[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PATCH 31/37] target/i386: reimplement 0x0f 0x28-0x2f, add AVX
From: |
Paolo Bonzini |
Subject: |
[PATCH 31/37] target/i386: reimplement 0x0f 0x28-0x2f, add AVX |
Date: |
Mon, 12 Sep 2022 01:04:11 +0200 |
Here the code is a bit uglier due to the truncation and extension
of registers to and from 32-bit. Otherwise there is nothing special
going on.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
target/i386/tcg/decode-new.c.inc | 54 ++++++++++++++
target/i386/tcg/emit.c.inc | 120 +++++++++++++++++++++++++++++++
target/i386/tcg/translate.c | 1 +
3 files changed, 175 insertions(+)
diff --git a/target/i386/tcg/decode-new.c.inc b/target/i386/tcg/decode-new.c.inc
index 268ccb886f..383a425ccd 100644
--- a/target/i386/tcg/decode-new.c.inc
+++ b/target/i386/tcg/decode-new.c.inc
@@ -620,6 +620,51 @@ static void decode_0F16(DisasContext *s, CPUX86State *env,
X86OpEntry *entry, ui
}
}
+static void decode_0F2A(DisasContext *s, CPUX86State *env, X86OpEntry *entry,
uint8_t *b)
+{
+ static const X86OpEntry opcodes_0F2A[4] = {
+ X86_OP_ENTRY3(CVTPI2Px, V,x, None,None, Q,q, vex4),
+ X86_OP_ENTRY3(CVTPI2Px, V,x, None,None, Q,q, vex4),
+ X86_OP_ENTRY3(VCVTSI2Sx, V,x, H,x, E,y, vex3),
+ X86_OP_ENTRY3(VCVTSI2Sx, V,x, H,x, E,y, vex3),
+ };
+ *entry = *decode_by_prefix(s, opcodes_0F2A);
+}
+
+static void decode_0F2B(DisasContext *s, CPUX86State *env, X86OpEntry *entry,
uint8_t *b)
+{
+ static const X86OpEntry opcodes_0F2B[4] = {
+ X86_OP_ENTRY3(MOVDQ, M,x, None,None, V,x, vex4), /* MOVNTPS */
+ X86_OP_ENTRY3(MOVDQ, M,x, None,None, V,x, vex4), /* MOVNTPD */
+ X86_OP_ENTRY3(VMOVSS_st, M,ss, None,None, V,x, vex4),
+ X86_OP_ENTRY3(VMOVLPx_st, M,sd, None,None, V,x, vex4), /* MOVSD */
+ };
+
+ *entry = *decode_by_prefix(s, opcodes_0F2B);
+}
+
+static void decode_0F2C(DisasContext *s, CPUX86State *env, X86OpEntry *entry,
uint8_t *b)
+{
+ static const X86OpEntry opcodes_0F2C[4] = {
+ X86_OP_ENTRY3(CVTTPx2PI, P,q, None,None, W,x, vex4),
+ X86_OP_ENTRY3(CVTTPx2PI, P,q, None,None, W,x, vex4),
+ X86_OP_ENTRY3(VCVTTSx2SI, G,y, None,None, W,x, vex3),
+ X86_OP_ENTRY3(VCVTTSx2SI, G,y, None,None, W,x, vex3),
+ };
+ *entry = *decode_by_prefix(s, opcodes_0F2C);
+}
+
+static void decode_0F2D(DisasContext *s, CPUX86State *env, X86OpEntry *entry,
uint8_t *b)
+{
+ static const X86OpEntry opcodes_0F2D[4] = {
+ X86_OP_ENTRY3(CVTPx2PI, P,q, None,None, W,x, vex4),
+ X86_OP_ENTRY3(CVTPx2PI, P,q, None,None, W,x, vex4),
+ X86_OP_ENTRY3(VCVTSx2SI, G,y, None,None, W,x, vex3),
+ X86_OP_ENTRY3(VCVTSx2SI, G,y, None,None, W,x, vex3),
+ };
+ *entry = *decode_by_prefix(s, opcodes_0F2D);
+}
+
static void decode_sse_unary(DisasContext *s, CPUX86State *env, X86OpEntry
*entry, uint8_t *b)
{
if (!(s->prefix & (PREFIX_REPZ | PREFIX_REPNZ))) {
@@ -672,6 +717,15 @@ static const X86OpEntry opcodes_0F[256] = {
[0x76] = X86_OP_ENTRY3(PCMPEQD, V,x, H,x, W,x, vex4 mmx avx2_256
p_00_66),
[0x77] = X86_OP_ENTRY0(EMMS_VZERO, vex8),
+ [0x28] = X86_OP_ENTRY3(MOVDQ, V,x, None,None, W,x, vex1 p_00_66), /*
MOVAPS */
+ [0x29] = X86_OP_ENTRY3(MOVDQ, W,x, None,None, V,x, vex1 p_00_66), /*
MOVAPS */
+ [0x2A] = X86_OP_GROUP0(0F2A),
+ [0x2B] = X86_OP_GROUP0(0F2B),
+ [0x2C] = X86_OP_GROUP0(0F2C),
+ [0x2D] = X86_OP_GROUP0(0F2D),
+ [0x2E] = X86_OP_ENTRY3(VUCOMI, None,None, V,x, W,x, vex4 p_00_66),
+ [0x2F] = X86_OP_ENTRY3(VCOMI, None,None, V,x, W,x, vex4 p_00_66),
+
[0x38] = X86_OP_GROUP0(0F38),
[0x3a] = X86_OP_GROUP0(0F3A),
diff --git a/target/i386/tcg/emit.c.inc b/target/i386/tcg/emit.c.inc
index 2319368cb5..d61b43f21c 100644
--- a/target/i386/tcg/emit.c.inc
+++ b/target/i386/tcg/emit.c.inc
@@ -921,6 +921,36 @@ static void gen_CRC32(DisasContext *s, CPUX86State *env,
X86DecodedInsn *decode)
gen_helper_crc32(s->T0, s->tmp2_i32, s->T1, tcg_const_i32(8 << ot));
}
+static void gen_CVTPI2Px(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ gen_helper_enter_mmx(cpu_env);
+ if (s->prefix & PREFIX_DATA) {
+ gen_helper_cvtpi2pd(cpu_env, s->ptr0, s->ptr2);
+ } else {
+ gen_helper_cvtpi2ps(cpu_env, s->ptr0, s->ptr2);
+ }
+}
+
+static void gen_CVTPx2PI(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ gen_helper_enter_mmx(cpu_env);
+ if (s->prefix & PREFIX_DATA) {
+ gen_helper_cvtpd2pi(cpu_env, s->ptr0, s->ptr2);
+ } else {
+ gen_helper_cvtps2pi(cpu_env, s->ptr0, s->ptr2);
+ }
+}
+
+static void gen_CVTTPx2PI(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ gen_helper_enter_mmx(cpu_env);
+ if (s->prefix & PREFIX_DATA) {
+ gen_helper_cvttpd2pi(cpu_env, s->ptr0, s->ptr2);
+ } else {
+ gen_helper_cvttps2pi(cpu_env, s->ptr0, s->ptr2);
+ }
+}
+
static void gen_EMMS_VZERO(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
{
if (!(s->prefix & PREFIX_VEX)) {
@@ -1863,6 +1893,14 @@ static inline void gen_VCMP(DisasContext *s, CPUX86State
*env, X86DecodedInsn *d
gen_helper_cmp_funcs[index][b](cpu_env, s->ptr0, s->ptr1, s->ptr2);
}
+static void gen_VCOMI(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ SSEFunc_0_epp fn;
+ fn = s->prefix & PREFIX_DATA ? gen_helper_comisd : gen_helper_comiss;
+ fn(cpu_env, s->ptr1, s->ptr2);
+ set_cc_op(s, CC_OP_EFLAGS);
+}
+
static void gen_VCVTfp2fp(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
{
gen_unary_fp_sse(s, env, decode,
@@ -1871,6 +1909,80 @@ static void gen_VCVTfp2fp(DisasContext *s, CPUX86State
*env, X86DecodedInsn *dec
gen_helper_cvtsd2ss, gen_helper_cvtss2sd);
}
+static void gen_VCVTSI2Sx(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ int vec_len = sse_vec_len(s, decode);
+ MemOp ot = decode->op[2].ot;
+ TCGv_i32 in;
+
+ tcg_gen_gvec_mov(MO_64, decode->op[0].offset, decode->op[1].offset,
vec_len, vec_len);
+#ifdef TARGET_X86_64
+ if (ot == MO_64) {
+ if (s->prefix & PREFIX_REPNZ) {
+ gen_helper_cvtsq2sd(cpu_env, s->ptr0, s->T1);
+ } else {
+ gen_helper_cvtsq2ss(cpu_env, s->ptr0, s->T1);
+ }
+ return;
+ }
+ in = s->tmp2_i32;
+ tcg_gen_trunc_tl_i32(in, s->T1);
+#else
+ in = s->T1;
+#endif
+
+ if (s->prefix & PREFIX_REPNZ) {
+ gen_helper_cvtsi2sd(cpu_env, s->ptr0, in);
+ } else {
+ gen_helper_cvtsi2ss(cpu_env, s->ptr0, in);
+ }
+}
+
+static inline void gen_VCVTtSx2SI(DisasContext *s, CPUX86State *env,
X86DecodedInsn *decode,
+ SSEFunc_i_ep ss2si, SSEFunc_l_ep ss2sq,
+ SSEFunc_i_ep sd2si, SSEFunc_l_ep sd2sq)
+{
+ MemOp ot = decode->op[0].ot;
+ TCGv_i32 out;
+
+#ifdef TARGET_X86_64
+ if (ot == MO_64) {
+ if (s->prefix & PREFIX_REPNZ) {
+ sd2sq(s->T0, cpu_env, s->ptr2);
+ } else {
+ ss2sq(s->T0, cpu_env, s->ptr2);
+ }
+ return;
+ }
+
+ out = s->tmp2_i32;
+#else
+ out = s->T0;
+#endif
+ if (s->prefix & PREFIX_REPNZ) {
+ sd2si(out, cpu_env, s->ptr2);
+ } else {
+ ss2si(out, cpu_env, s->ptr2);
+ }
+#ifdef TARGET_X86_64
+ tcg_gen_extu_i32_tl(s->T0, out);
+#endif
+}
+
+static void gen_VCVTSx2SI(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ gen_VCVTtSx2SI(s, env, decode,
+ gen_helper_cvtss2si, gen_helper_cvtss2sq,
+ gen_helper_cvtsd2si, gen_helper_cvtsd2sq);
+}
+
+static void gen_VCVTTSx2SI(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ gen_VCVTtSx2SI(s, env, decode,
+ gen_helper_cvttss2si, gen_helper_cvttss2sq,
+ gen_helper_cvttsd2si, gen_helper_cvttsd2sq);
+}
+
static void gen_VCVTpd_dq(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
{
SSEFunc_0_epp fn = NULL;
@@ -2162,4 +2274,12 @@ static inline void gen_VSHUF(DisasContext *s,
CPUX86State *env, X86DecodedInsn *
tcg_temp_free_i32(imm);
}
+static void gen_VUCOMI(DisasContext *s, CPUX86State *env, X86DecodedInsn
*decode)
+{
+ SSEFunc_0_epp fn;
+ fn = s->prefix & PREFIX_DATA ? gen_helper_ucomisd : gen_helper_ucomiss;
+ fn(cpu_env, s->ptr1, s->ptr2);
+ set_cc_op(s, CC_OP_EFLAGS);
+}
+
#define gen_VXOR gen_PXOR
diff --git a/target/i386/tcg/translate.c b/target/i386/tcg/translate.c
index bb5f74140c..f312663110 100644
--- a/target/i386/tcg/translate.c
+++ b/target/i386/tcg/translate.c
@@ -4669,6 +4669,7 @@ static target_ulong disas_insn(DisasContext *s, CPUState
*cpu)
if (use_new &&
(b == 0x138 || b == 0x13a ||
(b >= 0x110 && b <= 0x117) ||
+ (b >= 0x128 && b <= 0x12f) ||
(b >= 0x150 && b <= 0x17f) ||
b == 0x1c2 || (b >= 0x1c4 && b <= 0x1c6) ||
(b >= 0x1d0 && b <= 0x1ff))) {
--
2.37.2
- Re: [PATCH 27/37] target/i386: Use tcg gvec ops for pmovmskb, (continued)
[PATCH 26/37] target/i386: reimplement 0x0f 0x3a, add AVX, Paolo Bonzini, 2022/09/11
[PATCH 28/37] target/i386: reimplement 0x0f 0x38, add AVX, Paolo Bonzini, 2022/09/11
[PATCH 29/37] target/i386: reimplement 0x0f 0xc2, 0xc4-0xc6, add AVX, Paolo Bonzini, 2022/09/11
[PATCH 31/37] target/i386: reimplement 0x0f 0x28-0x2f, add AVX,
Paolo Bonzini <=
[PATCH 30/37] target/i386: reimplement 0x0f 0x10-0x17, add AVX, Paolo Bonzini, 2022/09/11
Re: [PATCH 30/37] target/i386: reimplement 0x0f 0x10-0x17, add AVX, Richard Henderson, 2022/09/13
[PATCH 32/37] target/i386: implement XSAVE and XRSTOR of AVX registers, Paolo Bonzini, 2022/09/11
[PATCH 33/37] target/i386: Enable AVX cpuid bits when using TCG, Paolo Bonzini, 2022/09/11