Allow RED internal format for CopyTextureCHROMIUM.

Allow RED internal format for CopyTextureCHROMIUM.

In crrev.com/1434453008 we changed the internal format for images
from R8 to RED but we forgot to change glCopyTextureCHROMIUM internal
format validation.

This breaks video to WebGL on GL Core Profile since APPLE_ycbcr_422 is
not avaiable, we fall back to three R8 planes, and we then fail to validate
RED as internal format.

BUG=587158
CQ_INCLUDE_TRYBOTS=tryserver.chromium.win:win_optional_gpu_tests_rel

Committed: https://crrev.com/c98f7529306942a721fac81ec024f553ba179dba
Cr-Commit-Position: refs/heads/master@{#376026}

[Daniele Castagna, 2016-02-17 18:59:49]

Description was changed from

==========
Allow RED internal format for CopyTextureCHROMIUM.

In crrev.com/1434453008 we changed the internal format for images
from R8 to RED but we forgot to change glCopyTextureCHROMIUM internal
format validation.

This breaks video to WebGL on GL Core Profile since APPLE_ycbcr_422 is
not avaiable, we fall back to three R8 planes, and we then fail to validate
RED as internal format.

BUG=587158
==========

to

==========
Allow RED internal format for CopyTextureCHROMIUM.

In crrev.com/1434453008 we changed the internal format for images
from R8 to RED but we forgot to change glCopyTextureCHROMIUM internal
format validation.

This breaks video to WebGL on GL Core Profile since APPLE_ycbcr_422 is
not avaiable, we fall back to three R8 planes, and we then fail to validate
RED as internal format.

BUG=587158
CQ_INCLUDE_TRYBOTS=tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[Daniele Castagna, 2016-02-17 19:06:27]

dcastagna@chromium.org changed reviewers:
+ ccameron@chromium.org, kbr@chromium.org

[Ken Russell (switch to Gerrit), 2016-02-17 19:15:15]

kbr@chromium.org changed reviewers:
+ zmo@chromium.org

[Ken Russell (switch to Gerrit), 2016-02-17 19:15:15]

zmo's an OWNER and should review this.

The change looks OK, but I'm a little concerned that the previous change moved
from using a sized internalformat to an unsized one. In general, unsized
internalformats are discouraged in ES 3.0.

[Zhenyao Mo, 2016-02-17 19:22:10]

This lgtm if it fixes your bug

In theory you could just allow both RED and R8, assuming when you call
TexImage*(), we already validated whichever formats are supported.

[Daniele Castagna, 2016-02-17 20:09:46]

On 2016/02/17 at 19:15:15, kbr wrote:
> zmo's an OWNER and should review this.
> 
> The change looks OK, but I'm a little concerned that the previous change moved
from using a sized internalformat to an unsized one. In general, unsized
internalformats are discouraged in ES 3.0.

Do we run on ES 1.1? IIRC ES 1.1 expects format to be the same as internalFormat
therefore R8 can't be used.
I honestly don't remember why we switched from R8 to RED. I can try to switch
everything to R8 and see if it breaks anything.

I'd land this patch first to keep the code consistent.

[Daniele Castagna, 2016-02-17 20:27:14]

On 2016/02/17 at 19:22:10, zmo wrote:
> This lgtm if it fixes your bug
> 
> In theory you could just allow both RED and R8, assuming when you call
TexImage*(), we already validated whichever formats are supported.

The internalformat passed by the client side is not really used anywhere if not
for validation on the service side. The actual internalformat
passed to CGLTexImage2d for example is determined based on the GMB format.

Since this is explicitly used with GMBs and CopyTextureCHROMIUM, I'd rather pick
one format and stick with that instead of allowing them both.
In case you're OK with that, I can follow up with another patch that uses R8 if
you prefer, or leave it as it is.

[Daniele Castagna, 2016-02-17 20:33:15]

The CQ bit was checked by dcastagna@chromium.org

[commit-bot: I haz the power, 2016-02-17 20:38:00]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1703153002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1703153002/1

[Zhenyao Mo, 2016-02-17 20:39:30]

On 2016/02/17 20:27:14, Daniele Castagna wrote:
> On 2016/02/17 at 19:22:10, zmo wrote:
> > This lgtm if it fixes your bug
> > 
> > In theory you could just allow both RED and R8, assuming when you call
> TexImage*(), we already validated whichever formats are supported.
> 
> The internalformat passed by the client side is not really used anywhere if
not
> for validation on the service side. The actual internalformat
> passed to CGLTexImage2d for example is determined based on the GMB format.
> 
> Since this is explicitly used with GMBs and CopyTextureCHROMIUM, I'd rather
pick
> one format and stick with that instead of allowing them both.
> In case you're OK with that, I can follow up with another patch that uses R8
if
> you prefer, or leave it as it is.

You can land the patch as is, but it might not work on top of ES3 where RED may
not exist as a valid internal_format.

[Daniele Castagna, 2016-02-17 20:44:25]

On 2016/02/17 at 20:39:30, zmo wrote:
> On 2016/02/17 20:27:14, Daniele Castagna wrote:
> > On 2016/02/17 at 19:22:10, zmo wrote:
> > > This lgtm if it fixes your bug
> > > 
> > > In theory you could just allow both RED and R8, assuming when you call
> > TexImage*(), we already validated whichever formats are supported.
> > 
> > The internalformat passed by the client side is not really used anywhere if
not
> > for validation on the service side. The actual internalformat
> > passed to CGLTexImage2d for example is determined based on the GMB format.
> > 
> > Since this is explicitly used with GMBs and CopyTextureCHROMIUM, I'd rather
pick
> > one format and stick with that instead of allowing them both.
> > In case you're OK with that, I can follow up with another patch that uses R8
if
> > you prefer, or leave it as it is.
> 
> You can land the patch as is, but it might not work on top of ES3 where RED
may not exist as a valid internal_format.

We can pick the correct internalformat based on GL implementation service side;
the image decides the internalformat passed to GL.
I'll follow up with another CL trying to bring back R8 instead of RED and we can
continue the discussion there.

Thanks. :)

[commit-bot: I haz the power, 2016-02-17 23:24:27]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-02-17 23:25:54]

Description was changed from

==========
Allow RED internal format for CopyTextureCHROMIUM.

In crrev.com/1434453008 we changed the internal format for images
from R8 to RED but we forgot to change glCopyTextureCHROMIUM internal
format validation.

This breaks video to WebGL on GL Core Profile since APPLE_ycbcr_422 is
not avaiable, we fall back to three R8 planes, and we then fail to validate
RED as internal format.

BUG=587158
CQ_INCLUDE_TRYBOTS=tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
Allow RED internal format for CopyTextureCHROMIUM.

In crrev.com/1434453008 we changed the internal format for images
from R8 to RED but we forgot to change glCopyTextureCHROMIUM internal
format validation.

This breaks video to WebGL on GL Core Profile since APPLE_ycbcr_422 is
not avaiable, we fall back to three R8 planes, and we then fail to validate
RED as internal format.

BUG=587158
CQ_INCLUDE_TRYBOTS=tryserver.chromium.win:win_optional_gpu_tests_rel

Committed: https://crrev.com/c98f7529306942a721fac81ec024f553ba179dba
Cr-Commit-Position: refs/heads/master@{#376026}
==========

[commit-bot: I haz the power, 2016-02-17 23:25:55]

Patchset 1 (id:??) landed as
https://crrev.com/c98f7529306942a721fac81ec024f553ba179dba
Cr-Commit-Position: refs/heads/master@{#376026}