2000-11-10 03:18:19

by H. Peter Anvin

[permalink] [raw]
Subject: CPU detection revamp (Request for comments)

Hi guys,

I wanted to give you a preview of the CPUID revision patch; it is not
100% ready yet in that (a) it still has a bunch of debugging printk() and
(b) I haven't ported over mtrr.c yet.

This code took a lot longer to write than I had expected, mostly because
I kept running into various forms of bugs with the *old* code. In fact,
it looks like some of the bugs involving CPUID feature flags in the past
have been due to the inconsistent handling of these flags in Linux. I
believe the new code should be a lot better in this aspect, and should
avoid bugs as much as possible.

I would very much like this code to be tested and get success/fail
reports on various CPUs, *especially* AMD (Athlon especially) and Cyrix
chips.

As I mentioned, you will have to compile without MTRR support in this
version. I should have that fixed soon.

The patch is at:

ftp://ftp.kernel.org/pub/linux/kernel/people/hpa/cpuid-2.4.0-test11-pre2-1.diff

In case of success, please send me the output of the new /proc/cpuinfo on
your system. In case of failure, please compile and run the following
program and send me the output, as well as /proc/cpuinfo from an
unpatched kernel version:

ftp://ftp.kernel.org/pub/linux/kernel/people/hpa/cpuid.c

Many thanks,

-hpa

--
<[email protected]> at work, <[email protected]> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt