Hello,
I was trying to delete all my .gnome* files and found one
of the subdirectories had a couple of corrupted entries.
I've a suse linux 7.2 installation and this partition was
formatted with a tea hash and 3.5.x format.
A first pass of reiserfsck gave a lot of these entries
(cat -v reiserfsck.txt):
[...]
bad_directory_item: block 559823: item 203453 203647 0x1 DIR, len 314, entry count 12,
fsck need 0, format old has entry "." with wrong deh_state 147174
bad_directory_item: block 559823: item 203453 203647 0x1 DIR, len 314, entry count 12,
fsck need 0, format old has entry ".." with wrong deh_state 147406
bad_directory_item: block 778847: item 203798 203799 0x1 DIR, len 171, entry count 7,
fsck need 0, format old has entry "." with wrong deh_state 147016
bad_directory_item: block 778847: item 203798 203799 0x1 DIR, len 171, entry count 7,
fsck need 0, format old has entry ".." with wrong deh_state 147406
bad_directory_item: block 639932: item 204129 204166 0x1 DIR, len 35, entry count 2,
fsck need 0, format old has entry "." with wrong deh_state 145706
bad_directory_item: block 639932: item 204129 204166 0x1 DIR, len 35, entry count 2,
fsck need 0, format old has entry ".." with wrong deh_state 147424
ok
Comparing bitmaps..free block count 54701 mismatches with a correct one 54733.
on-disk bitmap does not match to the correct one. 10 bytes differ
ok
Checking Semantic tree...check_semantic_pass: hash mismatch detected (M-H3&^HM-
M-jM-^D@^P)
check_semantic_pass: hash mismatch detected (M-^Hs4^HM-
M-^E@^P)
ok
Note that the lines referring to hash mismatch are those two corrupted
entries in my .gnome directory.
A second pass with the -o option gave this (cat -v reiserfsck2.txt):
[...]
bad_directory_item: block 204701: item 178967 178968 0x1 DIR, len 63, entry count 3,
fsck need 0, format old has entry "." with wrong deh_state 147476 - fixed
bad_directory_item: block 204701: item 178967 178968 0x1 DIR, len 63, entry count 3,
fsck need 0, format old has entry ".." with wrong deh_state 141676 - fixed
bad_directory_item: block 559055: item 203798 203799 0x1 DIR, len 171, entry count 7,
fsck need 0, format old has entry "." with wrong deh_state 147016 - fixed
bad_directory_item: block 559055: item 203798 203799 0x1 DIR, len 171, entry count 7,
fsck need 0, format old has entry ".." with wrong deh_state 147406 - fixed
bad_directory_item: block 639792: item 204129 204166 0x1 DIR, len 35, entry count 2,
fsck need 0, format old has entry "." with wrong deh_state 145706 - fixed
bad_directory_item: block 639792: item 204129 204166 0x1 DIR, len 35, entry count 2,
fsck need 0, format old has entry ".." with wrong deh_state 147424 - fixed
ok
Comparing bitmaps..free block count 656612 mismatches with a correct one 656644.
on-disk bitmap does not match to the correct one. 10 bytes differ
ok
Checking Semantic tree...check_semantic_pass: hash mismatch detected (M-H3&^HM-
M-jM-^D@^P)
check_semantic_pass: hash mismatch detected (M-^Hs4^HM-
M-^E@^P)
ok
A third pass with the --rebuild-tree gave this
(cat -v reiserfsck4.txt):
####### Pass 0 #######
pass0: block 718765, item 6584 44416 0x1 DIR, len 85, entry count 4, fsck need 0,
format old: 2 entries were deleted of
not set got 2 hits
"tea" got 29112 hits
####### Pass 1 #######
####### Pass 2 #######
####### Pass 3 #########
dir 6584 44416 has wrong sd_size 85, has to be 35
dir 1 2 has wrong sd_size 145, has to be 171
####### Pass 3a (lost+found pass) #########
The two corrupted entries are gone.
This is the first time I got a problem with reiserfs (besides one with
rm long time ago when it still was experimental). Can someone explain what
is the meaning of the errors for the ".", ".." directories? and if this
is something I should worry about.
I don't remember why I used tea for this partition, it was a while ago, does
it matter?
PS. I've put all the outputs in this directory if you want to look at them
in full:
http://www.neuronet.pitt.edu/~raffo/reiserfsck/
Thanks.
--
Rafael
