[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Re[4]: general protection: 4f00



On Thu, 20 Aug 1998, Shlomi Fish wrote:

> 
> If you consider fatal buffer overflows a bug, then i386 Linux is not

What should one consider it, a feature? A birthday surprise? A practical
joke?

> bug-free. AFAIK, one could say the same thing for Solaris x86, FreeBSD,

Do you know about overflows in 2.0.35 kernel? Did you report it to
linux-kernel and other relevant lists and to kernel maintainer (Alan Cox
presently, AFAIR)?

> However, an overflow that is fatal on one system is often not fatal on
> another. Recently, a hacker used a (then) recently discovered overflow in
> bind to methodically crash hundreds of Linux hosts in Israel.

This has nothing to do with linux or any OS at all. This is BIND, this is
user program. BTW, one may take a look on bind-chroot - a chrooted BIND.
Should make remote BIND roots less easy.

> If you ask me, Linux (and UNIX in general) should have a system call with
> which root can give permissions for another user (and optionally -
> process group) to open a socket on a port below 1024, in-order to
> eliminate those incumbresses. Is anybody with me on this?

Capabilites. Should be in 2.2, IIRC.
-- 
frodo@sharat.co.il	\/  There shall be counsels taken
Stanislav Malyshev	/\  Stronger than Morgul-spells
phone +972-2-6245112	/\  		JRRT LotR.
http://sharat.co.il/frodo/	whois:!SM8333