1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
|
#! /bin/sh
. ./common.sh
cd mnt
# See fs/ext2/ialloc.c, the comment above ext2_new_inode, about ext2's
# location policy. We're probably still doing things slightly wrong.
# (E.g. we're currently ignoring the effect that existing files, like
# the root directory and /lost+found, have on allocation.)
dd_bs=4k
n_big=10
dd_big_count=100
dev="`df . | sed -n 2p | awk '{ print $1 }'`"
if [ -r "$dev" ]; then :; else
echo "$dev not readable; exiting" >&2
exit 1
fi
blks_per_grp=`LANG=C /sbin/tune2fs -l "$dev" 2>/dev/null | sed -n 's/^Blocks per group: *//p'`
blk_count=`LANG=C /sbin/tune2fs -l "$dev" 2>/dev/null | sed -n 's/^Block count: *//p'`
first_block=`LANG=C /sbin/tune2fs -l "$dev" 2>/dev/null | sed -n 's/^First block: *//p'`
n_grps=`expr $blk_count - $first_block - 1`
n_grps=`expr $n_grps / $blks_per_grp + 1`
n_sm=`expr 2 \* $n_grps`
for i in `seq $n_sm`; do
r=`expr $i % 255 + 1`
# Create file sm$i consisting entirely of character code $r.
# The idea of using a different repeated character for each file
# is so that defrag errors are more likely to be picked up.
# The idea of using repeated characters, rather than /dev/urandom,
# is to make the filesystem compressible, to allow a cmp.
# [For related amusement, search the linux-kernel mailing list
# archives for `/dev/monkeys'.]
mkdir dir$i
dd if=/dev/zero of=- bs=$dd_bs count=1 2>/dev/null | tr \\0 \\`printf '%o' $r` > dir$i/sm$i
done
# Create some holes in the allocation, so that we can create large
# files that are fragmented.
n_rm=$n_grps
for i in `seq $n_rm`; do
rm dir$i/sm$i
done
# Create fragmented big files.
for i in `seq $n_big`; do
# Note that $n_big can be bigger than $n_grps, hence mkdir.
mkdir -p dir$i
r=`expr 255 - $i % 255`
dd if=/dev/zero of=- bs=$dd_bs count=$dd_big_count 2>/dev/null | tr \\0 \\`printf '%o' $r` > dir$i/big$i
done
# Check that both fast & slow symlinks work.
# (In ext2fs, symlinks whose target name is 60 (?) bytes or shorter
# store the target name where the blocks are usually found. E2defrag
# mustn't treat the target name of fast symlinks as block data.)
# (Handled by defrag.c:optimise_inode.)
ln -s fast fastlink
ln -s this-is-a/-slow-symlink/-needing-its/-own-block/-allocated/-for-it slowlink
sync
|
