Locally on my linux desktop, I have run this:
Tools/Scripts/run-webkit-tests -t Release fast/canvas/ --order=random

for over 10 times, and the above test never fails.

Could you please provide a bit more info:
1. Is it under linux?
2. Is this a debug build?
3. If it is not debug build, is dcheck on?
4. Any other info?

Thank you.

It was on Ubuntu. I don't think it was a debug build and not sure about dcheck - I used the default build settings. CCing qyearsley@

From my data it expected PASS but got TEXT in 9 out of 117 test runs.

OK, default build setting should not have dcheck on. Maybe I need more amount of runs.

Another question, I assume that when you ran the test, you ran the entire fast/canvas/ suite, instead of just that test?

Yep, I was running all of LayoutTests each time. One possibility is that the failures could be caused by resource contention, so it might be more likely to fail in the middle of a long test run.

confirmed.

Found an easier way to repro this:

third_party/WebKit/Tools/Scripts/run-webkit-tests -t Release fast/canvas/canvas-drawImage-live-video.html --repeat-each=200

It fails 3 times.

I didn't know about --repeat-each! Added it to the doc under useful flags.

I looked at the test and I don't know why it could fail. Maybe when the time advances, the video has not updated its frame yet?

I am unable to repro locally.

xidachen@: could you run this experiment for me?

Modify the test by inserting the following code at line 60:

      if (imagesAreTheSame) {
        console.log(JSON.stringify(imageData.data));
      }

Then look at the test results for the runs that fail.  I want to know whether the console dump is giving all black pixels.  This will tell us which of the following two situations is occurring:
a) video draws are failing.
b) frame is not advancing.

I have attached the console output. It is not all black pixels, I think it is more likely that the frame is not advancing.

Deleted: console output.txt 1.5 KB

console output.txt 1.5 KB View Download

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b034403ab50b3a27bda35d6cf72efec57f2f1217

commit b034403ab50b3a27bda35d6cf72efec57f2f1217
Author: qyearsley <qyearsley@chromium.org>
Date: Tue Feb 07 17:50:47 2017

Remove flaky expectations with update-test-expectations script

The update-test-expectations script is meant to remove
flaky expectations when it appears that according to the
most recent results on the flakiness dashboard for that test,
the test is no longer flaky.

BUG= 491764 , 498539 , 575766 , 600248 ,658305, 663838 , 663840 ,663848, 663851 , 663853 , 663872 , 663874 , 663877 , 663879 ,664817,664839, 664840 , 664841 , 664842 , 664846 , 664850 , 664855 , 664856 , 666991 , 671480 , 671618 , 672204 , 673296 , 673632 , 674720 ,674858, 678488 , 683800 

Review-Url: https://codereview.chromium.org/2679543005
Cr-Commit-Position: refs/heads/master@{#448663}

[modify] https://crrev.com/b034403ab50b3a27bda35d6cf72efec57f2f1217/third_party/WebKit/LayoutTests/TestExpectations

The flaky expectation has now been removed, because in recent runs on the continuous builders it doesn't appear to be flaky -- this can probably be closed?

Confirmed no failures recently.