On Mon, Jan 24, 2005 at 03:48:49PM -0800, List Mail User wrote:
> 	It just looks like DEC did something very weird on the Alphas in
> general.
> 
Well, from an alphastation 255:
000:06:0: Symbios Logic 53c810 (SCSI mass storage, revision 0x11)
000:07:0: Intel 82378ZB System I/O (SIO) (miscellaneous prehistoric, revision 0x43)
000:11:0: 3Com 3c905C-TX 10/100 Ethernet with mngmt (ethernet network, revision 0x78)
000:12:0: S3 ViRGE/DX (VGA display, revision 0x01)
000:14:0: Digital Equipment DECchip 21040 ("Tulip") Ethernet (ethernet network, revision 0x24)

The onboard 53c810 and on-board tulip are at 0xff, the 3com and S3 are at
0xf8. Note that the S3 has its maximum latency set to 0xff. The 53c810
has it at 0x40, the 3com 0x0a, the 82378ZB and 21040 0x0

> 	Could someone else check some other non-x86 machines, both old and
> relatively current to see the behavior (I thinking of Suns and/or SGI boxes).

On a Sun Ultra/5:
# pcictl pci1 list
001:01:0: Sun Microsystems, Inc. PCIO Ebus2 (miscellaneous bridge, revision 0x01)
001:01:1: Sun Microsystems, Inc. PCIO Happy Meal Ethernet (ethernet network, revision 0x01)
001:02:0: ATI Technologies 3D Rage Pro (VGA display, revision 0x5c)
001:03:0: CMD Technology PCI0646 (IDE mass storage, interface 0x8f, revision 0x03)
# pcictl pci1 dump -d 1 -f 0 |grep -i lat
    Latency Timer: 0x52
    Maximum Latency: 0x19
# pcictl pci1 dump -d 1 -f 1 |grep -i lat
    Latency Timer: 0x52
    Maximum Latency: 0x05
# pcictl pci1 dump -d 2 |grep -i lat     
    Latency Timer: 0x42
    Maximum Latency: 0x00
# pcictl pci1 dump -d 3 |grep -i lat
    Latency Timer: 0x10
    Maximum Latency: 0x04
# pcictl pci2 list
002:01:0: Promise Technology PDC20269 Ultra/133 IDE controller (miscellaneous mass storage, interface 0x85, revision 0x02)
002:03:0: Symbios Logic 53c875/876 (SCSI mass storage, revision 0x03)
# pcictl pci2 dump -d 1 |grep -i lat
    Latency Timer: 0x20
    Maximum Latency: 0x12
# pcictl pci2 dump -d 3 |grep -i lat
    Latency Timer: 0x8c
    Maximum Latency: 0x40

So, it looks like the openfirware did something (these are not default
values), but they don't look right either ...

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--