[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [RFC PATCH v2 37/39] target/i386: introduce SSE code genera
From: |
Jan Bobek |
Subject: |
[Qemu-devel] [RFC PATCH v2 37/39] target/i386: introduce SSE code generators |
Date: |
Sat, 10 Aug 2019 00:12:53 -0400 |
Introduce code generators required by SSE instructions.
Signed-off-by: Jan Bobek <address@hidden>
---
target/i386/translate.c | 440 ++++++++++++++++++++++++++++++++++++++++
1 file changed, 440 insertions(+)
diff --git a/target/i386/translate.c b/target/i386/translate.c
index 12d2ac2eb5..681fa1aee2 100644
--- a/target/i386/translate.c
+++ b/target/i386/translate.c
@@ -4943,6 +4943,9 @@ INSNOP_LDST_UNIFY(Wdq, Udq, UdqMdq)
#define GEN_INSN_WR_GVEC_MM(mnem, gvec, opW1, opR1, vece) \
GEN_INSN_WR_GVEC(mnem, gvec, opW1, opR1, vece, \
sizeof(MMXReg), sizeof(MMXReg))
+#define GEN_INSN_WR_GVEC_XMM(mnem, gvec, opW1, opR1, vece) \
+ GEN_INSN_WR_GVEC(mnem, gvec, opW1, opR1, vece, \
+ sizeof(XMMReg), sizeof(XMMReg))
#define GEN_INSN_WRR_GVEC(mnem, gvec, opW1, opR1, opR2, vece, oprsz, maxsz) \
static void gen_insn_wrr(mnem, opW1, opR1, opR2)( \
@@ -4954,6 +4957,9 @@ INSNOP_LDST_UNIFY(Wdq, Udq, UdqMdq)
#define GEN_INSN_WRR_GVEC_MM(mnem, gvec, opW1, opR1, opR2, vece) \
GEN_INSN_WRR_GVEC(mnem, gvec, opW1, opR1, opR2, vece, \
sizeof(MMXReg), sizeof(MMXReg))
+#define GEN_INSN_WRR_GVEC_XMM(mnem, gvec, opW1, opR1, opR2, vece) \
+ GEN_INSN_WRR_GVEC(mnem, gvec, opW1, opR1, opR2, vece, \
+ sizeof(XMMReg), sizeof(XMMReg))
static void gen_insn_wr(movq, Eq, Pq)(CPUX86State *env, DisasContext *s,
insnop_t(Eq) ret, insnop_t(Pq) arg1)
@@ -4986,6 +4992,101 @@ static void gen_insn_wr(movd, Pq, Ed)(CPUX86State *env,
DisasContext *s,
GEN_INSN_WR_GVEC_MM(movq, mov, Pq, Qq, MO_64)
GEN_INSN_WR_GVEC_MM(movq, mov, Qq, Pq, MO_64)
+GEN_INSN_WR_GVEC_XMM(movaps, mov, Vdq, Wdq, MO_64)
+GEN_INSN_WR_GVEC_XMM(movaps, mov, Wdq, Vdq, MO_64)
+GEN_INSN_WR_GVEC_XMM(movups, mov, Vdq, Wdq, MO_64)
+GEN_INSN_WR_GVEC_XMM(movups, mov, Wdq, Vdq, MO_64)
+
+static void gen_insn_wr(movss, Wd, Vd)(CPUX86State *env, DisasContext *s,
+ insnop_t(Wd) ret, insnop_t(Vd) arg1)
+{
+ const size_t ofs = offsetof(ZMMReg, ZMM_L(0));
+ gen_op_movl(s, ret + ofs, arg1 + ofs);
+}
+
+static void gen_insn_wrr(movss, Vdq, Vdq, UdMd)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Vdq) ret,
+ insnop_t(Vdq) arg1,
+ insnop_t(UdMd) arg2)
+{
+ assert(ret == arg1);
+
+ if (arg2.is_mem) {
+ const size_t ofs0 = offsetof(ZMMReg, ZMM_Q(0));
+ const size_t ofs1 = offsetof(ZMMReg, ZMM_Q(1));
+
+ tcg_gen_movi_i64(s->tmp1_i64, 0);
+ tcg_gen_st_i64(s->tmp1_i64, cpu_env, ret + ofs0);
+ tcg_gen_st_i64(s->tmp1_i64, cpu_env, ret + ofs1);
+ }
+ gen_insn_wr(movss, Wd, Vd)(env, s, ret, arg2.op_reg);
+}
+
+static void gen_insn_wr(movlps, Mq, Vq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Mq) ret, insnop_t(Vq) arg1)
+{
+ assert(ret == s->A0);
+ gen_stq_env_A0(s, arg1 + offsetof(ZMMReg, ZMM_Q(0)));
+}
+
+static void gen_insn_wr(movhlps, Vq, UdqMq)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Vq) ret,
+ insnop_t(UdqMq) arg1)
+{
+ const size_t dofs = offsetof(ZMMReg, ZMM_Q(0));
