1325661 - 4 new failing reftests from merging m-c -> graphics at m-c cset bbbd2f7539f2

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect, P3)

Description

[Kartikaya Gupta (email:[e-mail])]

The merge from m-c at cset bbbd2f7539f2 [1] caused 4 reftest failures on the graphics branch.

layout/reftests/svg/paint-on-maskLayer-1a.html
layout/reftests/svg/paint-on-maskLayer-1b.html
layout/reftests/xul/inactive-fixed-bg-bug1205630.xul
layout/reftests/xul/inactive-fixed-bg-bug1272525.xul

The first two on the list are actually new reftests introduced by bug 1318266. I suspect the other to were also regressed by the same bug, but I'm doing a try push to confirm.

This is likely an issue with the way we convert WebRenderLayer instances to wr_* API calls, since we kind of ignore the transform on the layers in some places and probably shouldn't be doing that.

[1] https://treeherder.mozilla.org/#/jobs?repo=graphics&revision=f241b4570f5396628738da4f0543f981bcc549d7

Comment 1

[u459114]

paint-on-maskLayer-1x.html was evolve form inactive-fixed-bg-bug1205630.xul(see bug 1318266 comment 48).

Comment 2

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/projects/graphics/rev/a731b493e8c7
Mark 4 new reftests as failing with webrender after merge from m-c. r=gfx?

Comment 3

[Kartikaya Gupta (email:[e-mail])]

(In reply to C.J. Ku[:cjku](UTC+8) from comment #1)
> paint-on-maskLayer-1x.html was evolve form
> inactive-fixed-bg-bug1205630.xul(see bug 1318266 comment 48).

Ah, good to know, thanks. So likely all four reftest failures are from the same bug.

Comment 4

[Kartikaya Gupta (email:[e-mail])]

Try push at https://treeherder.mozilla.org/#/jobs?repo=try&revision=9eebdf91a4158134a8c189f381a2e9f367d02a52 also confirmed that the failures came from the same bug.

Comment 5

[Markus Stange [:mstange]]

These tests may go green again once bug 1325550 lands and merges to the graphics branch, because the patch there will again cause the masking to happen on the content side instead of on the compositor.

Comment 6

[Kartikaya Gupta (email:[e-mail])]

(In reply to Markus Stange [:mstange] from comment #5)
> These tests may go green again once bug 1325550 lands and merges to the
> graphics branch, because the patch there will again cause the masking to
> happen on the content side instead of on the compositor.

Indeed, that seems to be the case with the latest merge from m-c

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/projects/graphics/rev/545975b950e8
Back out cset a731b493e8c7 as the tests are passing again. r=gfx?