Closed
Bug 1130914
Opened 9 years ago
Closed 9 years ago
crash in OOM | unknown | NS_ABORT_OOM(unsigned int) | PL_DHashTableAdd(PLDHashTable*, void const*) | CCGraphBuilder::AddNode(void*, nsCycleCollectionParticipant*)
Categories
(Core :: XPCOM, defect)
Tracking
()
RESOLVED
FIXED
People
(Reporter: away, Unassigned)
References
Details
(Keywords: crash)
Crash Data
This bug was filed from the Socorro interface and is report bp-5fbc3697-b015-4d91-9776-67d2e2150207. ============================================================= 0 xul.dll NS_ABORT_OOM(unsigned int) xpcom/base/nsDebugImpl.cpp 1 xul.dll PL_DHashTableAdd(PLDHashTable*, void const*) xpcom/glue/pldhash.cpp 2 xul.dll CCGraphBuilder::AddNode(void*, nsCycleCollectionParticipant*) xpcom/base/nsCycleCollector.cpp 3 xul.dll CCGraphBuilder::NoteRoot(void*, nsCycleCollectionParticipant*) xpcom/base/nsCycleCollector.cpp 4 xul.dll CCGraphBuilder::NoteJSRoot(void*) xpcom/base/nsCycleCollector.cpp The really strange thing is that it's "OOM | unknown" meaning the size parameter was zero. How could that be?
Flags: needinfo?(n.nethercote)
Comment 1•9 years ago
|
||
Hash tables are lazily initialized. If the hashtable is empty, and we attempt to initialize it, then EntryCount() will be zero, and we'll report zero. I think the better way to handle this would be to return whatever size we need for the initial table if the entrycount is currently zero, and assume that any initialized table can hold at least two things.
Updated•9 years ago
|
Component: General → XPCOM
Comment 2•9 years ago
|
||
Yeah, to paraphrase mccr8, we report the size of the table before we tried to add to it, but it would be better to report the size we attempted to allocate. BTW, did you used to see some OOM aborts in CCGraph::Init()? Because this OOM abort should have replaced those ones.
Flags: needinfo?(n.nethercote)
Comment 3•9 years ago
|
||
I've seen crashes creating the CC graph hashtable, though I don't recall what the signature was.
Comment 4•9 years ago
|
||
David, is this crash happening a lot, or did the 0 of the allocation size just draw your attention?
Kind of both. It's one of the top OOMs in recent nightlies (which doesn't say much since nightly is pretty low-OOM) but yeah the 0 is what made it stand out.
Comment 6•9 years ago
|
||
Given that bug 1050035's patches have been backed out and we have bug 1131097, I think this can be closed.
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•