[Mesa-dev] [Mesa3d-dev] patches for various problems when playing with egl

[Ian Romanick]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 08/01/2011 10:40 AM, Ian Romanick wrote:
> On 07/31/2011 02:51 PM, kristof.ralovich at gmail.com wrote:
>> Please apply!
> 
> Send patches using git-send-email.  People cannot reply with review
> comments to patches sent as attachments.

And don't use mesa3d-dev at lists.sourceforge.net.  That list is dead.  Use
mesa-dev at freedesktop.org.

> patch 1:
> 
> +   if (!gbm)
> +   {
> +      free(dri2_dpy);
> +      return EGL_FALSE;
> +   }
> 
> should be
> 
> +   if (!gbm) {
> +      free(dri2_dpy);
> +      return EGL_FALSE;
> +   }
> 
> I think 'gbm == NULL' is the more common idiom in Mesa, but either way
> works for most people.
> 
> It may be even better to fold this hunk with the following if-statement:
> 
>     gbm = (struct gbm_device *) disp->PlatformDisplay;
> -   if (strcmp(gbm_device_get_backend_name(gbm), "drm") != 0) {
> +   if (!gbm || strcmp(gbm_device_get_backend_name(gbm), "drm") != 0) {
>        free(dri2_dpy);
>        return EGL_FALSE;
> 
> patch 2:
> 
> I don't think always linking with -lm is quite correct either.  The
> comment at the top says that backends should define GBM_LIBS, but then
> nothing under src/gdb uses GBM_LIBS.
> 
> Also, I think everyone hates the commit message style of starting a
> sentence in the "brief" line and continuing it in the body.  A better
> message is:
> 
> gbm: link gbm_gallium_drm.so against math library
> 
> This avoids the following runtime error with EGL on platforms that
> require linking with libm for nontrivial math functions:
> 
> failed to load module: /xorg/lib64/gbm/gbm_gallium_drm.so: undefined
> symbol: powf
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/

iEYEARECAAYFAk425TMACgkQX1gOwKyEAw8/4gCgg75nOjsK/QAjPxQG5kWc6ZB8
eE8An2auNAnwbn3HrswuGNBs/0WR6mqG
=1HFY
-----END PGP SIGNATURE-----