[PATCH] fix ras irq handlers

[Nathan Fontenot]

Everything looks good to me.

I did add three changes you may (or may not) want to include.  These
are pretty minor things I noticed in looking over the code with your
patch applied.  I included an updated patch with these updates.

- Added a comment (/* EPOW Events */) before the irq initialization of
   epow-events to atch te internal errors comment.

- Changed ras_epow_interrupt to use tha same irq manipulation,
   (virt_irq_to_real(irq_offset_down(irq)), you used in
   ras_error_interrupt for consistancy.

- Moved the taking of the log_lock in rtas_epow_interrupt to after the
   first rtas_call.  We don't need it until the second rtas call and I
   don't see any need to hold it longer than need be.

> One of the changes I have made is to check the #interrupt-cells
> applying to the interrupts property.  Without this I think we will
> incorrectly try to get interrupt 0 or 1, since when #interrupt-cells
> is 2, the second cell is the edge/level indicator.

Anton had opened a LTC bug (9508) for this issue.  I posted a patch
there awhile back but I like your implementation better.  I'll update
the bug with your patch if thats ok.

--
Nathan Fontenot
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ras_irq_update.patch
Type: text/x-patch
Size: 8967 bytes
Desc: not available
Url : http://ozlabs.org/pipermail/linuxppc64-dev/attachments/20040702/9e1a15c7/attachment.bin