Just my fp16 work, plus some small stuff for the sbsa-ref board; but my rule of thumb is to send a pullreq once I get over about 30 patches... -- PMM The following changes since commit 2f4c51c0f384d7888a04b4815861e6d5fd244d75: Merge remote-tracking branch 'remotes/kraxel/tags/usb-20200831-pull-request' into staging (2020-08-31 19:39:13 +0100) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20200901 for you to fetch changes up to 3f462bf0f6ea6382dd1502d4eb1fcd33c8e774f5: hw/arm/sbsa-ref : Add embedded controller in secure memory (2020-09-01 14:01:34 +0100) ---------------------------------------------------------------- target-arm queue: * Implement fp16 support for AArch32 VFP and Neon * hw/arm/sbsa-ref: add "reg" property to DT cpu nodes * hw/arm/sbsa-ref : Add embedded controller in secure memory ---------------------------------------------------------------- Graeme Gregory (2): hw/misc/sbsa_ec : Add an embedded controller for sbsa-ref hw/arm/sbsa-ref : Add embedded controller in secure memory Leif Lindholm (1): hw/arm/sbsa-ref: add "reg" property to DT cpu nodes Peter Maydell (44): target/arm: Remove local definitions of float constants target/arm: Use correct ID register check for aa32_fp16_arith target/arm: Implement VFP fp16 for VFP_BINOP operations target/arm: Implement VFP fp16 VMLA, VMLS, VNMLS, VNMLA, VNMUL target/arm: Macroify trans functions for VFMA, VFMS, VFNMA, VFNMS target/arm: Implement VFP fp16 for fused-multiply-add target/arm: Macroify uses of do_vfp_2op_sp() and do_vfp_2op_dp() target/arm: Implement VFP fp16 for VABS, VNEG, VSQRT target/arm: Implement VFP fp16 for VMOV immediate target/arm: Implement VFP fp16 VCMP target/arm: Implement VFP fp16 VLDR and VSTR target/arm: Implement VFP fp16 VCVT between float and integer target/arm: Make VFP_CONV_FIX macros take separate float type and float size target/arm: Use macros instead of open-coding fp16 conversion helpers target/arm: Implement VFP fp16 VCVT between float and fixed-point target/arm: Implement VFP vp16 VCVT-with-specified-rounding-mode target/arm: Implement VFP fp16 VSEL target/arm: Implement VFP fp16 VRINT* target/arm: Implement new VFP fp16 insn VINS target/arm: Implement new VFP fp16 insn VMOVX target/arm: Implement VFP fp16 VMOV between gp and halfprec registers target/arm: Implement FP16 for Neon VADD, VSUB, VABD, VMUL target/arm: Implement fp16 for Neon VRECPE, VRSQRTE using gvec target/arm: Implement fp16 for Neon VABS, VNEG of floats target/arm: Implement fp16 for VCEQ, VCGE, VCGT comparisons target/arm: Implement fp16 for VACGE, VACGT target/arm: Implement fp16 for Neon VMAX, VMIN target/arm: Implement fp16 for Neon VMAXNM, VMINNM target/arm: Implement fp16 for Neon VMLA, VMLS operations target/arm: Implement fp16 for Neon VFMA, VMFS target/arm: Implement fp16 for Neon fp compare-vs-0 target/arm: Implement fp16 for Neon VRECPS target/arm: Implement fp16 for Neon VRSQRTS target/arm: Implement fp16 for Neon pairwise fp ops target/arm: Implement fp16 for Neon float-integer VCVT target/arm: Convert Neon VCVT fixed-point to gvec target/arm: Implement fp16 for Neon VCVT fixed-point target/arm: Implement fp16 for Neon VCVT with rounding modes target/arm: Implement fp16 for Neon VRINT-with-specified-rounding-mode target/arm: Implement fp16 for Neon VRINTX target/arm/vec_helper: Handle oprsz less than 16 bytes in indexed operations target/arm/vec_helper: Add gvec fp indexed multiply-and-add operations target/arm: Implement fp16 for Neon VMUL, VMLA, VMLS target/arm: Enable FP16 in '-cpu max' target/arm/cpu.h | 7 +- target/arm/helper.h | 133 ++++++- target/arm/neon-dp.decode | 8 +- target/arm/vfp-uncond.decode | 27 +- target/arm/vfp.decode | 34 +- hw/arm/sbsa-ref.c | 43 ++- hw/misc/sbsa_ec.c | 98 +++++ target/arm/cpu.c | 3 +- target/arm/cpu64.c | 10 +- target/arm/helper-a64.c | 11 - target/arm/translate-sve.c | 4 - target/arm/vec_helper.c | 431 ++++++++++++++++++++- target/arm/vfp_helper.c | 244 +++++------- hw/misc/meson.build | 2 + target/arm/translate-neon.c.inc | 755 +++++++++++++------------------------ target/arm/translate-vfp.c.inc | 810 ++++++++++++++++++++++++++++++++++++---- 16 files changed, 1819 insertions(+), 801 deletions(-) create mode 100644 hw/misc/sbsa_ec.c In several places the target/arm code defines local float constants for 2, 3 and 1.5, which are also provided by include/fpu/softfloat.h. Remove the unnecessary local duplicate versions. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-2-peter.maydell@linaro.org --- target/arm/helper-a64.c | 11 ----------- target/arm/translate-sve.c | 4 ---- target/arm/vfp_helper.c | 4 ---- 3 files changed, 19 deletions(-) diff --git a/target/arm/helper-a64.c b/target/arm/helper-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-a64.c +++ b/target/arm/helper-a64.c @@ -XXX,XX +XXX,XX @@ uint64_t HELPER(neon_cgt_f64)(float64 a, float64 b, void *fpstp) * versions, these do a fully fused multiply-add or * multiply-add-and-halve. */ -#define float16_two make_float16(0x4000) -#define float16_three make_float16(0x4200) -#define float16_one_point_five make_float16(0x3e00) - -#define float32_two make_float32(0x40000000) -#define float32_three make_float32(0x40400000) -#define float32_one_point_five make_float32(0x3fc00000) - -#define float64_two make_float64(0x4000000000000000ULL) -#define float64_three make_float64(0x4008000000000000ULL) -#define float64_one_point_five make_float64(0x3FF8000000000000ULL) uint32_t HELPER(recpsf_f16)(uint32_t a, uint32_t b, void *fpstp) { diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_##NAME##_zpzi(DisasContext *s, arg_rpri_esz *a) \ return true; \ } -#define float16_two make_float16(0x4000) -#define float32_two make_float32(0x40000000) -#define float64_two make_float64(0x4000000000000000ULL) - DO_FP_IMM(FADD, fadds, half, one) DO_FP_IMM(FSUB, fsubs, half, one) DO_FP_IMM(FMUL, fmuls, half, two) diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(vfp_fcvt_f64_to_f16)(float64 a, void *fpstp, uint32_t ahp_mode) return r; } -#define float32_two make_float32(0x40000000) -#define float32_three make_float32(0x40400000) -#define float32_one_point_five make_float32(0x3fc00000) - float32 HELPER(recps_f32)(CPUARMState *env, float32 a, float32 b) { float_status *s = &env->vfp.standard_fp_status; -- 2.20.1 The aa32_fp16_arith feature check function currently looks at the AArch64 ID_AA64PFR0 register. This is (as the comment notes) not correct. The bogus check was put in mostly to allow testing of the fp16 variants of the VCMLA instructions and it was something of a mistake that we allowed them to exist in master. Switch the feature check function to testing VMFR1.FPHP, which is what it ought to be. This will remove emulation of the VCMLA and VCADD insns from AArch32 code running on an AArch64 '-cpu max' using system emulation. (They were never enabled for aarch32 linux-user and system-emulation.) Since we weren't advertising their existence via the AArch32 ID register, well-behaved guests wouldn't have been using them anyway. Once we have implemented all the AArch32 support for the FP16 extension we will advertise it in the MVFR1 ID register field, which will reenable these insns along with all the others. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-3-peter.maydell@linaro.org --- target/arm/cpu.h | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa32_predinv(const ARMISARegisters *id) static inline bool isar_feature_aa32_fp16_arith(const ARMISARegisters *id) { - /* - * This is a placeholder for use by VCMA until the rest of - * the ARMv8.2-FP16 extension is implemented for aa32 mode. - * At which point we can properly set and check MVFR1.FPHP. - */ - return FIELD_EX64(id->id_aa64pfr0, ID_AA64PFR0, FP) == 1; + return FIELD_EX32(id->mvfr1, MVFR1, FPHP) >= 3; } static inline bool isar_feature_aa32_vfp_simd(const ARMISARegisters *id) -- 2.20.1 Implmeent VFP fp16 support for simple binary-operator VFP insns VADD, VSUB, VMUL, VDIV, VMINNM and VMAXNM: * make the VFP_BINOP() macro generate float16 helpers as well as float32 and float64 * implement a do_vfp_3op_hp() function similar to the existing do_vfp_3op_sp() * add decode for the half-precision insn patterns Note that the VFP_BINOP macro use creates a couple of unused helper functions vfp_maxh and vfp_minh, but they're small so it's not worth splitting the BINOP operations into "needs halfprec" and "no halfprec" groups. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-4-peter.maydell@linaro.org --- target/arm/helper.h | 8 ++++ target/arm/vfp-uncond.decode | 3 ++ target/arm/vfp.decode | 4 ++ target/arm/vfp_helper.c | 5 ++ target/arm/translate-vfp.c.inc | 86 ++++++++++++++++++++++++++++++++++ 5 files changed, 106 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(probe_access, TCG_CALL_NO_WG, void, env, tl, i32, i32, i32) DEF_HELPER_1(vfp_get_fpscr, i32, env) DEF_HELPER_2(vfp_set_fpscr, void, env, i32) +DEF_HELPER_3(vfp_addh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_adds, f32, f32, f32, ptr) DEF_HELPER_3(vfp_addd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_subh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_subs, f32, f32, f32, ptr) DEF_HELPER_3(vfp_subd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_mulh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_muls, f32, f32, f32, ptr) DEF_HELPER_3(vfp_muld, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_divh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_divs, f32, f32, f32, ptr) DEF_HELPER_3(vfp_divd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_maxh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_maxs, f32, f32, f32, ptr) DEF_HELPER_3(vfp_maxd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_minh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_mins, f32, f32, f32, ptr) DEF_HELPER_3(vfp_mind, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_maxnumh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_maxnums, f32, f32, f32, ptr) DEF_HELPER_3(vfp_maxnumd, f64, f64, f64, ptr) +DEF_HELPER_3(vfp_minnumh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_minnums, f32, f32, f32, ptr) DEF_HELPER_3(vfp_minnumd, f64, f64, f64, ptr) DEF_HELPER_1(vfp_negs, f32, f32) diff --git a/target/arm/vfp-uncond.decode b/target/arm/vfp-uncond.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp-uncond.decode +++ b/target/arm/vfp-uncond.decode @@ -XXX,XX +XXX,XX @@ VSEL 1111 1110 0. cc:2 .... .... 1010 .0.0 .... \ VSEL 1111 1110 0. cc:2 .... .... 1011 .0.0 .... \ vm=%vm_dp vn=%vn_dp vd=%vd_dp dp=1 +VMAXNM_hp 1111 1110 1.00 .... .... 1001 .0.0 .... @vfp_dnm_s +VMINNM_hp 1111 1110 1.00 .... .... 1001 .1.0 .... @vfp_dnm_s + VMAXNM_sp 1111 1110 1.00 .... .... 1010 .0.0 .... @vfp_dnm_s VMINNM_sp 1111 1110 1.00 .... .... 1010 .1.0 .... @vfp_dnm_s diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VNMLS_dp ---- 1110 0.01 .... .... 1011 .0.0 .... @vfp_dnm_d VNMLA_sp ---- 1110 0.01 .... .... 1010 .1.0 .... @vfp_dnm_s VNMLA_dp ---- 1110 0.01 .... .... 1011 .1.0 .... @vfp_dnm_d +VMUL_hp ---- 1110 0.10 .... .... 1001 .0.0 .... @vfp_dnm_s VMUL_sp ---- 1110 0.10 .... .... 1010 .0.0 .... @vfp_dnm_s VMUL_dp ---- 1110 0.10 .... .... 1011 .0.0 .... @vfp_dnm_d VNMUL_sp ---- 1110 0.10 .... .... 1010 .1.0 .... @vfp_dnm_s VNMUL_dp ---- 1110 0.10 .... .... 1011 .1.0 .... @vfp_dnm_d +VADD_hp ---- 1110 0.11 .... .... 1001 .0.0 .... @vfp_dnm_s VADD_sp ---- 1110 0.11 .... .... 1010 .0.0 .... @vfp_dnm_s VADD_dp ---- 1110 0.11 .... .... 1011 .0.0 .... @vfp_dnm_d +VSUB_hp ---- 1110 0.11 .... .... 1001 .1.0 .... @vfp_dnm_s VSUB_sp ---- 1110 0.11 .... .... 1010 .1.0 .... @vfp_dnm_s VSUB_dp ---- 1110 0.11 .... .... 1011 .1.0 .... @vfp_dnm_d +VDIV_hp ---- 1110 1.00 .... .... 1001 .0.0 .... @vfp_dnm_s VDIV_sp ---- 1110 1.00 .... .... 1010 .0.0 .... @vfp_dnm_s VDIV_dp ---- 1110 1.00 .... .... 1011 .0.0 .... @vfp_dnm_d diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ void vfp_set_fpscr(CPUARMState *env, uint32_t val) #define VFP_HELPER(name, p) HELPER(glue(glue(vfp_,name),p)) #define VFP_BINOP(name) \ +dh_ctype_f16 VFP_HELPER(name, h)(dh_ctype_f16 a, dh_ctype_f16 b, void *fpstp) \ +{ \ + float_status *fpst = fpstp; \ + return float16_ ## name(a, b, fpst); \ +} \ float32 VFP_HELPER(name, s)(float32 a, float32 b, void *fpstp) \ { \ float_status *fpst = fpstp; \ diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool do_vfp_3op_sp(DisasContext *s, VFPGen3OpSPFn *fn, return true; } +static bool do_vfp_3op_hp(DisasContext *s, VFPGen3OpSPFn *fn, + int vd, int vn, int vm, bool reads_vd) +{ + /* + * Do a half-precision operation. Functionally this is + * the same as do_vfp_3op_sp(), except: + * - it uses the FPST_FPCR_F16 + * - it doesn't need the VFP vector handling (fp16 is a + * v8 feature, and in v8 VFP vectors don't exist) + * - it does the aa32_fp16_arith feature test + */ + TCGv_i32 f0, f1, fd; + TCGv_ptr fpst; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (s->vec_len != 0 || s->vec_stride != 0) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + f0 = tcg_temp_new_i32(); + f1 = tcg_temp_new_i32(); + fd = tcg_temp_new_i32(); + fpst = fpstatus_ptr(FPST_FPCR_F16); + + neon_load_reg32(f0, vn); + neon_load_reg32(f1, vm); + + if (reads_vd) { + neon_load_reg32(fd, vd); + } + fn(fd, f0, f1, fpst); + neon_store_reg32(fd, vd); + + tcg_temp_free_i32(f0); + tcg_temp_free_i32(f1); + tcg_temp_free_i32(fd); + tcg_temp_free_ptr(fpst); + + return true; +} + static bool do_vfp_3op_dp(DisasContext *s, VFPGen3OpDPFn *fn, int vd, int vn, int vm, bool reads_vd) { @@ -XXX,XX +XXX,XX @@ static bool trans_VNMLA_dp(DisasContext *s, arg_VNMLA_dp *a) return do_vfp_3op_dp(s, gen_VNMLA_dp, a->vd, a->vn, a->vm, true); } +static bool trans_VMUL_hp(DisasContext *s, arg_VMUL_sp *a) +{ + return do_vfp_3op_hp(s, gen_helper_vfp_mulh, a->vd, a->vn, a->vm, false); +} + static bool trans_VMUL_sp(DisasContext *s, arg_VMUL_sp *a) { return do_vfp_3op_sp(s, gen_helper_vfp_muls, a->vd, a->vn, a->vm, false); @@ -XXX,XX +XXX,XX @@ static bool trans_VNMUL_dp(DisasContext *s, arg_VNMUL_dp *a) return do_vfp_3op_dp(s, gen_VNMUL_dp, a->vd, a->vn, a->vm, false); } +static bool trans_VADD_hp(DisasContext *s, arg_VADD_sp *a) +{ + return do_vfp_3op_hp(s, gen_helper_vfp_addh, a->vd, a->vn, a->vm, false); +} + static bool trans_VADD_sp(DisasContext *s, arg_VADD_sp *a) { return do_vfp_3op_sp(s, gen_helper_vfp_adds, a->vd, a->vn, a->vm, false); @@ -XXX,XX +XXX,XX @@ static bool trans_VADD_dp(DisasContext *s, arg_VADD_dp *a) return do_vfp_3op_dp(s, gen_helper_vfp_addd, a->vd, a->vn, a->vm, false); } +static bool trans_VSUB_hp(DisasContext *s, arg_VSUB_sp *a) +{ + return do_vfp_3op_hp(s, gen_helper_vfp_subh, a->vd, a->vn, a->vm, false); +} + static bool trans_VSUB_sp(DisasContext *s, arg_VSUB_sp *a) { return do_vfp_3op_sp(s, gen_helper_vfp_subs, a->vd, a->vn, a->vm, false); @@ -XXX,XX +XXX,XX @@ static bool trans_VSUB_dp(DisasContext *s, arg_VSUB_dp *a) return do_vfp_3op_dp(s, gen_helper_vfp_subd, a->vd, a->vn, a->vm, false); } +static bool trans_VDIV_hp(DisasContext *s, arg_VDIV_sp *a) +{ + return do_vfp_3op_hp(s, gen_helper_vfp_divh, a->vd, a->vn, a->vm, false); +} + static bool trans_VDIV_sp(DisasContext *s, arg_VDIV_sp *a) { return do_vfp_3op_sp(s, gen_helper_vfp_divs, a->vd, a->vn, a->vm, false); @@ -XXX,XX +XXX,XX @@ static bool trans_VDIV_dp(DisasContext *s, arg_VDIV_dp *a) return do_vfp_3op_dp(s, gen_helper_vfp_divd, a->vd, a->vn, a->vm, false); } +static bool trans_VMINNM_hp(DisasContext *s, arg_VMINNM_sp *a) +{ + if (!dc_isar_feature(aa32_vminmaxnm, s)) { + return false; + } + return do_vfp_3op_hp(s, gen_helper_vfp_minnumh, + a->vd, a->vn, a->vm, false); +} + +static bool trans_VMAXNM_hp(DisasContext *s, arg_VMAXNM_sp *a) +{ + if (!dc_isar_feature(aa32_vminmaxnm, s)) { + return false; + } + return do_vfp_3op_hp(s, gen_helper_vfp_maxnumh, + a->vd, a->vn, a->vm, false); +} + static bool trans_VMINNM_sp(DisasContext *s, arg_VMINNM_sp *a) { if (!dc_isar_feature(aa32_vminmaxnm, s)) { -- 2.20.1 Implement fp16 versions of the VFP VMLA, VMLS, VNMLS, VNMLA, VNMUL instructions. (These are all the remaining ones which we implement via do_vfp_3op_[hsd]p().) Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-5-peter.maydell@linaro.org --- target/arm/helper.h | 1 + target/arm/vfp.decode | 5 ++ target/arm/vfp_helper.c | 5 ++ target/arm/translate-vfp.c.inc | 84 ++++++++++++++++++++++++++++++++++ 4 files changed, 95 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_3(vfp_maxnumd, f64, f64, f64, ptr) DEF_HELPER_3(vfp_minnumh, f16, f16, f16, ptr) DEF_HELPER_3(vfp_minnums, f32, f32, f32, ptr) DEF_HELPER_3(vfp_minnumd, f64, f64, f64, ptr) +DEF_HELPER_1(vfp_negh, f16, f16) DEF_HELPER_1(vfp_negs, f32, f32) DEF_HELPER_1(vfp_negd, f64, f64) DEF_HELPER_1(vfp_abss, f32, f32) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VLDM_VSTM_dp ---- 1101 0.1 l:1 rn:4 .... 1011 imm:8 \ vd=%vd_dp p=1 u=0 w=1 # 3-register VFP data-processing; bits [23,21:20,6] identify the operation. +VMLA_hp ---- 1110 0.00 .... .... 1001 .0.0 .... @vfp_dnm_s VMLA_sp ---- 1110 0.00 .... .... 1010 .0.0 .... @vfp_dnm_s VMLA_dp ---- 1110 0.00 .... .... 1011 .0.0 .... @vfp_dnm_d +VMLS_hp ---- 1110 0.00 .... .... 1001 .1.0 .... @vfp_dnm_s VMLS_sp ---- 1110 0.00 .... .... 1010 .1.0 .... @vfp_dnm_s VMLS_dp ---- 1110 0.00 .... .... 1011 .1.0 .... @vfp_dnm_d +VNMLS_hp ---- 1110 0.01 .... .... 1001 .0.0 .... @vfp_dnm_s VNMLS_sp ---- 1110 0.01 .... .... 1010 .0.0 .... @vfp_dnm_s VNMLS_dp ---- 1110 0.01 .... .... 1011 .0.0 .... @vfp_dnm_d +VNMLA_hp ---- 1110 0.01 .... .... 1001 .1.0 .... @vfp_dnm_s VNMLA_sp ---- 1110 0.01 .... .... 1010 .1.0 .... @vfp_dnm_s VNMLA_dp ---- 1110 0.01 .... .... 1011 .1.0 .... @vfp_dnm_d @@ -XXX,XX +XXX,XX @@ VMUL_hp ---- 1110 0.10 .... .... 1001 .0.0 .... @vfp_dnm_s VMUL_sp ---- 1110 0.10 .... .... 1010 .0.0 .... @vfp_dnm_s VMUL_dp ---- 1110 0.10 .... .... 1011 .0.0 .... @vfp_dnm_d +VNMUL_hp ---- 1110 0.10 .... .... 1001 .1.0 .... @vfp_dnm_s VNMUL_sp ---- 1110 0.10 .... .... 1010 .1.0 .... @vfp_dnm_s VNMUL_dp ---- 1110 0.10 .... .... 1011 .1.0 .... @vfp_dnm_d diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ VFP_BINOP(minnum) VFP_BINOP(maxnum) #undef VFP_BINOP +dh_ctype_f16 VFP_HELPER(neg, h)(dh_ctype_f16 a) +{ + return float16_chs(a); +} + float32 VFP_HELPER(neg, s)(float32 a) { return float32_chs(a); diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool do_vfp_2op_dp(DisasContext *s, VFPGen2OpDPFn *fn, int vd, int vm) return true; } +static void gen_VMLA_hp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) +{ + /* Note that order of inputs to the add matters for NaNs */ + TCGv_i32 tmp = tcg_temp_new_i32(); + + gen_helper_vfp_mulh(tmp, vn, vm, fpst); + gen_helper_vfp_addh(vd, vd, tmp, fpst); + tcg_temp_free_i32(tmp); +} + +static bool trans_VMLA_hp(DisasContext *s, arg_VMLA_sp *a) +{ + return do_vfp_3op_hp(s, gen_VMLA_hp, a->vd, a->vn, a->vm, true); +} + static void gen_VMLA_sp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) { /* Note that order of inputs to the add matters for NaNs */ @@ -XXX,XX +XXX,XX @@ static bool trans_VMLA_dp(DisasContext *s, arg_VMLA_dp *a) return do_vfp_3op_dp(s, gen_VMLA_dp, a->vd, a->vn, a->vm, true); } +static void gen_VMLS_hp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) +{ + /* + * VMLS: vd = vd + -(vn * vm) + * Note that order of inputs to the add matters for NaNs. + */ + TCGv_i32 tmp = tcg_temp_new_i32(); + + gen_helper_vfp_mulh(tmp, vn, vm, fpst); + gen_helper_vfp_negh(tmp, tmp); + gen_helper_vfp_addh(vd, vd, tmp, fpst); + tcg_temp_free_i32(tmp); +} + +static bool trans_VMLS_hp(DisasContext *s, arg_VMLS_sp *a) +{ + return do_vfp_3op_hp(s, gen_VMLS_hp, a->vd, a->vn, a->vm, true); +} + static void gen_VMLS_sp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) { /* @@ -XXX,XX +XXX,XX @@ static bool trans_VMLS_dp(DisasContext *s, arg_VMLS_dp *a) return do_vfp_3op_dp(s, gen_VMLS_dp, a->vd, a->vn, a->vm, true); } +static void gen_VNMLS_hp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) +{ + /* + * VNMLS: -fd + (fn * fm) + * Note that it isn't valid to replace (-A + B) with (B - A) or similar + * plausible looking simplifications because this will give wrong results + * for NaNs. + */ + TCGv_i32 tmp = tcg_temp_new_i32(); + + gen_helper_vfp_mulh(tmp, vn, vm, fpst); + gen_helper_vfp_negh(vd, vd); + gen_helper_vfp_addh(vd, vd, tmp, fpst); + tcg_temp_free_i32(tmp); +} + +static bool trans_VNMLS_hp(DisasContext *s, arg_VNMLS_sp *a) +{ + return do_vfp_3op_hp(s, gen_VNMLS_hp, a->vd, a->vn, a->vm, true); +} + static void gen_VNMLS_sp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) { /* @@ -XXX,XX +XXX,XX @@ static bool trans_VNMLS_dp(DisasContext *s, arg_VNMLS_dp *a) return do_vfp_3op_dp(s, gen_VNMLS_dp, a->vd, a->vn, a->vm, true); } +static void gen_VNMLA_hp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) +{ + /* VNMLA: -fd + -(fn * fm) */ + TCGv_i32 tmp = tcg_temp_new_i32(); + + gen_helper_vfp_mulh(tmp, vn, vm, fpst); + gen_helper_vfp_negh(tmp, tmp); + gen_helper_vfp_negh(vd, vd); + gen_helper_vfp_addh(vd, vd, tmp, fpst); + tcg_temp_free_i32(tmp); +} + +static bool trans_VNMLA_hp(DisasContext *s, arg_VNMLA_sp *a) +{ + return do_vfp_3op_hp(s, gen_VNMLA_hp, a->vd, a->vn, a->vm, true); +} + static void gen_VNMLA_sp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) { /* VNMLA: -fd + -(fn * fm) */ @@ -XXX,XX +XXX,XX @@ static bool trans_VMUL_dp(DisasContext *s, arg_VMUL_dp *a) return do_vfp_3op_dp(s, gen_helper_vfp_muld, a->vd, a->vn, a->vm, false); } +static void gen_VNMUL_hp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) +{ + /* VNMUL: -(fn * fm) */ + gen_helper_vfp_mulh(vd, vn, vm, fpst); + gen_helper_vfp_negh(vd, vd); +} + +static bool trans_VNMUL_hp(DisasContext *s, arg_VNMUL_sp *a) +{ + return do_vfp_3op_hp(s, gen_VNMUL_hp, a->vd, a->vn, a->vm, false); +} + static void gen_VNMUL_sp(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpst) { /* VNMUL: -(fn * fm) */ -- 2.20.1 Macroify creation of the trans functions for single and double precision VFMA, VFMS, VFNMA, VFNMS. The repetition was OK for two sizes, but we're about to add halfprec and it will get a bit more than seems reasonable. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-6-peter.maydell@linaro.org --- target/arm/translate-vfp.c.inc | 50 +++++++++------------------------- 1 file changed, 13 insertions(+), 37 deletions(-) diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool do_vfm_sp(DisasContext *s, arg_VFMA_sp *a, bool neg_n, bool neg_d) return true; } -static bool trans_VFMA_sp(DisasContext *s, arg_VFMA_sp *a) -{ - return do_vfm_sp(s, a, false, false); -} - -static bool trans_VFMS_sp(DisasContext *s, arg_VFMS_sp *a) -{ - return do_vfm_sp(s, a, true, false); -} - -static bool trans_VFNMA_sp(DisasContext *s, arg_VFNMA_sp *a) -{ - return do_vfm_sp(s, a, false, true); -} - -static bool trans_VFNMS_sp(DisasContext *s, arg_VFNMS_sp *a) -{ - return do_vfm_sp(s, a, true, true); -} - static bool do_vfm_dp(DisasContext *s, arg_VFMA_dp *a, bool neg_n, bool neg_d) { /* @@ -XXX,XX +XXX,XX @@ static bool do_vfm_dp(DisasContext *s, arg_VFMA_dp *a, bool neg_n, bool neg_d) return true; } -static bool trans_VFMA_dp(DisasContext *s, arg_VFMA_dp *a) -{ - return do_vfm_dp(s, a, false, false); -} +#define MAKE_ONE_VFM_TRANS_FN(INSN, PREC, NEGN, NEGD) \ + static bool trans_##INSN##_##PREC(DisasContext *s, \ + arg_##INSN##_##PREC *a) \ + { \ + return do_vfm_##PREC(s, a, NEGN, NEGD); \ + } -static bool trans_VFMS_dp(DisasContext *s, arg_VFMS_dp *a) -{ - return do_vfm_dp(s, a, true, false); -} +#define MAKE_VFM_TRANS_FNS(PREC) \ + MAKE_ONE_VFM_TRANS_FN(VFMA, PREC, false, false) \ + MAKE_ONE_VFM_TRANS_FN(VFMS, PREC, true, false) \ + MAKE_ONE_VFM_TRANS_FN(VFNMA, PREC, false, true) \ + MAKE_ONE_VFM_TRANS_FN(VFNMS, PREC, true, true) -static bool trans_VFNMA_dp(DisasContext *s, arg_VFNMA_dp *a) -{ - return do_vfm_dp(s, a, false, true); -} - -static bool trans_VFNMS_dp(DisasContext *s, arg_VFNMS_dp *a) -{ - return do_vfm_dp(s, a, true, true); -} +MAKE_VFM_TRANS_FNS(sp) +MAKE_VFM_TRANS_FNS(dp) static bool trans_VMOV_imm_sp(DisasContext *s, arg_VMOV_imm_sp *a) { -- 2.20.1 Implement VFP fp16 support for fused multiply-add insns VFNMA, VFNMS, VFMA, VFMS. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-7-peter.maydell@linaro.org --- target/arm/helper.h | 1 + target/arm/vfp.decode | 5 +++ target/arm/vfp_helper.c | 7 ++++ target/arm/translate-vfp.c.inc | 64 ++++++++++++++++++++++++++++++++++ 4 files changed, 77 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_3(vfp_fcvt_f64_to_f16, TCG_CALL_NO_RWG, f16, f64, ptr, i32) DEF_HELPER_4(vfp_muladdd, f64, f64, f64, f64, ptr) DEF_HELPER_4(vfp_muladds, f32, f32, f32, f32, ptr) +DEF_HELPER_4(vfp_muladdh, f16, f16, f16, f16, ptr) DEF_HELPER_3(recps_f32, f32, env, f32, f32) DEF_HELPER_3(rsqrts_f32, f32, env, f32, f32) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VDIV_hp ---- 1110 1.00 .... .... 1001 .0.0 .... @vfp_dnm_s VDIV_sp ---- 1110 1.00 .... .... 1010 .0.0 .... @vfp_dnm_s VDIV_dp ---- 1110 1.00 .... .... 1011 .0.0 .... @vfp_dnm_d +VFMA_hp ---- 1110 1.10 .... .... 1001 .0. 0 .... @vfp_dnm_s +VFMS_hp ---- 1110 1.10 .... .... 1001 .1. 0 .... @vfp_dnm_s +VFNMA_hp ---- 1110 1.01 .... .... 1001 .0. 0 .... @vfp_dnm_s +VFNMS_hp ---- 1110 1.01 .... .... 1001 .1. 0 .... @vfp_dnm_s + VFMA_sp ---- 1110 1.10 .... .... 1010 .0. 0 .... @vfp_dnm_s VFMS_sp ---- 1110 1.10 .... .... 1010 .1. 0 .... @vfp_dnm_s VFNMA_sp ---- 1110 1.01 .... .... 1010 .0. 0 .... @vfp_dnm_s diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(rsqrte_u32)(uint32_t a) } /* VFPv4 fused multiply-accumulate */ +dh_ctype_f16 VFP_HELPER(muladd, h)(dh_ctype_f16 a, dh_ctype_f16 b, + dh_ctype_f16 c, void *fpstp) +{ + float_status *fpst = fpstp; + return float16_muladd(a, b, c, 0, fpst); +} + float32 VFP_HELPER(muladd, s)(float32 a, float32 b, float32 c, void *fpstp) { float_status *fpst = fpstp; diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VMAXNM_dp(DisasContext *s, arg_VMAXNM_dp *a) a->vd, a->vn, a->vm, false); } +static bool do_vfm_hp(DisasContext *s, arg_VFMA_sp *a, bool neg_n, bool neg_d) +{ + /* + * VFNMA : fd = muladd(-fd, fn, fm) + * VFNMS : fd = muladd(-fd, -fn, fm) + * VFMA : fd = muladd( fd, fn, fm) + * VFMS : fd = muladd( fd, -fn, fm) + * + * These are fused multiply-add, and must be done as one floating + * point operation with no rounding between the multiplication and + * addition steps. NB that doing the negations here as separate + * steps is correct : an input NaN should come out with its sign + * bit flipped if it is a negated-input. + */ + TCGv_ptr fpst; + TCGv_i32 vn, vm, vd; + + /* + * Present in VFPv4 only, and only with the FP16 extension. + * Note that we can't rely on the SIMDFMAC check alone, because + * in a Neon-no-VFP core that ID register field will be non-zero. + */ + if (!dc_isar_feature(aa32_fp16_arith, s) || + !dc_isar_feature(aa32_simdfmac, s) || + !dc_isar_feature(aa32_fpsp_v2, s)) { + return false; + } + + if (s->vec_len != 0 || s->vec_stride != 0) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + vn = tcg_temp_new_i32(); + vm = tcg_temp_new_i32(); + vd = tcg_temp_new_i32(); + + neon_load_reg32(vn, a->vn); + neon_load_reg32(vm, a->vm); + if (neg_n) { + /* VFNMS, VFMS */ + gen_helper_vfp_negh(vn, vn); + } + neon_load_reg32(vd, a->vd); + if (neg_d) { + /* VFNMA, VFNMS */ + gen_helper_vfp_negh(vd, vd); + } + fpst = fpstatus_ptr(FPST_FPCR_F16); + gen_helper_vfp_muladdh(vd, vn, vm, vd, fpst); + neon_store_reg32(vd, a->vd); + + tcg_temp_free_ptr(fpst); + tcg_temp_free_i32(vn); + tcg_temp_free_i32(vm); + tcg_temp_free_i32(vd); + + return true; +} + static bool do_vfm_sp(DisasContext *s, arg_VFMA_sp *a, bool neg_n, bool neg_d) { /* @@ -XXX,XX +XXX,XX @@ static bool do_vfm_dp(DisasContext *s, arg_VFMA_dp *a, bool neg_n, bool neg_d) MAKE_ONE_VFM_TRANS_FN(VFNMA, PREC, false, true) \ MAKE_ONE_VFM_TRANS_FN(VFNMS, PREC, true, true) +MAKE_VFM_TRANS_FNS(hp) MAKE_VFM_TRANS_FNS(sp) MAKE_VFM_TRANS_FNS(dp) -- 2.20.1 Macroify the uses of do_vfp_2op_sp() and do_vfp_2op_dp(); this will make it easier to add the halfprec support. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-8-peter.maydell@linaro.org --- target/arm/translate-vfp.c.inc | 49 ++++++++++------------------------ 1 file changed, 14 insertions(+), 35 deletions(-) diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VMOV_imm_dp(DisasContext *s, arg_VMOV_imm_dp *a) return true; } -static bool trans_VMOV_reg_sp(DisasContext *s, arg_VMOV_reg_sp *a) -{ - return do_vfp_2op_sp(s, tcg_gen_mov_i32, a->vd, a->vm); -} +#define DO_VFP_2OP(INSN, PREC, FN) \ + static bool trans_##INSN##_##PREC(DisasContext *s, \ + arg_##INSN##_##PREC *a) \ + { \ + return do_vfp_2op_##PREC(s, FN, a->vd, a->vm); \ + } -static bool trans_VMOV_reg_dp(DisasContext *s, arg_VMOV_reg_dp *a) -{ - return do_vfp_2op_dp(s, tcg_gen_mov_i64, a->vd, a->vm); -} +DO_VFP_2OP(VMOV_reg, sp, tcg_gen_mov_i32) +DO_VFP_2OP(VMOV_reg, dp, tcg_gen_mov_i64) -static bool trans_VABS_sp(DisasContext *s, arg_VABS_sp *a) -{ - return do_vfp_2op_sp(s, gen_helper_vfp_abss, a->vd, a->vm); -} +DO_VFP_2OP(VABS, sp, gen_helper_vfp_abss) +DO_VFP_2OP(VABS, dp, gen_helper_vfp_absd) -static bool trans_VABS_dp(DisasContext *s, arg_VABS_dp *a) -{ - return do_vfp_2op_dp(s, gen_helper_vfp_absd, a->vd, a->vm); -} - -static bool trans_VNEG_sp(DisasContext *s, arg_VNEG_sp *a) -{ - return do_vfp_2op_sp(s, gen_helper_vfp_negs, a->vd, a->vm); -} - -static bool trans_VNEG_dp(DisasContext *s, arg_VNEG_dp *a) -{ - return do_vfp_2op_dp(s, gen_helper_vfp_negd, a->vd, a->vm); -} +DO_VFP_2OP(VNEG, sp, gen_helper_vfp_negs) +DO_VFP_2OP(VNEG, dp, gen_helper_vfp_negd) static void gen_VSQRT_sp(TCGv_i32 vd, TCGv_i32 vm) { gen_helper_vfp_sqrts(vd, vm, cpu_env); } -static bool trans_VSQRT_sp(DisasContext *s, arg_VSQRT_sp *a) -{ - return do_vfp_2op_sp(s, gen_VSQRT_sp, a->vd, a->vm); -} - static void gen_VSQRT_dp(TCGv_i64 vd, TCGv_i64 vm) { gen_helper_vfp_sqrtd(vd, vm, cpu_env); } -static bool trans_VSQRT_dp(DisasContext *s, arg_VSQRT_dp *a) -{ - return do_vfp_2op_dp(s, gen_VSQRT_dp, a->vd, a->vm); -} +DO_VFP_2OP(VSQRT, sp, gen_VSQRT_sp) +DO_VFP_2OP(VSQRT, dp, gen_VSQRT_dp) static bool trans_VCMP_sp(DisasContext *s, arg_VCMP_sp *a) { -- 2.20.1 Implement VFP fp16 for VABS, VNEG and VSQRT. This is all the fp16 insns that use the DO_VFP_2OP macro, because there is no fp16 version of VMOV_reg. Notes: * the gen_helper_vfp_negh already exists as we needed to create it for the fp16 multiply-add insns * as usual we need to use the f16 version of the fp_status; this is only relevant for VSQRT Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-9-peter.maydell@linaro.org --- target/arm/helper.h | 2 ++ target/arm/vfp.decode | 3 +++ target/arm/vfp_helper.c | 10 +++++++++ target/arm/translate-vfp.c.inc | 40 ++++++++++++++++++++++++++++++++++ 4 files changed, 55 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_3(vfp_minnumd, f64, f64, f64, ptr) DEF_HELPER_1(vfp_negh, f16, f16) DEF_HELPER_1(vfp_negs, f32, f32) DEF_HELPER_1(vfp_negd, f64, f64) +DEF_HELPER_1(vfp_absh, f16, f16) DEF_HELPER_1(vfp_abss, f32, f32) DEF_HELPER_1(vfp_absd, f64, f64) +DEF_HELPER_2(vfp_sqrth, f16, f16, env) DEF_HELPER_2(vfp_sqrts, f32, f32, env) DEF_HELPER_2(vfp_sqrtd, f64, f64, env) DEF_HELPER_3(vfp_cmps, void, f32, f32, env) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VMOV_imm_dp ---- 1110 1.11 .... .... 1011 0000 .... \ VMOV_reg_sp ---- 1110 1.11 0000 .... 1010 01.0 .... @vfp_dm_ss VMOV_reg_dp ---- 1110 1.11 0000 .... 1011 01.0 .... @vfp_dm_dd +VABS_hp ---- 1110 1.11 0000 .... 1001 11.0 .... @vfp_dm_ss VABS_sp ---- 1110 1.11 0000 .... 1010 11.0 .... @vfp_dm_ss VABS_dp ---- 1110 1.11 0000 .... 1011 11.0 .... @vfp_dm_dd +VNEG_hp ---- 1110 1.11 0001 .... 1001 01.0 .... @vfp_dm_ss VNEG_sp ---- 1110 1.11 0001 .... 1010 01.0 .... @vfp_dm_ss VNEG_dp ---- 1110 1.11 0001 .... 1011 01.0 .... @vfp_dm_dd +VSQRT_hp ---- 1110 1.11 0001 .... 1001 11.0 .... @vfp_dm_ss VSQRT_sp ---- 1110 1.11 0001 .... 1010 11.0 .... @vfp_dm_ss VSQRT_dp ---- 1110 1.11 0001 .... 1011 11.0 .... @vfp_dm_dd diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ float64 VFP_HELPER(neg, d)(float64 a) return float64_chs(a); } +dh_ctype_f16 VFP_HELPER(abs, h)(dh_ctype_f16 a) +{ + return float16_abs(a); +} + float32 VFP_HELPER(abs, s)(float32 a) { return float32_abs(a); @@ -XXX,XX +XXX,XX @@ float64 VFP_HELPER(abs, d)(float64 a) return float64_abs(a); } +dh_ctype_f16 VFP_HELPER(sqrt, h)(dh_ctype_f16 a, CPUARMState *env) +{ + return float16_sqrt(a, &env->vfp.fp_status_f16); +} + float32 VFP_HELPER(sqrt, s)(float32 a, CPUARMState *env) { return float32_sqrt(a, &env->vfp.fp_status); diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool do_vfp_2op_sp(DisasContext *s, VFPGen2OpSPFn *fn, int vd, int vm) return true; } +static bool do_vfp_2op_hp(DisasContext *s, VFPGen2OpSPFn *fn, int vd, int vm) +{ + /* + * Do a half-precision operation. Functionally this is + * the same as do_vfp_2op_sp(), except: + * - it doesn't need the VFP vector handling (fp16 is a + * v8 feature, and in v8 VFP vectors don't exist) + * - it does the aa32_fp16_arith feature test + */ + TCGv_i32 f0; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (s->vec_len != 0 || s->vec_stride != 0) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + f0 = tcg_temp_new_i32(); + neon_load_reg32(f0, vm); + fn(f0, f0); + neon_store_reg32(f0, vd); + tcg_temp_free_i32(f0); + + return true; +} + static bool do_vfp_2op_dp(DisasContext *s, VFPGen2OpDPFn *fn, int vd, int vm) { uint32_t delta_m = 0; @@ -XXX,XX +XXX,XX @@ static bool trans_VMOV_imm_dp(DisasContext *s, arg_VMOV_imm_dp *a) DO_VFP_2OP(VMOV_reg, sp, tcg_gen_mov_i32) DO_VFP_2OP(VMOV_reg, dp, tcg_gen_mov_i64) +DO_VFP_2OP(VABS, hp, gen_helper_vfp_absh) DO_VFP_2OP(VABS, sp, gen_helper_vfp_abss) DO_VFP_2OP(VABS, dp, gen_helper_vfp_absd) +DO_VFP_2OP(VNEG, hp, gen_helper_vfp_negh) DO_VFP_2OP(VNEG, sp, gen_helper_vfp_negs) DO_VFP_2OP(VNEG, dp, gen_helper_vfp_negd) +static void gen_VSQRT_hp(TCGv_i32 vd, TCGv_i32 vm) +{ + gen_helper_vfp_sqrth(vd, vm, cpu_env); +} + static void gen_VSQRT_sp(TCGv_i32 vd, TCGv_i32 vm) { gen_helper_vfp_sqrts(vd, vm, cpu_env); @@ -XXX,XX +XXX,XX @@ static void gen_VSQRT_dp(TCGv_i64 vd, TCGv_i64 vm) gen_helper_vfp_sqrtd(vd, vm, cpu_env); } +DO_VFP_2OP(VSQRT, hp, gen_VSQRT_hp) DO_VFP_2OP(VSQRT, sp, gen_VSQRT_sp) DO_VFP_2OP(VSQRT, dp, gen_VSQRT_dp) -- 2.20.1 Implement VFP fp16 support for the VMOV immediate insn. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-10-peter.maydell@linaro.org --- target/arm/vfp.decode | 2 ++ target/arm/translate-vfp.c.inc | 22 ++++++++++++++++++++++ 2 files changed, 24 insertions(+) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VFMS_dp ---- 1110 1.10 .... .... 1011 .1.0 .... @vfp_dnm_d VFNMA_dp ---- 1110 1.01 .... .... 1011 .0.0 .... @vfp_dnm_d VFNMS_dp ---- 1110 1.01 .... .... 1011 .1.0 .... @vfp_dnm_d +VMOV_imm_hp ---- 1110 1.11 .... .... 1001 0000 .... \ + vd=%vd_sp imm=%vmov_imm VMOV_imm_sp ---- 1110 1.11 .... .... 1010 0000 .... \ vd=%vd_sp imm=%vmov_imm VMOV_imm_dp ---- 1110 1.11 .... .... 1011 0000 .... \ diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ MAKE_VFM_TRANS_FNS(hp) MAKE_VFM_TRANS_FNS(sp) MAKE_VFM_TRANS_FNS(dp) +static bool trans_VMOV_imm_hp(DisasContext *s, arg_VMOV_imm_sp *a) +{ + TCGv_i32 fd; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (s->vec_len != 0 || s->vec_stride != 0) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + fd = tcg_const_i32(vfp_expand_imm(MO_16, a->imm)); + neon_store_reg32(fd, a->vd); + tcg_temp_free_i32(fd); + return true; +} + static bool trans_VMOV_imm_sp(DisasContext *s, arg_VMOV_imm_sp *a) { uint32_t delta_d = 0; -- 2.20.1 Implement fp16 version of VCMP. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-11-peter.maydell@linaro.org --- target/arm/helper.h | 2 ++ target/arm/vfp.decode | 2 ++ target/arm/vfp_helper.c | 15 +++++++------ target/arm/translate-vfp.c.inc | 39 ++++++++++++++++++++++++++++++++++ 4 files changed, 51 insertions(+), 7 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_1(vfp_absd, f64, f64) DEF_HELPER_2(vfp_sqrth, f16, f16, env) DEF_HELPER_2(vfp_sqrts, f32, f32, env) DEF_HELPER_2(vfp_sqrtd, f64, f64, env) +DEF_HELPER_3(vfp_cmph, void, f16, f16, env) DEF_HELPER_3(vfp_cmps, void, f32, f32, env) DEF_HELPER_3(vfp_cmpd, void, f64, f64, env) +DEF_HELPER_3(vfp_cmpeh, void, f16, f16, env) DEF_HELPER_3(vfp_cmpes, void, f32, f32, env) DEF_HELPER_3(vfp_cmped, void, f64, f64, env) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VSQRT_hp ---- 1110 1.11 0001 .... 1001 11.0 .... @vfp_dm_ss VSQRT_sp ---- 1110 1.11 0001 .... 1010 11.0 .... @vfp_dm_ss VSQRT_dp ---- 1110 1.11 0001 .... 1011 11.0 .... @vfp_dm_dd +VCMP_hp ---- 1110 1.11 010 z:1 .... 1001 e:1 1.0 .... \ + vd=%vd_sp vm=%vm_sp VCMP_sp ---- 1110 1.11 010 z:1 .... 1010 e:1 1.0 .... \ vd=%vd_sp vm=%vm_sp VCMP_dp ---- 1110 1.11 010 z:1 .... 1011 e:1 1.0 .... \ diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ static void softfloat_to_vfp_compare(CPUARMState *env, FloatRelation cmp) } /* XXX: check quiet/signaling case */ -#define DO_VFP_cmp(p, type) \ -void VFP_HELPER(cmp, p)(type a, type b, CPUARMState *env) \ +#define DO_VFP_cmp(P, FLOATTYPE, ARGTYPE, FPST) \ +void VFP_HELPER(cmp, P)(ARGTYPE a, ARGTYPE b, CPUARMState *env) \ { \ softfloat_to_vfp_compare(env, \ - type ## _compare_quiet(a, b, &env->vfp.fp_status)); \ + FLOATTYPE ## _compare_quiet(a, b, &env->vfp.FPST)); \ } \ -void VFP_HELPER(cmpe, p)(type a, type b, CPUARMState *env) \ +void VFP_HELPER(cmpe, P)(ARGTYPE a, ARGTYPE b, CPUARMState *env) \ { \ softfloat_to_vfp_compare(env, \ - type ## _compare(a, b, &env->vfp.fp_status)); \ + FLOATTYPE ## _compare(a, b, &env->vfp.FPST)); \ } -DO_VFP_cmp(s, float32) -DO_VFP_cmp(d, float64) +DO_VFP_cmp(h, float16, dh_ctype_f16, fp_status_f16) +DO_VFP_cmp(s, float32, float32, fp_status) +DO_VFP_cmp(d, float64, float64, fp_status) #undef DO_VFP_cmp /* Integer to float and float to integer conversions */ diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ DO_VFP_2OP(VSQRT, hp, gen_VSQRT_hp) DO_VFP_2OP(VSQRT, sp, gen_VSQRT_sp) DO_VFP_2OP(VSQRT, dp, gen_VSQRT_dp) +static bool trans_VCMP_hp(DisasContext *s, arg_VCMP_sp *a) +{ + TCGv_i32 vd, vm; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + /* Vm/M bits must be zero for the Z variant */ + if (a->z && a->vm != 0) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + vd = tcg_temp_new_i32(); + vm = tcg_temp_new_i32(); + + neon_load_reg32(vd, a->vd); + if (a->z) { + tcg_gen_movi_i32(vm, 0); + } else { + neon_load_reg32(vm, a->vm); + } + + if (a->e) { + gen_helper_vfp_cmpeh(vd, vm, cpu_env); + } else { + gen_helper_vfp_cmph(vd, vm, cpu_env); + } + + tcg_temp_free_i32(vd); + tcg_temp_free_i32(vm); + + return true; +} + static bool trans_VCMP_sp(DisasContext *s, arg_VCMP_sp *a) { TCGv_i32 vd, vm; -- 2.20.1 Implement the fp16 versions of the VFP VLDR/VSTR (immediate). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-12-peter.maydell@linaro.org --- target/arm/vfp.decode | 3 +-- target/arm/translate-vfp.c.inc | 35 ++++++++++++++++++++++++++++++++++ 2 files changed, 36 insertions(+), 2 deletions(-) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VMOV_single ---- 1110 000 l:1 .... rt:4 1010 . 001 0000 vn=%vn_sp VMOV_64_sp ---- 1100 010 op:1 rt2:4 rt:4 1010 00.1 .... vm=%vm_sp VMOV_64_dp ---- 1100 010 op:1 rt2:4 rt:4 1011 00.1 .... vm=%vm_dp -# Note that the half-precision variants of VLDR and VSTR are -# not part of this decodetree at all because they have bits [9:8] == 0b01 +VLDR_VSTR_hp ---- 1101 u:1 .0 l:1 rn:4 .... 1001 imm:8 vd=%vd_sp VLDR_VSTR_sp ---- 1101 u:1 .0 l:1 rn:4 .... 1010 imm:8 vd=%vd_sp VLDR_VSTR_dp ---- 1101 u:1 .0 l:1 rn:4 .... 1011 imm:8 vd=%vd_dp diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VMOV_64_dp(DisasContext *s, arg_VMOV_64_dp *a) return true; } +static bool trans_VLDR_VSTR_hp(DisasContext *s, arg_VLDR_VSTR_sp *a) +{ + uint32_t offset; + TCGv_i32 addr, tmp; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + /* imm8 field is offset/2 for fp16, unlike fp32 and fp64 */ + offset = a->imm << 1; + if (!a->u) { + offset = -offset; + } + + /* For thumb, use of PC is UNPREDICTABLE. */ + addr = add_reg_for_lit(s, a->rn, offset); + tmp = tcg_temp_new_i32(); + if (a->l) { + gen_aa32_ld16u(s, tmp, addr, get_mem_index(s)); + neon_store_reg32(tmp, a->vd); + } else { + neon_load_reg32(tmp, a->vd); + gen_aa32_st16(s, tmp, addr, get_mem_index(s)); + } + tcg_temp_free_i32(tmp); + tcg_temp_free_i32(addr); + + return true; +} + static bool trans_VLDR_VSTR_sp(DisasContext *s, arg_VLDR_VSTR_sp *a) { uint32_t offset; -- 2.20.1 Implement the fp16 versions of the VFP VCVT instruction forms which convert between floating point and integer. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-13-peter.maydell@linaro.org --- target/arm/vfp.decode | 4 +++ target/arm/translate-vfp.c.inc | 65 ++++++++++++++++++++++++++++++++++ 2 files changed, 69 insertions(+) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VCVT_sp ---- 1110 1.11 0111 .... 1010 11.0 .... @vfp_dm_ds VCVT_dp ---- 1110 1.11 0111 .... 1011 11.0 .... @vfp_dm_sd # VCVT from integer to floating point: Vm always single; Vd depends on size +VCVT_int_hp ---- 1110 1.11 1000 .... 1001 s:1 1.0 .... \ + vd=%vd_sp vm=%vm_sp VCVT_int_sp ---- 1110 1.11 1000 .... 1010 s:1 1.0 .... \ vd=%vd_sp vm=%vm_sp VCVT_int_dp ---- 1110 1.11 1000 .... 1011 s:1 1.0 .... \ @@ -XXX,XX +XXX,XX @@ VCVT_fix_dp ---- 1110 1.11 1.1. .... 1011 .1.0 .... \ vd=%vd_dp imm=%vm_sp opc=%vcvt_fix_op # VCVT float to integer (VCVT and VCVTR): Vd always single; Vd depends on size +VCVT_hp_int ---- 1110 1.11 110 s:1 .... 1001 rz:1 1.0 .... \ + vd=%vd_sp vm=%vm_sp VCVT_sp_int ---- 1110 1.11 110 s:1 .... 1010 rz:1 1.0 .... \ vd=%vd_sp vm=%vm_sp VCVT_dp_int ---- 1110 1.11 110 s:1 .... 1011 rz:1 1.0 .... \ diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VCVT_dp(DisasContext *s, arg_VCVT_dp *a) return true; } +static bool trans_VCVT_int_hp(DisasContext *s, arg_VCVT_int_sp *a) +{ + TCGv_i32 vm; + TCGv_ptr fpst; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + vm = tcg_temp_new_i32(); + neon_load_reg32(vm, a->vm); + fpst = fpstatus_ptr(FPST_FPCR_F16); + if (a->s) { + /* i32 -> f16 */ + gen_helper_vfp_sitoh(vm, vm, fpst); + } else { + /* u32 -> f16 */ + gen_helper_vfp_uitoh(vm, vm, fpst); + } + neon_store_reg32(vm, a->vd); + tcg_temp_free_i32(vm); + tcg_temp_free_ptr(fpst); + return true; +} + static bool trans_VCVT_int_sp(DisasContext *s, arg_VCVT_int_sp *a) { TCGv_i32 vm; @@ -XXX,XX +XXX,XX @@ static bool trans_VCVT_fix_dp(DisasContext *s, arg_VCVT_fix_dp *a) return true; } +static bool trans_VCVT_hp_int(DisasContext *s, arg_VCVT_sp_int *a) +{ + TCGv_i32 vm; + TCGv_ptr fpst; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + fpst = fpstatus_ptr(FPST_FPCR_F16); + vm = tcg_temp_new_i32(); + neon_load_reg32(vm, a->vm); + + if (a->s) { + if (a->rz) { + gen_helper_vfp_tosizh(vm, vm, fpst); + } else { + gen_helper_vfp_tosih(vm, vm, fpst); + } + } else { + if (a->rz) { + gen_helper_vfp_touizh(vm, vm, fpst); + } else { + gen_helper_vfp_touih(vm, vm, fpst); + } + } + neon_store_reg32(vm, a->vd); + tcg_temp_free_i32(vm); + tcg_temp_free_ptr(fpst); + return true; +} + static bool trans_VCVT_sp_int(DisasContext *s, arg_VCVT_sp_int *a) { TCGv_i32 vm; -- 2.20.1 Currently the VFP_CONV_FIX macros take a single fsz argument for the size of the float type, which is used both to select the name of the functions to call (eg float32_is_any_nan()) and also for the type to use for the float inputs and outputs (eg float32). Separate these into fsz and ftype arguments, so that we can use them for fp16, which uses 'float16' in the function names but is still passing inputs and outputs in a 32-bit sized type. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-14-peter.maydell@linaro.org --- target/arm/vfp_helper.c | 46 ++++++++++++++++++++--------------------- 1 file changed, 23 insertions(+), 23 deletions(-) diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ float32 VFP_HELPER(fcvts, d)(float64 x, CPUARMState *env) } /* VFP3 fixed point conversion. */ -#define VFP_CONV_FIX_FLOAT(name, p, fsz, isz, itype) \ -float##fsz HELPER(vfp_##name##to##p)(uint##isz##_t x, uint32_t shift, \ +#define VFP_CONV_FIX_FLOAT(name, p, fsz, ftype, isz, itype) \ +ftype HELPER(vfp_##name##to##p)(uint##isz##_t x, uint32_t shift, \ void *fpstp) \ { return itype##_to_##float##fsz##_scalbn(x, -shift, fpstp); } -#define VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, ROUND, suff) \ -uint##isz##_t HELPER(vfp_to##name##p##suff)(float##fsz x, uint32_t shift, \ +#define VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, ftype, isz, itype, ROUND, suff) \ +uint##isz##_t HELPER(vfp_to##name##p##suff)(ftype x, uint32_t shift, \ void *fpst) \ { \ if (unlikely(float##fsz##_is_any_nan(x))) { \ @@ -XXX,XX +XXX,XX @@ uint##isz##_t HELPER(vfp_to##name##p##suff)(float##fsz x, uint32_t shift, \ return float##fsz##_to_##itype##_scalbn(x, ROUND, shift, fpst); \ } -#define VFP_CONV_FIX(name, p, fsz, isz, itype) \ -VFP_CONV_FIX_FLOAT(name, p, fsz, isz, itype) \ -VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, \ +#define VFP_CONV_FIX(name, p, fsz, ftype, isz, itype) \ +VFP_CONV_FIX_FLOAT(name, p, fsz, ftype, isz, itype) \ +VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, ftype, isz, itype, \ float_round_to_zero, _round_to_zero) \ -VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, \ +VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, ftype, isz, itype, \ get_float_rounding_mode(fpst), ) -#define VFP_CONV_FIX_A64(name, p, fsz, isz, itype) \ -VFP_CONV_FIX_FLOAT(name, p, fsz, isz, itype) \ -VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, isz, itype, \ +#define VFP_CONV_FIX_A64(name, p, fsz, ftype, isz, itype) \ +VFP_CONV_FIX_FLOAT(name, p, fsz, ftype, isz, itype) \ +VFP_CONV_FLOAT_FIX_ROUND(name, p, fsz, ftype, isz, itype, \ get_float_rounding_mode(fpst), ) -VFP_CONV_FIX(sh, d, 64, 64, int16) -VFP_CONV_FIX(sl, d, 64, 64, int32) -VFP_CONV_FIX_A64(sq, d, 64, 64, int64) -VFP_CONV_FIX(uh, d, 64, 64, uint16) -VFP_CONV_FIX(ul, d, 64, 64, uint32) -VFP_CONV_FIX_A64(uq, d, 64, 64, uint64) -VFP_CONV_FIX(sh, s, 32, 32, int16) -VFP_CONV_FIX(sl, s, 32, 32, int32) -VFP_CONV_FIX_A64(sq, s, 32, 64, int64) -VFP_CONV_FIX(uh, s, 32, 32, uint16) -VFP_CONV_FIX(ul, s, 32, 32, uint32) -VFP_CONV_FIX_A64(uq, s, 32, 64, uint64) +VFP_CONV_FIX(sh, d, 64, float64, 64, int16) +VFP_CONV_FIX(sl, d, 64, float64, 64, int32) +VFP_CONV_FIX_A64(sq, d, 64, float64, 64, int64) +VFP_CONV_FIX(uh, d, 64, float64, 64, uint16) +VFP_CONV_FIX(ul, d, 64, float64, 64, uint32) +VFP_CONV_FIX_A64(uq, d, 64, float64, 64, uint64) +VFP_CONV_FIX(sh, s, 32, float32, 32, int16) +VFP_CONV_FIX(sl, s, 32, float32, 32, int32) +VFP_CONV_FIX_A64(sq, s, 32, float32, 64, int64) +VFP_CONV_FIX(uh, s, 32, float32, 32, uint16) +VFP_CONV_FIX(ul, s, 32, float32, 32, uint32) +VFP_CONV_FIX_A64(uq, s, 32, float32, 64, uint64) #undef VFP_CONV_FIX #undef VFP_CONV_FIX_FLOAT -- 2.20.1 Now the VFP_CONV_FIX macros can handle fp16's distinction between the width of the operation and the width of the type used to pass operands, use the macros rather than the open-coded functions. This creates an extra six helper functions, all of which we are going to need for the AArch32 VFP fp16 instructions. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-15-peter.maydell@linaro.org --- target/arm/helper.h | 6 +++ target/arm/vfp_helper.c | 86 +++-------------------------------------- 2 files changed, 12 insertions(+), 80 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(vfp_tosizh, s32, f16, ptr) DEF_HELPER_2(vfp_tosizs, s32, f32, ptr) DEF_HELPER_2(vfp_tosizd, s32, f64, ptr) +DEF_HELPER_3(vfp_toshh_round_to_zero, i32, f16, i32, ptr) +DEF_HELPER_3(vfp_toslh_round_to_zero, i32, f16, i32, ptr) +DEF_HELPER_3(vfp_touhh_round_to_zero, i32, f16, i32, ptr) +DEF_HELPER_3(vfp_toulh_round_to_zero, i32, f16, i32, ptr) DEF_HELPER_3(vfp_toshs_round_to_zero, i32, f32, i32, ptr) DEF_HELPER_3(vfp_tosls_round_to_zero, i32, f32, i32, ptr) DEF_HELPER_3(vfp_touhs_round_to_zero, i32, f32, i32, ptr) @@ -XXX,XX +XXX,XX @@ DEF_HELPER_3(vfp_sqtod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_uhtod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_ultod, f64, i64, i32, ptr) DEF_HELPER_3(vfp_uqtod, f64, i64, i32, ptr) +DEF_HELPER_3(vfp_shtoh, f16, i32, i32, ptr) +DEF_HELPER_3(vfp_uhtoh, f16, i32, i32, ptr) DEF_HELPER_3(vfp_sltoh, f16, i32, i32, ptr) DEF_HELPER_3(vfp_ultoh, f16, i32, i32, ptr) DEF_HELPER_3(vfp_sqtoh, f16, i64, i32, ptr) diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ VFP_CONV_FIX_A64(sq, s, 32, float32, 64, int64) VFP_CONV_FIX(uh, s, 32, float32, 32, uint16) VFP_CONV_FIX(ul, s, 32, float32, 32, uint32) VFP_CONV_FIX_A64(uq, s, 32, float32, 64, uint64) +VFP_CONV_FIX(sh, h, 16, dh_ctype_f16, 32, int16) +VFP_CONV_FIX(sl, h, 16, dh_ctype_f16, 32, int32) +VFP_CONV_FIX_A64(sq, h, 16, dh_ctype_f16, 64, int64) +VFP_CONV_FIX(uh, h, 16, dh_ctype_f16, 32, uint16) +VFP_CONV_FIX(ul, h, 16, dh_ctype_f16, 32, uint32) +VFP_CONV_FIX_A64(uq, h, 16, dh_ctype_f16, 64, uint64) #undef VFP_CONV_FIX #undef VFP_CONV_FIX_FLOAT #undef VFP_CONV_FLOAT_FIX_ROUND #undef VFP_CONV_FIX_A64 -uint32_t HELPER(vfp_sltoh)(uint32_t x, uint32_t shift, void *fpst) -{ - return int32_to_float16_scalbn(x, -shift, fpst); -} - -uint32_t HELPER(vfp_ultoh)(uint32_t x, uint32_t shift, void *fpst) -{ - return uint32_to_float16_scalbn(x, -shift, fpst); -} - -uint32_t HELPER(vfp_sqtoh)(uint64_t x, uint32_t shift, void *fpst) -{ - return int64_to_float16_scalbn(x, -shift, fpst); -} - -uint32_t HELPER(vfp_uqtoh)(uint64_t x, uint32_t shift, void *fpst) -{ - return uint64_to_float16_scalbn(x, -shift, fpst); -} - -uint32_t HELPER(vfp_toshh)(uint32_t x, uint32_t shift, void *fpst) -{ - if (unlikely(float16_is_any_nan(x))) { - float_raise(float_flag_invalid, fpst); - return 0; - } - return float16_to_int16_scalbn(x, get_float_rounding_mode(fpst), - shift, fpst); -} - -uint32_t HELPER(vfp_touhh)(uint32_t x, uint32_t shift, void *fpst) -{ - if (unlikely(float16_is_any_nan(x))) { - float_raise(float_flag_invalid, fpst); - return 0; - } - return float16_to_uint16_scalbn(x, get_float_rounding_mode(fpst), - shift, fpst); -} - -uint32_t HELPER(vfp_toslh)(uint32_t x, uint32_t shift, void *fpst) -{ - if (unlikely(float16_is_any_nan(x))) { - float_raise(float_flag_invalid, fpst); - return 0; - } - return float16_to_int32_scalbn(x, get_float_rounding_mode(fpst), - shift, fpst); -} - -uint32_t HELPER(vfp_toulh)(uint32_t x, uint32_t shift, void *fpst) -{ - if (unlikely(float16_is_any_nan(x))) { - float_raise(float_flag_invalid, fpst); - return 0; - } - return float16_to_uint32_scalbn(x, get_float_rounding_mode(fpst), - shift, fpst); -} - -uint64_t HELPER(vfp_tosqh)(uint32_t x, uint32_t shift, void *fpst) -{ - if (unlikely(float16_is_any_nan(x))) { - float_raise(float_flag_invalid, fpst); - return 0; - } - return float16_to_int64_scalbn(x, get_float_rounding_mode(fpst), - shift, fpst); -} - -uint64_t HELPER(vfp_touqh)(uint32_t x, uint32_t shift, void *fpst) -{ - if (unlikely(float16_is_any_nan(x))) { - float_raise(float_flag_invalid, fpst); - return 0; - } - return float16_to_uint64_scalbn(x, get_float_rounding_mode(fpst), - shift, fpst); -} - /* Set the current fp rounding mode and return the old one. * The argument is a softfloat float_round_ value. */ -- 2.20.1 Implement the fp16 versions of the VFP VCVT instruction forms which convert between floating point and fixed-point. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-16-peter.maydell@linaro.org --- target/arm/vfp.decode | 2 ++ target/arm/translate-vfp.c.inc | 59 ++++++++++++++++++++++++++++++++++ 2 files changed, 61 insertions(+) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VJCVT ---- 1110 1.11 1001 .... 1011 11.0 .... @vfp_dm_sd # We assemble bits 18 (op), 16 (u) and 7 (sx) into a single opc field # for the convenience of the trans_VCVT_fix functions. %vcvt_fix_op 18:1 16:1 7:1 +VCVT_fix_hp ---- 1110 1.11 1.1. .... 1001 .1.0 .... \ + vd=%vd_sp imm=%vm_sp opc=%vcvt_fix_op VCVT_fix_sp ---- 1110 1.11 1.1. .... 1010 .1.0 .... \ vd=%vd_sp imm=%vm_sp opc=%vcvt_fix_op VCVT_fix_dp ---- 1110 1.11 1.1. .... 1011 .1.0 .... \ diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VJCVT(DisasContext *s, arg_VJCVT *a) return true; } +static bool trans_VCVT_fix_hp(DisasContext *s, arg_VCVT_fix_sp *a) +{ + TCGv_i32 vd, shift; + TCGv_ptr fpst; + int frac_bits; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + frac_bits = (a->opc & 1) ? (32 - a->imm) : (16 - a->imm); + + vd = tcg_temp_new_i32(); + neon_load_reg32(vd, a->vd); + + fpst = fpstatus_ptr(FPST_FPCR_F16); + shift = tcg_const_i32(frac_bits); + + /* Switch on op:U:sx bits */ + switch (a->opc) { + case 0: + gen_helper_vfp_shtoh(vd, vd, shift, fpst); + break; + case 1: + gen_helper_vfp_sltoh(vd, vd, shift, fpst); + break; + case 2: + gen_helper_vfp_uhtoh(vd, vd, shift, fpst); + break; + case 3: + gen_helper_vfp_ultoh(vd, vd, shift, fpst); + break; + case 4: + gen_helper_vfp_toshh_round_to_zero(vd, vd, shift, fpst); + break; + case 5: + gen_helper_vfp_toslh_round_to_zero(vd, vd, shift, fpst); + break; + case 6: + gen_helper_vfp_touhh_round_to_zero(vd, vd, shift, fpst); + break; + case 7: + gen_helper_vfp_toulh_round_to_zero(vd, vd, shift, fpst); + break; + default: + g_assert_not_reached(); + } + + neon_store_reg32(vd, a->vd); + tcg_temp_free_i32(vd); + tcg_temp_free_i32(shift); + tcg_temp_free_ptr(fpst); + return true; +} + static bool trans_VCVT_fix_sp(DisasContext *s, arg_VCVT_fix_sp *a) { TCGv_i32 vd, shift; -- 2.20.1 Implement the fp16 versions of the VFP VCVT instruction forms which convert between floating point and integer with a specified rounding mode. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-17-peter.maydell@linaro.org --- target/arm/vfp-uncond.decode | 6 ++++-- target/arm/translate-vfp.c.inc | 32 ++++++++++++++++++++++++-------- 2 files changed, 28 insertions(+), 10 deletions(-) diff --git a/target/arm/vfp-uncond.decode b/target/arm/vfp-uncond.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp-uncond.decode +++ b/target/arm/vfp-uncond.decode @@ -XXX,XX +XXX,XX @@ VRINT 1111 1110 1.11 10 rm:2 .... 1011 01.0 .... \ vm=%vm_dp vd=%vd_dp dp=1 # VCVT float to int with specified rounding mode; Vd is always single-precision +VCVT 1111 1110 1.11 11 rm:2 .... 1001 op:1 1.0 .... \ + vm=%vm_sp vd=%vd_sp sz=1 VCVT 1111 1110 1.11 11 rm:2 .... 1010 op:1 1.0 .... \ - vm=%vm_sp vd=%vd_sp dp=0 + vm=%vm_sp vd=%vd_sp sz=2 VCVT 1111 1110 1.11 11 rm:2 .... 1011 op:1 1.0 .... \ - vm=%vm_dp vd=%vd_sp dp=1 + vm=%vm_dp vd=%vd_sp sz=3 diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VRINT(DisasContext *s, arg_VRINT *a) static bool trans_VCVT(DisasContext *s, arg_VCVT *a) { uint32_t rd, rm; - bool dp = a->dp; + int sz = a->sz; TCGv_ptr fpst; TCGv_i32 tcg_rmode, tcg_shift; int rounding = fp_decode_rm[a->rm]; @@ -XXX,XX +XXX,XX @@ static bool trans_VCVT(DisasContext *s, arg_VCVT *a) return false; } - if (dp && !dc_isar_feature(aa32_fpdp_v2, s)) { + if (sz == 3 && !dc_isar_feature(aa32_fpdp_v2, s)) { + return false; + } + + if (sz == 1 && !dc_isar_feature(aa32_fp16_arith, s)) { return false; } /* UNDEF accesses to D16-D31 if they don't exist */ - if (dp && !dc_isar_feature(aa32_simd_r32, s) && (a->vm & 0x10)) { + if (sz == 3 && !dc_isar_feature(aa32_simd_r32, s) && (a->vm & 0x10)) { return false; } @@ -XXX,XX +XXX,XX @@ static bool trans_VCVT(DisasContext *s, arg_VCVT *a) return true; } - fpst = fpstatus_ptr(FPST_FPCR); + if (sz == 1) { + fpst = fpstatus_ptr(FPST_FPCR_F16); + } else { + fpst = fpstatus_ptr(FPST_FPCR); + } tcg_shift = tcg_const_i32(0); tcg_rmode = tcg_const_i32(arm_rmode_to_sf(rounding)); gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst); - if (dp) { + if (sz == 3) { TCGv_i64 tcg_double, tcg_res; TCGv_i32 tcg_tmp; tcg_double = tcg_temp_new_i64(); @@ -XXX,XX +XXX,XX @@ static bool trans_VCVT(DisasContext *s, arg_VCVT *a) tcg_single = tcg_temp_new_i32(); tcg_res = tcg_temp_new_i32(); neon_load_reg32(tcg_single, rm); - if (is_signed) { - gen_helper_vfp_tosls(tcg_res, tcg_single, tcg_shift, fpst); + if (sz == 1) { + if (is_signed) { + gen_helper_vfp_toslh(tcg_res, tcg_single, tcg_shift, fpst); + } else { + gen_helper_vfp_toulh(tcg_res, tcg_single, tcg_shift, fpst); + } } else { - gen_helper_vfp_touls(tcg_res, tcg_single, tcg_shift, fpst); + if (is_signed) { + gen_helper_vfp_tosls(tcg_res, tcg_single, tcg_shift, fpst); + } else { + gen_helper_vfp_touls(tcg_res, tcg_single, tcg_shift, fpst); + } } neon_store_reg32(tcg_res, rd); tcg_temp_free_i32(tcg_res); -- 2.20.1 Implement the fp16 versions of the VFP VSEL instruction. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-18-peter.maydell@linaro.org --- target/arm/vfp-uncond.decode | 6 ++++-- target/arm/translate-vfp.c.inc | 16 ++++++++++++---- 2 files changed, 16 insertions(+), 6 deletions(-) diff --git a/target/arm/vfp-uncond.decode b/target/arm/vfp-uncond.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp-uncond.decode +++ b/target/arm/vfp-uncond.decode @@ -XXX,XX +XXX,XX @@ @vfp_dnm_s ................................ vm=%vm_sp vn=%vn_sp vd=%vd_sp @vfp_dnm_d ................................ vm=%vm_dp vn=%vn_dp vd=%vd_dp +VSEL 1111 1110 0. cc:2 .... .... 1001 .0.0 .... \ + vm=%vm_sp vn=%vn_sp vd=%vd_sp sz=1 VSEL 1111 1110 0. cc:2 .... .... 1010 .0.0 .... \ - vm=%vm_sp vn=%vn_sp vd=%vd_sp dp=0 + vm=%vm_sp vn=%vn_sp vd=%vd_sp sz=2 VSEL 1111 1110 0. cc:2 .... .... 1011 .0.0 .... \ - vm=%vm_dp vn=%vn_dp vd=%vd_dp dp=1 + vm=%vm_dp vn=%vn_dp vd=%vd_dp sz=3 VMAXNM_hp 1111 1110 1.00 .... .... 1001 .0.0 .... @vfp_dnm_s VMINNM_hp 1111 1110 1.00 .... .... 1001 .1.0 .... @vfp_dnm_s diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool vfp_access_check(DisasContext *s) static bool trans_VSEL(DisasContext *s, arg_VSEL *a) { uint32_t rd, rn, rm; - bool dp = a->dp; + int sz = a->sz; if (!dc_isar_feature(aa32_vsel, s)) { return false; } - if (dp && !dc_isar_feature(aa32_fpdp_v2, s)) { + if (sz == 3 && !dc_isar_feature(aa32_fpdp_v2, s)) { + return false; + } + + if (sz == 1 && !dc_isar_feature(aa32_fp16_arith, s)) { return false; } /* UNDEF accesses to D16-D31 if they don't exist */ - if (dp && !dc_isar_feature(aa32_simd_r32, s) && + if (sz == 3 && !dc_isar_feature(aa32_simd_r32, s) && ((a->vm | a->vn | a->vd) & 0x10)) { return false; } @@ -XXX,XX +XXX,XX @@ static bool trans_VSEL(DisasContext *s, arg_VSEL *a) return true; } - if (dp) { + if (sz == 3) { TCGv_i64 frn, frm, dest; TCGv_i64 tmp, zero, zf, nf, vf; @@ -XXX,XX +XXX,XX @@ static bool trans_VSEL(DisasContext *s, arg_VSEL *a) tcg_temp_free_i32(tmp); break; } + /* For fp16 the top half is always zeroes */ + if (sz == 1) { + tcg_gen_andi_i32(dest, dest, 0xffff); + } neon_store_reg32(dest, rd); tcg_temp_free_i32(frn); tcg_temp_free_i32(frm); -- 2.20.1 Implement the fp16 version of the VFP VRINT* insns. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-19-peter.maydell@linaro.org --- target/arm/helper.h | 2 + target/arm/vfp-uncond.decode | 6 ++- target/arm/vfp.decode | 3 ++ target/arm/vfp_helper.c | 21 ++++++++ target/arm/translate-vfp.c.inc | 98 +++++++++++++++++++++++++++++++--- 5 files changed, 122 insertions(+), 8 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_3(shr_cc, i32, env, i32, i32) DEF_HELPER_3(sar_cc, i32, env, i32, i32) DEF_HELPER_3(ror_cc, i32, env, i32, i32) +DEF_HELPER_FLAGS_2(rinth_exact, TCG_CALL_NO_RWG, f16, f16, ptr) DEF_HELPER_FLAGS_2(rints_exact, TCG_CALL_NO_RWG, f32, f32, ptr) DEF_HELPER_FLAGS_2(rintd_exact, TCG_CALL_NO_RWG, f64, f64, ptr) +DEF_HELPER_FLAGS_2(rinth, TCG_CALL_NO_RWG, f16, f16, ptr) DEF_HELPER_FLAGS_2(rints, TCG_CALL_NO_RWG, f32, f32, ptr) DEF_HELPER_FLAGS_2(rintd, TCG_CALL_NO_RWG, f64, f64, ptr) diff --git a/target/arm/vfp-uncond.decode b/target/arm/vfp-uncond.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp-uncond.decode +++ b/target/arm/vfp-uncond.decode @@ -XXX,XX +XXX,XX @@ VMINNM_sp 1111 1110 1.00 .... .... 1010 .1.0 .... @vfp_dnm_s VMAXNM_dp 1111 1110 1.00 .... .... 1011 .0.0 .... @vfp_dnm_d VMINNM_dp 1111 1110 1.00 .... .... 1011 .1.0 .... @vfp_dnm_d +VRINT 1111 1110 1.11 10 rm:2 .... 1001 01.0 .... \ + vm=%vm_sp vd=%vd_sp sz=1 VRINT 1111 1110 1.11 10 rm:2 .... 1010 01.0 .... \ - vm=%vm_sp vd=%vd_sp dp=0 + vm=%vm_sp vd=%vd_sp sz=2 VRINT 1111 1110 1.11 10 rm:2 .... 1011 01.0 .... \ - vm=%vm_dp vd=%vd_dp dp=1 + vm=%vm_dp vd=%vd_dp sz=3 # VCVT float to int with specified rounding mode; Vd is always single-precision VCVT 1111 1110 1.11 11 rm:2 .... 1001 op:1 1.0 .... \ diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VCVT_f16_f32 ---- 1110 1.11 0011 .... 1010 t:1 1.0 .... \ VCVT_f16_f64 ---- 1110 1.11 0011 .... 1011 t:1 1.0 .... \ vd=%vd_sp vm=%vm_dp +VRINTR_hp ---- 1110 1.11 0110 .... 1001 01.0 .... @vfp_dm_ss VRINTR_sp ---- 1110 1.11 0110 .... 1010 01.0 .... @vfp_dm_ss VRINTR_dp ---- 1110 1.11 0110 .... 1011 01.0 .... @vfp_dm_dd +VRINTZ_hp ---- 1110 1.11 0110 .... 1001 11.0 .... @vfp_dm_ss VRINTZ_sp ---- 1110 1.11 0110 .... 1010 11.0 .... @vfp_dm_ss VRINTZ_dp ---- 1110 1.11 0110 .... 1011 11.0 .... @vfp_dm_dd +VRINTX_hp ---- 1110 1.11 0111 .... 1001 01.0 .... @vfp_dm_ss VRINTX_sp ---- 1110 1.11 0111 .... 1010 01.0 .... @vfp_dm_ss VRINTX_dp ---- 1110 1.11 0111 .... 1011 01.0 .... @vfp_dm_dd diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ float64 VFP_HELPER(muladd, d)(float64 a, float64 b, float64 c, void *fpstp) } /* ARMv8 round to integral */ +dh_ctype_f16 HELPER(rinth_exact)(dh_ctype_f16 x, void *fp_status) +{ + return float16_round_to_int(x, fp_status); +} + float32 HELPER(rints_exact)(float32 x, void *fp_status) { return float32_round_to_int(x, fp_status); @@ -XXX,XX +XXX,XX @@ float64 HELPER(rintd_exact)(float64 x, void *fp_status) return float64_round_to_int(x, fp_status); } +dh_ctype_f16 HELPER(rinth)(dh_ctype_f16 x, void *fp_status) +{ + int old_flags = get_float_exception_flags(fp_status), new_flags; + float16 ret; + + ret = float16_round_to_int(x, fp_status); + + /* Suppress any inexact exceptions the conversion produced */ + if (!(old_flags & float_flag_inexact)) { + new_flags = get_float_exception_flags(fp_status); + set_float_exception_flags(new_flags & ~float_flag_inexact, fp_status); + } + + return ret; +} + float32 HELPER(rints)(float32 x, void *fp_status) { int old_flags = get_float_exception_flags(fp_status), new_flags; diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static const uint8_t fp_decode_rm[] = { static bool trans_VRINT(DisasContext *s, arg_VRINT *a) { uint32_t rd, rm; - bool dp = a->dp; + int sz = a->sz; TCGv_ptr fpst; TCGv_i32 tcg_rmode; int rounding = fp_decode_rm[a->rm]; @@ -XXX,XX +XXX,XX @@ static bool trans_VRINT(DisasContext *s, arg_VRINT *a) return false; } - if (dp && !dc_isar_feature(aa32_fpdp_v2, s)) { + if (sz == 3 && !dc_isar_feature(aa32_fpdp_v2, s)) { + return false; + } + + if (sz == 1 && !dc_isar_feature(aa32_fp16_arith, s)) { return false; } /* UNDEF accesses to D16-D31 if they don't exist */ - if (dp && !dc_isar_feature(aa32_simd_r32, s) && + if (sz == 3 && !dc_isar_feature(aa32_simd_r32, s) && ((a->vm | a->vd) & 0x10)) { return false; } @@ -XXX,XX +XXX,XX @@ static bool trans_VRINT(DisasContext *s, arg_VRINT *a) return true; } - fpst = fpstatus_ptr(FPST_FPCR); + if (sz == 1) { + fpst = fpstatus_ptr(FPST_FPCR_F16); + } else { + fpst = fpstatus_ptr(FPST_FPCR); + } tcg_rmode = tcg_const_i32(arm_rmode_to_sf(rounding)); gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst); - if (dp) { + if (sz == 3) { TCGv_i64 tcg_op; TCGv_i64 tcg_res; tcg_op = tcg_temp_new_i64(); @@ -XXX,XX +XXX,XX @@ static bool trans_VRINT(DisasContext *s, arg_VRINT *a) tcg_op = tcg_temp_new_i32(); tcg_res = tcg_temp_new_i32(); neon_load_reg32(tcg_op, rm); - gen_helper_rints(tcg_res, tcg_op, fpst); + if (sz == 1) { + gen_helper_rinth(tcg_res, tcg_op, fpst); + } else { + gen_helper_rints(tcg_res, tcg_op, fpst); + } neon_store_reg32(tcg_res, rd); tcg_temp_free_i32(tcg_op); tcg_temp_free_i32(tcg_res); @@ -XXX,XX +XXX,XX @@ static bool trans_VCVT_f16_f64(DisasContext *s, arg_VCVT_f16_f64 *a) return true; } +static bool trans_VRINTR_hp(DisasContext *s, arg_VRINTR_sp *a) +{ + TCGv_ptr fpst; + TCGv_i32 tmp; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + tmp = tcg_temp_new_i32(); + neon_load_reg32(tmp, a->vm); + fpst = fpstatus_ptr(FPST_FPCR_F16); + gen_helper_rinth(tmp, tmp, fpst); + neon_store_reg32(tmp, a->vd); + tcg_temp_free_ptr(fpst); + tcg_temp_free_i32(tmp); + return true; +} + static bool trans_VRINTR_sp(DisasContext *s, arg_VRINTR_sp *a) { TCGv_ptr fpst; @@ -XXX,XX +XXX,XX @@ static bool trans_VRINTR_dp(DisasContext *s, arg_VRINTR_dp *a) return true; } +static bool trans_VRINTZ_hp(DisasContext *s, arg_VRINTZ_sp *a) +{ + TCGv_ptr fpst; + TCGv_i32 tmp; + TCGv_i32 tcg_rmode; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + tmp = tcg_temp_new_i32(); + neon_load_reg32(tmp, a->vm); + fpst = fpstatus_ptr(FPST_FPCR_F16); + tcg_rmode = tcg_const_i32(float_round_to_zero); + gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst); + gen_helper_rinth(tmp, tmp, fpst); + gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst); + neon_store_reg32(tmp, a->vd); + tcg_temp_free_ptr(fpst); + tcg_temp_free_i32(tcg_rmode); + tcg_temp_free_i32(tmp); + return true; +} + static bool trans_VRINTZ_sp(DisasContext *s, arg_VRINTZ_sp *a) { TCGv_ptr fpst; @@ -XXX,XX +XXX,XX @@ static bool trans_VRINTZ_dp(DisasContext *s, arg_VRINTZ_dp *a) return true; } +static bool trans_VRINTX_hp(DisasContext *s, arg_VRINTX_sp *a) +{ + TCGv_ptr fpst; + TCGv_i32 tmp; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + tmp = tcg_temp_new_i32(); + neon_load_reg32(tmp, a->vm); + fpst = fpstatus_ptr(FPST_FPCR_F16); + gen_helper_rinth_exact(tmp, tmp, fpst); + neon_store_reg32(tmp, a->vd); + tcg_temp_free_ptr(fpst); + tcg_temp_free_i32(tmp); + return true; +} + static bool trans_VRINTX_sp(DisasContext *s, arg_VRINTX_sp *a) { TCGv_ptr fpst; -- 2.20.1 The fp16 extension includes a new instruction VINS, which copies the lower 16 bits of a 32-bit source VFP register into the upper 16 bits of the destination. Implement it. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-20-peter.maydell@linaro.org --- target/arm/vfp-uncond.decode | 3 +++ target/arm/translate-vfp.c.inc | 28 ++++++++++++++++++++++++++++ 2 files changed, 31 insertions(+) diff --git a/target/arm/vfp-uncond.decode b/target/arm/vfp-uncond.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp-uncond.decode +++ b/target/arm/vfp-uncond.decode @@ -XXX,XX +XXX,XX @@ VCVT 1111 1110 1.11 11 rm:2 .... 1010 op:1 1.0 .... \ vm=%vm_sp vd=%vd_sp sz=2 VCVT 1111 1110 1.11 11 rm:2 .... 1011 op:1 1.0 .... \ vm=%vm_dp vd=%vd_sp sz=3 + +VINS 1111 1110 1.11 0000 .... 1010 11 . 0 .... \ + vd=%vd_sp vm=%vm_sp diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_NOCP(DisasContext *s, arg_NOCP *a) return false; } + +static bool trans_VINS(DisasContext *s, arg_VINS *a) +{ + TCGv_i32 rd, rm; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (s->vec_len != 0 || s->vec_stride != 0) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + /* Insert low half of Vm into high half of Vd */ + rm = tcg_temp_new_i32(); + rd = tcg_temp_new_i32(); + neon_load_reg32(rm, a->vm); + neon_load_reg32(rd, a->vd); + tcg_gen_deposit_i32(rd, rd, rm, 16, 16); + neon_store_reg32(rd, a->vd); + tcg_temp_free_i32(rm); + tcg_temp_free_i32(rd); + return true; +} -- 2.20.1 The fp16 extension includes a new instruction VMOVX, which copies the upper 16 bits of a 32-bit source VFP register into the lower 16 bits of the destination and zeroes the high half of the destination. Implement it. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-21-peter.maydell@linaro.org --- target/arm/vfp-uncond.decode | 3 +++ target/arm/translate-vfp.c.inc | 25 +++++++++++++++++++++++++ 2 files changed, 28 insertions(+) diff --git a/target/arm/vfp-uncond.decode b/target/arm/vfp-uncond.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp-uncond.decode +++ b/target/arm/vfp-uncond.decode @@ -XXX,XX +XXX,XX @@ VCVT 1111 1110 1.11 11 rm:2 .... 1010 op:1 1.0 .... \ VCVT 1111 1110 1.11 11 rm:2 .... 1011 op:1 1.0 .... \ vm=%vm_dp vd=%vd_sp sz=3 +VMOVX 1111 1110 1.11 0000 .... 1010 01 . 0 .... \ + vd=%vd_sp vm=%vm_sp + VINS 1111 1110 1.11 0000 .... 1010 11 . 0 .... \ vd=%vd_sp vm=%vm_sp diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VINS(DisasContext *s, arg_VINS *a) tcg_temp_free_i32(rd); return true; } + +static bool trans_VMOVX(DisasContext *s, arg_VINS *a) +{ + TCGv_i32 rm; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (s->vec_len != 0 || s->vec_stride != 0) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + /* Set Vd to high half of Vm */ + rm = tcg_temp_new_i32(); + neon_load_reg32(rm, a->vm); + tcg_gen_shri_i32(rm, rm, 16); + neon_store_reg32(rm, a->vd); + tcg_temp_free_i32(rm); + return true; +} -- 2.20.1 Implement the VFP fp16 variant of VMOV that transfers a 16-bit value between a general purpose register and a VFP register. Note that Rt == 15 is UNPREDICTABLE; since this insn is v8 and later only we have no need to replicate the old "updates CPSR.NZCV" behaviour that the singleprec version of this insn does. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-22-peter.maydell@linaro.org --- target/arm/vfp.decode | 1 + target/arm/translate-vfp.c.inc | 34 ++++++++++++++++++++++++++++++++++ 2 files changed, 35 insertions(+) diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp.decode +++ b/target/arm/vfp.decode @@ -XXX,XX +XXX,XX @@ VDUP ---- 1110 1 b:1 q:1 0 .... rt:4 1011 . 0 e:1 1 0000 \ vn=%vn_dp VMSR_VMRS ---- 1110 111 l:1 reg:4 rt:4 1010 0001 0000 +VMOV_half ---- 1110 000 l:1 .... rt:4 1001 . 001 0000 vn=%vn_sp VMOV_single ---- 1110 000 l:1 .... rt:4 1010 . 001 0000 vn=%vn_sp VMOV_64_sp ---- 1100 010 op:1 rt2:4 rt:4 1010 00.1 .... vm=%vm_sp diff --git a/target/arm/translate-vfp.c.inc b/target/arm/translate-vfp.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c.inc +++ b/target/arm/translate-vfp.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VMSR_VMRS(DisasContext *s, arg_VMSR_VMRS *a) return true; } +static bool trans_VMOV_half(DisasContext *s, arg_VMOV_single *a) +{ + TCGv_i32 tmp; + + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + if (a->rt == 15) { + /* UNPREDICTABLE; we choose to UNDEF */ + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + if (a->l) { + /* VFP to general purpose register */ + tmp = tcg_temp_new_i32(); + neon_load_reg32(tmp, a->vn); + tcg_gen_andi_i32(tmp, tmp, 0xffff); + store_reg(s, a->rt, tmp); + } else { + /* general purpose register to VFP */ + tmp = load_reg(s, a->rt); + tcg_gen_andi_i32(tmp, tmp, 0xffff); + neon_store_reg32(tmp, a->vn); + tcg_temp_free_i32(tmp); + } + + return true; +} + static bool trans_VMOV_single(DisasContext *s, arg_VMOV_single *a) { TCGv_i32 tmp; -- 2.20.1 Implement FP16 support for the Neon insns which use the DO_3S_FP_GVEC macro: VADD, VSUB, VABD, VMUL. For VABD this requires us to implement a new gvec_fabd_h helper using the machinery we have already for the other helpers. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-24-peter.maydell@linaro.org --- target/arm/helper.h | 1 + target/arm/vec_helper.c | 6 ++++++ target/arm/translate-neon.c.inc | 36 +++++++++++++++++---------------- 3 files changed, 26 insertions(+), 17 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmul_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmul_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmul_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fabd_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fabd_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ static float64 float64_ftsmul(float64 op1, uint64_t op2, float_status *stat) return result; } +static float16 float16_abd(float16 op1, float16 op2, float_status *stat) +{ + return float16_abs(float16_sub(op1, op2, stat)); +} + static float32 float32_abd(float32 op1, float32 op2, float_status *stat) { return float32_abs(float32_sub(op1, op2, stat)); @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_ftsmul_h, float16_ftsmul, float16) DO_3OP(gvec_ftsmul_s, float32_ftsmul, float32) DO_3OP(gvec_ftsmul_d, float64_ftsmul, float64) +DO_3OP(gvec_fabd_h, float16_abd, float16) DO_3OP(gvec_fabd_s, float32_abd, float32) #ifdef TARGET_AARCH64 diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool do_3same_fp(DisasContext *s, arg_3same *a, VFPGen3OpSPFn *fn, return true; } -/* - * For all the functions using this macro, size == 1 means fp16, - * which is an architecture extension we don't implement yet. - */ -#define DO_3S_FP_GVEC(INSN,FUNC) \ - static void gen_##INSN##_3s(unsigned vece, uint32_t rd_ofs, \ - uint32_t rn_ofs, uint32_t rm_ofs, \ - uint32_t oprsz, uint32_t maxsz) \ +#define WRAP_FP_GVEC(WRAPNAME, FPST, FUNC) \ + static void WRAPNAME(unsigned vece, uint32_t rd_ofs, \ + uint32_t rn_ofs, uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ { \ - TCGv_ptr fpst = fpstatus_ptr(FPST_STD); \ + TCGv_ptr fpst = fpstatus_ptr(FPST); \ tcg_gen_gvec_3_ptr(rd_ofs, rn_ofs, rm_ofs, fpst, \ oprsz, maxsz, 0, FUNC); \ tcg_temp_free_ptr(fpst); \ - } \ + } + +#define DO_3S_FP_GVEC(INSN,SFUNC,HFUNC) \ + WRAP_FP_GVEC(gen_##INSN##_fp32_3s, FPST_STD, SFUNC) \ + WRAP_FP_GVEC(gen_##INSN##_fp16_3s, FPST_STD_F16, HFUNC) \ static bool trans_##INSN##_fp_3s(DisasContext *s, arg_3same *a) \ { \ if (a->size != 0) { \ - /* TODO fp16 support */ \ - return false; \ + if (!dc_isar_feature(aa32_fp16_arith, s)) { \ + return false; \ + } \ + return do_3same(s, a, gen_##INSN##_fp16_3s); \ } \ - return do_3same(s, a, gen_##INSN##_3s); \ + return do_3same(s, a, gen_##INSN##_fp32_3s); \ } -DO_3S_FP_GVEC(VADD, gen_helper_gvec_fadd_s) -DO_3S_FP_GVEC(VSUB, gen_helper_gvec_fsub_s) -DO_3S_FP_GVEC(VABD, gen_helper_gvec_fabd_s) -DO_3S_FP_GVEC(VMUL, gen_helper_gvec_fmul_s) +DO_3S_FP_GVEC(VADD, gen_helper_gvec_fadd_s, gen_helper_gvec_fadd_h) +DO_3S_FP_GVEC(VSUB, gen_helper_gvec_fsub_s, gen_helper_gvec_fsub_h) +DO_3S_FP_GVEC(VABD, gen_helper_gvec_fabd_s, gen_helper_gvec_fabd_h) +DO_3S_FP_GVEC(VMUL, gen_helper_gvec_fmul_s, gen_helper_gvec_fmul_h) /* * For all the functions using this macro, size == 1 means fp16, -- 2.20.1 We already have gvec helpers for floating point VRECPE and VRQSRTE, so convert the Neon decoder to use them and add the fp16 support. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-25-peter.maydell@linaro.org --- target/arm/translate-neon.c.inc | 31 +++++++++++++++++++++++++++++-- 1 file changed, 29 insertions(+), 2 deletions(-) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool do_2misc_fp(DisasContext *s, arg_2misc *a, return do_2misc_fp(s, a, FUNC); \ } -DO_2MISC_FP(VRECPE_F, gen_helper_recpe_f32) -DO_2MISC_FP(VRSQRTE_F, gen_helper_rsqrte_f32) DO_2MISC_FP(VCVT_FS, gen_helper_vfp_sitos) DO_2MISC_FP(VCVT_FU, gen_helper_vfp_uitos) DO_2MISC_FP(VCVT_SF, gen_helper_vfp_tosizs) DO_2MISC_FP(VCVT_UF, gen_helper_vfp_touizs) +#define DO_2MISC_FP_VEC(INSN, HFUNC, SFUNC) \ + static void gen_##INSN(unsigned vece, uint32_t rd_ofs, \ + uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ + { \ + static gen_helper_gvec_2_ptr * const fns[4] = { \ + NULL, HFUNC, SFUNC, NULL, \ + }; \ + TCGv_ptr fpst; \ + fpst = fpstatus_ptr(vece == MO_16 ? FPST_STD_F16 : FPST_STD); \ + tcg_gen_gvec_2_ptr(rd_ofs, rm_ofs, fpst, oprsz, maxsz, 0, \ + fns[vece]); \ + tcg_temp_free_ptr(fpst); \ + } \ + static bool trans_##INSN(DisasContext *s, arg_2misc *a) \ + { \ + if (a->size == MO_16) { \ + if (!dc_isar_feature(aa32_fp16_arith, s)) { \ + return false; \ + } \ + } else if (a->size != MO_32) { \ + return false; \ + } \ + return do_2misc_vec(s, a, gen_##INSN); \ + } + +DO_2MISC_FP_VEC(VRECPE_F, gen_helper_gvec_frecpe_h, gen_helper_gvec_frecpe_s) +DO_2MISC_FP_VEC(VRSQRTE_F, gen_helper_gvec_frsqrte_h, gen_helper_gvec_frsqrte_s) + static bool trans_VRINTX(DisasContext *s, arg_2misc *a) { if (!arm_dc_feature(s, ARM_FEATURE_V8)) { -- 2.20.1 Rewrite Neon VABS/VNEG of floats to use gvec logical AND and XOR, so that we can implement the fp16 version of the insns. