56762 - one-time leak from nsRange of 4 nsVoidArray and 1 PRMonitor

Status: RESOLVED FIXED

(Core :: DOM: Selection, defect, P2)

Description

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

Code in nsRange.cpp allocates and never destroys the following static variables:

PRMonitor*   nsRange::mMonitor
nsVoidArray* nsRange::mStartAncestors
nsVoidArray* nsRange::mEndAncestors
nsVoidArray* nsRange::mStartAncestorOffsets
nsVoidArray* nsRange::mEndAncestorOffsets

This can be fixed easily by a static shutdown function called from the module
destructor (nsLayoutModule::Shutdown() I guess).

Comment 1

[Kathleen :Brade]

Setting to Future; I don't think this will meet rtm criteria at this stage
unless someone can explain make a case for it.

Comment 2

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

Attachment: proposed fix

Comment 3

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

Assigning to self since I attached proposed fix.

mjudge - Could you have a look at the attached fix.  Does it seem reasonable?
In particular, is there a way you can guarantee that mMonitor is not entered
when the layout module dtor is called (it certainly shouldn't be during XPCOM
shutdown, but can you guarantee it?).

Comment 4

[mjudge]

the monitor should not be entered if events or JS is not processed while xpcom 
is shutdown. looks good to me. r=mjudge

Comment 5

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

Attachment: oops - add a null check for mMonitor in nsRange::Shutdown

Comment 6

[Scott Collins]

Naturally, I prefer |0| over |nsnull|, but that is not something I would will on
others as a requirement.

sr=scc

Comment 7

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

Fix checked in 2000-10-29 13:38-0800.