mirrors/qemu
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

target/arm: Convert FMLA, FMLS to decodetree

Browse Source 
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240524232121.284515-25-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This commit is contained in:
Richard Henderson 2024-05-24 16:20:38 -07:00 committed by Peter Maydell
parent 69cefabcac
commit 2d558efbf5
4 changed files with 163 additions and 116 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
target/arm/helper.h
View File

@ -770,9 +770,11 @@ 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_vfma_d, 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_vfms_d, 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)

22
target/arm/tcg/a64.decode
View File

@ -742,12 +742,26 @@ FMINNM_v 0.00 1110 1.1 ..... 11000 1 ..... ..... @qrrr_sd
FMULX_v 0.00 1110 010 ..... 00011 1 ..... ..... @qrrr_h
FMULX_v 0.00 1110 0.1 ..... 11011 1 ..... ..... @qrrr_sd
FMLA_v 0.00 1110 010 ..... 00001 1 ..... ..... @qrrr_h
FMLA_v 0.00 1110 0.1 ..... 11001 1 ..... ..... @qrrr_sd
FMLS_v 0.00 1110 110 ..... 00001 1 ..... ..... @qrrr_h
FMLS_v 0.00 1110 1.1 ..... 11001 1 ..... ..... @qrrr_sd
### Advanced SIMD scalar x indexed element
FMUL_si 0101 1111 00 .. .... 1001 . 0 ..... ..... @rrx_h
FMUL_si 0101 1111 10 . ..... 1001 . 0 ..... ..... @rrx_s
FMUL_si 0101 1111 11 0 ..... 1001 . 0 ..... ..... @rrx_d
FMLA_si 0101 1111 00 .. .... 0001 . 0 ..... ..... @rrx_h
FMLA_si 0101 1111 10 .. .... 0001 . 0 ..... ..... @rrx_s
FMLA_si 0101 1111 11 0. .... 0001 . 0 ..... ..... @rrx_d
FMLS_si 0101 1111 00 .. .... 0101 . 0 ..... ..... @rrx_h
FMLS_si 0101 1111 10 .. .... 0101 . 0 ..... ..... @rrx_s
FMLS_si 0101 1111 11 0. .... 0101 . 0 ..... ..... @rrx_d
FMULX_si 0111 1111 00 .. .... 1001 . 0 ..... ..... @rrx_h
FMULX_si 0111 1111 10 . ..... 1001 . 0 ..... ..... @rrx_s
FMULX_si 0111 1111 11 0 ..... 1001 . 0 ..... ..... @rrx_d
@ -758,6 +772,14 @@ FMUL_vi 0.00 1111 00 .. .... 1001 . 0 ..... ..... @qrrx_h
FMUL_vi 0.00 1111 10 . ..... 1001 . 0 ..... ..... @qrrx_s
FMUL_vi 0.00 1111 11 0 ..... 1001 . 0 ..... ..... @qrrx_d
FMLA_vi 0.00 1111 00 .. .... 0001 . 0 ..... ..... @qrrx_h
FMLA_vi 0.00 1111 10 . ..... 0001 . 0 ..... ..... @qrrx_s
FMLA_vi 0.00 1111 11 0 ..... 0001 . 0 ..... ..... @qrrx_d
FMLS_vi 0.00 1111 00 .. .... 0101 . 0 ..... ..... @qrrx_h
FMLS_vi 0.00 1111 10 . ..... 0101 . 0 ..... ..... @qrrx_s
FMLS_vi 0.00 1111 11 0 ..... 0101 . 0 ..... ..... @qrrx_d
FMULX_vi 0.10 1111 00 .. .... 1001 . 0 ..... ..... @qrrx_h
FMULX_vi 0.10 1111 10 . ..... 1001 . 0 ..... ..... @qrrx_s
FMULX_vi 0.10 1111 11 0 ..... 1001 . 0 ..... ..... @qrrx_d

