Re: SiI 3112 ATA Controller (and UDMA33 vs SATA150)

[Dave Hayes <dave@xxxxxxxxxxx>]

Joerg Sonnenberger <joerg@xxxxxxxxxxxxxxxxx> writes:
> On Tue, Aug 23, 2005 at 01:38:27PM -0700, Dave Hayes wrote:
>> DragonFly 1.3-preview (from the CD preview image I got from some
>> .de site) tells me this:
>> 
>> pci6: <unknown card> (vendor=0x1095, dev=0x0240) at 2.0 irq 9
>
> CMDTECH AAR-1210SA? That could it be, it would need some PCI IDs and
> init-code in ata.

I thought it was an SiI 3112? FreeBSD 5.4 thinks it is, and if I
presume correctly that you are inferring this name and model from the
"vendor" and "dev" fields, there's some inconsistency here between
your call and 5.4's call. (Of course, I'm probably just ignorant of
these things and there's data I don't have.)

If someone needs to access the box to test code, I'm willing to assist
by providing an SSH login. I haven't done kernel device drivers since
college, and while I am a quite capable C programmer, I would prefer
if someone with more experience did the work.

Alternatively, is lifting the code that "properly" identifies this
from FreeBSD 5.4 ok?

>> I've had to turn ACPI off (more exactly: boot with ACPI disabled in
>> the menu that posts an ascii dragonfly) because of:
>> 
>> kernel: acpi_bus_number: can't get _ADR
>
> Does it panic or otherwise stop?

No, with the caveat that I haven't put it under heavy load yet. 

>> kernel: ad7: DMA limited to UDMA33, non-ATA66 cable or device
> Those might be harmless, depending on which controller or device it is.
> It can be fully ignored for SATA controllers.

How can I test this? :) ("dd" comes to mind...)
------
Dave Hayes - Consultant - Altadena CA, USA - dave@xxxxxxxxxxx 
>>> The opinions expressed above are entirely my own <<<

History is not usually what has happened.

History is what some people have thought to be significant.