New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 709435 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Email to this user bounced
Closed: Apr 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 1
Type: Bug-Regression



Sign in to add a comment

WebRtcCaptureFromElementBrowserTest.VerifyCanvasWebGLCaptureColor failing reliably on Android Tests (dbg) (L Nexus6) WebRTC bot

Project Member Reported by guidou@chromium.org, Apr 7 2017

Issue description

First failed run: 
https://build.chromium.org/p/chromium.webrtc.fyi/builders/Android%20Tests%20%28dbg%29%20%28L%20Nexus6%29/builds/14774

Last good run:
https://build.chromium.org/p/chromium.webrtc.fyi/builders/Android%20Tests%20%28dbg%29%20%28L%20Nexus6%29/builds/14773

There were no WebRTC commits between the failed runs.
Chrome version for the last good run is 462254
Chrome version for the first bad run is 462264

The error is detected on this line:
https://cs.chromium.org/chromium/src/content/test/data/media/canvas_capture_color.html?q=canvas_capture_color.html&sq=package:chromium&dr&l=90

Sample log line:
C  858.462s Main  [INFO:CONSOLE(90)] "Expected 0,0,255,229.5, got 0,0,27,229", source: http://127.0.0.1:52887/media/canvas_capture_color.html (90)

Which is repeated many times.
 

Comment 1 Deleted

I didn't find any obvious commit in the Chromium range that might break this .
I speculatively reverted https://codereview.chromium.org/2797973002, which is in the range, but it did not fix the problem, so I relanded it.
Hmm, the blame list is:
https://chromium.googlesource.com/chromium/src/+log/12af4f53a6107c5c80252666c804fb5a262b01bf..0738354f2a76055b590c45b691e48c19c6281fbf?pretty=fuller
(created by looking at parent_got_cr_revision for the last green+first red builds)

https://codereview.chromium.org/2797973002 is not even in this range, but I understand if you ended up with the wrong revision range since the got_cr_revision properties can be misleading ( bug 670856  which I have a fix in progress for).

Can you see something suspicious in the range above?
My best bet is 
https://chromium.googlesource.com/chromium/src/+/65a58f0cd7291f90e7edd10ea42c7c8de24f3d1a
or the skia roll in 
https://chromium.googlesource.com/chromium/src/+/84e5378049c4ae1c361c0bfacfb11ecec487ef2a

I also don't understand why this fails only on our bot, it runs on the regular bots, like Nexus 5Xs in https://build.chromium.org/p/chromium.linux/builders/Android%20Tests/builds/40316
Do you think it's specific to Nexus 6 and/or the particular Android version on that bot?

Comment 4 by guidou@chromium.org, Apr 10 2017

I am clueless as to why this fails only on that particular bot.

Comment 5 by guidou@chromium.org, Apr 10 2017

Owner: olka@chromium.org
Status: Assigned (was: Untriaged)

Comment 6 by olka@webrtc.org, Apr 10 2017

The skia roll has "reed remove android legacy flags". That looks like something which can affect specific versions of Android.

Comment 7 by olka@chromium.org, Apr 10 2017

Cc: guidou@chromium.org olka@chromium.org
Owner: reed@chromium.org
reed@ could you PTAL?

Comment 8 by guidou@chromium.org, Apr 11 2017

Status: Fixed (was: Assigned)
Marking as fixed since the bot has now been green for the last five runs.
Was something done to specifically to fix this bug, or was it a nice side effect of some other work?

Sign in to add a comment