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
69
70
|
GNU cp used to detect files that contain 0-filled holes and save disk
space by skipping them with lseek when writing a file and thus not allocating
disk blocks.
Unfortunately, as people using Slackware 2.2 can notice, it does no
longer. So I wrote a program to make holes in existing files. To use
it, run
zum file1 file2 ...
or, if you want to process all files on your system (recommened), do:
find / -xdev -type f -print | zum
It should work find under Linux. If it runs out of disk space when
processing files (it has to make a copy of each before replacing it), just
Ctrl-C and delete all files that end with __zum__ (eq find / -xdev -name
'*__zum__' -print | xargs rm). After you free some space, it's safe to run it
from the beginning one more time. Nevertheless shell scripts in this package
modify your files and I am not responsible for anything that might happen
(hey, you have source code!)
By the way, don't try to do this on other UNIX varieties without
preliminary investigation. For example, SunOS can't boot from vmlinux with
holes (it's Ok to zum shared libraries however, because I unlink programs
before overwriting them).
While I was at it, I wrote some more scripts to save disk
space. finddup finds all the duplicate files in a subtree rooted in current
directory. Run it as:
cd /
/some/dir/finddup > /tmp/duplist
It takes quite a while to run. At the end, /tmp/duplist has groups of
duplicate files sorted in the order of decreasing size (so you can look at the
most interesting ones first). They can be merged with hard links:
cd /
/some/dir/nodup < /tmp/duplist
However you shouldn't merge all of them. Instead edit the duplist and
see what they are. For example, /root/.zshenv and /home/snowcat/.zshenv
shouldn't be merged even if they are identical, because in future I may want
to edit my .zshenv without changing root's setup. Likewise, don't link
/etc/nntpserver and /usr/adm/messages even if both consist of a single newline
character.
Finally, findstrip will find all unstripped files and write them to
stdout, line by line. Remember that you can strip only real executables, but
not shared libraries, objects and some other things like .do files in Andrew
toolkit. findstrip filters out everything I know about, but if you don't edit
the list before stripping it, you are quite likely to get in trouble.
On full Slackware 2.2 distribution (w/o Tex), these 3 scripts together
can save about 15M. It may be not much but at least it's free and doesn't have
any performance penalty :) Anyone willing to include them as part of
installation process?
Also, it's useful to make holes in files on rescue floppies.
Oleg Kibirev <oleg@gd.cs.CSUFresno.EDU>
PS: tzx is covered by GPL unless someone can give me a compeling reason to
relax restrictions.
|
