1033493 - Hangs/long loops on nsUrlClassifierDBServiceWorker thread, pages and its subresources take ages to load

Status: RESOLVED DUPLICATE of bug 1164518

(Toolkit :: Safe Browsing, defect)

Description

[Honza Bambas (:mayhemer)]

Nightly, 83c09fe3a658, BuildID=20140701030202, Win8.

Started to manifest on visit of https://www.conrad.cz/kosik/ (with having something in the basket of course) and pressing the 'Pokračovat' button and then the browsr back button.  Not sure however these are reliable STR.

Symptoms:
- page loading slowed *rapidly* down, appeared like an ISP problem for me first
- I tried to restart Firefox
=> shutdown hang, with one background thread spinning (some 75% of one of the cores):


Main thread:
 	xul.dll!nsThread::Shutdown()  Line 598 + 0xa bytes	C++
 	xul.dll!nsUrlClassifierDBService::Shutdown()  Line 1547	C++
 	xul.dll!nsUrlClassifierDBService::Observe(nsISupports * aSubject, const char * aTopic, const wchar_t * aData)  Line 1502	C++
 	xul.dll!nsObserverList::NotifyObservers(nsISupports * aSubject, const char * aTopic, const wchar_t * someData)  Line 96 + 0x14 bytes	C++
 	xul.dll!nsObserverService::NotifyObservers(nsISupports * aSubject, const char * aTopic, const wchar_t * someData)  Line 303 + 0xf bytes	C++
 	xul.dll!nsXREDirProvider::DoShutdown()  Line 871	C++
 	xul.dll!ScopedXPCOMStartup::~ScopedXPCOMStartup()  Line 1197	C++
 	xul.dll!ScopedXPCOMStartup::`scalar deleting destructor'()  + 0x8 bytes	C++
 	xul.dll!XREMain::XRE_main(int argc, char * * argv, const nsXREAppData * aAppData)  Line 4114	C++
 	xul.dll!XRE_main(int argc, char * * argv, const nsXREAppData * aAppData, unsigned int aFlags)  Line 4298 + 0xd bytes	C++



Classifier thread:
	xul.dll!mozilla::safebrowsing::AddPrefix::Compare<mozilla::safebrowsing::AddPrefix>(const mozilla::safebrowsing::AddPrefix & other)  Line 153 + 0x4 bytes	C++
 	xul.dll!mozilla::safebrowsing::EntryCompare<mozilla::safebrowsing::AddPrefix>::Compare(const void * e1, const void * e2)  Line 259	C++
 	msvcr100.dll!shortsort(char * lo, char * hi, unsigned int width, int (const void *, const void *)* comp)  Line 355 + 0x5 bytes	C
 	msvcr100.dll!qsort(void * base, unsigned int num, unsigned int width, int (const void *, const void *)* comp)  Line 136 + 0xd bytes	C
 	xul.dll!mozilla::safebrowsing::EntrySort<mozilla::safebrowsing::AddPrefix,nsTArrayFallibleAllocator>(nsTArray_Impl<mozilla::safebrowsing::AddPrefix,nsTArrayFallibleAllocator> & aArray)  Line 272 + 0x15 bytes	C++
 	xul.dll!mozilla::safebrowsing::Merge<mozilla::safebrowsing::AddPrefix>(mozilla::safebrowsing::ChunkSet * aStoreChunks, FallibleTArray<mozilla::safebrowsing::AddPrefix> * aStorePrefixes, mozilla::safebrowsing::ChunkSet & aUpdateChunks, FallibleTArray<mozilla::safebrowsing::AddPrefix> & aUpdatePrefixes, bool aAllowMerging)  Line 481	C++
 	xul.dll!mozilla::safebrowsing::HashStore::ApplyUpdate(mozilla::safebrowsing::TableUpdate & update)  Line 497 + 0x14 bytes	C++
 	xul.dll!mozilla::safebrowsing::Classifier::ApplyTableUpdates(nsTArray<mozilla::safebrowsing::TableUpdate *> * aUpdates, const nsACString_internal & aTable)  Line 631	C++
 	xul.dll!mozilla::safebrowsing::Classifier::ApplyUpdates(nsTArray<mozilla::safebrowsing::TableUpdate *> * aUpdates)  Line 329	C++
 	xul.dll!nsUrlClassifierDBServiceWorker::CacheCompletions(nsTArray<mozilla::safebrowsing::CacheResult> * results)  Line 709	C++
 	xul.dll!UrlClassifierDBServiceWorkerProxy::CacheCompletionsRunnable::Run()  Line 166	C++
 	xul.dll!nsThread::ProcessNextEvent(bool aMayWait, bool * aResult)  Line 766 + 0x8 bytes	C++
 	xul.dll!NS_ProcessNextEvent(nsIThread * aThread, bool aMayWait)  Line 284 + 0x1f bytes	C++
 	xul.dll!mozilla::ipc::MessagePumpForNonMainThreads::Run(base::MessagePump::Delegate * aDelegate)  Line 337 + 0xa bytes	C++
 	xul.dll!MessageLoop::RunHandler()  Line 223	C++
 	xul.dll!MessageLoop::Run()  Line 197	C++
 	xul.dll!nsThread::ThreadFunc(void * aArg)  Line 355	C++
 	nss3.dll!_PR_NativeRunThread(void * arg)  Line 419	C
 	msvcr100.dll!_getptd_noexit()  Line 498 + 0x7 bytes	C
 	msvcr100.dll!_callthreadstartex()  Line 314 + 0x6 bytes	C
 	msvcr100.dll!_threadstartex(void * ptd)  Line 292 + 0x5 bytes	C
 	kernel32.dll!@BaseThreadInitThunk@12()  + 0x12 bytes	
 	ntdll.dll!__RtlUserThreadStart()  + 0x4e bytes	
 	ntdll.dll!__RtlUserThreadStart@8()  + 0x1c bytes	


