target/avr: Optimize various functions using extract opcode

[PATCH] target/avr: Optimize various functions using extract opcode

Posted by Philippe Mathieu-Daudé 4 years, 4 months ago

When running the scripts/coccinelle/tcg_gen_extract.cocci
Coccinelle semantic patch on target/avr/, we get:

  [DBG] candidate at target/avr/translate.c:228
  [DBG] candidate at target/avr/translate.c:266
  [DBG] candidate at target/avr/translate.c:885
  [DBG] candidate at target/avr/translate.c:924
  [DBG] candidate at target/avr/translate.c:962

Manually inspect and replace combinations of (shri, andi)
opcodes by the extract opcode.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
---
 target/avr/translate.c | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)

diff --git a/target/avr/translate.c b/target/avr/translate.c
index 438e7b13c18..246cbfba1cd 100644
--- a/target/avr/translate.c
+++ b/target/avr/translate.c
@@ -225,8 +225,7 @@ static void gen_add_CHf(TCGv R, TCGv Rd, TCGv Rr)
     tcg_gen_or_tl(t1, t1, t3);
 
     tcg_gen_shri_tl(cpu_Cf, t1, 7); /* Cf = t1(7) */
-    tcg_gen_shri_tl(cpu_Hf, t1, 3); /* Hf = t1(3) */
-    tcg_gen_andi_tl(cpu_Hf, cpu_Hf, 1);
+    tcg_gen_extract_tl(cpu_Hf, t1, 3, 1); /* Hf = t1(3) */
 
     tcg_temp_free_i32(t3);
     tcg_temp_free_i32(t2);
@@ -263,8 +262,7 @@ static void gen_sub_CHf(TCGv R, TCGv Rd, TCGv Rr)
     tcg_gen_or_tl(t2, t2, t3); /* t2 = ~Rd & Rr | ~Rd & R | R & Rr */
 
     tcg_gen_shri_tl(cpu_Cf, t2, 7); /* Cf = t2(7) */
-    tcg_gen_shri_tl(cpu_Hf, t2, 3); /* Hf = t2(3) */
-    tcg_gen_andi_tl(cpu_Hf, cpu_Hf, 1);
+    tcg_gen_extract_tl(cpu_Hf, t2, 3, 1); /* Hf = t2(3) */
 
     tcg_temp_free_i32(t3);
     tcg_temp_free_i32(t2);
@@ -882,9 +880,7 @@ static bool trans_FMUL(DisasContext *ctx, arg_FMUL *a)
     /* update output registers */
     tcg_gen_shli_tl(R, R, 1);
     tcg_gen_andi_tl(R0, R, 0xff);
-    tcg_gen_shri_tl(R1, R, 8);
-    tcg_gen_andi_tl(R1, R1, 0xff);
-
+    tcg_gen_extract_tl(R1, R, 8, 8);
 
     tcg_temp_free_i32(R);
 
@@ -921,8 +917,7 @@ static bool trans_FMULS(DisasContext *ctx, arg_FMULS *a)
     /* update output registers */
     tcg_gen_shli_tl(R, R, 1);
     tcg_gen_andi_tl(R0, R, 0xff);
-    tcg_gen_shri_tl(R1, R, 8);
-    tcg_gen_andi_tl(R1, R1, 0xff);
+    tcg_gen_extract_tl(R1, R, 8, 8);
 
     tcg_temp_free_i32(t1);
     tcg_temp_free_i32(t0);
@@ -959,8 +954,7 @@ static bool trans_FMULSU(DisasContext *ctx, arg_FMULSU *a)
     /* update output registers */
     tcg_gen_shli_tl(R, R, 1);
     tcg_gen_andi_tl(R0, R, 0xff);
-    tcg_gen_shri_tl(R1, R, 8);
-    tcg_gen_andi_tl(R1, R1, 0xff);
+    tcg_gen_extract_tl(R1, R, 8, 8);
 
     tcg_temp_free_i32(t0);
     tcg_temp_free_i32(R);
-- 
2.31.1

Re: [PATCH] target/avr: Optimize various functions using extract opcode

Posted by Richard Henderson 4 years, 4 months ago

On 10/3/21 10:21 AM, Philippe Mathieu-Daudé wrote:
> When running the scripts/coccinelle/tcg_gen_extract.cocci
> Coccinelle semantic patch on target/avr/, we get:
> 
>    [DBG] candidate at target/avr/translate.c:228
>    [DBG] candidate at target/avr/translate.c:266
>    [DBG] candidate at target/avr/translate.c:885
>    [DBG] candidate at target/avr/translate.c:924
>    [DBG] candidate at target/avr/translate.c:962
> 
> Manually inspect and replace combinations of (shri, andi)
> opcodes by the extract opcode.
> 
> Signed-off-by: Philippe Mathieu-Daudé<f4bug@amsat.org>
> ---
>   target/avr/translate.c | 16 +++++-----------
>   1 file changed, 5 insertions(+), 11 deletions(-)

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

r~

Re: [PATCH] target/avr: Optimize various functions using extract opcode

Posted by Philippe Mathieu-Daudé 4 years, 3 months ago

Hi Richard,

On 10/3/21 17:24, Richard Henderson wrote:
> On 10/3/21 10:21 AM, Philippe Mathieu-Daudé wrote:
>> When running the scripts/coccinelle/tcg_gen_extract.cocci
>> Coccinelle semantic patch on target/avr/, we get:
>>
>>    [DBG] candidate at target/avr/translate.c:228
>>    [DBG] candidate at target/avr/translate.c:266
>>    [DBG] candidate at target/avr/translate.c:885
>>    [DBG] candidate at target/avr/translate.c:924
>>    [DBG] candidate at target/avr/translate.c:962
>>
>> Manually inspect and replace combinations of (shri, andi)
>> opcodes by the extract opcode.
>>
>> Signed-off-by: Philippe Mathieu-Daudé<f4bug@amsat.org>
>> ---
>>   target/avr/translate.c | 16 +++++-----------
>>   1 file changed, 5 insertions(+), 11 deletions(-)
> 
> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

Do you mind taking this patch via tcg-next?

Re: [PATCH] target/avr: Optimize various functions using extract opcode

Posted by Michael Rolnik 4 years, 3 months ago

hi Philippe

how was it tested?

On Wed, Oct 27, 2021 at 7:42 AM Philippe Mathieu-Daudé <f4bug@amsat.org>
wrote:

> Hi Richard,
>
> On 10/3/21 17:24, Richard Henderson wrote:
> > On 10/3/21 10:21 AM, Philippe Mathieu-Daudé wrote:
> >> When running the scripts/coccinelle/tcg_gen_extract.cocci
> >> Coccinelle semantic patch on target/avr/, we get:
> >>
> >>    [DBG] candidate at target/avr/translate.c:228
> >>    [DBG] candidate at target/avr/translate.c:266
> >>    [DBG] candidate at target/avr/translate.c:885
> >>    [DBG] candidate at target/avr/translate.c:924
> >>    [DBG] candidate at target/avr/translate.c:962
> >>
> >> Manually inspect and replace combinations of (shri, andi)
> >> opcodes by the extract opcode.
> >>
> >> Signed-off-by: Philippe Mathieu-Daudé<f4bug@amsat.org>
> >> ---
> >>   target/avr/translate.c | 16 +++++-----------
> >>   1 file changed, 5 insertions(+), 11 deletions(-)
> >
> > Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
>
> Do you mind taking this patch via tcg-next?
>


-- 
Best Regards,
Michael Rolnik