Finding sendmail during configure
rra at stanford.edu
Thu Sep 4 17:54:55 UTC 2003
Erik Klavon <erik at eriq.org> writes:
> Most people use sendmail and feel that, when properly configured, it in
> general poses no greater threat then other programs with similar
> privileges. So I think there is no harm in looking in the same places
> for sendmail as everyone else does, and using it if found. This behavior
> should be documented in INSTALL in an obvious place, of course.
What do you think about the question of whether to look along the user's
path or not? One of the problems that I had with the original behavior
(and have had since the beginning) is that the stated reason for it was to
avoid using some random version of sendmail that the sysadmin didn't know
they had installed, but we looked for sendmail on the path of the user
doing the compile. I think that's *far* more likely to pick up some
broken sendmail than looking in /usr/sbin and /usr/lib.
It seems to me that if there's a broken sendmail in /usr/sbin or /usr/lib,
the user is going to have a lot more problems than just an outside chance
of a security hole via INN, given the amount of software that just assumes
something that's command-line-compatible with sendmail is there
(particularly in /usr/lib).
Russ Allbery (rra at stanford.edu) <http://www.eyrie.org/~eagle/>
Please send questions to the list rather than mailing me directly.
<http://www.eyrie.org/~eagle/faqs/questions.html> explains why.
More information about the inn-workers