From: Jeff Dike <jdike@ka...> - 2002-11-14 04:36:14
> I'd like to know what the state of stability UML is in. I understand
> that there are two different modes of operation, tt (old), and skas
> (in dev). What is the stability rating (say on a 1-10 scale, where 10
> = the stability of current x86 code) for tt, and skas.
I'm not going to do the 0-10 thing, but UML between 2.4.19-6 and when I started
in on highmem support and SMP are what I would use in production at the moment.
Bill Stearns has a UML server with (last I checked) 28 2.4.19-6 UMLs running.
They've been going since around Sept 21.
The one thing we've seen (twice that I know of) is the network driver leaving
its interrupt disabled, effectively knocking the UML off the net. This is
non-fatal and can be fixed without rebooting the UML.
Other than that, there have been no problems with the UMLs.
I would not use a skas UML in production right now. It's brand-new, there
are still bugs, and there has been a lot of code changed. The improved
performance and security are definitely tempting, so experiment with it
and put it in production when you feel comfortable with it.
From: Jeff Dike <jdike@ka...> - 2002-11-14 04:44:07
> To add another data point, i've encountered "i'm tracing myself and
> can't get out" lockups in UMLs that were using the 2.4.19-6 kernel,
You're sure that -6, and not -5 or earlier? I fixed such a bug in -6, and
I haven't heard of it happening since.
From: Jeff Dike <jdike@ka...> - 2002-11-15 14:58:32
> Don't know if Jeff will be including it in the next release of
Yeah, I've been staring at that patch. It's definitely a bug, and it definitely
needs fixing. I just need to refamiliarize myself with that code to figure
out what the right fix is.