Icecast dev - May 2010 - Memory leak on Icecast 2.3.2 / Debian ?

On 09/12/2009 16:59, Romain Beauxis wrote:
> 	Hi !
> 
> Le mardi 8 d?cembre 2009 19:59:23, Karl Heyes a ?crit :
>>> But since you asked, attached are the missing frees for
icecast2_2.3.2-4
>>> in a totally untested drive-by-patching manner. Like I said the
-kh17 I
>>> also happened to have did not have this specific code at all and
the
>>> xmlEncodeEntitiesReentrant() calls were all fine.
>> The patch is fine, both kh and the trunk code use different code so 
>> those should not have this problem but feel free to report any issues 
>> like this.
> 
> Ok, thanks !
> 
> Giles, could you confirm that this patch fixes your memory leak ?
> 
> 
> Romain

Hi,

it has been running for 3 days, and hit hard for stress test, it seems 
fine now with this patch. I guess it would be good for debian to 
backport that indeed (and I'm surprised nobody was impacted by this in a 
noticeable enough way so as to report it).

As I'm now facing a required recompile/upgrade anyway, I'm wondering 
whether I should just patch this, or go for the trunk code.. How 
reliable is it?

Cheers, and thanks a lot to you guys
Gilou

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256


re all,

this patch has  not been integrated in Debian  yet, hence the Icecast2
package still segfaults after filling up all the memory with its leak.

as this  is a  critical bug,  while it seems  that the  current Debian
mantainer for the icecast2 package  has abandoned its duty, i'm kindly
asking the attention of debian  developers included in this mail, will
provide them with more details in private.

ciao

On Sat, Dec 12, 2009 at 04:26:04AM +0100, Gilles PIETRI
wrote:
> On 09/12/2009 16:59, Romain Beauxis wrote:
> > 	Hi !
> >
> > Le mardi 8 d?cembre 2009 19:59:23, Karl Heyes a ?crit :
> >>> But since you asked, attached are the missing frees for
icecast2_2.3.2-4
> >>> in a totally untested drive-by-patching manner. Like I said
the -kh17 I
> >>> also happened to have did not have this specific code at all
and the
> >>> xmlEncodeEntitiesReentrant() calls were all fine.
> >> The patch is fine, both kh and the trunk code use different code
so
> >> those should not have this problem but feel free to report any
issues
> >> like this.
> >
> > Ok, thanks !
> >
> > Giles, could you confirm that this patch fixes your memory leak ?
> >
> >
> > Romain
>
>
> Hi,
>
> it has been running for 3 days, and hit hard for stress test, it seems
> fine now with this patch. I guess it would be good for debian to
> backport that indeed (and I'm surprised nobody was impacted by this in
a
> noticeable enough way so as to report it).
>
> As I'm now facing a required recompile/upgrade anyway, I'm
wondering
> whether I should just patch this, or go for the trunk code.. How
> reliable is it?
>
> Cheers, and thanks a lot to you guys
> Gilou
> _______________________________________________
> Icecast-dev mailing list
> Icecast-dev at xiph.org
> http://lists.xiph.org/mailman/listinfo/icecast-dev
>
- --
jaromil, dyne.org developer, http://jaromil.dyne.org

GPG: B2D9 9376 BFB2 60B7 601F  5B62 F6D3 FBD9 C2B6 8E39
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
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=0GlH
-----END PGP SIGNATURE-----

Le jeudi 13 mai 2010 05:46:12, jaromil a ?crit :
>   re all,
	Hi,
 
> this patch has  not been integrated in Debian  yet, hence the Icecast2
> package still segfaults after filling up all the memory with its leak.
> 
> as this  is a  critical bug,  while it seems  that the  current Debian
> mantainer for the icecast2 package  has abandoned its duty, i'm kindly
> asking the attention of debian  developers included in this mail, will
> provide them with more details in private.
icecast2 (2.3.2-5) unstable; urgency=high

   * Added patch to fix a memory leak.
     Thanks to Gilles Pietri for reporting
     and Jussi Kukkonen for providing a patch.
   * Bumped standards version to 3.8.3
   * Added doc-base registration.
 -- Romain Beauxis <toots at rastageeks.org>  Mon, 14 Dec 2009 10:05:47
-0600

A memory leak is not a security issue so the package in stable (lenny) has not 
been modified. However, a backport has been prepared, see:
  http://www.backports.org/
  http://www.backports.org/debian/pool/main/i/icecast2/

I do not exactly call that abandoning its duty ;-)


Romain