Hi
pac@xxxxxxxxxxxxxx [pac@xxxxxxxxxxxxxx] wrote:
> * The maximum accessible file offset of a Linux XFS file is 16 Terabytes on 4K
> page size and 64 Terabytes on 16K page size.
Is this true?
I was able to create up to 8 exabyte sized files (with a big hole). But it
produces fs corruption.
[root@segv /root]# mkfs.xfs -f /dev/hda4
meta-data=/dev/hda4 isize=256 agcount=8, agsize=63256 blks
data = bsize=4096 blocks=506047, imaxpct=25
= sunit=0 swidth=0 blks, unwritten=0
naming =version 2 bsize=4096
log =internal log bsize=4096 blocks=1200
realtime =none extsz=65536 blocks=0, rtextents=0
[root@segv /root]# mount /dev/hda4 /mnt/
[root@segv /root]# cd /mnt/
[root@segv /mnt]# seq 1 100 | dd of=BIG1 bs=1024 seek=9007199254740991
0+1 records in
0+1 records out
[root@segv /mnt]# ls -l
total 4
-rw-r--r-- 1 root root 9223372036854775076 Oct 16 02:26 BIG1
[root@segv /mnt]# ls -lh
total 4.0k
-rw-r--r-- 1 root root 8.0E Oct 16 02:26 BIG1
[root@segv /mnt]# tail BIG1
91
92
93
94
95
96
97
98
99
100
[root@segv /mnt]#
looks good so far.
[root@segv /mnt]# cd
[root@segv /root]# umount /mnt/
[root@segv /root]# xfs_check /dev/hda4
bad agf magic # 0 in ag 0
bad agf version # 0 in ag 0
block 0/0 expected type unknown got sb
bad agi magic # 0 in ag 0
bad agi version # 0 in ag 0
bad magic # 0x58465342 in btbno block 0/0
bad magic # 0x58465342 in btcnt block 0/0
bad magic # 0x58465342 in inobt block 0/0
agi unlinked bucket 0 is 0 in ag 0 (inode=0)
agi unlinked bucket 1 is 0 in ag 0 (inode=0)
agi unlinked bucket 2 is 0 in ag 0 (inode=0)
[...]
agi unlinked bucket 61 is 0 in ag 0 (inode=0)
agi unlinked bucket 62 is 0 in ag 0 (inode=0)
agi unlinked bucket 63 is 0 in ag 0 (inode=0)
I first tried it on my system partition. It trap in kdb (X running and no
serial console). recovery crashed. xfs_repair from the 1.0.1 installer
repaired it. 465 files from 221237 were in lost+found, most from kernel
sources and a few from tuxracer. I verify the system with rpm -Va, looks
good so far. A good working repair tool is great!
Then i tried this with some kernel versions on a spare partition (hda4).
2.4.10 till todays CVS had this fs corruption.
2.4.2 till 2.4.10-pre13 had "Bad write on page 0xc13110cc0" (address varies)
errors on umount but no corruption.
All kernels were compiled with kgcc. System is a K6-500 256MB RAM.
utz
|