Re: netbsd-5 system hangs while booting since recent kernel update

[Matthew Mondor <mm_lists%pulsar-zone.net@localhost>]

On Wed, 13 Jan 2010 22:33:11 +0100
Matthias Drochner <M.Drochner%fz-juelich.de@localhost> wrote:

> j+nbsd%2010.salmi.ch@localhost said:
> > I'm not familiar with the code; what might be the problem with it?
> 
> A PCI device driver should never touch the PCI_COMMAND_IO_ENABLE
> and PCI_COMMAND_MEM_ENABLE bits in PCI_COMMAND_STATUS_REG.
> 
> >   auich0: can't map codec i/o space

The related PR was kern/39652, the patch was confirmed by me to fix the
issue with a thinkpad I no longer own on which the driver would fail to
work otherwise.  It even was pulled-up to netbsd-5 already.

> If the IO space is not enabled this is for a reason which is
> outside the reach of the driver. Either is is impossible due to
> address space shortage, or a lazy BIOS didn't care to map it.
> For the latter case we have these *FIXUP* options.
> If the BIOS didn't map the range, chances are high that the
> base address register is still zero; enabling IO access will
> immediately cause a decoding conflict at low addresses where
> the standard ISA components are mapped.
> 
> This change is wrong and should be backed out.

I had no problems with other auich systems which previously worked and
still do, but if the patch is wrong I have no objection if it's back it
out either.  Perhaps the *FIXUP* options you describe could provide a
better alternative, but I unfortunately won't be able to test it, or at
least not easily as I'd have to borrow again said laptop which I gave
away.

Thanks,
-- 
Matt