target/arm: Implement SVE bitwise shift by wide elements (predicated)

This commit is contained in:
Lioncash 2018-05-20 03:09:27 -04:00
parent 5aa51a3a74
commit 878b862a04
No known key found for this signature in database
GPG key ID: 4E3C3CC1031BA9C7
7 changed files with 113 additions and 0 deletions

View file

@ -3287,6 +3287,9 @@
#define helper_sve_andv_d helper_sve_andv_d_aarch64 #define helper_sve_andv_d helper_sve_andv_d_aarch64
#define helper_sve_andv_h helper_sve_andv_h_aarch64 #define helper_sve_andv_h helper_sve_andv_h_aarch64
#define helper_sve_andv_s helper_sve_andv_s_aarch64 #define helper_sve_andv_s helper_sve_andv_s_aarch64
#define helper_sve_asr_zpzw_b helper_sve_asr_zpzw_b_aarch64
#define helper_sve_asr_zpzw_h helper_sve_asr_zpzw_h_aarch64
#define helper_sve_asr_zpzw_s helper_sve_asr_zpzw_s_aarch64
#define helper_sve_asr_zpzz_b helper_sve_asr_zpzz_b_aarch64 #define helper_sve_asr_zpzz_b helper_sve_asr_zpzz_b_aarch64
#define helper_sve_asr_zpzz_d helper_sve_asr_zpzz_d_aarch64 #define helper_sve_asr_zpzz_d helper_sve_asr_zpzz_d_aarch64
#define helper_sve_asr_zpzz_h helper_sve_asr_zpzz_h_aarch64 #define helper_sve_asr_zpzz_h helper_sve_asr_zpzz_h_aarch64
@ -3321,6 +3324,9 @@
#define helper_sve_lsl_zpzi_d helper_sve_lsl_zpzi_d_aarch64 #define helper_sve_lsl_zpzi_d helper_sve_lsl_zpzi_d_aarch64
#define helper_sve_lsl_zpzi_h helper_sve_lsl_zpzi_h_aarch64 #define helper_sve_lsl_zpzi_h helper_sve_lsl_zpzi_h_aarch64
#define helper_sve_lsl_zpzi_s helper_sve_lsl_zpzi_s_aarch64 #define helper_sve_lsl_zpzi_s helper_sve_lsl_zpzi_s_aarch64
#define helper_sve_lsl_zpzw_b helper_sve_lsl_zpzw_b_aarch64
#define helper_sve_lsl_zpzw_h helper_sve_lsl_zpzw_h_aarch64
#define helper_sve_lsl_zpzw_s helper_sve_lsl_zpzw_s_aarch64
#define helper_sve_lsl_zpzz_b helper_sve_lsl_zpzz_b_aarch64 #define helper_sve_lsl_zpzz_b helper_sve_lsl_zpzz_b_aarch64
#define helper_sve_lsl_zpzz_d helper_sve_lsl_zpzz_d_aarch64 #define helper_sve_lsl_zpzz_d helper_sve_lsl_zpzz_d_aarch64
#define helper_sve_lsl_zpzz_h helper_sve_lsl_zpzz_h_aarch64 #define helper_sve_lsl_zpzz_h helper_sve_lsl_zpzz_h_aarch64
@ -3329,6 +3335,9 @@
#define helper_sve_lsr_zpzi_d helper_sve_lsr_zpzi_d_aarch64 #define helper_sve_lsr_zpzi_d helper_sve_lsr_zpzi_d_aarch64
#define helper_sve_lsr_zpzi_h helper_sve_lsr_zpzi_h_aarch64 #define helper_sve_lsr_zpzi_h helper_sve_lsr_zpzi_h_aarch64
#define helper_sve_lsr_zpzi_s helper_sve_lsr_zpzi_s_aarch64 #define helper_sve_lsr_zpzi_s helper_sve_lsr_zpzi_s_aarch64
#define helper_sve_lsr_zpzw_b helper_sve_lsr_zpzw_b_aarch64
#define helper_sve_lsr_zpzw_h helper_sve_lsr_zpzw_h_aarch64
#define helper_sve_lsr_zpzw_s helper_sve_lsr_zpzw_s_aarch64
#define helper_sve_lsr_zpzz_b helper_sve_lsr_zpzz_b_aarch64 #define helper_sve_lsr_zpzz_b helper_sve_lsr_zpzz_b_aarch64
#define helper_sve_lsr_zpzz_d helper_sve_lsr_zpzz_d_aarch64 #define helper_sve_lsr_zpzz_d helper_sve_lsr_zpzz_d_aarch64
#define helper_sve_lsr_zpzz_h helper_sve_lsr_zpzz_h_aarch64 #define helper_sve_lsr_zpzz_h helper_sve_lsr_zpzz_h_aarch64

View file

@ -3287,6 +3287,9 @@
#define helper_sve_andv_d helper_sve_andv_d_aarch64eb #define helper_sve_andv_d helper_sve_andv_d_aarch64eb
#define helper_sve_andv_h helper_sve_andv_h_aarch64eb #define helper_sve_andv_h helper_sve_andv_h_aarch64eb
#define helper_sve_andv_s helper_sve_andv_s_aarch64eb #define helper_sve_andv_s helper_sve_andv_s_aarch64eb
#define helper_sve_asr_zpzw_b helper_sve_asr_zpzw_b_aarch64eb
#define helper_sve_asr_zpzw_h helper_sve_asr_zpzw_h_aarch64eb
#define helper_sve_asr_zpzw_s helper_sve_asr_zpzw_s_aarch64eb
#define helper_sve_asr_zpzz_b helper_sve_asr_zpzz_b_aarch64eb #define helper_sve_asr_zpzz_b helper_sve_asr_zpzz_b_aarch64eb
#define helper_sve_asr_zpzz_d helper_sve_asr_zpzz_d_aarch64eb #define helper_sve_asr_zpzz_d helper_sve_asr_zpzz_d_aarch64eb
#define helper_sve_asr_zpzz_h helper_sve_asr_zpzz_h_aarch64eb #define helper_sve_asr_zpzz_h helper_sve_asr_zpzz_h_aarch64eb
@ -3321,6 +3324,9 @@
#define helper_sve_lsl_zpzi_d helper_sve_lsl_zpzi_d_aarch64eb #define helper_sve_lsl_zpzi_d helper_sve_lsl_zpzi_d_aarch64eb
#define helper_sve_lsl_zpzi_h helper_sve_lsl_zpzi_h_aarch64eb #define helper_sve_lsl_zpzi_h helper_sve_lsl_zpzi_h_aarch64eb
#define helper_sve_lsl_zpzi_s helper_sve_lsl_zpzi_s_aarch64eb #define helper_sve_lsl_zpzi_s helper_sve_lsl_zpzi_s_aarch64eb
#define helper_sve_lsl_zpzw_b helper_sve_lsl_zpzw_b_aarch64eb
#define helper_sve_lsl_zpzw_h helper_sve_lsl_zpzw_h_aarch64eb
#define helper_sve_lsl_zpzw_s helper_sve_lsl_zpzw_s_aarch64eb
#define helper_sve_lsl_zpzz_b helper_sve_lsl_zpzz_b_aarch64eb #define helper_sve_lsl_zpzz_b helper_sve_lsl_zpzz_b_aarch64eb
#define helper_sve_lsl_zpzz_d helper_sve_lsl_zpzz_d_aarch64eb #define helper_sve_lsl_zpzz_d helper_sve_lsl_zpzz_d_aarch64eb
#define helper_sve_lsl_zpzz_h helper_sve_lsl_zpzz_h_aarch64eb #define helper_sve_lsl_zpzz_h helper_sve_lsl_zpzz_h_aarch64eb
@ -3329,6 +3335,9 @@
#define helper_sve_lsr_zpzi_d helper_sve_lsr_zpzi_d_aarch64eb #define helper_sve_lsr_zpzi_d helper_sve_lsr_zpzi_d_aarch64eb
#define helper_sve_lsr_zpzi_h helper_sve_lsr_zpzi_h_aarch64eb #define helper_sve_lsr_zpzi_h helper_sve_lsr_zpzi_h_aarch64eb
#define helper_sve_lsr_zpzi_s helper_sve_lsr_zpzi_s_aarch64eb #define helper_sve_lsr_zpzi_s helper_sve_lsr_zpzi_s_aarch64eb
#define helper_sve_lsr_zpzw_b helper_sve_lsr_zpzw_b_aarch64eb
#define helper_sve_lsr_zpzw_h helper_sve_lsr_zpzw_h_aarch64eb
#define helper_sve_lsr_zpzw_s helper_sve_lsr_zpzw_s_aarch64eb
#define helper_sve_lsr_zpzz_b helper_sve_lsr_zpzz_b_aarch64eb #define helper_sve_lsr_zpzz_b helper_sve_lsr_zpzz_b_aarch64eb
#define helper_sve_lsr_zpzz_d helper_sve_lsr_zpzz_d_aarch64eb #define helper_sve_lsr_zpzz_d helper_sve_lsr_zpzz_d_aarch64eb
#define helper_sve_lsr_zpzz_h helper_sve_lsr_zpzz_h_aarch64eb #define helper_sve_lsr_zpzz_h helper_sve_lsr_zpzz_h_aarch64eb

View file

@ -3308,6 +3308,9 @@ aarch64_symbols = (
'helper_sve_andv_d', 'helper_sve_andv_d',
'helper_sve_andv_h', 'helper_sve_andv_h',
'helper_sve_andv_s', 'helper_sve_andv_s',
'helper_sve_asr_zpzw_b',
'helper_sve_asr_zpzw_h',
'helper_sve_asr_zpzw_s',
'helper_sve_asr_zpzz_b', 'helper_sve_asr_zpzz_b',
'helper_sve_asr_zpzz_d', 'helper_sve_asr_zpzz_d',
'helper_sve_asr_zpzz_h', 'helper_sve_asr_zpzz_h',
@ -3342,6 +3345,9 @@ aarch64_symbols = (
'helper_sve_lsl_zpzi_d', 'helper_sve_lsl_zpzi_d',
'helper_sve_lsl_zpzi_h', 'helper_sve_lsl_zpzi_h',
'helper_sve_lsl_zpzi_s', 'helper_sve_lsl_zpzi_s',
'helper_sve_lsl_zpzw_b',
'helper_sve_lsl_zpzw_h',
'helper_sve_lsl_zpzw_s',
'helper_sve_lsl_zpzz_b', 'helper_sve_lsl_zpzz_b',
'helper_sve_lsl_zpzz_d', 'helper_sve_lsl_zpzz_d',
'helper_sve_lsl_zpzz_h', 'helper_sve_lsl_zpzz_h',
@ -3350,6 +3356,9 @@ aarch64_symbols = (
'helper_sve_lsr_zpzi_d', 'helper_sve_lsr_zpzi_d',
'helper_sve_lsr_zpzi_h', 'helper_sve_lsr_zpzi_h',
'helper_sve_lsr_zpzi_s', 'helper_sve_lsr_zpzi_s',
'helper_sve_lsr_zpzw_b',
'helper_sve_lsr_zpzw_h',
'helper_sve_lsr_zpzw_s',
'helper_sve_lsr_zpzz_b', 'helper_sve_lsr_zpzz_b',
'helper_sve_lsr_zpzz_d', 'helper_sve_lsr_zpzz_d',
'helper_sve_lsr_zpzz_h', 'helper_sve_lsr_zpzz_h',

View file

@ -195,6 +195,27 @@ DEF_HELPER_FLAGS_5(sve_lsl_zpzz_s, TCG_CALL_NO_RWG,
DEF_HELPER_FLAGS_5(sve_lsl_zpzz_d, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_5(sve_lsl_zpzz_d, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32) void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_asr_zpzw_b, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_asr_zpzw_h, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_asr_zpzw_s, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_lsr_zpzw_b, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_lsr_zpzw_h, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_lsr_zpzw_s, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_lsl_zpzw_b, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_lsl_zpzw_h, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_5(sve_lsl_zpzw_s, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_3(sve_orv_b, TCG_CALL_NO_RWG, i64, ptr, ptr, i32) DEF_HELPER_FLAGS_3(sve_orv_b, TCG_CALL_NO_RWG, i64, ptr, ptr, i32)
DEF_HELPER_FLAGS_3(sve_orv_h, TCG_CALL_NO_RWG, i64, ptr, ptr, i32) DEF_HELPER_FLAGS_3(sve_orv_h, TCG_CALL_NO_RWG, i64, ptr, ptr, i32)
DEF_HELPER_FLAGS_3(sve_orv_s, TCG_CALL_NO_RWG, i64, ptr, ptr, i32) DEF_HELPER_FLAGS_3(sve_orv_s, TCG_CALL_NO_RWG, i64, ptr, ptr, i32)

View file

@ -157,6 +157,12 @@ ASR_zpzz 00000100 .. 010 100 100 ... ..... ..... @rdm_pg_rn # ASRR
LSR_zpzz 00000100 .. 010 101 100 ... ..... ..... @rdm_pg_rn # LSRR LSR_zpzz 00000100 .. 010 101 100 ... ..... ..... @rdm_pg_rn # LSRR
LSL_zpzz 00000100 .. 010 111 100 ... ..... ..... @rdm_pg_rn # LSLR LSL_zpzz 00000100 .. 010 111 100 ... ..... ..... @rdm_pg_rn # LSLR
# SVE bitwise shift by wide elements (predicated)
# Note these require size != 3.
ASR_zpzw 00000100 .. 011 000 100 ... ..... ..... @rdn_pg_rm
LSR_zpzw 00000100 .. 011 001 100 ... ..... ..... @rdn_pg_rm
LSL_zpzw 00000100 .. 011 011 100 ... ..... ..... @rdn_pg_rm
### SVE Logical - Unpredicated Group ### SVE Logical - Unpredicated Group
# SVE bitwise logical operations (unpredicated) # SVE bitwise logical operations (unpredicated)

View file

@ -464,6 +464,41 @@ DO_ZPZZ_D(sve_lsl_zpzz_d, uint64_t, DO_LSL)
#undef DO_ZPZZ #undef DO_ZPZZ
#undef DO_ZPZZ_D #undef DO_ZPZZ_D
/* Three-operand expander, controlled by a predicate, in which the
* third operand is "wide". That is, for D = N op M, the same 64-bit
* value of M is used with all of the narrower values of N.
*/
#define DO_ZPZW(NAME, TYPE, TYPEW, H, OP) \
void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \
{ \
intptr_t i, opr_sz = simd_oprsz(desc); \
for (i = 0; i < opr_sz; ) { \
uint8_t pg = *(uint8_t *)(vg + H1(i >> 3)); \
TYPEW mm = *(TYPEW *)(vm + i); \
do { \
if (pg & 1) { \
TYPE nn = *(TYPE *)(vn + H(i)); \
*(TYPE *)(vd + H(i)) = OP(nn, mm); \
} \
i += sizeof(TYPE), pg >>= sizeof(TYPE); \
} while (i & 7); \
} \
}
DO_ZPZW(sve_asr_zpzw_b, int8_t, uint64_t, H1, DO_ASR)
DO_ZPZW(sve_lsr_zpzw_b, uint8_t, uint64_t, H1, DO_LSR)
DO_ZPZW(sve_lsl_zpzw_b, uint8_t, uint64_t, H1, DO_LSL)
DO_ZPZW(sve_asr_zpzw_h, int16_t, uint64_t, H1_2, DO_ASR)
DO_ZPZW(sve_lsr_zpzw_h, uint16_t, uint64_t, H1_2, DO_LSR)
DO_ZPZW(sve_lsl_zpzw_h, uint16_t, uint64_t, H1_2, DO_LSL)
DO_ZPZW(sve_asr_zpzw_s, int32_t, uint64_t, H1_4, DO_ASR)
DO_ZPZW(sve_lsr_zpzw_s, uint32_t, uint64_t, H1_4, DO_LSR)
DO_ZPZW(sve_lsl_zpzw_s, uint32_t, uint64_t, H1_4, DO_LSL)
#undef DO_ZPZW
/* Two-operand reduction expander, controlled by a predicate. /* Two-operand reduction expander, controlled by a predicate.
* The difference between TYPERED and TYPERET has to do with * The difference between TYPERED and TYPERET has to do with
* sign-extension. E.g. for SMAX, TYPERED must be signed, * sign-extension. E.g. for SMAX, TYPERED must be signed,

View file

@ -506,6 +506,30 @@ static bool trans_ASRD(DisasContext *s, arg_rpri_esz *a, uint32_t insn)
} }
} }
/*
*** SVE Bitwise Shift - Predicated Group
*/
#define DO_ZPZW(NAME, name) \
static bool trans_##NAME##_zpzw(DisasContext *s, arg_rprr_esz *a, \
uint32_t insn) \
{ \
static gen_helper_gvec_4 * const fns[3] = { \
gen_helper_sve_##name##_zpzw_b, gen_helper_sve_##name##_zpzw_h, \
gen_helper_sve_##name##_zpzw_s, \
}; \
if (a->esz < 0 || a->esz >= 3) { \
return false; \
} \
return do_zpzz_ool(s, a, fns[a->esz]); \
}
DO_ZPZW(ASR, asr)
DO_ZPZW(LSR, lsr)
DO_ZPZW(LSL, lsl)
#undef DO_ZPZW
/* /*
*** SVE Predicate Logical Operations Group *** SVE Predicate Logical Operations Group
*/ */