> 1) The 6x86MX kernel oops in time.c which I first reported on 11/03.
> With a workaround which needs a kernel patch.
>
> 2) The Cyrix hidden CLI bug, first reported by Erik Corry on 11/08. A
> few hours later on 11/08 I posted both a gcc demo program and a proposed
> workaround. This workaround may be implemented as a kernel patch _or_ as
> a user-space rc.cyrix script, called from rc.local.
>
> So, a new patch is needed, since the otherwise excellent 2.1.39 patch by
> Mike Jagdis does not account for these two issues at all.
>
> Mike has already signaled that he is not willing to maintain the 2.1.39
> Cyrix patch. Personally, I DO NOT wish to maintain Mike's patch either.
> It is much beyond my competence and time availability.
>
> I am perfectly willing to start afresh from my proposed small patch
> (attached), which can be found on my Linux/6x86 site:
> http://www.tux.org/~balsa/linux/cyrix and already addresses issue 1).
> The patch is properly documented on the News page.
>
> I would add features as I deem necessary, accepting input from Linux
> users that effectively have Cyrix processors. This excludes a lot of
> people who have posted to the previous thread, and I wish in the future
> they would keep out of this thread too. I am certainly not out to say
> what the Alpha code should do without having an Alpha myself, and the
> same applies to Cyrix processors, scaled down, of course.
I've already included a heavily modified version of the Cyrix patch in my
CPU detection & features code included now in the VGER CVS tree. It certainly
needs some more work (like adding fixes for the timer problems etc.), but not
too much. If you have any other features to include, please send them to me.
Have a nice fortnight
-- Martin `MJ' Mares <mj@gts.cz> http://atrey.karlin.mff.cuni.cz/~mj/ Faculty of Math and Physics, Charles University, Prague, Czech Rep., Earth "Disc space, the final frontier!"