Closed
Bug 491735
Opened 15 years ago
Closed 12 years ago
random leak test orange: ASSERTION: imgIRequest::STATUS_FRAME_COMPLETE not set
Categories
(Core :: Layout: Images, Video, and HTML Frames, defect)
Core
Layout: Images, Video, and HTML Frames
Tracking
()
RESOLVED
WORKSFORME
People
(Reporter: dbaron, Unassigned)
References
Details
(Keywords: intermittent-failure)
Attachments
(1 file)
10.69 KB,
text/plain
|
Details |
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241640344.1241640845.10586.gz Linux mozilla-central leak test build on 2009/05/06 13:05:44 shows random orange during the leak test with the assertion: ###!!! ASSERTION: imgIRequest::STATUS_FRAME_COMPLETE not set: '(status & imgIRequest::STATUS_FRAME_COMPLETE)', file /builds/moz2_slave/mozilla-central-linux-debug/build/layout/base/nsImageLoader.cpp, line 169 This assertion is also reported in other contexts in bug 458275 and bug 472590.
Reporter | ||
Updated•15 years ago
|
Whiteboard: [orange]
Showed up on trunk Linux and Mac OS X tinderboxes tonight. Linux: http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241670340.1241670798.20998.gz&fulltext=1 Mac OS X: http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241667126.1241669271.19118.gz&fulltext=1
Linux trunk again, see timestamp on this comment http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241673489.1241673894.25115.gz
Comment 4•15 years ago
|
||
So, did something change with this code recently? This is a lot in a short period of time... http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241680963.1241682391.3127.gz&fulltext=1 OS X 10.5.2 mozilla-central leak test build on 2009/05/07 00:22:43 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241719061.1241721423.5015.gz&fulltext=1 Linux mozilla-central leak test build on 2009/05/07 10:57:41 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241680095.1241682300.2982.gz&fulltext=1 Linux mozilla-central leak test build on 2009/05/07 00:08:15 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241702913.1241705636.7887.gz&fulltext=1 OS X 10.5.2 mozilla-central leak test build on 2009/05/07 06:28:33 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241690080.1241691929.16528.gz&fulltext=1 OS X 10.5.2 mozilla-central leak test build on 2009/05/07 02:54:40 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241697280.1241701076.31581.gz&fulltext=1 OS X 10.5.2 mozilla-central leak test build on 2009/05/07 04:54:40
Comment 5•15 years ago
|
||
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241721422.1241724688.10642.gz&fulltext=1 OS X 10.5.2 mozilla-central leak test build on 2009/05/07 11:37:02
The stack trace in comment #2 is bogus.
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241743463.1241746380.10492.gz OS X 10.5.2 mozilla-central leak test build on 2009/05/07 17:44:23
For the record, there are 3 bugs on file with this assertion: this, 472590, and 458275.
I loaded up a bunch of logs and looked for things that precede the assertion in all cases, this output precedes every assertion on Mac OS X and Windows. I don't know if it is relevant, just pointing out the pattern. localhost - - [07/May/2009 12:28:52] code 404, message File not found localhost - - [07/May/2009 12:28:52] "GET /favicon.ico HTTP/1.1" 404 - .. localhost - - [07/May/2009 12:28:55] code 404, message File not found localhost - - [07/May/2009 12:28:55] "GET /favicon.ico HTTP/1.1" 404 -
Comment 10•15 years ago
|
||
From grepping over the logs for the leak builders, I find * mac mozilla-central - 8 of 30 builds assert, starting from http://hg.mozilla.org/mozilla-central/rev/89574e2e0381 built Wed May 6 17:14:13 2009 * linux mozilla-central - 10 of 62 builds, starting from http://hg.mozilla.org/mozilla-central/rev/f13ca2edd374 built Tue May 5 16:27:16 2009 * mac mozilla-1.9.1 - 7 of 37 builds, starting from http://hg.mozilla.org/releases/mozilla-1.9.1/rev/38c2817eba84 built Tue May 5 17:19:58 2009 * linux mozilla-1.9.1 - 8 of 40 builds, starting from http://hg.mozilla.org/releases/mozilla-1.9.1/rev/38c2817eba84 built Tue May 5 15:19:58 2009 * not seen on windows on m-c or m-1.9.1 in last 5 days So something before f13ca2edd374 on m-c, here's a fairly broad range http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=9c7d85c8390c&tochange=f13ca2edd374 and before 38c2817eba84 on 1.9.1, most probably in this http://hg.mozilla.org/releases/mozilla-1.9.1/pushloghtml?fromchange=8eb9de885e58&tochange=38c2817eba84 Need to find the intersection.