241
target/arm/tcg/translate-a64.c
View File

@ -5066,6 +5066,20 @@ static gen_helper_gvec_3_ptr * const f_vector_fmulx[3] = {
};
TRANS(FMULX_v, do_fp3_vector, a, f_vector_fmulx)
static gen_helper_gvec_3_ptr * const f_vector_fmla[3] = {
gen_helper_gvec_vfma_h,
gen_helper_gvec_vfma_s,
gen_helper_gvec_vfma_d,
};
TRANS(FMLA_v, do_fp3_vector, a, f_vector_fmla)
static gen_helper_gvec_3_ptr * const f_vector_fmls[3] = {
gen_helper_gvec_vfms_h,
gen_helper_gvec_vfms_s,
gen_helper_gvec_vfms_d,
};
TRANS(FMLS_v, do_fp3_vector, a, f_vector_fmls)
/*
* Advanced SIMD scalar/vector x indexed element
*/
@ -5115,6 +5129,64 @@ static bool do_fp3_scalar_idx(DisasContext *s, arg_rrx_e *a, const FPScalar *f)
TRANS(FMUL_si, do_fp3_scalar_idx, a, &f_scalar_fmul)
TRANS(FMULX_si, do_fp3_scalar_idx, a, &f_scalar_fmulx)
static bool do_fmla_scalar_idx(DisasContext *s, arg_rrx_e *a, bool neg)
{
switch (a->esz) {
case MO_64:
if (fp_access_check(s)) {
TCGv_i64 t0 = read_fp_dreg(s, a->rd);
TCGv_i64 t1 = read_fp_dreg(s, a->rn);
TCGv_i64 t2 = tcg_temp_new_i64();
read_vec_element(s, t2, a->rm, a->idx, MO_64);
if (neg) {
gen_vfp_negd(t1, t1);
}
gen_helper_vfp_muladdd(t0, t1, t2, t0, fpstatus_ptr(FPST_FPCR));
write_fp_dreg(s, a->rd, t0);
}
break;
case MO_32:
if (fp_access_check(s)) {
TCGv_i32 t0 = read_fp_sreg(s, a->rd);
TCGv_i32 t1 = read_fp_sreg(s, a->rn);
TCGv_i32 t2 = tcg_temp_new_i32();
read_vec_element_i32(s, t2, a->rm, a->idx, MO_32);
if (neg) {
gen_vfp_negs(t1, t1);
}
gen_helper_vfp_muladds(t0, t1, t2, t0, fpstatus_ptr(FPST_FPCR));
write_fp_sreg(s, a->rd, t0);
}
break;
case MO_16:
if (!dc_isar_feature(aa64_fp16, s)) {
return false;
}
if (fp_access_check(s)) {
TCGv_i32 t0 = read_fp_hreg(s, a->rd);
TCGv_i32 t1 = read_fp_hreg(s, a->rn);
TCGv_i32 t2 = tcg_temp_new_i32();
read_vec_element_i32(s, t2, a->rm, a->idx, MO_16);
if (neg) {
gen_vfp_negh(t1, t1);
}
gen_helper_advsimd_muladdh(t0, t1, t2, t0,
fpstatus_ptr(FPST_FPCR_F16));
write_fp_sreg(s, a->rd, t0);
}
break;
default:
g_assert_not_reached();
}
return true;
}
TRANS(FMLA_si, do_fmla_scalar_idx, a, false)
TRANS(FMLS_si, do_fmla_scalar_idx, a, true)
static bool do_fp3_vector_idx(DisasContext *s, arg_qrrx_e *a,
gen_helper_gvec_3_ptr * const fns[3])
{
@ -5157,6 +5229,42 @@ static gen_helper_gvec_3_ptr * const f_vector_idx_fmulx[3] = {
};
TRANS(FMULX_vi, do_fp3_vector_idx, a, f_vector_idx_fmulx)
static bool do_fmla_vector_idx(DisasContext *s, arg_qrrx_e *a, bool neg)
{
static gen_helper_gvec_4_ptr * const fns[3] = {
gen_helper_gvec_fmla_idx_h,
gen_helper_gvec_fmla_idx_s,
gen_helper_gvec_fmla_idx_d,
};
MemOp esz = a->esz;
switch (esz) {
case MO_64:
if (!a->q) {
return false;
}
break;
case MO_32:
break;
case MO_16:
if (!dc_isar_feature(aa64_fp16, s)) {
return false;
}
break;
default:
g_assert_not_reached();
}
if (fp_access_check(s)) {
gen_gvec_op4_fpst(s, a->q, a->rd, a->rn, a->rm, a->rd,
esz == MO_16, (a->idx << 1) | neg,
fns[esz - 1]);
}
return true;
}
TRANS(FMLA_vi, do_fmla_vector_idx, a, false)
TRANS(FMLS_vi, do_fmla_vector_idx, a, true)
/* Shift a TCGv src by TCGv shift_amount, put result in dst.
* Note that it is the caller's responsibility to ensure that the
@ -9119,15 +9227,6 @@ static void handle_3same_float(DisasContext *s, int size, int elements,
read_vec_element(s, tcg_op2, rm, pass, MO_64);
switch (fpopcode) {
case 0x39: /* FMLS */
/* As usual for ARM, separate negation for fused multiply-add */
gen_vfp_negd(tcg_op1, tcg_op1);
/* fall through */
case 0x19: /* FMLA */
read_vec_element(s, tcg_res, rd, pass, MO_64);
gen_helper_vfp_muladdd(tcg_res, tcg_op1, tcg_op2,
tcg_res, fpst);
break;
case 0x1c: /* FCMEQ */
gen_helper_neon_ceq_f64(tcg_res, tcg_op1, tcg_op2, fpst);
break;
@ -9155,10 +9254,12 @@ static void handle_3same_float(DisasContext *s, int size, int elements,
break;
default:
case 0x18: /* FMAXNM */
case 0x19: /* FMLA */
case 0x1a: /* FADD */
case 0x1b: /* FMULX */
case 0x1e: /* FMAX */
case 0x38: /* FMINNM */
case 0x39: /* FMLS */
case 0x3a: /* FSUB */
case 0x3e: /* FMIN */
case 0x5b: /* FMUL */
@ -9177,15 +9278,6 @@ static void handle_3same_float(DisasContext *s, int size, int elements,
read_vec_element_i32(s, tcg_op2, rm, pass, MO_32);
switch (fpopcode) {
case 0x39: /* FMLS */
/* As usual for ARM, separate negation for fused multiply-add */
gen_vfp_negs(tcg_op1, tcg_op1);
/* fall through */
case 0x19: /* FMLA */
read_vec_element_i32(s, tcg_res, rd, pass, MO_32);
gen_helper_vfp_muladds(tcg_res, tcg_op1, tcg_op2,
tcg_res, fpst);
break;
case 0x1c: /* FCMEQ */
gen_helper_neon_ceq_f32(tcg_res, tcg_op1, tcg_op2, fpst);
break;
@ -9213,10 +9305,12 @@ static void handle_3same_float(DisasContext *s, int size, int elements,
break;
default:
case 0x18: /* FMAXNM */
case 0x19: /* FMLA */
case 0x1a: /* FADD */
case 0x1b: /* FMULX */
case 0x1e: /* FMAX */
case 0x38: /* FMINNM */
case 0x39: /* FMLS */
case 0x3a: /* FSUB */
case 0x3e: /* FMIN */
case 0x5b: /* FMUL */
@ -11140,8 +11234,6 @@ static void disas_simd_3same_float(DisasContext *s, uint32_t insn)
case 0x3f: /* FRSQRTS */
case 0x5d: /* FACGE */
case 0x7d: /* FACGT */
case 0x19: /* FMLA */
case 0x39: /* FMLS */
case 0x1c: /* FCMEQ */
case 0x5c: /* FCMGE */
case 0x7a: /* FABD */
@ -11174,10 +11266,12 @@ static void disas_simd_3same_float(DisasContext *s, uint32_t insn)
default:
case 0x18: /* FMAXNM */
case 0x19: /* FMLA */
case 0x1a: /* FADD */
case 0x1b: /* FMULX */
case 0x1e: /* FMAX */
case 0x38: /* FMINNM */
case 0x39: /* FMLS */
case 0x3a: /* FSUB */
case 0x3e: /* FMIN */
case 0x5b: /* FMUL */
@ -11523,10 +11617,8 @@ static void disas_simd_three_reg_same_fp16(DisasContext *s, uint32_t insn)
int pass;
switch (fpopcode) {
case 0x1: /* FMLA */
case 0x4: /* FCMEQ */
case 0x7: /* FRECPS */
case 0x9: /* FMLS */
case 0xf: /* FRSQRTS */
case 0x14: /* FCMGE */
case 0x15: /* FACGE */
@ -11544,10 +11636,12 @@ static void disas_simd_three_reg_same_fp16(DisasContext *s, uint32_t insn)
break;
default:
case 0x0: /* FMAXNM */
case 0x1: /* FMLA */
case 0x2: /* FADD */
case 0x3: /* FMULX */
case 0x6: /* FMAX */
case 0x8: /* FMINNM */
case 0x9: /* FMLS */
case 0xa: /* FSUB */
case 0xe: /* FMIN */
case 0x13: /* FMUL */
@ -11617,24 +11711,12 @@ static void disas_simd_three_reg_same_fp16(DisasContext *s, uint32_t insn)
read_vec_element_i32(s, tcg_op2, rm, pass, MO_16);
switch (fpopcode) {
case 0x1: /* FMLA */
read_vec_element_i32(s, tcg_res, rd, pass, MO_16);
gen_helper_advsimd_muladdh(tcg_res, tcg_op1, tcg_op2, tcg_res,
fpst);
break;
case 0x4: /* FCMEQ */
gen_helper_advsimd_ceq_f16(tcg_res, tcg_op1, tcg_op2, fpst);
break;
case 0x7: /* FRECPS */
gen_helper_recpsf_f16(tcg_res, tcg_op1, tcg_op2, fpst);
break;
case 0x9: /* FMLS */
/* As usual for ARM, separate negation for fused multiply-add */
tcg_gen_xori_i32(tcg_op1, tcg_op1, 0x8000);
read_vec_element_i32(s, tcg_res, rd, pass, MO_16);
gen_helper_advsimd_muladdh(tcg_res, tcg_op1, tcg_op2, tcg_res,
fpst);
break;
case 0xf: /* FRSQRTS */
gen_helper_rsqrtsf_f16(tcg_res, tcg_op1, tcg_op2, fpst);
break;
