Collect driver vendor in linux through glXGetClientString.

Collect driver vendor in linux through glXGetClientString.

This helps in the situation where libpci is unavailable: at least we can identify a few vendors.

Unfortunately we can not run this in preliminary GPU info collection because gl binding hasn't been initialized yet.

BUG=78526
TEST=bots green
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=80876

[Zhenyao Mo, 2011-04-06 21:45:53]

Please review.

[piman, 2011-04-06 22:10:00]

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
File content/gpu/gpu_info_collector_linux.cc (right):

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
content/gpu/gpu_info_collector_linux.cc:172: const char* vendor =
glXGetClientString(display, GLX_VENDOR);
I suspect you'll want to copy that into the std::string before you close the
display. Since it doesn't have an explicit lifetime, I assume the string is
owned by the driver, and could be released when you call XCloseDisplay.
Other thing: this code only works with glx which may not be present or
functional (e.g. on ARM devices, we use egl instead). Also, it needs to
explicitly link to libGL which may not exist (e.g. on ARM). You need to use the
GL implementation system instead, which dynamically loads the correct library
instead.

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
content/gpu/gpu_info_collector_linux.cc:183: // TODO(zmo): find a way to
identify Intel cards.
The problem is that the client vendor string for Intel GPUs and other opensource
drivers (e.g. nouveau) is going to say Mesa and SGI. To differentiate, you'll
need the OpenGL renderer string that will say for Intel something like "Mesa DRI
Intel(R) IGD GEM 20100330"

[zmo, 2011-04-06 23:14:34]

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
File content/gpu/gpu_info_collector_linux.cc (right):

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
content/gpu/gpu_info_collector_linux.cc:172: const char* vendor =
glXGetClientString(display, GLX_VENDOR);
On 2011/04/06 22:10:00, piman wrote:
> I suspect you'll want to copy that into the std::string before you close the
> display. Since it doesn't have an explicit lifetime, I assume the string is
> owned by the driver, and could be released when you call XCloseDisplay.

Got it.  Will fix.

> Other thing: this code only works with glx which may not be present or
> functional (e.g. on ARM devices, we use egl instead). Also, it needs to
> explicitly link to libGL which may not exist (e.g. on ARM). You need to use
the
> GL implementation system instead, which dynamically loads the correct library
> instead.

This is through the GL implementation system.  Only I am not sure what happens
if glx does not exist.  Should I test if the function pointer is null or not?

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
content/gpu/gpu_info_collector_linux.cc:183: // TODO(zmo): find a way to
identify Intel cards.
On 2011/04/06 22:10:00, piman wrote:
> The problem is that the client vendor string for Intel GPUs and other
opensource
> drivers (e.g. nouveau) is going to say Mesa and SGI. To differentiate, you'll
> need the OpenGL renderer string that will say for Intel something like "Mesa
DRI
> Intel(R) IGD GEM 20100330"

This is trying to collect vendor without creating a context.  So GL Renderer
string won't be available.  At the moment I am not sure if there will be a way
to identify Intel GPU solely through glx vendor.

[piman, 2011-04-06 23:18:01]

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
File content/gpu/gpu_info_collector_linux.cc (right):

http://codereview.chromium.org/6803024/diff/1/content/gpu/gpu_info_collector_...
content/gpu/gpu_info_collector_linux.cc:172: const char* vendor =
glXGetClientString(display, GLX_VENDOR);
> This is through the GL implementation system.  Only I am not sure what happens

Ah, ok, that should be fine then

> if glx does not exist.  Should I test if the function pointer is null or not?

you probably only want to run this if the GL implementation is GLX. then the
function pointer ought to be there.

[Zhenyao Mo, 2011-04-08 00:29:57]

Revised.  Please review again.  The linux bot is red, but I think the failed
test has nothing to do with this CL.

[piman, 2011-04-08 00:31:39]

LGTM