yuzu/unicorn
1
0
Fork 0
mirror of https://github.com/yuzu-emu/unicorn.git synced 2025-03-24 22:15:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

target/arm: Implement ARMv8.2-DotProd

Browse source 
We've already added the helpers with an SVE patch, all that remains
is to wire up the aa64 and aa32 translators. Enable the feature
within -cpu max for CONFIG_USER_ONLY.

Backports commit 26c470a7bb4233454137de1062341ad48947f252 from qemu
This commit is contained in:
Richard Henderson 2018-07-03 04:53:29 -04:00 committed by Lioncash
parent d4a8f1bfcb
commit a325de6685
No known key found for this signature in database
GPG key ID: 4E3C3CC1031BA9C7
5 changed files with 94 additions and 21 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
qemu/target/arm/cpu.c
View file

@ -1551,6 +1551,7 @@ static void arm_max_initfn(struct uc_struct *uc, Object *obj, void *opaque)
set_feature(&cpu->env, ARM_FEATURE_V8_PMULL); set_feature(&cpu->env, ARM_FEATURE_V8_PMULL);
set_feature(&cpu->env, ARM_FEATURE_CRC); set_feature(&cpu->env, ARM_FEATURE_CRC);
set_feature(&cpu->env, ARM_FEATURE_V8_RDM); set_feature(&cpu->env, ARM_FEATURE_V8_RDM);
set_feature(&cpu->env, ARM_FEATURE_V8_DOTPROD);
set_feature(&cpu->env, ARM_FEATURE_V8_FCMA); set_feature(&cpu->env, ARM_FEATURE_V8_FCMA);
} }
#endif #endif

1
qemu/target/arm/cpu.h
View file

@ -1420,6 +1420,7 @@ enum arm_features {
ARM_FEATURE_V8_SM4, /* implements SM4 part of v8 Crypto Extensions */ ARM_FEATURE_V8_SM4, /* implements SM4 part of v8 Crypto Extensions */
ARM_FEATURE_V8_ATOMICS, /* ARMv8.1-Atomics feature */ ARM_FEATURE_V8_ATOMICS, /* ARMv8.1-Atomics feature */
ARM_FEATURE_V8_RDM, /* implements v8.1 simd round multiply */ ARM_FEATURE_V8_RDM, /* implements v8.1 simd round multiply */
ARM_FEATURE_V8_DOTPROD, /* implements v8.2 simd dot product */
ARM_FEATURE_V8_FP16, /* implements v8.2 half-precision float */ ARM_FEATURE_V8_FP16, /* implements v8.2 half-precision float */
ARM_FEATURE_V8_FCMA, /* has complex number part of v8.3 extensions. */ ARM_FEATURE_V8_FCMA, /* has complex number part of v8.3 extensions. */
ARM_FEATURE_M_MAIN, /* M profile Main Extension */ ARM_FEATURE_M_MAIN, /* M profile Main Extension */

1
qemu/target/arm/cpu64.c
View file

@ -223,6 +223,7 @@ static void aarch64_max_initfn(struct uc_struct *uc, Object *obj, void *opaque)
set_feature(&cpu->env, ARM_FEATURE_CRC); set_feature(&cpu->env, ARM_FEATURE_CRC);
set_feature(&cpu->env, ARM_FEATURE_V8_ATOMICS); set_feature(&cpu->env, ARM_FEATURE_V8_ATOMICS);
set_feature(&cpu->env, ARM_FEATURE_V8_RDM); set_feature(&cpu->env, ARM_FEATURE_V8_RDM);
set_feature(&cpu->env, ARM_FEATURE_V8_DOTPROD);
set_feature(&cpu->env, ARM_FEATURE_V8_FP16); set_feature(&cpu->env, ARM_FEATURE_V8_FP16);
set_feature(&cpu->env, ARM_FEATURE_V8_FCMA); set_feature(&cpu->env, ARM_FEATURE_V8_FCMA);
set_feature(&cpu->env, ARM_FEATURE_SVE); set_feature(&cpu->env, ARM_FEATURE_SVE);