Comment 11•15 years ago
|
||
The only bug in both of those ranges is 426930 and I doubt that is it, it is gnomestripe only.
Comment 12•15 years ago
|
||
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241783431.1241785215.1868.gz&fulltext=1 Linux mozilla-central leak test build on 2009/05/08 04:50:31 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241791539.1241793464.19110.gz&fulltext=1 Linux mozilla-central leak test build on 2009/05/08 07:05:39
Reporter | ||
Comment 13•15 years ago
|
||
The leak test is testing live pages off the Web, in particular: "http://www.mozilla.org", "http://www.mozilla.org", "http://www.mozilla.org/newlayout/samples/test2.html", "http://www.mozilla.org/newlayout/samples/test8.html", "http://www.mozilla.org/newlayout/samples/test6.html", "http://www.mozilla.org/newlayout/samples/test8.html" (see bloatcycle.html). So it's possible this could have been triggered by a change to one of the pages. (Once we get debug unit test tinderboxes we won't need to run this test anymore.)
Comment 14•15 years ago
|
||
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1241804776.1241806996.9663.gz OS X 10.5.2 mozilla-central leak test build on 2009/05/08 10:46:16
Comment 15•15 years ago
|
||
(In reply to comment #13) I don't see anything on bonsai for www.mozilla.org that would have affected those pages - http://tinyurl.com/phus6m
Reporter | ||
Comment 16•15 years ago
|
||
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox3.5/1242233337.1242237001.3556.gz OS X 10.5.2 mozilla-1.9.1 leak test build on 2009/05/13 09:48:57
Comment 17•15 years ago
|
||
More on OS X leak test: http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1242232957.1242233320.30213.gz http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1242233871.1242234393.31773.gz
Comment 18•15 years ago
|
||
Bug 472590 has an testcase which reliably reproduces the assertion.
Comment 19•15 years ago
|
||
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1242331541.1242331930.27960.gz OS X 10.5.2 mozilla-central leak test build on 2009/05/14 13:05:41 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1242332204.1242332659.29192.gz OS X 10.5.2 mozilla-central leak test build on 2009/05/14 13:16:44
Comment 20•15 years ago
|
||
Linux: http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1242332204.1242333541.30621.gz
Comment 21•15 years ago
|
||
I suspect that the below yellow is the same problem, but now that bug 472590 is fixed, we don't get the false alarm for imgIRequest::STATUS_FRAME_COMPLETE. http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1242360246.1242362838.13699.gz (I could be wrong, but in that case this bug should be closed)
Summary: random leak test orange: ASSERTION: imgIRequest::STATUS_FRAME_COMPLETE not set → random leak test orange: ASSERTION: Class name and proto chain interface name mismatch!: 'nsCRT::strcmp(CutPrefix(name), mData->mName) == 0'
Reporter | ||
Comment 22•15 years ago
|
||
What makes you think it's the same problem? Is it happening at the same frequency and on the same distribution of platforms? There are a whole bunch of existing bugs on assertions during the leak test.
Reporter | ||
Comment 23•15 years ago
|
||
The assertion this bug was originally about generally happened (sample size of 4, all 4 times) after one occurrence of: localhost.localdomain - - [14/May/2009 13:37:39] "GET /bloatcycle.html HTTP/1.1" 200 - two occurrences of: localhost.localdomain - - [08/May/2009 05:19:21] "GET /favicon.ico HTTP/1.1" 404 - and then some plugin loading. The log in comment 21 has an assertion earlier -- after the first "favicon.ico" line. So I really don't think it's something later in the process exposed by fixing the old one, since it seems earlier in the process. Retitling the bug back to avoid any further confusion (since morphing bugs makes things very confusing very quickly). We should have a separate bug on this other assertion.
Summary: random leak test orange: ASSERTION: Class name and proto chain interface name mismatch!: 'nsCRT::strcmp(CutPrefix(name), mData->mName) == 0' → random leak test orange: ASSERTION: imgIRequest::STATUS_FRAME_COMPLETE not set
I think the current "Class name and proto chain interface name mismatch" assert showing up on leak boxes ins just a regression from a change that I'm about to back out (bug 492219 in particular).
Comment 25•15 years ago
|
||
Sorry, I was hasty.
Updated•12 years ago
|
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → WORKSFORME
Assignee | ||
Updated•12 years ago
|
Keywords: intermittent-failure
Assignee | ||
Updated•12 years ago
|
Whiteboard: [orange]
Updated•6 years ago
|
Product: Core → Core Graveyard
Updated•6 years ago
|
Product: Core Graveyard → Core
You need to log in
before you can comment on or make changes to this bug.
Description
•