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-26-peter.maydell@linaro.org --- target/arm/translate-neon.c.inc | 34 +++++++++++++++++++++++++++------ 1 file changed, 28 insertions(+), 6 deletions(-) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VCNT(DisasContext *s, arg_2misc *a) return do_2misc(s, a, gen_helper_neon_cnt_u8); } +static void gen_VABS_F(unsigned vece, uint32_t rd_ofs, uint32_t rm_ofs, + uint32_t oprsz, uint32_t maxsz) +{ + tcg_gen_gvec_andi(vece, rd_ofs, rm_ofs, + vece == MO_16 ? 0x7fff : 0x7fffffff, + oprsz, maxsz); +} + static bool trans_VABS_F(DisasContext *s, arg_2misc *a) { - if (a->size != 2) { + if (a->size == MO_16) { + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + } else if (a->size != MO_32) { return false; } - /* TODO: FP16 : size == 1 */ - return do_2misc(s, a, gen_helper_vfp_abss); + return do_2misc_vec(s, a, gen_VABS_F); +} + +static void gen_VNEG_F(unsigned vece, uint32_t rd_ofs, uint32_t rm_ofs, + uint32_t oprsz, uint32_t maxsz) +{ + tcg_gen_gvec_xori(vece, rd_ofs, rm_ofs, + vece == MO_16 ? 0x8000 : 0x80000000, + oprsz, maxsz); } static bool trans_VNEG_F(DisasContext *s, arg_2misc *a) { - if (a->size != 2) { + if (a->size == MO_16) { + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + } else if (a->size != MO_32) { return false; } - /* TODO: FP16 : size == 1 */ - return do_2misc(s, a, gen_helper_vfp_negs); + return do_2misc_vec(s, a, gen_VNEG_F); } static bool trans_VRECPE(DisasContext *s, arg_2misc *a) -- 2.20.1 Convert the Neon floating-point vector comparison ops VCEQ, VCGE and VCGT over to using a gvec helper and use this to implement the fp16 case. (We put the float16_ceq() etc functions above the DO_2OP() macro definition because later when we convert the compare-against-zero instructions we'll want their definitions to be visible at that point in the source file.) Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-27-peter.maydell@linaro.org --- target/arm/helper.h | 9 +++++++ target/arm/vec_helper.c | 44 +++++++++++++++++++++++++++++++++ target/arm/translate-neon.c.inc | 6 ++--- 3 files changed, 56 insertions(+), 3 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmul_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fabd_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fabd_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fceq_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fceq_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_fcge_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcge_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_fcgt_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcgt_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_ftsmul_s, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(gvec_fcmlad)(void *vd, void *vn, void *vm, clear_tail(d, opr_sz, simd_maxsz(desc)); } +/* + * Floating point comparisons producing an integer result (all 1s or all 0s). + * Note that EQ doesn't signal InvalidOp for QNaNs but GE and GT do. + * Softfloat routines return 0/1, which we convert to the 0/-1 Neon requires. + */ +static uint16_t float16_ceq(float16 op1, float16 op2, float_status *stat) +{ + return -float16_eq_quiet(op1, op2, stat); +} + +static uint32_t float32_ceq(float32 op1, float32 op2, float_status *stat) +{ + return -float32_eq_quiet(op1, op2, stat); +} + +static uint16_t float16_cge(float16 op1, float16 op2, float_status *stat) +{ + return -float16_le(op2, op1, stat); +} + +static uint32_t float32_cge(float32 op1, float32 op2, float_status *stat) +{ + return -float32_le(op2, op1, stat); +} + +static uint16_t float16_cgt(float16 op1, float16 op2, float_status *stat) +{ + return -float16_lt(op2, op1, stat); +} + +static uint32_t float32_cgt(float32 op1, float32 op2, float_status *stat) +{ + return -float32_lt(op2, op1, stat); +} + #define DO_2OP(NAME, FUNC, TYPE) \ void HELPER(NAME)(void *vd, void *vn, void *stat, uint32_t desc) \ { \ @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_ftsmul_d, float64_ftsmul, float64) DO_3OP(gvec_fabd_h, float16_abd, float16) DO_3OP(gvec_fabd_s, float32_abd, float32) +DO_3OP(gvec_fceq_h, float16_ceq, float16) +DO_3OP(gvec_fceq_s, float32_ceq, float32) + +DO_3OP(gvec_fcge_h, float16_cge, float16) +DO_3OP(gvec_fcge_s, float32_cge, float32) + +DO_3OP(gvec_fcgt_h, float16_cgt, float16) +DO_3OP(gvec_fcgt_s, float32_cgt, float32) + #ifdef TARGET_AARCH64 DO_3OP(gvec_recps_h, helper_recpsf_f16, float16) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VADD, gen_helper_gvec_fadd_s, gen_helper_gvec_fadd_h) DO_3S_FP_GVEC(VSUB, gen_helper_gvec_fsub_s, gen_helper_gvec_fsub_h) DO_3S_FP_GVEC(VABD, gen_helper_gvec_fabd_s, gen_helper_gvec_fabd_h) DO_3S_FP_GVEC(VMUL, gen_helper_gvec_fmul_s, gen_helper_gvec_fmul_h) +DO_3S_FP_GVEC(VCEQ, gen_helper_gvec_fceq_s, gen_helper_gvec_fceq_h) +DO_3S_FP_GVEC(VCGE, gen_helper_gvec_fcge_s, gen_helper_gvec_fcge_h) +DO_3S_FP_GVEC(VCGT, gen_helper_gvec_fcgt_s, gen_helper_gvec_fcgt_h) /* * For all the functions using this macro, size == 1 means fp16, @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VMUL, gen_helper_gvec_fmul_s, gen_helper_gvec_fmul_h) return do_3same_fp(s, a, FUNC, READS_VD); \ } -DO_3S_FP(VCEQ, gen_helper_neon_ceq_f32, false) -DO_3S_FP(VCGE, gen_helper_neon_cge_f32, false) -DO_3S_FP(VCGT, gen_helper_neon_cgt_f32, false) DO_3S_FP(VACGE, gen_helper_neon_acge_f32, false) DO_3S_FP(VACGT, gen_helper_neon_acgt_f32, false) DO_3S_FP(VMAX, gen_helper_vfp_maxs, false) -- 2.20.1 Convert the neon floating-point vector absolute comparison ops VACGE and VACGT over to using a gvec hepler and use this to implement the fp16 case. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-28-peter.maydell@linaro.org --- target/arm/helper.h | 6 ++++++ target/arm/vec_helper.c | 26 ++++++++++++++++++++++++++ target/arm/translate-neon.c.inc | 4 ++-- 3 files changed, 34 insertions(+), 2 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fcge_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fcgt_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fcgt_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_facge_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_facge_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_facgt_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_facgt_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_ftsmul_s, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ static uint32_t float32_cgt(float32 op1, float32 op2, float_status *stat) return -float32_lt(op2, op1, stat); } +static uint16_t float16_acge(float16 op1, float16 op2, float_status *stat) +{ + return -float16_le(float16_abs(op2), float16_abs(op1), stat); +} + +static uint32_t float32_acge(float32 op1, float32 op2, float_status *stat) +{ + return -float32_le(float32_abs(op2), float32_abs(op1), stat); +} + +static uint16_t float16_acgt(float16 op1, float16 op2, float_status *stat) +{ + return -float16_lt(float16_abs(op2), float16_abs(op1), stat); +} + +static uint32_t float32_acgt(float32 op1, float32 op2, float_status *stat) +{ + return -float32_lt(float32_abs(op2), float32_abs(op1), stat); +} + #define DO_2OP(NAME, FUNC, TYPE) \ void HELPER(NAME)(void *vd, void *vn, void *stat, uint32_t desc) \ { \ @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_fcge_s, float32_cge, float32) DO_3OP(gvec_fcgt_h, float16_cgt, float16) DO_3OP(gvec_fcgt_s, float32_cgt, float32) +DO_3OP(gvec_facge_h, float16_acge, float16) +DO_3OP(gvec_facge_s, float32_acge, float32) + +DO_3OP(gvec_facgt_h, float16_acgt, float16) +DO_3OP(gvec_facgt_s, float32_acgt, float32) + #ifdef TARGET_AARCH64 DO_3OP(gvec_recps_h, helper_recpsf_f16, float16) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VMUL, gen_helper_gvec_fmul_s, gen_helper_gvec_fmul_h) DO_3S_FP_GVEC(VCEQ, gen_helper_gvec_fceq_s, gen_helper_gvec_fceq_h) DO_3S_FP_GVEC(VCGE, gen_helper_gvec_fcge_s, gen_helper_gvec_fcge_h) DO_3S_FP_GVEC(VCGT, gen_helper_gvec_fcgt_s, gen_helper_gvec_fcgt_h) +DO_3S_FP_GVEC(VACGE, gen_helper_gvec_facge_s, gen_helper_gvec_facge_h) +DO_3S_FP_GVEC(VACGT, gen_helper_gvec_facgt_s, gen_helper_gvec_facgt_h) /* * For all the functions using this macro, size == 1 means fp16, @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VCGT, gen_helper_gvec_fcgt_s, gen_helper_gvec_fcgt_h) return do_3same_fp(s, a, FUNC, READS_VD); \ } -DO_3S_FP(VACGE, gen_helper_neon_acge_f32, false) -DO_3S_FP(VACGT, gen_helper_neon_acgt_f32, false) DO_3S_FP(VMAX, gen_helper_vfp_maxs, false) DO_3S_FP(VMIN, gen_helper_vfp_mins, false) -- 2.20.1 Convert the Neon float-point VMAX and VMIN insns over to using a gvec helper, and use this to implement the fp16 case. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-29-peter.maydell@linaro.org --- target/arm/helper.h | 6 ++++++ target/arm/vec_helper.c | 6 ++++++ target/arm/translate-neon.c.inc | 5 ++--- 3 files changed, 14 insertions(+), 3 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_facge_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_facgt_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_facgt_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmax_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmax_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_fmin_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmin_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_ftsmul_s, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_facge_s, float32_acge, float32) DO_3OP(gvec_facgt_h, float16_acgt, float16) DO_3OP(gvec_facgt_s, float32_acgt, float32) +DO_3OP(gvec_fmax_h, float16_max, float16) +DO_3OP(gvec_fmax_s, float32_max, float32) + +DO_3OP(gvec_fmin_h, float16_min, float16) +DO_3OP(gvec_fmin_s, float32_min, float32) + #ifdef TARGET_AARCH64 DO_3OP(gvec_recps_h, helper_recpsf_f16, float16) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VCGE, gen_helper_gvec_fcge_s, gen_helper_gvec_fcge_h) DO_3S_FP_GVEC(VCGT, gen_helper_gvec_fcgt_s, gen_helper_gvec_fcgt_h) DO_3S_FP_GVEC(VACGE, gen_helper_gvec_facge_s, gen_helper_gvec_facge_h) DO_3S_FP_GVEC(VACGT, gen_helper_gvec_facgt_s, gen_helper_gvec_facgt_h) +DO_3S_FP_GVEC(VMAX, gen_helper_gvec_fmax_s, gen_helper_gvec_fmax_h) +DO_3S_FP_GVEC(VMIN, gen_helper_gvec_fmin_s, gen_helper_gvec_fmin_h) /* * For all the functions using this macro, size == 1 means fp16, @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VACGT, gen_helper_gvec_facgt_s, gen_helper_gvec_facgt_h) return do_3same_fp(s, a, FUNC, READS_VD); \ } -DO_3S_FP(VMAX, gen_helper_vfp_maxs, false) -DO_3S_FP(VMIN, gen_helper_vfp_mins, false) - static void gen_VMLA_fp_3s(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, TCGv_ptr fpstatus) { -- 2.20.1 Convert the Neon floating point VMAXNM and VMINNM insns to using a gvec helper and use this to implement the fp16 case. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-30-peter.maydell@linaro.org --- target/arm/helper.h | 6 ++++++ target/arm/vec_helper.c | 6 ++++++ target/arm/translate-neon.c.inc | 23 +++++++++++++++-------- 3 files changed, 27 insertions(+), 8 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmax_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmin_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmin_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmaxnum_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmaxnum_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_fminnum_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fminnum_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_ftsmul_s, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_fmax_s, float32_max, float32) DO_3OP(gvec_fmin_h, float16_min, float16) DO_3OP(gvec_fmin_s, float32_min, float32) +DO_3OP(gvec_fmaxnum_h, float16_maxnum, float16) +DO_3OP(gvec_fmaxnum_s, float32_maxnum, float32) + +DO_3OP(gvec_fminnum_h, float16_minnum, float16) +DO_3OP(gvec_fminnum_s, float32_minnum, float32) + #ifdef TARGET_AARCH64 DO_3OP(gvec_recps_h, helper_recpsf_f16, float16) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static void gen_VMLS_fp_3s(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, DO_3S_FP(VMLA, gen_VMLA_fp_3s, true) DO_3S_FP(VMLS, gen_VMLS_fp_3s, true) +WRAP_FP_GVEC(gen_VMAXNM_fp32_3s, FPST_STD, gen_helper_gvec_fmaxnum_s) +WRAP_FP_GVEC(gen_VMAXNM_fp16_3s, FPST_STD_F16, gen_helper_gvec_fmaxnum_h) +WRAP_FP_GVEC(gen_VMINNM_fp32_3s, FPST_STD, gen_helper_gvec_fminnum_s) +WRAP_FP_GVEC(gen_VMINNM_fp16_3s, FPST_STD_F16, gen_helper_gvec_fminnum_h) + static bool trans_VMAXNM_fp_3s(DisasContext *s, arg_3same *a) { if (!arm_dc_feature(s, ARM_FEATURE_V8)) { @@ -XXX,XX +XXX,XX @@ static bool trans_VMAXNM_fp_3s(DisasContext *s, arg_3same *a) } if (a->size != 0) { - /* TODO fp16 support */ - return false; + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + return do_3same(s, a, gen_VMAXNM_fp16_3s); } - - return do_3same_fp(s, a, gen_helper_vfp_maxnums, false); + return do_3same(s, a, gen_VMAXNM_fp32_3s); } static bool trans_VMINNM_fp_3s(DisasContext *s, arg_3same *a) @@ -XXX,XX +XXX,XX @@ static bool trans_VMINNM_fp_3s(DisasContext *s, arg_3same *a) } if (a->size != 0) { - /* TODO fp16 support */ - return false; + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + return do_3same(s, a, gen_VMINNM_fp16_3s); } - - return do_3same_fp(s, a, gen_helper_vfp_minnums, false); + return do_3same(s, a, gen_VMINNM_fp32_3s); } WRAP_ENV_FN(gen_VRECPS_tramp, gen_helper_recps_f32) -- 2.20.1 Convert the Neon floating-point VMLA and VMLS insns over to using a gvec helper, and use this to implement the fp16 case. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-31-peter.maydell@linaro.org --- target/arm/helper.h | 6 +++++ target/arm/vec_helper.c | 42 +++++++++++++++++++++++++++++++++ target/arm/translate-neon.c.inc | 33 ++------------------------ 3 files changed, 50 insertions(+), 31 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmaxnum_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i3 DEF_HELPER_FLAGS_5(gvec_fminnum_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fminnum_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmla_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmla_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_fmls_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmls_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_ftsmul_s, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_rsqrts_d, helper_rsqrtsf_f64, float64) #endif #undef DO_3OP +/* Non-fused multiply-add (unlike float16_muladd etc, which are fused) */ +static float16 float16_muladd_nf(float16 dest, float16 op1, float16 op2, + float_status *stat) +{ + return float16_add(dest, float16_mul(op1, op2, stat), stat); +} + +static float32 float32_muladd_nf(float32 dest, float32 op1, float32 op2, + float_status *stat) +{ + return float32_add(dest, float32_mul(op1, op2, stat), stat); +} + +static float16 float16_mulsub_nf(float16 dest, float16 op1, float16 op2, + float_status *stat) +{ + return float16_sub(dest, float16_mul(op1, op2, stat), stat); +} + +static float32 float32_mulsub_nf(float32 dest, float32 op1, float32 op2, + float_status *stat) +{ + return float32_sub(dest, float32_mul(op1, op2, stat), stat); +} + +#define DO_MULADD(NAME, FUNC, TYPE) \ +void HELPER(NAME)(void *vd, void *vn, void *vm, void *stat, uint32_t desc) \ +{ \ + intptr_t i, oprsz = simd_oprsz(desc); \ + TYPE *d = vd, *n = vn, *m = vm; \ + for (i = 0; i < oprsz / sizeof(TYPE); i++) { \ + d[i] = FUNC(d[i], n[i], m[i], stat); \ + } \ + clear_tail(d, oprsz, simd_maxsz(desc)); \ +} + +DO_MULADD(gvec_fmla_h, float16_muladd_nf, float16) +DO_MULADD(gvec_fmla_s, float32_muladd_nf, float32) + +DO_MULADD(gvec_fmls_h, float16_mulsub_nf, float16) +DO_MULADD(gvec_fmls_s, float32_mulsub_nf, float32) + /* For the indexed ops, SVE applies the index per 128-bit vector segment. * For AdvSIMD, there is of course only one such vector segment. */ diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VACGE, gen_helper_gvec_facge_s, gen_helper_gvec_facge_h) DO_3S_FP_GVEC(VACGT, gen_helper_gvec_facgt_s, gen_helper_gvec_facgt_h) DO_3S_FP_GVEC(VMAX, gen_helper_gvec_fmax_s, gen_helper_gvec_fmax_h) DO_3S_FP_GVEC(VMIN, gen_helper_gvec_fmin_s, gen_helper_gvec_fmin_h) - -/* - * For all the functions using this macro, size == 1 means fp16, - * which is an architecture extension we don't implement yet. - */ -#define DO_3S_FP(INSN,FUNC,READS_VD) \ - static bool trans_##INSN##_fp_3s(DisasContext *s, arg_3same *a) \ - { \ - if (a->size != 0) { \ - /* TODO fp16 support */ \ - return false; \ - } \ - return do_3same_fp(s, a, FUNC, READS_VD); \ - } - -static void gen_VMLA_fp_3s(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, - TCGv_ptr fpstatus) -{ - gen_helper_vfp_muls(vn, vn, vm, fpstatus); - gen_helper_vfp_adds(vd, vd, vn, fpstatus); -} - -static void gen_VMLS_fp_3s(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, - TCGv_ptr fpstatus) -{ - gen_helper_vfp_muls(vn, vn, vm, fpstatus); - gen_helper_vfp_subs(vd, vd, vn, fpstatus); -} - -DO_3S_FP(VMLA, gen_VMLA_fp_3s, true) -DO_3S_FP(VMLS, gen_VMLS_fp_3s, true) +DO_3S_FP_GVEC(VMLA, gen_helper_gvec_fmla_s, gen_helper_gvec_fmla_h) +DO_3S_FP_GVEC(VMLS, gen_helper_gvec_fmls_s, gen_helper_gvec_fmls_h) WRAP_FP_GVEC(gen_VMAXNM_fp32_3s, FPST_STD, gen_helper_gvec_fmaxnum_s) WRAP_FP_GVEC(gen_VMAXNM_fp16_3s, FPST_STD_F16, gen_helper_gvec_fmaxnum_h) -- 2.20.1 Convert the neon floating-point vector operations VFMA and VFMS to use a gvec helper, and use this to implement the fp16 case. This is the last use of do_3same_fp() so we can now delete that function. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-32-peter.maydell@linaro.org --- target/arm/helper.h | 6 +++ target/arm/vec_helper.c | 33 +++++++++++- target/arm/translate-neon.c.inc | 92 +-------------------------------- 3 files changed, 40 insertions(+), 91 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmla_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmls_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmls_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_vfma_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_vfma_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_vfms_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_vfms_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_ftsmul_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_ftsmul_s, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ static float32 float32_mulsub_nf(float32 dest, float32 op1, float32 op2, return float32_sub(dest, float32_mul(op1, op2, stat), stat); } -#define DO_MULADD(NAME, FUNC, TYPE) \ +/* Fused versions; these have the semantics Neon VFMA/VFMS want */ +static float16 float16_muladd_f(float16 dest, float16 op1, float16 op2, + float_status *stat) +{ + return float16_muladd(op1, op2, dest, 0, stat); +} + +static float32 float32_muladd_f(float32 dest, float32 op1, float32 op2, + float_status *stat) +{ + return float32_muladd(op1, op2, dest, 0, stat); +} + +static float16 float16_mulsub_f(float16 dest, float16 op1, float16 op2, + float_status *stat) +{ + return float16_muladd(float16_chs(op1), op2, dest, 0, stat); +} + +static float32 float32_mulsub_f(float32 dest, float32 op1, float32 op2, + float_status *stat) +{ + return float32_muladd(float32_chs(op1), op2, dest, 0, stat); +} + +#define DO_MULADD(NAME, FUNC, TYPE) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *stat, uint32_t desc) \ { \ intptr_t i, oprsz = simd_oprsz(desc); \ @@ -XXX,XX +XXX,XX @@ DO_MULADD(gvec_fmla_s, float32_muladd_nf, float32) DO_MULADD(gvec_fmls_h, float16_mulsub_nf, float16) DO_MULADD(gvec_fmls_s, float32_mulsub_nf, float32) +DO_MULADD(gvec_vfma_h, float16_muladd_f, float16) +DO_MULADD(gvec_vfma_s, float32_muladd_f, float32) + +DO_MULADD(gvec_vfms_h, float16_mulsub_f, float16) +DO_MULADD(gvec_vfms_s, float32_mulsub_f, float32) + /* For the indexed ops, SVE applies the index per 128-bit vector segment. * For AdvSIMD, there is of course only one such vector segment. */ diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_3SAME_PAIR(VPADD, padd_u) DO_3SAME_VQDMULH(VQDMULH, qdmulh) DO_3SAME_VQDMULH(VQRDMULH, qrdmulh) -static bool do_3same_fp(DisasContext *s, arg_3same *a, VFPGen3OpSPFn *fn, - bool reads_vd) -{ - /* - * FP operations handled elementwise 32 bits at a time. - * If reads_vd is true then the old value of Vd will be - * loaded before calling the callback function. This is - * used for multiply-accumulate type operations. - */ - TCGv_i32 tmp, tmp2; - int pass; - - if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { - return false; - } - - /* UNDEF accesses to D16-D31 if they don't exist. */ - if (!dc_isar_feature(aa32_simd_r32, s) && - ((a->vd | a->vn | a->vm) & 0x10)) { - return false; - } - - if ((a->vn | a->vm | a->vd) & a->q) { - return false; - } - - if (!vfp_access_check(s)) { - return true; - } - - TCGv_ptr fpstatus = fpstatus_ptr(FPST_STD); - for (pass = 0; pass < (a->q ? 4 : 2); pass++) { - tmp = neon_load_reg(a->vn, pass); - tmp2 = neon_load_reg(a->vm, pass); - if (reads_vd) { - TCGv_i32 tmp_rd = neon_load_reg(a->vd, pass); - fn(tmp_rd, tmp, tmp2, fpstatus); - neon_store_reg(a->vd, pass, tmp_rd); - tcg_temp_free_i32(tmp); - } else { - fn(tmp, tmp, tmp2, fpstatus); - neon_store_reg(a->vd, pass, tmp); - } - tcg_temp_free_i32(tmp2); - } - tcg_temp_free_ptr(fpstatus); - return true; -} - #define WRAP_FP_GVEC(WRAPNAME, FPST, FUNC) \ static void WRAPNAME(unsigned vece, uint32_t rd_ofs, \ uint32_t rn_ofs, uint32_t rm_ofs, \ @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VMAX, gen_helper_gvec_fmax_s, gen_helper_gvec_fmax_h) DO_3S_FP_GVEC(VMIN, gen_helper_gvec_fmin_s, gen_helper_gvec_fmin_h) DO_3S_FP_GVEC(VMLA, gen_helper_gvec_fmla_s, gen_helper_gvec_fmla_h) DO_3S_FP_GVEC(VMLS, gen_helper_gvec_fmls_s, gen_helper_gvec_fmls_h) +DO_3S_FP_GVEC(VFMA, gen_helper_gvec_vfma_s, gen_helper_gvec_vfma_h) +DO_3S_FP_GVEC(VFMS, gen_helper_gvec_vfms_s, gen_helper_gvec_vfms_h) WRAP_FP_GVEC(gen_VMAXNM_fp32_3s, FPST_STD, gen_helper_gvec_fmaxnum_s) WRAP_FP_GVEC(gen_VMAXNM_fp16_3s, FPST_STD_F16, gen_helper_gvec_fmaxnum_h) @@ -XXX,XX +XXX,XX @@ static bool trans_VRSQRTS_fp_3s(DisasContext *s, arg_3same *a) return do_3same(s, a, gen_VRSQRTS_fp_3s); } -static void gen_VFMA_fp_3s(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, - TCGv_ptr fpstatus) -{ - gen_helper_vfp_muladds(vd, vn, vm, vd, fpstatus); -} - -static bool trans_VFMA_fp_3s(DisasContext *s, arg_3same *a) -{ - if (!dc_isar_feature(aa32_simdfmac, s)) { - return false; - } - - if (a->size != 0) { - /* TODO fp16 support */ - return false; - } - - return do_3same_fp(s, a, gen_VFMA_fp_3s, true); -} - -static void gen_VFMS_fp_3s(TCGv_i32 vd, TCGv_i32 vn, TCGv_i32 vm, - TCGv_ptr fpstatus) -{ - gen_helper_vfp_negs(vn, vn); - gen_helper_vfp_muladds(vd, vn, vm, vd, fpstatus); -} - -static bool trans_VFMS_fp_3s(DisasContext *s, arg_3same *a) -{ - if (!dc_isar_feature(aa32_simdfmac, s)) { - return false; - } - - if (a->size != 0) { - /* TODO fp16 support */ - return false; - } - - return do_3same_fp(s, a, gen_VFMS_fp_3s, true); -} - static bool do_3same_fp_pair(DisasContext *s, arg_3same *a, VFPGen3OpSPFn *fn) { /* FP operations handled pairwise 32 bits at a time */ -- 2.20.1 Convert the neon floating-point vector compare-vs-0 insns VCEQ0, VCGT0, VCLE0, VCGE0 and VCLT0 to use a gvec helper, and use this to implement the fp16 case. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-33-peter.maydell@linaro.org --- target/arm/helper.h | 15 +++++++++++++++ target/arm/vec_helper.c | 25 +++++++++++++++++++++++++ target/arm/translate-neon.c.inc | 33 +++++---------------------------- 3 files changed, 45 insertions(+), 28 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(gvec_frsqrte_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frsqrte_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frsqrte_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_fcgt0_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_fcgt0_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_4(gvec_fcge0_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_fcge0_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_4(gvec_fceq0_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_fceq0_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_4(gvec_fcle0_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_fcle0_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_4(gvec_fclt0_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_fclt0_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_fadd_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fadd_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fadd_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_2OP(gvec_frsqrte_h, helper_rsqrte_f16, float16) DO_2OP(gvec_frsqrte_s, helper_rsqrte_f32, float32) DO_2OP(gvec_frsqrte_d, helper_rsqrte_f64, float64) +#define WRAP_CMP0_FWD(FN, CMPOP, TYPE) \ + static TYPE TYPE##_##FN##0(TYPE op, float_status *stat) \ + { \ + return TYPE##_##CMPOP(op, TYPE##_zero, stat); \ + } + +#define WRAP_CMP0_REV(FN, CMPOP, TYPE) \ + static TYPE TYPE##_##FN##0(TYPE op, float_status *stat) \ + { \ + return TYPE##_##CMPOP(TYPE##_zero, op, stat); \ + } + +#define DO_2OP_CMP0(FN, CMPOP, DIRN) \ + WRAP_CMP0_##DIRN(FN, CMPOP, float16) \ + WRAP_CMP0_##DIRN(FN, CMPOP, float32) \ + DO_2OP(gvec_f##FN##0_h, float16_##FN##0, float16) \ + DO_2OP(gvec_f##FN##0_s, float32_##FN##0, float32) + +DO_2OP_CMP0(cgt, cgt, FWD) +DO_2OP_CMP0(cge, cge, FWD) +DO_2OP_CMP0(ceq, ceq, FWD) +DO_2OP_CMP0(clt, cgt, REV) +DO_2OP_CMP0(cle, cge, REV) + #undef DO_2OP +#undef DO_2OP_CMP0 /* Floating-point trigonometric starting value. * See the ARM ARM pseudocode function FPTrigSMul. diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_2MISC_FP(VCVT_UF, gen_helper_vfp_touizs) DO_2MISC_FP_VEC(VRECPE_F, gen_helper_gvec_frecpe_h, gen_helper_gvec_frecpe_s) DO_2MISC_FP_VEC(VRSQRTE_F, gen_helper_gvec_frsqrte_h, gen_helper_gvec_frsqrte_s) +DO_2MISC_FP_VEC(VCGT0_F, gen_helper_gvec_fcgt0_h, gen_helper_gvec_fcgt0_s) +DO_2MISC_FP_VEC(VCGE0_F, gen_helper_gvec_fcge0_h, gen_helper_gvec_fcge0_s) +DO_2MISC_FP_VEC(VCEQ0_F, gen_helper_gvec_fceq0_h, gen_helper_gvec_fceq0_s) +DO_2MISC_FP_VEC(VCLT0_F, gen_helper_gvec_fclt0_h, gen_helper_gvec_fclt0_s) +DO_2MISC_FP_VEC(VCLE0_F, gen_helper_gvec_fcle0_h, gen_helper_gvec_fcle0_s) static bool trans_VRINTX(DisasContext *s, arg_2misc *a) { @@ -XXX,XX +XXX,XX @@ static bool trans_VRINTX(DisasContext *s, arg_2misc *a) return do_2misc_fp(s, a, gen_helper_rints_exact); } -#define WRAP_FP_CMP0_FWD(WRAPNAME, FUNC) \ - static void WRAPNAME(TCGv_i32 d, TCGv_i32 m, TCGv_ptr fpst) \ - { \ - TCGv_i32 zero = tcg_const_i32(0); \ - FUNC(d, m, zero, fpst); \ - tcg_temp_free_i32(zero); \ - } -#define WRAP_FP_CMP0_REV(WRAPNAME, FUNC) \ - static void WRAPNAME(TCGv_i32 d, TCGv_i32 m, TCGv_ptr fpst) \ - { \ - TCGv_i32 zero = tcg_const_i32(0); \ - FUNC(d, zero, m, fpst); \ - tcg_temp_free_i32(zero); \ - } - -#define DO_FP_CMP0(INSN, FUNC, REV) \ - WRAP_FP_CMP0_##REV(gen_##INSN, FUNC) \ - static bool trans_##INSN(DisasContext *s, arg_2misc *a) \ - { \ - return do_2misc_fp(s, a, gen_##INSN); \ - } - -DO_FP_CMP0(VCGT0_F, gen_helper_neon_cgt_f32, FWD) -DO_FP_CMP0(VCGE0_F, gen_helper_neon_cge_f32, FWD) -DO_FP_CMP0(VCEQ0_F, gen_helper_neon_ceq_f32, FWD) -DO_FP_CMP0(VCLE0_F, gen_helper_neon_cge_f32, REV) -DO_FP_CMP0(VCLT0_F, gen_helper_neon_cgt_f32, REV) - static bool do_vrint(DisasContext *s, arg_2misc *a, int rmode) { /* -- 2.20.1 Convert the Neon VRECPS insn to using a gvec helper, and use this to implement the fp16 case. The phrasing of the new float32_recps_nf() is slightly different from the old recps_f32() so that it parallels the f16 version; for f16 we can't assume that flush-to-zero is always enabled. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-34-peter.maydell@linaro.org --- target/arm/helper.h | 4 +++- target/arm/vec_helper.c | 31 +++++++++++++++++++++++++++++++ target/arm/vfp_helper.c | 13 ------------- target/arm/translate-neon.c.inc | 21 +-------------------- 4 files changed, 35 insertions(+), 34 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_4(vfp_muladdd, f64, f64, f64, f64, ptr) DEF_HELPER_4(vfp_muladds, f32, f32, f32, f32, ptr) DEF_HELPER_4(vfp_muladdh, f16, f16, f16, f16, ptr) -DEF_HELPER_3(recps_f32, f32, env, f32, f32) DEF_HELPER_3(rsqrts_f32, f32, env, f32, f32) DEF_HELPER_FLAGS_2(recpe_f16, TCG_CALL_NO_RWG, f16, f16, ptr) DEF_HELPER_FLAGS_2(recpe_f32, TCG_CALL_NO_RWG, f32, f32, ptr) @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmaxnum_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i3 DEF_HELPER_FLAGS_5(gvec_fminnum_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fminnum_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_recps_nf_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_recps_nf_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_fmla_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmla_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ static float32 float32_abd(float32 op1, float32 op2, float_status *stat) return float32_abs(float32_sub(op1, op2, stat)); } +/* + * Reciprocal step. These are the AArch32 version which uses a + * non-fused multiply-and-subtract. + */ +static float16 float16_recps_nf(float16 op1, float16 op2, float_status *stat) +{ + op1 = float16_squash_input_denormal(op1, stat); + op2 = float16_squash_input_denormal(op2, stat); + + if ((float16_is_infinity(op1) && float16_is_zero(op2)) || + (float16_is_infinity(op2) && float16_is_zero(op1))) { + return float16_two; + } + return float16_sub(float16_two, float16_mul(op1, op2, stat), stat); +} + +static float32 float32_recps_nf(float32 op1, float32 op2, float_status *stat) +{ + op1 = float32_squash_input_denormal(op1, stat); + op2 = float32_squash_input_denormal(op2, stat); + + if ((float32_is_infinity(op1) && float32_is_zero(op2)) || + (float32_is_infinity(op2) && float32_is_zero(op1))) { + return float32_two; + } + return float32_sub(float32_two, float32_mul(op1, op2, stat), stat); +} + #define DO_3OP(NAME, FUNC, TYPE) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *stat, uint32_t desc) \ { \ @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_fmaxnum_s, float32_maxnum, float32) DO_3OP(gvec_fminnum_h, float16_minnum, float16) DO_3OP(gvec_fminnum_s, float32_minnum, float32) +DO_3OP(gvec_recps_nf_h, float16_recps_nf, float16) +DO_3OP(gvec_recps_nf_s, float32_recps_nf, float32) + #ifdef TARGET_AARCH64 DO_3OP(gvec_recps_h, helper_recpsf_f16, float16) diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(vfp_fcvt_f64_to_f16)(float64 a, void *fpstp, uint32_t ahp_mode) return r; } -float32 HELPER(recps_f32)(CPUARMState *env, float32 a, float32 b) -{ - float_status *s = &env->vfp.standard_fp_status; - if ((float32_is_infinity(a) && float32_is_zero_or_denormal(b)) || - (float32_is_infinity(b) && float32_is_zero_or_denormal(a))) { - if (!(float32_is_zero(a) || float32_is_zero(b))) { - float_raise(float_flag_input_denormal, s); - } - return float32_two; - } - return float32_sub(float32_two, float32_mul(a, b, s), s); -} - float32 HELPER(rsqrts_f32)(CPUARMState *env, float32 a, float32 b) { float_status *s = &env->vfp.standard_fp_status; diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VMLA, gen_helper_gvec_fmla_s, gen_helper_gvec_fmla_h) DO_3S_FP_GVEC(VMLS, gen_helper_gvec_fmls_s, gen_helper_gvec_fmls_h) DO_3S_FP_GVEC(VFMA, gen_helper_gvec_vfma_s, gen_helper_gvec_vfma_h) DO_3S_FP_GVEC(VFMS, gen_helper_gvec_vfms_s, gen_helper_gvec_vfms_h) +DO_3S_FP_GVEC(VRECPS, gen_helper_gvec_recps_nf_s, gen_helper_gvec_recps_nf_h) WRAP_FP_GVEC(gen_VMAXNM_fp32_3s, FPST_STD, gen_helper_gvec_fmaxnum_s) WRAP_FP_GVEC(gen_VMAXNM_fp16_3s, FPST_STD_F16, gen_helper_gvec_fmaxnum_h) @@ -XXX,XX +XXX,XX @@ static bool trans_VMINNM_fp_3s(DisasContext *s, arg_3same *a) return do_3same(s, a, gen_VMINNM_fp32_3s); } -WRAP_ENV_FN(gen_VRECPS_tramp, gen_helper_recps_f32) - -static void gen_VRECPS_fp_3s(unsigned vece, uint32_t rd_ofs, - uint32_t rn_ofs, uint32_t rm_ofs, - uint32_t oprsz, uint32_t maxsz) -{ - static const GVecGen3 ops = { .fni4 = gen_VRECPS_tramp }; - tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, oprsz, maxsz, &ops); -} - -static bool trans_VRECPS_fp_3s(DisasContext *s, arg_3same *a) -{ - if (a->size != 0) { - /* TODO fp16 support */ - return false; - } - - return do_3same(s, a, gen_VRECPS_fp_3s); -} - WRAP_ENV_FN(gen_VRSQRTS_tramp, gen_helper_rsqrts_f32) static void gen_VRSQRTS_fp_3s(unsigned vece, uint32_t rd_ofs, -- 2.20.1 Convert the Neon VRSQRTS insn to using a gvec helper, and use this to implement the fp16 case. As with VRECPS, we adjust the phrasing of the new implementation slightly so that the fp32 version parallels the fp16 one. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-35-peter.maydell@linaro.org --- target/arm/helper.h | 4 +++- target/arm/vec_helper.c | 30 ++++++++++++++++++++++++++++++ target/arm/vfp_helper.c | 15 --------------- target/arm/translate-neon.c.inc | 21 +-------------------- 4 files changed, 34 insertions(+), 36 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_4(vfp_muladdd, f64, f64, f64, f64, ptr) DEF_HELPER_4(vfp_muladds, f32, f32, f32, f32, ptr) DEF_HELPER_4(vfp_muladdh, f16, f16, f16, f16, ptr) -DEF_HELPER_3(rsqrts_f32, f32, env, f32, f32) DEF_HELPER_FLAGS_2(recpe_f16, TCG_CALL_NO_RWG, f16, f16, ptr) DEF_HELPER_FLAGS_2(recpe_f32, TCG_CALL_NO_RWG, f32, f32, ptr) DEF_HELPER_FLAGS_2(recpe_f64, TCG_CALL_NO_RWG, f64, f64, ptr) @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fminnum_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i3 DEF_HELPER_FLAGS_5(gvec_recps_nf_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_recps_nf_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_rsqrts_nf_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_rsqrts_nf_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_5(gvec_fmla_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(gvec_fmla_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ static float32 float32_recps_nf(float32 op1, float32 op2, float_status *stat) return float32_sub(float32_two, float32_mul(op1, op2, stat), stat); } +/* Reciprocal square-root step. AArch32 non-fused semantics. */ +static float16 float16_rsqrts_nf(float16 op1, float16 op2, float_status *stat) +{ + op1 = float16_squash_input_denormal(op1, stat); + op2 = float16_squash_input_denormal(op2, stat); + + if ((float16_is_infinity(op1) && float16_is_zero(op2)) || + (float16_is_infinity(op2) && float16_is_zero(op1))) { + return float16_one_point_five; + } + op1 = float16_sub(float16_three, float16_mul(op1, op2, stat), stat); + return float16_div(op1, float16_two, stat); +} + +static float32 float32_rsqrts_nf(float32 op1, float32 op2, float_status *stat) +{ + op1 = float32_squash_input_denormal(op1, stat); + op2 = float32_squash_input_denormal(op2, stat); + + if ((float32_is_infinity(op1) && float32_is_zero(op2)) || + (float32_is_infinity(op2) && float32_is_zero(op1))) { + return float32_one_point_five; + } + op1 = float32_sub(float32_three, float32_mul(op1, op2, stat), stat); + return float32_div(op1, float32_two, stat); +} + #define DO_3OP(NAME, FUNC, TYPE) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *stat, uint32_t desc) \ { \ @@ -XXX,XX +XXX,XX @@ DO_3OP(gvec_fminnum_s, float32_minnum, float32) DO_3OP(gvec_recps_nf_h, float16_recps_nf, float16) DO_3OP(gvec_recps_nf_s, float32_recps_nf, float32) +DO_3OP(gvec_rsqrts_nf_h, float16_rsqrts_nf, float16) +DO_3OP(gvec_rsqrts_nf_s, float32_rsqrts_nf, float32) + #ifdef TARGET_AARCH64 DO_3OP(gvec_recps_h, helper_recpsf_f16, float16) diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(vfp_fcvt_f64_to_f16)(float64 a, void *fpstp, uint32_t ahp_mode) return r; } -float32 HELPER(rsqrts_f32)(CPUARMState *env, float32 a, float32 b) -{ - float_status *s = &env->vfp.standard_fp_status; - float32 product; - if ((float32_is_infinity(a) && float32_is_zero_or_denormal(b)) || - (float32_is_infinity(b) && float32_is_zero_or_denormal(a))) { - if (!(float32_is_zero(a) || float32_is_zero(b))) { - float_raise(float_flag_input_denormal, s); - } - return float32_one_point_five; - } - product = float32_mul(a, b, s); - return float32_div(float32_sub(float32_three, product, s), float32_two, s); -} - /* NEON helpers. */ /* Constants 256 and 512 are used in some helpers; we avoid relying on diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_3S_FP_GVEC(VMLS, gen_helper_gvec_fmls_s, gen_helper_gvec_fmls_h) DO_3S_FP_GVEC(VFMA, gen_helper_gvec_vfma_s, gen_helper_gvec_vfma_h) DO_3S_FP_GVEC(VFMS, gen_helper_gvec_vfms_s, gen_helper_gvec_vfms_h) DO_3S_FP_GVEC(VRECPS, gen_helper_gvec_recps_nf_s, gen_helper_gvec_recps_nf_h) +DO_3S_FP_GVEC(VRSQRTS, gen_helper_gvec_rsqrts_nf_s, gen_helper_gvec_rsqrts_nf_h) WRAP_FP_GVEC(gen_VMAXNM_fp32_3s, FPST_STD, gen_helper_gvec_fmaxnum_s) WRAP_FP_GVEC(gen_VMAXNM_fp16_3s, FPST_STD_F16, gen_helper_gvec_fmaxnum_h) @@ -XXX,XX +XXX,XX @@ static bool trans_VMINNM_fp_3s(DisasContext *s, arg_3same *a) return do_3same(s, a, gen_VMINNM_fp32_3s); } -WRAP_ENV_FN(gen_VRSQRTS_tramp, gen_helper_rsqrts_f32) - -static void gen_VRSQRTS_fp_3s(unsigned vece, uint32_t rd_ofs, - uint32_t rn_ofs, uint32_t rm_ofs, - uint32_t oprsz, uint32_t maxsz) -{ - static const GVecGen3 ops = { .fni4 = gen_VRSQRTS_tramp }; - tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, oprsz, maxsz, &ops); -} - -static bool trans_VRSQRTS_fp_3s(DisasContext *s, arg_3same *a) -{ - if (a->size != 0) { - /* TODO fp16 support */ - return false; - } - - return do_3same(s, a, gen_VRSQRTS_fp_3s); -} - static bool do_3same_fp_pair(DisasContext *s, arg_3same *a, VFPGen3OpSPFn *fn) { /* FP operations handled pairwise 32 bits at a time */ -- 2.20.1 Convert the Neon pairwise fp ops to use a single gvic-style helper to do the full operation instead of one helper call for each 32-bit part. This allows us to use the same framework to implement the fp16. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-36-peter.maydell@linaro.org --- target/arm/helper.h | 7 +++++ target/arm/vec_helper.c | 45 +++++++++++++++++++++++++++++++++ target/arm/translate-neon.c.inc | 42 ++++++++++++------------------ 3 files changed, 68 insertions(+), 26 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fcmlas_idx, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_5(gvec_fcmlad, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(neon_paddh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(neon_pmaxh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(neon_pminh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(neon_padds, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(neon_pmaxs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(neon_pmins, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(gvec_frecpe_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_ABA(gvec_uaba_s, uint32_t) DO_ABA(gvec_uaba_d, uint64_t) #undef DO_ABA + +#define DO_NEON_PAIRWISE(NAME, OP) \ + void HELPER(NAME##s)(void *vd, void *vn, void *vm, \ + void *stat, uint32_t oprsz) \ + { \ + float_status *fpst = stat; \ + float32 *d = vd; \ + float32 *n = vn; \ + float32 *m = vm; \ + float32 r0, r1; \ + \ + /* Read all inputs before writing outputs in case vm == vd */ \ + r0 = float32_##OP(n[H4(0)], n[H4(1)], fpst); \ + r1 = float32_##OP(m[H4(0)], m[H4(1)], fpst); \ + \ + d[H4(0)] = r0; \ + d[H4(1)] = r1; \ + } \ + \ + void HELPER(NAME##h)(void *vd, void *vn, void *vm, \ + void *stat, uint32_t oprsz) \ + { \ + float_status *fpst = stat; \ + float16 *d = vd; \ + float16 *n = vn; \ + float16 *m = vm; \ + float16 r0, r1, r2, r3; \ + \ + /* Read all inputs before writing outputs in case vm == vd */ \ + r0 = float16_##OP(n[H2(0)], n[H2(1)], fpst); \ + r1 = float16_##OP(n[H2(2)], n[H2(3)], fpst); \ + r2 = float16_##OP(m[H2(0)], m[H2(1)], fpst); \ + r3 = float16_##OP(m[H2(2)], m[H2(3)], fpst); \ + \ + d[H4(0)] = r0; \ + d[H4(1)] = r1; \ + d[H4(2)] = r2; \ + d[H4(3)] = r3; \ + } + +DO_NEON_PAIRWISE(neon_padd, add) +DO_NEON_PAIRWISE(neon_pmax, max) +DO_NEON_PAIRWISE(neon_pmin, min) + +#undef DO_NEON_PAIRWISE diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VMINNM_fp_3s(DisasContext *s, arg_3same *a) return do_3same(s, a, gen_VMINNM_fp32_3s); } -static bool do_3same_fp_pair(DisasContext *s, arg_3same *a, VFPGen3OpSPFn *fn) +static bool do_3same_fp_pair(DisasContext *s, arg_3same *a, + gen_helper_gvec_3_ptr *fn) { - /* FP operations handled pairwise 32 bits at a time */ - TCGv_i32 tmp, tmp2, tmp3; + /* FP pairwise operations */ TCGv_ptr fpstatus; if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { @@ -XXX,XX +XXX,XX @@ static bool do_3same_fp_pair(DisasContext *s, arg_3same *a, VFPGen3OpSPFn *fn) assert(a->q == 0); /* enforced by decode patterns */ - /* - * Note that we have to be careful not to clobber the source operands - * in the "vm == vd" case by storing the result of the first pass too - * early. Since Q is 0 there are always just two passes, so instead - * of a complicated loop over each pass we just unroll. - */ - fpstatus = fpstatus_ptr(FPST_STD); - tmp = neon_load_reg(a->vn, 0); - tmp2 = neon_load_reg(a->vn, 1); - fn(tmp, tmp, tmp2, fpstatus); - tcg_temp_free_i32(tmp2); - tmp3 = neon_load_reg(a->vm, 0); - tmp2 = neon_load_reg(a->vm, 1); - fn(tmp3, tmp3, tmp2, fpstatus); - tcg_temp_free_i32(tmp2); + fpstatus = fpstatus_ptr(a->size != 0 ? FPST_STD_F16 : FPST_STD); + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, a->vd), + vfp_reg_offset(1, a->vn), + vfp_reg_offset(1, a->vm), + fpstatus, 8, 8, 0, fn); tcg_temp_free_ptr(fpstatus); - neon_store_reg(a->vd, 0, tmp); - neon_store_reg(a->vd, 1, tmp3); return true; } @@ -XXX,XX +XXX,XX @@ static bool do_3same_fp_pair(DisasContext *s, arg_3same *a, VFPGen3OpSPFn *fn) static bool trans_##INSN##_fp_3s(DisasContext *s, arg_3same *a) \ { \ if (a->size != 0) { \ - /* TODO fp16 support */ \ - return false; \ + if (!dc_isar_feature(aa32_fp16_arith, s)) { \ + return false; \ + } \ + return do_3same_fp_pair(s, a, FUNC##h); \ } \ - return do_3same_fp_pair(s, a, FUNC); \ + return do_3same_fp_pair(s, a, FUNC##s); \ } -DO_3S_FP_PAIR(VPADD, gen_helper_vfp_adds) -DO_3S_FP_PAIR(VPMAX, gen_helper_vfp_maxs) -DO_3S_FP_PAIR(VPMIN, gen_helper_vfp_mins) +DO_3S_FP_PAIR(VPADD, gen_helper_neon_padd) +DO_3S_FP_PAIR(VPMAX, gen_helper_neon_pmax) +DO_3S_FP_PAIR(VPMIN, gen_helper_neon_pmin) static bool do_vector_2sh(DisasContext *s, arg_2reg_shift *a, GVecGen2iFn *fn) { -- 2.20.1 Convert the Neon float-integer VCVT insns to gvec, and use this to implement fp16 support for them. Note that unlike the VFP int<->fp16 VCVT insns we converted earlier and which convert to/from a 32-bit integer, these Neon insns convert to/from 16-bit integers. So we can use the existing vfp conversion helpers for the f32<->u32/i32 case but need to provide our own for f16<->u16/i16. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-37-peter.maydell@linaro.org --- target/arm/helper.h | 9 +++++++++ target/arm/vec_helper.c | 29 +++++++++++++++++++++++++++++ target/arm/translate-neon.c.inc | 15 ++++----------- 3 files changed, 42 insertions(+), 11 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(neon_padds, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(neon_pmaxs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(neon_pmins, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_sstoh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_sitos, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_ustoh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_uitos, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_tosszh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_tosizs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_touszh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_touizs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(gvec_frecpe_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ static uint32_t float32_acgt(float32 op1, float32 op2, float_status *stat) return -float32_lt(float32_abs(op2), float32_abs(op1), stat); } +static int16_t vfp_tosszh(float16 x, void *fpstp) +{ + float_status *fpst = fpstp; + if (float16_is_any_nan(x)) { + float_raise(float_flag_invalid, fpst); + return 0; + } + return float16_to_int16_round_to_zero(x, fpst); +} + +static uint16_t vfp_touszh(float16 x, void *fpstp) +{ + float_status *fpst = fpstp; + if (float16_is_any_nan(x)) { + float_raise(float_flag_invalid, fpst); + return 0; + } + return float16_to_uint16_round_to_zero(x, fpst); +} + #define DO_2OP(NAME, FUNC, TYPE) \ void HELPER(NAME)(void *vd, void *vn, void *stat, uint32_t desc) \ { \ @@ -XXX,XX +XXX,XX @@ DO_2OP(gvec_frsqrte_h, helper_rsqrte_f16, float16) DO_2OP(gvec_frsqrte_s, helper_rsqrte_f32, float32) DO_2OP(gvec_frsqrte_d, helper_rsqrte_f64, float64) +DO_2OP(gvec_sitos, helper_vfp_sitos, int32_t) +DO_2OP(gvec_uitos, helper_vfp_uitos, uint32_t) +DO_2OP(gvec_tosizs, helper_vfp_tosizs, float32) +DO_2OP(gvec_touizs, helper_vfp_touizs, float32) +DO_2OP(gvec_sstoh, int16_to_float16, int16_t) +DO_2OP(gvec_ustoh, uint16_to_float16, uint16_t) +DO_2OP(gvec_tosszh, vfp_tosszh, float16) +DO_2OP(gvec_touszh, vfp_touszh, float16) + #define WRAP_CMP0_FWD(FN, CMPOP, TYPE) \ static TYPE TYPE##_##FN##0(TYPE op, float_status *stat) \ { \ diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool do_2misc_fp(DisasContext *s, arg_2misc *a, return true; } -#define DO_2MISC_FP(INSN, FUNC) \ - static bool trans_##INSN(DisasContext *s, arg_2misc *a) \ - { \ - return do_2misc_fp(s, a, FUNC); \ - } - -DO_2MISC_FP(VCVT_FS, gen_helper_vfp_sitos) -DO_2MISC_FP(VCVT_FU, gen_helper_vfp_uitos) -DO_2MISC_FP(VCVT_SF, gen_helper_vfp_tosizs) -DO_2MISC_FP(VCVT_UF, gen_helper_vfp_touizs) - #define DO_2MISC_FP_VEC(INSN, HFUNC, SFUNC) \ static void gen_##INSN(unsigned vece, uint32_t rd_ofs, \ uint32_t rm_ofs, \ @@ -XXX,XX +XXX,XX @@ DO_2MISC_FP_VEC(VCGE0_F, gen_helper_gvec_fcge0_h, gen_helper_gvec_fcge0_s) DO_2MISC_FP_VEC(VCEQ0_F, gen_helper_gvec_fceq0_h, gen_helper_gvec_fceq0_s) DO_2MISC_FP_VEC(VCLT0_F, gen_helper_gvec_fclt0_h, gen_helper_gvec_fclt0_s) DO_2MISC_FP_VEC(VCLE0_F, gen_helper_gvec_fcle0_h, gen_helper_gvec_fcle0_s) +DO_2MISC_FP_VEC(VCVT_FS, gen_helper_gvec_sstoh, gen_helper_gvec_sitos) +DO_2MISC_FP_VEC(VCVT_FU, gen_helper_gvec_ustoh, gen_helper_gvec_uitos) +DO_2MISC_FP_VEC(VCVT_SF, gen_helper_gvec_tosszh, gen_helper_gvec_tosizs) +DO_2MISC_FP_VEC(VCVT_UF, gen_helper_gvec_touszh, gen_helper_gvec_touizs) static bool trans_VRINTX(DisasContext *s, arg_2misc *a) { -- 2.20.1 Convert the Neon VCVT float<->fixed-point insns to a gvec style, in preparation for adding fp16 support. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-38-peter.maydell@linaro.org --- target/arm/helper.h | 5 +++++ target/arm/vec_helper.c | 20 +++++++++++++++++++ target/arm/translate-neon.c.inc | 35 +++++++++++++++++---------------- 3 files changed, 43 insertions(+), 17 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(gvec_tosizs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_touszh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_touizs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_sf, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_uf, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_fs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_fu, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(gvec_frecpe_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_NEON_PAIRWISE(neon_pmax, max) DO_NEON_PAIRWISE(neon_pmin, min) #undef DO_NEON_PAIRWISE + +#define DO_VCVT_FIXED(NAME, FUNC, TYPE) \ + void HELPER(NAME)(void *vd, void *vn, void *stat, uint32_t desc) \ + { \ + intptr_t i, oprsz = simd_oprsz(desc); \ + int shift = simd_data(desc); \ + TYPE *d = vd, *n = vn; \ + float_status *fpst = stat; \ + for (i = 0; i < oprsz / sizeof(TYPE); i++) { \ + d[i] = FUNC(n[i], shift, fpst); \ + } \ + clear_tail(d, oprsz, simd_maxsz(desc)); \ + } + +DO_VCVT_FIXED(gvec_vcvt_sf, helper_vfp_sltos, uint32_t) +DO_VCVT_FIXED(gvec_vcvt_uf, helper_vfp_ultos, uint32_t) +DO_VCVT_FIXED(gvec_vcvt_fs, helper_vfp_tosls_round_to_zero, uint32_t) +DO_VCVT_FIXED(gvec_vcvt_fu, helper_vfp_touls_round_to_zero, uint32_t) + +#undef DO_VCVT_FIXED diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VSHLL_U_2sh(DisasContext *s, arg_2reg_shift *a) } static bool do_fp_2sh(DisasContext *s, arg_2reg_shift *a, - NeonGenTwoSingleOpFn *fn) + gen_helper_gvec_2_ptr *fn) { /* FP operations in 2-reg-and-shift group */ - TCGv_i32 tmp, shiftv; - TCGv_ptr fpstatus; - int pass; + int vec_size = a->q ? 16 : 8; + int rd_ofs = neon_reg_offset(a->vd, 0); + int rm_ofs = neon_reg_offset(a->vm, 0); + TCGv_ptr fpst; if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { return false; } + if (a->size != 0) { + if (!dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + } + /* UNDEF accesses to D16-D31 if they don't exist. */ if (!dc_isar_feature(aa32_simd_r32, s) && ((a->vd | a->vm) & 0x10)) { @@ -XXX,XX +XXX,XX @@ static bool do_fp_2sh(DisasContext *s, arg_2reg_shift *a, return true; } - fpstatus = fpstatus_ptr(FPST_STD); - shiftv = tcg_const_i32(a->shift); - for (pass = 0; pass < (a->q ? 4 : 2); pass++) { - tmp = neon_load_reg(a->vm, pass); - fn(tmp, tmp, shiftv, fpstatus); - neon_store_reg(a->vd, pass, tmp); - } - tcg_temp_free_ptr(fpstatus); - tcg_temp_free_i32(shiftv); + fpst = fpstatus_ptr(a->size ? FPST_STD_F16 : FPST_STD); + tcg_gen_gvec_2_ptr(rd_ofs, rm_ofs, fpst, vec_size, vec_size, a->shift, fn); + tcg_temp_free_ptr(fpst); return true; } @@ -XXX,XX +XXX,XX @@ static bool do_fp_2sh(DisasContext *s, arg_2reg_shift *a, return do_fp_2sh(s, a, FUNC); \ } -DO_FP_2SH(VCVT_SF, gen_helper_vfp_sltos) -DO_FP_2SH(VCVT_UF, gen_helper_vfp_ultos) -DO_FP_2SH(VCVT_FS, gen_helper_vfp_tosls_round_to_zero) -DO_FP_2SH(VCVT_FU, gen_helper_vfp_touls_round_to_zero) +DO_FP_2SH(VCVT_SF, gen_helper_gvec_vcvt_sf) +DO_FP_2SH(VCVT_UF, gen_helper_gvec_vcvt_uf) +DO_FP_2SH(VCVT_FS, gen_helper_gvec_vcvt_fs) +DO_FP_2SH(VCVT_FU, gen_helper_gvec_vcvt_fu) static uint64_t asimd_imm_const(uint32_t imm, int cmode, int op) { -- 2.20.1 Implement fp16 for the Neon VCVT insns which convert between float and fixed-point. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-39-peter.maydell@linaro.org --- target/arm/helper.h | 5 +++++ target/arm/neon-dp.decode | 8 +++++++- target/arm/vec_helper.c | 4 ++++ target/arm/translate-neon.c.inc | 5 +++++ 4 files changed, 21 insertions(+), 1 deletion(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(gvec_vcvt_uf, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vcvt_fs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vcvt_fu, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_sh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_uh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_hs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_hu, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(gvec_frecpe_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ VMINNM_fp_3s 1111 001 1 0 . 1 . .... .... 1111 ... 1 .... @3same_fp # We use size=0 for fp32 and size=1 for fp16 to match the 3-same encodings. @2reg_vcvt .... ... . . . 1 ..... .... .... . q:1 . . .... \ &2reg_shift vm=%vm_dp vd=%vd_dp size=0 shift=%neon_rshift_i5 +@2reg_vcvt_f16 .... ... . . . 11 .... .... .... . q:1 . . .... \ + &2reg_shift vm=%vm_dp vd=%vd_dp size=1 shift=%neon_rshift_i4 VSHR_S_2sh 1111 001 0 1 . ...... .... 0000 . . . 1 .... @2reg_shr_d VSHR_S_2sh 1111 001 0 1 . ...... .... 0000 . . . 1 .... @2reg_shr_s @@ -XXX,XX +XXX,XX @@ VSHLL_U_2sh 1111 001 1 1 . ...... .... 1010 . 0 . 1 .... @2reg_shll_h VSHLL_U_2sh 1111 001 1 1 . ...... .... 1010 . 0 . 1 .... @2reg_shll_b # VCVT fixed<->float conversions -# TODO: FP16 fixed<->float conversions are opc==0b1100 and 0b1101 +VCVT_SH_2sh 1111 001 0 1 . ...... .... 1100 0 . . 1 .... @2reg_vcvt_f16 +VCVT_UH_2sh 1111 001 1 1 . ...... .... 1100 0 . . 1 .... @2reg_vcvt_f16 +VCVT_HS_2sh 1111 001 0 1 . ...... .... 1101 0 . . 1 .... @2reg_vcvt_f16 +VCVT_HU_2sh 1111 001 1 1 . ...... .... 1101 0 . . 1 .... @2reg_vcvt_f16 + VCVT_SF_2sh 1111 001 0 1 . ...... .... 1110 0 . . 1 .... @2reg_vcvt VCVT_UF_2sh 1111 001 1 1 . ...... .... 1110 0 . . 1 .... @2reg_vcvt VCVT_FS_2sh 1111 001 0 1 . ...... .... 1111 0 . . 1 .... @2reg_vcvt diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_VCVT_FIXED(gvec_vcvt_sf, helper_vfp_sltos, uint32_t) DO_VCVT_FIXED(gvec_vcvt_uf, helper_vfp_ultos, uint32_t) DO_VCVT_FIXED(gvec_vcvt_fs, helper_vfp_tosls_round_to_zero, uint32_t) DO_VCVT_FIXED(gvec_vcvt_fu, helper_vfp_touls_round_to_zero, uint32_t) +DO_VCVT_FIXED(gvec_vcvt_sh, helper_vfp_shtoh, uint16_t) +DO_VCVT_FIXED(gvec_vcvt_uh, helper_vfp_uhtoh, uint16_t) +DO_VCVT_FIXED(gvec_vcvt_hs, helper_vfp_toshh_round_to_zero, uint16_t) +DO_VCVT_FIXED(gvec_vcvt_hu, helper_vfp_touhh_round_to_zero, uint16_t) #undef DO_VCVT_FIXED diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_FP_2SH(VCVT_UF, gen_helper_gvec_vcvt_uf) DO_FP_2SH(VCVT_FS, gen_helper_gvec_vcvt_fs) DO_FP_2SH(VCVT_FU, gen_helper_gvec_vcvt_fu) +DO_FP_2SH(VCVT_SH, gen_helper_gvec_vcvt_sh) +DO_FP_2SH(VCVT_UH, gen_helper_gvec_vcvt_uh) +DO_FP_2SH(VCVT_HS, gen_helper_gvec_vcvt_hs) +DO_FP_2SH(VCVT_HU, gen_helper_gvec_vcvt_hu) + static uint64_t asimd_imm_const(uint32_t imm, int cmode, int op) { /* -- 2.20.1 Convert the Neon VCVT with-specified-rounding-mode instructions to gvec, and use this to implement fp16 support for them. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-40-peter.maydell@linaro.org --- target/arm/helper.h | 5 ++ target/arm/vec_helper.c | 23 +++++++ target/arm/translate-neon.c.inc | 105 ++++++++++++-------------------- 3 files changed, 66 insertions(+), 67 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(gvec_vcvt_uh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vcvt_hs, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vcvt_hu, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_rm_ss, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_rm_us, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_rm_sh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vcvt_rm_uh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(gvec_frecpe_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_VCVT_FIXED(gvec_vcvt_hs, helper_vfp_toshh_round_to_zero, uint16_t) DO_VCVT_FIXED(gvec_vcvt_hu, helper_vfp_touhh_round_to_zero, uint16_t) #undef DO_VCVT_FIXED + +#define DO_VCVT_RMODE(NAME, FUNC, TYPE) \ + void HELPER(NAME)(void *vd, void *vn, void *stat, uint32_t desc) \ + { \ + float_status *fpst = stat; \ + intptr_t i, oprsz = simd_oprsz(desc); \ + uint32_t rmode = simd_data(desc); \ + uint32_t prev_rmode = get_float_rounding_mode(fpst); \ + TYPE *d = vd, *n = vn; \ + set_float_rounding_mode(rmode, fpst); \ + for (i = 0; i < oprsz / sizeof(TYPE); i++) { \ + d[i] = FUNC(n[i], 0, fpst); \ + } \ + set_float_rounding_mode(prev_rmode, fpst); \ + clear_tail(d, oprsz, simd_maxsz(desc)); \ + } + +DO_VCVT_RMODE(gvec_vcvt_rm_ss, helper_vfp_tosls, uint32_t) +DO_VCVT_RMODE(gvec_vcvt_rm_us, helper_vfp_touls, uint32_t) +DO_VCVT_RMODE(gvec_vcvt_rm_sh, helper_vfp_toshh, uint16_t) +DO_VCVT_RMODE(gvec_vcvt_rm_uh, helper_vfp_touhh, uint16_t) + +#undef DO_VCVT_RMODE diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ DO_VRINT(VRINTZ, FPROUNDING_ZERO) DO_VRINT(VRINTM, FPROUNDING_NEGINF) DO_VRINT(VRINTP, FPROUNDING_POSINF) -static bool do_vcvt(DisasContext *s, arg_2misc *a, int rmode, bool is_signed) -{ - /* - * Handle a VCVT* operation by iterating 32 bits at a time, - * with a specified rounding mode in operation. - */ - int pass; - TCGv_ptr fpst; - TCGv_i32 tcg_rmode, tcg_shift; - - if (!arm_dc_feature(s, ARM_FEATURE_NEON) || - !arm_dc_feature(s, ARM_FEATURE_V8)) { - return false; +#define DO_VEC_RMODE(INSN, RMODE, OP) \ + static void gen_##INSN(unsigned vece, uint32_t rd_ofs, \ + uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ + { \ + static gen_helper_gvec_2_ptr * const fns[4] = { \ + NULL, \ + gen_helper_gvec_##OP##h, \ + gen_helper_gvec_##OP##s, \ + NULL, \ + }; \ + TCGv_ptr fpst; \ + fpst = fpstatus_ptr(vece == 1 ? FPST_STD_F16 : FPST_STD); \ + tcg_gen_gvec_2_ptr(rd_ofs, rm_ofs, fpst, oprsz, maxsz, \ + arm_rmode_to_sf(RMODE), fns[vece]); \ + tcg_temp_free_ptr(fpst); \ + } \ + static bool trans_##INSN(DisasContext *s, arg_2misc *a) \ + { \ + if (!arm_dc_feature(s, ARM_FEATURE_V8)) { \ + return false; \ + } \ + if (a->size == MO_16) { \ + if (!dc_isar_feature(aa32_fp16_arith, s)) { \ + return false; \ + } \ + } else if (a->size != MO_32) { \ + return false; \ + } \ + return do_2misc_vec(s, a, gen_##INSN); \ } - /* UNDEF accesses to D16-D31 if they don't exist. */ - if (!dc_isar_feature(aa32_simd_r32, s) && - ((a->vd | a->vm) & 0x10)) { - return false; - } - - if (a->size != 2) { - /* TODO: FP16 will be the size == 1 case */ - return false; - } - - if ((a->vd | a->vm) & a->q) { - return false; - } - - if (!vfp_access_check(s)) { - return true; - } - - fpst = fpstatus_ptr(FPST_STD); - tcg_shift = tcg_const_i32(0); - tcg_rmode = tcg_const_i32(arm_rmode_to_sf(rmode)); - gen_helper_set_neon_rmode(tcg_rmode, tcg_rmode, cpu_env); - for (pass = 0; pass < (a->q ? 4 : 2); pass++) { - TCGv_i32 tmp = neon_load_reg(a->vm, pass); - if (is_signed) { - gen_helper_vfp_tosls(tmp, tmp, tcg_shift, fpst); - } else { - gen_helper_vfp_touls(tmp, tmp, tcg_shift, fpst); - } - neon_store_reg(a->vd, pass, tmp); - } - gen_helper_set_neon_rmode(tcg_rmode, tcg_rmode, cpu_env); - tcg_temp_free_i32(tcg_rmode); - tcg_temp_free_i32(tcg_shift); - tcg_temp_free_ptr(fpst); - - return true; -} - -#define DO_VCVT(INSN, RMODE, SIGNED) \ - static bool trans_##INSN(DisasContext *s, arg_2misc *a) \ - { \ - return do_vcvt(s, a, RMODE, SIGNED); \ - } - -DO_VCVT(VCVTAU, FPROUNDING_TIEAWAY, false) -DO_VCVT(VCVTAS, FPROUNDING_TIEAWAY, true) -DO_VCVT(VCVTNU, FPROUNDING_TIEEVEN, false) -DO_VCVT(VCVTNS, FPROUNDING_TIEEVEN, true) -DO_VCVT(VCVTPU, FPROUNDING_POSINF, false) -DO_VCVT(VCVTPS, FPROUNDING_POSINF, true) -DO_VCVT(VCVTMU, FPROUNDING_NEGINF, false) -DO_VCVT(VCVTMS, FPROUNDING_NEGINF, true) +DO_VEC_RMODE(VCVTAU, FPROUNDING_TIEAWAY, vcvt_rm_u) +DO_VEC_RMODE(VCVTAS, FPROUNDING_TIEAWAY, vcvt_rm_s) +DO_VEC_RMODE(VCVTNU, FPROUNDING_TIEEVEN, vcvt_rm_u) +DO_VEC_RMODE(VCVTNS, FPROUNDING_TIEEVEN, vcvt_rm_s) +DO_VEC_RMODE(VCVTPU, FPROUNDING_POSINF, vcvt_rm_u) +DO_VEC_RMODE(VCVTPS, FPROUNDING_POSINF, vcvt_rm_s) +DO_VEC_RMODE(VCVTMU, FPROUNDING_NEGINF, vcvt_rm_u) +DO_VEC_RMODE(VCVTMS, FPROUNDING_NEGINF, vcvt_rm_s) static bool trans_VSWP(DisasContext *s, arg_2misc *a) { -- 2.20.1 Convert the Neon VRINT-with-specified-rounding-mode insns to gvec, and use this to implement the fp16 versions. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-41-peter.maydell@linaro.org --- target/arm/helper.h | 4 +- target/arm/vec_helper.c | 21 +++++++++++ target/arm/vfp_helper.c | 17 --------- target/arm/translate-neon.c.inc | 67 +++------------------------------ 4 files changed, 30 insertions(+), 79 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_3(vfp_sqtoh, f16, i64, i32, ptr) DEF_HELPER_3(vfp_uqtoh, f16, i64, i32, ptr) DEF_HELPER_FLAGS_2(set_rmode, TCG_CALL_NO_RWG, i32, i32, ptr) -DEF_HELPER_FLAGS_2(set_neon_rmode, TCG_CALL_NO_RWG, i32, i32, env) DEF_HELPER_FLAGS_3(vfp_fcvt_f16_to_f32, TCG_CALL_NO_RWG, f32, f16, ptr, i32) DEF_HELPER_FLAGS_3(vfp_fcvt_f32_to_f16, TCG_CALL_NO_RWG, f16, f32, ptr, i32) @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(gvec_vcvt_rm_us, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vcvt_rm_sh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vcvt_rm_uh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vrint_rm_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vrint_rm_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(gvec_frecpe_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_VCVT_RMODE(gvec_vcvt_rm_sh, helper_vfp_toshh, uint16_t) DO_VCVT_RMODE(gvec_vcvt_rm_uh, helper_vfp_touhh, uint16_t) #undef DO_VCVT_RMODE + +#define DO_VRINT_RMODE(NAME, FUNC, TYPE) \ + void HELPER(NAME)(void *vd, void *vn, void *stat, uint32_t desc) \ + { \ + float_status *fpst = stat; \ + intptr_t i, oprsz = simd_oprsz(desc); \ + uint32_t rmode = simd_data(desc); \ + uint32_t prev_rmode = get_float_rounding_mode(fpst); \ + TYPE *d = vd, *n = vn; \ + set_float_rounding_mode(rmode, fpst); \ + for (i = 0; i < oprsz / sizeof(TYPE); i++) { \ + d[i] = FUNC(n[i], fpst); \ + } \ + set_float_rounding_mode(prev_rmode, fpst); \ + clear_tail(d, oprsz, simd_maxsz(desc)); \ + } + +DO_VRINT_RMODE(gvec_vrint_rm_h, helper_rinth, uint16_t) +DO_VRINT_RMODE(gvec_vrint_rm_s, helper_rints, uint32_t) + +#undef DO_VRINT_RMODE diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vfp_helper.c +++ b/target/arm/vfp_helper.c @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(set_rmode)(uint32_t rmode, void *fpstp) return prev_rmode; } -/* Set the current fp rounding mode in the standard fp status and return - * the old one. This is for NEON instructions that need to change the - * rounding mode but wish to use the standard FPSCR values for everything - * else. Always set the rounding mode back to the correct value after - * modifying it. - * The argument is a softfloat float_round_ value. - */ -uint32_t HELPER(set_neon_rmode)(uint32_t rmode, CPUARMState *env) -{ - float_status *fp_status = &env->vfp.standard_fp_status; - - uint32_t prev_rmode = get_float_rounding_mode(fp_status); - set_float_rounding_mode(rmode, fp_status); - - return prev_rmode; -} - /* Half precision conversions. */ float32 HELPER(vfp_fcvt_f16_to_f32)(uint32_t a, void *fpstp, uint32_t ahp_mode) { diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VRINTX(DisasContext *s, arg_2misc *a) return do_2misc_fp(s, a, gen_helper_rints_exact); } -static bool do_vrint(DisasContext *s, arg_2misc *a, int rmode) -{ - /* - * Handle a VRINT* operation by iterating 32 bits at a time, - * with a specified rounding mode in operation. - */ - int pass; - TCGv_ptr fpst; - TCGv_i32 tcg_rmode; - - if (!arm_dc_feature(s, ARM_FEATURE_NEON) || - !arm_dc_feature(s, ARM_FEATURE_V8)) { - return false; - } - - /* UNDEF accesses to D16-D31 if they don't exist. */ - if (!dc_isar_feature(aa32_simd_r32, s) && - ((a->vd | a->vm) & 0x10)) { - return false; - } - - if (a->size != 2) { - /* TODO: FP16 will be the size == 1 case */ - return false; - } - - if ((a->vd | a->vm) & a->q) { - return false; - } - - if (!vfp_access_check(s)) { - return true; - } - - fpst = fpstatus_ptr(FPST_STD); - tcg_rmode = tcg_const_i32(arm_rmode_to_sf(rmode)); - gen_helper_set_neon_rmode(tcg_rmode, tcg_rmode, cpu_env); - for (pass = 0; pass < (a->q ? 4 : 2); pass++) { - TCGv_i32 tmp = neon_load_reg(a->vm, pass); - gen_helper_rints(tmp, tmp, fpst); - neon_store_reg(a->vd, pass, tmp); - } - gen_helper_set_neon_rmode(tcg_rmode, tcg_rmode, cpu_env); - tcg_temp_free_i32(tcg_rmode); - tcg_temp_free_ptr(fpst); - - return true; -} - -#define DO_VRINT(INSN, RMODE) \ - static bool trans_##INSN(DisasContext *s, arg_2misc *a) \ - { \ - return do_vrint(s, a, RMODE); \ - } - -DO_VRINT(VRINTN, FPROUNDING_TIEEVEN) -DO_VRINT(VRINTA, FPROUNDING_TIEAWAY) -DO_VRINT(VRINTZ, FPROUNDING_ZERO) -DO_VRINT(VRINTM, FPROUNDING_NEGINF) -DO_VRINT(VRINTP, FPROUNDING_POSINF) - #define DO_VEC_RMODE(INSN, RMODE, OP) \ static void gen_##INSN(unsigned vece, uint32_t rd_ofs, \ uint32_t rm_ofs, \ @@ -XXX,XX +XXX,XX @@ DO_VEC_RMODE(VCVTPS, FPROUNDING_POSINF, vcvt_rm_s) DO_VEC_RMODE(VCVTMU, FPROUNDING_NEGINF, vcvt_rm_u) DO_VEC_RMODE(VCVTMS, FPROUNDING_NEGINF, vcvt_rm_s) +DO_VEC_RMODE(VRINTN, FPROUNDING_TIEEVEN, vrint_rm_) +DO_VEC_RMODE(VRINTA, FPROUNDING_TIEAWAY, vrint_rm_) +DO_VEC_RMODE(VRINTZ, FPROUNDING_ZERO, vrint_rm_) +DO_VEC_RMODE(VRINTM, FPROUNDING_NEGINF, vrint_rm_) +DO_VEC_RMODE(VRINTP, FPROUNDING_POSINF, vrint_rm_) + static bool trans_VSWP(DisasContext *s, arg_2misc *a) { TCGv_i64 rm, rd; -- 2.20.1 Convert the Neon VRINTX insn to use gvec, and use this to implement fp16 support for it. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-42-peter.maydell@linaro.org --- target/arm/helper.h | 3 +++ target/arm/vec_helper.c | 3 +++ target/arm/translate-neon.c.inc | 45 +++------------------------------ 3 files changed, 9 insertions(+), 42 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(gvec_vcvt_rm_uh, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vrint_rm_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_vrint_rm_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vrintx_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(gvec_vrintx_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(gvec_frecpe_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(gvec_frecpe_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_2OP(gvec_frsqrte_h, helper_rsqrte_f16, float16) DO_2OP(gvec_frsqrte_s, helper_rsqrte_f32, float32) DO_2OP(gvec_frsqrte_d, helper_rsqrte_f64, float64) +DO_2OP(gvec_vrintx_h, float16_round_to_int, float16) +DO_2OP(gvec_vrintx_s, float32_round_to_int, float32) + DO_2OP(gvec_sitos, helper_vfp_sitos, int32_t) DO_2OP(gvec_uitos, helper_vfp_uitos, uint32_t) DO_2OP(gvec_tosizs, helper_vfp_tosizs, float32) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VQNEG(DisasContext *s, arg_2misc *a) return do_2misc(s, a, fn[a->size]); } -static bool do_2misc_fp(DisasContext *s, arg_2misc *a, - NeonGenOneSingleOpFn *fn) -{ - int pass; - TCGv_ptr fpst; - - /* Handle a 2-reg-misc operation by iterating 32 bits at a time */ - if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { - return false; - } - - /* UNDEF accesses to D16-D31 if they don't exist. */ - if (!dc_isar_feature(aa32_simd_r32, s) && - ((a->vd | a->vm) & 0x10)) { - return false; - } - - if (a->size != 2) { - /* TODO: FP16 will be the size == 1 case */ - return false; - } - - if ((a->vd | a->vm) & a->q) { - return false; - } - - if (!vfp_access_check(s)) { - return true; - } - - fpst = fpstatus_ptr(FPST_STD); - for (pass = 0; pass < (a->q ? 4 : 2); pass++) { - TCGv_i32 tmp = neon_load_reg(a->vm, pass); - fn(tmp, tmp, fpst); - neon_store_reg(a->vd, pass, tmp); - } - tcg_temp_free_ptr(fpst); - - return true; -} - #define DO_2MISC_FP_VEC(INSN, HFUNC, SFUNC) \ static void gen_##INSN(unsigned vece, uint32_t rd_ofs, \ uint32_t rm_ofs, \ @@ -XXX,XX +XXX,XX @@ DO_2MISC_FP_VEC(VCVT_FU, gen_helper_gvec_ustoh, gen_helper_gvec_uitos) DO_2MISC_FP_VEC(VCVT_SF, gen_helper_gvec_tosszh, gen_helper_gvec_tosizs) DO_2MISC_FP_VEC(VCVT_UF, gen_helper_gvec_touszh, gen_helper_gvec_touizs) +DO_2MISC_FP_VEC(VRINTX_impl, gen_helper_gvec_vrintx_h, gen_helper_gvec_vrintx_s) + static bool trans_VRINTX(DisasContext *s, arg_2misc *a) { if (!arm_dc_feature(s, ARM_FEATURE_V8)) { return false; } - return do_2misc_fp(s, a, gen_helper_rints_exact); + return trans_VRINTX_impl(s, a); } #define DO_VEC_RMODE(INSN, RMODE, OP) \ -- 2.20.1 In the gvec helper functions for indexed operations, for AArch32 Neon the oprsz (total size of the vector) can be less than 16 bytes if the operation is on a D reg. Since the inner loop in these helpers always goes from 0 to segment, we must clamp it based on oprsz to avoid processing a full 16 byte segment when asked to handle an 8 byte wide vector. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-43-peter.maydell@linaro.org --- target/arm/vec_helper.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_MULADD(gvec_vfms_s, float32_mulsub_f, float32) #define DO_MUL_IDX(NAME, TYPE, H) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ - intptr_t i, j, oprsz = simd_oprsz(desc), segment = 16 / sizeof(TYPE); \ + intptr_t i, j, oprsz = simd_oprsz(desc); \ + intptr_t segment = MIN(16, oprsz) / sizeof(TYPE); \ intptr_t idx = simd_data(desc); \ TYPE *d = vd, *n = vn, *m = vm; \ for (i = 0; i < oprsz / sizeof(TYPE); i += segment) { \ @@ -XXX,XX +XXX,XX @@ DO_MUL_IDX(gvec_mul_idx_d, uint64_t, ) #define DO_MLA_IDX(NAME, TYPE, OP, H) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, uint32_t desc) \ { \ - intptr_t i, j, oprsz = simd_oprsz(desc), segment = 16 / sizeof(TYPE); \ + intptr_t i, j, oprsz = simd_oprsz(desc); \ + intptr_t segment = MIN(16, oprsz) / sizeof(TYPE); \ intptr_t idx = simd_data(desc); \ TYPE *d = vd, *n = vn, *m = vm, *a = va; \ for (i = 0; i < oprsz / sizeof(TYPE); i += segment) { \ @@ -XXX,XX +XXX,XX @@ DO_MLA_IDX(gvec_mls_idx_d, uint64_t, -, ) #define DO_FMUL_IDX(NAME, TYPE, H) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *stat, uint32_t desc) \ { \ - intptr_t i, j, oprsz = simd_oprsz(desc), segment = 16 / sizeof(TYPE); \ + intptr_t i, j, oprsz = simd_oprsz(desc); \ + intptr_t segment = MIN(16, oprsz) / sizeof(TYPE); \ intptr_t idx = simd_data(desc); \ TYPE *d = vd, *n = vn, *m = vm; \ for (i = 0; i < oprsz / sizeof(TYPE); i += segment) { \ @@ -XXX,XX +XXX,XX @@ DO_FMUL_IDX(gvec_fmul_idx_d, float64, ) void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, \ void *stat, uint32_t desc) \ { \ - intptr_t i, j, oprsz = simd_oprsz(desc), segment = 16 / sizeof(TYPE); \ + intptr_t i, j, oprsz = simd_oprsz(desc); \ + intptr_t segment = MIN(16, oprsz) / sizeof(TYPE); \ TYPE op1_neg = extract32(desc, SIMD_DATA_SHIFT, 1); \ intptr_t idx = desc >> (SIMD_DATA_SHIFT + 1); \ TYPE *d = vd, *n = vn, *m = vm, *a = va; \ -- 2.20.1 Add gvec helpers for doing Neon-style indexed non-fused fp multiply-and-accumulate operations. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20200828183354.27913-44-peter.maydell@linaro.org --- target/arm/helper.h | 10 ++++++++++ target/arm/vec_helper.c | 27 ++++++++++++++++++++++----- 2 files changed, 32 insertions(+), 5 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmul_idx_s, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_5(gvec_fmul_idx_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmla_nf_idx_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmla_nf_idx_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_fmls_nf_idx_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fmls_nf_idx_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_6(gvec_fmla_idx_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_6(gvec_fmla_idx_s, TCG_CALL_NO_RWG, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ DO_MLA_IDX(gvec_mls_idx_d, uint64_t, -, ) #undef DO_MLA_IDX -#define DO_FMUL_IDX(NAME, TYPE, H) \ +#define DO_FMUL_IDX(NAME, ADD, TYPE, H) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *stat, uint32_t desc) \ { \ intptr_t i, j, oprsz = simd_oprsz(desc); \ @@ -XXX,XX +XXX,XX @@ void HELPER(NAME)(void *vd, void *vn, void *vm, void *stat, uint32_t desc) \ for (i = 0; i < oprsz / sizeof(TYPE); i += segment) { \ TYPE mm = m[H(i + idx)]; \ for (j = 0; j < segment; j++) { \ - d[i + j] = TYPE##_mul(n[i + j], mm, stat); \ + d[i + j] = TYPE##_##ADD(d[i + j], \ + TYPE##_mul(n[i + j], mm, stat), stat); \ } \ } \ clear_tail(d, oprsz, simd_maxsz(desc)); \ } -DO_FMUL_IDX(gvec_fmul_idx_h, float16, H2) -DO_FMUL_IDX(gvec_fmul_idx_s, float32, H4) -DO_FMUL_IDX(gvec_fmul_idx_d, float64, ) +#define float16_nop(N, M, S) (M) +#define float32_nop(N, M, S) (M) +#define float64_nop(N, M, S) (M) +DO_FMUL_IDX(gvec_fmul_idx_h, nop, float16, H2) +DO_FMUL_IDX(gvec_fmul_idx_s, nop, float32, H4) +DO_FMUL_IDX(gvec_fmul_idx_d, nop, float64, ) + +/* + * Non-fused multiply-accumulate operations, for Neon. NB that unlike + * the fused ops below they assume accumulate both from and into Vd. + */ +DO_FMUL_IDX(gvec_fmla_nf_idx_h, add, float16, H2) +DO_FMUL_IDX(gvec_fmla_nf_idx_s, add, float32, H4) +DO_FMUL_IDX(gvec_fmls_nf_idx_h, sub, float16, H2) +DO_FMUL_IDX(gvec_fmls_nf_idx_s, sub, float32, H4) + +#undef float16_nop +#undef float32_nop +#undef float64_nop #undef DO_FMUL_IDX #define DO_FMLA_IDX(NAME, TYPE, H) \ -- 2.20.1 Convert the Neon floating-point VMUL, VMLA and VMLS to use gvec, and use this to implement fp16 support. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-45-peter.maydell@linaro.org --- target/arm/translate-neon.c.inc | 114 ++++++++++++++++---------------- 1 file changed, 57 insertions(+), 57 deletions(-) diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.c.inc +++ b/target/arm/translate-neon.c.inc @@ -XXX,XX +XXX,XX @@ static bool trans_VMLS_2sc(DisasContext *s, arg_2scalar *a) return do_2scalar(s, a, opfn[a->size], accfn[a->size]); } -/* - * Rather than have a float-specific version of do_2scalar just for - * three insns, we wrap a NeonGenTwoSingleOpFn to turn it into - * a NeonGenTwoOpFn. - */ -#define WRAP_FP_FN(WRAPNAME, FUNC) \ - static void WRAPNAME(TCGv_i32 rd, TCGv_i32 rn, TCGv_i32 rm) \ - { \ - TCGv_ptr fpstatus = fpstatus_ptr(FPST_STD); \ - FUNC(rd, rn, rm, fpstatus); \ - tcg_temp_free_ptr(fpstatus); \ +static bool do_2scalar_fp_vec(DisasContext *s, arg_2scalar *a, + gen_helper_gvec_3_ptr *fn) +{ + /* Two registers and a scalar, using gvec */ + int vec_size = a->q ? 16 : 8; + int rd_ofs = neon_reg_offset(a->vd, 0); + int rn_ofs = neon_reg_offset(a->vn, 0); + int rm_ofs; + int idx; + TCGv_ptr fpstatus; + + if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { + return false; } -WRAP_FP_FN(gen_VMUL_F_mul, gen_helper_vfp_muls) -WRAP_FP_FN(gen_VMUL_F_add, gen_helper_vfp_adds) -WRAP_FP_FN(gen_VMUL_F_sub, gen_helper_vfp_subs) + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd | a->vn | a->vm) & 0x10)) { + return false; + } -static bool trans_VMUL_F_2sc(DisasContext *s, arg_2scalar *a) -{ - static NeonGenTwoOpFn * const opfn[] = { - NULL, - NULL, /* TODO: fp16 support */ - gen_VMUL_F_mul, - NULL, - }; + if (!fn) { + /* Bad size (including size == 3, which is a different insn group) */ + return false; + } - return do_2scalar(s, a, opfn[a->size], NULL); + if (a->q && ((a->vd | a->vn) & 1)) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + /* a->vm is M:Vm, which encodes both register and index */ + idx = extract32(a->vm, a->size + 2, 2); + a->vm = extract32(a->vm, 0, a->size + 2); + rm_ofs = neon_reg_offset(a->vm, 0); + + fpstatus = fpstatus_ptr(a->size == 1 ? FPST_STD_F16 : FPST_STD); + tcg_gen_gvec_3_ptr(rd_ofs, rn_ofs, rm_ofs, fpstatus, + vec_size, vec_size, idx, fn); + tcg_temp_free_ptr(fpstatus); + return true; } -static bool trans_VMLA_F_2sc(DisasContext *s, arg_2scalar *a) -{ - static NeonGenTwoOpFn * const opfn[] = { - NULL, - NULL, /* TODO: fp16 support */ - gen_VMUL_F_mul, - NULL, - }; - static NeonGenTwoOpFn * const accfn[] = { - NULL, - NULL, /* TODO: fp16 support */ - gen_VMUL_F_add, - NULL, - }; +#define DO_VMUL_F_2sc(NAME, FUNC) \ + static bool trans_##NAME##_F_2sc(DisasContext *s, arg_2scalar *a) \ + { \ + static gen_helper_gvec_3_ptr * const opfn[] = { \ + NULL, \ + gen_helper_##FUNC##_h, \ + gen_helper_##FUNC##_s, \ + NULL, \ + }; \ + if (a->size == MO_16 && !dc_isar_feature(aa32_fp16_arith, s)) { \ + return false; \ + } \ + return do_2scalar_fp_vec(s, a, opfn[a->size]); \ + } - return do_2scalar(s, a, opfn[a->size], accfn[a->size]); -} - -static bool trans_VMLS_F_2sc(DisasContext *s, arg_2scalar *a) -{ - static NeonGenTwoOpFn * const opfn[] = { - NULL, - NULL, /* TODO: fp16 support */ - gen_VMUL_F_mul, - NULL, - }; - static NeonGenTwoOpFn * const accfn[] = { - NULL, - NULL, /* TODO: fp16 support */ - gen_VMUL_F_sub, - NULL, - }; - - return do_2scalar(s, a, opfn[a->size], accfn[a->size]); -} +DO_VMUL_F_2sc(VMUL, gvec_fmul_idx) +DO_VMUL_F_2sc(VMLA, gvec_fmla_nf_idx) +DO_VMUL_F_2sc(VMLS, gvec_fmls_nf_idx) WRAP_ENV_FN(gen_VQDMULH_16, gen_helper_neon_qdmulh_s16) WRAP_ENV_FN(gen_VQDMULH_32, gen_helper_neon_qdmulh_s32) -- 2.20.1 Set the MVFR1 ID register FPHP and SIMDHP fields to indicate that our "-cpu max" has v8.2-FP16. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200828183354.27913-46-peter.maydell@linaro.org --- target/arm/cpu.c | 3 ++- target/arm/cpu64.c | 10 ++++------ 2 files changed, 6 insertions(+), 7 deletions(-) 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_max_initfn(Object *obj) cpu->isar.id_isar6 = t; t = cpu->isar.mvfr1; - t = FIELD_DP32(t, MVFR1, FPHP, 2); /* v8.0 FP support */ + t = FIELD_DP32(t, MVFR1, FPHP, 3); /* v8.2-FP16 */ + t = FIELD_DP32(t, MVFR1, SIMDHP, 2); /* v8.2-FP16 */ cpu->isar.mvfr1 = t; t = cpu->isar.mvfr2; diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu64.c +++ b/target/arm/cpu64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) u = FIELD_DP32(u, ID_DFR0, PERFMON, 5); /* v8.4-PMU */ cpu->isar.id_dfr0 = u; - /* - * FIXME: We do not yet support ARMv8.2-fp16 for AArch32 yet, - * so do not set MVFR1.FPHP. Strictly speaking this is not legal, - * but it is also not legal to enable SVE without support for FP16, - * and enabling SVE in system mode is more useful in the short term. - */ + u = cpu->isar.mvfr1; + u = FIELD_DP32(u, MVFR1, FPHP, 3); /* v8.2-FP16 */ + u = FIELD_DP32(u, MVFR1, SIMDHP, 2); /* v8.2-FP16 */ + cpu->isar.mvfr1 = u; #ifdef CONFIG_USER_ONLY /* For usermode -cpu max we can use a larger and more efficient DCZ -- 2.20.1 From: Leif Lindholm <leif@nuviainc.com> The sbsa-ref platform uses a minimal device tree to pass amount of memory as well as number of cpus to the firmware. However, when dumping that minimal dtb (with -M sbsa-virt,dumpdtb=<file>), the resulting blob generates a warning when decompiled by dtc due to lack of reg property. Add a simple reg property per cpu, representing a 64-bit MPIDR_EL1. This also ends up being cleaner than having the firmware calculating its own IDs for generating APCI. Signed-off-by: Leif Lindholm <leif@nuviainc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Message-id: 20200827124335.30586-1-leif@nuviainc.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/sbsa-ref.c | 29 +++++++++++++++++++++++------ 1 file changed, 23 insertions(+), 6 deletions(-) diff --git a/hw/arm/sbsa-ref.c b/hw/arm/sbsa-ref.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/sbsa-ref.c +++ b/hw/arm/sbsa-ref.c @@ -XXX,XX +XXX,XX @@ static const int sbsa_ref_irqmap[] = { [SBSA_EHCI] = 11, }; +static uint64_t sbsa_ref_cpu_mp_affinity(SBSAMachineState *sms, int idx) +{ + uint8_t clustersz = ARM_DEFAULT_CPUS_PER_CLUSTER; + return arm_cpu_mp_affinity(idx, clustersz); +} + /* * Firmware on this machine only uses ACPI table to load OS, these limited * device tree nodes are just to let firmware know the info which varies from @@ -XXX,XX +XXX,XX @@ static void create_fdt(SBSAMachineState *sms) g_free(matrix); } + /* + * From Documentation/devicetree/bindings/arm/cpus.yaml + * On ARM v8 64-bit systems this property is required + * and matches the MPIDR_EL1 register affinity bits. + * + * * If cpus node's #address-cells property is set to 2 + * + * The first reg cell bits [7:0] must be set to + * bits [39:32] of MPIDR_EL1. + * + * The second reg cell bits [23:0] must be set to + * bits [23:0] of MPIDR_EL1. + */ qemu_fdt_add_subnode(sms->fdt, "/cpus"); + qemu_fdt_setprop_cell(sms->fdt, "/cpus", "#address-cells", 2); + qemu_fdt_setprop_cell(sms->fdt, "/cpus", "#size-cells", 0x0); for (cpu = sms->smp_cpus - 1; cpu >= 0; cpu--) { char *nodename = g_strdup_printf("/cpus/cpu@%d", cpu); ARMCPU *armcpu = ARM_CPU(qemu_get_cpu(cpu)); CPUState *cs = CPU(armcpu); + uint64_t mpidr = sbsa_ref_cpu_mp_affinity(sms, cpu); qemu_fdt_add_subnode(sms->fdt, nodename); + qemu_fdt_setprop_u64(sms->fdt, nodename, "reg", mpidr); if (ms->possible_cpus->cpus[cs->cpu_index].props.has_node_id) { qemu_fdt_setprop_cell(sms->fdt, nodename, "numa-node-id", @@ -XXX,XX +XXX,XX @@ static void sbsa_ref_init(MachineState *machine) arm_load_kernel(ARM_CPU(first_cpu), machine, &sms->bootinfo); } -static uint64_t sbsa_ref_cpu_mp_affinity(SBSAMachineState *sms, int idx) -{ - uint8_t clustersz = ARM_DEFAULT_CPUS_PER_CLUSTER; - return arm_cpu_mp_affinity(idx, clustersz); -} - static const CPUArchIdList *sbsa_ref_possible_cpu_arch_ids(MachineState *ms) { unsigned int max_cpus = ms->smp.max_cpus; -- 2.20.1 From: Graeme Gregory <graeme@nuviainc.com> A difference between sbsa platform and the virt platform is PSCI is handled by ARM-TF in the sbsa platform. This means that the PSCI code there needs to communicate some of the platform power changes down to the qemu code for things like shutdown/reset control. Space has been left to extend the EC if we find other use cases in future where ARM-TF and qemu need to communicate. Signed-off-by: Graeme Gregory <graeme@nuviainc.com> Reviewed-by: Leif Lindholm <leif@nuviainc.com> Tested-by: Leif Lindholm <leif@nuviainc.com> Message-id: 20200826141952.136164-2-graeme@nuviainc.com Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/misc/sbsa_ec.c | 98 +++++++++++++++++++++++++++++++++++++++++++++ hw/misc/meson.build | 2 + 2 files changed, 100 insertions(+) create mode 100644 hw/misc/sbsa_ec.c diff --git a/hw/misc/sbsa_ec.c b/hw/misc/sbsa_ec.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/misc/sbsa_ec.c @@ -XXX,XX +XXX,XX @@ +/* + * ARM SBSA Reference Platform Embedded Controller + * + * A device to allow PSCI running in the secure side of sbsa-ref machine + * to communicate platform power states to qemu. + * + * Copyright (c) 2020 Nuvia Inc + * Written by Graeme Gregory <graeme@nuviainc.com> + * + * SPDX-License-Identifer: GPL-2.0-or-later + */ + +#include "qemu/osdep.h" +#include "qemu-common.h" +#include "qemu/log.h" +#include "hw/sysbus.h" +#include "sysemu/runstate.h" + +typedef struct { + SysBusDevice parent_obj; + MemoryRegion iomem; +} SECUREECState; + +#define TYPE_SBSA_EC "sbsa-ec" +#define SECURE_EC(obj) OBJECT_CHECK(SECUREECState, (obj), TYPE_SBSA_EC) + +enum sbsa_ec_powerstates { + SBSA_EC_CMD_POWEROFF = 0x01, + SBSA_EC_CMD_REBOOT = 0x02, +}; + +static uint64_t sbsa_ec_read(void *opaque, hwaddr offset, unsigned size) +{ + /* No use for this currently */ + qemu_log_mask(LOG_GUEST_ERROR, "sbsa-ec: no readable registers"); + return 0; +} + +static void sbsa_ec_write(void *opaque, hwaddr offset, + uint64_t value, unsigned size) +{ + if (offset == 0) { /* PSCI machine power command register */ + switch (value) { + case SBSA_EC_CMD_POWEROFF: + qemu_system_shutdown_request(SHUTDOWN_CAUSE_GUEST_SHUTDOWN); + break; + case SBSA_EC_CMD_REBOOT: + qemu_system_reset_request(SHUTDOWN_CAUSE_GUEST_RESET); + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, + "sbsa-ec: unknown power command"); + } + } else { + qemu_log_mask(LOG_GUEST_ERROR, "sbsa-ec: unknown EC register"); + } +} + +static const MemoryRegionOps sbsa_ec_ops = { + .read = sbsa_ec_read, + .write = sbsa_ec_write, + .endianness = DEVICE_NATIVE_ENDIAN, + .valid.min_access_size = 4, + .valid.max_access_size = 4, +}; + +static void sbsa_ec_init(Object *obj) +{ + SECUREECState *s = SECURE_EC(obj); + SysBusDevice *dev = SYS_BUS_DEVICE(obj); + + memory_region_init_io(&s->iomem, obj, &sbsa_ec_ops, s, "sbsa-ec", + 0x1000); + sysbus_init_mmio(dev, &s->iomem); +} + +static void sbsa_ec_class_init(ObjectClass *klass, void *data) +{ + DeviceClass *dc = DEVICE_CLASS(klass); + + /* No vmstate or reset required: device has no internal state */ + dc->user_creatable = false; +} + +static const TypeInfo sbsa_ec_info = { + .name = TYPE_SBSA_EC, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(SECUREECState), + .instance_init = sbsa_ec_init, + .class_init = sbsa_ec_class_init, +}; + +static void sbsa_ec_register_type(void) +{ + type_register_static(&sbsa_ec_info); +} + +type_init(sbsa_ec_register_type); diff --git a/hw/misc/meson.build b/hw/misc/meson.build index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/meson.build +++ b/hw/misc/meson.build @@ -XXX,XX +XXX,XX @@ specific_ss.add(when: 'CONFIG_MAC_VIA', if_true: files('mac_via.c')) specific_ss.add(when: 'CONFIG_MIPS_CPS', if_true: files('mips_cmgcr.c', 'mips_cpc.c')) specific_ss.add(when: 'CONFIG_MIPS_ITU', if_true: files('mips_itu.c')) + +specific_ss.add(when: 'CONFIG_SBSA_REF', if_true: files('sbsa_ec.c')) -- 2.20.1 From: Graeme Gregory <graeme@nuviainc.com> Add the previously created sbsa-ec device to the sbsa-ref machine in secure memory so the PSCI implementation in ARM-TF can access it, but not expose it to non secure firmware or OS except by via ARM-TF. Signed-off-by: Graeme Gregory <graeme@nuviainc.com> Reviewed-by: Leif Lindholm <leif@nuviainc.com> Tested-by: Leif Lindholm <leif@nuviainc.com> Message-id: 20200826141952.136164-3-graeme@nuviainc.com Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/sbsa-ref.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/hw/arm/sbsa-ref.c b/hw/arm/sbsa-ref.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/sbsa-ref.c +++ b/hw/arm/sbsa-ref.c @@ -XXX,XX +XXX,XX @@ enum { SBSA_CPUPERIPHS, SBSA_GIC_DIST, SBSA_GIC_REDIST, + SBSA_SECURE_EC, SBSA_SMMU, SBSA_UART, SBSA_RTC, @@ -XXX,XX +XXX,XX @@ static const MemMapEntry sbsa_ref_memmap[] = { [SBSA_CPUPERIPHS] = { 0x40000000, 0x00040000 }, [SBSA_GIC_DIST] = { 0x40060000, 0x00010000 }, [SBSA_GIC_REDIST] = { 0x40080000, 0x04000000 }, + [SBSA_SECURE_EC] = { 0x50000000, 0x00001000 }, [SBSA_UART] = { 0x60000000, 0x00001000 }, [SBSA_RTC] = { 0x60010000, 0x00001000 }, [SBSA_GPIO] = { 0x60020000, 0x00001000 }, @@ -XXX,XX +XXX,XX @@ static void *sbsa_ref_dtb(const struct arm_boot_info *binfo, int *fdt_size) return board->fdt; } +static void create_secure_ec(MemoryRegion *mem) +{ + hwaddr base = sbsa_ref_memmap[SBSA_SECURE_EC].base; + DeviceState *dev = qdev_new("sbsa-ec"); + SysBusDevice *s = SYS_BUS_DEVICE(dev); + + memory_region_add_subregion(mem, base, + sysbus_mmio_get_region(s, 0)); +} + static void sbsa_ref_init(MachineState *machine) { unsigned int smp_cpus = machine->smp.cpus; @@ -XXX,XX +XXX,XX @@ static void sbsa_ref_init(MachineState *machine) create_pcie(sms); + create_secure_ec(secure_sysmem); + sms->bootinfo.ram_size = machine->ram_size; sms->bootinfo.nb_cpus = smp_cpus; sms->bootinfo.board_id = -1; -- 2.20.1 I don't have anything else queued up at the moment, so this is just Richard's SME patches. -- PMM The following changes since commit 63b38f6c85acd312c2cab68554abf33adf4ee2b3: Merge tag 'pull-target-arm-20220707' of https://git.linaro.org/people/pmaydell/qemu-arm into staging (2022-07-08 06:17:11 +0530) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20220711 for you to fetch changes up to f9982ceaf26df27d15547a3a7990a95019e9e3a8: linux-user/aarch64: Add SME related hwcap entries (2022-07-11 13:43:52 +0100) ---------------------------------------------------------------- target-arm: * Implement SME emulation, for both system and linux-user ---------------------------------------------------------------- Richard Henderson (45): target/arm: Handle SME in aarch64_cpu_dump_state target/arm: Add infrastructure for disas_sme target/arm: Trap non-streaming usage when Streaming SVE is active target/arm: Mark ADR as non-streaming target/arm: Mark RDFFR, WRFFR, SETFFR as non-streaming target/arm: Mark BDEP, BEXT, BGRP, COMPACT, FEXPA, FTSSEL as non-streaming target/arm: Mark PMULL, FMMLA as non-streaming target/arm: Mark FTSMUL, FTMAD, FADDA as non-streaming target/arm: Mark SMMLA, UMMLA, USMMLA as non-streaming target/arm: Mark string/histo/crypto as non-streaming target/arm: Mark gather/scatter load/store as non-streaming target/arm: Mark gather prefetch as non-streaming target/arm: Mark LDFF1 and LDNF1 as non-streaming target/arm: Mark LD1RO as non-streaming target/arm: Add SME enablement checks target/arm: Handle SME in sve_access_check target/arm: Implement SME RDSVL, ADDSVL, ADDSPL target/arm: Implement SME ZERO target/arm: Implement SME MOVA target/arm: Implement SME LD1, ST1 target/arm: Export unpredicated ld/st from translate-sve.c target/arm: Implement SME LDR, STR target/arm: Implement SME ADDHA, ADDVA target/arm: Implement FMOPA, FMOPS (non-widening) target/arm: Implement BFMOPA, BFMOPS target/arm: Implement FMOPA, FMOPS (widening) target/arm: Implement SME integer outer product target/arm: Implement PSEL target/arm: Implement REVD target/arm: Implement SCLAMP, UCLAMP target/arm: Reset streaming sve state on exception boundaries target/arm: Enable SME for -cpu max linux-user/aarch64: Clear tpidr2_el0 if CLONE_SETTLS linux-user/aarch64: Reset PSTATE.SM on syscalls linux-user/aarch64: Add SM bit to SVE signal context linux-user/aarch64: Tidy target_restore_sigframe error return linux-user/aarch64: Do not allow duplicate or short sve records linux-user/aarch64: Verify extra record lock succeeded linux-user/aarch64: Move sve record checks into restore linux-user/aarch64: Implement SME signal handling linux-user: Rename sve prctls linux-user/aarch64: Implement PR_SME_GET_VL, PR_SME_SET_VL target/arm: Only set ZEN in reset if SVE present target/arm: Enable SME for user-only linux-user/aarch64: Add SME related hwcap entries docs/system/arm/emulation.rst | 4 + linux-user/aarch64/target_cpu.h | 5 +- linux-user/aarch64/target_prctl.h | 62 +- target/arm/cpu.h | 7 + target/arm/helper-sme.h | 126 ++++ target/arm/helper-sve.h | 4 + target/arm/helper.h | 18 + target/arm/translate-a64.h | 45 ++ target/arm/translate.h | 16 + target/arm/sme-fa64.decode | 60 ++ target/arm/sme.decode | 88 +++ target/arm/sve.decode | 41 +- linux-user/aarch64/cpu_loop.c | 9 + linux-user/aarch64/signal.c | 243 ++++++-- linux-user/elfload.c | 20 + linux-user/syscall.c | 28 +- target/arm/cpu.c | 35 +- target/arm/cpu64.c | 11 + target/arm/helper.c | 56 +- target/arm/sme_helper.c | 1140 +++++++++++++++++++++++++++++++++++++ target/arm/sve_helper.c | 28 + target/arm/translate-a64.c | 103 +++- target/arm/translate-sme.c | 373 ++++++++++++ target/arm/translate-sve.c | 393 ++++++++++--- target/arm/translate-vfp.c | 12 + target/arm/translate.c | 2 + target/arm/vec_helper.c | 24 + target/arm/meson.build | 3 + 28 files changed, 2821 insertions(+), 135 deletions(-) create mode 100644 target/arm/sme-fa64.decode create mode 100644 target/arm/sme.decode create mode 100644 target/arm/translate-sme.c From: Richard Henderson <richard.henderson@linaro.org> Dump SVCR, plus use the correct access check for Streaming Mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-2-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) 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 aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) int i; int el = arm_current_el(env); const char *ns_status; + bool sve; qemu_fprintf(f, " PC=%016" PRIx64 " ", env->pc); for (i = 0; i < 32; i++) { @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) el, psr & PSTATE_SP ? 'h' : 't'); + if (cpu_isar_feature(aa64_sme, cpu)) { + qemu_fprintf(f, " SVCR=%08" PRIx64 " %c%c", + env->svcr, + (FIELD_EX64(env->svcr, SVCR, ZA) ? 'Z' : '-'), + (FIELD_EX64(env->svcr, SVCR, SM) ? 'S' : '-')); + } if (cpu_isar_feature(aa64_bti, cpu)) { qemu_fprintf(f, " BTYPE=%d", (psr & PSTATE_BTYPE) >> 10); } @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) qemu_fprintf(f, " FPCR=%08x FPSR=%08x\n", vfp_get_fpcr(env), vfp_get_fpsr(env)); - if (cpu_isar_feature(aa64_sve, cpu) && sve_exception_el(env, el) == 0) { + if (cpu_isar_feature(aa64_sme, cpu) && FIELD_EX64(env->svcr, SVCR, SM)) { + sve = sme_exception_el(env, el) == 0; + } else if (cpu_isar_feature(aa64_sve, cpu)) { + sve = sve_exception_el(env, el) == 0; + } else { + sve = false; + } + + if (sve) { int j, zcr_len = sve_vqm1_for_el(env, el); for (i = 0; i <= FFR_PRED_NUM; i++) { -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This includes the build rules for the decoder, and the new file for translation, but excludes any instructions. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-3-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 1 + target/arm/sme.decode | 20 ++++++++++++++++++++ target/arm/translate-a64.c | 7 ++++++- target/arm/translate-sme.c | 35 +++++++++++++++++++++++++++++++++++ target/arm/meson.build | 2 ++ 5 files changed, 64 insertions(+), 1 deletion(-) create mode 100644 target/arm/sme.decode create mode 100644 target/arm/translate-sme.c diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int pred_gvec_reg_size(DisasContext *s) } bool disas_sve(DisasContext *, uint32_t); +bool disas_sme(DisasContext *, uint32_t); void gen_gvec_rax1(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, uint32_t rm_ofs, uint32_t opr_sz, uint32_t max_sz); diff --git a/target/arm/sme.decode b/target/arm/sme.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ +# AArch64 SME instruction descriptions +# +# Copyright (c) 2022 Linaro, Ltd +# +# This library is free software; you can redistribute it and/or +# modify it under the terms of the GNU Lesser General Public +# License as published by the Free Software Foundation; either +# version 2.1 of the License, or (at your option) any later version. +# +# This library is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public +# License along with this library; if not, see <http://www.gnu.org/licenses/>. + +# +# This file is processed by scripts/decodetree.py +# 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 void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) } switch (extract32(insn, 25, 4)) { - case 0x0: case 0x1: case 0x3: /* UNALLOCATED */ + case 0x0: + if (!extract32(insn, 31, 1) || !disas_sme(s, insn)) { + unallocated_encoding(s); + } + break; + case 0x1: case 0x3: /* UNALLOCATED */ unallocated_encoding(s); break; case 0x2: diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ +/* + * AArch64 SME translation + * + * Copyright (c) 2022 Linaro, Ltd + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, see <http://www.gnu.org/licenses/>. + */ + +#include "qemu/osdep.h" +#include "cpu.h" +#include "tcg/tcg-op.h" +#include "tcg/tcg-op-gvec.h" +#include "tcg/tcg-gvec-desc.h" +#include "translate.h" +#include "exec/helper-gen.h" +#include "translate-a64.h" +#include "fpu/softfloat.h" + + +/* + * Include the generated decoder. + */ + +#include "decode-sme.c.inc" diff --git a/target/arm/meson.build b/target/arm/meson.build index XXXXXXX..XXXXXXX 100644 --- a/target/arm/meson.build +++ b/target/arm/meson.build @@ -XXX,XX +XXX,XX @@ gen = [ decodetree.process('sve.decode', extra_args: '--decode=disas_sve'), + decodetree.process('sme.decode', extra_args: '--decode=disas_sme'), decodetree.process('neon-shared.decode', extra_args: '--decode=disas_neon_shared'), decodetree.process('neon-dp.decode', extra_args: '--decode=disas_neon_dp'), decodetree.process('neon-ls.decode', extra_args: '--decode=disas_neon_ls'), @@ -XXX,XX +XXX,XX @@ arm_ss.add(when: 'TARGET_AARCH64', if_true: files( 'sme_helper.c', 'translate-a64.c', 'translate-sve.c', + 'translate-sme.c', )) arm_softmmu_ss = ss.source_set() -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This new behaviour is in the ARM pseudocode function AArch64.CheckFPAdvSIMDEnabled, which applies to AArch32 via AArch32.CheckAdvSIMDOrFPEnabled when the EL to which the trap would be delivered is in AArch64 mode. Given that ARMv9 drops support for AArch32 outside EL0, the trap EL detection ought to be trivially true, but the pseudocode still contains a number of conditions, and QEMU has not yet committed to dropping A32 support for EL[12] when v9 features are present. Since the computation of SME_TRAP_NONSTREAMING is necessarily different for the two modes, we might as well preserve bits within TBFLAG_ANY and allocate separate bits within TBFLAG_A32 and TBFLAG_A64 instead. Note that DDI0616A.a has typos for bits [22:21] of LD1RO in the table of instructions illegal in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-4-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 7 +++ target/arm/translate.h | 4 ++ target/arm/sme-fa64.decode | 90 ++++++++++++++++++++++++++++++++++++++ target/arm/helper.c | 41 +++++++++++++++++ target/arm/translate-a64.c | 40 ++++++++++++++++- target/arm/translate-vfp.c | 12 +++++ target/arm/translate.c | 2 + target/arm/meson.build | 1 + 8 files changed, 195 insertions(+), 2 deletions(-) create mode 100644 target/arm/sme-fa64.decode diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ FIELD(TBFLAG_A32, HSTR_ACTIVE, 9, 1) * the same thing as the current security state of the processor! */ FIELD(TBFLAG_A32, NS, 10, 1) +/* + * Indicates that SME Streaming mode is active, and SMCR_ELx.FA64 is not. + * This requires an SME trap from AArch32 mode when using NEON. + */ +FIELD(TBFLAG_A32, SME_TRAP_NONSTREAMING, 11, 1) /* * Bit usage when in AArch32 state, for M-profile only. @@ -XXX,XX +XXX,XX @@ FIELD(TBFLAG_A64, SMEEXC_EL, 20, 2) FIELD(TBFLAG_A64, PSTATE_SM, 22, 1) FIELD(TBFLAG_A64, PSTATE_ZA, 23, 1) FIELD(TBFLAG_A64, SVL, 24, 4) +/* Indicates that SME Streaming mode is active, and SMCR_ELx.FA64 is not. */ +FIELD(TBFLAG_A64, SME_TRAP_NONSTREAMING, 28, 1) /* * Helpers for using the above. diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ typedef struct DisasContext { bool pstate_sm; /* True if PSTATE.ZA is set. */ bool pstate_za; + /* True if non-streaming insns should raise an SME Streaming exception. */ + bool sme_trap_nonstreaming; + /* True if the current instruction is non-streaming. */ + bool is_nonstreaming; /* True if MVE insns are definitely not predicated by VPR or LTPSIZE */ bool mve_no_pred; /* diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ +# AArch64 SME allowed instruction decoding +# +# Copyright (c) 2022 Linaro, Ltd +# +# This library is free software; you can redistribute it and/or +# modify it under the terms of the GNU Lesser General Public +# License as published by the Free Software Foundation; either +# version 2.1 of the License, or (at your option) any later version. +# +# This library is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public +# License along with this library; if not, see <http://www.gnu.org/licenses/>. + +# +# This file is processed by scripts/decodetree.py +# + +# These patterns are taken from Appendix E1.1 of DDI0616 A.a, +# Arm Architecture Reference Manual Supplement, +# The Scalable Matrix Extension (SME), for Armv9-A + +{ + [ + OK 0-00 1110 0000 0001 0010 11-- ---- ---- # SMOV W|Xd,Vn.B[0] + OK 0-00 1110 0000 0010 0010 11-- ---- ---- # SMOV W|Xd,Vn.H[0] + OK 0100 1110 0000 0100 0010 11-- ---- ---- # SMOV Xd,Vn.S[0] + OK 0000 1110 0000 0001 0011 11-- ---- ---- # UMOV Wd,Vn.B[0] + OK 0000 1110 0000 0010 0011 11-- ---- ---- # UMOV Wd,Vn.H[0] + OK 0000 1110 0000 0100 0011 11-- ---- ---- # UMOV Wd,Vn.S[0] + OK 0100 1110 0000 1000 0011 11-- ---- ---- # UMOV Xd,Vn.D[0] + ] + FAIL 0--0 111- ---- ---- ---- ---- ---- ---- # Advanced SIMD vector operations +} + +{ + [ + OK 0101 1110 --1- ---- 11-1 11-- ---- ---- # FMULX/FRECPS/FRSQRTS (scalar) + OK 0101 1110 -10- ---- 00-1 11-- ---- ---- # FMULX/FRECPS/FRSQRTS (scalar, FP16) + OK 01-1 1110 1-10 0001 11-1 10-- ---- ---- # FRECPE/FRSQRTE/FRECPX (scalar) + OK 01-1 1110 1111 1001 11-1 10-- ---- ---- # FRECPE/FRSQRTE/FRECPX (scalar, FP16) + ] + FAIL 01-1 111- ---- ---- ---- ---- ---- ---- # Advanced SIMD single-element operations +} + +FAIL 0-00 110- ---- ---- ---- ---- ---- ---- # Advanced SIMD structure load/store +FAIL 1100 1110 ---- ---- ---- ---- ---- ---- # Advanced SIMD cryptography extensions +FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS + +# These are the "avoidance of doubt" final table of Illegal Advanced SIMD instructions +# We don't actually need to include these, as the default is OK. +# -001 111- ---- ---- ---- ---- ---- ---- # Scalar floating-point operations +# --10 110- ---- ---- ---- ---- ---- ---- # Load/store pair of FP registers +# --01 1100 ---- ---- ---- ---- ---- ---- # Load FP register (PC-relative literal) +# --11 1100 --0- ---- ---- ---- ---- ---- # Load/store FP register (unscaled imm) +# --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) +# --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) + +FAIL 0000 0100 --1- ---- 1010 ---- ---- ---- # ADR +FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA +FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT +FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS +FAIL 0010 0101 --10 1--- 1001 ---- ---- ---- # WRFFR, SETFFR +FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP +FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) +FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA +FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL +FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD +FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA +FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA +FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions +FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) +FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) +FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) +FAIL 1000 010- -01- ---- 1--- ---- ---- ---- # SVE 32-bit gather load (vector+imm) +FAIL 1000 0100 0-0- ---- 0--- ---- ---- ---- # SVE 32-bit gather load byte (scalar+vector) +FAIL 1000 0100 1--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load half (scalar+vector) +FAIL 1000 0101 0--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load word (scalar+vector) +FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) +FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) +FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) +FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) +FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch +FAIL 1110 010- -00- ---- 001- ---- ---- ---- # SVE2 64-bit scatter NT store (vector+scalar) +FAIL 1110 010- -10- ---- 001- ---- ---- ---- # SVE2 32-bit scatter NT store (vector+scalar) +FAIL 1110 010- ---- ---- 1-0- ---- ---- ---- # SVE scatter store (scalar+32-bit vector) +FAIL 1110 010- ---- ---- 101- ---- ---- ---- # SVE scatter store (misc) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ int sme_exception_el(CPUARMState *env, int el) return 0; } +/* This corresponds to the ARM pseudocode function IsFullA64Enabled(). */ +static bool sme_fa64(CPUARMState *env, int el) +{ + if (!cpu_isar_feature(aa64_sme_fa64, env_archcpu(env))) { + return false; + } + + if (el <= 1 && !el_is_in_host(env, el)) { + if (!FIELD_EX64(env->vfp.smcr_el[1], SMCR, FA64)) { + return false; + } + } + if (el <= 2 && arm_is_el2_enabled(env)) { + if (!FIELD_EX64(env->vfp.smcr_el[2], SMCR, FA64)) { + return false; + } + } + if (arm_feature(env, ARM_FEATURE_EL3)) { + if (!FIELD_EX64(env->vfp.smcr_el[3], SMCR, FA64)) { + return false; + } + } + + return true; +} + /* * Given that SVE is enabled, return the vector length for EL. */ @@ -XXX,XX +XXX,XX @@ static CPUARMTBFlags rebuild_hflags_a32(CPUARMState *env, int fp_el, DP_TBFLAG_ANY(flags, PSTATE__IL, 1); } + /* + * The SME exception we are testing for is raised via + * AArch64.CheckFPAdvSIMDEnabled(), as called from + * AArch32.CheckAdvSIMDOrFPEnabled(). + */ + if (el == 0 + && FIELD_EX64(env->svcr, SVCR, SM) + && (!arm_is_el2_enabled(env) + || (arm_el_is_aa64(env, 2) && !(env->cp15.hcr_el2 & HCR_TGE))) + && arm_el_is_aa64(env, 1) + && !sme_fa64(env, el)) { + DP_TBFLAG_A32(flags, SME_TRAP_NONSTREAMING, 1); + } + return rebuild_hflags_common_32(env, fp_el, mmu_idx, flags); } @@ -XXX,XX +XXX,XX @@ static CPUARMTBFlags rebuild_hflags_a64(CPUARMState *env, int el, int fp_el, } if (FIELD_EX64(env->svcr, SVCR, SM)) { DP_TBFLAG_A64(flags, PSTATE_SM, 1); + DP_TBFLAG_A64(flags, SME_TRAP_NONSTREAMING, !sme_fa64(env, el)); } DP_TBFLAG_A64(flags, PSTATE_ZA, FIELD_EX64(env->svcr, SVCR, ZA)); } 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 void do_vec_ld(DisasContext *s, int destidx, int element, * unallocated-encoding checks (otherwise the syndrome information * for the resulting exception will be incorrect). */ -static bool fp_access_check(DisasContext *s) +static bool fp_access_check_only(DisasContext *s) { if (s->fp_excp_el) { assert(!s->fp_access_checked); @@ -XXX,XX +XXX,XX @@ static bool fp_access_check(DisasContext *s) return true; } +static bool fp_access_check(DisasContext *s) +{ + if (!fp_access_check_only(s)) { + return false; + } + if (s->sme_trap_nonstreaming && s->is_nonstreaming) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_Streaming, false)); + return false; + } + return true; +} + /* Check that SVE access is enabled. If it is, return true. * If not, emit code to generate an appropriate exception and return false. */ @@ -XXX,XX +XXX,XX @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread, default: g_assert_not_reached(); } - if ((ri->type & ARM_CP_FPU) && !fp_access_check(s)) { + if ((ri->type & ARM_CP_FPU) && !fp_access_check_only(s)) { return; } else if ((ri->type & ARM_CP_SVE) && !sve_access_check(s)) { return; @@ -XXX,XX +XXX,XX @@ static void disas_data_proc_simd_fp(DisasContext *s, uint32_t insn) } } +/* + * Include the generated SME FA64 decoder. + */ + +#include "decode-sme-fa64.c.inc" + +static bool trans_OK(DisasContext *s, arg_OK *a) +{ + return true; +} + +static bool trans_FAIL(DisasContext *s, arg_OK *a) +{ + s->is_nonstreaming = true; + return true; +} + /** * is_guarded_page: * @env: The cpu environment @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase, dc->mte_active[1] = EX_TBFLAG_A64(tb_flags, MTE0_ACTIVE); dc->pstate_sm = EX_TBFLAG_A64(tb_flags, PSTATE_SM); dc->pstate_za = EX_TBFLAG_A64(tb_flags, PSTATE_ZA); + dc->sme_trap_nonstreaming = EX_TBFLAG_A64(tb_flags, SME_TRAP_NONSTREAMING); dc->vec_len = 0; dc->vec_stride = 0; dc->cp_regs = arm_cpu->cp_regs; @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) } } + s->is_nonstreaming = false; + if (s->sme_trap_nonstreaming) { + disas_sme_fa64(s, insn); + } + switch (extract32(insn, 25, 4)) { case 0x0: if (!extract32(insn, 31, 1) || !disas_sme(s, insn)) { diff --git a/target/arm/translate-vfp.c b/target/arm/translate-vfp.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c +++ b/target/arm/translate-vfp.c @@ -XXX,XX +XXX,XX @@ static bool vfp_access_check_a(DisasContext *s, bool ignore_vfp_enabled) return false; } + /* + * Note that rebuild_hflags_a32 has already accounted for being in EL0 + * and the higher EL in A64 mode, etc. Unlike A64 mode, there do not + * appear to be any insns which touch VFP which are allowed. + */ + if (s->sme_trap_nonstreaming) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_Streaming, + s->base.pc_next - s->pc_curr == 2)); + return false; + } + if (!s->vfp_enabled && !ignore_vfp_enabled) { assert(!arm_dc_feature(s, ARM_FEATURE_M)); unallocated_encoding(s); diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static void arm_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) dc->vec_len = EX_TBFLAG_A32(tb_flags, VECLEN); dc->vec_stride = EX_TBFLAG_A32(tb_flags, VECSTRIDE); } + dc->sme_trap_nonstreaming = + EX_TBFLAG_A32(tb_flags, SME_TRAP_NONSTREAMING); } dc->cp_regs = cpu->cp_regs; dc->features = env->features; diff --git a/target/arm/meson.build b/target/arm/meson.build index XXXXXXX..XXXXXXX 100644 --- a/target/arm/meson.build +++ b/target/arm/meson.build @@ -XXX,XX +XXX,XX @@ gen = [ decodetree.process('sve.decode', extra_args: '--decode=disas_sve'), decodetree.process('sme.decode', extra_args: '--decode=disas_sme'), + decodetree.process('sme-fa64.decode', extra_args: '--static-decode=disas_sme_fa64'), decodetree.process('neon-shared.decode', extra_args: '--decode=disas_neon_shared'), decodetree.process('neon-dp.decode', extra_args: '--decode=disas_neon_dp'), decodetree.process('neon-ls.decode', extra_args: '--decode=disas_neon_ls'), -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark ADR as a non-streaming instruction, which should trap if full a64 support is not enabled in streaming mode. Removing entries from sme-fa64.decode is an easy way to see what remains to be done. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-5-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.h | 7 +++++++ target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 8 ++++---- 3 files changed, 11 insertions(+), 5 deletions(-) diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ uint64_t asimd_imm_const(uint32_t imm, int cmode, int op); static bool trans_##NAME(DisasContext *s, arg_##NAME *a) \ { return dc_isar_feature(FEAT, s) && FUNC(s, __VA_ARGS__); } +#define TRANS_FEAT_NONSTREAMING(NAME, FEAT, FUNC, ...) \ + static bool trans_##NAME(DisasContext *s, arg_##NAME *a) \ + { \ + s->is_nonstreaming = true; \ + return dc_isar_feature(FEAT, s) && FUNC(s, __VA_ARGS__); \ + } + #endif /* TARGET_ARM_TRANSLATE_H */ diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0000 0100 --1- ---- 1010 ---- ---- ---- # ADR FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_adr(DisasContext *s, arg_rrri *a, gen_helper_gvec_3 *fn) return gen_gvec_ool_zzz(s, fn, a->rd, a->rn, a->rm, a->imm); } -TRANS_FEAT(ADR_p32, aa64_sve, do_adr, a, gen_helper_sve_adr_p32) -TRANS_FEAT(ADR_p64, aa64_sve, do_adr, a, gen_helper_sve_adr_p64) -TRANS_FEAT(ADR_s32, aa64_sve, do_adr, a, gen_helper_sve_adr_s32) -TRANS_FEAT(ADR_u32, aa64_sve, do_adr, a, gen_helper_sve_adr_u32) +TRANS_FEAT_NONSTREAMING(ADR_p32, aa64_sve, do_adr, a, gen_helper_sve_adr_p32) +TRANS_FEAT_NONSTREAMING(ADR_p64, aa64_sve, do_adr, a, gen_helper_sve_adr_p64) +TRANS_FEAT_NONSTREAMING(ADR_s32, aa64_sve, do_adr, a, gen_helper_sve_adr_s32) +TRANS_FEAT_NONSTREAMING(ADR_u32, aa64_sve, do_adr, a, gen_helper_sve_adr_u32) /* *** SVE Integer Misc - Unpredicated Group -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-6-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 9 ++++++--- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT -FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS -FAIL 0010 0101 --10 1--- 1001 ---- ---- ---- # WRFFR, SETFFR FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_predset(DisasContext *s, int esz, int rd, int pat, bool setflag) TRANS_FEAT(PTRUE, aa64_sve, do_predset, a->esz, a->rd, a->pat, a->s) /* Note pat == 31 is #all, to set all elements. */ -TRANS_FEAT(SETFFR, aa64_sve, do_predset, 0, FFR_PRED_NUM, 31, false) +TRANS_FEAT_NONSTREAMING(SETFFR, aa64_sve, + do_predset, 0, FFR_PRED_NUM, 31, false) /* Note pat == 32 is #unimp, to set no elements. */ TRANS_FEAT(PFALSE, aa64_sve, do_predset, 0, a->rd, 32, false) @@ -XXX,XX +XXX,XX @@ static bool trans_RDFFR_p(DisasContext *s, arg_RDFFR_p *a) .rd = a->rd, .pg = a->pg, .s = a->s, .rn = FFR_PRED_NUM, .rm = FFR_PRED_NUM, }; + + s->is_nonstreaming = true; return trans_AND_pppp(s, &alt_a); } -TRANS_FEAT(RDFFR, aa64_sve, do_mov_p, a->rd, FFR_PRED_NUM) -TRANS_FEAT(WRFFR, aa64_sve, do_mov_p, FFR_PRED_NUM, a->rn) +TRANS_FEAT_NONSTREAMING(RDFFR, aa64_sve, do_mov_p, a->rd, FFR_PRED_NUM) +TRANS_FEAT_NONSTREAMING(WRFFR, aa64_sve, do_mov_p, FFR_PRED_NUM, a->rn) static bool do_pfirst_pnext(DisasContext *s, arg_rr_esz *a, void (*gen_fn)(TCGv_i32, TCGv_ptr, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-7-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 22 ++++++++++++---------- 2 files changed, 12 insertions(+), 13 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA -FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT -FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_2 * const fexpa_fns[4] = { NULL, gen_helper_sve_fexpa_h, gen_helper_sve_fexpa_s, gen_helper_sve_fexpa_d, }; -TRANS_FEAT(FEXPA, aa64_sve, gen_gvec_ool_zz, - fexpa_fns[a->esz], a->rd, a->rn, 0) +TRANS_FEAT_NONSTREAMING(FEXPA, aa64_sve, gen_gvec_ool_zz, + fexpa_fns[a->esz], a->rd, a->rn, 0) static gen_helper_gvec_3 * const ftssel_fns[4] = { NULL, gen_helper_sve_ftssel_h, gen_helper_sve_ftssel_s, gen_helper_sve_ftssel_d, }; -TRANS_FEAT(FTSSEL, aa64_sve, gen_gvec_ool_arg_zzz, ftssel_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(FTSSEL, aa64_sve, gen_gvec_ool_arg_zzz, + ftssel_fns[a->esz], a, 0) /* *** SVE Predicate Logical Operations Group @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(TRN2_q, aa64_sve_f64mm, gen_gvec_ool_arg_zzz, static gen_helper_gvec_3 * const compact_fns[4] = { NULL, NULL, gen_helper_sve_compact_s, gen_helper_sve_compact_d }; -TRANS_FEAT(COMPACT, aa64_sve, gen_gvec_ool_arg_zpz, compact_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(COMPACT, aa64_sve, gen_gvec_ool_arg_zpz, + compact_fns[a->esz], a, 0) /* Call the helper that computes the ARM LastActiveElement pseudocode * function, scaled by the element size. This includes the not found @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_3 * const bext_fns[4] = { gen_helper_sve2_bext_b, gen_helper_sve2_bext_h, gen_helper_sve2_bext_s, gen_helper_sve2_bext_d, }; -TRANS_FEAT(BEXT, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bext_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BEXT, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bext_fns[a->esz], a, 0) static gen_helper_gvec_3 * const bdep_fns[4] = { gen_helper_sve2_bdep_b, gen_helper_sve2_bdep_h, gen_helper_sve2_bdep_s, gen_helper_sve2_bdep_d, }; -TRANS_FEAT(BDEP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bdep_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BDEP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bdep_fns[a->esz], a, 0) static gen_helper_gvec_3 * const bgrp_fns[4] = { gen_helper_sve2_bgrp_b, gen_helper_sve2_bgrp_h, gen_helper_sve2_bgrp_s, gen_helper_sve2_bgrp_d, }; -TRANS_FEAT(BGRP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bgrp_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BGRP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bgrp_fns[a->esz], a, 0) static gen_helper_gvec_3 * const cadd_fns[4] = { gen_helper_sve2_cadd_b, gen_helper_sve2_cadd_h, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-8-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 24 +++++++++++++++--------- 2 files changed, 15 insertions(+), 11 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) -FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_trans_pmull(DisasContext *s, arg_rrr_esz *a, bool sel) gen_helper_gvec_pmull_q, gen_helper_sve2_pmull_h, NULL, gen_helper_sve2_pmull_d, }; - if (a->esz == 0 - ? !dc_isar_feature(aa64_sve2_pmull128, s) - : !dc_isar_feature(aa64_sve, s)) { + + if (a->esz == 0) { + if (!dc_isar_feature(aa64_sve2_pmull128, s)) { + return false; + } + s->is_nonstreaming = true; + } else if (!dc_isar_feature(aa64_sve, s)) { return false; } return gen_gvec_ool_arg_zzz(s, fns[a->esz], a, sel); @@ -XXX,XX +XXX,XX @@ DO_ZPZZ_FP(FMINP, aa64_sve2, sve2_fminp_zpzz) * SVE Integer Multiply-Add (unpredicated) */ -TRANS_FEAT(FMMLA_s, aa64_sve_f32mm, gen_gvec_fpst_zzzz, gen_helper_fmmla_s, - a->rd, a->rn, a->rm, a->ra, 0, FPST_FPCR) -TRANS_FEAT(FMMLA_d, aa64_sve_f64mm, gen_gvec_fpst_zzzz, gen_helper_fmmla_d, - a->rd, a->rn, a->rm, a->ra, 0, FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FMMLA_s, aa64_sve_f32mm, gen_gvec_fpst_zzzz, + gen_helper_fmmla_s, a->rd, a->rn, a->rm, a->ra, + 0, FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FMMLA_d, aa64_sve_f64mm, gen_gvec_fpst_zzzz, + gen_helper_fmmla_d, a->rd, a->rn, a->rm, a->ra, + 0, FPST_FPCR) static gen_helper_gvec_4 * const sqdmlal_zzzw_fns[] = { NULL, gen_helper_sve2_sqdmlal_zzzw_h, @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(BFDOT_zzzz, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, TRANS_FEAT(BFDOT_zzxz, aa64_sve_bf16, gen_gvec_ool_arg_zzxz, gen_helper_gvec_bfdot_idx, a) -TRANS_FEAT(BFMMLA, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_bfmmla, a, 0) +TRANS_FEAT_NONSTREAMING(BFMMLA, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_bfmmla, a, 0) static bool do_BFMLAL_zzzw(DisasContext *s, arg_rrrr_esz *a, bool sel) { -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-9-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 15 +++++++++++---- 2 files changed, 11 insertions(+), 7 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL -FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD -FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_3_ptr * const ftmad_fns[4] = { NULL, gen_helper_sve_ftmad_h, gen_helper_sve_ftmad_s, gen_helper_sve_ftmad_d, }; -TRANS_FEAT(FTMAD, aa64_sve, gen_gvec_fpst_zzz, - ftmad_fns[a->esz], a->rd, a->rn, a->rm, a->imm, - a->esz == MO_16 ? FPST_FPCR_F16 : FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FTMAD, aa64_sve, gen_gvec_fpst_zzz, + ftmad_fns[a->esz], a->rd, a->rn, a->rm, a->imm, + a->esz == MO_16 ? FPST_FPCR_F16 : FPST_FPCR) /* *** SVE Floating Point Accumulating Reduction Group @@ -XXX,XX +XXX,XX @@ static bool trans_FADDA(DisasContext *s, arg_rprr_esz *a) if (a->esz == 0 || !dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_FADDA(DisasContext *s, arg_rprr_esz *a) DO_FP3(FADD_zzz, fadd) DO_FP3(FSUB_zzz, fsub) DO_FP3(FMUL_zzz, fmul) -DO_FP3(FTSMUL, ftsmul) DO_FP3(FRECPS, recps) DO_FP3(FRSQRTS, rsqrts) #undef DO_FP3 +static gen_helper_gvec_3_ptr * const ftsmul_fns[4] = { + NULL, gen_helper_gvec_ftsmul_h, + gen_helper_gvec_ftsmul_s, gen_helper_gvec_ftsmul_d +}; +TRANS_FEAT_NONSTREAMING(FTSMUL, aa64_sve, gen_gvec_fpst_arg_zzz, + ftsmul_fns[a->esz], a, 0) + /* *** SVE Floating Point Arithmetic - Predicated Group */ -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-10-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 12 ++++++------ 2 files changed, 6 insertions(+), 7 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(FMLALT_zzxw, aa64_sve2, do_FMLAL_zzxw, a, false, true) TRANS_FEAT(FMLSLB_zzxw, aa64_sve2, do_FMLAL_zzxw, a, true, false) TRANS_FEAT(FMLSLT_zzxw, aa64_sve2, do_FMLAL_zzxw, a, true, true) -TRANS_FEAT(SMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_smmla_b, a, 0) -TRANS_FEAT(USMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_usmmla_b, a, 0) -TRANS_FEAT(UMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_ummla_b, a, 0) +TRANS_FEAT_NONSTREAMING(SMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_smmla_b, a, 0) +TRANS_FEAT_NONSTREAMING(USMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_usmmla_b, a, 0) +TRANS_FEAT_NONSTREAMING(UMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_ummla_b, a, 0) TRANS_FEAT(BFDOT_zzzz, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, gen_helper_gvec_bfdot, a, 0) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-11-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 35 ++++++++++++++++++----------------- 2 files changed, 18 insertions(+), 18 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ DO_SVE2_ZZZ_NARROW(RSUBHNT, rsubhnt) static gen_helper_gvec_flags_4 * const match_fns[4] = { gen_helper_sve2_match_ppzz_b, gen_helper_sve2_match_ppzz_h, NULL, NULL }; -TRANS_FEAT(MATCH, aa64_sve2, do_ppzz_flags, a, match_fns[a->esz]) +TRANS_FEAT_NONSTREAMING(MATCH, aa64_sve2, do_ppzz_flags, a, match_fns[a->esz]) static gen_helper_gvec_flags_4 * const nmatch_fns[4] = { gen_helper_sve2_nmatch_ppzz_b, gen_helper_sve2_nmatch_ppzz_h, NULL, NULL }; -TRANS_FEAT(NMATCH, aa64_sve2, do_ppzz_flags, a, nmatch_fns[a->esz]) +TRANS_FEAT_NONSTREAMING(NMATCH, aa64_sve2, do_ppzz_flags, a, nmatch_fns[a->esz]) static gen_helper_gvec_4 * const histcnt_fns[4] = { NULL, NULL, gen_helper_sve2_histcnt_s, gen_helper_sve2_histcnt_d }; -TRANS_FEAT(HISTCNT, aa64_sve2, gen_gvec_ool_arg_zpzz, - histcnt_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(HISTCNT, aa64_sve2, gen_gvec_ool_arg_zpzz, + histcnt_fns[a->esz], a, 0) -TRANS_FEAT(HISTSEG, aa64_sve2, gen_gvec_ool_arg_zzz, - a->esz == 0 ? gen_helper_sve2_histseg : NULL, a, 0) +TRANS_FEAT_NONSTREAMING(HISTSEG, aa64_sve2, gen_gvec_ool_arg_zzz, + a->esz == 0 ? gen_helper_sve2_histseg : NULL, a, 0) DO_ZPZZ_FP(FADDP, aa64_sve2, sve2_faddp_zpzz) DO_ZPZZ_FP(FMAXNMP, aa64_sve2, sve2_fmaxnmp_zpzz) @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(SQRDCMLAH_zzzz, aa64_sve2, gen_gvec_ool_zzzz, TRANS_FEAT(USDOT_zzzz, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, a->esz == 2 ? gen_helper_gvec_usdot_b : NULL, a, 0) -TRANS_FEAT(AESMC, aa64_sve2_aes, gen_gvec_ool_zz, - gen_helper_crypto_aesmc, a->rd, a->rd, a->decrypt) +TRANS_FEAT_NONSTREAMING(AESMC, aa64_sve2_aes, gen_gvec_ool_zz, + gen_helper_crypto_aesmc, a->rd, a->rd, a->decrypt) -TRANS_FEAT(AESE, aa64_sve2_aes, gen_gvec_ool_arg_zzz, - gen_helper_crypto_aese, a, false) -TRANS_FEAT(AESD, aa64_sve2_aes, gen_gvec_ool_arg_zzz, - gen_helper_crypto_aese, a, true) +TRANS_FEAT_NONSTREAMING(AESE, aa64_sve2_aes, gen_gvec_ool_arg_zzz, + gen_helper_crypto_aese, a, false) +TRANS_FEAT_NONSTREAMING(AESD, aa64_sve2_aes, gen_gvec_ool_arg_zzz, + gen_helper_crypto_aese, a, true) -TRANS_FEAT(SM4E, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, - gen_helper_crypto_sm4e, a, 0) -TRANS_FEAT(SM4EKEY, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, - gen_helper_crypto_sm4ekey, a, 0) +TRANS_FEAT_NONSTREAMING(SM4E, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, + gen_helper_crypto_sm4e, a, 0) +TRANS_FEAT_NONSTREAMING(SM4EKEY, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, + gen_helper_crypto_sm4ekey, a, 0) -TRANS_FEAT(RAX1, aa64_sve2_sha3, gen_gvec_fn_arg_zzz, gen_gvec_rax1, a) +TRANS_FEAT_NONSTREAMING(RAX1, aa64_sve2_sha3, gen_gvec_fn_arg_zzz, + gen_gvec_rax1, a) TRANS_FEAT(FCVTNT_sh, aa64_sve2, gen_gvec_fpst_arg_zpz, gen_helper_sve2_fcvtnt_sh, a, 0, FPST_FPCR) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-12-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 9 --------- target/arm/translate-sve.c | 6 ++++++ 2 files changed, 6 insertions(+), 9 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) -FAIL 1000 010- -01- ---- 1--- ---- ---- ---- # SVE 32-bit gather load (vector+imm) -FAIL 1000 0100 0-0- ---- 0--- ---- ---- ---- # SVE 32-bit gather load byte (scalar+vector) -FAIL 1000 0100 1--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load half (scalar+vector) -FAIL 1000 0101 0--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load word (scalar+vector) FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch -FAIL 1110 010- -00- ---- 001- ---- ---- ---- # SVE2 64-bit scatter NT store (vector+scalar) -FAIL 1110 010- -10- ---- 001- ---- ---- ---- # SVE2 32-bit scatter NT store (vector+scalar) -FAIL 1110 010- ---- ---- 1-0- ---- ---- ---- # SVE scatter store (scalar+32-bit vector) -FAIL 1110 010- ---- ---- 101- ---- ---- ---- # SVE scatter store (misc) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LD1_zprz(DisasContext *s, arg_LD1_zprz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LD1_zpiz(DisasContext *s, arg_LD1_zpiz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LDNT1_zprz(DisasContext *s, arg_LD1_zprz *a) if (!dc_isar_feature(aa64_sve2, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_ST1_zprz(DisasContext *s, arg_ST1_zprz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_ST1_zpiz(DisasContext *s, arg_ST1_zpiz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STNT1_zprz(DisasContext *s, arg_ST1_zprz *a) if (!dc_isar_feature(aa64_sve2, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. In this case, introduce PRF_ns (prefetch non-streaming) to handle the checks. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-13-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/sve.decode | 10 +++++----- target/arm/translate-sve.c | 11 +++++++++++ 3 files changed, 16 insertions(+), 8 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) -FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) -FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ LD1RO_zpri 1010010 .. 01 0.... 001 ... ..... ..... \ @rpri_load_msz nreg=0 # SVE 32-bit gather prefetch (scalar plus 32-bit scaled offsets) -PRF 1000010 00 -1 ----- 0-- --- ----- 0 ---- +PRF_ns 1000010 00 -1 ----- 0-- --- ----- 0 ---- # SVE 32-bit gather prefetch (vector plus immediate) -PRF 1000010 -- 00 ----- 111 --- ----- 0 ---- +PRF_ns 1000010 -- 00 ----- 111 --- ----- 0 ---- # SVE contiguous prefetch (scalar plus immediate) PRF 1000010 11 1- ----- 0-- --- ----- 0 ---- @@ -XXX,XX +XXX,XX @@ LD1_zpiz 1100010 .. 01 ..... 1.. ... ..... ..... \ @rpri_g_load esz=3 # SVE 64-bit gather prefetch (scalar plus 64-bit scaled offsets) -PRF 1100010 00 11 ----- 1-- --- ----- 0 ---- +PRF_ns 1100010 00 11 ----- 1-- --- ----- 0 ---- # SVE 64-bit gather prefetch (scalar plus unpacked 32-bit scaled offsets) -PRF 1100010 00 -1 ----- 0-- --- ----- 0 ---- +PRF_ns 1100010 00 -1 ----- 0-- --- ----- 0 ---- # SVE 64-bit gather prefetch (vector plus immediate) -PRF 1100010 -- 00 ----- 111 --- ----- 0 ---- +PRF_ns 1100010 -- 00 ----- 111 --- ----- 0 ---- ### SVE Memory Store Group diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_PRF_rr(DisasContext *s, arg_PRF_rr *a) return true; } +static bool trans_PRF_ns(DisasContext *s, arg_PRF_ns *a) +{ + if (!dc_isar_feature(aa64_sve, s)) { + return false; + } + /* Prefetch is a nop within QEMU. */ + s->is_nonstreaming = true; + (void)sve_access_check(s); + return true; +} + /* * Move Prefix * -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-14-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) -FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LDFF1_zprr(DisasContext *s, arg_rprr_load *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), dtype_msz(a->dtype)); @@ -XXX,XX +XXX,XX @@ static bool trans_LDNF1_zpri(DisasContext *s, arg_rpri_load *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { int vsz = vec_full_reg_size(s); int elements = vsz >> dtype_esz[a->dtype]; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-15-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 2 ++ 2 files changed, 2 insertions(+), 3 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --0- ---- ---- ---- ---- ---- # Load/store FP register (unscaled imm) # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) - -FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) -FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LD1RO_zprr(DisasContext *s, arg_rprr_load *a) if (a->rm == 31) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), dtype_msz(a->dtype)); @@ -XXX,XX +XXX,XX @@ static bool trans_LD1RO_zpri(DisasContext *s, arg_rpri_load *a) if (!dc_isar_feature(aa64_sve_f64mm, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_addi_i64(addr, cpu_reg_sp(s, a->rn), a->imm * 32); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These functions will be used to verify that the cpu is in the correct state for a given instruction. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-16-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 21 +++++++++++++++++++++ target/arm/translate-a64.c | 34 ++++++++++++++++++++++++++++++++++ 2 files changed, 55 insertions(+) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ void write_fp_dreg(DisasContext *s, int reg, TCGv_i64 v); bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn, unsigned int imms, unsigned int immr); bool sve_access_check(DisasContext *s); +bool sme_enabled_check(DisasContext *s); +bool sme_enabled_check_with_svcr(DisasContext *s, unsigned); + +/* This function corresponds to CheckStreamingSVEEnabled. */ +static inline bool sme_sm_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_SM_MASK); +} + +/* This function corresponds to CheckSMEAndZAEnabled. */ +static inline bool sme_za_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_ZA_MASK); +} + +/* Note that this function corresponds to CheckStreamingSVEAndZAEnabled. */ +static inline bool sme_smza_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_SM_MASK | R_SVCR_ZA_MASK); +} + TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr); TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write, bool tag_checked, int log2_size); 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 sme_access_check(DisasContext *s) return true; } +/* This function corresponds to CheckSMEEnabled. */ +bool sme_enabled_check(DisasContext *s) +{ + /* + * Note that unlike sve_excp_el, we have not constrained sme_excp_el + * to be zero when fp_excp_el has priority. This is because we need + * sme_excp_el by itself for cpregs access checks. + */ + if (!s->fp_excp_el || s->sme_excp_el < s->fp_excp_el) { + s->fp_access_checked = true; + return sme_access_check(s); + } + return fp_access_check_only(s); +} + +/* Common subroutine for CheckSMEAnd*Enabled. */ +bool sme_enabled_check_with_svcr(DisasContext *s, unsigned req) +{ + if (!sme_enabled_check(s)) { + return false; + } + if (FIELD_EX64(req, SVCR, SM) && !s->pstate_sm) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_NotStreaming, false)); + return false; + } + if (FIELD_EX64(req, SVCR, ZA) && !s->pstate_za) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_InactiveZA, false)); + return false; + } + return true; +} + /* * This utility function is for doing register extension with an * optional shift. You will likely want to pass a temporary for the -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> The pseudocode for CheckSVEEnabled gains a check for Streaming SVE mode, and for SME present but SVE absent. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-17-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) 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 fp_access_check(DisasContext *s) return true; } -/* Check that SVE access is enabled. If it is, return true. +/* + * Check that SVE access is enabled. If it is, return true. * If not, emit code to generate an appropriate exception and return false. + * This function corresponds to CheckSVEEnabled(). */ bool sve_access_check(DisasContext *s) { - if (s->sve_excp_el) { - assert(!s->sve_access_checked); - s->sve_access_checked = true; - + if (s->pstate_sm || !dc_isar_feature(aa64_sve, s)) { + assert(dc_isar_feature(aa64_sme, s)); + if (!sme_sm_enabled_check(s)) { + goto fail_exit; + } + } else if (s->sve_excp_el) { gen_exception_insn_el(s, s->pc_curr, EXCP_UDEF, syn_sve_access_trap(), s->sve_excp_el); - return false; + goto fail_exit; } s->sve_access_checked = true; return fp_access_check(s); + + fail_exit: + /* Assert that we only raise one exception per instruction. */ + assert(!s->sve_access_checked); + s->sve_access_checked = true; + return false; } /* -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These SME instructions are nominally within the SVE decode space, so we add them to sve.decode and translate-sve.c. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-18-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 12 ++++++++++++ target/arm/sve.decode | 5 ++++- target/arm/translate-sve.c | 38 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 54 insertions(+), 1 deletion(-) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int vec_full_reg_size(DisasContext *s) return s->vl; } +/* Return the byte size of the vector register, SVL / 8. */ +static inline int streaming_vec_reg_size(DisasContext *s) +{ + return s->svl; +} + /* * Return the offset info CPUARMState of the predicate vector register Pn. * Note for this purpose, FFR is P16. @@ -XXX,XX +XXX,XX @@ static inline int pred_full_reg_size(DisasContext *s) return s->vl >> 3; } +/* Return the byte size of the predicate register, SVL / 64. */ +static inline int streaming_pred_reg_size(DisasContext *s) +{ + return s->svl >> 3; +} + /* * Round up the size of a register to a size allowed by * the tcg vector infrastructure. Any operation which uses this diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ INDEX_ri 00000100 esz:2 1 imm:s5 010001 rn:5 rd:5 # SVE index generation (register start, register increment) INDEX_rr 00000100 .. 1 ..... 010011 ..... ..... @rd_rn_rm -### SVE Stack Allocation Group +### SVE / Streaming SVE Stack Allocation Group # SVE stack frame adjustment ADDVL 00000100 001 ..... 01010 ...... ..... @rd_rn_i6 +ADDSVL 00000100 001 ..... 01011 ...... ..... @rd_rn_i6 ADDPL 00000100 011 ..... 01010 ...... ..... @rd_rn_i6 +ADDSPL 00000100 011 ..... 01011 ...... ..... @rd_rn_i6 # SVE stack frame size RDVL 00000100 101 11111 01010 imm:s6 rd:5 +RDSVL 00000100 101 11111 01011 imm:s6 rd:5 ### SVE Bitwise Shift - Unpredicated Group diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_ADDVL(DisasContext *s, arg_ADDVL *a) return true; } +static bool trans_ADDSVL(DisasContext *s, arg_ADDSVL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 rd = cpu_reg_sp(s, a->rd); + TCGv_i64 rn = cpu_reg_sp(s, a->rn); + tcg_gen_addi_i64(rd, rn, a->imm * streaming_vec_reg_size(s)); + } + return true; +} + static bool trans_ADDPL(DisasContext *s, arg_ADDPL *a) { if (!dc_isar_feature(aa64_sve, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_ADDPL(DisasContext *s, arg_ADDPL *a) return true; } +static bool trans_ADDSPL(DisasContext *s, arg_ADDSPL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 rd = cpu_reg_sp(s, a->rd); + TCGv_i64 rn = cpu_reg_sp(s, a->rn); + tcg_gen_addi_i64(rd, rn, a->imm * streaming_pred_reg_size(s)); + } + return true; +} + static bool trans_RDVL(DisasContext *s, arg_RDVL *a) { if (!dc_isar_feature(aa64_sve, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_RDVL(DisasContext *s, arg_RDVL *a) return true; } +static bool trans_RDSVL(DisasContext *s, arg_RDSVL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 reg = cpu_reg(s, a->rd); + tcg_gen_movi_i64(reg, a->imm * streaming_vec_reg_size(s)); + } + return true; +} + /* *** SVE Compute Vector Address Group */ -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-19-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 4 ++++ target/arm/sme_helper.c | 25 +++++++++++++++++++++++++ target/arm/translate-sme.c | 13 +++++++++++++ 4 files changed, 44 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(set_pstate_sm, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_2(set_pstate_za, TCG_CALL_NO_RWG, void, env, i32) + +DEF_HELPER_FLAGS_3(sme_zero, TCG_CALL_NO_RWG, void, env, i32, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ # # This file is processed by scripts/decodetree.py # + +### SME Misc + +ZERO 11000000 00 001 00000000000 imm:8 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void helper_set_pstate_za(CPUARMState *env, uint32_t i) memset(env->zarray, 0, sizeof(env->zarray)); } } + +void helper_sme_zero(CPUARMState *env, uint32_t imm, uint32_t svl) +{ + uint32_t i; + + /* + * Special case clearing the entire ZA space. + * This falls into the CONSTRAINED UNPREDICTABLE zeroing of any + * parts of the ZA storage outside of SVL. + */ + if (imm == 0xff) { + memset(env->zarray, 0, sizeof(env->zarray)); + return; + } + + /* + * Recall that ZAnH.D[m] is spread across ZA[n+8*m], + * so each row is discontiguous within ZA[]. + */ + for (i = 0; i < svl; i++) { + if (imm & (1 << (i % 8))) { + memset(&env->zarray[i], 0, svl); + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ */ #include "decode-sme.c.inc" + + +static bool trans_ZERO(DisasContext *s, arg_ZERO *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_za_enabled_check(s)) { + gen_helper_sme_zero(cpu_env, tcg_constant_i32(a->imm), + tcg_constant_i32(streaming_vec_reg_size(s))); + } + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can reuse the SVE functions for implementing moves to/from horizontal tile slices, but we need new ones for moves to/from vertical tile slices. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-20-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 12 +++ target/arm/helper-sve.h | 2 + target/arm/translate-a64.h | 8 ++ target/arm/translate.h | 5 ++ target/arm/sme.decode | 15 ++++ target/arm/sme_helper.c | 151 ++++++++++++++++++++++++++++++++++++- target/arm/sve_helper.c | 12 +++ target/arm/translate-sme.c | 127 +++++++++++++++++++++++++++++++ 8 files changed, 331 insertions(+), 1 deletion(-) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(set_pstate_sm, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_2(set_pstate_za, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_3(sme_zero, TCG_CALL_NO_RWG, void, env, i32, i32) + +/* Move to/from vertical array slices, i.e. columns, so 'c'. */ +DEF_HELPER_FLAGS_4(sme_mova_cz_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sve.h +++ b/target/arm/helper-sve.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sve_sel_zpzz_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sve_sel_zpzz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sve_sel_zpzz_q, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sve2_addp_zpzz_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int pred_gvec_reg_size(DisasContext *s) return size_for_gvec(pred_full_reg_size(s)); } +/* Return a newly allocated pointer to the predicate register. */ +static inline TCGv_ptr pred_full_reg_ptr(DisasContext *s, int regno) +{ + TCGv_ptr ret = tcg_temp_new_ptr(); + tcg_gen_addi_ptr(ret, cpu_env, pred_full_reg_offset(s, regno)); + return ret; +} + bool disas_sve(DisasContext *, uint32_t); bool disas_sme(DisasContext *, uint32_t); diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ static inline int plus_2(DisasContext *s, int x) return x + 2; } +static inline int plus_12(DisasContext *s, int x) +{ + return x + 12; +} + static inline int times_2(DisasContext *s, int x) { return x * 2; diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ### SME Misc ZERO 11000000 00 001 00000000000 imm:8 + +### SME Move into/from Array + +%mova_rs 13:2 !function=plus_12 +&mova esz rs pg zr za_imm v:bool to_vec:bool + +MOVA 11000000 esz:2 00000 0 v:1 .. pg:3 zr:5 0 za_imm:4 \ + &mova to_vec=0 rs=%mova_rs +MOVA 11000000 11 00000 1 v:1 .. pg:3 zr:5 0 za_imm:4 \ + &mova to_vec=0 rs=%mova_rs esz=4 + +MOVA 11000000 esz:2 00001 0 v:1 .. pg:3 0 za_imm:4 zr:5 \ + &mova to_vec=1 rs=%mova_rs +MOVA 11000000 11 00001 1 v:1 .. pg:3 0 za_imm:4 zr:5 \ + &mova to_vec=1 rs=%mova_rs esz=4 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "cpu.h" -#include "internals.h" +#include "tcg/tcg-gvec-desc.h" #include "exec/helper-proto.h" +#include "qemu/int128.h" +#include "vec_internal.h" /* ResetSVEState */ void arm_reset_sve_state(CPUARMState *env) @@ -XXX,XX +XXX,XX @@ void helper_sme_zero(CPUARMState *env, uint32_t imm, uint32_t svl) } } } + + +/* + * When considering the ZA storage as an array of elements of + * type T, the index within that array of the Nth element of + * a vertical slice of a tile can be calculated like this, + * regardless of the size of type T. This is because the tiles + * are interleaved, so if type T is size N bytes then row 1 of + * the tile is N rows away from row 0. The division by N to + * convert a byte offset into an array index and the multiplication + * by N to convert from vslice-index-within-the-tile to + * the index within the ZA storage cancel out. + */ +#define tile_vslice_index(i) ((i) * sizeof(ARMVectorReg)) + +/* + * When doing byte arithmetic on the ZA storage, the element + * byteoff bytes away in a tile vertical slice is always this + * many bytes away in the ZA storage, regardless of the + * size of the tile element, assuming that byteoff is a multiple + * of the element size. Again this is because of the interleaving + * of the tiles. For instance if we have 1 byte per element then + * each row of the ZA storage has one byte of the vslice data, + * and (counting from 0) byte 8 goes in row 8 of the storage + * at offset (8 * row-size-in-bytes). + * If we have 8 bytes per element then each row of the ZA storage + * has 8 bytes of the data, but there are 8 interleaved tiles and + * so byte 8 of the data goes into row 1 of the tile, + * which is again row 8 of the storage, so the offset is still + * (8 * row-size-in-bytes). Similarly for other element sizes. + */ +#define tile_vslice_offset(byteoff) ((byteoff) * sizeof(ARMVectorReg)) + + +/* + * Move Zreg vector to ZArray column. + */ +#define DO_MOVA_C(NAME, TYPE, H) \ +void HELPER(NAME)(void *za, void *vn, void *vg, uint32_t desc) \ +{ \ + int i, oprsz = simd_oprsz(desc); \ + for (i = 0; i < oprsz; ) { \ + uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ + do { \ + if (pg & 1) { \ + *(TYPE *)(za + tile_vslice_offset(i)) = *(TYPE *)(vn + H(i)); \ + } \ + i += sizeof(TYPE); \ + pg >>= sizeof(TYPE); \ + } while (i & 15); \ + } \ +} + +DO_MOVA_C(sme_mova_cz_b, uint8_t, H1) +DO_MOVA_C(sme_mova_cz_h, uint16_t, H1_2) +DO_MOVA_C(sme_mova_cz_s, uint32_t, H1_4) + +void HELPER(sme_mova_cz_d)(void *za, void *vn, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 8; + uint8_t *pg = vg; + uint64_t *n = vn; + uint64_t *a = za; + + for (i = 0; i < oprsz; i++) { + if (pg[H1(i)] & 1) { + a[tile_vslice_index(i)] = n[i]; + } + } +} + +void HELPER(sme_mova_cz_q)(void *za, void *vn, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 16; + uint16_t *pg = vg; + Int128 *n = vn; + Int128 *a = za; + + /* + * Int128 is used here simply to copy 16 bytes, and to simplify + * the address arithmetic. + */ + for (i = 0; i < oprsz; i++) { + if (pg[H2(i)] & 1) { + a[tile_vslice_index(i)] = n[i]; + } + } +} + +#undef DO_MOVA_C + +/* + * Move ZArray column to Zreg vector. + */ +#define DO_MOVA_Z(NAME, TYPE, H) \ +void HELPER(NAME)(void *vd, void *za, void *vg, uint32_t desc) \ +{ \ + int i, oprsz = simd_oprsz(desc); \ + for (i = 0; i < oprsz; ) { \ + uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ + do { \ + if (pg & 1) { \ + *(TYPE *)(vd + H(i)) = *(TYPE *)(za + tile_vslice_offset(i)); \ + } \ + i += sizeof(TYPE); \ + pg >>= sizeof(TYPE); \ + } while (i & 15); \ + } \ +} + +DO_MOVA_Z(sme_mova_zc_b, uint8_t, H1) +DO_MOVA_Z(sme_mova_zc_h, uint16_t, H1_2) +DO_MOVA_Z(sme_mova_zc_s, uint32_t, H1_4) + +void HELPER(sme_mova_zc_d)(void *vd, void *za, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 8; + uint8_t *pg = vg; + uint64_t *d = vd; + uint64_t *a = za; + + for (i = 0; i < oprsz; i++) { + if (pg[H1(i)] & 1) { + d[i] = a[tile_vslice_index(i)]; + } + } +} + +void HELPER(sme_mova_zc_q)(void *vd, void *za, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 16; + uint16_t *pg = vg; + Int128 *d = vd; + Int128 *a = za; + + /* + * Int128 is used here simply to copy 16 bytes, and to simplify + * the address arithmetic. + */ + for (i = 0; i < oprsz; i++, za += sizeof(ARMVectorReg)) { + if (pg[H2(i)] & 1) { + d[i] = a[tile_vslice_index(i)]; + } + } +} + +#undef DO_MOVA_Z 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 @@ void HELPER(sve_sel_zpzz_d)(void *vd, void *vn, void *vm, } } +void HELPER(sve_sel_zpzz_q)(void *vd, void *vn, void *vm, + void *vg, uint32_t desc) +{ + intptr_t i, opr_sz = simd_oprsz(desc) / 16; + Int128 *d = vd, *n = vn, *m = vm; + uint16_t *pg = vg; + + for (i = 0; i < opr_sz; i += 1) { + d[i] = (pg[H2(i)] & 1 ? n : m)[i]; + } +} + /* Two operand comparison controlled by a predicate. * ??? It is very tempting to want to be able to expand this inline * with x86 instructions, e.g. diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ #include "decode-sme.c.inc" +/* + * Resolve tile.size[index] to a host pointer, where tile and index + * are always decoded together, dependent on the element size. + */ +static TCGv_ptr get_tile_rowcol(DisasContext *s, int esz, int rs, + int tile_index, bool vertical) +{ + int tile = tile_index >> (4 - esz); + int index = esz == MO_128 ? 0 : extract32(tile_index, 0, 4 - esz); + int pos, len, offset; + TCGv_i32 tmp; + TCGv_ptr addr; + + /* Compute the final index, which is Rs+imm. */ + tmp = tcg_temp_new_i32(); + tcg_gen_trunc_tl_i32(tmp, cpu_reg(s, rs)); + tcg_gen_addi_i32(tmp, tmp, index); + + /* Prepare a power-of-two modulo via extraction of @len bits. */ + len = ctz32(streaming_vec_reg_size(s)) - esz; + + if (vertical) { + /* + * Compute the byte offset of the index within the tile: + * (index % (svl / size)) * size + * = (index % (svl >> esz)) << esz + * Perform the power-of-two modulo via extraction of the low @len bits. + * Perform the multiply by shifting left by @pos bits. + * Perform these operations simultaneously via deposit into zero. + */ + pos = esz; + tcg_gen_deposit_z_i32(tmp, tmp, pos, len); + + /* + * For big-endian, adjust the indexed column byte offset within + * the uint64_t host words that make up env->zarray[]. + */ + if (HOST_BIG_ENDIAN && esz < MO_64) { + tcg_gen_xori_i32(tmp, tmp, 8 - (1 << esz)); + } + } else { + /* + * Compute the byte offset of the index within the tile: + * (index % (svl / size)) * (size * sizeof(row)) + * = (index % (svl >> esz)) << (esz + log2(sizeof(row))) + */ + pos = esz + ctz32(sizeof(ARMVectorReg)); + tcg_gen_deposit_z_i32(tmp, tmp, pos, len); + + /* Row slices are always aligned and need no endian adjustment. */ + } + + /* The tile byte offset within env->zarray is the row. */ + offset = tile * sizeof(ARMVectorReg); + + /* Include the byte offset of zarray to make this relative to env. */ + offset += offsetof(CPUARMState, zarray); + tcg_gen_addi_i32(tmp, tmp, offset); + + /* Add the byte offset to env to produce the final pointer. */ + addr = tcg_temp_new_ptr(); + tcg_gen_ext_i32_ptr(addr, tmp); + tcg_temp_free_i32(tmp); + tcg_gen_add_ptr(addr, addr, cpu_env); + + return addr; +} + static bool trans_ZERO(DisasContext *s, arg_ZERO *a) { if (!dc_isar_feature(aa64_sme, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_ZERO(DisasContext *s, arg_ZERO *a) } return true; } + +static bool trans_MOVA(DisasContext *s, arg_MOVA *a) +{ + static gen_helper_gvec_4 * const h_fns[5] = { + gen_helper_sve_sel_zpzz_b, gen_helper_sve_sel_zpzz_h, + gen_helper_sve_sel_zpzz_s, gen_helper_sve_sel_zpzz_d, + gen_helper_sve_sel_zpzz_q + }; + static gen_helper_gvec_3 * const cz_fns[5] = { + gen_helper_sme_mova_cz_b, gen_helper_sme_mova_cz_h, + gen_helper_sme_mova_cz_s, gen_helper_sme_mova_cz_d, + gen_helper_sme_mova_cz_q, + }; + static gen_helper_gvec_3 * const zc_fns[5] = { + gen_helper_sme_mova_zc_b, gen_helper_sme_mova_zc_h, + gen_helper_sme_mova_zc_s, gen_helper_sme_mova_zc_d, + gen_helper_sme_mova_zc_q, + }; + + TCGv_ptr t_za, t_zr, t_pg; + TCGv_i32 t_desc; + int svl; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sme_smza_enabled_check(s)) { + return true; + } + + t_za = get_tile_rowcol(s, a->esz, a->rs, a->za_imm, a->v); + t_zr = vec_full_reg_ptr(s, a->zr); + t_pg = pred_full_reg_ptr(s, a->pg); + + svl = streaming_vec_reg_size(s); + t_desc = tcg_constant_i32(simd_desc(svl, svl, 0)); + + if (a->v) { + /* Vertical slice -- use sme mova helpers. */ + if (a->to_vec) { + zc_fns[a->esz](t_zr, t_za, t_pg, t_desc); + } else { + cz_fns[a->esz](t_za, t_zr, t_pg, t_desc); + } + } else { + /* Horizontal slice -- reuse sve sel helpers. */ + if (a->to_vec) { + h_fns[a->esz](t_zr, t_za, t_zr, t_pg, t_desc); + } else { + h_fns[a->esz](t_za, t_zr, t_za, t_pg, t_desc); + } + } + + tcg_temp_free_ptr(t_za); + tcg_temp_free_ptr(t_zr); + tcg_temp_free_ptr(t_pg); + + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We cannot reuse the SVE functions for LD[1-4] and ST[1-4], because those functions accept only a Zreg register number. For SME, we want to pass a pointer into ZA storage. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-21-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 82 +++++ target/arm/sme.decode | 9 + target/arm/sme_helper.c | 595 +++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 70 +++++ 4 files changed, 756 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(sme_mova_cz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_zc_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_cz_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_zc_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(sme_ld1b_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1h_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1s_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1d_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1q_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1b_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1h_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1s_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1d_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1q_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ MOVA 11000000 esz:2 00001 0 v:1 .. pg:3 0 za_imm:4 zr:5 \ &mova to_vec=1 rs=%mova_rs MOVA 11000000 11 00001 1 v:1 .. pg:3 0 za_imm:4 zr:5 \ &mova to_vec=1 rs=%mova_rs esz=4 + +### SME Memory + +&ldst esz rs pg rn rm za_imm v:bool st:bool + +LDST1 1110000 0 esz:2 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ + &ldst rs=%mova_rs +LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ + &ldst esz=4 rs=%mova_rs diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "cpu.h" +#include "internals.h" #include "tcg/tcg-gvec-desc.h" #include "exec/helper-proto.h" +#include "exec/cpu_ldst.h" +#include "exec/exec-all.h" #include "qemu/int128.h" #include "vec_internal.h" +#include "sve_ldst_internal.h" /* ResetSVEState */ void arm_reset_sve_state(CPUARMState *env) @@ -XXX,XX +XXX,XX @@ void HELPER(sme_mova_zc_q)(void *vd, void *za, void *vg, uint32_t desc) } #undef DO_MOVA_Z + +/* + * Clear elements in a tile slice comprising len bytes. + */ + +typedef void ClearFn(void *ptr, size_t off, size_t len); + +static void clear_horizontal(void *ptr, size_t off, size_t len) +{ + memset(ptr + off, 0, len); +} + +static void clear_vertical_b(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; ++i) { + *(uint8_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_h(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 2) { + *(uint16_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_s(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 4) { + *(uint32_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_d(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 8) { + *(uint64_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_q(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 16) { + memset(vptr + tile_vslice_offset(i + off), 0, 16); + } +} + +/* + * Copy elements from an array into a tile slice comprising len bytes. + */ + +typedef void CopyFn(void *dst, const void *src, size_t len); + +static void copy_horizontal(void *dst, const void *src, size_t len) +{ + memcpy(dst, src, len); +} + +static void copy_vertical_b(void *vdst, const void *vsrc, size_t len) +{ + const uint8_t *src = vsrc; + uint8_t *dst = vdst; + size_t i; + + for (i = 0; i < len; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_h(void *vdst, const void *vsrc, size_t len) +{ + const uint16_t *src = vsrc; + uint16_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 2; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_s(void *vdst, const void *vsrc, size_t len) +{ + const uint32_t *src = vsrc; + uint32_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 4; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_d(void *vdst, const void *vsrc, size_t len) +{ + const uint64_t *src = vsrc; + uint64_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 8; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_q(void *vdst, const void *vsrc, size_t len) +{ + for (size_t i = 0; i < len; i += 16) { + memcpy(vdst + tile_vslice_offset(i), vsrc + i, 16); + } +} + +/* + * Host and TLB primitives for vertical tile slice addressing. + */ + +#define DO_LD(NAME, TYPE, HOST, TLB) \ +static inline void sme_##NAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + TYPE val = HOST(host); \ + *(TYPE *)(za + tile_vslice_offset(off)) = val; \ +} \ +static inline void sme_##NAME##_v_tlb(CPUARMState *env, void *za, \ + intptr_t off, target_ulong addr, uintptr_t ra) \ +{ \ + TYPE val = TLB(env, useronly_clean_ptr(addr), ra); \ + *(TYPE *)(za + tile_vslice_offset(off)) = val; \ +} + +#define DO_ST(NAME, TYPE, HOST, TLB) \ +static inline void sme_##NAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + TYPE val = *(TYPE *)(za + tile_vslice_offset(off)); \ + HOST(host, val); \ +} \ +static inline void sme_##NAME##_v_tlb(CPUARMState *env, void *za, \ + intptr_t off, target_ulong addr, uintptr_t ra) \ +{ \ + TYPE val = *(TYPE *)(za + tile_vslice_offset(off)); \ + TLB(env, useronly_clean_ptr(addr), val, ra); \ +} + +/* + * The ARMVectorReg elements are stored in host-endian 64-bit units. + * For 128-bit quantities, the sequence defined by the Elem[] pseudocode + * corresponds to storing the two 64-bit pieces in little-endian order. + */ +#define DO_LDQ(HNAME, VNAME, BE, HOST, TLB) \ +static inline void HNAME##_host(void *za, intptr_t off, void *host) \ +{ \ + uint64_t val0 = HOST(host), val1 = HOST(host + 8); \ + uint64_t *ptr = za + off; \ + ptr[0] = BE ? val1 : val0, ptr[1] = BE ? val0 : val1; \ +} \ +static inline void VNAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + HNAME##_host(za, tile_vslice_offset(off), host); \ +} \ +static inline void HNAME##_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + uint64_t val0 = TLB(env, useronly_clean_ptr(addr), ra); \ + uint64_t val1 = TLB(env, useronly_clean_ptr(addr + 8), ra); \ + uint64_t *ptr = za + off; \ + ptr[0] = BE ? val1 : val0, ptr[1] = BE ? val0 : val1; \ +} \ +static inline void VNAME##_v_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + HNAME##_tlb(env, za, tile_vslice_offset(off), addr, ra); \ +} + +#define DO_STQ(HNAME, VNAME, BE, HOST, TLB) \ +static inline void HNAME##_host(void *za, intptr_t off, void *host) \ +{ \ + uint64_t *ptr = za + off; \ + HOST(host, ptr[BE]); \ + HOST(host + 1, ptr[!BE]); \ +} \ +static inline void VNAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + HNAME##_host(za, tile_vslice_offset(off), host); \ +} \ +static inline void HNAME##_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + uint64_t *ptr = za + off; \ + TLB(env, useronly_clean_ptr(addr), ptr[BE], ra); \ + TLB(env, useronly_clean_ptr(addr + 8), ptr[!BE], ra); \ +} \ +static inline void VNAME##_v_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + HNAME##_tlb(env, za, tile_vslice_offset(off), addr, ra); \ +} + +DO_LD(ld1b, uint8_t, ldub_p, cpu_ldub_data_ra) +DO_LD(ld1h_be, uint16_t, lduw_be_p, cpu_lduw_be_data_ra) +DO_LD(ld1h_le, uint16_t, lduw_le_p, cpu_lduw_le_data_ra) +DO_LD(ld1s_be, uint32_t, ldl_be_p, cpu_ldl_be_data_ra) +DO_LD(ld1s_le, uint32_t, ldl_le_p, cpu_ldl_le_data_ra) +DO_LD(ld1d_be, uint64_t, ldq_be_p, cpu_ldq_be_data_ra) +DO_LD(ld1d_le, uint64_t, ldq_le_p, cpu_ldq_le_data_ra) + +DO_LDQ(sve_ld1qq_be, sme_ld1q_be, 1, ldq_be_p, cpu_ldq_be_data_ra) +DO_LDQ(sve_ld1qq_le, sme_ld1q_le, 0, ldq_le_p, cpu_ldq_le_data_ra) + +DO_ST(st1b, uint8_t, stb_p, cpu_stb_data_ra) +DO_ST(st1h_be, uint16_t, stw_be_p, cpu_stw_be_data_ra) +DO_ST(st1h_le, uint16_t, stw_le_p, cpu_stw_le_data_ra) +DO_ST(st1s_be, uint32_t, stl_be_p, cpu_stl_be_data_ra) +DO_ST(st1s_le, uint32_t, stl_le_p, cpu_stl_le_data_ra) +DO_ST(st1d_be, uint64_t, stq_be_p, cpu_stq_be_data_ra) +DO_ST(st1d_le, uint64_t, stq_le_p, cpu_stq_le_data_ra) + +DO_STQ(sve_st1qq_be, sme_st1q_be, 1, stq_be_p, cpu_stq_be_data_ra) +DO_STQ(sve_st1qq_le, sme_st1q_le, 0, stq_le_p, cpu_stq_le_data_ra) + +#undef DO_LD +#undef DO_ST +#undef DO_LDQ +#undef DO_STQ + +/* + * Common helper for all contiguous predicated loads. + */ + +static inline QEMU_ALWAYS_INLINE +void sme_ld1(CPUARMState *env, void *za, uint64_t *vg, + const target_ulong addr, uint32_t desc, const uintptr_t ra, + const int esz, uint32_t mtedesc, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn, + ClearFn *clr_fn, + CopyFn *cpy_fn) +{ + const intptr_t reg_max = simd_oprsz(desc); + const intptr_t esize = 1 << esz; + intptr_t reg_off, reg_last; + SVEContLdSt info; + void *host; + int flags; + + /* Find the active elements. */ + if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, esize)) { + /* The entire predicate was false; no load occurs. */ + clr_fn(za, 0, reg_max); + return; + } + + /* Probe the page(s). Exit with exception for any invalid page. */ + sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_LOAD, ra); + + /* Handle watchpoints for all active elements. */ + sve_cont_ldst_watchpoints(&info, env, vg, addr, esize, esize, + BP_MEM_READ, ra); + + /* + * Handle mte checks for all active elements. + * Since TBI must be set for MTE, !mtedesc => !mte_active. + */ + if (mtedesc) { + sve_cont_ldst_mte_check(&info, env, vg, addr, esize, esize, + mtedesc, ra); + } + + flags = info.page[0].flags | info.page[1].flags; + if (unlikely(flags != 0)) { +#ifdef CONFIG_USER_ONLY + g_assert_not_reached(); +#else + /* + * At least one page includes MMIO. + * Any bus operation can fail with cpu_transaction_failed, + * which for ARM will raise SyncExternal. Perform the load + * into scratch memory to preserve register state until the end. + */ + ARMVectorReg scratch = { }; + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[1]; + if (reg_last < 0) { + reg_last = info.reg_off_split; + if (reg_last < 0) { + reg_last = info.reg_off_last[0]; + } + } + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + tlb_fn(env, &scratch, reg_off, addr + reg_off, ra); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + + cpy_fn(za, &scratch, reg_max); + return; +#endif + } + + /* The entire operation is in RAM, on valid pages. */ + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[0]; + host = info.page[0].host; + + if (!vertical) { + memset(za, 0, reg_max); + } else if (reg_off) { + clr_fn(za, 0, reg_off); + } + + while (reg_off <= reg_last) { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } else if (vertical) { + clr_fn(za, reg_off, esize); + } + reg_off += esize; + } while (reg_off <= reg_last && (reg_off & 63)); + } + + /* + * Use the slow path to manage the cross-page misalignment. + * But we know this is RAM and cannot trap. + */ + reg_off = info.reg_off_split; + if (unlikely(reg_off >= 0)) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + + reg_off = info.reg_off_first[1]; + if (unlikely(reg_off >= 0)) { + reg_last = info.reg_off_last[1]; + host = info.page[1].host; + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } else if (vertical) { + clr_fn(za, reg_off, esize); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + } +} + +static inline QEMU_ALWAYS_INLINE +void sme_ld1_mte(CPUARMState *env, void *za, uint64_t *vg, + target_ulong addr, uint32_t desc, uintptr_t ra, + const int esz, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn, + ClearFn *clr_fn, + CopyFn *cpy_fn) +{ + uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + int bit55 = extract64(addr, 55, 1); + + /* Remove mtedesc from the normal sve descriptor. */ + desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + + /* Perform gross MTE suppression early. */ + if (!tbi_check(desc, bit55) || + tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { + mtedesc = 0; + } + + sme_ld1(env, za, vg, addr, desc, ra, esz, mtedesc, vertical, + host_fn, tlb_fn, clr_fn, cpy_fn); +} + +#define DO_LD(L, END, ESZ) \ +void HELPER(sme_ld1##L##END##_h)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1(env, za, vg, addr, desc, GETPC(), ESZ, 0, false, \ + sve_ld1##L##L##END##_host, sve_ld1##L##L##END##_tlb, \ + clear_horizontal, copy_horizontal); \ +} \ +void HELPER(sme_ld1##L##END##_v)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1(env, za, vg, addr, desc, GETPC(), ESZ, 0, true, \ + sme_ld1##L##END##_v_host, sme_ld1##L##END##_v_tlb, \ + clear_vertical_##L, copy_vertical_##L); \ +} \ +void HELPER(sme_ld1##L##END##_h_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1_mte(env, za, vg, addr, desc, GETPC(), ESZ, false, \ + sve_ld1##L##L##END##_host, sve_ld1##L##L##END##_tlb, \ + clear_horizontal, copy_horizontal); \ +} \ +void HELPER(sme_ld1##L##END##_v_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1_mte(env, za, vg, addr, desc, GETPC(), ESZ, true, \ + sme_ld1##L##END##_v_host, sme_ld1##L##END##_v_tlb, \ + clear_vertical_##L, copy_vertical_##L); \ +} + +DO_LD(b, , MO_8) +DO_LD(h, _be, MO_16) +DO_LD(h, _le, MO_16) +DO_LD(s, _be, MO_32) +DO_LD(s, _le, MO_32) +DO_LD(d, _be, MO_64) +DO_LD(d, _le, MO_64) +DO_LD(q, _be, MO_128) +DO_LD(q, _le, MO_128) + +#undef DO_LD + +/* + * Common helper for all contiguous predicated stores. + */ + +static inline QEMU_ALWAYS_INLINE +void sme_st1(CPUARMState *env, void *za, uint64_t *vg, + const target_ulong addr, uint32_t desc, const uintptr_t ra, + const int esz, uint32_t mtedesc, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn) +{ + const intptr_t reg_max = simd_oprsz(desc); + const intptr_t esize = 1 << esz; + intptr_t reg_off, reg_last; + SVEContLdSt info; + void *host; + int flags; + + /* Find the active elements. */ + if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, esize)) { + /* The entire predicate was false; no store occurs. */ + return; + } + + /* Probe the page(s). Exit with exception for any invalid page. */ + sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_STORE, ra); + + /* Handle watchpoints for all active elements. */ + sve_cont_ldst_watchpoints(&info, env, vg, addr, esize, esize, + BP_MEM_WRITE, ra); + + /* + * Handle mte checks for all active elements. + * Since TBI must be set for MTE, !mtedesc => !mte_active. + */ + if (mtedesc) { + sve_cont_ldst_mte_check(&info, env, vg, addr, esize, esize, + mtedesc, ra); + } + + flags = info.page[0].flags | info.page[1].flags; + if (unlikely(flags != 0)) { +#ifdef CONFIG_USER_ONLY + g_assert_not_reached(); +#else + /* + * At least one page includes MMIO. + * Any bus operation can fail with cpu_transaction_failed, + * which for ARM will raise SyncExternal. We cannot avoid + * this fault and will leave with the store incomplete. + */ + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[1]; + if (reg_last < 0) { + reg_last = info.reg_off_split; + if (reg_last < 0) { + reg_last = info.reg_off_last[0]; + } + } + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + return; +#endif + } + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[0]; + host = info.page[0].host; + + while (reg_off <= reg_last) { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } + reg_off += 1 << esz; + } while (reg_off <= reg_last && (reg_off & 63)); + } + + /* + * Use the slow path to manage the cross-page misalignment. + * But we know this is RAM and cannot trap. + */ + reg_off = info.reg_off_split; + if (unlikely(reg_off >= 0)) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + + reg_off = info.reg_off_first[1]; + if (unlikely(reg_off >= 0)) { + reg_last = info.reg_off_last[1]; + host = info.page[1].host; + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } + reg_off += 1 << esz; + } while (reg_off & 63); + } while (reg_off <= reg_last); + } +} + +static inline QEMU_ALWAYS_INLINE +void sme_st1_mte(CPUARMState *env, void *za, uint64_t *vg, target_ulong addr, + uint32_t desc, uintptr_t ra, int esz, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn) +{ + uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + int bit55 = extract64(addr, 55, 1); + + /* Remove mtedesc from the normal sve descriptor. */ + desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + + /* Perform gross MTE suppression early. */ + if (!tbi_check(desc, bit55) || + tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { + mtedesc = 0; + } + + sme_st1(env, za, vg, addr, desc, ra, esz, mtedesc, + vertical, host_fn, tlb_fn); +} + +#define DO_ST(L, END, ESZ) \ +void HELPER(sme_st1##L##END##_h)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1(env, za, vg, addr, desc, GETPC(), ESZ, 0, false, \ + sve_st1##L##L##END##_host, sve_st1##L##L##END##_tlb); \ +} \ +void HELPER(sme_st1##L##END##_v)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1(env, za, vg, addr, desc, GETPC(), ESZ, 0, true, \ + sme_st1##L##END##_v_host, sme_st1##L##END##_v_tlb); \ +} \ +void HELPER(sme_st1##L##END##_h_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1_mte(env, za, vg, addr, desc, GETPC(), ESZ, false, \ + sve_st1##L##L##END##_host, sve_st1##L##L##END##_tlb); \ +} \ +void HELPER(sme_st1##L##END##_v_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1_mte(env, za, vg, addr, desc, GETPC(), ESZ, true, \ + sme_st1##L##END##_v_host, sme_st1##L##END##_v_tlb); \ +} + +DO_ST(b, , MO_8) +DO_ST(h, _be, MO_16) +DO_ST(h, _le, MO_16) +DO_ST(s, _be, MO_32) +DO_ST(s, _le, MO_32) +DO_ST(d, _be, MO_64) +DO_ST(d, _le, MO_64) +DO_ST(q, _be, MO_128) +DO_ST(q, _le, MO_128) + +#undef DO_ST diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool trans_MOVA(DisasContext *s, arg_MOVA *a) return true; } + +static bool trans_LDST1(DisasContext *s, arg_LDST1 *a) +{ + typedef void GenLdSt1(TCGv_env, TCGv_ptr, TCGv_ptr, TCGv, TCGv_i32); + + /* + * Indexed by [esz][be][v][mte][st], which is (except for load/store) + * also the order in which the elements appear in the function names, + * and so how we must concatenate the pieces. + */ + +#define FN_LS(F) { gen_helper_sme_ld1##F, gen_helper_sme_st1##F } +#define FN_MTE(F) { FN_LS(F), FN_LS(F##_mte) } +#define FN_HV(F) { FN_MTE(F##_h), FN_MTE(F##_v) } +#define FN_END(L, B) { FN_HV(L), FN_HV(B) } + + static GenLdSt1 * const fns[5][2][2][2][2] = { + FN_END(b, b), + FN_END(h_le, h_be), + FN_END(s_le, s_be), + FN_END(d_le, d_be), + FN_END(q_le, q_be), + }; + +#undef FN_LS +#undef FN_MTE +#undef FN_HV +#undef FN_END + + TCGv_ptr t_za, t_pg; + TCGv_i64 addr; + int svl, desc = 0; + bool be = s->be_data == MO_BE; + bool mte = s->mte_active[0]; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sme_smza_enabled_check(s)) { + return true; + } + + t_za = get_tile_rowcol(s, a->esz, a->rs, a->za_imm, a->v); + t_pg = pred_full_reg_ptr(s, a->pg); + addr = tcg_temp_new_i64(); + + tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), a->esz); + tcg_gen_add_i64(addr, addr, cpu_reg_sp(s, a->rn)); + + if (mte) { + desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s)); + desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid); + desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma); + desc = FIELD_DP32(desc, MTEDESC, WRITE, a->st); + desc = FIELD_DP32(desc, MTEDESC, SIZEM1, (1 << a->esz) - 1); + desc <<= SVE_MTEDESC_SHIFT; + } else { + addr = clean_data_tbi(s, addr); + } + svl = streaming_vec_reg_size(s); + desc = simd_desc(svl, svl, desc); + + fns[a->esz][be][a->v][mte][a->st](cpu_env, t_za, t_pg, addr, + tcg_constant_i32(desc)); + + tcg_temp_free_ptr(t_za); + tcg_temp_free_ptr(t_pg); + tcg_temp_free_i64(addr); + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Add a TCGv_ptr base argument, which will be cpu_env for SVE. We will reuse this for SME save and restore array insns. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-22-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 3 +++ target/arm/translate-sve.c | 48 ++++++++++++++++++++++++++++---------- 2 files changed, 39 insertions(+), 12 deletions(-) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ void gen_gvec_xar(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, uint32_t rm_ofs, int64_t shift, uint32_t opr_sz, uint32_t max_sz); +void gen_sve_ldr(DisasContext *s, TCGv_ptr, int vofs, int len, int rn, int imm); +void gen_sve_str(DisasContext *s, TCGv_ptr, int vofs, int len, int rn, int imm); + #endif /* TARGET_ARM_TRANSLATE_A64_H */ diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(UCVTF_dd, aa64_sve, gen_gvec_fpst_arg_zpz, * The load should begin at the address Rn + IMM. */ -static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) +void gen_sve_ldr(DisasContext *s, TCGv_ptr base, int vofs, + int len, int rn, int imm) { int len_align = QEMU_ALIGN_DOWN(len, 8); int len_remain = len % 8; @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) t0 = tcg_temp_new_i64(); for (i = 0; i < len_align; i += 8) { tcg_gen_qemu_ld_i64(t0, clean_addr, midx, MO_LEUQ); - tcg_gen_st_i64(t0, cpu_env, vofs + i); + tcg_gen_st_i64(t0, base, vofs + i); tcg_gen_addi_i64(clean_addr, clean_addr, 8); } tcg_temp_free_i64(t0); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) clean_addr = new_tmp_a64_local(s); tcg_gen_mov_i64(clean_addr, t0); + if (base != cpu_env) { + TCGv_ptr b = tcg_temp_local_new_ptr(); + tcg_gen_mov_ptr(b, base); + base = b; + } + gen_set_label(loop); t0 = tcg_temp_new_i64(); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_addi_i64(clean_addr, clean_addr, 8); tp = tcg_temp_new_ptr(); - tcg_gen_add_ptr(tp, cpu_env, i); + tcg_gen_add_ptr(tp, base, i); tcg_gen_addi_ptr(i, i, 8); tcg_gen_st_i64(t0, tp, vofs); tcg_temp_free_ptr(tp); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop); tcg_temp_free_ptr(i); + + if (base != cpu_env) { + tcg_temp_free_ptr(base); + assert(len_remain == 0); + } } /* @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) default: g_assert_not_reached(); } - tcg_gen_st_i64(t0, cpu_env, vofs + len_align); + tcg_gen_st_i64(t0, base, vofs + len_align); tcg_temp_free_i64(t0); } } /* Similarly for stores. */ -static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) +void gen_sve_str(DisasContext *s, TCGv_ptr base, int vofs, + int len, int rn, int imm) { int len_align = QEMU_ALIGN_DOWN(len, 8); int len_remain = len % 8; @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) t0 = tcg_temp_new_i64(); for (i = 0; i < len_align; i += 8) { - tcg_gen_ld_i64(t0, cpu_env, vofs + i); + tcg_gen_ld_i64(t0, base, vofs + i); tcg_gen_qemu_st_i64(t0, clean_addr, midx, MO_LEUQ); tcg_gen_addi_i64(clean_addr, clean_addr, 8); } @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) clean_addr = new_tmp_a64_local(s); tcg_gen_mov_i64(clean_addr, t0); + if (base != cpu_env) { + TCGv_ptr b = tcg_temp_local_new_ptr(); + tcg_gen_mov_ptr(b, base); + base = b; + } + gen_set_label(loop); t0 = tcg_temp_new_i64(); tp = tcg_temp_new_ptr(); - tcg_gen_add_ptr(tp, cpu_env, i); + tcg_gen_add_ptr(tp, base, i); tcg_gen_ld_i64(t0, tp, vofs); tcg_gen_addi_ptr(i, i, 8); tcg_temp_free_ptr(tp); @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop); tcg_temp_free_ptr(i); + + if (base != cpu_env) { + tcg_temp_free_ptr(base); + assert(len_remain == 0); + } } /* Predicate register stores can be any multiple of 2. */ if (len_remain) { t0 = tcg_temp_new_i64(); - tcg_gen_ld_i64(t0, cpu_env, vofs + len_align); + tcg_gen_ld_i64(t0, base, vofs + len_align); switch (len_remain) { case 2: @@ -XXX,XX +XXX,XX @@ static bool trans_LDR_zri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = vec_full_reg_size(s); int off = vec_full_reg_offset(s, a->rd); - do_ldr(s, off, size, a->rn, a->imm * size); + gen_sve_ldr(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LDR_pri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = pred_full_reg_size(s); int off = pred_full_reg_offset(s, a->rd); - do_ldr(s, off, size, a->rn, a->imm * size); + gen_sve_ldr(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STR_zri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = vec_full_reg_size(s); int off = vec_full_reg_offset(s, a->rd); - do_str(s, off, size, a->rn, a->imm * size); + gen_sve_str(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STR_pri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = pred_full_reg_size(s); int off = pred_full_reg_offset(s, a->rd); - do_str(s, off, size, a->rn, a->imm * size); + gen_sve_str(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can reuse the SVE functions for LDR and STR, passing in the base of the ZA vector and a zero offset. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-23-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme.decode | 7 +++++++ target/arm/translate-sme.c | 24 ++++++++++++++++++++++++ 2 files changed, 31 insertions(+) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ LDST1 1110000 0 esz:2 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ &ldst rs=%mova_rs LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ &ldst esz=4 rs=%mova_rs + +&ldstr rv rn imm +@ldstr ....... ... . ...... .. ... rn:5 . imm:4 \ + &ldstr rv=%mova_rs + +LDR 1110000 100 0 000000 .. 000 ..... 0 .... @ldstr +STR 1110000 100 1 000000 .. 000 ..... 0 .... @ldstr diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool trans_LDST1(DisasContext *s, arg_LDST1 *a) tcg_temp_free_i64(addr); return true; } + +typedef void GenLdStR(DisasContext *, TCGv_ptr, int, int, int, int); + +static bool do_ldst_r(DisasContext *s, arg_ldstr *a, GenLdStR *fn) +{ + int svl = streaming_vec_reg_size(s); + int imm = a->imm; + TCGv_ptr base; + + if (!sme_za_enabled_check(s)) { + return true; + } + + /* ZA[n] equates to ZA0H.B[n]. */ + base = get_tile_rowcol(s, MO_8, a->rv, imm, false); + + fn(s, base, 0, svl, a->rn, imm * svl); + + tcg_temp_free_ptr(base); + return true; +} + +TRANS_FEAT(LDR, aa64_sme, do_ldst_r, a, gen_sve_ldr) +TRANS_FEAT(STR, aa64_sme, do_ldst_r, a, gen_sve_str) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-24-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 5 +++ target/arm/sme.decode | 11 +++++ target/arm/sme_helper.c | 90 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 31 +++++++++++++ 4 files changed, 137 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_st1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i DEF_HELPER_FLAGS_5(sme_st1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) DEF_HELPER_FLAGS_5(sme_st1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) DEF_HELPER_FLAGS_5(sme_st1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_addha_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ LDR 1110000 100 0 000000 .. 000 ..... 0 .... @ldstr STR 1110000 100 1 000000 .. 000 ..... 0 .... @ldstr + +### SME Add Vector to Array + +&adda zad zn pm pn +@adda_32 ........ .. ..... . pm:3 pn:3 zn:5 ... zad:2 &adda +@adda_64 ........ .. ..... . pm:3 pn:3 zn:5 .. zad:3 &adda + +ADDHA_s 11000000 10 01000 0 ... ... ..... 000 .. @adda_32 +ADDVA_s 11000000 10 01000 1 ... ... ..... 000 .. @adda_32 +ADDHA_d 11000000 11 01000 0 ... ... ..... 00 ... @adda_64 +ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ DO_ST(q, _be, MO_128) DO_ST(q, _le, MO_128) #undef DO_ST + +void HELPER(sme_addha_s)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 4; + uint64_t *pn = vpn, *pm = vpm; + uint32_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ) { + uint64_t pa = pn[row >> 4]; + do { + if (pa & 1) { + for (col = 0; col < oprsz; ) { + uint64_t pb = pm[col >> 4]; + do { + if (pb & 1) { + zda[tile_vslice_index(row) + H4(col)] += zn[H4(col)]; + } + pb >>= 4; + } while (++col & 15); + } + } + pa >>= 4; + } while (++row & 15); + } +} + +void HELPER(sme_addha_d)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint8_t *pn = vpn, *pm = vpm; + uint64_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + zda[tile_vslice_index(row) + col] += zn[col]; + } + } + } + } +} + +void HELPER(sme_addva_s)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 4; + uint64_t *pn = vpn, *pm = vpm; + uint32_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ) { + uint64_t pa = pn[row >> 4]; + do { + if (pa & 1) { + uint32_t zn_row = zn[H4(row)]; + for (col = 0; col < oprsz; ) { + uint64_t pb = pm[col >> 4]; + do { + if (pb & 1) { + zda[tile_vslice_index(row) + H4(col)] += zn_row; + } + pb >>= 4; + } while (++col & 15); + } + } + pa >>= 4; + } while (++row & 15); + } +} + +void HELPER(sme_addva_d)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint8_t *pn = vpn, *pm = vpm; + uint64_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + uint64_t zn_row = zn[row]; + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + zda[tile_vslice_index(row) + col] += zn_row; + } + } + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool do_ldst_r(DisasContext *s, arg_ldstr *a, GenLdStR *fn) TRANS_FEAT(LDR, aa64_sme, do_ldst_r, a, gen_sve_ldr) TRANS_FEAT(STR, aa64_sme, do_ldst_r, a, gen_sve_str) + +static bool do_adda(DisasContext *s, arg_adda *a, MemOp esz, + gen_helper_gvec_4 *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, 0); + TCGv_ptr za, zn, pn, pm; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + + fn(za, zn, pn, pm, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + return true; +} + +TRANS_FEAT(ADDHA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addha_s) +TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) +TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) +TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-25-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 5 +++ target/arm/sme.decode | 9 +++++ target/arm/sme_helper.c | 69 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 32 ++++++++++++++++++ 4 files changed, 115 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_addha_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ADDHA_s 11000000 10 01000 0 ... ... ..... 000 .. @adda_32 ADDVA_s 11000000 10 01000 1 ... ... ..... 000 .. @adda_32 ADDHA_d 11000000 11 01000 0 ... ... ..... 00 ... @adda_64 ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 + +### SME Outer Product + +&op zad zn zm pm pn sub:bool +@op_32 ........ ... zm:5 pm:3 pn:3 zn:5 sub:1 .. zad:2 &op +@op_64 ........ ... zm:5 pm:3 pn:3 zn:5 sub:1 . zad:3 &op + +FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 +FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "exec/cpu_ldst.h" #include "exec/exec-all.h" #include "qemu/int128.h" +#include "fpu/softfloat.h" #include "vec_internal.h" #include "sve_ldst_internal.h" @@ -XXX,XX +XXX,XX @@ void HELPER(sme_addva_d)(void *vzda, void *vzn, void *vpn, } } } + +void HELPER(sme_fmopa_s)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) << 31; + uint16_t *pn = vpn, *pm = vpm; + float_status fpst; + + /* + * Make a copy of float_status because this operation does not + * update the cumulative fp exception status. It also produces + * default nans. + */ + fpst = *(float_status *)vst; + set_default_nan_mode(true, &fpst); + + for (row = 0; row < oprsz; ) { + uint16_t pa = pn[H2(row >> 4)]; + do { + if (pa & 1) { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)) ^ neg; + + for (col = 0; col < oprsz; ) { + uint16_t pb = pm[H2(col >> 4)]; + do { + if (pb & 1) { + uint32_t *a = vza_row + H1_4(col); + uint32_t *m = vzm + H1_4(col); + *a = float32_muladd(n, *m, *a, 0, vst); + } + col += 4; + pb >>= 4; + } while (col & 15); + } + } + row += 4; + pa >>= 4; + } while (row & 15); + } +} + +void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint64_t neg = (uint64_t)simd_data(desc) << 63; + uint64_t *za = vza, *zn = vzn, *zm = vzm; + uint8_t *pn = vpn, *pm = vpm; + float_status fpst = *(float_status *)vst; + + set_default_nan_mode(true, &fpst); + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + uint64_t *za_row = &za[tile_vslice_index(row)]; + uint64_t n = zn[row] ^ neg; + + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + uint64_t *a = &za_row[col]; + *a = float64_muladd(n, zm[col], *a, 0, &fpst); + } + } + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(ADDHA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addha_s) TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) + +static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, + gen_helper_gvec_5_ptr *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, a->sub); + TCGv_ptr za, zn, zm, pn, pm, fpst; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + zm = vec_full_reg_ptr(s, a->zm); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + fpst = fpstatus_ptr(FPST_FPCR); + + fn(za, zn, zm, pn, pm, fpst, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + tcg_temp_free_ptr(fpst); + return true; +} + +TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) +TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-26-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 2 ++ target/arm/sme_helper.c | 56 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 30 ++++++++++++++++++++ 4 files changed, 90 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 + +BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm, void *vpn, } } } + +/* + * Alter PAIR as needed for controlling predicates being false, + * and for NEG on an enabled row element. + */ +static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t neg) +{ + /* + * The pseudocode uses a conditional negate after the conditional zero. + * It is simpler here to unconditionally negate before conditional zero. + */ + pair ^= neg; + if (!(pg & 1)) { + pair &= 0xffff0000u; + } + if (!(pg & 4)) { + pair &= 0x0000ffffu; + } + return pair; +} + +void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) * 0x80008000u; + uint16_t *pn = vpn, *pm = vpm; + + for (row = 0; row < oprsz; ) { + uint16_t prow = pn[H2(row >> 4)]; + do { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)); + + n = f16mop_adj_pair(n, prow, neg); + + for (col = 0; col < oprsz; ) { + uint16_t pcol = pm[H2(col >> 4)]; + do { + if (prow & pcol & 0b0101) { + uint32_t *a = vza_row + H1_4(col); + uint32_t m = *(uint32_t *)(vzm + H1_4(col)); + + m = f16mop_adj_pair(m, pcol, 0); + *a = bfdotadd(*a, n, m); + + col += 4; + pcol >>= 4; + } + } while (col & 15); + } + row += 4; + prow >>= 4; + } while (row & 15); + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) +static bool do_outprod(DisasContext *s, arg_op *a, MemOp esz, + gen_helper_gvec_5 *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, a->sub); + TCGv_ptr za, zn, zm, pn, pm; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + zm = vec_full_reg_ptr(s, a->zm); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + + fn(za, zn, zm, pn, pm, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + return true; +} + static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, gen_helper_gvec_5_ptr *fn) { @@ -XXX,XX +XXX,XX @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) + +/* TODO: FEAT_EBF16 */ +TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-27-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 1 + target/arm/sme_helper.c | 74 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 1 + 4 files changed, 78 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_7(sme_fmopa_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 +FMOPA_h 10000001 101 ..... ... ... ..... . 00 .. @op_32 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t neg) return pair; } +static float32 f16_dotadd(float32 sum, uint32_t e1, uint32_t e2, + float_status *s_std, float_status *s_odd) +{ + float64 e1r = float16_to_float64(e1 & 0xffff, true, s_std); + float64 e1c = float16_to_float64(e1 >> 16, true, s_std); + float64 e2r = float16_to_float64(e2 & 0xffff, true, s_std); + float64 e2c = float16_to_float64(e2 >> 16, true, s_std); + float64 t64; + float32 t32; + + /* + * The ARM pseudocode function FPDot performs both multiplies + * and the add with a single rounding operation. Emulate this + * by performing the first multiply in round-to-odd, then doing + * the second multiply as fused multiply-add, and rounding to + * float32 all in one step. + */ + t64 = float64_mul(e1r, e2r, s_odd); + t64 = float64r32_muladd(e1c, e2c, t64, 0, s_std); + + /* This conversion is exact, because we've already rounded. */ + t32 = float64_to_float32(t64, s_std); + + /* The final accumulation step is not fused. */ + return float32_add(sum, t32, s_std); +} + +void HELPER(sme_fmopa_h)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) * 0x80008000u; + uint16_t *pn = vpn, *pm = vpm; + float_status fpst_odd, fpst_std; + + /* + * Make a copy of float_status because this operation does not + * update the cumulative fp exception status. It also produces + * default nans. Make a second copy with round-to-odd -- see above. + */ + fpst_std = *(float_status *)vst; + set_default_nan_mode(true, &fpst_std); + fpst_odd = fpst_std; + set_float_rounding_mode(float_round_to_odd, &fpst_odd); + + for (row = 0; row < oprsz; ) { + uint16_t prow = pn[H2(row >> 4)]; + do { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)); + + n = f16mop_adj_pair(n, prow, neg); + + for (col = 0; col < oprsz; ) { + uint16_t pcol = pm[H2(col >> 4)]; + do { + if (prow & pcol & 0b0101) { + uint32_t *a = vza_row + H1_4(col); + uint32_t m = *(uint32_t *)(vzm + H1_4(col)); + + m = f16mop_adj_pair(m, pcol, 0); + *a = f16_dotadd(*a, n, m, &fpst_std, &fpst_odd); + + col += 4; + pcol >>= 4; + } + } while (col & 15); + } + row += 4; + prow >>= 4; + } while (row & 15); + } +} + void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, void *vpm, uint32_t desc) { diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, return true; } +TRANS_FEAT(FMOPA_h, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_h) TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is SMOPA, SUMOPA, USMOPA_s, UMOPA, for both Int8 and Int16. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-28-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 16 ++++++++ target/arm/sme.decode | 10 +++++ target/arm/sme_helper.c | 82 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 10 +++++ 4 files changed, 118 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_smopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_umopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_sumopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_usmopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_smopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_umopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_sumopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_usmopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_h 10000001 101 ..... ... ... ..... . 00 .. @op_32 + +SMOPA_s 1010000 0 10 0 ..... ... ... ..... . 00 .. @op_32 +SUMOPA_s 1010000 0 10 1 ..... ... ... ..... . 00 .. @op_32 +USMOPA_s 1010000 1 10 0 ..... ... ... ..... . 00 .. @op_32 +UMOPA_s 1010000 1 10 1 ..... ... ... ..... . 00 .. @op_32 + +SMOPA_d 1010000 0 11 0 ..... ... ... ..... . 0 ... @op_64 +SUMOPA_d 1010000 0 11 1 ..... ... ... ..... . 0 ... @op_64 +USMOPA_d 1010000 1 11 0 ..... ... ... ..... . 0 ... @op_64 +UMOPA_d 1010000 1 11 1 ..... ... ... ..... . 0 ... @op_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, } while (row & 15); } } + +typedef uint64_t IMOPFn(uint64_t, uint64_t, uint64_t, uint8_t, bool); + +static inline void do_imopa(uint64_t *za, uint64_t *zn, uint64_t *zm, + uint8_t *pn, uint8_t *pm, + uint32_t desc, IMOPFn *fn) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + bool neg = simd_data(desc); + + for (row = 0; row < oprsz; ++row) { + uint8_t pa = pn[H1(row)]; + uint64_t *za_row = &za[tile_vslice_index(row)]; + uint64_t n = zn[row]; + + for (col = 0; col < oprsz; ++col) { + uint8_t pb = pm[H1(col)]; + uint64_t *a = &za_row[col]; + + *a = fn(n, zm[col], *a, pa & pb, neg); + } + } +} + +#define DEF_IMOP_32(NAME, NTYPE, MTYPE) \ +static uint64_t NAME(uint64_t n, uint64_t m, uint64_t a, uint8_t p, bool neg) \ +{ \ + uint32_t sum0 = 0, sum1 = 0; \ + /* Apply P to N as a mask, making the inactive elements 0. */ \ + n &= expand_pred_b(p); \ + sum0 += (NTYPE)(n >> 0) * (MTYPE)(m >> 0); \ + sum0 += (NTYPE)(n >> 8) * (MTYPE)(m >> 8); \ + sum0 += (NTYPE)(n >> 16) * (MTYPE)(m >> 16); \ + sum0 += (NTYPE)(n >> 24) * (MTYPE)(m >> 24); \ + sum1 += (NTYPE)(n >> 32) * (MTYPE)(m >> 32); \ + sum1 += (NTYPE)(n >> 40) * (MTYPE)(m >> 40); \ + sum1 += (NTYPE)(n >> 48) * (MTYPE)(m >> 48); \ + sum1 += (NTYPE)(n >> 56) * (MTYPE)(m >> 56); \ + if (neg) { \ + sum0 = (uint32_t)a - sum0, sum1 = (uint32_t)(a >> 32) - sum1; \ + } else { \ + sum0 = (uint32_t)a + sum0, sum1 = (uint32_t)(a >> 32) + sum1; \ + } \ + return ((uint64_t)sum1 << 32) | sum0; \ +} + +#define DEF_IMOP_64(NAME, NTYPE, MTYPE) \ +static uint64_t NAME(uint64_t n, uint64_t m, uint64_t a, uint8_t p, bool neg) \ +{ \ + uint64_t sum = 0; \ + /* Apply P to N as a mask, making the inactive elements 0. */ \ + n &= expand_pred_h(p); \ + sum += (NTYPE)(n >> 0) * (MTYPE)(m >> 0); \ + sum += (NTYPE)(n >> 16) * (MTYPE)(m >> 16); \ + sum += (NTYPE)(n >> 32) * (MTYPE)(m >> 32); \ + sum += (NTYPE)(n >> 48) * (MTYPE)(m >> 48); \ + return neg ? a - sum : a + sum; \ +} + +DEF_IMOP_32(smopa_s, int8_t, int8_t) +DEF_IMOP_32(umopa_s, uint8_t, uint8_t) +DEF_IMOP_32(sumopa_s, int8_t, uint8_t) +DEF_IMOP_32(usmopa_s, uint8_t, int8_t) + +DEF_IMOP_64(smopa_d, int16_t, int16_t) +DEF_IMOP_64(umopa_d, uint16_t, uint16_t) +DEF_IMOP_64(sumopa_d, int16_t, uint16_t) +DEF_IMOP_64(usmopa_d, uint16_t, int16_t) + +#define DEF_IMOPH(NAME) \ + void HELPER(sme_##NAME)(void *vza, void *vzn, void *vzm, void *vpn, \ + void *vpm, uint32_t desc) \ + { do_imopa(vza, vzn, vzm, vpn, vpm, desc, NAME); } + +DEF_IMOPH(smopa_s) +DEF_IMOPH(umopa_s) +DEF_IMOPH(sumopa_s) +DEF_IMOPH(usmopa_s) +DEF_IMOPH(smopa_d) +DEF_IMOPH(umopa_d) +DEF_IMOPH(sumopa_d) +DEF_IMOPH(usmopa_d) diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_f /* TODO: FEAT_EBF16 */ TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa) + +TRANS_FEAT(SMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_smopa_s) +TRANS_FEAT(UMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_umopa_s) +TRANS_FEAT(SUMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_sumopa_s) +TRANS_FEAT(USMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_usmopa_s) + +TRANS_FEAT(SMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_smopa_d) +TRANS_FEAT(UMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_umopa_d) +TRANS_FEAT(SUMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_sumopa_d) +TRANS_FEAT(USMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_usmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-29-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sve.decode | 20 +++++++++++++ target/arm/translate-sve.c | 57 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 77 insertions(+) diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ BFMLALT_zzxw 01100100 11 1 ..... 0100.1 ..... ..... @rrxr_3a esz=2 ### SVE2 floating-point bfloat16 dot-product (indexed) BFDOT_zzxz 01100100 01 1 ..... 010000 ..... ..... @rrxr_2 esz=2 + +### SVE broadcast predicate element + +&psel esz pd pn pm rv imm +%psel_rv 16:2 !function=plus_12 +%psel_imm_b 22:2 19:2 +%psel_imm_h 22:2 20:1 +%psel_imm_s 22:2 +%psel_imm_d 23:1 +@psel ........ .. . ... .. .. pn:4 . pm:4 . pd:4 \ + &psel rv=%psel_rv + +PSEL 00100101 .. 1 ..1 .. 01 .... 0 .... 0 .... \ + @psel esz=0 imm=%psel_imm_b +PSEL 00100101 .. 1 .10 .. 01 .... 0 .... 0 .... \ + @psel esz=1 imm=%psel_imm_h +PSEL 00100101 .. 1 100 .. 01 .... 0 .... 0 .... \ + @psel esz=2 imm=%psel_imm_s +PSEL 00100101 .1 1 000 .. 01 .... 0 .... 