Closed Bug 1089793 Opened 10 years ago Closed 10 years ago

Some reftest assertion failures on Android 4.0 are not reported in the log summary

Categories

(Testing :: Reftest, defect)

All
Android
defect
Not set
major

Tracking

(Not tracked)

RESOLVED DUPLICATE of bug 1000995

People

(Reporter: MatsPalmgren_bugz, Unassigned)

Details

https://treeherder.mozilla.org/ui/logviewer.html#?job_id=2731858&repo=try

...
10:55:00     INFO -  REFTEST TEST-LOAD | http://10.26.133.17:30530/tests/layout/reftests/font-face/download-3.html | 1000 / 1376 (72%)
10:55:00     INFO -  REFTEST TEST-PASS | http://10.26.133.17:30530/tests/layout/reftests/font-face/download-3-notref.html | image comparison (!=)
10:55:00     INFO -  REFTEST INFO | Loading a blank page
10:55:00     INFO -  REFTEST TEST-UNEXPECTED-FAIL | http://10.26.133.17:30530/tests/layout/reftests/font-face/download-3-notref.html | assertion count 4 is more than expected 2 assertions
10:55:00     INFO -  REFTEST TEST-END | http://10.26.133.17:30530/tests/layout/reftests/font-face/download-3-notref.html
...

You might think that the test causing the assertions is download-3-notref.html
but then you would be mistaken, the true answer lies further below in the
logcat output:

11:04:25     INFO -  10-27 11:02:45.109 I/Gecko   ( 2087): ++DOMWINDOW == 30 (0x67b386e0) [pid = 2087] [serial = 3511] [outer = 0x67b39e20]
11:04:25     INFO -  10-27 11:02:45.164 I/Gecko   ( 2087): [2087] ###!!! ASSERTION: NS_BLOCK_HAS_FIRST_LETTER_STYLE state out of sync: 'haveFirstLetterStyle == ((mState & NS_BLOCK_HAS_FIRST_LETTER_STYLE) != 0)', file /builds/slave/try-and-d-00000000000000000000/build/layout/generic/nsBlockFrame.cpp, line 6540
...
11:04:25     INFO -  10-27 11:03:10.960 I/Gecko   ( 2087): --DOMWINDOW == 43 (0x67b386e0) [pid = 2087] [serial = 3511] [outer = 0x0] [url = http://10.26.133.17:30530/tests/layout/reftests/forms/button/first-letter-1.html]
...

Note 0x67b386e0 connecting the ++DOMWINDOW with the --DOMWINDOW which will
give you the true URL for the failure.
Actually, after looking closer in the separate logcat file, it turns out that download-3.html /
download-3-notref.html did have unexpected assertions.  first-letter-1.html also had unexpected
assertions, but the log viewer only listed download-3* and did not mention first-letter-1.html
at all, which is why I was confused about which tests had failed.

(Also, there's no stacks for the assertions, but I assume that's a known problem for this platform.)
Summary: Reftest assertion failures on Android 4.0 blames the wrong test → Some reftest assertion failures on Android 4.0 are not reported in the log summary
Yes, on Android you need to look at the logcat (usually in a separate file) to find the assertions, and then there are no stacks. I would really like to see this handled better, but I don't know how to get there -- see discussion in the dup'd bug.
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.