Re: Debugging ahcisata?

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Thu, Aug 20, 2009 at 06:10:14PM -0700, Paul Goyette wrote:
> On Wed, 19 Aug 2009, Manuel Bouyer wrote:
> 
> >I guess that if the upper level code thinks we didn't read anything,
> >it's because ahci_cmd_done didn't set AT_XFDONE. This would be because
> >the controller didn't write back the transfer count in cmdh_prdbc.
> >Can you see if adding a delay() at the beggining of ahci_intr_port()
> >changes anyything (i.e changes regiters values above, and makes the drive
> >probe).
> 
> OK, I changed the delay to 10ms, and re-enabled the debugging messages. 
> Looking at the dmesg, everything appears identical to the no-delay 
> version, at least up to the point where it displays the ST506 for drive 
> type.
> 
> The dmesg output is attached.
> 
> Looks like there's something peculiar about this controller.  I'll see 
> if I can make sense of that OpenBSD link to pointed to yesterday.

I had a look and didn't see anything obvious. Looks like we're already
doing what the patch does.
You could print the content of the buffer after the command completes,
to see if despide the transfer count being 0, the IDENTIFY data was
received from the drive.

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--