1075530 - Win64 Nightly crash in nsHostResolver::FlushCache() on address 0x18

Status: RESOLVED DUPLICATE of bug 1024614

(Core :: Networking: DNS, defect)

Description

[Robert Kaiser]

This bug was filed from the Socorro interface and is 
report bp-d91951f7-8459-4173-9c07-073342141001.
=============================================================

Top frames:
0 	xul.dll 	nsHostResolver::FlushCache() 	netwerk/dns/nsHostResolver.cpp
1 	xul.dll 	nsDNSService::Observe(nsISupports*, char const*, wchar_t const*) 	netwerk/dns/nsDNSService2.cpp
2 	xul.dll 	nsObserverList::NotifyObservers(nsISupports*, char const*, wchar_t const*) 	xpcom/ds/nsObserverList.cpp
3 	xul.dll 	nsObserverService::NotifyObservers(nsISupports*, char const*, wchar_t const*) 	xpcom/ds/nsObserverService.cpp
4 	xul.dll 	nsNotifyAddrListener::ChangeEvent::Run() 	netwerk/system/win32/nsNotifyAddrListener.cpp
5 	xul.dll 	nsThread::ProcessNextEvent(bool, bool*) 	xpcom/threads/nsThread.cpp
[...]

From what I can see in the reports tab of https://crash-stats.mozilla.com/report/list?signature=nsHostResolver%3A%3AFlushCache%28%29 those reports are all on Win64 and EXCEPTION_ACCESS_VIOLATION_READ at 0x18. They're all on Win7 or higher.

This started with 2014-09-26 builds on Nightly 35.0a1, so some patch that landed on the 25th is probably to blame.

Comment 1

[Robert Kaiser]

Bug 939318 was pushed on the 25th and is the last change in both code lines of the top two frames in the stack, so I'm pretty sure it's connected to this.

Daniel, any idea on this?

Comment 2

[James Willcox (:snorp) ([e-mail]) (he/him)]

This should be fixed by one of the patches in bug 1024614.

Comment 3

[Daniel Stenberg [:bagder]]

The fix for bug 1024614 landed in m-c on Sep 29th, so if that fixed this problem we should be able to see it... But would that bug really get visualized as a crash _within_ FlushCache() as this crash dump suggests?

Comment 4

[James Willcox (:snorp) ([e-mail]) (he/him)]

(In reply to Daniel Stenberg [:bagder] from comment #3)
> The fix for bug 1024614 landed in m-c on Sep 29th, so if that fixed this
> problem we should be able to see it... 

I don't see any crashes past Sep 29, but I might not be driving soccoro correctly. :kairo does this look fixed to you?

> But would that bug really get
> visualized as a crash _within_ FlushCache() as this crash dump suggests?

Yup. When you call a method on a null object, nothing bad happens until you try to access instance data. In this case I guess it's probably the mLock in nsHostResolver that blows up.

Comment 5

[Robert Kaiser]

(In reply to James Willcox (:snorp) ([e-mail]) from comment #4)
> (In reply to Daniel Stenberg [:bagder] from comment #3)
> > The fix for bug 1024614 landed in m-c on Sep 29th, so if that fixed this
> > problem we should be able to see it... 
> 
> I don't see any crashes past Sep 29, but I might not be driving soccoro
> correctly. :kairo does this look fixed to you?

Oopps, you're right. Fun that a bug with "Android" in the summary would have fixed Win64, is that really the right number?