From owner-freebsd-current  Mon Nov 29 15:25:49 1999
Delivered-To: freebsd-current@freebsd.org
Received: from apollo.backplane.com (apollo.backplane.com [216.240.41.2])
	by hub.freebsd.org (Postfix) with ESMTP
	id 973C015631; Mon, 29 Nov 1999 15:25:22 -0800 (PST)
	(envelope-from dillon@apollo.backplane.com)
Received: (from dillon@localhost)
	by apollo.backplane.com (8.9.3/8.9.1) id PAA12356;
	Mon, 29 Nov 1999 15:25:21 -0800 (PST)
	(envelope-from dillon)
Date: Mon, 29 Nov 1999 15:25:21 -0800 (PST)
From: Matthew Dillon <dillon@apollo.backplane.com>
Message-Id: <199911292325.PAA12356@apollo.backplane.com>
To: Eivind Eklund <eivind@FreeBSD.org>
Cc: Julian Elischer <julian@whistle.com>,
	"Viren R.Shah" <viren@rstcorp.com>, Greg Lehey <grog@lemis.com>,
	freebsd-current@FreeBSD.org
Subject: Re: repeatable crash in -current (softupdates, NFS)
References: <14399.63511.296802.242618@jabberwock.rstcorp.com> <19991127191729.A53832@bitbox.follo.net> <14402.62122.461010.454021@jabberwock.rstcorp.com> <199911292152.NAA09656@apollo.backplane.com> <19991129235631.P60031@bitbox.follo.net> <199911292308.PAA12218@apollo.backplane.com> <19991130001724.S60031@bitbox.follo.net>
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG


:I think I (well, Alfred Perlstein) have found what the problem is - in
:nfs_symlink, newvp isn't initialized for NFSv2.  Unfortunately, I have
:zero clue about how to fix that - Alfred believes the checks for NFSv3
:may not be necessary - myself, I find the NFS code almost totally
:incomprehensible, and have tried to keep my fingers as much out of it
:as possible, but for the changes I'm working on now I have to touch it
::-(
:
:Eivind.

    Yes, I concur.  There are also problems with the ASSERT_VOP_*() macros...
    the filesystem code is not very good at NULLing out dead fields in namei()
    requests (it has caused me no end of trouble), so you can't assume
    that a non-null pointer is valid in the ASSERT's.  You *must* check the
    error return.  

    But that is not what caused the bug.  Alfred has it tagged.


        if (v3) {
                if (!error)
                        nfsm_mtofh(dvp, newvp, v3, gotvp);
                nfsm_wcc_data(dvp, wccflag);
        }

    nfsm_wcc_data() is an NFSv3 only mechanism, I believe.  But the 
    if (!error) nfsm_mtofh(...) can be moved to outside (before) that
    conditional.  I'll patch it in and test it.  Also, the nfsm macros
    are dangerous.

    I've added a little cleanup to this patch.  Viren, please try this 
    patch.

					-Matt
					Matthew Dillon 
					<dillon@backplane.com>

Index: nfs_vnops.c
===================================================================
RCS file: /FreeBSD/FreeBSD-CVS/src/sys/nfs/nfs_vnops.c,v
retrieving revision 1.146
diff -u -r1.146 nfs_vnops.c
--- nfs_vnops.c	1999/11/27 18:14:41	1.146
+++ nfs_vnops.c	1999/11/29 23:23:05
@@ -1806,11 +1806,10 @@
 		txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime);
 	}
 	nfsm_request(dvp, NFSPROC_SYMLINK, cnp->cn_proc, cnp->cn_cred);
-	if (v3) {
-		if (!error)
-			nfsm_mtofh(dvp, newvp, v3, gotvp);
+	if (!error)
+		nfsm_mtofh(dvp, newvp, v3, gotvp);
+	if (v3)
 		nfsm_wcc_data(dvp, wccflag);
-	}
 	nfsm_reqdone;
 	/*
 	 * Kludge: Map EEXIST => 0 assuming that it is a reply to a retry.
@@ -1821,8 +1820,9 @@
 	if (error) {
 		if (newvp)
 			vput(newvp);
-	} else
+	} else {
 		*ap->a_vpp = newvp;
+	}
 	VTONFS(dvp)->n_flag |= NMODIFIED;
 	if (!wccflag)
 		VTONFS(dvp)->n_attrstamp = 0;


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message