fix for glXGetProcAddress of egl functions in glx context

fix for glXGetProcAddress of egl functions in glx context
When using latest nvidia blob drivers on linux (width glvnd enabled) skia
crashes trying to cal eglQueryString in glx context while creating new opengl
target.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1707283002

[danvd, 2016-02-18 09:06:45]

Description was changed from

==========
fix for glXGetProcAddress of egl functions in glx context

BUG=skia:
==========

to

==========
fix for glXGetProcAddress of egl functions in glx context

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[danvd, 2016-02-18 09:08:25]

The CQ bit was checked by andrewderyabin@gmail.com

[danvd, 2016-02-18 09:08:26]

The CQ bit was unchecked by andrewderyabin@gmail.com

[danvd, 2016-02-18 09:11:19]

When using latest nvidia blob drivers on linux (width glvnd enabled) skia
crashes trying to cal eglQueryString in glx context while creating new opengl
target.
On 2016/02/18 09:06:45, andrewderyabin wrote:
> Description was changed from
> 
> ==========
> fix for glXGetProcAddress of egl functions in glx context
> 
> BUG=skia:
> ==========
> 
> to
> 
> ==========
> fix for glXGetProcAddress of egl functions in glx context
> 
> BUG=skia:
> GOLD_TRYBOT_URL=
>
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
> ==========

[danvd, 2016-02-18 09:17:20]

Description was changed from

==========
fix for glXGetProcAddress of egl functions in glx context

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
fix for glXGetProcAddress of egl functions in glx context
When using latest nvidia blob drivers on linux (width glvnd enabled) skia
crashes trying to cal eglQueryString in glx context while creating new opengl
target.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[danvd, 2016-02-18 09:17:20]

andrewderyabin@gmail.com changed reviewers:
+ bsalomon@google.com, halcanary@chromium.org, halcanary@google.com

[danvd, 2016-02-18 09:23:56]

Hi!
I'm on arch linux. After recent nvidia blob driver update my skia-based app
stopped working.

after making a call:
SkAutoTUnref<const GrGLInterface> interface(GrGLCreateNativeInterface());

there is a segmentation fault. After some research I found that there is a
function get_egl_query_and_display (GrGLAssembleInterface.cpp) which is called
even in glx context. 
After some time, in GrGLExtensions::init (GrGLExtensions.cpp) there is a line:
const char* extensions = queryString(eglDisplay, GR_EGL_EXTENSIONS);
It returns garbage in <extensions> variable, and as a reason next call to
eat_space_sep_strings produces a segfault. 

My patch discards all calls to glxGetProcAddress of egl-prefixed function names
and fixed described behavior. 

Thanks!

Regards,
Andrew

[hal.canary, 2016-02-18 15:45:04]

halcanary@google.com changed reviewers:
- halcanary@chromium.org, halcanary@google.com

[hal.canary, 2016-02-18 15:46:18]

I'm going to to take myself off this CL's reviewer list, since I don't
understand why things were the way they were.  bsalomon@ is the right person to
look at this.

[bsalomon, 2016-02-18 16:11:47]

On 2016/02/18 09:23:56, danvd wrote:
> Hi!
> I'm on arch linux. After recent nvidia blob driver update my skia-based app
> stopped working.
> 
> after making a call:
> SkAutoTUnref<const GrGLInterface> interface(GrGLCreateNativeInterface());
> 
> there is a segmentation fault. After some research I found that there is a
> function get_egl_query_and_display (GrGLAssembleInterface.cpp) which is called
> even in glx context. 
> After some time, in GrGLExtensions::init (GrGLExtensions.cpp) there is a line:
> const char* extensions = queryString(eglDisplay, GR_EGL_EXTENSIONS);
> It returns garbage in <extensions> variable, and as a reason next call to
> eat_space_sep_strings produces a segfault. 
> 
> My patch discards all calls to glxGetProcAddress of egl-prefixed function
names
> and fixed described behavior. 
> 
> Thanks!
> 
> Regards,
> Andrew

Does this mean that glXGetProcAddress is returning a non-null ptr when queried
for "eglQueryString"? That seems really odd.

Anyway, can you simplify this just a bit to be

if (strstr(name, "egl") == name) {
     return nullptr;
}

If name is nullptr then returning nullptr is ok, no need to guard against it.
Also, can you move the comment about the if?