0 .... \ + @psel esz=3 imm=%psel_imm_d diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_BFMLAL_zzxw(DisasContext *s, arg_rrxr_esz *a, bool sel) TRANS_FEAT(BFMLALB_zzxw, aa64_sve_bf16, do_BFMLAL_zzxw, a, false) TRANS_FEAT(BFMLALT_zzxw, aa64_sve_bf16, do_BFMLAL_zzxw, a, true) + +static bool trans_PSEL(DisasContext *s, arg_psel *a) +{ + int vl = vec_full_reg_size(s); + int pl = pred_gvec_reg_size(s); + int elements = vl >> a->esz; + TCGv_i64 tmp, didx, dbit; + TCGv_ptr ptr; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sve_access_check(s)) { + return true; + } + + tmp = tcg_temp_new_i64(); + dbit = tcg_temp_new_i64(); + didx = tcg_temp_new_i64(); + ptr = tcg_temp_new_ptr(); + + /* Compute the predicate element. */ + tcg_gen_addi_i64(tmp, cpu_reg(s, a->rv), a->imm); + if (is_power_of_2(elements)) { + tcg_gen_andi_i64(tmp, tmp, elements - 1); + } else { + tcg_gen_remu_i64(tmp, tmp, tcg_constant_i64(elements)); + } + + /* Extract the predicate byte and bit indices. */ + tcg_gen_shli_i64(tmp, tmp, a->esz); + tcg_gen_andi_i64(dbit, tmp, 7); + tcg_gen_shri_i64(didx, tmp, 3); + if (HOST_BIG_ENDIAN) { + tcg_gen_xori_i64(didx, didx, 7); + } + + /* Load the predicate word. */ + tcg_gen_trunc_i64_ptr(ptr, didx); + tcg_gen_add_ptr(ptr, ptr, cpu_env); + tcg_gen_ld8u_i64(tmp, ptr, pred_full_reg_offset(s, a->pm)); + + /* Extract the predicate bit and replicate to MO_64. */ + tcg_gen_shr_i64(tmp, tmp, dbit); + tcg_gen_andi_i64(tmp, tmp, 1); + tcg_gen_neg_i64(tmp, tmp); + + /* Apply to either copy the source, or write zeros. */ + tcg_gen_gvec_ands(MO_64, pred_full_reg_offset(s, a->pd), + pred_full_reg_offset(s, a->pn), tmp, pl, pl); + + tcg_temp_free_i64(tmp); + tcg_temp_free_i64(dbit); + tcg_temp_free_i64(didx); + tcg_temp_free_ptr(ptr); + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-30-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sve.h | 2 ++ target/arm/sve.decode | 1 + target/arm/sve_helper.c | 16 ++++++++++++++++ target/arm/translate-sve.c | 2 ++ 4 files changed, 21 insertions(+) diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sve.h +++ b/target/arm/helper-sve.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(sve_revh_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_revw_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_revd_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(sve_rbit_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_rbit_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_rbit_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ REVB 00000101 .. 1001 00 100 ... ..... ..... @rd_pg_rn REVH 00000101 .. 1001 01 100 ... ..... ..... @rd_pg_rn REVW 00000101 .. 1001 10 100 ... ..... ..... @rd_pg_rn RBIT 00000101 .. 1001 11 100 ... ..... ..... @rd_pg_rn +REVD 00000101 00 1011 10 100 ... ..... ..... @rd_pg_rn_e0 # SVE vector splice (predicated, destructive) SPLICE 00000101 .. 101 100 100 ... ..... ..... @rdn_pg_rm 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 @@ DO_ZPZ_D(sve_revh_d, uint64_t, hswap64) DO_ZPZ_D(sve_revw_d, uint64_t, wswap64) +void HELPER(sme_revd_q)(void *vd, void *vn, void *vg, uint32_t desc) +{ + intptr_t i, opr_sz = simd_oprsz(desc) / 8; + uint64_t *d = vd, *n = vn; + uint8_t *pg = vg; + + for (i = 0; i < opr_sz; i += 2) { + if (pg[H1(i)] & 1) { + uint64_t n0 = n[i + 0]; + uint64_t n1 = n[i + 1]; + d[i + 0] = n1; + d[i + 1] = n0; + } + } +} + DO_ZPZ(sve_rbit_b, uint8_t, H1, revbit8) DO_ZPZ(sve_rbit_h, uint16_t, H1_2, revbit16) DO_ZPZ(sve_rbit_s, uint32_t, H1_4, revbit32) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(REVH, aa64_sve, gen_gvec_ool_arg_zpz, revh_fns[a->esz], a, 0) TRANS_FEAT(REVW, aa64_sve, gen_gvec_ool_arg_zpz, a->esz == 3 ? gen_helper_sve_revw_d : NULL, a, 0) +TRANS_FEAT(REVD, aa64_sme, gen_gvec_ool_arg_zpz, gen_helper_sme_revd_q, a, 0) + TRANS_FEAT(SPLICE, aa64_sve, gen_gvec_ool_arg_zpzz, gen_helper_sve_splice, a, a->esz) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-31-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.h | 18 +++++++ target/arm/sve.decode | 5 ++ target/arm/translate-sve.c | 102 +++++++++++++++++++++++++++++++++++++ target/arm/vec_helper.c | 24 +++++++++ 4 files changed, 149 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_6(gvec_bfmlal, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_6(gvec_bfmlal_idx, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_b, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_uclamp_b, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + #ifdef TARGET_AARCH64 #include "helper-a64.h" #include "helper-sve.h" diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ PSEL 00100101 .. 1 100 .. 01 .... 0 .... 0 .... \ @psel esz=2 imm=%psel_imm_s PSEL 00100101 .1 1 000 .. 01 .... 0 .... 0 .... \ @psel esz=3 imm=%psel_imm_d + +### SVE clamp + +SCLAMP 01000100 .. 0 ..... 110000 ..... ..... @rda_rn_rm +UCLAMP 01000100 .. 0 ..... 110001 ..... ..... @rda_rn_rm diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_PSEL(DisasContext *s, arg_psel *a) tcg_temp_free_ptr(ptr); return true; } + +static void gen_sclamp_i32(TCGv_i32 d, TCGv_i32 n, TCGv_i32 m, TCGv_i32 a) +{ + tcg_gen_smax_i32(d, a, n); + tcg_gen_smin_i32(d, d, m); +} + +static void gen_sclamp_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m, TCGv_i64 a) +{ + tcg_gen_smax_i64(d, a, n); + tcg_gen_smin_i64(d, d, m); +} + +static void gen_sclamp_vec(unsigned vece, TCGv_vec d, TCGv_vec n, + TCGv_vec m, TCGv_vec a) +{ + tcg_gen_smax_vec(vece, d, a, n); + tcg_gen_smin_vec(vece, d, d, m); +} + +static void gen_sclamp(unsigned vece, uint32_t d, uint32_t n, uint32_t m, + uint32_t a, uint32_t oprsz, uint32_t maxsz) +{ + static const TCGOpcode vecop[] = { + INDEX_op_smin_vec, INDEX_op_smax_vec, 0 + }; + static const GVecGen4 ops[4] = { + { .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_b, + .opt_opc = vecop, + .vece = MO_8 }, + { .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_h, + .opt_opc = vecop, + .vece = MO_16 }, + { .fni4 = gen_sclamp_i32, + .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_s, + .opt_opc = vecop, + .vece = MO_32 }, + { .fni8 = gen_sclamp_i64, + .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_d, + .opt_opc = vecop, + .vece = MO_64, + .prefer_i64 = TCG_TARGET_REG_BITS == 64 } + }; + tcg_gen_gvec_4(d, n, m, a, oprsz, maxsz, &ops[vece]); +} + +TRANS_FEAT(SCLAMP, aa64_sme, gen_gvec_fn_arg_zzzz, gen_sclamp, a) + +static void gen_uclamp_i32(TCGv_i32 d, TCGv_i32 n, TCGv_i32 m, TCGv_i32 a) +{ + tcg_gen_umax_i32(d, a, n); + tcg_gen_umin_i32(d, d, m); +} + +static void gen_uclamp_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m, TCGv_i64 a) +{ + tcg_gen_umax_i64(d, a, n); + tcg_gen_umin_i64(d, d, m); +} + +static void gen_uclamp_vec(unsigned vece, TCGv_vec d, TCGv_vec n, + TCGv_vec m, TCGv_vec a) +{ + tcg_gen_umax_vec(vece, d, a, n); + tcg_gen_umin_vec(vece, d, d, m); +} + +static void gen_uclamp(unsigned vece, uint32_t d, uint32_t n, uint32_t m, + uint32_t a, uint32_t oprsz, uint32_t maxsz) +{ + static const TCGOpcode vecop[] = { + INDEX_op_umin_vec, INDEX_op_umax_vec, 0 + }; + static const GVecGen4 ops[4] = { + { .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_b, + .opt_opc = vecop, + .vece = MO_8 }, + { .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_h, + .opt_opc = vecop, + .vece = MO_16 }, + { .fni4 = gen_uclamp_i32, + .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_s, + .opt_opc = vecop, + .vece = MO_32 }, + { .fni8 = gen_uclamp_i64, + .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_d, + .opt_opc = vecop, + .vece = MO_64, + .prefer_i64 = TCG_TARGET_REG_BITS == 64 } + }; + tcg_gen_gvec_4(d, n, m, a, oprsz, maxsz, &ops[vece]); +} + +TRANS_FEAT(UCLAMP, aa64_sme, gen_gvec_fn_arg_zzzz, gen_uclamp, a) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(gvec_bfmlal_idx)(void *vd, void *vn, void *vm, } clear_tail(d, opr_sz, simd_maxsz(desc)); } + +#define DO_CLAMP(NAME, TYPE) \ +void HELPER(NAME)(void *d, void *n, void *m, void *a, uint32_t desc) \ +{ \ + intptr_t i, opr_sz = simd_oprsz(desc); \ + for (i = 0; i < opr_sz; i += sizeof(TYPE)) { \ + TYPE aa = *(TYPE *)(a + i); \ + TYPE nn = *(TYPE *)(n + i); \ + TYPE mm = *(TYPE *)(m + i); \ + TYPE dd = MIN(MAX(aa, nn), mm); \ + *(TYPE *)(d + i) = dd; \ + } \ + clear_tail(d, opr_sz, simd_maxsz(desc)); \ +} + +DO_CLAMP(gvec_sclamp_b, int8_t) +DO_CLAMP(gvec_sclamp_h, int16_t) +DO_CLAMP(gvec_sclamp_s, int32_t) +DO_CLAMP(gvec_sclamp_d, int64_t) + +DO_CLAMP(gvec_uclamp_b, uint8_t) +DO_CLAMP(gvec_uclamp_h, uint16_t) +DO_CLAMP(gvec_uclamp_s, uint32_t) +DO_CLAMP(gvec_uclamp_d, uint64_t) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can handle both exception entry and exception return by hooking into aarch64_sve_change_el. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-32-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ void aarch64_sve_change_el(CPUARMState *env, int old_el, return; } + old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64; + new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64; + + /* + * Both AArch64.TakeException and AArch64.ExceptionReturn + * invoke ResetSVEState when taking an exception from, or + * returning to, AArch32 state when PSTATE.SM is enabled. + */ + if (old_a64 != new_a64 && FIELD_EX64(env->svcr, SVCR, SM)) { + arm_reset_sve_state(env); + return; + } + /* * DDI0584A.d sec 3.2: "If SVE instructions are disabled or trapped * at ELx, or not available because the EL is in AArch32 state, then @@ -XXX,XX +XXX,XX @@ void aarch64_sve_change_el(CPUARMState *env, int old_el, * we already have the correct register contents when encountering the * vq0->vq0 transition between EL0->EL1. */ - old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64; old_len = (old_a64 && !sve_exception_el(env, old_el) ? sve_vqm1_for_el(env, old_el) : 0); - new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64; new_len = (new_a64 && !sve_exception_el(env, new_el) ? sve_vqm1_for_el(env, new_el) : 0); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Note that SME remains effectively disabled for user-only, because we do not yet set CPACR_EL1.SMEN. This needs to wait until the kernel ABI is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-33-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- docs/system/arm/emulation.rst | 4 ++++ target/arm/cpu64.c | 11 +++++++++++ 2 files changed, 15 insertions(+) diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/system/arm/emulation.rst +++ b/docs/system/arm/emulation.rst @@ -XXX,XX +XXX,XX @@ the following architecture extensions: - FEAT_SHA512 (Advanced SIMD SHA512 instructions) - FEAT_SM3 (Advanced SIMD SM3 instructions) - FEAT_SM4 (Advanced SIMD SM4 instructions) +- FEAT_SME (Scalable Matrix Extension) +- FEAT_SME_FA64 (Full A64 instruction set in Streaming SVE mode) +- FEAT_SME_F64F64 (Double-precision floating-point outer product instructions) +- FEAT_SME_I16I64 (16-bit to 64-bit integer widening outer product instructions) - FEAT_SPECRES (Speculation restriction instructions) - FEAT_SSBS (Speculative Store Bypass Safe) - FEAT_TLBIOS (TLB invalidate instructions in Outer Shareable domain) diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu64.c +++ b/target/arm/cpu64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) */ t = FIELD_DP64(t, ID_AA64PFR1, MTE, 3); /* FEAT_MTE3 */ t = FIELD_DP64(t, ID_AA64PFR1, RAS_FRAC, 0); /* FEAT_RASv1p1 + FEAT_DoubleFault */ + t = FIELD_DP64(t, ID_AA64PFR1, SME, 1); /* FEAT_SME */ t = FIELD_DP64(t, ID_AA64PFR1, CSV2_FRAC, 0); /* FEAT_CSV2_2 */ cpu->isar.id_aa64pfr1 = t; @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) t = FIELD_DP64(t, ID_AA64DFR0, PMUVER, 5); /* FEAT_PMUv3p4 */ cpu->isar.id_aa64dfr0 = t; + t = cpu->isar.id_aa64smfr0; + t = FIELD_DP64(t, ID_AA64SMFR0, F32F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, B16F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, F16F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, I8I32, 0xf); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, F64F64, 1); /* FEAT_SME_F64F64 */ + t = FIELD_DP64(t, ID_AA64SMFR0, I16I64, 0xf); /* FEAT_SME_I16I64 */ + t = FIELD_DP64(t, ID_AA64SMFR0, FA64, 1); /* FEAT_SME_FA64 */ + cpu->isar.id_aa64smfr0 = t; + /* Replicate the same data to the 32-bit id registers. */ aa32_max_features(cpu); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-34-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_cpu.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/target_cpu.h b/linux-user/aarch64/target_cpu.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_cpu.h +++ b/linux-user/aarch64/target_cpu.h @@ -XXX,XX +XXX,XX @@ static inline void cpu_clone_regs_parent(CPUARMState *env, unsigned flags) static inline void cpu_set_tls(CPUARMState *env, target_ulong newtls) { - /* Note that AArch64 Linux keeps the TLS pointer in TPIDR; this is + /* + * Note that AArch64 Linux keeps the TLS pointer in TPIDR; this is * different from AArch32 Linux, which uses TPIDRRO. */ env->cp15.tpidr_el[0] = newtls; + /* TPIDR2_EL0 is cleared with CLONE_SETTLS. */ + env->cp15.tpidr2_el0 = 0; } static inline abi_ulong get_sp_from_cpustate(CPUARMState *state) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-35-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/cpu_loop.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/linux-user/aarch64/cpu_loop.c b/linux-user/aarch64/cpu_loop.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/cpu_loop.c +++ b/linux-user/aarch64/cpu_loop.c @@ -XXX,XX +XXX,XX @@ void cpu_loop(CPUARMState *env) switch (trapnr) { case EXCP_SWI: + /* + * On syscall, PSTATE.ZA is preserved, along with the ZA matrix. + * PSTATE.SM is cleared, per SMSTOP, which does ResetSVEState. + */ + if (FIELD_EX64(env->svcr, SVCR, SM)) { + env->svcr = FIELD_DP64(env->svcr, SVCR, SM, 0); + arm_rebuild_hflags(env); + arm_reset_sve_state(env); + } ret = do_syscall(env, env->xregs[8], env->xregs[0], -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Make sure to zero the currently reserved fields. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-36-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ struct target_extra_context { struct target_sve_context { struct target_aarch64_ctx head; uint16_t vl; - uint16_t reserved[3]; + uint16_t flags; + uint16_t reserved[2]; /* The actual SVE data immediately follows. It is laid out * according to TARGET_SVE_SIG_{Z,P}REG_OFFSET, based off of * the original struct pointer. @@ -XXX,XX +XXX,XX @@ struct target_sve_context { #define TARGET_SVE_SIG_CONTEXT_SIZE(VQ) \ (TARGET_SVE_SIG_PREG_OFFSET(VQ, 17)) +#define TARGET_SVE_SIG_FLAG_SM 1 + struct target_rt_sigframe { struct target_siginfo info; struct target_ucontext uc; @@ -XXX,XX +XXX,XX @@ static void target_setup_sve_record(struct target_sve_context *sve, { int i, j; + memset(sve, 0, sizeof(*sve)); __put_user(TARGET_SVE_MAGIC, &sve->head.magic); __put_user(size, &sve->head.size); __put_user(vq * TARGET_SVE_VQ_BYTES, &sve->vl); + if (FIELD_EX64(env->svcr, SVCR, SM)) { + __put_user(TARGET_SVE_SIG_FLAG_SM, &sve->flags); + } /* Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian store -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Fold the return value setting into the goto, so each point of failure need not do both. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-37-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 26 +++++++++++--------------- 1 file changed, 11 insertions(+), 15 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, struct target_sve_context *sve = NULL; uint64_t extra_datap = 0; bool used_extra = false; - bool err = false; int vq = 0, sve_size = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, switch (magic) { case 0: if (size != 0) { - err = true; - goto exit; + goto err; } if (used_extra) { ctx = NULL; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, case TARGET_FPSIMD_MAGIC: if (fpsimd || size != sizeof(struct target_fpsimd_context)) { - err = true; - goto exit; + goto err; } fpsimd = (struct target_fpsimd_context *)ctx; break; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, break; } } - err = true; - goto exit; + goto err; case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { - err = true; - goto exit; + goto err; } __get_user(extra_datap, &((struct target_extra_context *)ctx)->datap); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, /* Unknown record -- we certainly didn't generate it. * Did we in fact get out of sync? */ - err = true; - goto exit; + goto err; } ctx = (void *)ctx + size; } @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, if (fpsimd) { target_restore_fpsimd_record(env, fpsimd); } else { - err = true; + goto err; } /* SVE data, if present, overwrites FPSIMD data. */ if (sve) { target_restore_sve_record(env, sve, vq); } - - exit: unlock_user(extra, extra_datap, 0); - return err; + return 0; + + err: + unlock_user(extra, extra_datap, 0); + return 1; } static abi_ulong get_sigframe(struct target_sigaction *ka, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> In parse_user_sigframe, the kernel rejects duplicate sve records, or records that are smaller than the header. We were silently allowing these cases to pass, dropping the record. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-38-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, break; case TARGET_SVE_MAGIC: + if (sve || size < sizeof(struct target_sve_context)) { + goto err; + } if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { vq = sve_vq(env); sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); - if (!sve && size == sve_size) { + if (size == sve_size) { sve = (struct target_sve_context *)ctx; break; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-39-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, __get_user(extra_size, &((struct target_extra_context *)ctx)->size); extra = lock_user(VERIFY_READ, extra_datap, extra_size, 0); + if (!extra) { + return 1; + } break; default: -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Move the checks out of the parsing loop and into the restore function. This more closely mirrors the code structure in the kernel, and is slightly clearer. Reject rather than silently skip incorrect VL and SVE record sizes, bringing our checks in to line with those the kernel does. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-40-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 51 +++++++++++++++++++++++++------------ 1 file changed, 35 insertions(+), 16 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static void target_restore_fpsimd_record(CPUARMState *env, } } -static void target_restore_sve_record(CPUARMState *env, - struct target_sve_context *sve, int vq) +static bool target_restore_sve_record(CPUARMState *env, + struct target_sve_context *sve, + int size) { - int i, j; + int i, j, vl, vq; - /* Note that SVE regs are stored as a byte stream, with each byte element + if (!cpu_isar_feature(aa64_sve, env_archcpu(env))) { + return false; + } + + __get_user(vl, &sve->vl); + vq = sve_vq(env); + + /* Reject mismatched VL. */ + if (vl != vq * TARGET_SVE_VQ_BYTES) { + return false; + } + + /* Accept empty record -- used to clear PSTATE.SM. */ + if (size <= sizeof(*sve)) { + return true; + } + + /* Reject non-empty but incomplete record. */ + if (size < TARGET_SVE_SIG_CONTEXT_SIZE(vq)) { + return false; + } + + /* + * Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian load * of our 64-bit hunks. */ @@ -XXX,XX +XXX,XX @@ static void target_restore_sve_record(CPUARMState *env, } } } + return true; } static int target_restore_sigframe(CPUARMState *env, @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, struct target_sve_context *sve = NULL; uint64_t extra_datap = 0; bool used_extra = false; - int vq = 0, sve_size = 0; + int sve_size = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, if (sve || size < sizeof(struct target_sve_context)) { goto err; } - if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { - vq = sve_vq(env); - sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); - if (size == sve_size) { - sve = (struct target_sve_context *)ctx; - break; - } - } - goto err; + sve = (struct target_sve_context *)ctx; + sve_size = size; + break; case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, } /* SVE data, if present, overwrites FPSIMD data. */ - if (sve) { - target_restore_sve_record(env, sve, vq); + if (sve && !target_restore_sve_record(env, sve, sve_size)) { + goto err; } unlock_user(extra, extra_datap, 0); return 0; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Set the SM bit in the SVE record on signal delivery, create the ZA record. Restore SM and ZA state according to the records present on return. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-41-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 167 +++++++++++++++++++++++++++++++++--- 1 file changed, 154 insertions(+), 13 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ struct target_sve_context { #define TARGET_SVE_SIG_FLAG_SM 1 +#define TARGET_ZA_MAGIC 0x54366345 + +struct target_za_context { + struct target_aarch64_ctx head; + uint16_t vl; + uint16_t reserved[3]; + /* The actual ZA data immediately follows. */ +}; + +#define TARGET_ZA_SIG_REGS_OFFSET \ + QEMU_ALIGN_UP(sizeof(struct target_za_context), TARGET_SVE_VQ_BYTES) +#define TARGET_ZA_SIG_ZAV_OFFSET(VQ, N) \ + (TARGET_ZA_SIG_REGS_OFFSET + (VQ) * TARGET_SVE_VQ_BYTES * (N)) +#define TARGET_ZA_SIG_CONTEXT_SIZE(VQ) \ + TARGET_ZA_SIG_ZAV_OFFSET(VQ, VQ * TARGET_SVE_VQ_BYTES) + struct target_rt_sigframe { struct target_siginfo info; struct target_ucontext uc; @@ -XXX,XX +XXX,XX @@ static void target_setup_end_record(struct target_aarch64_ctx *end) } static void target_setup_sve_record(struct target_sve_context *sve, - CPUARMState *env, int vq, int size) + CPUARMState *env, int size) { - int i, j; + int i, j, vq = sve_vq(env); memset(sve, 0, sizeof(*sve)); __put_user(TARGET_SVE_MAGIC, &sve->head.magic); @@ -XXX,XX +XXX,XX @@ static void target_setup_sve_record(struct target_sve_context *sve, } } +static void target_setup_za_record(struct target_za_context *za, + CPUARMState *env, int size) +{ + int vq = sme_vq(env); + int vl = vq * TARGET_SVE_VQ_BYTES; + int i, j; + + memset(za, 0, sizeof(*za)); + __put_user(TARGET_ZA_MAGIC, &za->head.magic); + __put_user(size, &za->head.size); + __put_user(vl, &za->vl); + + if (size == TARGET_ZA_SIG_CONTEXT_SIZE(0)) { + return; + } + assert(size == TARGET_ZA_SIG_CONTEXT_SIZE(vq)); + + /* + * Note that ZA vectors are stored as a byte stream, + * with each byte element at a subsequent address. + */ + for (i = 0; i < vl; ++i) { + uint64_t *z = (void *)za + TARGET_ZA_SIG_ZAV_OFFSET(vq, i); + for (j = 0; j < vq * 2; ++j) { + __put_user_e(env->zarray[i].d[j], z + j, le); + } + } +} + static void target_restore_general_frame(CPUARMState *env, struct target_rt_sigframe *sf) { @@ -XXX,XX +XXX,XX @@ static void target_restore_fpsimd_record(CPUARMState *env, static bool target_restore_sve_record(CPUARMState *env, struct target_sve_context *sve, - int size) + int size, int *svcr) { - int i, j, vl, vq; + int i, j, vl, vq, flags; + bool sm; - if (!cpu_isar_feature(aa64_sve, env_archcpu(env))) { + __get_user(vl, &sve->vl); + __get_user(flags, &sve->flags); + + sm = flags & TARGET_SVE_SIG_FLAG_SM; + + /* The cpu must support Streaming or Non-streaming SVE. */ + if (sm + ? !cpu_isar_feature(aa64_sme, env_archcpu(env)) + : !cpu_isar_feature(aa64_sve, env_archcpu(env))) { return false; } - __get_user(vl, &sve->vl); - vq = sve_vq(env); + /* + * Note that we cannot use sve_vq() because that depends on the + * current setting of PSTATE.SM, not the state to be restored. + */ + vq = sve_vqm1_for_el_sm(env, 0, sm) + 1; /* Reject mismatched VL. */ if (vl != vq * TARGET_SVE_VQ_BYTES) { @@ -XXX,XX +XXX,XX @@ static bool target_restore_sve_record(CPUARMState *env, return false; } + *svcr = FIELD_DP64(*svcr, SVCR, SM, sm); + /* * Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian load @@ -XXX,XX +XXX,XX @@ static bool target_restore_sve_record(CPUARMState *env, return true; } +static bool target_restore_za_record(CPUARMState *env, + struct target_za_context *za, + int size, int *svcr) +{ + int i, j, vl, vq; + + if (!cpu_isar_feature(aa64_sme, env_archcpu(env))) { + return false; + } + + __get_user(vl, &za->vl); + vq = sme_vq(env); + + /* Reject mismatched VL. */ + if (vl != vq * TARGET_SVE_VQ_BYTES) { + return false; + } + + /* Accept empty record -- used to clear PSTATE.ZA. */ + if (size <= TARGET_ZA_SIG_CONTEXT_SIZE(0)) { + return true; + } + + /* Reject non-empty but incomplete record. */ + if (size < TARGET_ZA_SIG_CONTEXT_SIZE(vq)) { + return false; + } + + *svcr = FIELD_DP64(*svcr, SVCR, ZA, 1); + + for (i = 0; i < vl; ++i) { + uint64_t *z = (void *)za + TARGET_ZA_SIG_ZAV_OFFSET(vq, i); + for (j = 0; j < vq * 2; ++j) { + __get_user_e(env->zarray[i].d[j], z + j, le); + } + } + return true; +} + static int target_restore_sigframe(CPUARMState *env, struct target_rt_sigframe *sf) { struct target_aarch64_ctx *ctx, *extra = NULL; struct target_fpsimd_context *fpsimd = NULL; struct target_sve_context *sve = NULL; + struct target_za_context *za = NULL; uint64_t extra_datap = 0; bool used_extra = false; int sve_size = 0; + int za_size = 0; + int svcr = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, sve_size = size; break; + case TARGET_ZA_MAGIC: + if (za || size < sizeof(struct target_za_context)) { + goto err; + } + za = (struct target_za_context *)ctx; + za_size = size; + break; + case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { goto err; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, } /* SVE data, if present, overwrites FPSIMD data. */ - if (sve && !target_restore_sve_record(env, sve, sve_size)) { + if (sve && !target_restore_sve_record(env, sve, sve_size, &svcr)) { goto err; } + if (za && !target_restore_za_record(env, za, za_size, &svcr)) { + goto err; + } + if (env->svcr != svcr) { + env->svcr = svcr; + arm_rebuild_hflags(env); + } unlock_user(extra, extra_datap, 0); return 0; @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, .total_size = offsetof(struct target_rt_sigframe, uc.tuc_mcontext.__reserved), }; - int fpsimd_ofs, fr_ofs, sve_ofs = 0, vq = 0, sve_size = 0; + int fpsimd_ofs, fr_ofs, sve_ofs = 0, za_ofs = 0; + int sve_size = 0, za_size = 0; struct target_rt_sigframe *frame; struct target_rt_frame_record *fr; abi_ulong frame_addr, return_addr; @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, &layout); /* SVE state needs saving only if it exists. */ - if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { - vq = sve_vq(env); - sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); + if (cpu_isar_feature(aa64_sve, env_archcpu(env)) || + cpu_isar_feature(aa64_sme, env_archcpu(env))) { + sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(sve_vq(env)), 16); sve_ofs = alloc_sigframe_space(sve_size, &layout); } + if (cpu_isar_feature(aa64_sme, env_archcpu(env))) { + /* ZA state needs saving only if it is enabled. */ + if (FIELD_EX64(env->svcr, SVCR, ZA)) { + za_size = TARGET_ZA_SIG_CONTEXT_SIZE(sme_vq(env)); + } else { + za_size = TARGET_ZA_SIG_CONTEXT_SIZE(0); + } + za_ofs = alloc_sigframe_space(za_size, &layout); + } if (layout.extra_ofs) { /* Reserve space for the extra end marker. The standard end marker @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, target_setup_end_record((void *)frame + layout.extra_end_ofs); } if (sve_ofs) { - target_setup_sve_record((void *)frame + sve_ofs, env, vq, sve_size); + target_setup_sve_record((void *)frame + sve_ofs, env, sve_size); + } + if (za_ofs) { + target_setup_za_record((void *)frame + za_ofs, env, za_size); } /* Set up the stack frame for unwinding. */ @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, env->btype = 2; } + /* + * Invoke the signal handler with both SM and ZA disabled. + * When clearing SM, ResetSVEState, per SMSTOP. + */ + if (FIELD_EX64(env->svcr, SVCR, SM)) { + arm_reset_sve_state(env); + } + if (env->svcr) { + env->svcr = 0; + arm_rebuild_hflags(env); + } + if (info) { tswap_siginfo(&frame->info, info); env->xregs[1] = frame_addr + offsetof(struct target_rt_sigframe, info); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Add "sve" to the sve prctl functions, to distinguish them from the coming "sme" prctls with similar names. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-42-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_prctl.h | 8 ++++---- linux-user/syscall.c | 12 ++++++------ 2 files changed, 10 insertions(+), 10 deletions(-) diff --git a/linux-user/aarch64/target_prctl.h b/linux-user/aarch64/target_prctl.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_prctl.h +++ b/linux-user/aarch64/target_prctl.h @@ -XXX,XX +XXX,XX @@ #ifndef AARCH64_TARGET_PRCTL_H #define AARCH64_TARGET_PRCTL_H -static abi_long do_prctl_get_vl(CPUArchState *env) +static abi_long do_prctl_sve_get_vl(CPUArchState *env) { ARMCPU *cpu = env_archcpu(env); if (cpu_isar_feature(aa64_sve, cpu)) { @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_get_vl(CPUArchState *env) } return -TARGET_EINVAL; } -#define do_prctl_get_vl do_prctl_get_vl +#define do_prctl_sve_get_vl do_prctl_sve_get_vl -static abi_long do_prctl_set_vl(CPUArchState *env, abi_long arg2) +static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) { /* * We cannot support either PR_SVE_SET_VL_ONEXEC or PR_SVE_VL_INHERIT. @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_set_vl(CPUArchState *env, abi_long arg2) } return -TARGET_EINVAL; } -#define do_prctl_set_vl do_prctl_set_vl +#define do_prctl_sve_set_vl do_prctl_sve_set_vl static abi_long do_prctl_reset_keys(CPUArchState *env, abi_long arg2) { diff --git a/linux-user/syscall.c b/linux-user/syscall.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_inval1(CPUArchState *env, abi_long arg2) #ifndef do_prctl_set_fp_mode #define do_prctl_set_fp_mode do_prctl_inval1 #endif -#ifndef do_prctl_get_vl -#define do_prctl_get_vl do_prctl_inval0 +#ifndef do_prctl_sve_get_vl +#define do_prctl_sve_get_vl do_prctl_inval0 #endif -#ifndef do_prctl_set_vl -#define do_prctl_set_vl do_prctl_inval1 +#ifndef do_prctl_sve_set_vl +#define do_prctl_sve_set_vl do_prctl_inval1 #endif #ifndef do_prctl_reset_keys #define do_prctl_reset_keys do_prctl_inval1 @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, case PR_SET_FP_MODE: return do_prctl_set_fp_mode(env, arg2); case PR_SVE_GET_VL: - return do_prctl_get_vl(env); + return do_prctl_sve_get_vl(env); case PR_SVE_SET_VL: - return do_prctl_set_vl(env, arg2); + return do_prctl_sve_set_vl(env, arg2); case PR_PAC_RESET_KEYS: if (arg3 || arg4 || arg5) { return -TARGET_EINVAL; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These prctl set the Streaming SVE vector length, which may be completely different from the Normal SVE vector length. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-43-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_prctl.h | 54 +++++++++++++++++++++++++++++++ linux-user/syscall.c | 16 +++++++++ 2 files changed, 70 insertions(+) diff --git a/linux-user/aarch64/target_prctl.h b/linux-user/aarch64/target_prctl.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_prctl.h +++ b/linux-user/aarch64/target_prctl.h @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_get_vl(CPUArchState *env) { ARMCPU *cpu = env_archcpu(env); if (cpu_isar_feature(aa64_sve, cpu)) { + /* PSTATE.SM is always unset on syscall entry. */ return sve_vq(env) * 16; } return -TARGET_EINVAL; @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) && arg2 >= 0 && arg2 <= 512 * 16 && !(arg2 & 15)) { uint32_t vq, old_vq; + /* PSTATE.SM is always unset on syscall entry. */ old_vq = sve_vq(env); /* @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) } #define do_prctl_sve_set_vl do_prctl_sve_set_vl +static abi_long do_prctl_sme_get_vl(CPUArchState *env) +{ + ARMCPU *cpu = env_archcpu(env); + if (cpu_isar_feature(aa64_sme, cpu)) { + return sme_vq(env) * 16; + } + return -TARGET_EINVAL; +} +#define do_prctl_sme_get_vl do_prctl_sme_get_vl + +static abi_long do_prctl_sme_set_vl(CPUArchState *env, abi_long arg2) +{ + /* + * We cannot support either PR_SME_SET_VL_ONEXEC or PR_SME_VL_INHERIT. + * Note the kernel definition of sve_vl_valid allows for VQ=512, + * i.e. VL=8192, even though the architectural maximum is VQ=16. + */ + if (cpu_isar_feature(aa64_sme, env_archcpu(env)) + && arg2 >= 0 && arg2 <= 512 * 16 && !(arg2 & 15)) { + int vq, old_vq; + + old_vq = sme_vq(env); + + /* + * Bound the value of vq, so that we know that it fits into + * the 4-bit field in SMCR_EL1. Because PSTATE.SM is cleared + * on syscall entry, we are not modifying the current SVE + * vector length. + */ + vq = MAX(arg2 / 16, 1); + vq = MIN(vq, 16); + env->vfp.smcr_el[1] = + FIELD_DP64(env->vfp.smcr_el[1], SMCR, LEN, vq - 1); + + /* Delay rebuilding hflags until we know if ZA must change. */ + vq = sve_vqm1_for_el_sm(env, 0, true) + 1; + + if (vq != old_vq) { + /* + * PSTATE.ZA state is cleared on any change to SVL. + * We need not call arm_rebuild_hflags because PSTATE.SM was + * cleared on syscall entry, so this hasn't changed VL. + */ + env->svcr = FIELD_DP64(env->svcr, SVCR, ZA, 0); + arm_rebuild_hflags(env); + } + return vq * 16; + } + return -TARGET_EINVAL; +} +#define do_prctl_sme_set_vl do_prctl_sme_set_vl + static abi_long do_prctl_reset_keys(CPUArchState *env, abi_long arg2) { ARMCPU *cpu = env_archcpu(env); diff --git a/linux-user/syscall.c b/linux-user/syscall.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -XXX,XX +XXX,XX @@ abi_long do_arch_prctl(CPUX86State *env, int code, abi_ulong addr) #ifndef PR_SET_SYSCALL_USER_DISPATCH # define PR_SET_SYSCALL_USER_DISPATCH 59 #endif +#ifndef PR_SME_SET_VL +# define PR_SME_SET_VL 63 +# define PR_SME_GET_VL 64 +# define PR_SME_VL_LEN_MASK 0xffff +# define PR_SME_VL_INHERIT (1 << 17) +#endif #include "target_prctl.h" @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_inval1(CPUArchState *env, abi_long arg2) #ifndef do_prctl_set_unalign #define do_prctl_set_unalign do_prctl_inval1 #endif +#ifndef do_prctl_sme_get_vl +#define do_prctl_sme_get_vl do_prctl_inval0 +#endif +#ifndef do_prctl_sme_set_vl +#define do_prctl_sme_set_vl do_prctl_inval1 +#endif static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, abi_long arg3, abi_long arg4, abi_long arg5) @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, return do_prctl_sve_get_vl(env); case PR_SVE_SET_VL: return do_prctl_sve_set_vl(env, arg2); + case PR_SME_GET_VL: + return do_prctl_sme_get_vl(env); + case PR_SME_SET_VL: + return do_prctl_sme_set_vl(env, arg2); case PR_PAC_RESET_KEYS: if (arg3 || arg4 || arg5) { return -TARGET_EINVAL; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> There's no reason to set CPACR_EL1.ZEN if SVE disabled. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-44-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) 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_reset(DeviceState *dev) /* and to the FP/Neon instructions */ env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, CPACR_EL1, FPEN, 3); - /* and to the SVE instructions */ - env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, - CPACR_EL1, ZEN, 3); - /* with reasonable vector length */ + /* and to the SVE instructions, with default vector length */ if (cpu_isar_feature(aa64_sve, cpu)) { + env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, + CPACR_EL1, ZEN, 3); env->vfp.zcr_el[1] = cpu->sve_default_vq - 1; } /* -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Enable SME, TPIDR2_EL0, and FA64 if supported by the cpu. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-45-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 11 +++++++++++ 1 file changed, 11 insertions(+) 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_reset(DeviceState *dev) CPACR_EL1, ZEN, 3); env->vfp.zcr_el[1] = cpu->sve_default_vq - 1; } + /* and for SME instructions, with default vector length, and TPIDR2 */ + if (cpu_isar_feature(aa64_sme, cpu)) { + env->cp15.sctlr_el[1] |= SCTLR_EnTP2; + env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, + CPACR_EL1, SMEN, 3); + env->vfp.smcr_el[1] = cpu->sme_default_vq - 1; + if (cpu_isar_feature(aa64_sme_fa64, cpu)) { + env->vfp.smcr_el[1] = FIELD_DP64(env->vfp.smcr_el[1], + SMCR, FA64, 1); + } + } /* * Enable 48-bit address space (TODO: take reserved_va into account). * Enable TBI0 but not TBI1. -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-46-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/elfload.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/linux-user/elfload.c b/linux-user/elfload.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/elfload.c +++ b/linux-user/elfload.c @@ -XXX,XX +XXX,XX @@ enum { ARM_HWCAP2_A64_RNG = 1 << 16, ARM_HWCAP2_A64_BTI = 1 << 17, ARM_HWCAP2_A64_MTE = 1 << 18, + ARM_HWCAP2_A64_ECV = 1 << 19, + ARM_HWCAP2_A64_AFP = 1 << 20, + ARM_HWCAP2_A64_RPRES = 1 << 21, + ARM_HWCAP2_A64_MTE3 = 1 << 22, + ARM_HWCAP2_A64_SME = 1 << 23, + ARM_HWCAP2_A64_SME_I16I64 = 1 << 24, + ARM_HWCAP2_A64_SME_F64F64 = 1 << 25, + ARM_HWCAP2_A64_SME_I8I32 = 1 << 26, + ARM_HWCAP2_A64_SME_F16F32 = 1 << 27, + ARM_HWCAP2_A64_SME_B16F32 = 1 << 28, + ARM_HWCAP2_A64_SME_F32F32 = 1 << 29, + ARM_HWCAP2_A64_SME_FA64 = 1 << 30, }; #define ELF_HWCAP get_elf_hwcap() @@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap2(void) GET_FEATURE_ID(aa64_rndr, ARM_HWCAP2_A64_RNG); GET_FEATURE_ID(aa64_bti, ARM_HWCAP2_A64_BTI); GET_FEATURE_ID(aa64_mte, ARM_HWCAP2_A64_MTE); + GET_FEATURE_ID(aa64_sme, (ARM_HWCAP2_A64_SME | + ARM_HWCAP2_A64_SME_F32F32 | + ARM_HWCAP2_A64_SME_B16F32 | + ARM_HWCAP2_A64_SME_F16F32 | + ARM_HWCAP2_A64_SME_I8I32)); + GET_FEATURE_ID(aa64_sme_f64f64, ARM_HWCAP2_A64_SME_F64F64); + GET_FEATURE_ID(aa64_sme_i16i64, ARM_HWCAP2_A64_SME_I16I64); + GET_FEATURE_ID(aa64_sme_fa64, ARM_HWCAP2_A64_SME_FA64); return hwcaps; } -- 2.25.1