[PATCH v2 16/19] target/i386: move operand load and writeback out of gen_cmovcc1

[Paolo Bonzini]

Similar to gen_setcc1, make gen_cmovcc1 receive TCGv.  This is more friendly
to simultaneous implementation in the old and the new decoder.

A small wart is that s->T0 of CMOV is currently the *second* argument (which
would ordinarily be in T1).  Therefore, the condition as to be inverted in
order to overwrite s->T0 with cpu_regs[reg] if the MOV is not performed.

This only applies to the old decoder, and this code will go away soon.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 target/i386/tcg/translate.c | 20 ++++++++------------
 1 file changed, 8 insertions(+), 12 deletions(-)

diff --git a/target/i386/tcg/translate.c b/target/i386/tcg/translate.c
index 9c799b5a980..2c4e680a69e 100644
--- a/target/i386/tcg/translate.c
+++ b/target/i386/tcg/translate.c
@@ -2503,26 +2503,20 @@ static void gen_jcc(DisasContext *s, int b, int diff)
     gen_jmp_rel(s, s->dflag, diff, 0);
 }
 
-static void gen_cmovcc1(CPUX86State *env, DisasContext *s, MemOp ot, int b,
-                        int modrm, int reg)
+static void gen_cmovcc1(DisasContext *s, int b, TCGv dest, TCGv src)
 {
     CCPrepare cc;
 
-    gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
-
-    cc = gen_prepare_cc(s, b, s->T1);
+    cc = gen_prepare_cc(s, b, s->tmp4);
     if (cc.mask != -1) {
-        TCGv t0 = tcg_temp_new();
-        tcg_gen_andi_tl(t0, cc.reg, cc.mask);
-        cc.reg = t0;
+        tcg_gen_andi_tl(s->tmp4, cc.reg, cc.mask);
+        cc.reg = s->tmp4;
     }
     if (!cc.use_reg2) {
         cc.reg2 = tcg_constant_tl(cc.imm);
     }
 
-    tcg_gen_movcond_tl(cc.cond, s->T0, cc.reg, cc.reg2,
-                       s->T0, cpu_regs[reg]);
-    gen_op_mov_reg_v(s, ot, reg, s->T0);
+    tcg_gen_movcond_tl(cc.cond, dest, cc.reg, cc.reg2, src, dest);
 }
 
 static inline void gen_op_movl_T0_seg(DisasContext *s, X86Seg seg_reg)
@@ -5265,7 +5259,9 @@ static bool disas_insn(DisasContext *s, CPUState *cpu)
         ot = dflag;
         modrm = x86_ldub_code(env, s);
         reg = ((modrm >> 3) & 7) | REX_R(s);
-        gen_cmovcc1(env, s, ot, b, modrm, reg);
+        gen_ldst_modrm(env, s, modrm, ot, OR_TMP0, 0);
+        gen_cmovcc1(s, b ^ 1, s->T0, cpu_regs[reg]);
+        gen_op_mov_reg_v(s, ot, reg, s->T0);
         break;
 
         /************************/
-- 
2.41.0