+ const size_t aofs = offsetof(ZMMReg, ZMM_Q(arg1.is_mem ? 0 : 1));
+ gen_op_movq(s, ret + dofs, arg1.op_reg + aofs);
+}
+
+static void gen_insn_wr(movhps, Mq, Vdq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Mq) ret, insnop_t(Vdq) arg1)
+{
+ assert(ret == s->A0);
+ gen_stq_env_A0(s, arg1 + offsetof(ZMMReg, ZMM_Q(1)));
+}
+
+static void gen_insn_wrr(movlhps, Vdq, Vq, UqMq)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Vdq) ret,
+ insnop_t(Vq) arg1,
+ insnop_t(UqMq) arg2)
+{
+ assert(ret == arg1);
+
+ const size_t dofs = offsetof(ZMMReg, ZMM_Q(1));
+ const size_t aofs = offsetof(ZMMReg, ZMM_Q(0));
+ gen_op_movq(s, ret + dofs, arg2.op_reg + aofs);
+}
+
+static void gen_insn_wr(pmovmskb, Gd, Nq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gd) ret, insnop_t(Nq) arg1)
+{
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, arg1);
+ gen_helper_pmovmskb_mmx(ret, cpu_env, s->ptr0);
+}
+
+static void gen_insn_wr(pmovmskb, Gq, Nq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gq) ret, insnop_t(Nq) arg1)
+{
+ gen_insn_wr(pmovmskb, Gd, Nq)(env, s, s->tmp2_i32, arg1);
+ tcg_gen_extu_i32_i64(ret, s->tmp2_i32);
+}
+
+static void gen_insn_wr(movmskps, Gd, Udq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gd) ret, insnop_t(Udq)
arg1)
+{
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, arg1);
+ gen_helper_movmskps(ret, cpu_env, s->ptr0);
+}
+
+static void gen_insn_wr(movmskps, Gq, Udq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gq) ret, insnop_t(Udq)
arg1)
+{
+ gen_insn_wr(movmskps, Gd, Udq)(env, s, s->tmp2_i32, arg1);
+ tcg_gen_extu_i32_i64(ret, s->tmp2_i32);
+}
GEN_INSN_WRR_GVEC_MM(paddb, add, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(paddw, add, Pq, Pq, Qq, MO_16)
@@ -4994,6 +5095,8 @@ GEN_INSN_WRR_GVEC_MM(paddsb, ssadd, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(paddsw, ssadd, Pq, Pq, Qq, MO_16)
GEN_INSN_WRR_GVEC_MM(paddusb, usadd, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(paddusw, usadd, Pq, Pq, Qq, MO_16)
+GEN_INSN_WRR_HELPER(addps, addps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(addss, addss, Vd, Vd, Wd)
GEN_INSN_WRR_GVEC_MM(psubb, sub, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(psubw, sub, Pq, Pq, Qq, MO_16)
@@ -5002,11 +5105,39 @@ GEN_INSN_WRR_GVEC_MM(psubsb, sssub, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(psubsw, sssub, Pq, Pq, Qq, MO_16)
GEN_INSN_WRR_GVEC_MM(psubusb, ussub, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(psubusw, ussub, Pq, Pq, Qq, MO_16)
+GEN_INSN_WRR_HELPER(subps, subps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(subss, subss, Vd, Vd, Wd)
GEN_INSN_WRR_HELPER(pmulhw, pmulhw_mmx, Pq, Pq, Qq)
GEN_INSN_WRR_HELPER(pmullw, pmullw_mmx, Pq, Pq, Qq)
+GEN_INSN_WRR_HELPER(pmulhuw, pmulhuw_mmx, Pq, Pq, Qq)
+GEN_INSN_WRR_HELPER(mulps, mulps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(mulss, mulss, Vd, Vd, Wd)
+
GEN_INSN_WRR_HELPER(pmaddwd, pmaddwd_mmx, Pq, Pq, Qq)
+GEN_INSN_WRR_HELPER(divps, divps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(divss, divss, Vd, Vd, Wd)
+
+GEN_INSN_WR_HELPER(rcpps, rcpps, Vdq, Wdq)
+GEN_INSN_WR_HELPER(rcpss, rcpss, Vd, Wd)
+GEN_INSN_WR_HELPER(sqrtps, sqrtps, Vdq, Wdq)
+GEN_INSN_WR_HELPER(sqrtss, sqrtss, Vd, Wd)
+GEN_INSN_WR_HELPER(rsqrtps, rsqrtps, Vdq, Wdq)
+GEN_INSN_WR_HELPER(rsqrtss, rsqrtss, Vd, Wd)
+
+GEN_INSN_WRR_GVEC_MM(pminub, umin, Pq, Pq, Qq, MO_8)
+GEN_INSN_WRR_GVEC_MM(pminsw, smin, Pq, Pq, Qq, MO_16)
+GEN_INSN_WRR_HELPER(minps, minps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(minss, minss, Vd, Vd, Wd)
+GEN_INSN_WRR_GVEC_MM(pmaxub, umax, Pq, Pq, Qq, MO_8)
+GEN_INSN_WRR_GVEC_MM(pmaxsw, smax, Pq, Pq, Qq, MO_16)
+GEN_INSN_WRR_HELPER(maxps, maxps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(maxss, maxss, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(pavgb, pavgb_mmx, Pq, Pq, Qq)
+GEN_INSN_WRR_HELPER(pavgw, pavgw_mmx, Pq, Pq, Qq)
+GEN_INSN_WRR_HELPER(psadbw, psadbw_mmx, Pq, Pq, Qq)
+
GEN_INSN_WRR_GVEC_MM(pcmpeqb, cmpeq, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(pcmpeqw, cmpeq, Pq, Pq, Qq, MO_16)
GEN_INSN_WRR_GVEC_MM(pcmpeqd, cmpeq, Pq, Pq, Qq, MO_32)
@@ -5014,10 +5145,108 @@ GEN_INSN_WRR_GVEC_MM(pcmpgtb, cmpgt, Pq, Pq, Qq, MO_8)
GEN_INSN_WRR_GVEC_MM(pcmpgtw, cmpgt, Pq, Pq, Qq, MO_16)
GEN_INSN_WRR_GVEC_MM(pcmpgtd, cmpgt, Pq, Pq, Qq, MO_32)
+GEN_INSN_WRR_HELPER(cmpeqps, cmpeqps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpeqss, cmpeqss, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(cmpltps, cmpltps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpltss, cmpltss, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(cmpleps, cmpleps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpless, cmpless, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(cmpunordps, cmpunordps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpunordss, cmpunordss, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(cmpneqps, cmpneqps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpneqss, cmpneqss, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(cmpnltps, cmpnltps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpnltss, cmpnltss, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(cmpnleps, cmpnleps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpnless, cmpnless, Vd, Vd, Wd)
+GEN_INSN_WRR_HELPER(cmpordps, cmpordps, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(cmpordss, cmpordss, Vd, Vd, Wd)
+
+static void gen_insn_wrrr(cmpps, Vdq, Vdq, Wdq, Ib)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Vdq) ret,
+ insnop_t(Vdq) arg1,
+ insnop_t(Wdq) arg2,
+ insnop_t(Ib) arg3)
+{
+ switch (arg3 & 7) {
+ case 0:
+ gen_insn_wrr(cmpeqps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ case 1:
+ gen_insn_wrr(cmpltps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ case 2:
+ gen_insn_wrr(cmpleps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ case 3:
+ gen_insn_wrr(cmpunordps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ case 4:
+ gen_insn_wrr(cmpneqps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ case 5:
+ gen_insn_wrr(cmpnltps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ case 6:
+ gen_insn_wrr(cmpnleps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ case 7:
+ gen_insn_wrr(cmpordps, Vdq, Vdq, Wdq)(env, s, ret, arg1, arg2);
+ return;
+ }
+
+ g_assert_not_reached();
+}
+
+static void gen_insn_wrrr(cmpss, Vd, Vd, Wd, Ib)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Vd) ret,
+ insnop_t(Vd) arg1,
+ insnop_t(Wd) arg2,
+ insnop_t(Ib) arg3)
+{
+ switch (arg3 & 7) {
+ case 0:
+ gen_insn_wrr(cmpeqss, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ case 1:
+ gen_insn_wrr(cmpltss, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ case 2:
+ gen_insn_wrr(cmpless, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ case 3:
+ gen_insn_wrr(cmpunordss, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ case 4:
+ gen_insn_wrr(cmpneqss, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ case 5:
+ gen_insn_wrr(cmpnltss, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ case 6:
+ gen_insn_wrr(cmpnless, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ case 7:
+ gen_insn_wrr(cmpordss, Vd, Vd, Wd)(env, s, ret, arg1, arg2);
+ return;
+ }
+
+ g_assert_not_reached();
+}
+
+GEN_INSN_WR_HELPER(comiss, comiss, Vd, Wd)
+GEN_INSN_WR_HELPER(ucomiss, ucomiss, Vd, Wd)
+
GEN_INSN_WRR_GVEC_MM(pand, and, Pq, Pq, Qq, MO_64)
+GEN_INSN_WRR_GVEC_XMM(andps, and, Vdq, Vdq, Wdq, MO_64)
GEN_INSN_WRR_GVEC_MM(pandn, andn, Pq, Pq, Qq, MO_64)
+GEN_INSN_WRR_GVEC_XMM(andnps, andn, Vdq, Vdq, Wdq, MO_64)
GEN_INSN_WRR_GVEC_MM(por, or, Pq, Pq, Qq, MO_64)
+GEN_INSN_WRR_GVEC_XMM(orps, or, Vdq, Vdq, Wdq, MO_64)
GEN_INSN_WRR_GVEC_MM(pxor, xor, Pq, Pq, Qq, MO_64)
+GEN_INSN_WRR_GVEC_XMM(xorps, xor, Vdq, Vdq, Wdq, MO_64)
GEN_INSN_WRR_HELPER(psllw, psllw_mmx, Pq, Pq, Qq)
GEN_INSN_WRR_HELPER(pslld, pslld_mmx, Pq, Pq, Qq)
@@ -5061,8 +5290,219 @@ GEN_INSN_WRR_HELPER(punpckhbw, punpckhbw_mmx, Pq, Pq,
Qq)
GEN_INSN_WRR_HELPER(punpckhwd, punpckhwd_mmx, Pq, Pq, Qq)
GEN_INSN_WRR_HELPER(punpckhdq, punpckhdq_mmx, Pq, Pq, Qq)
+GEN_INSN_WRR_HELPER(unpcklps, punpckldq_xmm, Vdq, Vdq, Wdq)
+GEN_INSN_WRR_HELPER(unpckhps, punpckhdq_xmm, Vdq, Vdq, Wdq)
+
GEN_INSN_HELPER(emms, emms)
+static void gen_insn_wrr(pshufw, Pq, Qq, Ib)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Pq) ret,
+ insnop_t(Qq) arg1,
+ insnop_t(Ib) arg2)
+{
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, ret);
+ tcg_gen_addi_ptr(s->ptr1, cpu_env, arg1);
+ tcg_gen_movi_i32(s->tmp2_i32, arg2);
+ gen_helper_pshufw_mmx(s->ptr0, s->ptr1, s->tmp2_i32);
+}
+
+static void gen_insn_wrrr(shufps, Vdq, Vdq, Wdq, Ib)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Vdq) ret,
+ insnop_t(Vdq) arg1,
+ insnop_t(Wdq) arg2,
+ insnop_t(Ib) arg3)
+{
+ assert(ret == arg1);
+
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, ret);
+ tcg_gen_addi_ptr(s->ptr1, cpu_env, arg2);
+ tcg_gen_movi_i32(s->tmp2_i32, arg3);
+ gen_helper_shufps(s->ptr0, s->ptr1, s->tmp2_i32);
+}
+
+static void gen_insn_wrrr(pinsrw, Pq, Pq, RdMw, Ib)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Pq) ret,
+ insnop_t(Pq) arg1,
+ insnop_t(RdMw) arg2,
+ insnop_t(Ib) arg3)
+{
+ assert(ret == arg1);
+
+ const size_t ofs = offsetof(MMXReg, MMX_W(arg3 & 3));
+ tcg_gen_st16_i32(arg2.op_reg, cpu_env, ret + ofs);
+}
+
