Shell Programming and Scripting

hello all.

I'm not asking about the algorithm - or how it goes about computing the checksum - I'm asking how it views/looks at the files it does the cksum on.

For instance:

Say you had a directory named "dir_A"

And within this directory you had some files.

So:

dir_A
- file1
- file2
- file3

Do you have to step through dir_A and compute a cksum for EACH individual file?

For instance:

% cd dir_A
% cksum file1
% return some number
% cksum file2
% return some number
% cksum file3
% return some number

Or, can I just compute the cksum for the whole directory?

For instance:

% cksum dir_A
% return some number

If one of the files within dir_A changes and I redo the cksum is it safe to say that a valid cksum will be computed for the whole directory? What if file1 WITHIN dir_A changed - will the cksum change?

Or, do you have to compute a cksum for each individual file?

Thanks!!!

Why didn't you try it to see???

You can't run checksum on a directory (or I couldn't).

To do a checksum on the files in the directory,

$ sum ./directory/*.*
Note that it won't do hidden files.

You can cksum or md5 a directory. I am able to accomplish this on FreeBSD, but under Slackware, it does not appear to work.
If you want you can redirect the md5 into a txt file, so you can check it later.
Changing a file in the directory won't change the cksum, md5. (correction), but adding or removing files will.

you can issue
cksum A_dir
md5 A_dir

LOL..

I did try it. It worked for me.

I'm running digital unix tru64 4.0f.

%cksum dir_A
% 12345678 8908 dir_A

I changed a file in dir_A and got another number.

I was just curious if the number computed can be reliable or am I missing something...

Thanks for the replies...

I believe that the cksum is being computed on the directory entry itself - i.e. the actual "file" on disk that contains a list of the files that the directory contains.

e.g. under HP-UX 10.20
If you do
$ sum dir_A
2022 2 dir_A
This is the same as
$ cat dir_A | sum
2022 2

Then,

Code:

$ sum dir_A
2022 2 dir_A
$ touch dir_A/foo
$ sum dir_A
2492 2 dir_A
$ vi dir_A/foo		# add two hundred lines of whatever
$ sum dir_A
2492 2 dir_A		# see - file has changed, but checksum hasn't
$ rm dir_A/foo
$ sum dir_A		
2022 2 dir_A		# now checksum has returned to previous value

So, from what I can see, only the addition and removal of files within the directory (thus changing the directory entry) will actually modify the checksum. If you change the contents of a file, this won't make a difference.

It looks you would have to do something like (as previously mentioned)
$ sum dir/* >list # or cksum dir/*, md5sum dir/*, whatever
To get a list of checksums and then periodically verify those.

Same results seem to appear with cksum.

I don't know if this helps/answers any questions, but it's a result of my probing around.

Oh, some versions of Linux/UNIX don't allow you to compute checksums on directories, also mentioned above....

That's what I was curious about.

I got mixed results when I tested it myself, but was unable to figure out what was going on.

That's very helpful.

Thanks.