Closed Bug 875194 Opened 11 years ago Closed 10 years ago

Intermittent Android test_video_crossorigin.html | uncaught exception - NS_ERROR_NOT_AVAILABLE: Component is not available at http://mochi.test:8888/tests/content/canvas/test/crossorigin/test_video_crossorigin.html:33

Categories

(Core :: Audio/Video, defect)

Product:
Core
Component:
Audio/Video
Platform:
ARM
Android
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: philor, Assigned: baku)

References

Details

(Keywords: intermittent-failure) 

This is rather odd, since this is the second flurry of it that I've seen, and it seems to act like a busted push or a bad build, both the 2.2 and 4.0 mochitest-1 runs will fail, maybe they'll both fail on several pushes, maybe because it's Android we'll clobber and it'll go away, but then maybe retriggers on what looked like a busted build will come back green.

This test isn't hitting the network for something that's intermittently down, is it?

https://tbpl.mozilla.org/php/getParsedLog.php?id=23296731&tree=Mozilla-Inbound
Android Tegra 250 mozilla-inbound opt test mochitest-1 on 2013-05-22 19:31:57 PDT for push 65d9aab57df9
slave: tegra-273

https://tbpl.mozilla.org/php/getParsedLog.php?id=23297328&tree=Mozilla-Inbound
Android 4.0 Panda mozilla-inbound opt test mochitest-1 on 2013-05-22 19:32:05 PDT for push 65d9aab57df9
slave: panda-0736

38278 ERROR TEST-UNEXPECTED-FAIL | /tests/content/canvas/test/crossorigin/test_video_crossorigin.html | uncaught exception - NS_ERROR_NOT_AVAILABLE: Component is not available at http://mochi.test:8888/tests/content/canvas/test/crossorigin/test_video_crossorigin.html:33
do we need to back out a patch or disable this test?
We absolutely do not need to disable this test - having a build either pass on both tegras and pandas or fail on both tegras and pandas means it's telling us that either we intermittently don't package something the test needs, or we intermittently miscompile. That is very much not something we want to sweep under the carpet.

What we want to have happen is to have the people associated with the test, in the product/component where the bug is now, tell us *what* that failure means isn't packaged/isn't properly compiled, and to look at what landed in the twenty or so pushes before the first instance in light of that, and find out whether it was a video patch or a fennec patch or a build config patch that broke it, and either fix it or back that out.
Andrea, as last to touch this test, could you take a look?

(In reply to Phil Ringnalda (:philor) from comment #59)
> We absolutely do not need to disable this test - having a build either pass
> on both tegras and pandas or fail on both tegras and pandas means it's
> telling us that either we intermittently don't package something the test
> needs, or we intermittently miscompile. That is very much not something we
> want to sweep under the carpet.
> 
> What we want to have happen is to have the people associated with the test,
> in the product/component where the bug is now, tell us *what* that failure
> means isn't packaged/isn't properly compiled, and to look at what landed in
> the twenty or so pushes before the first instance in light of that, and find
> out whether it was a video patch or a fennec patch or a build config patch
> that broke it, and either fix it or back that out.
Flags: needinfo?(amarchesini)
(In reply to Ed Morley [:edmorley UTC+1] from comment #60)
> Andrea, as last to touch this test, could you take a look?

Unfortunately I'm not able to compile gecko for android - I need to create the build environment. Then, what I did on this test was to rename crossorigin to crossOrigin :)

But I scheduled to spend time on this bug today. I hope to find a solution asap.
Thank you :-)
So, it seems that nsLayoutUtils::SurfaceFromElement() is still loading when we do |ctx.drawImage(v, 0, 0);|. This is the reason why we throw NS_ERROR_NOT_AVAILABLE.
Flags: needinfo?(amarchesini)
Assignee: nobody → amarchesini
Closing inactive keywords:intermittent-failure bugs where the TBPLbot has previously commented and the test isn't marked as disabled; filter on orange-cleanup-201401.
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.