Re: autri0: can't map memory space

[gson%gson.org@localhost (Andreas Gustafsson)]

Jared D. McNeill wrote:
> There's no reason that the autri driver cannot modify this register 
> itself. Can you try the attached patch with a GENERIC kernel and no 
> PCI_*_FIXUP options?

I tried the patch.  A kernel with the patch and no PCI_*_FIXUP hangs
during boot attaching fxp0 which is the next device after autri0,
right after printing the line "fxp0: interrupting at irq 10".

If I remove the patch (and keep PCI_*_FIXUP disabled), the machine
boots (but autri0 fails to attach, of course).

I can't see anything wrong with your patch, so it seems strange that
it should cause a hang during the attach of an unrelated device, but I
tested this several times with and without the patch and that's what
happened.

There are several other weird things going on with -current on this
machine, such as cbb0 reporting "bad Vcc request" and fxp0 reporting
"dmasync timeout" when I try to configure it with an IP address; maybe
the hang is somehow related to these.  In any case, with so many
things being broken in possibly interrelated ways, it's hard for me
to give more useful feedback right now.

The machine is a Toshiba Libretto L2, by the way.
-- 
Andreas Gustafsson, gson%gson.org@localhost