[OpenAFS-port-darwin] 1.2.10 and patches?

Ragnar Sundblad ragge@nada.kth.se
Thu, 28 Aug 2003 04:39:28 +0200


--On den 27 augusti 2003 17:43 +0200 Sebastian Hagedorn 
<Hagedorn@uni-koeln.de> wrote:

> CAVEAT
>      Access() is a potential security hole and should never be used.

If I understand the issue correctly, access is not a security
problem as such, but could be one if misused by a setuid
program, exactly as much as stat and similarly or more or
less as any other syscall.

It is about the possibility of a setuid program that first
checks permissions for the real user to access a file, and
then going ahead doing it, and the user could have replaced
that file between the time of the access and the time of
the doing (for example writing) with a link to another file
of the users choise, like /etc/passwd or similar.

Any program that handles security has to do it right, or the
system is not secure. That is no news, and the reason for
having that text in the man page seems to be that there was
a program that had the security problem describe above.

I'd say, that the "potential security hole" issue is not a
valid reason for Finder not using it.

/ragge