Date: Sat, 29 Jun 1996 14:06:32 +0200 (MET DST) From: Juergen Lock <nox@jelal.hb.north.de> To: nasim!FreeBSD.org!stable Cc: davidg@root.com, jhay@mikom.csir.co.za Subject: Re: lockups. Message-ID: <199606291206.OAA03333@saturn.hb.north.de> In-Reply-To: <no.id> from "nox" at Jun 28, 96 04:08:44 pm
next in thread | previous in thread | raw e-mail | index | archive | help
I wrote:
>
> Update: i'm running vm_glue.c 1.20.4.6 + vm_pageout.c 1.51.4.11 now,
> and performance is at least :) as good again as it was.
> problem solved. (apparently, unless there are new bugs...)
And just when you think... :)
tonights /etc/weekly ended in DDB, vm_pageout_object_deactivate_pages
had been called with a null pointer. i then put this in,
cvs diff: Diffing /home/cvs/src/src/sys/vm
Index: /home/cvs/src/src/sys/vm/vm_pageout.c
===================================================================
RCS file: /home/cvs/cvs/src/sys/vm/vm_pageout.c,v
retrieving revision 1.51.4.11
diff -u -r1.51.4.11 vm_pageout.c
--- vm_pageout.c 1996/06/27 10:29:09 1.51.4.11
+++ vm_pageout.c 1996/06/29 08:29:19
@@ -475,6 +475,13 @@
break;
if ((tmpe->is_sub_map == 0) && (tmpe->is_a_map == 0)) {
obj = tmpe->object.vm_object;
+#if 1
+ if (!obj) {
+#ifdef DIAGNOSTIC
+ printf("vm_pageout_map_deactivate_pages: object.vm_object null\n");
+#endif
+ } else
+#endif
vm_pageout_object_deactivate_pages(map, obj, desired, 0);
}
tmpe = tmpe->next;
and tried to reproduce the problem. no luck until i just restarted
/etc/weekly... there was the message!
the good news is now the system is still up after that so maybe it was
really only the null check that was missing?
cheers
Juergen
PS: now who dares to test this version on something like a real busy
news server? ;)
PPS: my `isp' is down today, i hope mail will only get delayed not lost...
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199606291206.OAA03333>