[PATCH] Use the incore inode size in xfs_file_readdir()

[PATCH] Use the incore inode size in xfs_file_readdir()

[Lachlan McIlroy]

We should be using the incore inode size here not the linux inode
size.  The incore inode size is always up to date for directories
whereas the linux inode size is not updated for directories.

We've hit assertions in xfs_bmap() and traced it back to the linux
inode size being zero here but the incore size being correct.

--- xfs-fix.orig/fs/xfs/linux-2.6/xfs_file.c
+++ xfs-fix/fs/xfs/linux-2.6/xfs_file.c
@@ -254,7 +254,7 @@ xfs_file_readdir(
  	 * point we can change the ->readdir prototype to include the
  	 * buffer size.
  	 */
-	bufsize = (size_t)min_t(loff_t, PAGE_SIZE, inode->i_size);
+	bufsize = (size_t)min_t(loff_t, PAGE_SIZE, ip->i_d.di_size);

  	error = xfs_readdir(ip, dirent, bufsize,
  				(xfs_off_t *)&filp->f_pos, filldir);

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH] Use the incore inode size in xfs_file_readdir()

[Christoph Hellwig]

On Thu, Dec 04, 2008 at 05:59:08PM +1100, Lachlan McIlroy wrote:
> We should be using the incore inode size here not the linux inode
> size.  The incore inode size is always up to date for directories
> whereas the linux inode size is not updated for directories.
> 
> We've hit assertions in xfs_bmap() and traced it back to the linux
> inode size being zero here but the incore size being correct.

Heh. Looks good, but you can still call ->readdir with a 0 inode size,
so you might want to check for that (actualyl I think other pathes
are goign to take care of it, but..)

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH] Use the incore inode size in xfs_file_readdir()

[Lachlan McIlroy]

Christoph Hellwig wrote:
> On Thu, Dec 04, 2008 at 05:59:08PM +1100, Lachlan McIlroy wrote:
>> We should be using the incore inode size here not the linux inode
>> size.  The incore inode size is always up to date for directories
>> whereas the linux inode size is not updated for directories.
>>
>> We've hit assertions in xfs_bmap() and traced it back to the linux
>> inode size being zero here but the incore size being correct.
> 
> Heh. Looks good, but you can still call ->readdir with a 0 inode size,
> so you might want to check for that (actualyl I think other pathes
> are goign to take care of it, but..)

What if we remove bufsize and pass PAGE_SIZE into xfs_readdir()?
No need to worry about a 0 inode size.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH] Use the incore inode size in xfs_file_readdir()

[Christoph Hellwig]

On Fri, Dec 05, 2008 at 02:22:55PM +1100, Lachlan McIlroy wrote:
> Christoph Hellwig wrote:
> > On Thu, Dec 04, 2008 at 05:59:08PM +1100, Lachlan McIlroy wrote:
> >> We should be using the incore inode size here not the linux inode
> >> size.  The incore inode size is always up to date for directories
> >> whereas the linux inode size is not updated for directories.
> >>
> >> We've hit assertions in xfs_bmap() and traced it back to the linux
> >> inode size being zero here but the incore size being correct.
> > 
> > Heh. Looks good, but you can still call ->readdir with a 0 inode size,
> > so you might want to check for that (actualyl I think other pathes
> > are goign to take care of it, but..)
> 
> What if we remove bufsize and pass PAGE_SIZE into xfs_readdir()?
> No need to worry about a 0 inode size.

It is used for sizing a memory allocation in xfs_dir2_leaf_getdents, and
avoiding over-allocations for small directories seems like a good idea.

So please go ahead with your original patch.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs