Bug 75763 – [chromium] Add setting for painting debug info onto tiles

• Status: RESOLVED
• Resolution: WONTFIX
• Component: New Bugs

[Adrienne Walker]

[chromium] Add setting for painting debug info onto tiles

[Adrienne Walker]

Created attachment 121528 [details]
Patch

[Adrienne Walker]

At paint time, this patch prints a block of text onto the layer with some debug information on top of whatever was painted into the layer.  It's useful to do it at paint time partially because you already have a GraphicsContext there, but also because it helps indicate whether invalidation or upload is failing.

This was quite helpful in tracking down http://crbug.com/107769.  I'm not totally sure it's worth piping through to about:flags, so I just left it as a compile-time only flag.

I'm not totally convinced that this should be committed, but jamesr asked for a patch.

[WebKit Review Bot]

Comment on attachment 121528 [details]
Patch

Attachment 121528 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/11184024

New failing tests:
http/tests/inspector-enabled/dedicated-workers-list.html
compositing/reflections/transform-inside-reflection.html
compositing/geometry/vertical-scroll-composited.html
compositing/geometry/layer-due-to-layer-children.html
compositing/geometry/ancestor-overflow-change.html
css2.1/20110323/abspos-containing-block-initial-004b.htm
compositing/reflections/nested-reflection-transition.html
compositing/direct-image-compositing.html
compositing/shadows/shadow-drawing.html
compositing/geometry/fixed-position-transform-composited-page-scale-down.html
css2.1/20110323/abspos-containing-block-initial-004d.htm
compositing/color-matching/image-color-matching.html
compositing/visibility/visibility-image-layers.html
compositing/geometry/layer-due-to-layer-children-deep.html
compositing/geometry/fixed-position-transform-composited-page-scale.html
compositing/scaling/tiled-layer-recursion.html

[Adrienne Walker]

Comment on attachment 121528 [details]
Patch

Removing review flag and leaving this patch up as an FYI for anybody who wants this debugging tool.  I don't think it's worth committing and having to carry forward through future refactorings.

[Adrienne Walker]

Created attachment 143121 [details]
Rebased

[Adrienne Walker]

Created attachment 144140 [details]
Cleanup

[Adrienne Walker]

(In reply to comment #6)
> Created an attachment (id=144140) [details]
> Cleanup

I found this debug info useful again.  Safari has something similar, although they just print out repaint counts.

I think it's worth landing this and afterwards adding a command line flag or yet another about:flags entry.

[James Robinson]

Comment on attachment 144140 [details]
Cleanup

This seems pretty reasonable.

[WebKit Review Bot]

Comment on attachment 144140 [details]
Cleanup

Clearing flags on attachment: 144140

Committed r118580: <http://trac.webkit.org/changeset/118580>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Ryosuke Niwa]

It appears that this patch caused TiledLayerChromiumTest.pushDirtyTiles to crash on cr-mac and cr-win:

http://build.webkit.org/builders/Chromium%20Mac%20Release%20%28Tests%29/builds/17130/steps/webkit-unit-tests/logs/stdio

http://build.webkit.org/builders/Chromium%20Mac%20Release%20%28Tests%29/builds/17131/steps/webkit-unit-tests/logs/stdio

blame list:
http://trac.webkit.org/log/?action=stop_on_copy&mode=stop_on_copy&rev=118600&stop_rev=118544&limit=100&verbose=on

[Marcus Bulach]

I confirmed this also crashes on linux canaries:

http://build.chromium.org/p/chromium.webkit/builders/Webkit%20Linux/builds/25948/steps/webkit_unit_tests/logs/stdio

(and a local build with BUILDTYPE=Release also fails with this patch..)

I'll ask the sherrifbot to roll out this patch so that I can roll webkit, I hope it won't cause any trouble..

[WebKit Review Bot]

Re-opened since this is blocked by 87647

[Adrienne Walker]

I'm just going to WONTFIX this and not spend the time addressing the unit test issues that showed up on non-Linux platforms.  It's going to be reverted anyway as a part of GTFO, so it doesn't seem worth the time.