124517 - Double deletes happen because we can make more than one nsMacMemoryCushion

Status: RESOLVED FIXED

(Core :: XUL, defect, P2)

Description

[Simon Fraser [no longer active]]

We've always worked under the assumption that the nsAppShell in Mac widget code 
is a singleton, and thus given it ownership of the nsMacMemoryCushion.

However, we can create multiple nsAppShell objects, since we make one for each 
nested event loop (e.g. modal dialogs, including the profile picker). This means 
that we can instantiate > 1 nsMacMemoryCushion, each of which install a 
GrowZoneProc, and dispose the buffer handle in their dtors .This is bad.

Comment 1

[Simon Fraser [no longer active]]

Attachment: Patch to ensure that nsMemoryCusion is a singleton

Comment 2

[Steve Dagley]

Nit: softFluffyCusion?  Don't you want a softFluffyCushion?  And the whitespace 
in nsAppShell::Create() seems off.  Other than that, r=sdagley

Comment 3

[Simon Fraser [no longer active]]

I've been having a bad speling day today

Comment 4

[Simon Fraser [no longer active]]

*** Bug 106283 has been marked as a duplicate of this bug. ***

Comment 5

[Simon Fraser [no longer active]]

*** Bug 88946 has been marked as a duplicate of this bug. ***

Comment 6

[Patrick C. Beard]

Comment on attachment 68679 [details] [diff] [review]
Patch to ensure that nsMemoryCusion is a singleton

sr=beard (r=sdagley already)
same nits...

Comment 7

[Steve Dagley]

*** Bug 114907 has been marked as a duplicate of this bug. ***

Comment 8

[Mike Pinkerton (not reading bugmail)]

what does introducting 

+std::auto_ptr<nsMacMemoryCushion> gMemoryCusion;

do to embedding clients? Will this ever get released? Doesn't seem to play nice.'

Comment 9

[Simon Fraser [no longer active]]

+std::auto_ptr<nsMacMemoryCushion> gMemoryCusion;

The delete will happen at library unload time, which seems correct.

Comment 10

[Simon Fraser [no longer active]]

Fix checked in.

Comment 11

[Simon Fraser [no longer active]]

*** Bug 93458 has been marked as a duplicate of this bug. ***

Comment 12

[Conrad Carlen (not reading bugmail)]

*** Bug 121664 has been marked as a duplicate of this bug. ***