Closed Bug 896684 Opened 11 years ago Closed 11 years ago

Intermittent B2G delay-test.html?restore-previous.gif,delay-test.html?clear.gif | image comparison (==), max difference: 255, number of differing pixels: 10000

Categories

(Core :: Graphics: ImageLib, defect)

Product:
Core
Component:
Graphics: ImageLib
Platform:
ARM
Gonk (Firefox OS)
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED FIXED
Milestone:
mozilla28
Tracking Flags:
Tracking Status
firefox26 --- fixed
firefox27 --- fixed
firefox28 --- fixed
firefox-esr24 --- fixed
b2g-v1.2 --- fixed

People

(Reporter: RyanVM, Assigned: seth)

References

Details

(Keywords: intermittent-failure)

Attachments

(2 files)

test screenshot
5.35 KB, image/png
Details
reference screenshot
5.35 KB, image/png
Details
Attached image test screenshot — 
https://tbpl.mozilla.org/php/getParsedLog.php?id=25558829&tree=Birch

b2g_emulator birch opt test reftest-1 on 2013-07-22 04:51:05 PDT for push 97cb863ad75a
slave: talos-r3-fed-035

05:13:31     INFO -  REFTEST TEST-START | http://10.0.2.2:8888/tests/image/test/reftest/animated/delay-test.html?restore-previous.gif
05:13:31     INFO -  REFTEST TEST-LOAD | http://10.0.2.2:8888/tests/image/test/reftest/animated/delay-test.html?restore-previous.gif | 641 / 1746 (36%)
05:13:31     INFO -  REFTEST TEST-UNEXPECTED-FAIL | http://10.0.2.2:8888/tests/image/test/reftest/animated/delay-test.html?restore-previous.gif | image comparison (==), max difference: 255, number of differing pixels: 10000
05:13:31     INFO -  REFTEST   IMAGE 1 (TEST): <see attached>
05:13:44     INFO -  REFTEST   IMAGE 2 (REFERENCE): <see attached>
05:13:44     INFO -  REFTEST INFO | Saved log: START http://10.0.2.2:8888/tests/image/test/reftest/animated/delay-test.html?restore-previous.gif
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering WaitForTestEnd
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] WaitForTestEnd: Adding listeners
05:13:44     INFO -  REFTEST INFO | Saved log: Initializing canvas snapshot
05:13:44     INFO -  REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: STATE_WAITING_TO_FIRE_INVALIDATE_EVENT
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: dispatching MozReftestInvalidate
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: STATE_WAITING_FOR_REFTEST_WAIT_REMOVAL
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: waiting for reftest-wait to be removed
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] AfterPaintListener in http://10.0.2.2:8888/tests/image/test/reftest/animated/delay-test.html?restore-previous.gif
05:13:44     INFO -  REFTEST INFO | Saved log: Updating entire canvas for invalidation
05:13:44     INFO -  REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] AttrModifiedListener fired
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] AfterPaintListener in http://10.0.2.2:8888/tests/image/test/reftest/animated/delay-test.html?restore-previous.gif
05:13:44     INFO -  REFTEST INFO | Saved log: Updating entire canvas for invalidation
05:13:44     INFO -  REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: STATE_WAITING_FOR_REFTEST_WAIT_REMOVAL
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: STATE_WAITING_FOR_SPELL_CHECKS
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: STATE_WAITING_TO_FINISH
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: Completed
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: STATE_COMPLETED
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] MakeProgress: STATE_COMPLETED
05:13:44     INFO -  REFTEST INFO | Saved log: [CONTENT] RecordResult fired
05:13:44     INFO -  REFTEST INFO | Saved log: RecordResult fired
05:13:44     INFO -  REFTEST INFO | Saved log: RecordResult fired
05:13:44     INFO -  REFTEST INFO | Loading a blank page
05:13:44     INFO -  REFTEST TEST-END | http://10.0.2.2:8888/tests/image/test/reftest/animated/delay-test.html?restore-previous.gif
Attached image reference screenshot — 

    
    

    
  
https://tbpl.mozilla.org/php/getParsedLog.php?id=25551681&tree=Mozilla-Inbound
https://tbpl.mozilla.org/php/getParsedLog.php?id=25557078&tree=Mozilla-Inbound
Summary: Intermittent delay-test.html?restore-previous.gif | image comparison (==), max difference: 255, number of differing pixels: 10000 → Intermittent delay-test.html?restore-previous.gif,clear.gif | image comparison (==), max difference: 255, number of differing pixels: 10000

    
  
