1385611 - Crash in nsInputStreamReadyEvent::Run

Status: RESOLVED DUPLICATE of bug 1363372

(Core :: Networking: HTTP, defect)

Description

[Marcia Knous [:marcia]]

This bug was filed from the Socorro interface and is 
report bp-b5612272-0b47-48ac-93be-d11800170720.
=============================================================

Seen while looking at nightly crash stats - crashes started back in 20170719030206: http://bit.ly/2hcoub6. 47 crashes/33 installs

Possible regression range based on Build ID: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=5e73b9798464c3f7106f0161dc9a49b234f42f9c&tochange=1b065ffd8a535a0ad4c39a912af18e948e6a42c1

Comment 1

[Andrew McCreight [:mccr8]]

These are all null derefs. Looking at the regression range in comment 0, bug 1379762 is the only one that stands out to me as having anything to do with streams or input events (part 3 is "Don't mess about with load blockers if our document is already in the COMPLETE readyState."). Boris, could this be a regression from your patch?

Comment 2

[Boris Zbarsky [:bzbarsky]]

Seems unlikely.  Note also that part 3 was backed out; was that reflected in the crash stats?

What bothers me is this: what's null?  "this" better not be null, right?  mCallback was null-checked.  mCallback's vtable could be null, I suppose, but we're holding a strong ref to mCallback, right?

The crash is at address 10 (decimal).  That's not an offset I'd expect to see for any of mCallback, mStream, or a vtable entry...

Comment 3

[Andrew McCreight [:mccr8]]

(In reply to Boris Zbarsky [:bz] from comment #2)
> Seems unlikely.  Note also that part 3 was backed out; was that reflected in
> the crash stats?

No, they are continuing through today's build, so I guess it is something else.

> What bothers me is this: what's null?  "this" better not be null, right? 
> mCallback was null-checked.  mCallback's vtable could be null, I suppose,
> but we're holding a strong ref to mCallback, right?
> 
> The crash is at address 10 (decimal).  That's not an offset I'd expect to
> see for any of mCallback, mStream, or a vtable entry...

Yeah, I don't know. These callback-ish crash stacks are often hard to interpret like this.

Comment 4

[Dragana Damjanovic [:dragana]]

This is TFO bug.

Comment 6

[Gerry Chang [:gchang]]

Mark 56 disabled as TCP Fast Open was disabled in 56.