bug 716638 added an OOMAllocationSize annotation for when we crash with out of memory. This will be in the raw JSON data. We should run a query against our crashes with empty signatures to find out what percentage of them have this annotation (and thus are OOM crashes), and what the distribution of values is for empty signature crashes with this annotation.
Specifically, this data will now be present on Aurora (12.0a2) and Nightly (13.0a1).
I think we should modify the signature generation algorithm so that when OOMAllocationSize is set but no other information is available, the signature is "mozalloc_abort" instead of "empty signature". This would match the cases where we do have a stack trace, such as bp-f9122be1-204d-4c5a-b49e-f0b5e2120304.
Depends on: 716638
(In reply to Jesse Ruderman from comment #2) > I think we should modify the signature generation algorithm so that when > OOMAllocationSize is set but no other information is available, the > signature is "mozalloc_abort" instead of "empty signature". 1) Please file a new bug for that. 2) I'd rather have something more explicit, like "EMPTY: OOM while allocating" or something like that (we don't get "empty signature" nowadays, BTW, we get "EMPTY: no crashing thread identified; corrupt dump").
Okay, submitted bug 734703.
This is probably not as interesting in light of bug 943051's significantly reducing our empty dump volume.
if still desired, file a new bug an we can look at implementing as a processor rule
Status: NEW → RESOLVED
Last Resolved: 11 months ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.