+static void gen_insn_wrr(pextrw, Gd, Nq, Ib)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Gd) ret,
+ insnop_t(Nq) arg1,
+ insnop_t(Ib) arg2)
+{
+ const size_t ofs = offsetof(MMXReg, MMX_W(arg2 & 3));
+ tcg_gen_ld16u_i32(ret, cpu_env, arg1 + ofs);
+}
+
+static void gen_insn_wrr(pextrw, Gq, Nq, Ib)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Gq) ret,
+ insnop_t(Nq) arg1,
+ insnop_t(Ib) arg2)
+{
+ const size_t ofs = offsetof(MMXReg, MMX_W(arg2 & 3));
+ tcg_gen_ld16u_i64(ret, cpu_env, arg1 + ofs);
+}
+
+GEN_INSN_WR_HELPER(cvtpi2ps, cvtpi2ps, Vdq, Qq)
+
+static void gen_insn_wr(cvtsi2ss, Vd, Ed)(CPUX86State *env, DisasContext *s,
+ insnop_t(Vd) ret, insnop_t(Ed) arg1)
+{
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, ret);
+ gen_helper_cvtsi2ss(cpu_env, s->ptr0, arg1);
+}
+
+static void gen_insn_wr(cvtsi2ss, Vd, Eq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Vd) ret, insnop_t(Eq) arg1)
+{
+#ifdef TARGET_X86_64
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, ret);
+ gen_helper_cvtsq2ss(cpu_env, s->ptr0, arg1);
+#else /* TARGET_X86_64 */
+ g_assert_not_reached();
+#endif /* TARGET_X86_64 */
+}
+
+GEN_INSN_WR_HELPER(cvtps2pi, cvtps2pi, Pq, Wq)
+
+static void gen_insn_wr(cvtss2si, Gd, Wd)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gd) ret, insnop_t(Wd) arg1)
+{
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, arg1);
+ gen_helper_cvtss2si(ret, cpu_env, s->ptr0);
+}
+
+static void gen_insn_wr(cvtss2si, Gq, Wd)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gq) ret, insnop_t(Wd) arg1)
+{
+#ifdef TARGET_X86_64
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, arg1);
+ gen_helper_cvtss2sq(ret, cpu_env, s->ptr0);
+#else /* TARGET_X86_64 */
+ g_assert_not_reached();
+#endif /* TARGET_X86_64 */
+}
+
+GEN_INSN_WR_HELPER(cvttps2pi, cvttps2pi, Pq, Wq)
+
+static void gen_insn_wr(cvttss2si, Gd, Wd)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gd) ret, insnop_t(Wd) arg1)
+{
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, arg1);
+ gen_helper_cvttss2si(ret, cpu_env, s->ptr0);
+}
+
+static void gen_insn_wr(cvttss2si, Gq, Wd)(CPUX86State *env, DisasContext *s,
+ insnop_t(Gq) ret, insnop_t(Wd) arg1)
+{
+#ifdef TARGET_X86_64
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, arg1);
+ gen_helper_cvttss2sq(ret, cpu_env, s->ptr0);
+#else /* TARGET_X86_64 */
+ g_assert_not_reached();
+#endif /* TARGET_X86_64 */
+}
+
+static void gen_insn_wr(maskmovq, Pq, Nq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Pq) ret, insnop_t(Nq) arg1)
+{
+ tcg_gen_mov_tl(s->A0, cpu_regs[R_EDI]);
+ gen_extu(s->aflag, s->A0);
+ gen_add_A0_ds_seg(s);
+
+ tcg_gen_addi_ptr(s->ptr0, cpu_env, ret);
+ tcg_gen_addi_ptr(s->ptr1, cpu_env, arg1);
+ gen_helper_maskmov_mmx(cpu_env, s->ptr0, s->ptr1, s->A0);
+}
+
+static void gen_insn_wr(movntq, Mq, Pq)(CPUX86State *env, DisasContext *s,
+ insnop_t(Mq) ret, insnop_t(Pq) arg1)
+{
+ assert(ret == s->A0);
+ gen_stq_env_A0(s, arg1);
+}
+
+static void gen_insn_wr(movntps, Mdq, Vdq)(CPUX86State *env,
+ DisasContext *s,
+ insnop_t(Mdq) ret,
+ insnop_t(Vdq) arg1)
+{
+ assert(ret == s->A0);
+ gen_sto_env_A0(s, arg1);
+}
+
