[123] in linux-security and linux-alert archive
Re: "Find all the SUID programs." Fine. So which *should* be SUID?
daemon@ATHENA.MIT.EDU (Jeff Uphoff)
Sun Mar 12 15:48:46 1995
Date: Sun, 12 Mar 1995 15:20:42 -0500
From: Jeff Uphoff <juphoff@tarsier.cv.nrao.edu>
To: linux-security@tarsier.cv.nrao.edu
In-Reply-To: Your message of Sat, March 11, 1995 23:25:44 -0800
Reply-To: linux-security@tarsier.cv.nrao.edu
"EL" == Elias Levy <elias@power.net> writes:
EL> On Sun, 12 Mar 1995, Benedikt Stockebrand wrote:
>> Coding the proper permissions inside the binaries. Make the program
>> check its own permissions upon startup and add an option like
>> "--check-own-permissions" to it.
>>
Well, that involves either getting every program-designer/coder to agree
to add this option to their program (simply because the Linux community
thinks that it would be a good idea), or having a Linux "permissions"
patch for every piece of software that's included in standard
distributions, etc. What's the use for striving towards compliance and
portability if we're going to try and implement something like this?
Not an option, IMHO...can you imagine trying to tell someone like
Richard Stallman that all the GNU software should have this feature
added for Linux? This idea would die (from politics if nothing else)
_very_ rapidly.
Even your idea of hacking GCC's startup module to do something like this
at program load would run into heavy resistance (and you're right in
saying that it would be a kludge)--I don't see H.J. Lu et al. adding
this to the GCC code any time soon...
And what if you later decide that you want, say 'rtin', to run setgid to
"news" because you're switching from a local threading/indexing database
to an NFS mounted one that requires this? Now you have to recompile
'rtin' so that it will accept running itself setgid, instead of just
changing the permissions and being done with the matter.
EL> And just who decides what are the proper permissionsfor every diferent
EL> package? Thats the real problem.
I think the real problem is the implementation, not the permissions
decisions, but I agree: who decides? Every installation is different in
at least some subtle way...
--Up.
