Fix a memory leak in ExtensionMessageService.

Fix a memory leak in ExtensionMessageService.

Also grasp at straws trying to fix a crash bug.

BUG=21201
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=31024

[Erik does not do reviews, 2009-11-04 17:00:33]

http://codereview.chromium.org/345058/diff/5/1004
File chrome/browser/extensions/extension_message_service.cc (right):

http://codereview.chromium.org/345058/diff/5/1004#newcode316
Line 316: CHECK(receiver.sender);
remove redundant CHECK

http://codereview.chromium.org/345058/diff/5/1005
File chrome/browser/extensions/extension_message_service.h (right):

http://codereview.chromium.org/345058/diff/5/1005#newcode132
Line 132: typedef std::map<int, MessageChannel*> MessageChannelMap;
I think I'm missing something about the change away from linked_ptr.  Isn't this
equivalent to the old code (but less automatic)?

[Matt Perry, 2009-11-04 18:47:46]

http://codereview.chromium.org/345058/diff/5/1005
File chrome/browser/extensions/extension_message_service.h (right):

http://codereview.chromium.org/345058/diff/5/1005#newcode132
Line 132: typedef std::map<int, MessageChannel*> MessageChannelMap;
On 2009/11/04 17:00:33, Erik Kay wrote:
> I think I'm missing something about the change away from linked_ptr.  Isn't
this
> equivalent to the old code (but less automatic)?

Yes, theoretically. This is the "grasping at straws" part. I'm testing if
there's something about my usage of linked_ptr that may have caused a double
free. I couldn't find anything else in this file that was suspect. Json parsing
and serialization held up to fuzzing under purify.

[Erik does not do reviews, 2009-11-04 19:43:37]

ok.  LGTM


On Wed, Nov 4, 2009 at 10:47 AM,  <mpcomplete@chromium.org> wrote:
>
> http://codereview.chromium.org/345058/diff/5/1005
> File chrome/browser/extensions/extension_message_service.h (right):
>
> http://codereview.chromium.org/345058/diff/5/1005#newcode132
> Line 132: typedef std::map<int, MessageChannel*> MessageChannelMap;
> On 2009/11/04 17:00:33, Erik Kay wrote:
>>
>> I think I'm missing something about the change away from linked_ptr.
>
> Isn't this
>>
>> equivalent to the old code (but less automatic)?
>
> Yes, theoretically. This is the "grasping at straws" part. I'm testing
> if there's something about my usage of linked_ptr that may have caused a
> double free. I couldn't find anything else in this file that was
> suspect. Json parsing and serialization held up to fuzzing under purify.
>
> http://codereview.chromium.org/345058
>