529 bytes, application/vnd.mozilla.xul+xml
23.02 KB, text/plain
2.47 KB, patch
Neil Deakin (not available until Aug 9): review+
Alexander Sack: approval1.8.0.next+
|Details | Diff | Splinter Review|
937 bytes, patch
|Details | Diff | Splinter Review|
Created attachment 268038 [details] testcase See testcase, which crashes Mozilla within 100ms. It also crashes branch builds, so marking security sensitive for now. I guess bug 279703 might fix things here. Talkback ID: TB33045810E 0x01eb9089 PresShell::AttributeChanged [mozilla/layout/base/nspresshell.cpp, line 4408] nsNodeUtils::AttributeChanged [mozilla/content/base/src/nsnodeutils.cpp, line 100] nsXULElement::UnsetAttr [mozilla/content/xul/content/src/nsxulelement.cpp, line 1316] nsXULElement::GetAttrInfo [mozilla/content/xul/content/src/nsxulelement.cpp, line 2108] 0x02dc3aa0 0x02c44be8 PresShell::AddRef [mozilla/layout/base/nspresshell.cpp, line 1394] 0x08244c8b
taking for now. Need to think this a bit. Ungeneratemenu() shouldn't be called in ::Destroy(). Probably need to call that asynchronously, but only if menu doesn't have a new frame.
This is worksforme now that the patch for bug 279703 went in. (tested with a tinderbox build)
Created attachment 272571 [details] Crash log from branch build on Mac This still affects a branch build. Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en-US; rv:22.214.171.124pre) Gecko/20070716 BonEcho/126.96.36.199pre Requesting blocking (though; should probably be changed to wanted).
Crashes rv:188.8.131.52 as well (Firefox 184.108.40.206). I can't see bug 279703 landing on the 1.8 branch. I also get a different stack on branch, "this" is a deleted object in nsCachedStyleData::GetStyleData(). Is there some way to wallpaper this for the branch? This looks familiar actually, maybe this testcase is triggering a different bug that's already fixed on trunk but not yet backported, and once we get that out of the way we'll see the nsPresShell::AttributeChanged issue on the branch. > nsCachedStyleData::GetStyleData() Line 210 C++ nsStyleContext::GetStyleData() Line 248 C++ nsIFrame::GetStyleData() Line 612 C++ nsIFrame::GetStyleDisplay() Line 90 C++ nsCSSFrameConstructor::AttributeChanged() Line 10752 C++ PresShell::AttributeChanged() Line 5499 C++ nsXULDocument::AttributeChanged() Line 1133 C++ nsXULElement::UnsetAttr() Line 1706 C++ nsMenuFrame::UngenerateMenu() Line 712 C++ nsMenuFrame::Destroy() Line 398 C++ nsLineBox::DeleteLineList() Line 325 C++ nsBlockFrame::Destroy() Line 303 C++ nsAreaFrame::Destroy() Line 155 C++ nsBoxFrame::RemoveFrame() Line 1178 C++ nsFrameManager::RemoveFrame() Line 717 C++ nsCSSFrameConstructor::ContentRemoved() Line 10141 C++ nsCSSFrameConstructor::RecreateFramesForContent() Line 12102 C++ nsCSSFrameConstructor::RestyleElement() Line 10600 C++ nsCSSFrameConstructor::ProcessOneRestyle() Line 14147 C++ nsCSSFrameConstructor::ProcessPendingRestyles() Line 14201 C++ nsCSSFrameConstructor::RestyleEvent::HandleEvent() Line 14265 C++ HandleRestyleEvent() Line 14274 C++ PL_HandleEvent() Line 688 C PL_ProcessPendingEvents() Line 623 C _md_EventReceiverProc() Line 1408 C 7e418724 7e418806 7e4189bd 7e4193f2 7e418a00 nsAppShell::Run() Line 133 C++ nsAppStartup::Run() Line 151 C++ XRE_main() Line 2711 C++ main() Line 61 C++ mainCRTStartup() Line 398 C 7c816fd7
This is a 1.8.1 branch blocker, but we can't take the 500K trunk patch ("redesign XUL popups"). Please see if there's a local fix for just this crash.
Created attachment 280653 [details] [diff] [review] WIP for 1.8 Fixes the crash, but I do see ###!!! ASSERTION: frame was not removed from primary frame map before destruction or was readded to map after being removed: 'Not Reached', file /home/smaug/mozilla/mozilla_cvs/18_branch/mozilla/layout/base/nsFrameManager.cpp, line 734 That assertion is (sort of) from bug Bug 265404.
Other possibility is to not to notify when removing the attribute. Same assertions happen then though. And not notifying means no mutation events when dom is mutated; that is hackish.
Are there any other possibilities?
Smaug: is this likely to get a branch patch, reviews and approvals by tomorrow night (220.127.116.11 code freeze)?
Comment on attachment 280653 [details] [diff] [review] WIP for 1.8 Roc, what do you think about this. Not perfect, but possibly enough for branch. Not sure if some problem occurs when recreating a frame for menu...
Comment on attachment 280653 [details] [diff] [review] WIP for 1.8 looks OK but Neil should take a look
Comment on attachment 280653 [details] [diff] [review] WIP for 1.8 approved for 18.104.22.168, a=dveditz
Fix checked in for Olli
Thanks Daniel. Somehow I didn't notice this got reviews.
verified fixed 22.214.171.124 using Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:126.96.36.199) Gecko/2007100816 Firefox/188.8.131.52 and Mozilla/5.0 (Macintosh; U; Intel Mac OS X; ja-JP-mac; rv:184.108.40.206) Gecko/2007100816 Firefox/220.127.116.11 and the testcase from this bug. -> no crash on testcase - adding verified keyword
10 years ago
Comment on attachment 280653 [details] [diff] [review] WIP for 1.8 a=asac for 18.104.22.168 (same patch shipped by distros for some time now)
checked in on 1.8.0 branch
crash test landed http://hg.mozilla.org/mozilla-central/rev/4457b7408447