Message ID | 20191111180957.23443-1-hch@lst.de (mailing list archive) |
---|---|
State | Accepted |
Headers | show |
Series | xfs: remove the unused m_chsize field | expand |
On Mon, Nov 11, 2019 at 07:09:57PM +0100, Christoph Hellwig wrote: > Signed-off-by: Christoph Hellwig <hch@lst.de> Heh, what was that even used for? Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> --D > --- > fs/xfs/xfs_mount.h | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h > index 2dceb446e651..43145a4ab690 100644 > --- a/fs/xfs/xfs_mount.h > +++ b/fs/xfs/xfs_mount.h > @@ -155,7 +155,6 @@ typedef struct xfs_mount { > int m_swidth; /* stripe width */ > uint8_t m_sectbb_log; /* sectlog - BBSHIFT */ > const struct xfs_nameops *m_dirnameops; /* vector of dir name ops */ > - uint m_chsize; /* size of next field */ > atomic_t m_active_trans; /* number trans frozen */ > struct xfs_mru_cache *m_filestream; /* per-mount filestream data */ > struct delayed_work m_reclaim_work; /* background inode reclaim */ > -- > 2.20.1 >
On 11/11/19 7:38 PM, Darrick J. Wong wrote: > On Mon, Nov 11, 2019 at 07:09:57PM +0100, Christoph Hellwig wrote: >> Signed-off-by: Christoph Hellwig <hch@lst.de> > > Heh, what was that even used for? > > Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Looks like it was there in the original CVS import, and got removed in commit da353b0d64e070ae7c5342a0d56ec20ae9ef5cfb Author: David Chinner <dgc@sgi.com> Date: Tue Aug 28 14:00:13 2007 +1000 [XFS] Radix tree based inode caching
On Mon, Nov 11, 2019 at 05:38:10PM -0800, Darrick J. Wong wrote: > On Mon, Nov 11, 2019 at 07:09:57PM +0100, Christoph Hellwig wrote: > > Signed-off-by: Christoph Hellwig <hch@lst.de> > > Heh, what was that even used for? > > - uint m_chsize; /* size of next field */ Inode cluster hash size. We used to keep a linked list of all the inodes in a cluster in a separate structure, so that we could easily iterate them when needed (writeback/cluster freeing). They were kept in a separate hash table so that they could be used as a quick reference inode cluster buffer cache as well (IIRC it was to speed up things like xfs_imap_to_bp()) without having the overhead of a full buffer cache lookup... The need for all this went away with the radix tree indexing and gang lookups - I guess I missed removing this last fragment. Cheers, Dave.
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h index 2dceb446e651..43145a4ab690 100644 --- a/fs/xfs/xfs_mount.h +++ b/fs/xfs/xfs_mount.h @@ -155,7 +155,6 @@ typedef struct xfs_mount { int m_swidth; /* stripe width */ uint8_t m_sectbb_log; /* sectlog - BBSHIFT */ const struct xfs_nameops *m_dirnameops; /* vector of dir name ops */ - uint m_chsize; /* size of next field */ atomic_t m_active_trans; /* number trans frozen */ struct xfs_mru_cache *m_filestream; /* per-mount filestream data */ struct delayed_work m_reclaim_work; /* background inode reclaim */
Signed-off-by: Christoph Hellwig <hch@lst.de> --- fs/xfs/xfs_mount.h | 1 - 1 file changed, 1 deletion(-)