- 1. [PATCH] Introduce SEEK_DATA/SEEK_HOLE to XFS V2 (score: 1)
- Author: Jeff Liu <jeff.liu@xxxxxxxxxx>
- Date: Tue, 22 Nov 2011 16:19:45 +0800
- Hello, This is the V2 attempt to add SEEK_DATA/SEEK_HOLE to XFS. Changes: == 1. Merge xfs_find_desired_extent() and xfs_seek_data_hole() into xfs_seek_extent(), and place it at xfs_file.c. 2. Using t
- /archives/xfs/2011-11/msg00459.html (14,169 bytes)
- 2. Re: [PATCH] Introduce SEEK_DATA/SEEK_HOLE to XFS V2 (score: 1)
- Author: Jeff Liu <jeff.liu@xxxxxxxxxx>
- Date: Tue, 22 Nov 2011 16:30:53 +0800
- CC to Dave. Thanks, -Jeff
- /archives/xfs/2011-11/msg00460.html (16,929 bytes)
- 3. Re: [PATCH] Introduce SEEK_DATA/SEEK_HOLE to XFS V2 (score: 1)
- Author: Christoph Hellwig <hch@xxxxxxxxxxxxx>
- Date: Wed, 23 Nov 2011 04:40:49 -0500
- Thanks Jeff, this looks pretty good for "simple" implementation, I only have a few rather cosmetic comments. Do you plan on updating Josef's old xfstests support patch for SEEK_DATA/SEEK_HOLE? Also i
- /archives/xfs/2011-11/msg00493.html (14,014 bytes)
- 4. Re: [PATCH] Introduce SEEK_DATA/SEEK_HOLE to XFS V2 (score: 1)
- Author: Jeff Liu <jeff.liu@xxxxxxxxxx>
- Date: Wed, 23 Nov 2011 22:00:28 +0800
- Thanks for your comments. Sure! Additionally, how about if I write two test cases, only to update Josef's patch, another to perform a copy tests. i.e, create a sparse file with dozens of holes, copy
- /archives/xfs/2011-11/msg00495.html (17,366 bytes)
- 5. Re: [PATCH] Introduce SEEK_DATA/SEEK_HOLE to XFS V2 (score: 1)
- Author: Dave Chinner <david@xxxxxxxxxxxxx>
- Date: Thu, 24 Nov 2011 14:23:31 +1100
- If the hole would overflow br_blockcount (32 bits of FSB units, 16TB by default), then we should get multiple consecutive hole records returned. I think that's incorrect. A data extent is limited in
- /archives/xfs/2011-11/msg00515.html (17,385 bytes)
- 6. Re: [PATCH] Introduce SEEK_DATA/SEEK_HOLE to XFS V2 (score: 1)
- Author: Christoph Hellwig <hch@xxxxxxxxxxxxx>
- Date: Thu, 24 Nov 2011 04:02:20 -0500
- Right, the XFS_FILBLKS_MIN in xfs_bmapi_read will limit it, and we'll it the same case again in the loop. So yes, we'll need it; and we should have a test to verify this case. Indeed. Add fragmented
- /archives/xfs/2011-11/msg00516.html (12,611 bytes)
This search system is powered by
Namazu