Date: Thu, 21 Feb 2002 11:08:30 -0800 (PST) From: Matthew Dillon <dillon@apollo.backplane.com> To: Kirk McKusick <mckusick@mckusick.com> Cc: Mike Silbersack <silby@silby.com>, Valentin Nechayev <netch@iv.nn.kiev.ua>, "David W. Chapman Jr." <dwcjr@inethouston.net>, stable@FreeBSD.ORG Subject: Re: Softupdates failure during buffer syncing at shutdown (was Re: cvs commit: src/sys/ufs/ffs ffs_softdep.c) Message-ID: <200202211908.g1LJ8UK98232@apollo.backplane.com> References: <200202211243.g1LChpi21455@beastie.mckusick.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Great! Thanks for tracking this down for us, Kirk, it would have
taken me a lot longer.
I don't want to mess around with the sysdaemon shutdown code so what
I have done is incorporated the io_sync call into the shutdown code.
Considering the potential complexity in softupdates I wait until
normal flushing is as done as it is going to get (just in case
softupdate's sync crashes or panics) and then I call bioops.io_sync in
the remaining loop iterations.
I have tested it and it works like a charm! I now get:
syncing disks... 110 21 16 14 13 12 11 11 11 11 11 11 11
done
And, bang, on the 6th iteration after it can't sync any more it calls
io_sync and its happy.
This will be going into -current today and -stable in three days.
Index: kern/kern_shutdown.c
===================================================================
RCS file: /home/ncvs/src/sys/kern/kern_shutdown.c,v
retrieving revision 1.72.2.11
diff -u -r1.72.2.11 kern_shutdown.c
--- kern/kern_shutdown.c 4 Feb 2002 13:09:24 -0000 1.72.2.11
+++ kern/kern_shutdown.c 21 Feb 2002 18:52:03 -0000
@@ -257,6 +257,8 @@
if (nbusy < pbusy)
iter = 0;
pbusy = nbusy;
+ if (iter > 5 && bioops.io_sync)
+ (*bioops.io_sync)(NULL);
sync(&proc0, NULL);
DELAY(50000 * iter);
}
-Matt
Matthew Dillon
<dillon@backplane.com>
:I have had an opportunity to log into test1.backplane.com to look
:at the problem (thanks to the wireless network at the conference
:here in Helsinki :-) The buffers that are being held locked
:contain the indirect block pointers for files that are to be
:...
:syncer daemon has been stopped. They will be used and freed
:if the syncer daemon is permitted to run. I am not sure what
:...
: /*
: * Do soft update processing.
: */
: if (bioops.io_sync)
: (*bioops.io_sync)(NULL);
:
:This call needs to be done after all the other buffers have been
:written as the writes of some of the other buffers cause the
:deletes related to these buffers to be put on the soft updates
:...
: Kirk McKusick
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200202211908.g1LJ8UK98232>