"Matt D. Robinson" wrote:
> Larry Cohen wrote:
> >
> > In smp_send_stop(), I commented out the call to disable_local_APIC(),
> >
> > The core dumps now succeed.
> >
> > Note. I just read the archives and saw that the kernel patch was
> > updated. I had already noticed
> > the __SMP__ bug and fixed my local copy (forgot to mention this when
> > talking with Matt). So I was
> > really calling smp_send_stop() which called disable_local_APIC().
> >
> > Any ideas on why this is causing me grief?
> > Is commenting out the routine going to cause other major problems?
>
> I don't know -- I'll have to look at it more closely to
> find out why (I'm tracking down another lcrash bug at the
> moment from Mark Price). Unless the APIC is affecting the
> IDE driver directly ...
Unfortunately commenting out disable_local_APIC() did not work as well as
I thought.
It worked fine when I had pulled one of my cpu's out of the motherboard as an
experiment
(there really is no way to disable a cpu from the bios ??? ). When
I plugged the cpu back and then
tried a crash, the system would panic in flush_tbl_others() because:
if ((cpumask & cpu_online_map) != cpumask)
BUG();
Its getting closer but I'm not there yet. ....
>
>
> > Also, when I finally go to use lcrash I noticed that there were no
> > stack traces. Am I missing
> > something here?
>
> No stack traces at all, or no stack traces for the failing
> process? I corrected a KDB patch issue earlier which caused
> problems for the failing process, but not every stack trace.
>
No stack trace for any process. I saw the KDB patch and applied it. Did not
make any difference (should it have ?).
Thanks again for your help Matt,
-Larry
|