1816661 - Replace the address with the adjusted address (and provide the raw address?)

Status: RESOLVED DUPLICATE of bug 1827495

(Socorro :: Processor, enhancement)

Description

[Gabriele Svelto [:gsvelto]]

The stackwalker has been wired up to detect scenarios where the crashing address is different from what is reported by the exception handler, either because it wasn't reported or because it was adjusted before access. In those cases the stackwalker adds a field to the crash_info structure called adjusted_address. When this field is present the crash address shown by Socorro should be adjusted like so:

• If crash_info.adjusted_address.kind is the "non-canonical" string then crash_info.adjusted_address.address should be used instead of crash_info.address
• If crash_info.adjusted_address.kind is the "null-pointer" string then the null pointer should be used instead of crash_info.address (that is 0x00000000 on 32-bit builds and 0x0000000000000000 on 64-bit ones).

For extra clarity towards however is analyzing the crash an additional field should be added to the crash report holding the address that was replaced (we could call it raw_address, exception_address or something along the lines).

Comment 1

[Will Kahn-Greene [:willkg] ET needinfo? me]

This bug covers what gets shown in the report view for a crash report in Crash Stats. I'm pretty sure we need to update the stackwalker before working on this.

Comment 2

[Gabriele Svelto [:gsvelto]]

Yes, I've cut a new release for this. Should I file a separate bug to pick up the new release? There's a few changes to the output but they boil down to new fields we added, none of the old JSON fields were changed.

Comment 3

[Will Kahn-Greene [:willkg] ET needinfo? me]

We had agreed that I wouldn't be updating the stackwalker until you asked me to which reduces the maintenance burden on me and makes it easier for us to coordinate on the state of the stackwalker vs. what's running in the Socorro processor. I'll write up a bug for this update. Going forward, you can either write up a bug or let me know.

Comment 4

[Gabriele Svelto [:gsvelto]]

Thanks!

Comment 5

[Will Kahn-Greene [:willkg] ET needinfo? me]

Oops--looks like we had two of these bugs. I'm going to dupe this to the one with the work in it.