Less X connections.

Less X connections.

With Aura, we are using GLSurfaces on the UI thread. So we have 2 X connections, one for the MessagePumpX and another for the GLSurfaces. Using shawnsingh's scissoring patch, I've noticed that sometimes these race. This fixes the problem by using only one connection.

BUG=104248
TEST=Enable shawnssingh's scissoring (comment out "m_capabilities.usingPartialSwap = false" in LayerRendererChromium.cpp). Launch Aura chrome repeatedly on hardware (e.g. Intel graphics) supporting GLX_MESA_copy_sub_buffer. There should be no undrawn regions on the screen.

[jonathan.backer, 2011-12-08 16:02:13]

One potential problem is that MessagePumpX closes the X connection in the
destructor. This is as it should be because we'll trigger any outstanding X
connections (http://codereview.chromium.org/8873018).

For both the browser and the gpu processes, MessagePumpX is one of the last
things to be destroyed so we should be good here.

[apatrick_chromium, 2011-12-08 20:16:22]

LGTM with one nit.

http://codereview.chromium.org/8873018/diff/4/ui/gfx/gl/gl_surface_egl.cc
File ui/gfx/gl/gl_surface_egl.cc (right):

http://codereview.chromium.org/8873018/diff/4/ui/gfx/gl/gl_surface_egl.cc#new...
ui/gfx/gl/gl_surface_egl.cc:174: static EGLNativeDisplayType g_native_display =
NULL;
nit: drop the g_ for a local variable.

[commit-bot: I haz the power, 2011-12-08 20:26:55]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/backer@chromium.org/8873018/4001

[jonathan.backer, 2011-12-09 14:34:37]

Superceded by piman@'s patch http://codereview.chromium.org/8890042/

Closing.