NATA testing

[Thomas Nikolajsen <thomas.nikolajsen@xxxxxxx>]

I have tested NATA on two systems running HEAD:
Both systems works fine with the old ATA driver.

 1) non-portable: amd64
atapci0: <VIA Apollo ATA controller> port 0xc800-0xc80f at device 15.0 on pci0
atapci0@pci0:15:0:	class=0x01018a card=0x92231509 chip=0x05711106 rev=0x06 hdr=0x00

Works fine; apart from missing core dump noted below.

 2) laptop: TP X32
atapci0: <Intel ICH4 ATA100 controller> port 0x1860-0x186f,0-0x3,0-0x7,0-0x3,0-0x7 at device 31.1 on pci0
atapci0@pci0:31:1:	class=0x01018a card=0x052d1014 chip=0x24ca8086 rev=0x01 hdr=0x00

Problems at boot: init fails; single user works until loading sh(1).
Disabling DMA (hw.ata.ata_dma=0) makes it work.

On reboot ACPI don't shut down system; power isn't turned off;
same problem with freebsd 6.2. ATA kernels doesn't have this problem.


On both systems core dumps at a panic I experience don't work;
it just succeds in "no" time w/o the usual disk activity.
Manual core dump ("panic" in ddb) does work.
Reverting to ATA kernel makes core dumps at panic work again.

 -thomas