At Sun, 10 Jun 2001 20:11:53 +0100 (BST),
Robert Swindells wrote:
> 
> 
> >Robert Swindells wrote:
> >> The one remaining problem is with the interrupt code, which is different
> >> between hpcarm and arm32. Could any of the hpcarm implementors describe
> >> why they did things differently to the arm32 derived ports ?
> 
> >Are you talking about the difference between irq_claim and
> >sa11x0_intr_establish APIs?  Or irq_entry implementation in sa11x0_irq.S?
> 
> No, the lower level code in the arm32/arm32 and hpcarm/hpcarm directories.
> 
> >I don't see any problem with these difference as they affect SA11x0
> >integrated device driver code only, AFAIK.  Could you explain the
> >situation bit more?
> 
> The SA1111 driver depends on hpcarm/softintr.c which fails to link as
> ipl_to_spl() is only in hpcarm/stubs.c not arm32/stubs.c

The cvs log have some info wrt. these changes.
I wanted to use softintr for SA1111 to implement interrupt priorities.
IMO, we should have something like current_ipl_level instead of
currnet_spl_level and avoid copying hardware interrupt masks.
I think similar change should happen on arm32 arches also.

--
IWAMOTO Toshihiro