663229 - e10s HTTP: initialize prnetaddrs correctly in child

Status: RESOLVED FIXED

(Core :: Networking: HTTP, defect)

Description

[Jason Duell]

Spin-off of bug 661158.  Just minor codefix to move some initializers that should now be in HttpBaseChannel, not nsHttpChannel.

Comment 1

[Jason Duell]

Attachment: Approved patch copied from bug 661158

Copying jdm's +r.

Includes breakpoint to cause parent to crash (intentionally) if we see it trying to pass in a Unix domain socket enum over IPDL.

Comment 2

[Jason Duell]

http://hg.mozilla.org/integration/mozilla-inbound/rev/a19e5f55f67c

Comment 3

[Mounir Lamouri (:mounir)]

http://hg.mozilla.org/mozilla-central/rev/a19e5f55f67c

Comment 4

[Josh Matthews [:jdm]]

Comment on attachment 538347 [details] [diff] [review]
Approved patch copied from bug 661158

So, this patch does not actually fix the topcrash that is tracked by bug 661158. In fact, it arguably makes the crash strictly worse, since it changes a content process crash into a browser crash. On the other hand, this change should make us crash at the point when we start sending bad data over the wire, and this is good because Jason and I can't figure out why this would ever happen. Obviously it is, and it's happening fairly frequently, so this kind of data would be valuable to us.

Comment 5

[Jason Duell]

Comment on attachment 538347 [details] [diff] [review]
Approved patch copied from bug 661158

> this change should make us crash at the point when we start sending bad 
> data over the wire, and this is good 

+1

Comment 6

[Asa Dotzler [:asa]]

We didn't have a mobile rep at today's triage but desktop has no concerns here so I'm delegating the yes/no to which ever of the Mobile drivers gets here first.

Comment 7

[Brad Lassey [:blassey] (use needinfo?)]

Comment on attachment 538347 [details] [diff] [review]
Approved patch copied from bug 661158

since this doesn't actually fix anything I don't think we want to take this. If my understanding is incorrect, please explain and re-nom