Read framebuffer should have color image attached for ReadPixels/Copy{Sub}TexImage2D

Read framebuffer should have color image attached for ReadPixels/Copy{Sub}TexImage2D

Apparently some desktop drivers do not generate an INVALID_OPERATION, so we
should enforce it in GPU command buffer.

BUG=391953
TEST=conformance_renderbuffers_framebuffer_object_attachment passing on all GPU bots
R=kbr@chromium.org,bajones@chromium.org
NOTRY=true

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=282085

[Zhenyao Mo, 2014-07-08 19:44:30]

bajones, kbr: please review.

[bajones, 2014-07-08 19:47:34]

lgtm

[Ken Russell (switch to Gerrit), 2014-07-09 00:02:15]

lgtm

[Ken Russell (switch to Gerrit), 2014-07-09 00:04:14]

Question: was this error already being produced on Windows, where ANGLE is being
used? I assume this is a difference in behavior between desktop GL and GLES. If
it turns out that GLES implementations don't generate this error either then
we'll have to be careful to add this check to ANGLE when it subsumes the command
buffer's validation.

[Zhenyao Mo, 2014-07-09 00:16:10]

On 2014/07/09 00:04:14, Ken Russell wrote:
> Question: was this error already being produced on Windows, where ANGLE is
being
> used? I assume this is a difference in behavior between desktop GL and GLES.
If
> it turns out that GLES implementations don't generate this error either then
> we'll have to be careful to add this check to ANGLE when it subsumes the
command
> buffer's validation.

GLES3 spec clearly states that this is the behavior, while I can't find an
explicit text in the GLES2 spec.  However, this seems what it implies.

At the moment ANGLE doesn't generate such an error, nor does the Android bot
(not sure if it's ES2 or ES3).

Only Mac desktop GL generates an error on this.

So we should check this in ANGLE, but still needs to land this CL, so we can get
back the conformance_renderbuffers_framebuffer_object_attachment test coverage.

[Zhenyao Mo, 2014-07-09 00:20:39]

Filed an ANGLE bug: https://code.google.com/p/angleproject/issues/detail?id=693

[Ken Russell (switch to Gerrit), 2014-07-09 18:08:40]

On 2014/07/09 00:20:39, Zhenyao Mo wrote:
> Filed an ANGLE bug:
https://code.google.com/p/angleproject/issues/detail?id=693

Great, thanks. Let's certainly commit this fix for now.

[Zhenyao Mo, 2014-07-09 18:15:45]

The CQ bit was checked by zmo@chromium.org

[Zhenyao Mo, 2014-07-09 18:15:49]

The CQ bit was unchecked by zmo@chromium.org

[Zhenyao Mo, 2014-07-09 18:16:07]

The CQ bit was checked by zmo@chromium.org

[commit-bot: I haz the power, 2014-07-09 18:18:30]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/zmo@chromium.org/379623002/1

[commit-bot: I haz the power, 2014-07-09 18:20:36]

Change committed as 282085