Re: Multiple Processors on OS_MCA

From: Russ Anderson <rja_at_sgi.com>
Date: 2004-08-04 07:08:09
Tony Luck wrote:
> 
> >2.6.7 kernel/mca_asm.S serializes processors at the entry and exit of
> >the OS_MCA with a spinloop, hence makes sure only one 
> >processor performs the OS_MCA at a time.
> >A processor toggles the spinloop var before exiting OS_MCA so that
> >another processor spinning on the entry can proceed to OS_MCA.
> 
> That wasn't intended to be a final solution ... just a temporary fix
> to make sure that two processors couldn't enter the kernel at the
> same time (which would lead to chaos in the current implementation
> because we only have one block of memory declared to save registers,
> use a stack etc.).

Is there a plan for the next step towards the final solution?

For example, would it make sense to have per CPU register & stack
save areas (etc) to handle multiple OS_MCAs concurrently?

-- 
Russ Anderson, OS RAS/Partitioning Project Lead  
SGI - Silicon Graphics Inc          rja@sgi.com
-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Received on Tue Aug 3 17:09:48 2004

This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:29 EST