[Mesa-dev] [PATCH] gallium: reflect likely outcome in execution flow

[Ilia Mirkin]

On Wed, May 21, 2014 at 6:02 PM, Timothy Arceri <t_arceri at yahoo.com.au> wrote:
> Unless we run out of memory the old if statement would always fail so reflect the more likely outcome. Should be be faster most of the time and slightly cleaner looking code.
>
> Signed-off-by: Timothy Arceri <t_arceri at yahoo.com.au>
> ---
>  src/gallium/auxiliary/util/u_upload_mgr.c | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
>
> diff --git a/src/gallium/auxiliary/util/u_upload_mgr.c b/src/gallium/auxiliary/util/u_upload_mgr.c
> index 744ea2e..99f9a08 100644
> --- a/src/gallium/auxiliary/util/u_upload_mgr.c
> +++ b/src/gallium/auxiliary/util/u_upload_mgr.c
> @@ -247,11 +247,10 @@ enum pipe_error u_upload_data( struct u_upload_mgr *upload,
>     enum pipe_error ret = u_upload_alloc(upload, min_out_offset, size,
>                                          out_offset, outbuf,
>                                          (void**)&ptr);
> -   if (ret != PIPE_OK)
> -      return ret;
> +   if (ret == PIPE_OK)
> +      memcpy(ptr, data, size);
>
> -   memcpy(ptr, data, size);
> -   return PIPE_OK;
> +   return ret;
>  }

Have you actually compared the generated code? I find it can be fairly
instructive to do so... btw, there's a likely() (and unlikely() )
primitive (in p_compiler.h) which will indicate to gcc whether a
particular condition is likely or unlikely, so that it can adjust its
code generation accordingly.

  -ilia