John Rose johnrose at
Wed Feb 23 03:24:16 EST 2005

> The reason we request them in the first place is to prevent drivers from
> mucking around with those bits of hardware no ? (so they fail their own
> request_region).
> I recently added a platform hook to deal with that in a slightly more
> suitable way, though not all legacy drivers have been fixed yet to use
> it.

For the case of dynamically adding a PHB to a system that booted with no
physical I/O, I need to pick a starting virtual address for the I/O
region.  I could hard-code to start past these regions, but I'm not sure
it makes sense for these to exist in the first place.  Would legacy
drivers even come into play on a purely virtual system?


More information about the Linuxppc64-dev mailing list