[Ocfs2-devel] [PATCH 23/56] ocfs2: Implementation of local and global quota file handling

[Andrew Morton]

On Wed, 24 Dec 2008 16:29:23 -0800 Mark Fasheh <mfasheh at suse.com> wrote:

> > 
> > put_bh() is more efficient and modern, in the case where bh is known to
> > not be NULL.
> 
> How about __brelse()? Won't we lose the ref counting check if we go straight
> to put_bh()?
> 

That would work, if you value the debug check.

>
> ...
>
> > > +/* Write to quotafile (we know the transaction is already started and has
> > > + * enough credits) */
> > > +ssize_t ocfs2_quota_write(struct super_block *sb, int type,
> > > +			  const char *data, size_t len, loff_t off)
> > > +{
> > > +	struct mem_dqinfo *info = sb_dqinfo(sb, type);
> > > +	struct ocfs2_mem_dqinfo *oinfo = info->dqi_priv;
> > > +	struct inode *gqinode = oinfo->dqi_gqinode;
> > > +	int offset = off & (sb->s_blocksize - 1);
> > > +	sector_t blk = off >> sb->s_blocksize_bits;
> > 
> > does ocfs2 attempt to support CONFIG_LBD=n?
> 
> It should... What's the problem here?

Idle curiosity.  I noticed that the above expression could result in
truncation when writing a 64-bit value into a 32-bit one, which makes
one wonder whether this all works and is tested, etc.