> Hello,
>
> > If really we want a PCI config method to be selected in such a situation,
> > then, the best one is PCI DIRECT, in my opinion, at least for x386
> > platforms. The PCI BIOS should be considered as some work-around for too
> > broken PCI chipsets we just donnot want to deal with.
>
> The most reliable thing we can do with the current PCI code is to enable
> both BIOS and direct access -- if the BIOS at least correctly reports basic
> setup information, we use it just for detecting which hardware access
> method should we use, avoiding misdetection of that method. This is what
> we have by default in recent 2.1 kernels.
If you mean that it is the DIRECT method which will be used when both
methods (BIOS+DIRECT) are available, then I agree 100% with your code.
It seems that on x386, using a DIRECT method is more safe than
trusting PCI BIOSes.
As an example of that, it seems that some not old AWARD bioses donnot
detect additionnal functions of PCI multi-function devices, but using
a DIRECT method is OK. As a result, users that have this crappy PCI
BIOS are just stuck with PCI multi-function devices under linux-2.0.
Regards,
Gerard.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/