@ -11656,10 +11738,12 @@ static void disas_simd_three_reg_same_fp16(DisasContext *s, uint32_t insn)
break;
default:
case 0x0: /* FMAXNM */
case 0x1: /* FMLA */
case 0x2: /* FADD */
case 0x3: /* FMULX */
case 0x6: /* FMAX */
case 0x8: /* FMINNM */
case 0x9: /* FMLS */
case 0xa: /* FSUB */
case 0xe: /* FMIN */
case 0x13: /* FMUL */
@ -12880,10 +12964,6 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
case 0x0c: /* SQDMULH */
case 0x0d: /* SQRDMULH */
break;
case 0x01: /* FMLA */
case 0x05: /* FMLS */
is_fp = 1;
break;
case 0x1d: /* SQRDMLAH */
case 0x1f: /* SQRDMLSH */
if (!dc_isar_feature(aa64_rdm, s)) {
@ -12950,6 +13030,8 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
/* is_fp, but we pass tcg_env not fp_status. */
break;
default:
case 0x01: /* FMLA */
case 0x05: /* FMLS */
case 0x09: /* FMUL */
case 0x19: /* FMULX */
unallocated_encoding(s);
@ -12958,20 +13040,8 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
switch (is_fp) {
case 1: /* normal fp */
/* convert insn encoded size to MemOp size */
switch (size) {
case 0: /* half-precision */
size = MO_16;
is_fp16 = true;
break;
case MO_32: /* single precision */
case MO_64: /* double precision */
break;
default:
unallocated_encoding(s);
return;
}
break;
unallocated_encoding(s); /* in decodetree */
return;
case 2: /* complex fp */
/* Each indexable element is a complex pair. */
@ -13150,38 +13220,7 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
}
if (size == 3) {
TCGv_i64 tcg_idx = tcg_temp_new_i64();
int pass;
assert(is_fp && is_q && !is_long);
read_vec_element(s, tcg_idx, rm, index, MO_64);
for (pass = 0; pass < (is_scalar ? 1 : 2); pass++) {
TCGv_i64 tcg_op = tcg_temp_new_i64();
TCGv_i64 tcg_res = tcg_temp_new_i64();
read_vec_element(s, tcg_op, rn, pass, MO_64);
switch (16 * u + opcode) {
case 0x05: /* FMLS */
/* As usual for ARM, separate negation for fused multiply-add */
gen_vfp_negd(tcg_op, tcg_op);
/* fall through */
case 0x01: /* FMLA */
read_vec_element(s, tcg_res, rd, pass, MO_64);
gen_helper_vfp_muladdd(tcg_res, tcg_op, tcg_idx, tcg_res, fpst);
break;
default:
case 0x09: /* FMUL */
case 0x19: /* FMULX */
g_assert_not_reached();
}
write_vec_element(s, tcg_res, rd, pass, MO_64);
}
clear_vec_high(s, !is_scalar, rd);
g_assert_not_reached();
} else if (!is_long) {
/* 32 bit floating point, or 16 or 32 bit integer.
* For the 16 bit scalar case we use the usual Neon helpers and
@ -13237,38 +13276,6 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
genfn(tcg_res, tcg_op, tcg_res);
break;
}
case 0x05: /* FMLS */
case 0x01: /* FMLA */
read_vec_element_i32(s, tcg_res, rd, pass,
is_scalar ? size : MO_32);
switch (size) {
case 1:
if (opcode == 0x5) {
/* As usual for ARM, separate negation for fused
* multiply-add */
tcg_gen_xori_i32(tcg_op, tcg_op, 0x80008000);
}
if (is_scalar) {
gen_helper_advsimd_muladdh(tcg_res, tcg_op, tcg_idx,
tcg_res, fpst);
} else {
gen_helper_advsimd_muladd2h(tcg_res, tcg_op, tcg_idx,
tcg_res, fpst);
}
break;
case 2:
if (opcode == 0x5) {
/* As usual for ARM, separate negation for
* fused multiply-add */
tcg_gen_xori_i32(tcg_op, tcg_op, 0x80000000);
}
gen_helper_vfp_muladds(tcg_res, tcg_op, tcg_idx,
tcg_res, fpst);
break;
default:
g_assert_not_reached();
}
break;
case 0x0c: /* SQDMULH */
if (size == 1) {
gen_helper_neon_qdmulh_s16(tcg_res, tcg_env,
@ -13310,6 +13317,8 @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn)
}
break;
default:
case 0x01: /* FMLA */
case 0x05: /* FMLS */
case 0x09: /* FMUL */
case 0x19: /* FMULX */
g_assert_not_reached();

