Messages in this thread | | | Date | Fri, 9 May 1997 11:35:38 -0500 (CDT) | From | Jason Burrell <> | Subject | Re: Linux passwd references. |
| |
On Fri, 9 May 1997, Matthew Haas wrote:
> Hello. > > I was wondering- how many programs rely on the presence of /etc/passwd ? > If it were to be changed in the source code to reside in a different > location under a different name, the kernel shouldn't have any problem, > but what about programs like chfn? I was thinking that if I were to do > this, it would be harder for a hacker to get to my passwd file, as 1) they > don't know where it is, and 2) they don't know what it is. > > Not traditional Unix structure, I know...but for good security I think I > could handle it. > > So my main questions are: > > Would anything break (that is known of) if I were to do this?
This is something that gets asked from time to time, and it's a bad idea. This will break anything that references the passwords of users, happens to delve into /etc/passwd to get UIDs or GIDs, anything that tries to get the finger information of users, anything that tries to get the home directories of users, etc. A good number of processes like to open /etc/passwd for no really good reason too.
> Where in the source would this change be located? I have combed it over > looking for reference to 'passwd', but have had no luck, I have both done > a find and a grep.
That isn't a kernel thing. The kernel doesn't care about /etc/passwd. It only cares about UIDs and GIDs.
I think what you really want to do is install shadow passwording. That will remove the crypt()ed password strings from /etc/passwd, while still letting most things work. I believe altering source of the things that don't work to go somewhere else, like PAM, for passwords is trivial, though I've never tried it. If you're running RedHat, I think they have an RPM to set it up.
-- Good government. Good government. Sit. Stay.
| |