Re: [patch 04/30] ACPI driver support for pata

[Tejun Heo <htejun@gmail.com>]

Alan Cox wrote:
>> using separate pata_acpi driver (we can definitely implement pata_acpi
>> using the helpers).  It will reduce general confusion and allow
>> combining acpi cable detection with specialized device handling (e.g.
>> ACPI cable detection combined with ADMA command operation).
> 
> Read the code Tejun, I did exactly that. pata_acpi is a driver which uses
> a nice clean set of helper methods I added to libata-acpi. pata_acpi
> itself knows nothing about ACPI other than calling into libata-acpi.

Actually, I have.  I was thinking of higher level helpers.  e.g.
pacpi_cable_detect() in libata-acpi.c such that sata_nv's cable_detect
can do.

int nv_cable_detect(struct ata_port *ap)
{
	if (ck804) {
		if (pacpi_cable_detect(ap) == 0)
			return 0;
	}

	/* do original nv cable detection */
	return rc;
}

So that it's more robust && we don't have to alternate between sata_nv,
pata_acpi depending on BIOS/kernel configuration, and possibly combine
other low level features with ACPI support.

-- 
tejun