http://oss.sgi.com/bugzilla/show_bug.cgi?id=728
------- Additional Comments From sandeen-xfs@xxxxxxxxxxx 2006-11-25 12:08 CST
-------
Ok, running commentary. rtc's new patch on the RH bug seems to show a genuine
problem. It is writing staggered blocks of "1's" but when read back we get
zeros. Further, if we bmap the file:
EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL FLAGS
0: [0..255]: 192..447 0 (192..447) 256 10000
1: [256..10239]: hole 9984
2: [10240..10495]: 10432..10687 0 (10432..10687) 256 10000
3: [10496..20479]: hole 9984
...
looks like unwritten extents aren't getting converted?
[root@magnesium test]# ls -i file
133 file
[root@magnesium test]# xfs_db -r /dev/hda8
xfs_db> inode 133
xfs_db> p
...
xfs_db> fsblock 11543
xfs_db> type bmapbtd
xfs_db> p
...
recs[1-40] = [startoff,startblock,blockcount,extentflag] 1:[0,24,16,1]
2:[16,40,16,0] 3:[1280,1304,16,1] 4:[1296,1320,16,0] 5:[2560,2584,16,1]
6:[2576,2600,16,0] 7:[3840,3864,16,1] 8:[3856,3880,16,0] 9:[5120,5144,16,1]
10:[5136,5160,16,0] 11:[6400,6424,16,1] 12:[6416,6440,16,0] 13:[7680,7704,16,1]
14:[7696,7720,16,0] 15:[8960,8984,16,1] 16:[8976,9000,16,0]
17:[10240,10264,16,1] 18:[10256,10280,16,0] 19:[11520,11544,16,1]
20:[11536,11560,16,0] 21:[12800,12824,16,1] 22:[12816,12840,16,0]
23:[14080,14104,16,1] 24:[14096,14120,16,0] 25:[15360,15384,16,1]
26:[15376,15400,16,0] 27:[16640,16664,16,1] 28:[16656,16680,16,0]
29:[17920,17944,16,1] 30:[17936,17960,16,0] 31:[19200,19224,16,1]
32:[19216,19240,16,0] 33:[20480,20504,16,1] 34:[20496,20520,16,0]
35:[21760,21784,16,1] 36:[21776,21800,16,0] 37:[23040,23064,16,1]
38:[23056,23080,16,0] 39:[24320,24344,16,1] 40:[24336,24360,16,0]
xfs_db> fsblock 24
xfs_db> p
000: 01010101 01010101 01010101 01010101 01010101 01010101 01010101 01010101
020: 01010101 01010101 01010101 01010101 01010101 01010101 01010101 01010101
040: 01010101 01010101 01010101 01010101 01010101 01010101 01010101 01010101
...
so data is there but unwritten extents weren't unflagged.
If I make the filesystem with -dunwritten=0, it passes.
FWIW, this new test fails on an older kernel too (2.6.16.1) but passes on ext3.
--
Configure bugmail: http://oss.sgi.com/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
|