[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Yet another profiler crash [Was: Re: Issues for 24.3]
From: |
YAMAMOTO Mitsuharu |
Subject: |
Re: Yet another profiler crash [Was: Re: Issues for 24.3] |
Date: |
Mon, 28 Jan 2013 17:38:58 +0900 |
User-agent: |
Wanderlust/2.14.0 (Africa) SEMI/1.14.6 (Maruoka) FLIM/1.14.8 (Shijō) APEL/10.6 Emacs/22.3 (sparc-sun-solaris2.8) MULE/5.0 (SAKAKI) |
>>>>> On Mon, 28 Jan 2013 09:37:14 +0400, Dmitry Antipov <address@hidden> said:
>>> Crash related to new profiler?
>>> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=13545
>>
>> A recipe was requested, but no response.
> This crash is likely caused by C stack overflow and may be really
> hard to reproduce.
Can we make sure that keys (backtraces?) in the hash tables used in
the profiler are always acyclic? Otherwise, using the `equal' test
for such keys are wrong and may lead to stack overflow, because an
error signal triggers another hash table operation in the profiler.
I noticed that the emacs-24 branch used the `equal' test, but the
trunk used a specialized function (cmpfn_profiler) for comparing keys.
Probably the issue is fixed in the trunk.
YAMAMOTO Mitsuharu
address@hidden