kern/122833: [snapshots] [patch] mountd fails on nmount() after UFS snapshot creation with mount

[Craig Rodrigues]

The following reply was made to PR kern/122833; it has been noted by GNATS.

From: Craig Rodrigues <rodrigc at crodrigues.org>
To: Jaakko Heinonen <jh at saunalahti.fi>
Cc: bug-followup at FreeBSD.org,
	Ga??l Roualland <gael.roualland at dial.oleane.com>,
	leon.kos at lecad.fs.uni-lj.si, rodrigc at FreeBSD.org,
	Yar Tikhiy <yar at comp.chem.msu.su>
Subject: Re: kern/122833: [snapshots] [patch] mountd fails on nmount() after UFS snapshot creation with mount
Date: Wed, 14 May 2008 18:56:40 +0000

 On Wed, May 14, 2008 at 11:18:52AM +0300, Jaakko Heinonen wrote:
 > Following change is a possible quick fix for the bug:
 > 
 > Index: ffs_vfsops.c
 > ===================================================================
 > RCS file: /home/ncvs/src/sys/ufs/ffs/ffs_vfsops.c,v
 > retrieving revision 1.336
 > diff -p -u -r1.336 ffs_vfsops.c
 > --- ffs_vfsops.c	24 Feb 2008 16:38:58 -0000	1.336
 > +++ ffs_vfsops.c	13 May 2008 17:08:20 -0000
 > @@ -183,9 +183,6 @@ ffs_mount(struct mount *mp, struct threa
 >  	if (vfs_getopt(mp->mnt_optnew, "noclusterw", NULL, NULL) == 0)
 >  		mntorflags |= MNT_NOCLUSTERW;
 >  
 > -	if (vfs_getopt(mp->mnt_optnew, "snapshot", NULL, NULL) == 0)
 > -		mntorflags |= MNT_SNAPSHOT;
 > -
 >  	MNT_ILOCK(mp);
 >  	mp->mnt_flag = (mp->mnt_flag | mntorflags) & ~mntandnotflags;
 >  	MNT_IUNLOCK(mp);
 
 
 That fix is wrong.  The better fix would be to come up
 with a function that traverses a vfs_optlist and deletes a string
 option from the list.  Something like:
 
 int vfs_delopt(struct vfsoptlist *optlist, const char *optname);
 
 Then the above code would change to:
 	if (vfs_getopt(mp->mnt_optnew, "snapshot", NULL, NULL) == 0) {
 		mntorflags |= MNT_SNAPSHOT;
 		vfs_delopt(mp->optnew, "snapshot");
 	}
 
 -- 
 Craig Rodrigues
 rodrigc at crodrigues.org