> pcibios_find_device is usually not broken, but the PCI BIOS itself is. I've
> already seen BIOSes that crash when asked to do less common things (like reading
> config data from a non-existing card). So I vote for not using the BIOS whenever
> possible.
Martin, reading/writing config space for a non-existing PCI card from the
PCI BIOS read/write functions is _not_ specified by PCI BIOS
specifications.
If you are using the PCI BIOS, you must scan devices using a 'find'
function and then you can only access the config space of boards you
'found' this way using the PCI BIOS read/write functions.
Scanning the PCI BUS using PCI BIOS read/write functions is a
_non-existing_ feature and so cannot be broken.
If you are willing to do the job of the PCI BIOS or just are pissing it
off as hard as it merits, you must use a configuration mechanism. These
mechanism _are_ specified and so exist, but BTW, they may be broken.
I agree with you that we should'nt use the PCI BIOS when a configuration
mechanism is working ok. My thought is that it is the case most of the
time (always?). Finally, the need of PCI configuration knowledge in a
decent kernel is far beyond the informations PCI BIOS services are able
to return, so I do think that:
We should _always_ use a configuration mechanism in Linux and discard
definitely the PCI BIOS.
Gerard.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu