Bug 73621 – [chromium] Use ANGLE's texture_usage and texture_storage extensions when allocating compositor textures

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

[Vangelis Kokkevis]

GL_ANGLE_texture_usage (http://angleproject.googlecode.com/svn/trunk/extensions/ANGLE_texture_usage.txt)
By calling glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_USAGE_ANGLE, GL_FRAMEBUFFER_ATTACHMENT_ANGLE) on textures used by RenderSurfaces, ANGLE can eliminate the allocation of a system memory buffer for the texture.

GL_EXT_texture_storage (http://angleproject.googlecode.com/svn/trunk/extensions/EXT_texture_storage.txt)
By switching texture allocation in the compositor to using  glTexStorage2DEXT instead of glTexImage2D we can eliminate the unnecessary allocation of additional mip levels for each texture.


The two extensions are exposed to the command buffer via:  http://codereview.chromium.org/8772033/ which must land first in the chromium tree.

[Vangelis Kokkevis]

Created attachment 117568 [details]
Patch

[WebKit Review Bot]

Please wait for approval from [e-mail] before submitting because this patch contains changes to the Chromium public API.

[Darin Fisher (:fishd, Google)]

Comment on attachment 117568 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=117568&action=review

> Source/WebKit/chromium/public/platform/WebGraphicsContext3D.h:385
> +    virtual void texStorage2DEXT(WGC3Denum target, WGC3Dint levels, WGC3Duint internalformat,

I defer to kbr@ for review of WebGC3D, but I see no issues with this change ;-)

[WebKit Review Bot]

Comment on attachment 117568 [details]
Patch

Attachment 117568 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/10703308

[Vangelis Kokkevis]

Ken, James: ping?

[James Robinson]

What's the compile failure on the cr-linux bot? I saw that and assumed it'd be addressed somehow so I haven't looked at the actual patch yet.

[Kenneth Russell]

Comment on attachment 117568 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=117568&action=review

Looks good overall. One minor issue that will fix the build failure. Please upload a revised patch and we can r+/cq+ it.

> Source/WebKit/chromium/public/platform/WebGraphicsContext3D.h:386
> +                                 WGC3Dint width, WGC3Dint height) = 0;

Give this an empty body like the method above. That will fix the cr-linux build failure. I gather from fishd@ that this is OK to leave in essentially permanently.

[Vangelis Kokkevis]

Created attachment 117910 [details]
Patch

[Vangelis Kokkevis]

(In reply to comment #8)
> Created an attachment (id=117910) [details]
> Patch

Fixed compile issue as per Ken's suggestion. James, would you mind taking a look at the compositor change?  Thanks.

[James Robinson]

Comment on attachment 117910 [details]
Patch

Compositor changes LGTM

[Kenneth Russell]

Comment on attachment 117910 [details]
Patch

Looks fine; let this clear the cr-linux EWS bot before committing. r=me

[WebKit Review Bot]

Comment on attachment 117910 [details]
Patch

Attachment 117910 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/10729807

New failing tests:
svg/custom/linking-uri-01-b.svg

[Vangelis Kokkevis]

Committed r102055: <http://trac.webkit.org/changeset/102055>