Comment #5 on issue 6423 by w...@chromium.org: Chrome: Crash Report - Stack  
Signature: URLRequestHttpJob::ContinueDespiteLastError()-87EC89
http://code.google.com/p/chromium/issues/detail?id=6423

It always bothers me to fix a crash by adding a null pointer
check without knowing why we get the null pointer.  It's
also strange that we only get this crash in the Dev channel
2.0.x.y builds but not in the Stable channel 1.0.154.x
builds.  I didn't find any difference in the relevant source
code between 1.0.154.x and 2.0.x.y (the trunk).

A possible scenario for this crash is that we ignore a
certificate error, so the requests for the subresources
on that page are automatically continued despite the
certificate error, and we close the tab while we're still
loading the subresources.

What I found is that a URLRequest is cancelled and deleted
in two tasks.  So it is possible for the
ContinueDespiteLastError task to sneak in between those
two tasks, finding a URLRequestJob in the
cancelled-but-not-deleted state.  In addition, if the
info->resource_handler->OnResponseCompleted call in
ResourceDispatcherHost::OnResponseCompleted returns false
(which could happen for
CrossSiteResourceHandler::OnResponseCompleted), the
deletion of the URLRequest is further delayed.

In URLRequestHttpJob, I found null pointer checks for
transaction_ in the Kill and OnStartedCompleted methods.
The null pointer check in Kill was added to allow us
to call Kill more than once, so it's for a different
purpose.  But the null pointer check in OnStartedCompleted
may have been added to fix a similar crash.

This is all I can find now.  I'm going to proceed with
a null pointer check.

--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

--~--~---------~--~----~------------~-------~--~----~
Automated mail from issue updates at http://crbug.com/
Subscription options: http://groups.google.com/group/chromium-bugs
-~----------~----~----~----~------~----~------~--~---

Reply via email to