On Aug 30, 2007, at 2:18 PM, Chris Gilbert wrote:

> Manuel Bouyer wrote:
>> On Thu, Aug 30, 2007 at 11:18:22PM +0900, Izumi Tsutsui wrote:
>>> chris@dokein.co.uk wrote:
>>>> Does this look ok?  I'd like to check in the chage in then next  
>>>> couple
>>>> of days.
>>> If the flag is cats specific, it's better to use proplib(3)
>>> to specify such quirks rather than config flags,
>>> as iyonix/autoconf.c, sgimips/autoconf.c etc. do
>>> for their onboard PCI devices.
>>
>> Yes, this looks better to me as well
>>
>
> Ok, attached is the diff using proplib instead of flags.  Does this  
> look ok?
>
> Thanks,
> Chris
> Index: arch/cats/cats/autoconf.c
> ===================================================================
> RCS file: /cvsroot/src/sys/arch/cats/cats/autoconf.c,v
> retrieving revision 1.11.6.1
> diff -u -p -r1.11.6.1 autoconf.c
> --- arch/cats/cats/autoconf.c	11 Aug 2007 21:14:52 -0000	1.11.6.1
> +++ arch/cats/cats/autoconf.c	30 Aug 2007 21:08:56 -0000
> @@ -152,5 +150,18 @@ cpu_configure(void)
> void
> device_register(struct device *dev, void *aux)
> {
> +	if (device_is_a(dev, "aceride"))
> +	{

you should consult the actual PCI bus/device/function numbers, the way  
other platforms fix-up MI drivers for MD quirks.

The name of the property should reflect that it's aceride-specific  
(and not have a name that ties it to the driver): ali1543-ide-force- 
compat-mode

>
> +		prop_bool_t force_compat_mode;
> +
> +		force_compat_mode = prop_bool_create(true);
> +
> +		if (prop_dictionary_set(device_properties(dev),
> +					"force-compat-mode",
> +				       	force_compat_mode) == false)
> +			printf("WARNING: unable to set force-compat-mode "
> +	       				"property for %s\n", dev->dv_xname);
> +	     	prop_object_release(force_compat_mode);
> +	}
> }
> /* End of autoconf.c */
> Index: dev/pci/aceride.c
> ===================================================================
> RCS file: /cvsroot/src/sys/dev/pci/aceride.c,v
> retrieving revision 1.23
> diff -u -p -r1.23 aceride.c
> --- dev/pci/aceride.c	9 Feb 2007 21:55:27 -0000	1.23
> +++ dev/pci/aceride.c	30 Aug 2007 21:08:56 -0000
> @@ -167,7 +167,30 @@ acer_chip_map(struct pciide_softc *sc, s
> 	    ~ACER_CHANSTATUSREGS_RO);
> 	cr = pci_conf_read(sc->sc_pc, sc->sc_tag, PCI_CLASS_REG);
> 	cr |= (PCIIDE_CHANSTATUS_EN << PCI_INTERFACE_SHIFT);
> +	
> +	{
> +		/*
> +		 * some BIOSes (port-cats ABLE) enable native mode, but don't
> +		 * setup everything correctly, so allow the forcing of
> +		 * compat mode
> +		 */
> +		prop_bool_t pn;
> +		pn = prop_dictionary_get(
> +				device_properties(&sc->sc_wdcdev.sc_atac.atac_dev),
> +				"force-compat-mode");
> +		if (pn != NULL) {
> +			KASSERT(prop_object_type(pn) == PROP_TYPE_BOOL);
> +			if (prop_bool_true(pn))
> +			{
> +				cr &= ~((PCIIDE_INTERFACE_PCI(0)
> +					| PCIIDE_INTERFACE_PCI(1))
> +					<< PCI_INTERFACE_SHIFT);
> +			}
> +		}
> +	}
> +
> 	pci_conf_write(sc->sc_pc, sc->sc_tag, PCI_CLASS_REG, cr);
> +
> 	/* Don't use cr, re-read the real register content instead */
> 	interface = PCI_INTERFACE(pci_conf_read(sc->sc_pc, sc->sc_tag,
> 	    PCI_CLASS_REG));

-- thorpej