[PATCH 13/13] target/i386: avoid using s->tmp0 for add to implicit registers

Paolo Bonzini posted 13 patches 1 year, 1 month ago
[PATCH 13/13] target/i386: avoid using s->tmp0 for add to implicit registers
Posted by Paolo Bonzini 1 year, 1 month ago
For updates to implicit registers (RCX in LOOP instructions, RSI or RDI
in string instructions, or the stack pointer) do the add directly using
the registers (with no temporary) if 32-bit or 64-bit, or use a temporary
created for the occasion if 16-bit.  This is more efficient and removes
move instructions for the MO_TL case.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 target/i386/tcg/translate.c | 23 +++++++++++++++--------
 1 file changed, 15 insertions(+), 8 deletions(-)

diff --git a/target/i386/tcg/translate.c b/target/i386/tcg/translate.c
index 4b652cc23e1..8de506927b0 100644
--- a/target/i386/tcg/translate.c
+++ b/target/i386/tcg/translate.c
@@ -504,17 +504,24 @@ static inline void gen_op_jmp_v(DisasContext *s, TCGv dest)
     s->pc_save = -1;
 }
 
+static inline void gen_op_add_reg(DisasContext *s, MemOp size, int reg, TCGv val)
+{
+    /* Using cpu_regs[reg] does not work for xH registers.  */
+    assert(size >= MO_16);
+    if (size == MO_16) {
+        TCGv temp = tcg_temp_new();
+        tcg_gen_add_tl(temp, cpu_regs[reg], val);
+        gen_op_mov_reg_v(s, size, reg, temp);
+    } else {
+        tcg_gen_add_tl(cpu_regs[reg], cpu_regs[reg], val);
+        tcg_gen_ext_tl(cpu_regs[reg], cpu_regs[reg], size);
+    }
+}
+
 static inline
 void gen_op_add_reg_im(DisasContext *s, MemOp size, int reg, int32_t val)
 {
-    tcg_gen_addi_tl(s->tmp0, cpu_regs[reg], val);
-    gen_op_mov_reg_v(s, size, reg, s->tmp0);
-}
-
-static inline void gen_op_add_reg(DisasContext *s, MemOp size, int reg, TCGv val)
-{
-    tcg_gen_add_tl(s->tmp0, cpu_regs[reg], val);
-    gen_op_mov_reg_v(s, size, reg, s->tmp0);
+    gen_op_add_reg(s, size, reg, tcg_constant_tl(val));
 }
 
 static inline void gen_op_ld_v(DisasContext *s, int idx, TCGv t0, TCGv a0)
-- 
2.47.1
Re: [PATCH 13/13] target/i386: avoid using s->tmp0 for add to implicit registers
Posted by Richard Henderson 1 year, 1 month ago
On 12/15/24 03:06, Paolo Bonzini wrote:
> For updates to implicit registers (RCX in LOOP instructions, RSI or RDI
> in string instructions, or the stack pointer) do the add directly using
> the registers (with no temporary) if 32-bit or 64-bit, or use a temporary
> created for the occasion if 16-bit.  This is more efficient and removes
> move instructions for the MO_TL case.
> 
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>   target/i386/tcg/translate.c | 23 +++++++++++++++--------
>   1 file changed, 15 insertions(+), 8 deletions(-)
> 
> diff --git a/target/i386/tcg/translate.c b/target/i386/tcg/translate.c
> index 4b652cc23e1..8de506927b0 100644
> --- a/target/i386/tcg/translate.c
> +++ b/target/i386/tcg/translate.c
> @@ -504,17 +504,24 @@ static inline void gen_op_jmp_v(DisasContext *s, TCGv dest)
>       s->pc_save = -1;
>   }
>   
> +static inline void gen_op_add_reg(DisasContext *s, MemOp size, int reg, TCGv val)

Drop the inline.  Otherwise, yay!

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~