- 1. emove and xfs_rmdir (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Thu, 10 Apr 2008 20:44:13 +0200
- VFS always has an inode reference when we call these functions. So we only need to grab a signle reference to each inode that's joined to a transaction - all the other bumping and dropping is as
- /archives/xfs/2008-04/msg00236.html (13,506 bytes)
- 2. er_process_iunlinks (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 10:24:43 +1000
- ode. This seems rather strange to
- /archives/xfs/2008-04/msg00239.html (9,881 bytes)
- 3. aligned filesystems (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 14:40:09 +1000
- s has extra bits set in it xfs_bmap output when the filesyste is using stripe alignment. Fix the test to only look at the unwritten flag. Capture the xfs_bmap output as well so failures can be de
- /archives/xfs/2008-04/msg00245.html (11,116 bytes)
- 4. d HAVE_SPLICE macro (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Fri, 11 Apr 2008 09:37:30 +0200
- l more 2.4 build remnants to get rid of... HAVE_SPLICE was part of the infrastructure for building 2.4 and 2.6 kernels out of the same tree. Now we don't build 2.4 kernels this way anymore its un
- /archives/xfs/2008-04/msg00247.html (9,010 bytes)
- 5. al log and barriers (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 18:40:53 +1000
- to mention an interesting thing I've tested : using a fast SSD (MTRON 7000) to hold the filesystem log for a big RAID array gives an high boost for small IO on the filesystem, somewhat similar t
- /archives/xfs/2008-04/msg00251.html (9,726 bytes)
- 6. xfs_lock_for_rename (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Fri, 11 Apr 2008 10:46:46 +0200
- ri, 11 Apr 2008 17:37:30 +1000, Christoph Hellwig <hch@xxxxxx> wrote: It seems that do_rename() in namei.c calls lookup_hash/__lookup_hash which does an inode->i_op->lookup() to get the dentry. T
- /archives/xfs/2008-04/msg00252.html (8,597 bytes)
- 7. 2/2] kill usesless IHOLD calls in xfs_remove and xfs_rmdir (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Thu, 10 Apr 2008 20:44:13 +0200
- fs-kill-xfs_dir_lookup_int here, which gets rid of a lock roundtrip, and lots of stack space. Also kill the di_mode == 0 check that has been done in xfs_iget for a few years now. Signed-off
- /archives/xfs/2008-04/msg00832.html (13,506 bytes)
- 8. xfs_remove (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 10:24:43 +1000
- r bumping and dropping is as
- /archives/xfs/2008-04/msg00835.html (9,881 bytes)
- 9. ch] xfsqa - 166 - fix filter for stripe aligned filesystems (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 14:40:09 +1000
- , back in the days of dir1 (1994) when xfs_dir_removename() was expanded to take a block count, only xfs_rename() passed a value in (MAX_EXT_NEEDED). Both xfs_rmdir and xfs_remove passed z
- /archives/xfs/2008-04/msg00841.html (11,116 bytes)
- 10. fs_lock_for_rename (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Fri, 11 Apr 2008 09:37:30 +0200
- extra bits set in it xfs_bmap output when the filesyste is using stripe alignment. Fix the test to only look at the unwritten flag. Capture the xfs_bmap output as well so failures can be de
- /archives/xfs/2008-04/msg00843.html (9,010 bytes)
- 11. ve unused HAVE_SPLICE macro (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 18:40:53 +1000
- s un
- /archives/xfs/2008-04/msg00847.html (9,726 bytes)
- 12. g and barriers (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Fri, 11 Apr 2008 10:46:46 +0200
- th
- /archives/xfs/2008-04/msg00848.html (8,597 bytes)
- 13. [PATCH 2/2] simplify xfs_lock_for_rename (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Thu, 10 Apr 2008 20:44:13 +0200
- as with lookup merge what's left of xfs_dir_lookup_int into it, kill a useless lock roundtrip left over from IRIX and kill a bunch of useless variables. Signed-off-by: Christoph Hellwig <hch@xxxxxx>
- /archives/xfs/2008-04/msg01428.html (13,506 bytes)
- 14. Re: [PATCH 2/2] simplify xfs_lock_for_rename (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 10:24:43 +1000
- as with lookup merge what's left of xfs_dir_lookup_int into it, kill a useless lock roundtrip left over from IRIX and kill a bunch of useless variables. Signed-off-by: Christoph Hellwig <hch@xxxxxx>
- /archives/xfs/2008-04/msg01431.html (10,009 bytes)
- 15. Re: [PATCH 2/2] simplify xfs_lock_for_rename (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 14:40:09 +1000
- as with lookup merge what's left of xfs_dir_lookup_int into it, kill a useless lock roundtrip left over from IRIX and kill a bunch of useless variables. Signed-off-by: Christoph Hellwig <hch@xxxxxx>
- /archives/xfs/2008-04/msg01437.html (11,164 bytes)
- 16. Re: [PATCH 2/2] simplify xfs_lock_for_rename (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Fri, 11 Apr 2008 09:37:30 +0200
- You're right. Back to the drawing board to make rename even simpler. None of the other patches I sent yesterday depends on this one so they can go in independently. And I'd especially like to see th
- /archives/xfs/2008-04/msg01439.html (9,089 bytes)
- 17. Re: [PATCH 2/2] simplify xfs_lock_for_rename (score: 1)
- Author: "Barry Naujok" <bnaujok@xxxxxxx>
- Date: Fri, 11 Apr 2008 18:40:53 +1000
- It seems that do_rename() in namei.c calls lookup_hash/__lookup_hash which does an inode->i_op->lookup() to get the dentry. This will do the xfs_lookup anyway, taking a refernence if the target exis
- /archives/xfs/2008-04/msg01443.html (9,718 bytes)
- 18. Re: [PATCH 2/2] simplify xfs_lock_for_rename (score: 1)
- Author: Christoph Hellwig <hch@xxxxxx>
- Date: Fri, 11 Apr 2008 10:46:46 +0200
- I haveåa new rename patch ready aswell, it's half-way through xfsqa so far.
- /archives/xfs/2008-04/msg01444.html (8,730 bytes)
This search system is powered by
Namazu