diff options
Diffstat (limited to 'tcg/i386/tcg-target.c.inc')
| -rw-r--r-- | tcg/i386/tcg-target.c.inc | 112 |
1 files changed, 67 insertions, 45 deletions
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index 46e856f442..050f3cb0b1 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -942,8 +942,8 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece, return true; } -static void tcg_out_dupi_vec(TCGContext *s, TCGType type, - TCGReg ret, tcg_target_long arg) +static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece, + TCGReg ret, int64_t arg) { int vex_l = (type == TCG_TYPE_V256 ? P_VEXL : 0); @@ -956,7 +956,14 @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, return; } - if (TCG_TARGET_REG_BITS == 64) { + if (TCG_TARGET_REG_BITS == 32 && vece < MO_64) { + if (have_avx2) { + tcg_out_vex_modrm_pool(s, OPC_VPBROADCASTD + vex_l, ret); + } else { + tcg_out_vex_modrm_pool(s, OPC_VBROADCASTSS, ret); + } + new_pool_label(s, arg, R_386_32, s->code_ptr - 4, 0); + } else { if (type == TCG_TYPE_V64) { tcg_out_vex_modrm_pool(s, OPC_MOVQ_VqWq, ret); } else if (have_avx2) { @@ -964,40 +971,39 @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, } else { tcg_out_vex_modrm_pool(s, OPC_MOVDDUP, ret); } - new_pool_label(s, arg, R_386_PC32, s->code_ptr - 4, -4); - } else { - if (have_avx2) { - tcg_out_vex_modrm_pool(s, OPC_VPBROADCASTD + vex_l, ret); + if (TCG_TARGET_REG_BITS == 64) { + new_pool_label(s, arg, R_386_PC32, s->code_ptr - 4, -4); } else { - tcg_out_vex_modrm_pool(s, OPC_VBROADCASTSS, ret); + new_pool_l2(s, R_386_32, s->code_ptr - 4, 0, arg, arg >> 32); } - new_pool_label(s, arg, R_386_32, s->code_ptr - 4, 0); } } -static void tcg_out_movi(TCGContext *s, TCGType type, - TCGReg ret, tcg_target_long arg) +static void tcg_out_movi_vec(TCGContext *s, TCGType type, + TCGReg ret, tcg_target_long arg) { - tcg_target_long diff; - - switch (type) { - case TCG_TYPE_I32: -#if TCG_TARGET_REG_BITS == 64 - case TCG_TYPE_I64: -#endif - if (ret < 16) { - break; - } - /* fallthru */ - case TCG_TYPE_V64: - case TCG_TYPE_V128: - case TCG_TYPE_V256: - tcg_debug_assert(ret >= 16); - tcg_out_dupi_vec(s, type, ret, arg); + if (arg == 0) { + tcg_out_vex_modrm(s, OPC_PXOR, ret, ret, ret); return; - default: - g_assert_not_reached(); } + if (arg == -1) { + tcg_out_vex_modrm(s, OPC_PCMPEQB, ret, ret, ret); + return; + } + + int rexw = (type == TCG_TYPE_I32 ? 0 : P_REXW); + tcg_out_vex_modrm_pool(s, OPC_MOVD_VyEy + rexw, ret); + if (TCG_TARGET_REG_BITS == 64) { + new_pool_label(s, arg, R_386_PC32, s->code_ptr - 4, -4); + } else { + new_pool_label(s, arg, R_386_32, s->code_ptr - 4, 0); + } +} + +static void tcg_out_movi_int(TCGContext *s, TCGType type, + TCGReg ret, tcg_target_long arg) +{ + tcg_target_long diff; if (arg == 0) { tgen_arithr(s, ARITH_XOR, ret, ret); @@ -1027,6 +1033,25 @@ static void tcg_out_movi(TCGContext *s, TCGType type, tcg_out64(s, arg); } +static void tcg_out_movi(TCGContext *s, TCGType type, + TCGReg ret, tcg_target_long arg) +{ + switch (type) { + case TCG_TYPE_I32: +#if TCG_TARGET_REG_BITS == 64 + case TCG_TYPE_I64: +#endif + if (ret < 16) { + tcg_out_movi_int(s, type, ret, arg); + } else { + tcg_out_movi_vec(s, type, ret, arg); + } + break; + default: + g_assert_not_reached(); + } +} + static inline void tcg_out_pushi(TCGContext *s, tcg_target_long val) { if (val == (int8_t)val) { @@ -2641,8 +2666,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); @@ -2928,7 +2951,6 @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ - case INDEX_op_dupi_vec: /* Always emitted via tcg_out_movi. */ case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ default: g_assert_not_reached(); @@ -3464,7 +3486,7 @@ static void expand_vec_rotv(TCGType type, unsigned vece, TCGv_vec v0, static void expand_vec_mul(TCGType type, unsigned vece, TCGv_vec v0, TCGv_vec v1, TCGv_vec v2) { - TCGv_vec t1, t2, t3, t4; + TCGv_vec t1, t2, t3, t4, zero; tcg_debug_assert(vece == MO_8); @@ -3482,11 +3504,11 @@ static void expand_vec_mul(TCGType type, unsigned vece, case TCG_TYPE_V64: t1 = tcg_temp_new_vec(TCG_TYPE_V128); t2 = tcg_temp_new_vec(TCG_TYPE_V128); - tcg_gen_dup16i_vec(t2, 0); + zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0); vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8, - tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t2)); + tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero)); vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8, - tcgv_vec_arg(t2), tcgv_vec_arg(t2), tcgv_vec_arg(v2)); + tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2)); tcg_gen_mul_vec(MO_16, t1, t1, t2); tcg_gen_shri_vec(MO_16, t1, t1, 8); vec_gen_3(INDEX_op_x86_packus_vec, TCG_TYPE_V128, MO_8, @@ -3501,15 +3523,15 @@ static void expand_vec_mul(TCGType type, unsigned vece, t2 = tcg_temp_new_vec(type); t3 = tcg_temp_new_vec(type); t4 = tcg_temp_new_vec(type); - tcg_gen_dup16i_vec(t4, 0); + zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0); vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8, - tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t4)); + tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero)); vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8, - tcgv_vec_arg(t2), tcgv_vec_arg(t4), tcgv_vec_arg(v2)); + tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2)); vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8, - tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(t4)); + tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(zero)); vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8, - tcgv_vec_arg(t4), tcgv_vec_arg(t4), tcgv_vec_arg(v2)); + tcgv_vec_arg(t4), tcgv_vec_arg(zero), tcgv_vec_arg(v2)); tcg_gen_mul_vec(MO_16, t1, t1, t2); tcg_gen_mul_vec(MO_16, t3, t3, t4); tcg_gen_shri_vec(MO_16, t1, t1, 8); @@ -3537,7 +3559,7 @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0, NEED_UMIN = 8, NEED_UMAX = 16, }; - TCGv_vec t1, t2; + TCGv_vec t1, t2, t3; uint8_t fixup; switch (cond) { @@ -3608,9 +3630,9 @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0, } else if (fixup & NEED_BIAS) { t1 = tcg_temp_new_vec(type); t2 = tcg_temp_new_vec(type); - tcg_gen_dupi_vec(vece, t2, 1ull << ((8 << vece) - 1)); - tcg_gen_sub_vec(vece, t1, v1, t2); - tcg_gen_sub_vec(vece, t2, v2, t2); + t3 = tcg_constant_vec(type, vece, 1ull << ((8 << vece) - 1)); + tcg_gen_sub_vec(vece, t1, v1, t3); + tcg_gen_sub_vec(vece, t2, v2, t3); v1 = t1; v2 = t2; cond = tcg_signed_cond(cond); |