[libvirt] [PATCH 2/3] fdstream: avoid child process leak on error

Eric Blake eblake at redhat.com
Thu Jul 14 17:48:21 UTC 2011

On 07/14/2011 11:42 AM, Matthias Bolte wrote:
> 2011/7/12 Eric Blake <eblake at redhat.com>:
>> By requesting the pid in virCommandRunAsync, fdstream was claiming
>> that it would manually wait for the process.  But on the failure
>> path, the child process was being leaked.
> This difference in behavior between virCommandRunAsync(..., NULL) and
> virCommandRunAsync(..., &pid) is not documented on virCommandRunAsync
> itself but only on virCommandFree.

That was the case in libvirt.git right now, but I thought I addressed
that properly in patch 1/3.

> Also virCommandFree refers to
> waitpid instead of you're newly added virPidWait.

Oh, good point.  I'll tweak patch 1 to fix that before I push.

> Finally I noticed
> that the comments in command.h and command.c are out of sync (at least
> for virCommandFree). Why are you documenting in two places making it
> prone to get out of sync?

Which file is better, the .c or the .h?  I'm fine with dropping docs
from one of the two locations, if we have a preference on which file is
more likely to be referenced.

PRO - the documentation is next to the implementation, and hopefully
easier to keep the two in sync
PRO - matches how we do things in libvirt.h vs. libvirt.c
CON - much larger file to search through when you are looking up the
PRO - smaller file
CON - further away from implementation, could go stale

so I'm leaning towards .c only.

> ACK, to this patch.

Eric Blake   eblake at redhat.com    +1-801-349-2682
Libvirt virtualization library http://libvirt.org

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 619 bytes
Desc: OpenPGP digital signature
URL: <http://listman.redhat.com/archives/libvir-list/attachments/20110714/40c2dfa6/attachment-0001.sig>

More information about the libvir-list mailing list