I attached a debugger, paused (symbols loaded, no sources) and tried to put breakpoints up to NS_ProcessNextEvent.  It well popped up to that frame on re-run and then started to loop (process messages) again, like the code would be posting to the classifier thread on and on other events or there already where a huge number scheduled before and each was taking enormous amount of time to process.  Please be aware that (now I realize) I didn't check the classifier thread was the one calling NS_ProcessNextEvent on and on, so it could well be main thread, sorry :/



Second problem:

After kill and restart, I went to the same page (session store) and tried to complete my online purchase once more.  And it started again, same symptoms.  Another debugging sessions shows one stack:

	ntdll.dll!_KiFastSystemCallRet@0() 	
 	ntdll.dll!_ZwSetInformationFile@20()  + 0xa bytes	
 	KernelBase.dll!MoveFileWithProgressTransactedW()  + 0x1d2 bytes	
 	KernelBase.dll!_MoveFileExW@12()  + 0x18 bytes	
 	xul.dll!nsLocalFile::CopySingleFile(nsIFile * aSourceFile, nsIFile * aDestParent, const nsAString_internal & aNewName, unsigned int aOptions)  Line 1910	C++
 	xul.dll!nsLocalFile::CopyMove(nsIFile * aParentDir, const nsAString_internal & aNewName, unsigned int aOptions)  Line 2030	C++
 	xul.dll!nsLocalFile::MoveTo(nsIFile * aNewParentDir, const nsAString_internal & aNewName)  Line 2205	C++
 	xul.dll!nsAtomicFileOutputStream::Finish()  Line 944 + 0x11 bytes	C++
 	xul.dll!nsSafeFileOutputStream::Finish()  Line 984 + 0x6 bytes	C++
 	xul.dll!nsCheckSummedOutputStream::Finish()  Line 47 + 0x6 bytes	C++
 	xul.dll!mozilla::safebrowsing::HashStore::WriteFile()  Line 857 + 0x9 bytes	C++
 	xul.dll!mozilla::safebrowsing::Classifier::ApplyTableUpdates(nsTArray<mozilla::safebrowsing::TableUpdate *> * aUpdates, const nsACString_internal & aTable)  Line 679	C++
 	xul.dll!mozilla::safebrowsing::Classifier::ApplyUpdates(nsTArray<mozilla::safebrowsing::TableUpdate *> * aUpdates)  Line 329	C++
 	xul.dll!nsUrlClassifierDBServiceWorker::CacheCompletions(nsTArray<mozilla::safebrowsing::CacheResult> * results)  Line 709	C++
 	xul.dll!UrlClassifierDBServiceWorkerProxy::CacheCompletionsRunnable::Run()  Line 166	C++
 	xul.dll!nsThread::ProcessNextEvent(bool aMayWait, bool * aResult)  Line 766 + 0x8 bytes	C++
 	xul.dll!NS_ProcessNextEvent(nsIThread * aThread, bool aMayWait)  Line 284 + 0x1f bytes	C++
 	xul.dll!mozilla::ipc::MessagePumpForNonMainThreads::Run(base::MessagePump::Delegate * aDelegate)  Line 337 + 0xa bytes	C++
 	xul.dll!MessageLoop::RunHandler()  Line 223	C++
 	xul.dll!MessageLoop::Run()  Line 197	C++
 	xul.dll!nsThread::ThreadFunc(void * aArg)  Line 355	C++
 	nss3.dll!_PR_NativeRunThread(void * arg)  Line 419	C
 	nss3.dll!pr_root(void * arg)  Line 90 + 0xc bytes	C
 	msvcr100.dll!_callthreadstartex()  Line 314 + 0x6 bytes	C
 	msvcr100.dll!_threadstartex(void * ptd)  Line 292 + 0x5 bytes	C
 	kernel32.dll!@BaseThreadInitThunk@12()  + 0x12 bytes	
 	ntdll.dll!__RtlUserThreadStart()  + 0x4e bytes	
 	ntdll.dll!__RtlUserThreadStart@8()  + 0x1c bytes	


After a while (actually just few seconds) it stopped and I could browse normally.  Now I cannot reproduce, but I'm OK to try again.

Comment 1

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

https://crash-stats.mozilla.com/report/index/8c083214-6b5f-4fc0-aa02-676142160210#allthreads

Any ideas gcp?

Comment 2

[Gian-Carlo Pascutto [:gcp]]

This looks like another dupe of bug 1164518.

Francois, that bug is starting to look very serious and urgent. I wonder if we risk having these more now that Google is deploying v4 which is less likely to suffer from these malware vs propular site collisions.

Comment 3

[Gian-Carlo Pascutto [:gcp]]

Duping based on "page takes very long to load" and the stack being in a complete DB update caused by caching a completion.