On Thu, Feb 20, 2014 at 05:41:02PM +0800, Jeff Liu wrote:
>
> On 02/20 2014 08:27 AM, Dave Chinner wrote:
> > On Wed, Feb 19, 2014 at 03:56:54PM +0800, Jeff Liu wrote:
> >> Hi Eric,
> >>
> >> I read the previous comments from Dave about using defines for CRC offsets,
> >> and with a grep search after applying this patch, looks there have another
> >> two places maybe we should switch them to the macros as well:
> >>
> >> fs/xfs/xfs_log.c:
> >> Do we need a log record crc offset macros for offsetof(struct xlog_rec_header, h_crc))?
> >>
> >> xfs_dinode.h:
> >> we added the XFS_DINODE_CRC_OFF, just use it at below routine?
> >>
> >> static inline uint xfs_dinode_size(int version)
> >> {
> >> if (version == 3)
> >> return sizeof(struct xfs_dinode);
> >> return offsetof(struct xfs_dinode, di_crc);
> >> }
> >
> > No, that's a different case - it's not being used for determining
> > the offset of a CRC varaible - it's being used to calculate the size
> > of the version 2 inode core. Hence it should remain open coded like
> > because it has a different purpose in life....
>
> Thanks for the clarification, so we don't need that for the second inode
> case, but the first case is used to determine the log record crc offset
> to generate the crc for record header, shouldn't we make it consistent
> with others?
Probably should. Eric, can you send a separate patch that converts
the log crc fileds to use the same convention?
Cheers,
Dave.
--
Dave Chinner
david@xxxxxxxxxxxxx
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
