>Have you tried 2.1.116 or laters. There is an explicit test in the Cyrix
>code for both the presence of the DIR registers, and later for a 6x86
>class cpu before attempting any cpuid games (and all done in C). If this
>is not working I'll be very interested I finding out why, since the test
>procedures are the Cyrix Inc. recommended ones.
Actually, it seems you may be correct. These systems are mainly production
systems and I'm not really big on running development kernels on them. I
had tried 2.1.119 just for testing and it didn't work, but it was a kernel
that had been for an Intel machine in the lab and I don't know if SMP=1 was
remarked out or not. While looking at the latest 2.1.125 I noticed the new
code and basically it does the equivilant of the patches I made for
2.0.36pre15 so it should probablt work. It tests for 6x86 BEFORE trying to
enable CPUID. The problems does still appear with 2.0.36pre15. I added a
few lines of code to bugs.h to test for 6x86 before playing with CPUID and
changed the order in head.S to do the same. It's pretty illogical to try to
enable something that you're not sure is there before testing for chips that
you know don't have the feature. Anyway, the other solution may be to port
the 2.1.125 CPU stuff to 2.0.xx. Any hope of that?
Later,
Tom
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/