[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: libc bug Question
From:       Michael Pyne <mpyne () purinchu ! net>
Date:       2010-01-08 3:29:05
Message-ID: 201001072229.11231.mpyne () purinchu ! net
[Download RAW message or body]

[Attachment #2 (multipart/signed)]


On Thursday 07 January 2010 18:13:02 Andreas Pakulat wrote:
> On 08.01.10 09:06:21, Lindsay Mathieson wrote:
> > I'm working/testing with self compiled KDE trunk on Karmic (backports and
> > all updates) - not sure if the libc bug is fixed on that.
> >
> >
> > What's the story with the MALLOC_CHECK_ var? I have it set to "1", does
> > this workaround the problem? and is it a runtime thing or do I have to
> > build KDE with it set?
> 
> Its a runtime thing, but you have to make sure its set before the
> crashing apps are started (this mostly hits multi-threaded apps,
> single-threaded ones are often ok). And better than setting it to 1 is
> setting it to 0, as the actual printing of the memory-warning causes the
> race-condition (as far as I understood).

Printing out the error message may make it easier to trip across the bug but 
I've also reproduced SIGABRT even when the malloc check was enabled but output 
was disabled.

In other words, the best course of action is to simply "unset MALLOC_CHECK_" 
in your shell

Regards,
 - Michael Pyne

["signature.asc" (application/pgp-signature)]

>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<


[prev in list] [next in list] [prev in thread] [next in thread]