Avoid crash in trying to access profile while shutting down.

Avoid crash in trying to access profile while shutting down.

See http://code.google.com/p/chromium/issues/detail?id=128422#c9 for details.

BUG=128422
TEST=Watch the crash rate. I don't have a reproducible test case.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=141554

[sreeram, 2012-06-11 22:03:58]

Please review.

[Jói, 2012-06-11 22:44:47]

I still recommend continuing to try to understand the root cause and
experimenting with changing the shutdown order as I suggested in the bug.

This change LGTM once my doubt below is addressed.

Cheers,
Jói

https://chromiumcodereview.appspot.com/10536106/diff/1/chrome/browser/ui/brow...
File chrome/browser/ui/browser.cc (right):

https://chromiumcodereview.appspot.com/10536106/diff/1/chrome/browser/ui/brow...
chrome/browser/ui/browser.cc:3907: if
(browser_shutdown::ShuttingDownWithoutClosingBrowsers() ||
Don't you want the inverse condition?

[sreeram, 2012-06-11 22:53:03]

http://codereview.chromium.org/10536106/diff/1/chrome/browser/ui/browser.cc
File chrome/browser/ui/browser.cc (right):

http://codereview.chromium.org/10536106/diff/1/chrome/browser/ui/browser.cc#n...
chrome/browser/ui/browser.cc:3907: if
(browser_shutdown::ShuttingDownWithoutClosingBrowsers() ||
On 2012/06/11 22:44:47, Jói wrote:
> Don't you want the inverse condition?

No. If we are shutting down, the first clause returns true, so we don't check
the second clause (which is the one that uses profile and crashes). We'll then
delete the InstantController object, which is fine since we are shutting down.

In the normal case, the first clause returns false, and things behave as before.

[sreeram, 2012-06-11 22:53:56]

On 2012/06/11 22:44:47, Jói wrote:
> I still recommend continuing to try to understand the root cause and
> experimenting with changing the shutdown order as I suggested in the bug.

Sure. This change is meant for merging into M20. I'll continue to work on the
bug as you recommended.

[Jói, 2012-06-11 22:55:34]

LGTM, thanks for the clarification.

On Mon, Jun 11, 2012 at 10:53 PM,  <sreeram@chromium.org> wrote:
> On 2012/06/11 22:44:47, Jói wrote:
>>
>> I still recommend continuing to try to understand the root cause and
>> experimenting with changing the shutdown order as I suggested in the bug.
>
>
> Sure. This change is meant for merging into M20. I'll continue to work on
> the
> bug as you recommended.
>
> http://codereview.chromium.org/10536106/

[sky, 2012-06-11 23:16:34]

LGTM