xfs
[Top] [All Lists]

Re: [PATCH] xfs: Initialize all quota inodes to be NULLFSINO

To: Chandra Seetharaman <sekharan@xxxxxxxxxx>
Subject: Re: [PATCH] xfs: Initialize all quota inodes to be NULLFSINO
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Thu, 18 Jul 2013 13:43:21 +1000
Cc: XFS mailing list <xfs@xxxxxxxxxxx>
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <1374097330.8941.12.camel@xxxxxxxxxxxxxxxxxx>
References: <1373593665.20769.10.camel@xxxxxxxxxxxxxxxxxx> <20130713021349.GH3438@dastard> <1373928885.20769.42.camel@xxxxxxxxxxxxxxxxxx> <20130716002352.GA3920@dastard> <1374097330.8941.12.camel@xxxxxxxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Wed, Jul 17, 2013 at 04:42:10PM -0500, Chandra Seetharaman wrote:
> On Tue, 2013-07-16 at 10:23 +1000, Dave Chinner wrote:
> > On Mon, Jul 15, 2013 at 05:54:45PM -0500, Chandra Seetharaman wrote:
> > > On Sat, 2013-07-13 at 12:13 +1000, Dave Chinner wrote:
> > > > On Thu, Jul 11, 2013 at 08:47:45PM -0500, Chandra Seetharaman wrote:
> > > > > 
> > > > > mkfs doesn't initialize the quota inodes to NULLFSINO as it
> > > > > does for the other internal inodes. This leads to check for two
> > > > > values (0 and NULLFSINO) to make sure if a quota inode is
> > > > > valid.
> > > > > 
> > > > > Solve that problem by initializing the values to NULLFSINO
> > > > > if they are 0.
> > > > > 
> > > > > Note that these values are not written back to on-disk
> > > > > superblock unless some quota is enabled on the filesystem.
> > > > > 
> > > > > Signed-off-by: Chandra Seetharaman <sekharan@xxxxxxxxxx>
> > > > > ---
> > > > >  fs/xfs/xfs_mount.c |   12 ++++++++++++
> > > > >  1 files changed, 12 insertions(+), 0 deletions(-)
> > > > > 
> > > > > diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
> > > > > index 2b0ba35..8b95933 100644
> > > > > --- a/fs/xfs/xfs_mount.c
> > > > > +++ b/fs/xfs/xfs_mount.c
> > > > > @@ -572,6 +572,18 @@ out_unwind:
> > > > >  static void
> > > > >  xfs_sb_quota_from_disk(struct xfs_sb *sbp)
> > > > >  {
> > > > > +     /*
> > > > > +      * older mkfs doesn't initialize quota inodes to NULLFSINO,
> > > > > +      * which leads to two values for a quota inode to be invalid:
> > > > > +      * 0 and NULLFSINO. Fix it.
> > > > > +      */
> > > > > +     if (sbp->sb_uquotino == 0)
> > > > > +             sbp->sb_uquotino = NULLFSINO;
> > > > > +     if (sbp->sb_gquotino == 0)
> > > > > +             sbp->sb_gquotino = NULLFSINO;
> > > > > +     if (sbp->sb_pquotino == 0)
> > > > > +             sbp->sb_pquotino = NULLFSINO;
> > > > 
> > > > There coment needs to point out that this is changing the in-memory
> > > > superblock value right here, so we don't need to make changing
> > > > sb_pquotino dependent on the superblock having support for this
> > > > feature.
> > > 
> > > But, we are not just changing just for sb_pquotino, right ?
> > 
> > Sure, but those other two fields are always present in the
> > superblock...
> > 
> > > Since we are changing all the fields independent of superblock
> > > supporting sb_pquotino, I thought the comment should be generic.
> > 
> > So why is it safe to modify this when sb_pquotino isn't valid in the
> > superblock (i.e. v4 superblock) and the only valid on-disk value for
> > an unused section of the superblock is zero?
> > 
> > That's what the comment needs to explain.
> 
> How about this:
> 
> /*
>  * older mkfs doesn't initialize quota inodes to NULLFSINO
>  * This leads to in-core values having two different values
>  * for a quota inode to be invalid: 0 and NULLFSINO.
>  * Change it to a single value NULLFSINO.
>  *
>  * Note that this change affect only the in-core values. 
>  * These values are not written back to disk unless any
>  * quota information is written to the disk. Even in that
>  * case, sb_pquotino field is not written to disk unless
>  * the superblock supports pquotino.
>  */

Nice explanation. Expand it out to 80 columns and it's golden :)

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>