Open Bug 1707879 Opened 4 years ago Updated 2 years ago

nvd3d9wrapx.pdb@0x28bb is probably a wrong signature

Categories

(Eliot :: Symbolication, defect, P2)

Tracking

(Not tracked)

People

(Reporter: willkg, Unassigned)

Details

When symbolicating crash pings using fx-crash-sig 0.1.12 and siggen 1.0.6, we end up with signatures like:

nvd3d9wrapx.pdb@0x28bb

That looks like a frame that wasn't symbolicated. We have an awful lot of them compared to other signatures in the crash ping data.

When we look at Crash Stats, we don't see that at all:

https://crash-stats.mozilla.org/search/?signature=~nvd3d9wrapx&date=%3E%3D2021-01-27T12%3A39%3A00.000Z&date=%3C2021-04-27T12%3A39%3A00.000Z&_facets=signature&_sort=-date&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-signature

I think something's not working right here and there's a problem with fx-crash-sig (library that puts the requests together), Eliot (symbolication api), or siggen such that the resulting signature for a given crash is different between crash ping data and crash report data.

Here's a crash report where the first frame is what we're seeing with the signature from the crash ping:

https://crash-stats.mozilla.org/report/index/88133179-a8d7-4495-8891-c471f0210427

I can probably create a test case from that.

Mmm... I looked at that crash report and converted it to a stack and that symbolicated fine, but the way I convert it isn't equivalent to what we'd get in a crash ping.

Jim: Can you get me one of the crash pings that leads to that signature?

Flags: needinfo?(jmathies)

I remember these showing up a while back under the GetInfoForSelf signature, but they feel off the radar about a month ago or so. If I can find one I'll dump the report and post back.

Flags: needinfo?(jmathies)

Moving bugs into new Eliot product.

Product: Tecken → Eliot
You need to log in before you can comment on or make changes to this bug.