>
> Steve Lord <lord@xxxxxxx> wrote:
>
> > It is all a cunning ploy by Redhat, they always sneak things up on us.
> > Given that this is a security update I suspect there will be an update
> > to 7.2 when it comes out - but to what I don't know.
> >
> > If you are worried about security then you could download a later patch
> > from our ftp site and build your own kernel until we have something in
> > a more packaged form - 2.4.10 should do it.
>
> Uhh, I would be very careful about using any Linus kernel in the
> 2.4.10 - 2.4.12 range. That was right in the middle of the radical VM
> changes that went in, and several things were broken (HIGHMEM was
> certainly broken in many of those kernels, for example).
2.4.10 seems to hold up OK under non-esoteric loads, yes I realize it is
the new VM. We have an internal development team on another project
using several highmem XFS systems with this kernel. So far they are
happy - I do want to move them on again once there is another good
kernel, the partition problems in 11/12 + the symlink thing are all
a pain in the neck.
>
> I still worry about some things being broken, since there seems rather
> sloppy testing going on before each kernel is put out, and other things
> always slip in after something is known to work.
>
> (can you tell I disapprove?)
>
>
> I've been trying to find a newer than 2.4.6-ish kernel to settle on
> (2.4.7 would be ok, the redhat patch you guys have seems interesting!),
> and some of the XFS snapshots have had leaks, and just by running a
> kernel compile loop (make clean ; make bzImage) for a few days I could
> eat up much of the available memory on my 768MB machine.
>
The 2.4.7 rpm is based on rawhide - so it is an ac kernel + the usual
thousand and one patches. Up until the arrival of the new kernel from
Redhat this was believed to be what would ship in 7.2, it may still be,
along with an instant update on their ftp site. The package is built
with kgcc by the way, still waiting for a fix from redhat on the
compiler bug.
>
> On the other hand, the recent XFS CVS of 2.4.13-pre3 seems both to work
> fine through several days of testing (I have no machine big enough to
> test HIGHMEM, though I have SMP) and several build issues were resolved
> for various modules... the common complaint about the i2o move and
> other things seem resolved.
>
> I was going to try out 2.4.13-pre5 today.
>
>
> BTW, is the patched new RH kernel based on 1.0.1, or some XFS "snapshot"?
See above.
Steve
>
>
> --
> Erich Stefan Boleyn <erich@xxxxxxxx> http://www.uruk.org/
> "Reality is truly stranger than fiction; Probably why fiction is so popular"
|