432385 - [client] tinderbox should run fix-linux-stack.pl / fix-macosx-stack.pl over trace-malloc leak diffs

Status: RESOLVED FIXED

(Release Engineering :: General, enhancement)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Tinderbox should run fix-linux-stack.pl or fix-macosx-stack.pl over the trace-malloc leak differences output.  (This is related to bug 385248, which requests running it over the log.)

I just modified diffbloatdump.pl so that this can be done pretty efficiently, though it requires generating the leak stats differences a little differently from the way we do now.  In particular, what we should save after each run is the result of:

diffbloatdump.pl --use-address --depth=20 /dev/null sdleak.log | fix-linux-stack.pl

and then we should call diffbloatdump.pl on the result of that call for the previous two runs to get the diff that goes in the output.

This would make the leak stats differences much more useful.

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: untested patch

A rough idea of what I think needs to be done.

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Apparently bhearsum is rewriting the automation that runs this test for buildbot for the mozilla-central tinderboxes.  (The patch I referred to above hasn't been merged to mozilla-central yet, though.)

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

This was done in the buildbot code for mozilla-central in bug 422296; we might want to take the tinderbox patch if it works for 1.9, though...

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Moving bugs that were filed in Webtools/Tinderbox before there was a separate component for bugs in tinderbox/buildbot client to a more appropriate component.

Comment 5

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

Should this bug move to Core:BuildConfig, just like bug#385248 ?

Comment 6

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I think at this point we shuold just say this is fixed (see comment 3).  I don't think we're going to fix this in the 1.9.0.* automation at this point.