Re: pci_enable_msi() for everyone?

[Jeff Garzik <jgarzik@pobox.com>]

Benjamin Herrenschmidt wrote:
>>Honestly I can think of situations where one driver would want a bit per 
>>BAR, and many others would just need a single MMIO bit.  Don't forget 
>>legacy decoding too:  with -only- a bit per BAR, the driver cannot tell 
>>the PCI layer that disabling IO means disabling a legacy ISA region 
>>that's not listed in the PCI BARs.
> 
> 
> VGA is too much of a special case here. I'm currently working on a VGA
> arbitrer but it will need a separate API (along with a userland
> interface). Maybe the kernel side of this API could be folded in that
> pci_enable() thing though, I'll have to give it a though...

I was in fact thinking of IDE not VGA :)

Let's keep it simple.

Just need to make sure that, if we use an enable-by-PCI-BAR bitmap, it 
is still possible to let the driver make decisions about enable/disable 
of the IO and MMIO bits.  You could have a PCI BAR bitmap and then two 
additional "don't touch <xxx>" bits, for example.

	Jeff