[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[leafnode-list] Re: Fetchnews - glibc detected *** double free or corruption - aborted

Stroller <linux.luser@xxxxxxxxxxxxx> writes:

> fetchnews: 0 articles and 0 headers fetched, 0 killed, 0 posted, in  
> 18 seconds
> # echo $_MALLOC_CHECK
> 2
> # echo $MALLOC_CHECK_
> 2

With most POSIX-like shells (such as bash, sh, ksh, pdksh), you need to
either "export" (in your shell, literally) or "setenv" these variables.

Thus, you would need

     export MALLOC_CHECK_

unless your valgrind run below showed that the bug is indeed fixed.

To complete this, in order to un-export the variable, "unset" will work
in many shells:

     unset MALLOC_CHECK_

> So I recompiled manually with:
>   # cd leafnode-2.0.0.alpha20060706a && CFLAGS="-ggdb3 -O1" make &&  
> make install && valgrind --leak-check=yes --show-reachable=yes --num- 
> callers=16 --quiet fetchnews -vvv 2>&1 | tee fetchnews.log
> On the previous occasion I started seeing ==22628== output almost  
> immediately; this time, not until the run was completed. I will send  
> a copy of the output (331 lines) direct to Matthias but interested  
> parties can also get a copy from <http://linux.stroller.uk.eu.org/ 
> fetchnews.log.gz>.

Well, I'll see if I can fix the "definitely lost".

The "still reachable" are not a major reason for concern for programs
that run a few seconds to minutes and then quit.

About those that contain getpwnam() and other glibc stuff, these may be
leaks inside glibc that fetchnews can perhaps not work around.

> Well, it seems to be fixed.  :D
> Thanks very much for your help.

Glad it works. You're welcome.

Happy usenetting,

Matthias Andree
leafnode-list mailing list