_cpu_has_work; cc->dump_state = openrisc_cpu_dump_state; cc->set_pc = openrisc_cpu_set_pc; + cc->get_pc = openrisc_cpu_get_pc; cc->gdb_read_register = openrisc_cpu_gdb_read_register; cc->gdb_write_register = openrisc_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/ppc/cpu_init.c b/target/ppc/cpu_init.c index XXXXXXX..XXXXXXX 100644 --- a/target/ppc/cpu_init.c +++ b/target/ppc/cpu_init.c @@ -XXX,XX +XXX,XX @@ static void ppc_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.nip = value; } +static vaddr ppc_cpu_get_pc(CPUState *cs) +{ + PowerPCCPU *cpu = POWERPC_CPU(cs); + + return cpu->env.nip; +} + static bool ppc_cpu_has_work(CPUState *cs) { PowerPCCPU *cpu = POWERPC_CPU(cs); @@ -XXX,XX +XXX,XX @@ static void ppc_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = ppc_cpu_has_work; cc->dump_state = ppc_cpu_dump_state; cc->set_pc = ppc_cpu_set_pc; + cc->get_pc = ppc_cpu_get_pc; cc->gdb_read_register = ppc_cpu_gdb_read_register; cc->gdb_write_register = ppc_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/riscv/cpu.c +++ b/target/riscv/cpu.c @@ -XXX,XX +XXX,XX @@ static void riscv_cpu_set_pc(CPUState *cs, vaddr value) } } +static vaddr riscv_cpu_get_pc(CPUState *cs) +{ + RISCVCPU *cpu = RISCV_CPU(cs); + CPURISCVState *env = &cpu->env; + + /* Match cpu_get_tb_cpu_state. */ + if (env->xl == MXL_RV32) { + return env->pc & UINT32_MAX; + } + return env->pc; +} + static void riscv_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void riscv_cpu_class_init(ObjectClass *c, void *data) cc->has_work = riscv_cpu_has_work; cc->dump_state = riscv_cpu_dump_state; cc->set_pc = riscv_cpu_set_pc; + cc->get_pc = riscv_cpu_get_pc; cc->gdb_read_register = riscv_cpu_gdb_read_register; cc->gdb_write_register = riscv_cpu_gdb_write_register; cc->gdb_num_core_regs = 33; diff --git a/target/rx/cpu.c b/target/rx/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/rx/cpu.c +++ b/target/rx/cpu.c @@ -XXX,XX +XXX,XX @@ static void rx_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.pc = value; } +static vaddr rx_cpu_get_pc(CPUState *cs) +{ + RXCPU *cpu = RX_CPU(cs); + + return cpu->env.pc; +} + static void rx_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void rx_cpu_class_init(ObjectClass *klass, void *data) cc->has_work = rx_cpu_has_work; cc->dump_state = rx_cpu_dump_state; cc->set_pc = rx_cpu_set_pc; + cc->get_pc = rx_cpu_get_pc; #ifndef CONFIG_USER_ONLY cc->sysemu_ops = &rx_sysemu_ops; diff --git a/target/s390x/cpu.c b/target/s390x/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/s390x/cpu.c +++ b/target/s390x/cpu.c @@ -XXX,XX +XXX,XX @@ static void s390_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.psw.addr = value; } +static vaddr s390_cpu_get_pc(CPUState *cs) +{ + S390CPU *cpu = S390_CPU(cs); + + return cpu->env.psw.addr; +} + static bool s390_cpu_has_work(CPUState *cs) { S390CPU *cpu = S390_CPU(cs); @@ -XXX,XX +XXX,XX @@ static void s390_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = s390_cpu_has_work; cc->dump_state = s390_cpu_dump_state; cc->set_pc = s390_cpu_set_pc; + cc->get_pc = s390_cpu_get_pc; cc->gdb_read_register = s390_cpu_gdb_read_register; cc->gdb_write_register = s390_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/sh4/cpu.c b/target/sh4/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/sh4/cpu.c +++ b/target/sh4/cpu.c @@ -XXX,XX +XXX,XX @@ static void superh_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.pc = value; } +static vaddr superh_cpu_get_pc(CPUState *cs) +{ + SuperHCPU *cpu = SUPERH_CPU(cs); + + return cpu->env.pc; +} + static void superh_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void superh_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = superh_cpu_has_work; cc->dump_state = superh_cpu_dump_state; cc->set_pc = superh_cpu_set_pc; + cc->get_pc = superh_cpu_get_pc; cc->gdb_read_register = superh_cpu_gdb_read_register; cc->gdb_write_register = superh_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/sparc/cpu.c b/target/sparc/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/sparc/cpu.c +++ b/target/sparc/cpu.c @@ -XXX,XX +XXX,XX @@ static void sparc_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.npc = value + 4; } +static vaddr sparc_cpu_get_pc(CPUState *cs) +{ + SPARCCPU *cpu = SPARC_CPU(cs); + + return cpu->env.pc; +} + static void sparc_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void sparc_cpu_class_init(ObjectClass *oc, void *data) cc->memory_rw_debug = sparc_cpu_memory_rw_debug; #endif cc->set_pc = sparc_cpu_set_pc; + cc->get_pc = sparc_cpu_get_pc; cc->gdb_read_register = sparc_cpu_gdb_read_register; cc->gdb_write_register = sparc_cpu_gdb_write_register; #ifndef CONFIG_USER_ONLY diff --git a/target/tricore/cpu.c b/target/tricore/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/tricore/cpu.c +++ b/target/tricore/cpu.c @@ -XXX,XX +XXX,XX @@ static void tricore_cpu_set_pc(CPUState *cs, vaddr value) env->PC = value & ~(target_ulong)1; } +static vaddr tricore_cpu_get_pc(CPUState *cs) +{ + TriCoreCPU *cpu = TRICORE_CPU(cs); + CPUTriCoreState *env = &cpu->env; + + return env->PC; +} + static void tricore_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) { @@ -XXX,XX +XXX,XX @@ static void tricore_cpu_class_init(ObjectClass *c, void *data) cc->dump_state = tricore_cpu_dump_state; cc->set_pc = tricore_cpu_set_pc; + cc->get_pc = tricore_cpu_get_pc; cc->sysemu_ops = &tricore_sysemu_ops; cc->tcg_ops = &tricore_tcg_ops; } diff --git a/target/xtensa/cpu.c b/target/xtensa/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/xtensa/cpu.c +++ b/target/xtensa/cpu.c @@ -XXX,XX +XXX,XX @@ static void xtensa_cpu_set_pc(CPUState *cs, vaddr value) cpu->env.pc = value; } +static vaddr xtensa_cpu_get_pc(CPUState *cs) +{ + XtensaCPU *cpu = XTENSA_CPU(cs); + + return cpu->env.pc; +} + static bool xtensa_cpu_has_work(CPUState *cs) { #ifndef CONFIG_USER_ONLY @@ -XXX,XX +XXX,XX @@ static void xtensa_cpu_class_init(ObjectClass *oc, void *data) cc->has_work = xtensa_cpu_has_work; cc->dump_state = xtensa_cpu_dump_state; cc->set_pc = xtensa_cpu_set_pc; + cc->get_pc = xtensa_cpu_get_pc; cc->gdb_read_register = xtensa_cpu_gdb_read_register; cc->gdb_write_register = xtensa_cpu_gdb_write_register; cc->gdb_stop_before_watchpoint = true; -- 2.34.1 The availability of tb->pc will shortly be conditional. Introduce accessor functions to minimize ifdefs. Pass around a known pc to places like tcg_gen_code, where the caller must already have the value. