Re: [PATCH] kexec for 2.5.51....

Eric W. Biederman (ebiederm@xmission.com)
15 Dec 2002 17:41:35 -0700


Ed Tomlinson <tomlins@cam.org> writes:

> On December 15, 2002 04:03 pm, Eric W. Biederman wrote:
> > Ed Tomlinson <tomlins@cam.org> writes:
> > > Why not include this info in kexec -h ? Bet it would prevent a few
> > > failure reports...
> >
> > I will look, at that.
> >
> > > Two more possible additions to the kexec command.
> > >
> > > 1. kexec -q which returns rc=1 and types the pending selection and
> > > its command/append string if one exists and returns rc=0 if nothing
> > > is pending.
> >
> > This would require effort to little purpose. If you just call kexec
> > it loads the kernel and then calls shutdown -r now. So the loaded kernel
> > should be a transient entity anyway.
>
> Consider, you are not sure what kexec has been setup to do (maybe
> some other admin has something setup to take a crash dump etc). You
> do not want to destroy this setup, so you do kexec -q
>
> Think being able to query the pending kexec is very usefull. Also
> using an rc means that scripts can use it too.

I think setting up something like /etc/lilo.conf say /etc/kexec.conf
is probably practical. You have already seen how difficult it is to
bring up the hardware into a sane state from an arbitrary point. So
we can cross the crash dump bridge when we come to it.

If this was simply a case of supporting linux the problem would be
some easier. But I also intend to support booting other operating
systems.

Now I think there is some advantage in setting up a new set of page
tables and just switching to those at kexec time, it slightly reduces
the amount of work I need to do at kexec time, enhancing the
reliability of something like kexec on panic. And that could
probably be exported in /proc/<pid>/mem. So if and when I go this
route I will keep your request in mind.

> > > 2. kexec -c which clears any pending kernels.
> >
> > This I can and should do. The kernel side is already implemented.

Oh, and of course I accept patches so you don't have to wait for me to
implement something.

Eric
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/