cvs commit: src/sys/dev/pci pci.c

[Doug Rabson]

On Mon, 2003-09-01 at 17:41, John-Mark Gurney wrote:
> Alexey Dokuchaev wrote this message on Mon, Sep 01, 2003 at 23:02 +0700:
> > On Mon, Sep 01, 2003 at 04:40:40PM +0100, Doug Rabson wrote:
> > > I've tried healthd's ISA method as well as SMB. I know that the board
> > > uses a Winbond W83782D chip which healthd supports but the ISA ports all
> > > read 0xff and the SMB ports seem to read 0x00. I have my doubts as to
> > > whether the kernel SMBus driver works at all though.
> > 
> > I could not get kernel SMBus to work either, and also have doubts that
> > it works at all (GA 6OXET mobo, i815, ich 82801BA (ichsmb(4)) -- I think it
> > listed as supported though). :-(
> 
> alipm also appears broken.  I want to get it working since it exists
> on the sparc64 box I have.  Last I looked at it the driver was broken
> by design in that it tried to map resources that didn't exist.

Most of the pm drivers use non-standard PCI resources. They use
bus_set_resource to point out where their registers are (amdpm's
registers use RID 0x58, I seem to remember intpm sometimes uses 0x90). I
just fixed a problem in the pci code which stopped the amdpm driver from
probing since that device doesn't support standard i/o enable - it has
its own special stupid way of doing it. Why can't people build standard
pci components. Grumble :-)