Re: Nvidia MCP67 SATA controller

[Dieter Baron <dillo%danbala.tuwien.ac.at@localhost>]

On Wed, Jun 18, 2008 at 08:50:21PM +0200, Manuel Bouyer wrote:
> On Wed, Jun 18, 2008 at 08:23:04PM +0200, Dieter Baron wrote:
> [...]
> > > This is almost a duplicate of ahci_reset(), which is the first thing done
> > > in ahci_attach(). I think you don't need this function at all,
> > > just set the AHCI_GHC_AE bit in the force_ahci case, and let ahci_attach()
> > > to the reset.
> > 
> >   I tried that first, but then it never worked the first time after
> > power on.  I had to do a reset when the first boot hung, then the
> > second boot would succeed.
> 
> It's probably the same issue that cause your drive to not be probed
> properly some times. It's most probably timing-related and I guess it works
> with your patch only because of the 1s delay it adds.
> Could you see if the attached patch would help ?

  I've removed the reset (and 1s delay) and applied your patch, and it
seems to work -- it came up from a power-on and probed the drives
correctly.

  So it's time to clean the patch up and commit it.  As David Holland
has noted, this is needed for more than just the MCP67, so force_ahci
should be determined by a table.  Is there a generic framework for
quirks, or do I have to roll my own?

                                        yours,
                                        dillo