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/internal.h | 6 ++++ include/exec/exec-all.h | 6 ++++ include/tcg/tcg.h | 2 +- accel/tcg/cpu-exec.c | 46 ++++++++++++++----------- accel/tcg/translate-all.c | 37 +++++++++++--------- target/arm/cpu.c | 4 +-- target/avr/cpu.c | 2 +- target/hexagon/cpu.c | 2 +- target/hppa/cpu.c | 4 +-- target/i386/tcg/tcg-cpu.c | 2 +- target/loongarch/cpu.c | 2 +- target/microblaze/cpu.c | 2 +- target/mips/tcg/exception.c | 2 +- target/mips/tcg/sysemu/special_helper.c | 2 +- target/openrisc/cpu.c | 2 +- target/riscv/cpu.c | 4 +-- target/rx/cpu.c | 2 +- target/sh4/cpu.c | 4 +-- target/sparc/cpu.c | 2 +- target/tricore/cpu.c | 2 +- tcg/tcg.c | 8 ++--- 21 files changed, 82 insertions(+), 61 deletions(-) diff --git a/accel/tcg/internal.h b/accel/tcg/internal.h index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/internal.h +++ b/accel/tcg/internal.h @@ -XXX,XX +XXX,XX @@ G_NORETURN void cpu_io_recompile(CPUState *cpu, uintptr_t retaddr); void page_init(void); void tb_htable_init(void); +/* Return the current PC from CPU, which may be cached in TB. */ +static inline target_ulong log_pc(CPUState *cpu, const TranslationBlock *tb) +{ + return tb_pc(tb); +} + #endif /* ACCEL_TCG_INTERNAL_H */ diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/exec-all.h +++ b/include/exec/exec-all.h @@ -XXX,XX +XXX,XX @@ struct TranslationBlock { uintptr_t jmp_dest[2]; }; +/* Hide the read to avoid ifdefs for TARGET_TB_PCREL. */ +static inline target_ulong tb_pc(const TranslationBlock *tb) +{ + return tb->pc; +} + /* Hide the qatomic_read to make code a little easier on the eyes */ static inline uint32_t tb_cflags(const TranslationBlock *tb) { 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); +int tcg_gen_code(TCGContext *s, TranslationBlock *tb, target_ulong pc_start); void tcg_set_frame(TCGContext *s, TCGReg reg, intptr_t start, intptr_t size); diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cpu-exec.c +++ b/accel/tcg/cpu-exec.c @@ -XXX,XX +XXX,XX @@ static bool tb_lookup_cmp(const void *p, const void *d) const TranslationBlock *tb = p; const struct tb_desc *desc = d; - if (tb->pc == desc->pc && + if (tb_pc(tb) == desc->pc && tb->page_addr[0] == desc->page_addr0 && tb->cs_base == desc->cs_base && tb->flags == desc->flags && @@ -XXX,XX +XXX,XX @@ static inline TranslationBlock *tb_lookup(CPUState *cpu, target_ulong pc, return tb; } -static inline void log_cpu_exec(target_ulong pc, CPUState *cpu, - const TranslationBlock *tb) +static void log_cpu_exec(target_ulong pc, CPUState *cpu, + const TranslationBlock *tb) { - if (unlikely(qemu_loglevel_mask(CPU_LOG_TB_CPU | CPU_LOG_EXEC)) - && qemu_log_in_addr_range(pc)) { - + if (qemu_log_in_addr_range(pc)) { qemu_log_mask(CPU_LOG_EXEC, "Trace %d: %p [" TARGET_FMT_lx "/" TARGET_FMT_lx "/%08x/%08x] %s\n", @@ -XXX,XX +XXX,XX @@ const void *HELPER(lookup_tb_ptr)(CPUArchState *env) return tcg_code_gen_epilogue; } - log_cpu_exec(pc, cpu, tb); + if (qemu_loglevel_mask(CPU_LOG_TB_CPU | CPU_LOG_EXEC)) { + log_cpu_exec(pc, cpu, tb); + } return tb->tc.ptr; } @@ -XXX,XX +XXX,XX @@ cpu_tb_exec(CPUState *cpu, TranslationBlock *itb, int *tb_exit) TranslationBlock *last_tb; const void *tb_ptr = itb->tc.ptr; - log_cpu_exec(itb->pc, cpu, itb); + if (qemu_loglevel_mask(CPU_LOG_TB_CPU | CPU_LOG_EXEC)) { + log_cpu_exec(log_pc(cpu, itb), cpu, itb); + } qemu_thread_jit_execute(); ret = tcg_qemu_tb_exec(env, tb_ptr); @@ -XXX,XX +XXX,XX @@ cpu_tb_exec(CPUState *cpu, TranslationBlock *itb, int *tb_exit) * of the start of the TB. */ CPUClass *cc = CPU_GET_CLASS(cpu); - qemu_log_mask_and_addr(CPU_LOG_EXEC, last_tb->pc, - "Stopped execution of TB chain before %p [" - TARGET_FMT_lx "] %s\n", - last_tb->tc.ptr, last_tb->pc, - lookup_symbol(last_tb->pc)); + if (cc->tcg_ops->synchronize_from_tb) { cc->tcg_ops->synchronize_from_tb(cpu, last_tb); } else { assert(cc->set_pc); - cc->set_pc(cpu, last_tb->pc); + cc->set_pc(cpu, tb_pc(last_tb)); + } + if (qemu_loglevel_mask(CPU_LOG_EXEC)) { + target_ulong pc = log_pc(cpu, last_tb); + if (qemu_log_in_addr_range(pc)) { + qemu_log("Stopped execution of TB chain before %p [" + TARGET_FMT_lx "] %s\n", + last_tb->tc.ptr, pc, lookup_symbol(pc)); + } } } @@ -XXX,XX +XXX,XX @@ static inline void tb_add_jump(TranslationBlock *tb, int n, qemu_spin_unlock(&tb_next->jmp_lock); - qemu_log_mask_and_addr(CPU_LOG_EXEC, tb->pc, - "Linking TBs %p [" TARGET_FMT_lx - "] index %d -> %p [" TARGET_FMT_lx "]\n", - tb->tc.ptr, tb->pc, n, - tb_next->tc.ptr, tb_next->pc); + qemu_log_mask(CPU_LOG_EXEC, "Linking TBs %p index %d -> %p\n", + tb->tc.ptr, n, tb_next->tc.ptr); return; out_unlock_next: @@ -XXX,XX +XXX,XX @@ static inline bool cpu_handle_interrupt(CPUState *cpu, } static inline void cpu_loop_exec_tb(CPUState *cpu, TranslationBlock *tb, + target_ulong pc, TranslationBlock **last_tb, int *tb_exit) { int32_t insns_left; - trace_exec_tb(tb, tb->pc); + trace_exec_tb(tb, pc); tb = cpu_tb_exec(cpu, tb, tb_exit); if (*tb_exit != TB_EXIT_REQUESTED) { *last_tb = tb; @@ -XXX,XX +XXX,XX @@ int cpu_exec(CPUState *cpu) tb_add_jump(last_tb, tb_exit, tb); } - cpu_loop_exec_tb(cpu, tb, &last_tb, &tb_exit); + cpu_loop_exec_tb(cpu, tb, pc, &last_tb, &tb_exit); /* Try to align the host and virtual clocks if the guest is in advance */ 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 @@ static int encode_search(TranslationBlock *tb, uint8_t *block) for (j = 0; j < TARGET_INSN_START_WORDS; ++j) { if (i == 0) { - prev = (j == 0 ? tb->pc : 0); + prev = (j == 0 ? tb_pc(tb) : 0); } else { prev = tcg_ctx->gen_insn_data[i - 1][j]; } @@ -XXX,XX +XXX,XX @@ static int encode_search(TranslationBlock *tb, uint8_t *block) static int cpu_restore_state_from_tb(CPUState *cpu, TranslationBlock *tb, uintptr_t searched_pc, bool reset_icount) { - target_ulong data[TARGET_INSN_START_WORDS] = { tb->pc }; + target_ulong data[TARGET_INSN_START_WORDS] = { tb_pc(tb) }; uintptr_t host_pc = (uintptr_t)tb->tc.ptr; CPUArchState *env = cpu->env_ptr; const uint8_t *p = tb->tc.ptr + tb->tc.size; @@ -XXX,XX +XXX,XX @@ static bool tb_cmp(const void *ap, const void *bp) const TranslationBlock *a = ap; const TranslationBlock *b = bp; - return a->pc == b->pc && + return tb_pc(a) == tb_pc(b) && a->cs_base == b->cs_base && a->flags == b->flags && (tb_cflags(a) & ~CF_INVALID) == (tb_cflags(b) & ~CF_INVALID) && @@ -XXX,XX +XXX,XX @@ static void do_tb_invalidate_check(void *p, uint32_t hash, void *userp) TranslationBlock *tb = p; target_ulong addr = *(target_ulong *)userp; - if (!(addr + TARGET_PAGE_SIZE <= tb->pc || addr >= tb->pc + tb->size)) { + if (!(addr + TARGET_PAGE_SIZE <= tb_pc(tb) || + addr >= tb_pc(tb) + tb->size)) { printf("ERROR invalidate: address=" TARGET_FMT_lx - " PC=%08lx size=%04x\n", addr, (long)tb->pc, tb->size); + " PC=%08lx size=%04x\n", addr, (long)tb_pc(tb), tb->size); } } @@ -XXX,XX +XXX,XX @@ static void do_tb_page_check(void *p, uint32_t hash, void *userp) TranslationBlock *tb = p; int flags1, flags2; - flags1 = page_get_flags(tb->pc); - flags2 = page_get_flags(tb->pc + tb->size - 1); + flags1 = page_get_flags(tb_pc(tb)); + flags2 = page_get_flags(tb_pc(tb) + tb->size - 1); if ((flags1 & PAGE_WRITE) || (flags2 & PAGE_WRITE)) { printf("ERROR page flags: PC=%08lx size=%04x f1=%x f2=%x\n", - (long)tb->pc, tb->size, flags1, flags2); + (long)tb_pc(tb), tb->size, flags1, flags2); } } @@ -XXX,XX +XXX,XX @@ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list) /* remove the TB from the hash list */ phys_pc = tb->page_addr[0]; - h = tb_hash_func(phys_pc, tb->pc, tb->flags, orig_cflags, + h = tb_hash_func(phys_pc, tb_pc(tb), tb->flags, orig_cflags, tb->trace_vcpu_dstate); if (!qht_remove(&tb_ctx.htable, tb, h)) { return; @@ -XXX,XX +XXX,XX @@ tb_link_page(TranslationBlock *tb, tb_page_addr_t phys_pc, } /* add in the hash table */ - h = tb_hash_func(phys_pc, tb->pc, tb->flags, tb->cflags, + h = tb_hash_func(phys_pc, tb_pc(tb), tb->flags, tb->cflags, tb->trace_vcpu_dstate); qht_insert(&tb_ctx.htable, tb, h, &existing_tb); @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, tcg_ctx->cpu = NULL; max_insns = tb->icount; - trace_translate_block(tb, tb->pc, tb->tc.ptr); + trace_translate_block(tb, pc, tb->tc.ptr); /* generate machine code */ tb->jmp_reset_offset[0] = TB_JMP_RESET_OFFSET_INVALID; @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, ti = profile_getclock(); #endif - gen_code_size = tcg_gen_code(tcg_ctx, tb); + gen_code_size = tcg_gen_code(tcg_ctx, tb, pc); if (unlikely(gen_code_size < 0)) { error_return: switch (gen_code_size) { @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, #ifdef DEBUG_DISAS if (qemu_loglevel_mask(CPU_LOG_TB_OUT_ASM) && - qemu_log_in_addr_range(tb->pc)) { + qemu_log_in_addr_range(pc)) { FILE *logfile = qemu_log_trylock(); if (logfile) { int code_size, data_size; @@ -XXX,XX +XXX,XX @@ void cpu_io_recompile(CPUState *cpu, uintptr_t retaddr) */ cpu->cflags_next_tb = curr_cflags(cpu) | CF_MEMI_ONLY | CF_LAST_IO | n; - qemu_log_mask_and_addr(CPU_LOG_EXEC, tb->pc, - "cpu_io_recompile: rewound execution of TB to " - TARGET_FMT_lx "\n", tb->pc); + if (qemu_loglevel_mask(CPU_LOG_EXEC)) { + target_ulong pc = log_pc(cpu, tb); + if (qemu_log_in_addr_range(pc)) { + qemu_log("cpu_io_recompile: rewound execution of TB to " + TARGET_FMT_lx "\n", pc); + } + } cpu_loop_exit_noexc(cpu); } diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ void arm_cpu_synchronize_from_tb(CPUState *cs, * never possible for an AArch64 TB to chain to an AArch32 TB. */ if (is_a64(env)) { - env->pc = tb->pc; + env->pc = tb_pc(tb); } else { - env->regs[15] = tb->pc; + env->regs[15] = tb_pc(tb); } } #endif /* CONFIG_TCG */ diff --git a/target/avr/cpu.c b/target/avr/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/avr/cpu.c +++ b/target/avr/cpu.c @@ -XXX,XX +XXX,XX @@ static void avr_cpu_synchronize_from_tb(CPUState *cs, AVRCPU *cpu = AVR_CPU(cs); CPUAVRState *env = &cpu->env; - env->pc_w = tb->pc / 2; /* internally PC points to words */ + env->pc_w = tb_pc(tb) / 2; /* internally PC points to words */ } static void avr_cpu_reset(DeviceState *ds) diff --git a/target/hexagon/cpu.c b/target/hexagon/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/hexagon/cpu.c +++ b/target/hexagon/cpu.c @@ -XXX,XX +XXX,XX @@ static void hexagon_cpu_synchronize_from_tb(CPUState *cs, { HexagonCPU *cpu = HEXAGON_CPU(cs); CPUHexagonState *env = &cpu->env; - env->gpr[HEX_REG_PC] = tb->pc; + env->gpr[HEX_REG_PC] = tb_pc(tb); } static bool hexagon_cpu_has_work(CPUState *cs) diff --git a/target/hppa/cpu.c b/target/hppa/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/hppa/cpu.c +++ b/target/hppa/cpu.c @@ -XXX,XX +XXX,XX @@ static void hppa_cpu_synchronize_from_tb(CPUState *cs, HPPACPU *cpu = HPPA_CPU(cs); #ifdef CONFIG_USER_ONLY - cpu->env.iaoq_f = tb->pc; + cpu->env.iaoq_f = tb_pc(tb); cpu->env.iaoq_b = tb->cs_base; #else /* Recover the IAOQ values from the GVA + PRIV. */ @@ -XXX,XX +XXX,XX @@ static void hppa_cpu_synchronize_from_tb(CPUState *cs, int32_t diff = cs_base; cpu->env.iasq_f = iasq_f; - cpu->env.iaoq_f = (tb->pc & ~iasq_f) + priv; + cpu->env.iaoq_f = (tb_pc(tb) & ~iasq_f) + priv; if (diff) { cpu->env.iaoq_b = cpu->env.iaoq_f + diff; } diff --git a/target/i386/tcg/tcg-cpu.c b/target/i386/tcg/tcg-cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/i386/tcg/tcg-cpu.c +++ b/target/i386/tcg/tcg-cpu.c @@ -XXX,XX +XXX,XX @@ static void x86_cpu_synchronize_from_tb(CPUState *cs, { X86CPU *cpu = X86_CPU(cs); - cpu->env.eip = tb->pc - tb->cs_base; + cpu->env.eip = tb_pc(tb) - tb->cs_base; } #ifndef CONFIG_USER_ONLY diff --git a/target/loongarch/cpu.c b/target/loongarch/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/loongarch/cpu.c +++ b/target/loongarch/cpu.c @@ -XXX,XX +XXX,XX @@ static void loongarch_cpu_synchronize_from_tb(CPUState *cs, LoongArchCPU *cpu = LOONGARCH_CPU(cs); CPULoongArchState *env = &cpu->env; - env->pc = tb->pc; + env->pc = tb_pc(tb); } #endif /* CONFIG_TCG */ diff --git a/target/microblaze/cpu.c b/target/microblaze/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/microblaze/cpu.c +++ b/target/microblaze/cpu.c @@ -XXX,XX +XXX,XX @@ static void mb_cpu_synchronize_from_tb(CPUState *cs, { MicroBlazeCPU *cpu = MICROBLAZE_CPU(cs); - cpu->env.pc = tb->pc; + cpu->env.pc = tb_pc(tb); cpu->env.iflags = tb->flags & IFLAGS_TB_MASK; } diff --git a/target/mips/tcg/exception.c b/target/mips/tcg/exception.c index XXXXXXX..XXXXXXX 100644 --- a/target/mips/tcg/exception.c +++ b/target/mips/tcg/exception.c @@ -XXX,XX +XXX,XX @@ void mips_cpu_synchronize_from_tb(CPUState *cs, const TranslationBlock *tb) MIPSCPU *cpu = MIPS_CPU(cs); CPUMIPSState *env = &cpu->env; - env->active_tc.PC = tb->pc; + env->active_tc.PC = tb_pc(tb); env->hflags &= ~MIPS_HFLAG_BMASK; env->hflags |= tb->flags & MIPS_HFLAG_BMASK; } diff --git a/target/mips/tcg/sysemu/special_helper.c b/target/mips/tcg/sysemu/special_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/mips/tcg/sysemu/special_helper.c +++ b/target/mips/tcg/sysemu/special_helper.c @@ -XXX,XX +XXX,XX @@ bool mips_io_recompile_replay_branch(CPUState *cs, const TranslationBlock *tb) CPUMIPSState *env = &cpu->env; if ((env->hflags & MIPS_HFLAG_BMASK) != 0 - && env->active_tc.PC != tb->pc) { + && env->active_tc.PC != tb_pc(tb)) { env->active_tc.PC -= (env->hflags & MIPS_HFLAG_B16 ? 2 : 4); env->hflags &= ~MIPS_HFLAG_BMASK; return true; diff --git a/target/openrisc/cpu.c b/target/openrisc/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/openrisc/cpu.c +++ b/target/openrisc/cpu.c @@ -XXX,XX +XXX,XX @@ static void openrisc_cpu_synchronize_from_tb(CPUState *cs, { OpenRISCCPU *cpu = OPENRISC_CPU(cs); - cpu->env.pc = tb->pc; + cpu->env.pc = tb_pc(tb); } diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/riscv/cpu.c +++ b/target/riscv/cpu.c @@ -XXX,XX +XXX,XX @@ static void riscv_cpu_synchronize_from_tb(CPUState *cs, RISCVMXL xl = FIELD_EX32(tb->flags, TB_FLAGS, XL); if (xl == MXL_RV32) { - env->pc = (int32_t)tb->pc; + env->pc = (int32_t)tb_pc(tb); } else { - env->pc = tb->pc; + env->pc = tb_pc(tb); } } diff --git a/target/rx/cpu.c b/target/rx/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/rx/cpu.c +++ b/target/rx/cpu.c @@ -XXX,XX +XXX,XX @@ static void rx_cpu_synchronize_from_tb(CPUState *cs, { RXCPU *cpu = RX_CPU(cs); - cpu->env.pc = tb->pc; + cpu->env.pc = tb_pc(tb); } static bool rx_cpu_has_work(CPUState *cs) diff --git a/target/sh4/cpu.c b/target/sh4/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/sh4/cpu.c +++ b/target/sh4/cpu.c @@ -XXX,XX +XXX,XX @@ static void superh_cpu_synchronize_from_tb(CPUState *cs, { SuperHCPU *cpu = SUPERH_CPU(cs); - cpu->env.pc = tb->pc; + cpu->env.pc = tb_pc(tb); cpu->env.flags = tb->flags & TB_FLAG_ENVFLAGS_MASK; } @@ -XXX,XX +XXX,XX @@ static bool superh_io_recompile_replay_branch(CPUState *cs, CPUSH4State *env = &cpu->env; if ((env->flags & ((DELAY_SLOT | DELAY_SLOT_CONDITIONAL))) != 0 - && env->pc != tb->pc) { + && env->pc != tb_pc(tb)) { env->pc -= 2; env->flags &= ~(DELAY_SLOT | DELAY_SLOT_CONDITIONAL); return true; diff --git a/target/sparc/cpu.c b/target/sparc/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/sparc/cpu.c +++ b/target/sparc/cpu.c @@ -XXX,XX +XXX,XX @@ static void sparc_cpu_synchronize_from_tb(CPUState *cs, { SPARCCPU *cpu = SPARC_CPU(cs); - cpu->env.pc = tb->pc; + cpu->env.pc = tb_pc(tb); cpu->env.npc = tb->cs_base; } diff --git a/target/tricore/cpu.c b/target/tricore/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/tricore/cpu.c +++ b/target/tricore/cpu.c @@ -XXX,XX +XXX,XX @@ static void tricore_cpu_synchronize_from_tb(CPUState *cs, TriCoreCPU *cpu = TRICORE_CPU(cs); CPUTriCoreState *env = &cpu->env; - env->PC = tb->pc; + env->PC = tb_pc(tb); } static void tricore_cpu_reset(DeviceState *dev) 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) +int tcg_gen_code(TCGContext *s, TranslationBlock *tb, target_ulong pc_start) { #ifdef CONFIG_PROFILER TCGProfile *prof = &s->prof; @@ -XXX,XX +XXX,XX @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb) #ifdef DEBUG_DISAS if (unlikely(qemu_loglevel_mask(CPU_LOG_TB_OP) - && qemu_log_in_addr_range(tb->pc))) { + && qemu_log_in_addr_range(pc_start))) { FILE *logfile = qemu_log_trylock(); if (logfile) { fprintf(logfile, "OP:\n"); @@ -XXX,XX +XXX,XX @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb) if (s->nb_indirects > 0) { #ifdef DEBUG_DISAS if (unlikely(qemu_loglevel_mask(CPU_LOG_TB_OP_IND) - && qemu_log_in_addr_range(tb->pc))) { + && qemu_log_in_addr_range(pc_start))) { FILE *logfile = qemu_log_trylock(); if (logfile) { fprintf(logfile, "OP before indirect lowering:\n"); @@ -XXX,XX +XXX,XX @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb) #ifdef DEBUG_DISAS if (unlikely(qemu_loglevel_mask(CPU_LOG_TB_OP_OPT) - && qemu_log_in_addr_range(tb->pc))) { + && qemu_log_in_addr_range(pc_start))) { FILE *logfile = qemu_log_trylock(); if (logfile) { fprintf(logfile, "OP after optimization and liveness analysis:\n"); -- 2.34.1 Prepare for targets to be able to produce TBs that can run in more than one virtual context. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- accel/tcg/internal.h | 4 +++ accel/tcg/tb-jmp-cache.h | 41 +++++++++++++++++++++++++ include/exec/cpu-defs.h | 3 ++ include/exec/exec-all.h | 32 ++++++++++++++++++-- accel/tcg/cpu-exec.c | 16 ++++++---- accel/tcg/translate-all.c | 64 ++++++++++++++++++++++++++------------- 6 files changed, 131 insertions(+), 29 deletions(-) diff --git a/accel/tcg/internal.h b/accel/tcg/internal.h index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/internal.h +++ b/accel/tcg/internal.h @@ -XXX,XX +XXX,XX @@ void tb_htable_init(void); /* Return the current PC from CPU, which may be cached in TB. */ static inline target_ulong log_pc(CPUState *cpu, const TranslationBlock *tb) { +#if TARGET_TB_PCREL + return cpu->cc->get_pc(cpu); +#else return tb_pc(tb); +#endif } #endif /* ACCEL_TCG_INTERNAL_H */ diff --git a/accel/tcg/tb-jmp-cache.h b/accel/tcg/tb-jmp-cache.h index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/tb-jmp-cache.h +++ b/accel/tcg/tb-jmp-cache.h @@ -XXX,XX +XXX,XX @@ /* * Accessed in parallel; all accesses to 'tb' must be atomic. + * For TARGET_TB_PCREL, accesses to 'pc' must be protected by + * a load_acquire/store_release to 'tb'. */ struct CPUJumpCache { struct { TranslationBlock *tb; +#if TARGET_TB_PCREL + target_ulong pc; +#endif } array[TB_JMP_CACHE_SIZE]; }; +static inline TranslationBlock * +tb_jmp_cache_get_tb(CPUJumpCache *jc, uint32_t hash) +{ +#if TARGET_TB_PCREL + /* Use acquire to ensure current load of pc from jc. */ + return qatomic_load_acquire(&jc->array[hash].tb); +#else + /* Use rcu_read to ensure current load of pc from *tb. */ + return qatomic_rcu_read(&jc->array[hash].tb); +#endif +} + +static inline target_ulong +tb_jmp_cache_get_pc(CPUJumpCache *jc, uint32_t hash, TranslationBlock *tb) +{ +#if TARGET_TB_PCREL + return jc->array[hash].pc; +#else + return tb_pc(tb); +#endif +} + +static inline void +tb_jmp_cache_set(CPUJumpCache *jc, uint32_t hash, + TranslationBlock *tb, target_ulong pc) +{ +#if TARGET_TB_PCREL + jc->array[hash].pc = pc; + /* Use store_release on tb to ensure pc is written first. */ + qatomic_store_release(&jc->array[hash].tb, tb); +#else + /* Use the pc value already stored in tb->pc. */ + qatomic_set(&jc->array[hash].tb, tb); +#endif +} + #endif /* ACCEL_TCG_TB_JMP_CACHE_H */ diff --git a/include/exec/cpu-defs.h b/include/exec/cpu-defs.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/cpu-defs.h +++ b/include/exec/cpu-defs.h @@ -XXX,XX +XXX,XX @@ # error TARGET_PAGE_BITS must be defined in cpu-param.h # endif #endif +#ifndef TARGET_TB_PCREL +# define TARGET_TB_PCREL 0 +#endif #define TARGET_LONG_SIZE (TARGET_LONG_BITS / 8) diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h index XXXXXXX..XXXXXXX 100644 --- a/include/exec/exec-all.h +++ b/include/exec/exec-all.h @@ -XXX,XX +XXX,XX @@ struct tb_tc { }; struct TranslationBlock { - target_ulong pc; /* simulated PC corresponding to this block (EIP + CS base) */ - target_ulong cs_base; /* CS base for this block */ +#if !TARGET_TB_PCREL + /* + * Guest PC corresponding to this block. This must be the true + * virtual address. Therefore e.g. x86 stores EIP + CS_BASE, and + * targets like Arm, MIPS, HP-PA, which reuse low bits for ISA or + * privilege, must store those bits elsewhere. + * + * If TARGET_TB_PCREL, the opcodes for the TranslationBlock are + * written such that the TB is associated only with the physical + * page and may be run in any virtual address context. In this case, + * PC must always be taken from ENV in a target-specific manner. + * Unwind information is taken as offsets from the page, to be + * deposited into the "current" PC. + */ + target_ulong pc; +#endif + + /* + * Target-specific data associated with the TranslationBlock, e.g.: + * x86: the original user, the Code Segment virtual base, + * arm: an extension of tb->flags, + * s390x: instruction data for EXECUTE, + * sparc: the next pc of the instruction queue (for delay slots). + */ + target_ulong cs_base; + uint32_t flags; /* flags defining in which context the code was generated */ uint32_t cflags; /* compile flags */ @@ -XXX,XX +XXX,XX @@ struct TranslationBlock { /* Hide the read to avoid ifdefs for TARGET_TB_PCREL. */ static inline target_ulong tb_pc(const TranslationBlock *tb) { +#if TARGET_TB_PCREL + qemu_build_not_reached(); +#else return tb->pc; +#endif } /* Hide the qatomic_read to make code a little easier on the eyes */ diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c index XXXXXXX..XXXXXXX 100644 --- a/accel/tcg/cpu-exec.c +++ b/accel/tcg/cpu-exec.c @@ -XXX,XX +XXX,XX @@ static bool tb_lookup_cmp(const void *p, const void *d) const TranslationBlock *tb = p; const struct tb_desc *desc = d; - if (tb_pc(tb) == desc->pc && + if ((TARGET_TB_PCREL || tb_pc(tb) == desc->pc) && tb->page_addr[0] == desc->page_addr0 && tb->cs_base == desc->cs_base && tb->flags == desc->flags && @@ -XXX,XX +XXX,XX @@ static TranslationBlock *tb_htable_lookup(CPUState *cpu, target_ulong pc, return NULL; } desc.page_addr0 = phys_pc; - h = tb_hash_func(phys_pc, pc, flags, cflags, *cpu->trace_dstate); + h = tb_hash_func(phys_pc, (TARGET_TB_PCREL ? 