397854 - need leak tests for measuring kernel, gdi, and user handles

Status: REOPENED

(Testing :: General, defect)

Description

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

We should have a leak measurement tool (parallel to, or perhaps attached to one of the nsTraceRefcnt or trace-malloc ways of measuring leaks) that measures leaks of User, GDI, and Kernel handles on Windows.  See

http://msdn2.microsoft.com/en-us/library/ms724457.aspx

for some basic documentation and

The based way to debug them may be to hook all the functions that allocate and release them (like trace-malloc does for the allocation functions).  These functions are:

for GDI object handles:
http://msdn2.microsoft.com/en-us/library/ms724291.aspx

for User object handles:
http://msdn2.microsoft.com/en-us/library/ms725486.aspx

for Kernel object handles:
http://msdn2.microsoft.com/en-us/library/ms724485.aspx

Comment 1

[Geoff Brown [:gbrown]]

Mass closing bugs with no activity in 2+ years. If this bug is important to you, please re-open.

Comment 2

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Is this still worthwhile, or has it become less relevant over time?

Comment 3

[Bas Schouten (:bas.schouten)]

(In reply to David Baron :dbaron: ⌚️UTC-8 from comment #2)
> Is this still worthwhile, or has it become less relevant over time?

Essentially this has become a lot less relevant since we don't use cairo anymore to access GDI, in other words we're using significantly less 'large' handle based objects. However of course leaking kernel handles could still be bad, and one might argue our increased use of shmems may have increased our usage there somewhat.

Comment 4

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

OK... I think this is still a known gap in our testing that we should consider working on, so I'm going to reopen.