38
qemu/target/arm/translate-a64.c
View file

@ -684,6 +684,18 @@ static void gen_gvec_op3(DisasContext *s, bool is_q, int rd,
vec_full_reg_size(s), gvec_op); vec_full_reg_size(s), gvec_op);
} }
/* Expand a 3-operand operation using an out-of-line helper. */
static void gen_gvec_op3_ool(DisasContext *s, bool is_q, int rd,
int rn, int rm, int data, gen_helper_gvec_3 *fn)
{
TCGContext *tcg_ctx = s->uc->tcg_ctx;
tcg_gen_gvec_3_ool(tcg_ctx, vec_full_reg_offset(s, rd),
vec_full_reg_offset(s, rn),
vec_full_reg_offset(s, rm),
is_q ? 16 : 8, vec_full_reg_size(s), data, fn);
}
/* Expand a 3-operand + env pointer operation using /* Expand a 3-operand + env pointer operation using
* an out-of-line helper. * an out-of-line helper.
*/ */
@ -11487,6 +11499,14 @@ static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn)
} }
feature = ARM_FEATURE_V8_RDM; feature = ARM_FEATURE_V8_RDM;
break; break;
case 0x02: /* SDOT (vector) */
case 0x12: /* UDOT (vector) */
if (size != MO_32) {
unallocated_encoding(s);
return;
}
feature = ARM_FEATURE_V8_DOTPROD;
break;
case 0x8: /* FCMLA, #0 */ case 0x8: /* FCMLA, #0 */
case 0x9: /* FCMLA, #90 */ case 0x9: /* FCMLA, #90 */
case 0xa: /* FCMLA, #180 */ case 0xa: /* FCMLA, #180 */
@ -11540,6 +11560,11 @@ static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn)
} }
return; return;
case 0x2: /* SDOT / UDOT */
gen_gvec_op3_ool(s, is_q, rd, rn, rm, 0,
u ? gen_helper_gvec_udot_b : gen_helper_gvec_sdot_b);
return;
case 0x8: /* FCMLA, #0 */ case 0x8: /* FCMLA, #0 */
case 0x9: /* FCMLA, #90 */ case 0x9: /* FCMLA, #90 */
case 0xa: /* FCMLA, #180 */ case 0xa: /* FCMLA, #180 */
@ -12726,6 +12751,13 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
return; return;
} }
break; break;
case 0x0e: /* SDOT */
case 0x1e: /* UDOT */
if (size != MO_32 || !arm_dc_feature(s, ARM_FEATURE_V8_DOTPROD)) {
unallocated_encoding(s);
return;
}
break;
case 0x11: /* FCMLA #0 */ case 0x11: /* FCMLA #0 */
case 0x13: /* FCMLA #90 */ case 0x13: /* FCMLA #90 */
case 0x15: /* FCMLA #180 */ case 0x15: /* FCMLA #180 */
@ -12823,6 +12855,12 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
} }
switch (16 * u + opcode) { switch (16 * u + opcode) {
case 0x0e: /* SDOT */
case 0x1e: /* UDOT */
gen_gvec_op3_ool(s, is_q, rd, rn, rm, index,
u ? gen_helper_gvec_udot_idx_b
: gen_helper_gvec_sdot_idx_b);
return;
case 0x11: /* FCMLA #0 */ case 0x11: /* FCMLA #0 */
case 0x13: /* FCMLA #90 */ case 0x13: /* FCMLA #90 */
case 0x15: /* FCMLA #180 */ case 0x15: /* FCMLA #180 */

74
qemu/target/arm/translate.c
View file

@ -7914,9 +7914,10 @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn)
static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn)
{ {
TCGContext *tcg_ctx = s->uc->tcg_ctx; TCGContext *tcg_ctx = s->uc->tcg_ctx;
gen_helper_gvec_3_ptr *fn_gvec_ptr; gen_helper_gvec_3 *fn_gvec = NULL;
int rd, rn, rm, rot, size, opr_sz; gen_helper_gvec_3_ptr *fn_gvec_ptr = NULL;
TCGv_ptr fpst; int rd, rn, rm, opr_sz;
int data = 0;
bool q; bool q;
q = extract32(insn, 6, 1); q = extract32(insn, 6, 1);
@ -7929,8 +7930,8 @@ static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn)
if ((insn & 0xfe200f10) == 0xfc200800) { if ((insn & 0xfe200f10) == 0xfc200800) {
/* VCMLA -- 1111 110R R.1S .... .... 1000 ...0 .... */ /* VCMLA -- 1111 110R R.1S .... .... 1000 ...0 .... */
size = extract32(insn, 20, 1); int size = extract32(insn, 20, 1);
rot = extract32(insn, 23, 2); data = extract32(insn, 23, 2); /* rot */
if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA) if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA)
|| (!size && !arm_dc_feature(s, ARM_FEATURE_V8_FP16))) { || (!size && !arm_dc_feature(s, ARM_FEATURE_V8_FP16))) {
return 1; return 1;
@ -7938,13 +7939,20 @@ static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn)
fn_gvec_ptr = size ? gen_helper_gvec_fcmlas : gen_helper_gvec_fcmlah; fn_gvec_ptr = size ? gen_helper_gvec_fcmlas : gen_helper_gvec_fcmlah;
} else if ((insn & 0xfea00f10) == 0xfc800800) { } else if ((insn & 0xfea00f10) == 0xfc800800) {
/* VCADD -- 1111 110R 1.0S .... .... 1000 ...0 .... */ /* VCADD -- 1111 110R 1.0S .... .... 1000 ...0 .... */
size = extract32(insn, 20, 1); int size = extract32(insn, 20, 1);
rot = extract32(insn, 24, 1); data = extract32(insn, 24, 1); /* rot */
if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA) if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA)
|| (!size && !arm_dc_feature(s, ARM_FEATURE_V8_FP16))) { || (!size && !arm_dc_feature(s, ARM_FEATURE_V8_FP16))) {
return 1; return 1;
} }
fn_gvec_ptr = size ? gen_helper_gvec_fcadds : gen_helper_gvec_fcaddh; fn_gvec_ptr = size ? gen_helper_gvec_fcadds : gen_helper_gvec_fcaddh;
} else if ((insn & 0xfeb00f00) == 0xfc200d00) {
/* V[US]DOT -- 1111 1100 0.10 .... .... 1101 .Q.U .... */
bool u = extract32(insn, 4, 1);
if (!arm_dc_feature(s, ARM_FEATURE_V8_DOTPROD)) {
return 1;
}
fn_gvec = u ? gen_helper_gvec_udot_b : gen_helper_gvec_sdot_b;
} else { } else {
return 1; return 1;
} }
@ -7959,12 +7967,19 @@ static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn)
} }
opr_sz = (1 + q) * 8; opr_sz = (1 + q) * 8;
fpst = get_fpstatus_ptr(s, 1); if (fn_gvec_ptr) {
tcg_gen_gvec_3_ptr(tcg_ctx, vfp_reg_offset(1, rd), TCGv_ptr fpst = get_fpstatus_ptr(s, 1);
vfp_reg_offset(1, rn), tcg_gen_gvec_3_ptr(tcg_ctx, vfp_reg_offset(1, rd),
vfp_reg_offset(1, rm), fpst, vfp_reg_offset(1, rn),
opr_sz, opr_sz, rot, fn_gvec_ptr); vfp_reg_offset(1, rm), fpst,
tcg_temp_free_ptr(tcg_ctx, fpst); opr_sz, opr_sz, data, fn_gvec_ptr);
tcg_temp_free_ptr(tcg_ctx, fpst);
} else {
tcg_gen_gvec_3_ool(tcg_ctx, vfp_reg_offset(1, rd),
vfp_reg_offset(1, rn),
vfp_reg_offset(1, rm),
opr_sz, opr_sz, data, fn_gvec);
}
return 0; return 0;
} }
@ -7979,9 +7994,9 @@ static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn)
static int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn) static int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn)
{ {
TCGContext *tcg_ctx = s->uc->tcg_ctx; TCGContext *tcg_ctx = s->uc->tcg_ctx;
gen_helper_gvec_3_ptr *fn_gvec_ptr; gen_helper_gvec_3 *fn_gvec = NULL;
gen_helper_gvec_3_ptr *fn_gvec_ptr = NULL;
int rd, rn, rm, opr_sz, data; int rd, rn, rm, opr_sz, data;
TCGv_ptr fpst;
bool q; bool q;
q = extract32(insn, 6, 1); q = extract32(insn, 6, 1);
@ -8015,6 +8030,16 @@ static int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn)
data = (index << 2) | rot; data = (index << 2) | rot;
fn_gvec_ptr = (size ? gen_helper_gvec_fcmlas_idx fn_gvec_ptr = (size ? gen_helper_gvec_fcmlas_idx
: gen_helper_gvec_fcmlah_idx); : gen_helper_gvec_fcmlah_idx);
} else if ((insn & 0xffb00f00) == 0xfe200d00) {
/* V[US]DOT -- 1111 1110 0.10 .... .... 1101 .Q.U .... */
int u = extract32(insn, 4, 1);
if (!arm_dc_feature(s, ARM_FEATURE_V8_DOTPROD)) {
return 1;
}
fn_gvec = u ? gen_helper_gvec_udot_idx_b : gen_helper_gvec_sdot_idx_b;
/* rm is just Vm, and index is M. */
data = extract32(insn, 5, 1); /* index */
rm = extract32(insn, 0, 4);
} else { } else {
return 1; return 1;
} }
@ -8029,12 +8054,19 @@ static int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn)
} }
opr_sz = (1 + q) * 8; opr_sz = (1 + q) * 8;
fpst = get_fpstatus_ptr(s, 1); if (fn_gvec_ptr) {
tcg_gen_gvec_3_ptr(tcg_ctx, vfp_reg_offset(1, rd), TCGv_ptr fpst = get_fpstatus_ptr(s, 1);
vfp_reg_offset(1, rn), tcg_gen_gvec_3_ptr(tcg_ctx, vfp_reg_offset(1, rd),
vfp_reg_offset(1, rm), fpst, vfp_reg_offset(1, rn),
opr_sz, opr_sz, data, fn_gvec_ptr); vfp_reg_offset(1, rm), fpst,
tcg_temp_free_ptr(tcg_ctx, fpst); opr_sz, opr_sz, data, fn_gvec_ptr);
tcg_temp_free_ptr(tcg_ctx, fpst);
} else {
tcg_gen_gvec_3_ool(tcg_ctx, vfp_reg_offset(1, rd),
vfp_reg_offset(1, rn),
vfp_reg_offset(1, rm),
opr_sz, opr_sz, data, fn_gvec);
}
return 0; return 0;
} }