14
target/arm/tcg/vec_helper.c
View File

@ -1309,6 +1309,12 @@ static float32 float32_muladd_f(float32 dest, float32 op1, float32 op2,
return float32_muladd(op1, op2, dest, 0, stat);
}
static float64 float64_muladd_f(float64 dest, float64 op1, float64 op2,
float_status *stat)
{
return float64_muladd(op1, op2, dest, 0, stat);
}
static float16 float16_mulsub_f(float16 dest, float16 op1, float16 op2,
float_status *stat)
{
@ -1321,6 +1327,12 @@ static float32 float32_mulsub_f(float32 dest, float32 op1, float32 op2,
return float32_muladd(float32_chs(op1), op2, dest, 0, stat);
}
static float64 float64_mulsub_f(float64 dest, float64 op1, float64 op2,
float_status *stat)
{
return float64_muladd(float64_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) \
{ \
@ -1340,9 +1352,11 @@ 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_vfma_d, float64_muladd_f, float64)
DO_MULADD(gvec_vfms_h, float16_mulsub_f, float16)
DO_MULADD(gvec_vfms_s, float32_mulsub_f, float32)
DO_MULADD(gvec_vfms_d, float64_mulsub_f, float64)
/* For the indexed ops, SVE applies the index per 128-bit vector segment.
* For AdvSIMD, there is of course only one such vector segment.