0 : pc), + flags, cflags, *cpu->trace_dstate); return qht_lookup_custom(&tb_ctx.htable, &desc, h, tb_lookup_cmp); } @@ -XXX,XX +XXX,XX @@ static inline TranslationBlock *tb_lookup(CPUState *cpu, target_ulong pc, uint32_t flags, uint32_t cflags) { TranslationBlock *tb; + CPUJumpCache *jc; uint32_t hash; /* we should never be trying to look up an INVALID tb */ tcg_debug_assert(!(cflags & CF_INVALID)); hash = tb_jmp_cache_hash_func(pc); - tb = qatomic_rcu_read(&cpu->tb_jmp_cache->array[hash].tb); + jc = cpu->tb_jmp_cache; + tb = tb_jmp_cache_get_tb(jc, hash); if (likely(tb && - tb->pc == pc && + tb_jmp_cache_get_pc(jc, hash, tb) == pc && tb->cs_base == cs_base && tb->flags == flags && tb->trace_vcpu_dstate == *cpu->trace_dstate && @@ -XXX,XX +XXX,XX @@ static inline TranslationBlock *tb_lookup(CPUState *cpu, target_ulong pc, if (tb == NULL) { return NULL; } - qatomic_set(&cpu->tb_jmp_cache->array[hash].tb, tb); + tb_jmp_cache_set(jc, hash, tb, pc); return tb; } @@ -XXX,XX +XXX,XX @@ cpu_tb_exec(CPUState *cpu, TranslationBlock *itb, int *tb_exit) if (cc->tcg_ops->synchronize_from_tb) { cc->tcg_ops->synchronize_from_tb(cpu, last_tb); } else { + assert(!TARGET_TB_PCREL); assert(cc->set_pc); cc->set_pc(cpu, tb_pc(last_tb)); } @@ -XXX,XX +XXX,XX @@ int cpu_exec(CPUState *cpu) * for the fast lookup */ h = tb_jmp_cache_hash_func(pc); - qatomic_set(&cpu->tb_jmp_cache->array[h].tb, tb); + tb_jmp_cache_set(cpu->tb_jmp_cache, h, tb, pc); } #ifndef CONFIG_USER_ONLY 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 @@ static int encode_search(TranslationBlock *tb, uint8_t *block) for (j = 0; j < TARGET_INSN_START_WORDS; ++j) { if (i == 0) { - prev = (j == 0 ? tb_pc(tb) : 0); + prev = (!TARGET_TB_PCREL && j == 0 ? tb_pc(tb) : 0); } else { prev = tcg_ctx->gen_insn_data[i - 1][j]; } @@ -XXX,XX +XXX,XX @@ static int encode_search(TranslationBlock *tb, uint8_t *block) static int cpu_restore_state_from_tb(CPUState *cpu, TranslationBlock *tb, uintptr_t searched_pc, bool reset_icount) { - target_ulong data[TARGET_INSN_START_WORDS] = { tb_pc(tb) }; + target_ulong data[TARGET_INSN_START_WORDS]; uintptr_t host_pc = (uintptr_t)tb->tc.ptr; CPUArchState *env = cpu->env_ptr; const uint8_t *p = tb->tc.ptr + tb->tc.size; @@ -XXX,XX +XXX,XX @@ static int cpu_restore_state_from_tb(CPUState *cpu, TranslationBlock *tb, return -1; } + memset(data, 0, sizeof(data)); + if (!TARGET_TB_PCREL) { + data[0] = tb_pc(tb); + } + /* Reconstruct the stored insn data while looking for the point at which the end of the insn exceeds the searched_pc. */ for (i = 0; i < num_insns; ++i) { @@ -XXX,XX +XXX,XX @@ static bool tb_cmp(const void *ap, const void *bp) const TranslationBlock *a = ap; const TranslationBlock *b = bp; - return tb_pc(a) == tb_pc(b) && - a->cs_base == b->cs_base && - a->flags == b->flags && - (tb_cflags(a) & ~CF_INVALID) == (tb_cflags(b) & ~CF_INVALID) && - a->trace_vcpu_dstate == b->trace_vcpu_dstate && - a->page_addr[0] == b->page_addr[0] && - a->page_addr[1] == b->page_addr[1]; + return ((TARGET_TB_PCREL || tb_pc(a) == tb_pc(b)) && + a->cs_base == b->cs_base && + a->flags == b->flags && + (tb_cflags(a) & ~CF_INVALID) == (tb_cflags(b) & ~CF_INVALID) && + a->trace_vcpu_dstate == b->trace_vcpu_dstate && + a->page_addr[0] == b->page_addr[0] && + a->page_addr[1] == b->page_addr[1]); } void tb_htable_init(void) @@ -XXX,XX +XXX,XX @@ static inline void tb_jmp_unlink(TranslationBlock *dest) qemu_spin_unlock(&dest->jmp_lock); } +static void tb_jmp_cache_inval_tb(TranslationBlock *tb) +{ + CPUState *cpu; + + if (TARGET_TB_PCREL) { + /* A TB may be at any virtual address */ + CPU_FOREACH(cpu) { + tcg_flush_jmp_cache(cpu); + } + } else { + uint32_t h = tb_jmp_cache_hash_func(tb_pc(tb)); + + CPU_FOREACH(cpu) { + CPUJumpCache *jc = cpu->tb_jmp_cache; + + if (qatomic_read(&jc->array[h].tb) == tb) { + qatomic_set(&jc->array[h].tb, NULL); + } + } + } +} + /* * In user-mode, call with mmap_lock held. * In !user-mode, if @rm_from_page_list is set, call with the TB's pages' @@ -XXX,XX +XXX,XX @@ static inline void tb_jmp_unlink(TranslationBlock *dest) */ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list) { - CPUState *cpu; PageDesc *p; uint32_t h; tb_page_addr_t phys_pc; @@ -XXX,XX +XXX,XX @@ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list) /* remove the TB from the hash list */ phys_pc = tb->page_addr[0]; - h = tb_hash_func(phys_pc, tb_pc(tb), tb->flags, orig_cflags, - tb->trace_vcpu_dstate); + h = tb_hash_func(phys_pc, (TARGET_TB_PCREL ? 0 : tb_pc(tb)), + tb->flags, orig_cflags, tb->trace_vcpu_dstate); if (!qht_remove(&tb_ctx.htable, tb, h)) { return; } @@ -XXX,XX +XXX,XX @@ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list) } /* remove the TB from the hash list */ - h = tb_jmp_cache_hash_func(tb->pc); - CPU_FOREACH(cpu) { - CPUJumpCache *jc = cpu->tb_jmp_cache; - if (qatomic_read(&jc->array[h].tb) == tb) { - qatomic_set(&jc->array[h].tb, NULL); - } - } + tb_jmp_cache_inval_tb(tb); /* suppress this TB from the two jump lists */ tb_remove_from_jmp_list(tb, 0); @@ -XXX,XX +XXX,XX @@ tb_link_page(TranslationBlock *tb, tb_page_addr_t phys_pc, } /* add in the hash table */ - h = tb_hash_func(phys_pc, tb_pc(tb), tb->flags, tb->cflags, - tb->trace_vcpu_dstate); + h = tb_hash_func(phys_pc, (TARGET_TB_PCREL ? 0 : tb_pc(tb)), + tb->flags, tb->cflags, tb->trace_vcpu_dstate); qht_insert(&tb_ctx.htable, tb, h, &existing_tb); /* remove TB from the page(s) if we couldn't insert it */ @@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu, gen_code_buf = tcg_ctx->code_gen_ptr; tb->tc.ptr = tcg_splitwx_to_rx(gen_code_buf); +#if !TARGET_TB_PCREL tb->pc = pc; +#endif tb->cs_base = cs_base; tb->flags = flags; tb->cflags = cflags; -- 2.34.1 From: Leandro Lupori <leandro.lupori@eldorado.org.br> PowerPC64 processors handle direct branches better than indirect ones, resulting in less stalled cycles and branch misses. However, PPC's tb_target_set_jmp_target() was only using direct branches for 16-bit jumps, while PowerPC64's unconditional branch instructions are able to handle displacements of up to 26 bits. To take advantage of this, now jumps whose displacements fit in between 17 and 26 bits are also converted to direct branches. Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Leandro Lupori <leandro.lupori@eldorado.org.br> [rth: Expanded some commentary.] Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- tcg/ppc/tcg-target.c.inc | 119 +++++++++++++++++++++++++++++---------- 1 file changed, 88 insertions(+), 31 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 void tcg_out_mb(TCGContext *s, TCGArg a0) tcg_out32(s, insn); } +static inline uint64_t make_pair(tcg_insn_unit i1, tcg_insn_unit i2) +{ + if (HOST_BIG_ENDIAN) { + return (uint64_t)i1 << 32 | i2; + } + return (uint64_t)i2 << 32 | i1; +} + +static inline void ppc64_replace2(uintptr_t rx, uintptr_t rw, + tcg_insn_unit i0, tcg_insn_unit i1) +{ +#if TCG_TARGET_REG_BITS == 64 + qatomic_set((uint64_t *)rw, make_pair(i0, i1)); + flush_idcache_range(rx, rw, 8); +#else + qemu_build_not_reached(); +#endif +} + +static inline void ppc64_replace4(uintptr_t rx, uintptr_t rw, + tcg_insn_unit i0, tcg_insn_unit i1, + tcg_insn_unit i2, tcg_insn_unit i3) +{ + uint64_t p[2]; + + p[!HOST_BIG_ENDIAN] = make_pair(i0, i1); + p[HOST_BIG_ENDIAN] = make_pair(i2, i3); + + /* + * There's no convenient way to get the compiler to allocate a pair + * of registers at an even index, so copy into r6/r7 and clobber. + */ + asm("mr %%r6, %1\n\t" + "mr %%r7, %2\n\t" + "stq %%r6, %0" + : "=Q"(*(__int128 *)rw) : "r"(p[0]), "r"(p[1]) : "r6", "r7"); + flush_idcache_range(rx, rw, 16); +} + void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx, uintptr_t jmp_rw, uintptr_t addr) { - if (TCG_TARGET_REG_BITS == 64) { - tcg_insn_unit i1, i2; - intptr_t tb_diff = addr - tc_ptr; - intptr_t br_diff = addr - (jmp_rx + 4); - uint64_t pair; + tcg_insn_unit i0, i1, i2, i3; + intptr_t tb_diff = addr - tc_ptr; + intptr_t br_diff = addr - (jmp_rx + 4); + intptr_t lo, hi; - /* This does not exercise the range of the branch, but we do - still need to be able to load the new value of TCG_REG_TB. - But this does still happen quite often. */ - if (tb_diff == (int16_t)tb_diff) { - i1 = ADDI | TAI(TCG_REG_TB, TCG_REG_TB, tb_diff); - i2 = B | (br_diff & 0x3fffffc); - } else { - intptr_t lo = (int16_t)tb_diff; - intptr_t hi = (int32_t)(tb_diff - lo); - assert(tb_diff == hi + lo); - i1 = ADDIS | TAI(TCG_REG_TB, TCG_REG_TB, hi >> 16); - i2 = ADDI | TAI(TCG_REG_TB, TCG_REG_TB, lo); - } -#if HOST_BIG_ENDIAN - pair = (uint64_t)i1 << 32 | i2; -#else - pair = (uint64_t)i2 << 32 | i1; -#endif - - /* As per the enclosing if, this is ppc64. Avoid the _Static_assert - within qatomic_set that would fail to build a ppc32 host. */ - qatomic_set__nocheck((uint64_t *)jmp_rw, pair); - flush_idcache_range(jmp_rx, jmp_rw, 8); - } else { + if (TCG_TARGET_REG_BITS == 32) { intptr_t diff = addr - jmp_rx; tcg_debug_assert(in_range_b(diff)); qatomic_set((uint32_t *)jmp_rw, B | (diff & 0x3fffffc)); flush_idcache_range(jmp_rx, jmp_rw, 4); + return; } + + /* + * For 16-bit displacements, we can use a single add + branch. + * This happens quite often. + */ + if (tb_diff == (int16_t)tb_diff) { + i0 = ADDI | TAI(TCG_REG_TB, TCG_REG_TB, tb_diff); + i1 = B | (br_diff & 0x3fffffc); + ppc64_replace2(jmp_rx, jmp_rw, i0, i1); + return; + } + + lo = (int16_t)tb_diff; + hi = (int32_t)(tb_diff - lo); + assert(tb_diff == hi + lo); + i0 = ADDIS | TAI(TCG_REG_TB, TCG_REG_TB, hi >> 16); + i1 = ADDI | TAI(TCG_REG_TB, TCG_REG_TB, lo); + + /* + * Without stq from 2.07, we can only update two insns, + * and those must be the ones that load the target address. + */ + if (!have_isa_2_07) { + ppc64_replace2(jmp_rx, jmp_rw, i0, i1); + return; + } + + /* + * For 26-bit displacements, we can use a direct branch. + * Otherwise we still need the indirect branch, which we + * must restore after a potential direct branch write. + */ + br_diff -= 4; + if (in_range_b(br_diff)) { + i2 = B | (br_diff & 0x3fffffc); + i3 = NOP; + } else { + i2 = MTSPR | RS(TCG_REG_TB) | CTR; + i3 = BCCTR | BO_ALWAYS; + } + ppc64_replace4(jmp_rx, jmp_rw, i0, i1, i2, i3); } static void tcg_out_call_int(TCGContext *s, int lk, @@ -XXX,XX +XXX,XX @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, if (s->tb_jmp_insn_offset) { /* Direct jump. */ if (TCG_TARGET_REG_BITS == 64) { - /* Ensure the next insns are 8-byte aligned. */ - if ((uintptr_t)s->code_ptr & 7) { + /* Ensure the next insns are 8 or 16-byte aligned. */ + while ((uintptr_t)s->code_ptr & (have_isa_2_07 ? 15 : 7)) { tcg_out32(s, NOP); } s->tb_jmp_insn_offset[args[0]] = tcg_current_code_size(s); -- 2.34.1 The value previously chosen overlaps GUSA_MASK. Rename all DELAY_SLOT_* and GUSA_* defines to emphasize that they are included in TB_FLAGs. Add aliases for the FPSCR and SR bits that are included in TB_FLAGS, so that we don't accidentally reassign those bits. Fixes: 4da06fb3062 ("target/sh4: Implement prctl_unalign_sigbus") Resolves: https://gitlab.com/qemu-project/qemu/-/issues/856 Reviewed-by: Yoshinori Sato <ysato@users.sourceforge.jp> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- target/sh4/cpu.h | 56 +++++++++++++------------ linux-user/sh4/signal.c | 6 +-- target/sh4/cpu.c | 6 +-- target/sh4/helper.c | 6 +-- target/sh4/translate.c | 90 ++++++++++++++++++++++------------------- 5 files changed, 88 insertions(+), 76 deletions(-) diff --git a/target/sh4/cpu.h b/target/sh4/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/sh4/cpu.h +++ b/target/sh4/cpu.h @@ -XXX,XX +XXX,XX @@ #define FPSCR_RM_NEAREST (0 << 0) #define FPSCR_RM_ZERO (1 << 0) -#define DELAY_SLOT_MASK 0x7 -#define DELAY_SLOT (1 << 0) -#define DELAY_SLOT_CONDITIONAL (1 << 1) -#define DELAY_SLOT_RTE (1 << 2) +#define TB_FLAG_DELAY_SLOT (1 << 0) +#define TB_FLAG_DELAY_SLOT_COND (1 << 1) +#define TB_FLAG_DELAY_SLOT_RTE (1 << 2) +#define TB_FLAG_PENDING_MOVCA (1 << 3) +#define TB_FLAG_GUSA_SHIFT 4 /* [11:4] */ +#define TB_FLAG_GUSA_EXCLUSIVE (1 << 12) +#define TB_FLAG_UNALIGN (1 << 13) +#define TB_FLAG_SR_FD (1 << SR_FD) /* 15 */ +#define TB_FLAG_FPSCR_PR FPSCR_PR /* 19 */ +#define TB_FLAG_FPSCR_SZ FPSCR_SZ /* 20 */ +#define TB_FLAG_FPSCR_FR FPSCR_FR /* 21 */ +#define TB_FLAG_SR_RB (1 << SR_RB) /* 29 */ +#define TB_FLAG_SR_MD (1 << SR_MD) /* 30 */ -#define TB_FLAG_PENDING_MOVCA (1 << 3) -#define TB_FLAG_UNALIGN (1 << 4) - -#define GUSA_SHIFT 4 -#ifdef CONFIG_USER_ONLY -#define GUSA_EXCLUSIVE (1 << 12) -#define GUSA_MASK ((0xff << GUSA_SHIFT) | GUSA_EXCLUSIVE) -#else -/* Provide dummy versions of the above to allow tests against tbflags - to be elided while avoiding ifdefs. */ -#define GUSA_EXCLUSIVE 0 -#define GUSA_MASK 0 -#endif - -#define TB_FLAG_ENVFLAGS_MASK (DELAY_SLOT_MASK | GUSA_MASK) +#define TB_FLAG_DELAY_SLOT_MASK (TB_FLAG_DELAY_SLOT | \ + TB_FLAG_DELAY_SLOT_COND | \ + TB_FLAG_DELAY_SLOT_RTE) +#define TB_FLAG_GUSA_MASK ((0xff << TB_FLAG_GUSA_SHIFT) | \ + TB_FLAG_GUSA_EXCLUSIVE) +#define TB_FLAG_FPSCR_MASK (TB_FLAG_FPSCR_PR | \ + TB_FLAG_FPSCR_SZ | \ + TB_FLAG_FPSCR_FR) +#define TB_FLAG_SR_MASK (TB_FLAG_SR_FD | \ + TB_FLAG_SR_RB | \ + TB_FLAG_SR_MD) +#define TB_FLAG_ENVFLAGS_MASK (TB_FLAG_DELAY_SLOT_MASK | \ + TB_FLAG_GUSA_MASK) typedef struct tlb_t { uint32_t vpn; /* virtual page number */ @@ -XXX,XX +XXX,XX @@ static inline int cpu_mmu_index (CPUSH4State *env, bool ifetch) { /* The instruction in a RTE delay slot is fetched in privileged mode, but executed in user mode. */ - if (ifetch && (env->flags & DELAY_SLOT_RTE)) { + if (ifetch && (env->flags & TB_FLAG_DELAY_SLOT_RTE)) { return 0; } else { return (env->sr & (1u << SR_MD)) == 0 ? 