[parisc-linux] The continuing SMP problems

[parisc-linux] The continuing SMP problems

[Matthew Wilcox]

I've tracked down one problem a little bit.  The L2k hangs with an SMP
kernel (even with only one CPU enabled) after:

Linux NET4.0 for Linux 2.4                                                      
Based upon Swansea University Computer Society NET3.039                         

What it's doing here is trying to run the keyboard_tasklet, which seems
to be disabled, so it reschedules it... ad infinitum.  More investigation
required, but I'm on the track, I think.

-- 
Revolutions do not require corporate support.

Re: [parisc-linux] The continuing SMP problems

[Grant Grundler]

Matthew Wilcox wrote:
> What it's doing here is trying to run the keyboard_tasklet, which seems
> to be disabled, so it reschedules it... ad infinitum.  More investigation
> required, but I'm on the track, I think.

I was browsing commits made between -pa19 (I think worked) and -pa30 
(know it didn't work...could have been -pa32).  That got me looking
at -pa21 commit, changing the order of device detect and other init
calls.  Comments in include/asm-parisc/keyboard.h suggest the
initialization sequence is holy for kbd-ops usage. Still poking
around there.

BTW, rbrad doesn't think that commit got tested with CONFIG_SMP enabled
and noted that smp_init() is now *before* device detection.
Another lead to followup.

grant