Created attachment 760427 [details] css being applied with Stylish http://forum.userstyles.org/discussion/35738/stylish-1.3.2-is-proving-to-be-very-unstable-with-firefox-21-release I have a user who reports he can reliably crash Firefox when previewing a style with Stylish. He receives a hung thread notification, and says allowing the thread to continue is more likely to cause the crash than killing it. The CSS he's using is attached. I and a few others have tried to reproduce, but we can't get it to crash. http://crash-stats.mozilla.com/report/index/bp-910b8803-3639-4104-82ea-9af952130514 http://crash-stats.mozilla.com/report/index/bp-999482e3-7972-42c5-9a2a-fbe2a2130514 http://crash-stats.mozilla.com/report/index/bp-bf1e15b9-0e86-4392-bbc7-b73362130514 http://crash-stats.mozilla.com/report/index/bp-24c5c52b-aad4-4c3c-90a7-c7f2d2130514 http://crash-stats.mozilla.com/report/index/bp-68d21e37-28eb-4d00-b407-6216d2130515 http://crash-stats.mozilla.com/report/index/bp-e8d71d3b-b2b5-4a43-ba1b-1944d2130515 http://crash-stats.mozilla.com/report/index/bp-96366e75-b3f3-41cb-8b70-702942130515 http://crash-stats.mozilla.com/report/index/bp-4a48f80b-af6b-4f5d-88e5-773bd2130515
It first showed up in 21.0a1/20130213 but is discontinuous across builds. The regression range might be: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=36525224b14e&tochange=161a347bda5b
Crash Signature: [@ JSC::Yarr::Interpreter<wchar_t>::matchDisjunction(JSC::Yarr::ByteDisjunction*, JSC::Yarr::Interpreter<wchar_t>::DisjunctionContext*, bool) ] [@ @0x0 | JSC::Yarr::Interpreter<wchar_t>::matchNonZeroDisjunction(JSC::Yarr::ByteDisjunction*, JSC::Yarr::Int…
status-firefox21: --- → affected
status-firefox22: --- → affected
status-firefox23: --- → affected
status-firefox24: --- → affected
Keywords: regression, regressionwindow-wanted
Hardware: x86_64 → x86
I tried with FF24 on Win 7, I don't crash. Can the reporter test with a fresh profile and only Stylish+this style? (or post directly here)
I'm the person that reported the issue on userstyles.org. I did attempt to reproduce it with a new profile. Good news. Bad news. First the good news. I created a new profile and tried to reproduce it as requested. I was not successful. Now for the bad news. Because it didn't happen every time, I can't be sure it would happen if I had tried it just one more time. It is always very random. Sometimes it is constant. But the other day I was able to preview a style numerous times with no issue. I have found that while killing the thread at first seemed to do better than continuing it, that difference has almost vanished. I have some at least a little better reliability when I let the run for a moment before I kill it.
Something I just thought of: Could it be trying to kill a thread that isn't running anymore?
I can't reproduce the issue with Firefox 22. Did a patch already land in the release build?
Anyway it was already hard to reproduce it.
Happened with Firefox 25, same pattern: responding "continue" to a "do you want to continue this script" question box.
And here I was hoping I had a fix. :(
This Yarr crash is no longer relevant we removed Yarr in Firefox 32 (bug 976446).
Status: NEW → RESOLVED
Last Resolved: 4 years ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.