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

List:       sylpheed
Subject:    [sylpheed:15652] Re: gpg zombie
From:       Sergey Vlasov <vsu () altlinux ! ru>
Date:       2002-08-28 16:51:58
[Download RAW message or body]


On Wed, 28 Aug 2002 11:14:15 +0900
Hiroyuki Yamamoto <hiro-y@kcn.ne.jp> wrote:

> Hello,
> 
> On Tue, 27 Aug 2002 19:23:36 +0200
> Werner Koch <wk@gnupg.org> wrote:
> 
> > On Mon, 15 Jul 2002 23:14:33 +0200, Jens Oberender said:
> > 
> > > That's mostly Werner Koch's stuff. He is the expert, he coded it.
> > > Perhaps he comments on that.
> > > I have only one gpg defunc.
> > 
> > There should never be a zombie process for a long time of course.  It
> > should get cleaned up with the gpgme_release().  If you are using the
> > latest gpgme version and it still hapens, please report to bug-gpgme
> > at gnupg org.
> 
> It seems that two gpg zombies are always left after signature
> verification. They seem to increase for each verification.
> rfc2015.c::check_signature() calls gpgme_release() at the end, so I
> don't know why it happens.
> I'm using the official package of debian unstable (0.3.8-1).

Same thing here with 0.3.9.

It seems that child reaping is only done in gpgme_wait(), when the idle
function is called. gpgme_op_verify() uses _gpgme_wait_one(), which does
not call the idle function (BTW, it is also a problem for Sylpheed - the
GUI will be blocked completely while waiting for gpg).

To Hiro: Also, even if the idle function would be called, if gpg takes a
long time to complete (e.g. when it is fetching a key from a slow
keyserver), the idle function is called only once per second - too slow
for the GUI. Looks like we should move Sylpheed to the newer gpgme
interface for I/O callbacks, available since version 0.3.8.

To Werner Koch: Seems that some changes are needed to perform child
reaping - like a cleanup function, which should be usable in the I/O
callback mode. Or maybe adding more functions to the callback interface
to manage the subprocesses.

[Attachment #3 (application/pgp-signature)]

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