Date: Tue, 10 Feb 2004 10:56:04 +0100 From: Frode Nordahl <frode@nordahl.net> To: "Andrew P. Lentvorski, Jr." <bsder@allcaps.org> Cc: current@freebsd.org Subject: Re: rpc.lockd(8) seg faults on 5.2-RELEASE (patch, workaround) Message-ID: <56C69560-5BAF-11D8-A6E1-000A95A9A574@nordahl.net> In-Reply-To: <20040205231849.D41237@mail.allcaps.org> References: <0703C4CC-5805-11D8-951F-000A95A9A574@nordahl.net> <20040205231849.D41237@mail.allcaps.org>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
Hello,
This is not a sollution, but I have run with this workaround since
friday without incident.
I think there are two problems:
- freed items are sometimes not removed from the nfslocklist
- first element in blockedlocklist sometimes get wrong, causing a
infinite loop in retry_blockingfilelocklist()
Mvh,
Frode
[-- Attachment #2 --]
diff -ru rpc.lockd/lockd_lock.c rpc.lockd.patched/lockd_lock.c
--- rpc.lockd/lockd_lock.c Sun Oct 26 07:10:44 2003
+++ rpc.lockd.patched/lockd_lock.c Tue Feb 10 10:46:41 2004
@@ -385,7 +385,7 @@
if (newfl == NULL) {
return NULL;
}
- bzero(newfl, sizeof(newfl));
+ bzero(newfl, sizeof(*newfl));
newfl->client.oh.n_bytes = malloc(lockowner->n_len);
if (newfl->client.oh.n_bytes == NULL) {
@@ -438,6 +438,7 @@
void
deallocate_file_lock(struct file_lock *fl)
{
+ LIST_REMOVE(fl, nfslocklist);
free(fl->client.oh.n_bytes);
free(fl->client_cookie.n_bytes);
free(fl);
@@ -1246,6 +1247,12 @@
* current element has been moved
*/
nfl = LIST_NEXT(ifl, nfslocklist);
+
+ if (nfl == ifl) {
+ debuglog("nfl == ifl, resetting nfl\n");
+ nfl = NULL;
+ }
+
debuglog("Iterator choice %p\n",ifl);
debuglog("Prev iterator choice %p\n",pfl);
debuglog("Next iterator choice %p\n",nfl);
@@ -2152,7 +2159,7 @@
static struct nlm4_res retval4;
debuglog("About to send granted on blocked lock\n");
- sleep(1);
+ sleep(debugdelay);
debuglog("Blowing off return send\n");
cli = get_client(fl->addr,
[-- Attachment #3 --]
On Feb 6, 2004, at 08:21, Andrew P. Lentvorski, Jr. wrote:
> On Thu, 5 Feb 2004, Frode Nordahl wrote:
>
>> I also found this in send_granted(): lockd_lock.c:2161
>>
>> debuglog("About to send granted on blocked lock\n");
>> sleep(1);
>> debuglog("Blowing off return send\n");
>>
>> Anyone know what sleep(1) is good for here?
>
> The sleep() statements near debuglog() stuff are to work around the
> fact
> that syslog has bugs where it arbitrarily and randomly eats messages
> when
> you start sending too much data at it too quickly.
>
> By slowing down the logging, all of the messages get recorded.
>
> -a
> _______________________________________________
> freebsd-current@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to
> "freebsd-current-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?56C69560-5BAF-11D8-A6E1-000A95A9A574>