324325 - 100% CPU utilization exiting thunderbird; never finishes

Status: RESOLVED WORKSFORME

(Thunderbird :: General, defect)

Description

[Brad Cox]

User-Agent:       Mozilla/4.0 (compatible; MSIE 6.0; Mac_PowerPC Mac OS X; en) Opera 8.51
Build Identifier: version 1.5 (20051201)

Thunderbird sometimes hangs during exit, especially after running for several days. Apparently occurs during PREF_UnregisterCallback being slow or stuck in iinfinite loop. See enclosed snapshots.

Reproducible: Sometimes

Steps to Reproduce:
1. Launch Tbird
2. Run for several days
3. Exit

Actual Results:  
See enclosed snapshots. Exit never succeeds after many minutes.

Expected Results:  
Exits

Here's the first two of 12 ActivityMonitor Inspect/Sample outputs. All 12 are similar down to PREF_unregisterCallback and strcmp().

Why is strcmp being called here anyway? Could this be a performance bug, not infinite loop?

rocess 18463 each 10 msecs 100 times


REF_UnregisterCallback

Analysis of sampling pid 18463 every 10.000000 milliseconds
Call graph:
    233 Thread_0f0f
      233 start
        233 start
          233 XRE_main
            233 ScopedXPCOMStartup::~ScopedXPCOMStartup [unified]()
              233 NS_ShutdownXPCOM_P
                233 nsObserverService::NotifyObservers(nsISupports*, char const*, unsigned short const*)
                  233 nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)
                    233 nsPrefBranch::freeObserverList()
                      229 PREF_UnregisterCallback
                        116 dyld_stub_strcmp
                          116 dyld_stub_strcmp
                        109 strcmp
                          109 strcmp
                        4 PREF_UnregisterCallback
                      4 free
                        4 free
    233 Thread_1003
      233 _pthread_body
        233 PR_Select
          233 nsThread::Main(void*)
            233 TimerThread::Run()
              233 PR_WaitCondVar
                233 PR_Unlock
                  233 pthread_cond_timedwait
                    233 semaphore_timedwait_signal_trap
                      233 semaphore_timedwait_signal_trap

Total number in stack (recursive counted multiple, when >=5):

Sort by top of stack, same collapsed (when >= 5):
        semaphore_timedwait_signal_trap        233
        dyld_stub_strcmp        116
        strcmp        109
Sample analysis of process 18463 written to file /dev/stdout
Sampling process 18463 each 10 msecs 300 times

Analysis of sampling pid 18463 every 10.000000 milliseconds
Call graph:
    85 Thread_0f0f
      85 start
        85 start
          85 XRE_main
            85 ScopedXPCOMStartup::~ScopedXPCOMStartup [unified]()
              85 NS_ShutdownXPCOM_P
                85 nsObserverService::NotifyObservers(nsISupports*, char const*, unsigned short const*)
                  85 nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)
                    85 nsPrefBranch::freeObserverList()
                      84 PREF_UnregisterCallback
                        49 strcmp
                          49 strcmp
                        34 dyld_stub_strcmp
                          34 dyld_stub_strcmp
                        1 PREF_UnregisterCallback
                      1 free
                        1 free
    85 Thread_1003
      85 _pthread_body
        85 PR_Select
          85 nsThread::Main(void*)
            85 TimerThread::Run()
              85 PR_WaitCondVar
                85 PR_Unlock
                  85 pthread_cond_timedwait
                    85 semaphore_timedwait_signal_trap
                      85 semaphore_timedwait_signal_trap

Total number in stack (recursive counted multiple, when >=5):

Sort by top of stack, same collapsed (when >= 5):
        semaphore_timedwait_signal_trap        85
        strcmp        49
        dyld_stub_strcmp        34
Sample analysis of process 18463 written to file /dev/stdout
Sampling process 18463 each 10 msecs 100 times

Comment 1

[Wayne Mery (:wsmwk)]

Brad

is this gone in version 1.5.0.4?  http://www.mozilla.com/thunderbird/releases/1.5.0.4.html

if it still happens does can you test with a trunk build where it may well be fixed? http://ftp.mozilla.org/pub/mozilla.org/thunderbird/nightly/latest-trunk/

other hang bugs - Bug 326629, Bug 324196, Bug 340647

Comment 2

[Brad Cox]

I've not experienced this bug in the last 2-3 versions. Maybe its gone. Possibly my usage pattern changed; I no longer leave email running throughout the day but shut it down after checking.

Comment 3

[Brad Cox]

None.

Comment 4

[Rotis]

I want to notice that this bug is still alive and is obviously a duplicate of Bug 322343. If you look at Bug 322343 this is not a Mac problem only.

When I run Thunderbird for a few days (POP-Mail, RSS) and quit it, it closes the main windows but the process stays open for several minutes and takes 100 % CPU.

My system: TB 1.5.0.10 on MacOSX 10.4.9 on iBook PPC G4 with 1.42 GHz and 1.5 GB RAM

Comment 5

[David :Bienvenu]

Have you tried a 2.0 build? The 2.0 nightly build up there is what we plan on shipping...

ftp://ftp.mozilla.org/pub/mozilla.org/thunderbird/nightly/latest-mozilla1.8