Closed Bug 696302 Opened 13 years ago Closed 7 years ago

Automatically re-run valgrind with --track-origins=yes upon encountering an uninitialized-memory bug

Categories

(Release Engineering :: General, enhancement, P3)

x86
Linux
enhancement

Tracking

(Not tracked)

RESOLVED INCOMPLETE

People

(Reporter: jruderman, Unassigned)

References

Details

(Whiteboard: [valgrind])

When a Firefox-under-Valgrind process fails
  (with Memcheck:{Cond|Param|Value*})
the process should automatically be re-run with --track-origins=yes.

This will help track down uninitialized-memory bugs, as long as they're reproducible.
Does this require doing the entire thing again?  Is there a reason not to have this flag on every run?
Whiteboard: [valgrind
Priority: -- → P3
Whiteboard: [valgrind → [valgrind]
> Does this require doing the entire thing again?

The entire process, yes. So if Mochitest-2 is one process, you'd have to redo Mochitest-2.

> Is there a reason not to have this flag on every run?

It slows down Valgrind by a factor of two or so. And Valgrind is already pretty slow.

So assuming* most runs don't hit new uninitialized-memory bugs, it's probably best not to have --track-origins=yes on every run.

[*] This should be true once we fix bug 696305. I suspect it's not true now.
Depends on: 696305
Product: mozilla.org → Release Engineering
Component: Other → General Automation
QA Contact: catlee
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → INCOMPLETE
Component: General Automation → General
You need to log in before you can comment on or make changes to this bug.