Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 24 Oct 1995 21:20:41 -0700 (PDT)
From:      Julian Elischer <julian@ref.tfs.com>
To:        davidg@Root.COM
Cc:        mikebo@tellabs.com, hackers@freebsd.org, bugs@freebsd.org
Subject:   Re: 2.1.0-951020-SNAP: Major bug in NFS again!
Message-ID:  <199510250420.VAA16934@ref.tfs.com>
In-Reply-To: <199510250338.UAA27854@corbin.Root.COM> from "David Greenman" at Oct 24, 95 08:38:46 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> 
>    The client should ignore NFS packets from hosts that it's not talking to or
> doesn't know about, and that's what all 4.4BSD derived OSs do.

unfortunatly it doesn't gain you anything in security to do so however

>    This is obviously flamebait and I'm not going to respond to it.

I'm pretty sure FreeBSD can be made to do the same thing too, given the
right icmp packets

> 
> >Sorry for the rambling discourse, but I need this fixed or I can't
> >use FreeBSD. At the least, can the "Sun behavior" I need be added
> >as an option to the mount command?

like most misguided security attempts it should be optional.

> 
>    If you choose not to use my suggested work-around, then I guess you can't
> use FreeBSD. For the NFS server, FreeBSD (and all other 4.4BSD derived systems)
well 4.4BSD derived OS's comprise OSF/1, NetBSD, FreeBSD and BSD/OS.
Not exactly aearthshaking combination, and I wouldn't be surprised to
see that OSF1 might act differently.. (they are actually net2.5 based)

> keep an authentication list in the kernel that is constructed from
> /etc/exports. For the NFS client, FreeBSD requires that replies to its RPC
> requests come from the same address that they were issued to. If it didn't
> work this way, then *anyone* could send bogus udp datagrams with hand-tailored
> RPC calls/replies to you and as long as that someone can come up with a file
> handle (which is relatively easy), he can do unchecked file operations and
> bypass the system security.

So? I can make my machine have any address you wish.. and probably still get 
a packet to your machine.. I mean the source address is not looked
at for routing..

It's a security feature to keep out SIMPLE attacks but fails on 
any really dedicated attack.

anyway we're talking being a CLIENT here..
you're talking about being a server, with the exports list..
I didn't notice the exports list being involved in the client side of things..

I think that if we get a patch to make this optional, then we should 
allow it to be included.. 
certainly there should be some way to tell NFS that two addresses
are equivalent. A Mount option MIGHT work, but you'd have to
feed it an alternate IP address.. (not a name)
possibbly a routing table entry could be used to do it..

>    The best I could offer you would be a kernel option to disable this
> security, but I'll say right now that this *won't* be in the 2.1 release.
> 
> -DG
> 




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199510250420.VAA16934>