Remaining SATA (and other) issues TAKE 2

[Søren Schmidt]

Arno J. Klaassen wrote:
> I still test first on releng_6, but initial testing indicates
> it solves my problems.
>   
> Thank you very much!
>   
Now it would be interesting to know if both parts of the patch are 
needed, can you test that ?
That is just the part that changes the burst length, and just the part 
that hacks the SG list.
> NB, I still get an error in ata_pci_attach() when
> bus_alloc_resource_any() for ATA_BMADDR_RID :
>
>   pci0: child atapci0 requested type 4 for rid 0x20, but the BAR says it is an memio
>
> When I retry with "ctlr->r_type1 = SYS_RES_MEMORY" (diff attached)
> it says :
>
>   atapci0: Reserved 0x20000 bytes for rid 0x20 type 3 at 0xfba00000
>
> which it anyway says as well a bit later when attaching :
>
>   ioapic0: routing intpin 18 (PCI IRQ 18) to vector 49
>   atapci0: [MPSAFE]
>   atapci0: Reserved 0x20000 bytes for rid 0x20 type 3 at 0xfba00000
>   atapci0: Reserved 0x1000 bytes for rid 0x1c type 3 at 0xfbb00000
>   atapci0: [MPSAFE]
>
>
> Just if ever this info helps you; the failing bus_alloc_resource_any()
> for ATA_BMADDR_RID does not seem to have any influence.
>   
That's just harmless verbose chatter from the PCI subsystem. It needs to 
grow functionality to tell what kind of resources are there.
 I'll change ATA to take advantage of that when present, until then I 
have no intention of poking around PCI space registers on my own.
Keep snot and beard sorted as we say over here :)

-Søren