1 : 0; @@ -XXX,XX +XXX,XX @@ static inline void cpu_get_tb_cpu_state(CPUSH4State *env, target_ulong *pc, { *pc = env->pc; /* For a gUSA region, notice the end of the region. */ - *cs_base = env->flags & GUSA_MASK ? env->gregs[0] : 0; - *flags = env->flags /* TB_FLAG_ENVFLAGS_MASK: bits 0-2, 4-12 */ - | (env->fpscr & (FPSCR_FR | FPSCR_SZ | FPSCR_PR)) /* Bits 19-21 */ - | (env->sr & ((1u << SR_MD) | (1u << SR_RB))) /* Bits 29-30 */ - | (env->sr & (1u << SR_FD)) /* Bit 15 */ + *cs_base = env->flags & TB_FLAG_GUSA_MASK ? env->gregs[0] : 0; + *flags = env->flags + | (env->fpscr & TB_FLAG_FPSCR_MASK) + | (env->sr & TB_FLAG_SR_MASK) | (env->movcal_backup ? TB_FLAG_PENDING_MOVCA : 0); /* Bit 3 */ #ifdef CONFIG_USER_ONLY *flags |= TB_FLAG_UNALIGN * !env_cpu(env)->prctl_unalign_sigbus; diff --git a/linux-user/sh4/signal.c b/linux-user/sh4/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/sh4/signal.c +++ b/linux-user/sh4/signal.c @@ -XXX,XX +XXX,XX @@ static void restore_sigcontext(CPUSH4State *regs, struct target_sigcontext *sc) __get_user(regs->fpul, &sc->sc_fpul); regs->tra = -1; /* disable syscall checks */ - regs->flags &= ~(DELAY_SLOT_MASK | GUSA_MASK); + regs->flags = 0; } void setup_frame(int sig, struct target_sigaction *ka, @@ -XXX,XX +XXX,XX @@ void setup_frame(int sig, struct target_sigaction *ka, regs->gregs[5] = 0; regs->gregs[6] = frame_addr += offsetof(typeof(*frame), sc); regs->pc = (unsigned long) ka->_sa_handler; - regs->flags &= ~(DELAY_SLOT_MASK | GUSA_MASK); + regs->flags &= ~(TB_FLAG_DELAY_SLOT_MASK | TB_FLAG_GUSA_MASK); unlock_user_struct(frame, frame_addr, 1); return; @@ -XXX,XX +XXX,XX @@ void setup_rt_frame(int sig, struct target_sigaction *ka, regs->gregs[5] = frame_addr + offsetof(typeof(*frame), info); regs->gregs[6] = frame_addr + offsetof(typeof(*frame), uc); regs->pc = (unsigned long) ka->_sa_handler; - regs->flags &= ~(DELAY_SLOT_MASK | GUSA_MASK); + regs->flags &= ~(TB_FLAG_DELAY_SLOT_MASK | TB_FLAG_GUSA_MASK); unlock_user_struct(frame, frame_addr, 1); return; diff --git a/target/sh4/cpu.c b/target/sh4/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/sh4/cpu.c +++ b/target/sh4/cpu.c @@ -XXX,XX +XXX,XX @@ static void superh_cpu_synchronize_from_tb(CPUState *cs, SuperHCPU *cpu = SUPERH_CPU(cs); cpu->env.pc = tb_pc(tb); - cpu->env.flags = tb->flags & TB_FLAG_ENVFLAGS_MASK; + cpu->env.flags = tb->flags; } #ifndef CONFIG_USER_ONLY @@ -XXX,XX +XXX,XX @@ static bool superh_io_recompile_replay_branch(CPUState *cs, SuperHCPU *cpu = SUPERH_CPU(cs); CPUSH4State *env = &cpu->env; - if ((env->flags & ((DELAY_SLOT | DELAY_SLOT_CONDITIONAL))) != 0 + if ((env->flags & (TB_FLAG_DELAY_SLOT | TB_FLAG_DELAY_SLOT_COND)) && env->pc != tb_pc(tb)) { env->pc -= 2; - env->flags &= ~(DELAY_SLOT | DELAY_SLOT_CONDITIONAL); + env->flags &= ~(TB_FLAG_DELAY_SLOT | TB_FLAG_DELAY_SLOT_COND); return true; } return false; diff --git a/target/sh4/helper.c b/target/sh4/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/sh4/helper.c +++ b/target/sh4/helper.c @@ -XXX,XX +XXX,XX @@ void superh_cpu_do_interrupt(CPUState *cs) env->sr |= (1u << SR_BL) | (1u << SR_MD) | (1u << SR_RB); env->lock_addr = -1; - if (env->flags & DELAY_SLOT_MASK) { + if (env->flags & TB_FLAG_DELAY_SLOT_MASK) { /* Branch instruction should be executed again before delay slot. */ env->spc -= 2; /* Clear flags for exception/interrupt routine. */ - env->flags &= ~DELAY_SLOT_MASK; + env->flags &= ~TB_FLAG_DELAY_SLOT_MASK; } if (do_exp) { @@ -XXX,XX +XXX,XX @@ bool superh_cpu_exec_interrupt(CPUState *cs, int interrupt_request) CPUSH4State *env = &cpu->env; /* Delay slots are indivisible, ignore interrupts */ - if (env->flags & DELAY_SLOT_MASK) { + if (env->flags & TB_FLAG_DELAY_SLOT_MASK) { return false; } else { superh_cpu_do_interrupt(cs); diff --git a/target/sh4/translate.c b/target/sh4/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/sh4/translate.c +++ b/target/sh4/translate.c @@ -XXX,XX +XXX,XX @@ void superh_cpu_dump_state(CPUState *cs, FILE *f, int flags) i, env->gregs[i], i + 1, env->gregs[i + 1], i + 2, env->gregs[i + 2], i + 3, env->gregs[i + 3]); } - if (env->flags & DELAY_SLOT) { + if (env->flags & TB_FLAG_DELAY_SLOT) { qemu_printf("in delay slot (delayed_pc=0x%08x)\n", env->delayed_pc); - } else if (env->flags & DELAY_SLOT_CONDITIONAL) { + } else if (env->flags & TB_FLAG_DELAY_SLOT_COND) { qemu_printf("in conditional delay slot (delayed_pc=0x%08x)\n", env->delayed_pc); - } else if (env->flags & DELAY_SLOT_RTE) { + } else if (env->flags & TB_FLAG_DELAY_SLOT_RTE) { qemu_fprintf(f, "in rte delay slot (delayed_pc=0x%08x)\n", env->delayed_pc); } @@ -XXX,XX +XXX,XX @@ static inline void gen_save_cpu_state(DisasContext *ctx, bool save_pc) static inline bool use_exit_tb(DisasContext *ctx) { - return (ctx->tbflags & GUSA_EXCLUSIVE) != 0; + return (ctx->tbflags & TB_FLAG_GUSA_EXCLUSIVE) != 0; } static bool use_goto_tb(DisasContext *ctx, target_ulong dest) @@ -XXX,XX +XXX,XX @@ static void gen_conditional_jump(DisasContext *ctx, target_ulong dest, TCGLabel *l1 = gen_new_label(); TCGCond cond_not_taken = jump_if_true ? TCG_COND_EQ : TCG_COND_NE; - if (ctx->tbflags & GUSA_EXCLUSIVE) { + if (ctx->tbflags & TB_FLAG_GUSA_EXCLUSIVE) { /* When in an exclusive region, we must continue to the end. Therefore, exit the region on a taken branch, but otherwise fall through to the next instruction. */ tcg_gen_brcondi_i32(cond_not_taken, cpu_sr_t, 0, l1); - tcg_gen_movi_i32(cpu_flags, ctx->envflags & ~GUSA_MASK); + tcg_gen_movi_i32(cpu_flags, ctx->envflags & ~TB_FLAG_GUSA_MASK); /* Note that this won't actually use a goto_tb opcode because we disallow it in use_goto_tb, but it handles exit + singlestep. */ gen_goto_tb(ctx, 0, dest); @@ -XXX,XX +XXX,XX @@ static void gen_delayed_conditional_jump(DisasContext * ctx) tcg_gen_mov_i32(ds, cpu_delayed_cond); tcg_gen_discard_i32(cpu_delayed_cond); - if (ctx->tbflags & GUSA_EXCLUSIVE) { + if (ctx->tbflags & TB_FLAG_GUSA_EXCLUSIVE) { /* When in an exclusive region, we must continue to the end. Therefore, exit the region on a taken branch, but otherwise fall through to the next instruction. */ tcg_gen_brcondi_i32(TCG_COND_EQ, ds, 0, l1); /* Leave the gUSA region. */ - tcg_gen_movi_i32(cpu_flags, ctx->envflags & ~GUSA_MASK); + tcg_gen_movi_i32(cpu_flags, ctx->envflags & ~TB_FLAG_GUSA_MASK); gen_jump(ctx); gen_set_label(l1); @@ -XXX,XX +XXX,XX @@ static inline void gen_store_fpr64(DisasContext *ctx, TCGv_i64 t, int reg) #define XHACK(x) ((((x) & 1 ) << 4) | ((x) & 0xe)) #define CHECK_NOT_DELAY_SLOT \ - if (ctx->envflags & DELAY_SLOT_MASK) { \ - goto do_illegal_slot; \ + if (ctx->envflags & TB_FLAG_DELAY_SLOT_MASK) { \ + goto do_illegal_slot; \ } #define CHECK_PRIVILEGED \ @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) case 0x000b: /* rts */ CHECK_NOT_DELAY_SLOT tcg_gen_mov_i32(cpu_delayed_pc, cpu_pr); - ctx->envflags |= DELAY_SLOT; + ctx->envflags |= TB_FLAG_DELAY_SLOT; ctx->delayed_pc = (uint32_t) - 1; return; case 0x0028: /* clrmac */ @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) CHECK_NOT_DELAY_SLOT gen_write_sr(cpu_ssr); tcg_gen_mov_i32(cpu_delayed_pc, cpu_spc); - ctx->envflags |= DELAY_SLOT_RTE; + ctx->envflags |= TB_FLAG_DELAY_SLOT_RTE; ctx->delayed_pc = (uint32_t) - 1; ctx->base.is_jmp = DISAS_STOP; return; @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) return; case 0xe000: /* mov #imm,Rn */ #ifdef CONFIG_USER_ONLY - /* Detect the start of a gUSA region. If so, update envflags - and end the TB. This will allow us to see the end of the - region (stored in R0) in the next TB. */ + /* + * Detect the start of a gUSA region (mov #-n, r15). + * If so, update envflags and end the TB. This will allow us + * to see the end of the region (stored in R0) in the next TB. + */ if (B11_8 == 15 && B7_0s < 0 && (tb_cflags(ctx->base.tb) & CF_PARALLEL)) { - ctx->envflags = deposit32(ctx->envflags, GUSA_SHIFT, 8, B7_0s); + ctx->envflags = + deposit32(ctx->envflags, TB_FLAG_GUSA_SHIFT, 8, B7_0s); ctx->base.is_jmp = DISAS_STOP; } #endif @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) case 0xa000: /* bra disp */ CHECK_NOT_DELAY_SLOT ctx->delayed_pc = ctx->base.pc_next + 4 + B11_0s * 2; - ctx->envflags |= DELAY_SLOT; + ctx->envflags |= TB_FLAG_DELAY_SLOT; return; case 0xb000: /* bsr disp */ CHECK_NOT_DELAY_SLOT tcg_gen_movi_i32(cpu_pr, ctx->base.pc_next + 4); ctx->delayed_pc = ctx->base.pc_next + 4 + B11_0s * 2; - ctx->envflags |= DELAY_SLOT; + ctx->envflags |= TB_FLAG_DELAY_SLOT; return; } @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) CHECK_NOT_DELAY_SLOT tcg_gen_xori_i32(cpu_delayed_cond, cpu_sr_t, 1); ctx->delayed_pc = ctx->base.pc_next + 4 + B7_0s * 2; - ctx->envflags |= DELAY_SLOT_CONDITIONAL; + ctx->envflags |= TB_FLAG_DELAY_SLOT_COND; return; case 0x8900: /* bt label */ CHECK_NOT_DELAY_SLOT @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) CHECK_NOT_DELAY_SLOT tcg_gen_mov_i32(cpu_delayed_cond, cpu_sr_t); ctx->delayed_pc = ctx->base.pc_next + 4 + B7_0s * 2; - ctx->envflags |= DELAY_SLOT_CONDITIONAL; + ctx->envflags |= TB_FLAG_DELAY_SLOT_COND; return; case 0x8800: /* cmp/eq #imm,R0 */ tcg_gen_setcondi_i32(TCG_COND_EQ, cpu_sr_t, REG(0), B7_0s); @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) case 0x0023: /* braf Rn */ CHECK_NOT_DELAY_SLOT tcg_gen_addi_i32(cpu_delayed_pc, REG(B11_8), ctx->base.pc_next + 4); - ctx->envflags |= DELAY_SLOT; + ctx->envflags |= TB_FLAG_DELAY_SLOT; ctx->delayed_pc = (uint32_t) - 1; return; case 0x0003: /* bsrf Rn */ CHECK_NOT_DELAY_SLOT tcg_gen_movi_i32(cpu_pr, ctx->base.pc_next + 4); tcg_gen_add_i32(cpu_delayed_pc, REG(B11_8), cpu_pr); - ctx->envflags |= DELAY_SLOT; + ctx->envflags |= TB_FLAG_DELAY_SLOT; ctx->delayed_pc = (uint32_t) - 1; return; case 0x4015: /* cmp/pl Rn */ @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) case 0x402b: /* jmp @Rn */ CHECK_NOT_DELAY_SLOT tcg_gen_mov_i32(cpu_delayed_pc, REG(B11_8)); - ctx->envflags |= DELAY_SLOT; + ctx->envflags |= TB_FLAG_DELAY_SLOT; ctx->delayed_pc = (uint32_t) - 1; return; case 0x400b: /* jsr @Rn */ CHECK_NOT_DELAY_SLOT tcg_gen_movi_i32(cpu_pr, ctx->base.pc_next + 4); tcg_gen_mov_i32(cpu_delayed_pc, REG(B11_8)); - ctx->envflags |= DELAY_SLOT; + ctx->envflags |= TB_FLAG_DELAY_SLOT; ctx->delayed_pc = (uint32_t) - 1; return; case 0x400e: /* ldc Rm,SR */ @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) fflush(stderr); #endif do_illegal: - if (ctx->envflags & DELAY_SLOT_MASK) { + if (ctx->envflags & TB_FLAG_DELAY_SLOT_MASK) { do_illegal_slot: gen_save_cpu_state(ctx, true); gen_helper_raise_slot_illegal_instruction(cpu_env); @@ -XXX,XX +XXX,XX @@ static void _decode_opc(DisasContext * ctx) do_fpu_disabled: gen_save_cpu_state(ctx, true); - if (ctx->envflags & DELAY_SLOT_MASK) { + if (ctx->envflags & TB_FLAG_DELAY_SLOT_MASK) { gen_helper_raise_slot_fpu_disable(cpu_env); } else { gen_helper_raise_fpu_disable(cpu_env); @@ -XXX,XX +XXX,XX @@ static void decode_opc(DisasContext * ctx) _decode_opc(ctx); - if (old_flags & DELAY_SLOT_MASK) { + if (old_flags & TB_FLAG_DELAY_SLOT_MASK) { /* go out of the delay slot */ - ctx->envflags &= ~DELAY_SLOT_MASK; + ctx->envflags &= ~TB_FLAG_DELAY_SLOT_MASK; /* When in an exclusive region, we must continue to the end for conditional branches. */ - if (ctx->tbflags & GUSA_EXCLUSIVE - && old_flags & DELAY_SLOT_CONDITIONAL) { + if (ctx->tbflags & TB_FLAG_GUSA_EXCLUSIVE + && old_flags & TB_FLAG_DELAY_SLOT_COND) { gen_delayed_conditional_jump(ctx); return; } /* Otherwise this is probably an invalid gUSA region. Drop the GUSA bits so the next TB doesn't see them. */ - ctx->envflags &= ~GUSA_MASK; + ctx->envflags &= ~TB_FLAG_GUSA_MASK; tcg_gen_movi_i32(cpu_flags, ctx->envflags); - if (old_flags & DELAY_SLOT_CONDITIONAL) { + if (old_flags & TB_FLAG_DELAY_SLOT_COND) { gen_delayed_conditional_jump(ctx); } else { gen_jump(ctx); @@ -XXX,XX +XXX,XX @@ static void decode_gusa(DisasContext *ctx, CPUSH4State *env) } /* The entire region has been translated. */ - ctx->envflags &= ~GUSA_MASK; + ctx->envflags &= ~TB_FLAG_GUSA_MASK; ctx->base.pc_next = pc_end; ctx->base.num_insns += max_insns - 1; return; @@ -XXX,XX +XXX,XX @@ static void decode_gusa(DisasContext *ctx, CPUSH4State *env) /* Restart with the EXCLUSIVE bit set, within a TB run via cpu_exec_step_atomic holding the exclusive lock. */ - ctx->envflags |= GUSA_EXCLUSIVE; + ctx->envflags |= TB_FLAG_GUSA_EXCLUSIVE; gen_save_cpu_state(ctx, false); gen_helper_exclusive(cpu_env); ctx->base.is_jmp = DISAS_NORETURN; @@ -XXX,XX +XXX,XX @@ static void sh4_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) (tbflags & (1 << SR_RB))) * 0x10; ctx->fbank = tbflags & FPSCR_FR ? 0x10 : 0; - if (tbflags & GUSA_MASK) { +#ifdef CONFIG_USER_ONLY + if (tbflags & TB_FLAG_GUSA_MASK) { + /* In gUSA exclusive region. */ uint32_t pc = ctx->base.pc_next; uint32_t pc_end = ctx->base.tb->cs_base; - int backup = sextract32(ctx->tbflags, GUSA_SHIFT, 8); + int backup = sextract32(ctx->tbflags, TB_FLAG_GUSA_SHIFT, 8); int max_insns = (pc_end - pc) / 2; if (pc != pc_end + backup || max_insns < 2) { /* This is a malformed gUSA region. Don't do anything special, since the interpreter is likely to get confused. */ - ctx->envflags &= ~GUSA_MASK; - } else if (tbflags & GUSA_EXCLUSIVE) { + ctx->envflags &= ~TB_FLAG_GUSA_MASK; + } else if (tbflags & TB_FLAG_GUSA_EXCLUSIVE) { /* Regardless of single-stepping or the end of the page, we must complete execution of the gUSA region while holding the exclusive lock. */ @@ -XXX,XX +XXX,XX @@ static void sh4_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) return; } } +#endif /* Since the ISA is fixed-width, we can bound by the number of instructions remaining on the page. */ @@ -XXX,XX +XXX,XX @@ static void sh4_tr_translate_insn(DisasContextBase *dcbase, CPUState *cs) DisasContext *ctx = container_of(dcbase, DisasContext, base); #ifdef CONFIG_USER_ONLY - if (unlikely(ctx->envflags & GUSA_MASK) - && !(ctx->envflags & GUSA_EXCLUSIVE)) { + if (unlikely(ctx->envflags & TB_FLAG_GUSA_MASK) + && !(ctx->envflags & TB_FLAG_GUSA_EXCLUSIVE)) { /* We're in an gUSA region, and we have not already fallen back on using an exclusive region. Attempt to parse the region into a single supported atomic operation. Failure @@ -XXX,XX +XXX,XX @@ static void sh4_tr_tb_stop(DisasContextBase *dcbase, CPUState *cs) { DisasContext *ctx = container_of(dcbase, DisasContext, base); - if (ctx->tbflags & GUSA_EXCLUSIVE) { + if (ctx->tbflags & TB_FLAG_GUSA_EXCLUSIVE) { /* Ending the region of exclusivity. Clear the bits. */ - ctx->envflags &= ~GUSA_MASK; + ctx->envflags &= ~TB_FLAG_GUSA_MASK; } switch (ctx->base.is_jmp) { -- 2.34.1