"Ronald G. Minnich" <rminnich(a)lanl.gov> writes:
On Wed, 11 Dec 2002, Adam Agnew wrote:
we're going to have to fix that. I hoped not to do this but too many OSes
either can't do it or get it wrong.
I know we'd like it for the Bochs stuff. FreeBSD and Windows 9x series
certainly throw fits about it.
I'm working on sucking in the linux PIRQ code now. It took me some time to
figure out an approach but I think I know what to do.
Linux PCI code has diverged significantly from what we started with, which
is part of the fun.
And the Linux PCI code never sets the IRQ registers in PCI space
at any point. (Which is a good thing or we might have problems
with kexec). In any event per chipset part of the code is
only about 10 lines of code per vendor, so it should not be
to hard to handle.