Re: Defunct processes Clamav
On 10 Jul 2006 at 10:05, Thomas Cort wrote:
> On Mon, 10 Jul 2006 14:26:47 +0100
> "Dermot Paikkos" <dermot@sciencephoto.com> wrote:
>
> > Has anyone else seen this behaviour?
>
> http://sources.redhat.com/bugzilla/show_bug.cgi?id=1297
> http://bugs.gentoo.org/100259
> http://bugs.debian.org/325600
>
> As far as I can tell there are three solutions:
> 1) upgrade to a 2.6 kernel and compile glibc with only nptl (i.e. no
> linuxthreads) 2) use glibc-2.3.2 3) use the work around patch
> attached.
Well I didn't realise I had been using sources that were from
'testing' which is a massive slap on the back of the hand for me as I
only wanted to use stable stuff on this box.
If I stick to the principle of only using stable sources, I can
hardly upgrade to a 2.6 kernel or upgrade to glibc. And to be honest
it's a rather scary thing to do on a production system.
The patch seems incomplete? Was is really meant to end without
closing the comment?
===== 6003_all_alpha-glibc-2.3.5-linuxthreads.patch ======
--- linuxthreads/manager.c
+++ linuxthreads/manager.c
@@ -971,7 +971,7 @@
pid_t pid;
int status;
- while ((pid = waitpid_not_cancel(-1, &status, WNOHANG | __WCLONE)) > 0) {
+ while ((pid = wait3(&status, WNOHANG | __WCLONE, NULL)) > 0 ) {
pthread_exited(pid);
if (WIFSIGNALED(status)) {
/* If a thread died due to a signal, send the same signal to
==============
Is that all there was in the file? And just to make life more
interesting I can't find a manager.c on my system. That is the file
the patch is intended for isn't it?
Thanx.
Dp.
Reply to: