Re: /netbsd: panic: cpu0: ipi_send: couldn't send ipi to UPAID2

["Volkmar Seifert" <vs%nifelheim.info@localhost>]

> The only significant changes I have made, is adding a second graphics-card
> and a frontpanel with a usb-multi-card-reader / usb-hub in the meantime,
> and OHCI is failing with obscure errors. I haven't noticed those failures
> before adding that frontpanel, but I haven't looked for them, either.
> Though I don't see how, I wonder if this might be the source of the
> problem.

The OHCI is failing without that frontpanel, too, an outcome I expected to
see. Further test have turned out that (no wonder) it does not matter
what's to be compiled, as soon as there's compilation-action involved, at
least via distcc, the kernel starts to panic and dump core.
Error-message is always

cpu?: ipi_send: couldn't send ipi to UPAID? (tried 1000 times).

Where cpu? is either cpu0 or cpu1, and UPAID? always the UPAID of the
other cpu.

So far, I have not tried to run a build without pump/distcc, but may be I
will have the time to try that over the weekend, but I expect the kernel
to on that just as well, at least when setting -j to any value > 1.
Compilation of things is so far the most reliable way to provoke this issue.

What I don't understand here is why the machine did fine for months, with
heavy load, full ram, etc. with no problems whatsoever, and then suddenly
started to panic - and keeps doing so steadily. I cannot remember having
changed anything specific last friday, which was the day it all started.

- Volkmar

-- 
http://blog.nifelheim.info/tech