1135272 - crash in imgStatusTracker::SyncNotify(imgRequestProxy*)

Status: RESOLVED FIXED

(Core :: Graphics: ImageLib, defect)

Description

[Mats Palmgren (inactive)]

This bug was filed from the Socorro interface and is 
report bp-170a68e7-deed-465c-bca0-1c6cb2150213.
=============================================================

This signature is at #11 on "Top Crashers for Firefox 35.0.1",
with 4307 reported crashes per week in that channel (I suspect
the real number is higher).


imgStatusTracker::SyncNotify(imgRequestProxy*)
imgRequestProxy::SyncNotifyListener()
imgCacheValidator::OnStartRequest(nsIRequest*, nsISupports*)
mozilla::net::nsHttpChannel::CallOnStartRequest()
mozilla::net::nsHttpChannel::ContinueOnStartRequest2(tag_nsresult)
mozilla::net::nsHttpChannel::OnStartRequest(nsIRequest*, nsISupports*)
nsInputStreamPump::OnStateStart()
nsInputStreamPump::OnInputStreamReady(nsIAsyncInputStream*)
nsInputStreamReadyEvent::Run()
nsThread::ProcessNextEvent(bool, bool*)
NS_ProcessPendingEvents(nsIThread*, unsigned int)
nsBaseAppShell::NativeEventCallback()
...


It looks like imgStatusTracker was renamed mozilla::image::ProgressTracker
so the crashes continues in >35 under that name:
bp-49151938-e16e-4818-846d-4e7092150123

mozilla::image::ProgressTracker::SyncNotify(mozilla::image::IProgressObserver*)
imgRequestProxy::SyncNotifyListener()
imgCacheValidator::OnStartRequest(nsIRequest*, nsISupports*)
mozilla::net::nsHttpChannel::CallOnStartRequest()
mozilla::net::nsHttpChannel::ContinueOnStartRequest2(tag_nsresult)
mozilla::net::nsHttpChannel::OnStartRequest(nsIRequest*, nsISupports*)
nsInputStreamPump::OnStateStart()
nsInputStreamPump::OnInputStreamReady(nsIAsyncInputStream*)
nsInputStreamReadyEvent::Run()
nsThread::ProcessNextEvent(bool, bool*)
mozilla::ipc::MessagePump::Run(base::MessagePump::Delegate*)
MessageLoop::RunHandler()
MessageLoop::Run()
nsBaseAppShell::Run()
nsAppShell::Run()
nsAppStartup::Run()
XREMain::XRE_mainRun()
...

Comment 1

[Mats Palmgren (inactive)]

[Tracking Requested - why for this release]: topcrash

Comment 2

[Javaun Moradi [:javaun]]

Removing 1135274, I fat-fingered this. My bad

Comment 3

[Lawrence Mandel [:lmandel] (use needinfo)]

We've recently taken a substantial number of changes to ImageLib on 37. The changes have been on 38 for a while. Do we have any incidence of this crash signature on 38? We may wan to wait until 37 Beta 1 to see if this is still a topcrash.

Seth - Do you have an idea about whether you expect that this is still an issue with the fixes that you uplifted over the last two days?

Comment 4

[Sylvestre Ledru [:Sylvestre]]

That is also a wontfix for 36...

Comment 5

[Seth Fowler [:seth] [:s2h]]

(In reply to Lawrence Mandel [:lmandel] (use needinfo) from comment #3)
> Seth - Do you have an idea about whether you expect that this is still an
> issue with the fixes that you uplifted over the last two days?

I don't see any crashes with this signature after Firefox 35. It's unfortunate that we don't know what caused this or what resolved this, but it appears to be fixed.

I'm going to mark this as resolved; we can reopen if anyone hits this in Firefox 36 or above. Updating status flags to reflect this.

Comment 6

[Mats Palmgren (inactive)]

Seth, I guess you missed this note:

(In reply to Mats Palmgren (:mats) from comment #0)
> It looks like imgStatusTracker was renamed mozilla::image::ProgressTracker
> so the crashes continues in >35 under that name:
> bp-49151938-e16e-4818-846d-4e7092150123

Comment 7

[Seth Fowler [:seth] [:s2h]]

(In reply to Mats Palmgren (:mats) from comment #6)
> Seth, I guess you missed this note:

Argh, you're right. I didn't miss that note, but I did click on the wrong crashstats link. Thanks!

Comment 8

[Lawrence Mandel [:lmandel] (use needinfo)]

Marking 37+ as affected and tracking as this is still an active crash.

Comment 9

[Lawrence Mandel [:lmandel] (use needinfo)]

Seth - Only 3 Betas left in the 37 cycle. Can you spend some time on this to try and get a fix together for 37?

Comment 10

[Seth Fowler [:seth] [:s2h]]

(In reply to Lawrence Mandel [:lmandel] (use needinfo) from comment #9)
> Seth - Only 3 Betas left in the 37 cycle. Can you spend some time on this to
> try and get a fix together for 37?

I'll give you the details on IRC but I think the short answer is that this is probably wontfix for 37 as things stand now.

Comment 11

[Lawrence Mandel [:lmandel] (use needinfo)]

The short summary is that the real fix is going to be pretty large and probably too risky for Beta. Seth is going to investigate a more targeted fix that will likely have a perf hit for 37.

Comment 12

[Lawrence Mandel [:lmandel] (use needinfo)]

Should be fixed by bug 1142849.

Comment 13

[Robert Kaiser]

(In reply to Lawrence Mandel [:lmandel] (use needinfo) from comment #12)
> Should be fixed by bug 1142849.

Are we sure about that? This signature has been rising significantly on beta since 37.0b5 was released.

Comment 14

[Lawrence Mandel [:lmandel] (use needinfo)]

No. Not sure. Bug 1142849 has landed on m-c and m-a. The intention was to check the 3 day numbers to see if the crash volume decreases on those channels before uplifting to Beta. The fix is in Monday's Nightly build and Tuesday's Aurora build.

Comment 15

[Liz Henry (:lizzard) (relman/hg->git project)]

There's only one crash each in the last week with this signature, for 38 and 39: mozilla::image::ProgressTracker::SyncNotify(mozilla::image::IProgressObserver*)

No crashes in the last week for 37, 38 or 39 for [@ mozilla::image::ProgressTracker::SyncNotify(imgRequestProxy*) ]

No crashes for 37, 38, or 39 for [@ imgStatusTracker::SyncNotify(imgRequestProxy*) ]. 

So either this fix helped on nightly and Aurora, or something else did.

Comment 16

[Seth Fowler [:seth] [:s2h]]

Yeah, it looks like the fix helped. I've gone ahead and requested uplift for the patch in bug 1142849.

Comment 17

[Liz Henry (:lizzard) (relman/hg->git project)]

There are still a few signatures from 37 for [@ mozilla::image::ProgressTracker::SyncNotify(mozilla::image::IProgressObserver*) ] but the volume has dropped across all 3 associated crash signatures. I think we can call this fixed for now.