--ew6BAiZeqk4r7MaW
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Fri, Dec 19, 2003 at 10:55:52AM +0100, Martin Husemann wrote:
> On Thu, Dec 18, 2003 at 11:30:35PM +0100, R. Braun wrote:
> > The processes waste ressources, and mostly cpu time... The only way to =
get rid
> > of them is to reboot... It's not clean at all and should be handled cor=
rectly
> > everywhere imho.
>=20
> I have not looked at what realy catches the problem, but I supposed it
> would be the=20
>=20
> #ifdef DEBUG
> 	write_user_windows();=20
> 	...
> #endif
>=20
> call in syscall(). This is not cheap, and wasting it for every system=20
> call just to catch bogus programs is wrong, IMHO.
>=20
> We should, however, make sure the exit syscall can succeed even if the re=
turn
> path to userland is borked, and we should make sure such processes are=20
> killable.
>=20
> Martin
>=20

That's exactly what I meant by saying it "should be handled correctly". Eve=
ryone
can start a stupid inifinte loop, but if even root can't stop it, it can be
considered as a security problem.

Do you think it can be done before 1.6.2 release ?

--=20
R. Braun

--ew6BAiZeqk4r7MaW
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (NetBSD)

iD8DBQE/4s5+PEHFMcoF1vIRAoCfAKCDtdl0A9DQuX4jxoa6emr9FUhmdQCdFdDY
pvLHd8CdRjCLY557fvc8J4g=
=efwP
-----END PGP SIGNATURE-----

--ew6BAiZeqk4r7MaW--