Re: Mem corruption due to race? (Was: [BUG] : crash (perhaps gpg related))

[manu <eallaud yahoo fr>]

Le 09.12.2003 19:39:24, Albrecht Dreß a écrit :
> Am 08.12.03 15:28 schrieb(en) manu:
>> (balsa:2121): Gtk-CRITICAL **: file ../../gtk/gtktextbuffer.c: line  
>> 543 (gtk_text_buffer_emit_insert): assertion `g_utf8_validate (text,  
>> len, NULL)' failed
>> 
>> though I'm not sure it is realted to the crash.
>> I was only able to have this bt which does not look really helpful : 
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x4129e689 in __after_morecore_hook () from /lib/i686/libc.so.6
>> (gdb) bt
>> #0  0x4129e689 in __after_morecore_hook () from /lib/i686/libc.so.6
>> #1  0x40c4f54c in gtk_rc_scanner_new () from /usr/lib/libgtk-x11- 
>> 2.0.so.0
> 
> I also had some crashes, and I also did not send them to bugzilla as  
> they are completely irreproducible. Yesterday I managed to crash  
> balsa by fastly clicking over the new messages in the inbox (mbox  
> file), deleting most of them (I have the "remove immediately" option  
> checked).

OK now I am pretty sure that it is gpg realted, sorry ;-)
You just have to rapidly select a OpenPGP signed message and another  
one back and forth (you just have to be quick enough, hmm I guess  
nicing balsa can help a bit here). Then it will segfault with a short  
trace (always the same but not really useful, it is stuck in mallopt  
after being calld by malloc, meaning it is a memory allocation  
problem); more interesting is the fact that each time the invalid utf8  
message is printed on console. So basically there is something badly  
locked sonewhere that triggers mem corruption. But I still don't know  
the whole story.
Bye
Manu

PGP signature