On Thu, Apr 26, 2007 at 12:17:27AM +0200, Luc Verhaegen wrote:
On Thu, Apr 26, 2007 at 12:03:19AM +0200, Peter Stuge wrote:
Unfortunately none of the subsystem devices on this board seem like excellent picks for a unique board match (2b80 is listed as "AMD-8111 IDE [Quartet]" in my pci.ids, but if forced I would pick LPC and USB:
0x1022, 0x7468, 0x1022, 0x2b80, 0x1033, 0x00e0, 0x0e55, 0x2928
The IDE collision is no problem here. It's the whole set that needs to be matched and it seems as if it is isa bridge subsystem id versus ide controller main id. So no problem there.
Sorry, should have been more clear.
My point was that 1022:2b80 is listed as a subsystem for "Quartet" which is another mainboard. Yet it's also used on this board, from a different vendor, Iwill. Quartet could be a reference design and GPIOs could have been changed the dk8htx but subsystem ids still be from the reference design. This needs to go onto the hardware design gotchas wiki page that I'm working on. :p)
This seems as good a match as any, i hope that the other board Ning was mentioning differs from this.
This is my worry.
Well, pci subsystem ids are as reliable as the board vendors.
At best. PCI IDs are read from flash/eeprom sometimes and then they're not very useful anymore. :(
There are other options, but pci subsystem ids are very universal and easy to retrieve.
No doubt! It's much better than nothing, but I don't like heuristics unless it's the last resort.
Maybe it will be OK to have overlaps in the list though, and try all matching entries. Yes, who knows what the GPIOs will cause if twiddled the wrong way on the first match that's actually for another board, but we can't know until we've tried it.
The original BIOS will know, should we try to pick it's mind?
Do you have any experience from reading factory BIOS data structures found in RAM somewhere while working with the VIA boards?
//Peter