diff mbox series

[FFmpeg-devel] arm: Skip certain inline assembly functions if built without optimizations

Message ID 20220826083450.2867072-1-martin@martin.st
State Superseded
Headers show
Series [FFmpeg-devel] arm: Skip certain inline assembly functions if built without optimizations | expand

Commit Message

Martin Storsjö Aug. 26, 2022, 8:34 a.m. UTC
These inline assembly functions rely on being inlined into the
caller, so that the parameter "int p" can be a known assembly time
constant, instead of a variable parameter.

__OPTIMIZE__ is a built-in define which is set by both GCC and Clang
(the two main compilers supporting our inline assembly) when
optimizations are enabled.

This fixes building for arm targets with optimizations disabled.
---
With older GCC versions, the inline assembly in libavcodec/arm/aac.h
also might need to be disabled, but it does build successfully
with optimizations disabled with modern LLVM and GCC, so not touching
it immediately as part of this patch.
---
 libavutil/arm/intmath.h | 2 ++
 1 file changed, 2 insertions(+)

Comments

James Almer Aug. 26, 2022, 12:48 p.m. UTC | #1
On 8/26/2022 5:34 AM, Martin Storsjö wrote:
> These inline assembly functions rely on being inlined into the
> caller, so that the parameter "int p" can be a known assembly time
> constant, instead of a variable parameter.
> 
> __OPTIMIZE__ is a built-in define which is set by both GCC and Clang
> (the two main compilers supporting our inline assembly) when
> optimizations are enabled.
> 
> This fixes building for arm targets with optimizations disabled.

You could instead use av_builtin_constant_p() to check this at compile 
time. See how it's used for av_mod_uintp2_bmi2() in libavutil/x86/intmath.h

> ---
> With older GCC versions, the inline assembly in libavcodec/arm/aac.h
> also might need to be disabled, but it does build successfully
> with optimizations disabled with modern LLVM and GCC, so not touching
> it immediately as part of this patch.
> ---
>   libavutil/arm/intmath.h | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/libavutil/arm/intmath.h b/libavutil/arm/intmath.h
> index 5311a7d52b..67a8e72dc0 100644
> --- a/libavutil/arm/intmath.h
> +++ b/libavutil/arm/intmath.h
> @@ -62,6 +62,7 @@ static av_always_inline av_const int av_clip_int16_arm(int a)
>       return x;
>   }
>   
> +#ifdef __OPTIMIZE__
>   #define av_clip_intp2 av_clip_intp2_arm
>   static av_always_inline av_const int av_clip_intp2_arm(int a, int p)
>   {
> @@ -77,6 +78,7 @@ static av_always_inline av_const unsigned av_clip_uintp2_arm(int a, int p)
>       __asm__ ("usat %0, %2, %1" : "=r"(x) : "r"(a), "i"(p));
>       return x;
>   }
> +#endif
>   
>   #define av_sat_add32 av_sat_add32_arm
>   static av_always_inline int av_sat_add32_arm(int a, int b)
diff mbox series

Patch

diff --git a/libavutil/arm/intmath.h b/libavutil/arm/intmath.h
index 5311a7d52b..67a8e72dc0 100644
--- a/libavutil/arm/intmath.h
+++ b/libavutil/arm/intmath.h
@@ -62,6 +62,7 @@  static av_always_inline av_const int av_clip_int16_arm(int a)
     return x;
 }
 
+#ifdef __OPTIMIZE__
 #define av_clip_intp2 av_clip_intp2_arm
 static av_always_inline av_const int av_clip_intp2_arm(int a, int p)
 {
@@ -77,6 +78,7 @@  static av_always_inline av_const unsigned av_clip_uintp2_arm(int a, int p)
     __asm__ ("usat %0, %2, %1" : "=r"(x) : "r"(a), "i"(p));
     return x;
 }
+#endif
 
 #define av_sat_add32 av_sat_add32_arm
 static av_always_inline int av_sat_add32_arm(int a, int b)