Summary: Intermittent delay-test.html?restore-previous.gif,clear.gif | image comparison (==), max difference: 255, number of differing pixels: 10000 → Intermittent B2G delay-test.html restore-previous.gif,clear.gif | image comparison (==), max difference: 255, number of differing pixels: 10000

    
  
Summary: Intermittent B2G delay-test.html restore-previous.gif,clear.gif | image comparison (==), max difference: 255, number of differing pixels: 10000 → Intermittent B2G delay-test.html?restore-previous.gif,delay-test.html?clear.gif | image comparison (==), max difference: 255, number of differing pixels: 10000

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
Whoa wait what. Did this just start getting run or something? What caused this failure? This is really bad.
Flags: needinfo?(ryanvm)

    
    

    
  
Comment 0 reflects the first instance of this hitting our test infrastructure. If I knew what caused it, I would have backed the offending patch out already.
Flags: needinfo?(ryanvm)

    
    

    
      
  

    
    

    
  
Fair enough. And crappy.

    
    

    
      
  

    
  
Blocks: 897939

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
(In reply to Joe Drew (not getting mail) from comment #6)
> Whoa wait what. Did this just start getting run or something? What caused
> this failure? This is really bad.

Seth, sorry, I know you're overloaded, but I wanted to make sure this was on your radar based on Joe's comment above.
Flags: needinfo?(seth)

    
    

    
      
  

    
    

    
  
Thanks Ryan, consider it on my radar now. There is a similar bug (bug 900200) regarding APNGs, both of which may be related to some mistake that occurred in the switch to FrameAnimator. (I'm not 100% sure that the real cause has been identified yet in bug 900200, despite appearances.) I'll work on that bug while keeping this one in mind, and if they turn out to have different root causes, I'll tackle this one afterward.

I'm going to leave the needinfo on for now to remind me to check back.

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
Alright, bug 900200 has been pushed in. What I'm hoping to see is that this should drop off to zero over the next few days.

    
    

    
      
  

    
    

    
      
  

    
    

    
  
Comment 52 is definitely from after when bug 900200 merged around.

    
    

    
      
  

    
    

    
  
Gah, yes, seems to be a different issue. I'm going to go ahead and take this, although I don't have time to work on it immediately.
Assignee: nobody → seth
Flags: needinfo?(seth)

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
(In reply to Seth Fowler [:seth] from comment #55)
> Gah, yes, seems to be a different issue. I'm going to go ahead and take
> this, although I don't have time to work on it immediately.

Do you have a rough ETA for when you might be able to take a look? If you've got a lot on we could just disable on B2G for now if that's preferred?
Flags: needinfo?(seth)

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
@Ed, let me see if I can make some progress on this today. I'm hoping it's something relatively simple.

Here's a try job that tries to determine if the animation is actually happening (but just slowly) or not:

https://tbpl.mozilla.org/?tree=Try&rev=92ce69d23789
Flags: needinfo?(seth)

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
(In reply to Seth Fowler [:seth] from comment #262)
> @Ed, let me see if I can make some progress on this today. I'm hoping it's
> something relatively simple.
> 
> Here's a try job that tries to determine if the animation is actually
> happening (but just slowly) or not:
> 
> https://tbpl.mozilla.org/?tree=Try&rev=92ce69d23789

Thank you for looking at this :-)

I've retriggered the reftest-1 job a number of times on that push, to try and get it to reproduce.

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
So I think that test shows that the problem is either poorer-than-expected performance or just plain wonky timer behavior interacting poorly with the hard-coded timeout in this reftest. I've got a plan to fix this, after consultation with dbaron: we'll replace this set of reftests with mochitests that use repeated snapshotting for a longer period of time until a pass is found, like test-visited-reftests.html.

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
  
Depends on: 936720

    
    

    
  
I believe bug 936720 will solve this problem. It will, at a minimum, definitely get rid of this specific orange, since it removes this test entirely. Hopefully it also doesn't introduce any new ones. =)

Given the test removal, I'll be resolving this bug once bug 936720 lands.

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
      
  

    
    

    
  
Bug 936720 has now landed, and these tests have been removed. (They've been replaced with new tests which are much more robust to timing issues.) Marking this resolved.
Status: NEW → RESOLVED
Closed: 11 years ago
Resolution: --- → FIXED

    
  
Target Milestone: --- → mozilla28

    
  
No longer blocks: 897939

          You need to log in
          before you can comment on or make changes to this bug.
        






  

    
  







  

    

      
Attachment

      

      
      
      
      
    

    

      

        

          

            
General

            

              Creator: 



            

            
Created: 

            
Updated: 

            
Size: