Re: 2.2.0-pre1 worse than 2.1.132

MOLNAR Ingo (mingo@chiara.csoma.elte.hu)
Tue, 29 Dec 1998 12:31:29 +0100 (CET)


On Tue, 29 Dec 1998, George wrote:

> (gcc version egcs-2.91.60 19981201 (egcs-1.1.1 release)) #826 SMP Tue Dec 29 00:50:34 EST 1998
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

(it should work with egcs, but i'd suggest you try 2.7.2.3 too. The fact
that earlier kernels worked with your compiler means nothing, the SMP and
scheduling changes in pre5-2.1.133 (and pre-2.2) introduced some quite
optimized piece of APIC code)

i'll double check those changes now.

> Intel Pentium with F0 0F bug - workaround enabled.
> stuck on TLB IPI wait (CPU#0)
> per-CPU timeslice cutoff: 3702.36 usecs.

[what cachesize does /proc/cpuinfo report?]

> APIC delivery error (4).
> CPU #1 not responding. Removing from cpu_present_map.
> Error: only one processor found.

hm, have you enabled 'CONFIG_M686' by any chance (thats the default too)?
With those changes it's not safe to run a CONFIG_M686 kernel on
dual-Pentium boxes. (i'll post a patch today that makes this distinction
more visible and harder to get wrong)

-- mingo

-
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/