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

target/arm: Move sve zip high_ofs into simd_data

Browse Source 
This is in line with how we treat uzp, and will
eliminate the special case code during translation.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220527181907.189259-58-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This commit is contained in:
Richard Henderson 2022-05-27 11:18:10 -07:00 committed by Peter Maydell
parent 1d0fce4bd0
commit 09eb6d7025
2 changed files with 10 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
target/arm/sve_helper.c
View File

@ -3382,6 +3382,7 @@ void HELPER(sve_punpk_p)(void *vd, void *vn, uint32_t pred_desc)
void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \
{ \ { \
intptr_t oprsz = simd_oprsz(desc); \ intptr_t oprsz = simd_oprsz(desc); \
intptr_t odd_ofs = simd_data(desc); \
intptr_t i, oprsz_2 = oprsz / 2; \ intptr_t i, oprsz_2 = oprsz / 2; \
ARMVectorReg tmp_n, tmp_m; \ ARMVectorReg tmp_n, tmp_m; \
/* We produce output faster than we consume input. \ /* We produce output faster than we consume input. \
@ -3393,8 +3394,9 @@ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \
vm = memcpy(&tmp_m, vm, oprsz_2); \ vm = memcpy(&tmp_m, vm, oprsz_2); \
} \ } \
for (i = 0; i < oprsz_2; i += sizeof(TYPE)) { \ for (i = 0; i < oprsz_2; i += sizeof(TYPE)) { \
*(TYPE *)(vd + H(2 * i + 0)) = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(2 * i + 0)) = *(TYPE *)(vn + odd_ofs + H(i)); \
*(TYPE *)(vd + H(2 * i + sizeof(TYPE))) = *(TYPE *)(vm + H(i)); \ *(TYPE *)(vd + H(2 * i + sizeof(TYPE))) = \
*(TYPE *)(vm + odd_ofs + H(i)); \
} \ } \
if (sizeof(TYPE) == 16 && unlikely(oprsz & 16)) { \ if (sizeof(TYPE) == 16 && unlikely(oprsz & 16)) { \
memset(vd + oprsz - 16, 0, 16); \ memset(vd + oprsz - 16, 0, 16); \

12
target/arm/translate-sve.c
View File

@ -2298,9 +2298,9 @@ static bool do_zip(DisasContext *s, arg_rrr_esz *a, bool high)
unsigned vsz = vec_full_reg_size(s); unsigned vsz = vec_full_reg_size(s);
unsigned high_ofs = high ? vsz / 2 : 0; unsigned high_ofs = high ? vsz / 2 : 0;
tcg_gen_gvec_3_ool(vec_full_reg_offset(s, a->rd), tcg_gen_gvec_3_ool(vec_full_reg_offset(s, a->rd),
vec_full_reg_offset(s, a->rn) + high_ofs, vec_full_reg_offset(s, a->rn),
vec_full_reg_offset(s, a->rm) + high_ofs, vec_full_reg_offset(s, a->rm),
vsz, vsz, 0, fns[a->esz]); vsz, vsz, high_ofs, fns[a->esz]);
} }
return true; return true;
} }
@ -2324,9 +2324,9 @@ static bool do_zip_q(DisasContext *s, arg_rrr_esz *a, bool high)
unsigned vsz = vec_full_reg_size(s); unsigned vsz = vec_full_reg_size(s);
unsigned high_ofs = high ? QEMU_ALIGN_DOWN(vsz, 32) / 2 : 0; unsigned high_ofs = high ? QEMU_ALIGN_DOWN(vsz, 32) / 2 : 0;
tcg_gen_gvec_3_ool(vec_full_reg_offset(s, a->rd), tcg_gen_gvec_3_ool(vec_full_reg_offset(s, a->rd),
vec_full_reg_offset(s, a->rn) + high_ofs, vec_full_reg_offset(s, a->rn),
vec_full_reg_offset(s, a->rm) + high_ofs, vec_full_reg_offset(s, a->rm),
vsz, vsz, 0, gen_helper_sve2_zip_q); vsz, vsz, high_ofs, gen_helper_sve2_zip_q);
} }
return true; return true;
} }