+static void gen_insn(sfence)(CPUX86State *env, DisasContext *s)
+{
+ if (s->prefix & PREFIX_LOCK) {
+ gen_illegal_opcode(s);
+ } else {
+ tcg_gen_mb(TCG_MO_ST_ST | TCG_BAR_SC);
+ }
+}
+
+static void gen_insn_r(ldmxcsr, Md)(CPUX86State *env, DisasContext *s,
+ insnop_t(Md) arg1)
+{
+ if ((s->flags & HF_EM_MASK) || !(s->flags & HF_OSFXSR_MASK)) {
+ gen_illegal_opcode(s);
+ } else if (s->flags & HF_TS_MASK) {
+ gen_exception(s, EXCP07_PREX);
+ } else {
+ tcg_gen_qemu_ld_i32(s->tmp2_i32, arg1, s->mem_index, MO_LEUL);
+ gen_helper_ldmxcsr(cpu_env, s->tmp2_i32);
+ }
+}
+
+static void gen_insn_w(stmxcsr, Md)(CPUX86State *env, DisasContext *s,
+ insnop_t(Md) ret)
+{
+ if ((s->flags & HF_EM_MASK) || !(s->flags & HF_OSFXSR_MASK)) {
+ gen_illegal_opcode(s);
+ } else if (s->flags & HF_TS_MASK) {
+ gen_exception(s, EXCP07_PREX);
+ } else {
+ const size_t ofs = offsetof(CPUX86State, mxcsr);
+ tcg_gen_ld_i32(s->tmp2_i32, cpu_env, ofs);
+ tcg_gen_qemu_st_i32(s->tmp2_i32, ret, s->mem_index, MO_LEUL);
+ }
+}
+
+static void gen_insn_r(prefetcht0, Mb)(CPUX86State *env, DisasContext *s,
+ insnop_t(Mb) arg1)
+{
+ /* no-op */
+}
+
+static void gen_insn_r(prefetcht1, Mb)(CPUX86State *env, DisasContext *s,
+ insnop_t(Mb) arg1)
+{
+ /* no-op */
+}
+
+static void gen_insn_r(prefetcht2, Mb)(CPUX86State *env, DisasContext *s,
+ insnop_t(Mb) arg1)
+{
+ /* no-op */
+}
+
+static void gen_insn_r(prefetchnta, Mb)(CPUX86State *env, DisasContext *s,
+ insnop_t(Mb) arg1)
+{
+ /* no-op */
+}
+
/*
* Instruction translators
*/
--
2.20.1
- [Qemu-devel] [RFC PATCH v2 29/39] target/i386: introduce helper-based code generator macros, (continued)
- [Qemu-devel] [RFC PATCH v2 29/39] target/i386: introduce helper-based code generator macros, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 28/39] target/i386: introduce P*, N*, Q* (MMX) operands, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 30/39] target/i386: introduce gvec-based code generator macros, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 31/39] target/i386: introduce MMX translators, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 32/39] target/i386: introduce MMX code generators, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 33/39] target/i386: introduce MMX instructions to insn.h, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 34/39] target/i386: introduce V*, U*, W* (SSE/AVX) operands, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 35/39] target/i386: introduce UdqMq operand, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 36/39] target/i386: introduce SSE translators, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 39/39] target/i386: introduce memory-pointer operand read/write workarounds, Jan Bobek, 2019/08/10
- [Qemu-devel] [RFC PATCH v2 37/39] target/i386: introduce SSE code generators,
Jan Bobek <=
- [Qemu-devel] [RFC PATCH v2 38/39] target/i386: introduce SSE instructions to insn.h, Jan Bobek, 2019/08/10
- Re: [Qemu-devel] [RFC PATCH v2 00/39] rewrite MMX/SSE instruction translation, no-reply, 2019/08/10
- Re: [Qemu-devel] [RFC PATCH v2 00/39] rewrite MMX/SSE instruction translation, Richard Henderson, 2019/08/10