Delete copy_texture_to_parent_texture from GPU command buffer code.

Delete copy_texture_to_parent_texture from GPU command buffer code.

Efficient allocation of shared WebGL resource IDs (needed for m14).

Because we aren't using it for anything anymore.
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=96238

[apatrick_chromium, 2011-07-29 23:09:55]

codereview try results are all lies again. Here is linux:

http://build.chromium.org/p/tryserver.chromium/builders/linux/builds/40183

[jamesr, 2011-07-29 23:22:49]

Looks good, but hold off for a bit on landing prz

http://codereview.chromium.org/7538008/diff/2005/gpu/command_buffer/service/f...
File gpu/command_buffer/service/feature_info.cc (left):

http://codereview.chromium.org/7538008/diff/2005/gpu/command_buffer/service/f...
gpu/command_buffer/service/feature_info.cc:105:
AddExtensionString("GL_CHROMIUM_copy_texture_to_parent_texture");
On the WebKit side we're still checking for this string.  I have a patch to
delete that, but we can't land this before that side lands or I'm pretty sure
accel canvas 2d will completely break.

http://trac.webkit.org/browser/trunk/Source/WebCore/platform/graphics/chromiu...

[apatrick_chromium, 2011-08-08 22:30:03]

I updated this patch. PTAL.

[apatrick_chromium, 2011-08-09 01:03:43]

+gman since i'm not sure if jamesr is doing reviews.

[greggman, 2011-08-09 16:11:35]

http://codereview.chromium.org/7538008/diff/9047/gpu/command_buffer/common/gl...
File gpu/command_buffer/common/gles2_cmd_ids_autogen.h (left):

http://codereview.chromium.org/7538008/diff/9047/gpu/command_buffer/common/gl...
gpu/command_buffer/common/gles2_cmd_ids_autogen.h:204:
OP(CopyTextureToParentTextureCHROMIUM)                       /* 447 */ \
My bad. This isn't going to work. All the ids will be off by one.

The easiest fix is probably changing it to OP(id,value) and make the generator
pass the value in.

[apatrick_chromium, 2011-08-10 00:45:30]

Argh! I cobbled something together to replace the removed entry points with
placeholders that always return kUnknownCommand. We need something better
though. The absence of the command ID from the table should be enough to make
the GPU process return kUnknownCommand.

The change to make WebGL use non shared ID allocators is also mixed in here. At
the moment the WebKit GraphicsContext3D interface does not have a way to specify
whether sharing is enabled and what the share group is. That will take some
alternating WebKit / Chromium checkins. For now, it assumes that if
attribs.noExtensions is true, it's WebGL.

PTAL.

[greggman, 2011-08-10 04:27:26]

lgtm

Please add an issues for (a) fixing the kUnknownCommand stuff (b) stop using
noExtensions as the flag. Should probably discuss what to do about (b).