Re: [Nautilus-list] Nautilus News crashes pretty frequently

[Maciej Stachowiak <mjs noisehavoc org>]

On 23Jul2001 01:54PM (-0600), Dan Hensley wrote:
> It crashed on me again just now.  It crashes fairly frequently, but
> unfortunately whenever it does bug-buddy can't come up with a backtrace.
> Has anyone else noticed its instability (it's always been unstable for
> me)?  I guess I can try running nautilus-news in gdb whenever I launch
> Nautilus, but the crashes are pretty unpredictable.
> 

Whoops, I ran under the debugger some more and it looks like there are
still some crashes. 


Received Segmentation fault in LWP 3105 while waiting for SIGSTOP.

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 987076 (LWP 23189)]
0x409adbe1 in __libc_free (mem=0x40c70530) at malloc.c:3033
3033	malloc.c: No such file or directory.
(gdb) bt
#0  0x409adbe1 in __libc_free (mem=0x40c70530) at malloc.c:3033
#1  0x408ab943 in g_free (mem=0x40c00000) at gmem.c:411
#2  0x40ad6f97 in destroy_thread_state (state=0x40c70530)
    at gnome-vfs-thread-pool.c:97
#3  0x40ad7127 in thread_entry (cast_to_state=0x40c70530)
    at gnome-vfs-thread-pool.c:184
#4  0x40165d60 in pthread_start_thread_event (arg=0x473ffc00) at manager.c:274
(gdb) frame 2
#2  0x40ad6f97 in destroy_thread_state (state=0x40c70530)
    at gnome-vfs-thread-pool.c:97
(gdb) p state
$42 = (struct {...} *) 0x40c70530
(gdb) p *state
$43 = {thread_id = 0, waiting_for_work_lock = {__m_reserved = 0, 
    __m_count = 0, __m_owner = 0x0, __m_kind = 0, __m_lock = {__status = 0, 
      __spinlock = 0}}, waiting_for_work_lock_condition = {__c_lock = {
      __status = 0, __spinlock = 0}, __c_waiting = 0x0}, entry_point = 0, 
  entry_data = 0x0, exit_requested = 0}



This kind of smells of a double-free or the like so I'll try leaving
it running overnight under gdb with MALLOC_CHECK_=2 set and see if that
sheds light on the matter.

(This crash seems to take longer to manifest than the other one in the
http code).

 - Maciej