[LinuxBIOS] [PATCH] More vt8237r fixes

[Corey Osgood]

Rudolf Marek wrote:
> Corey Osgood wrote:
>> See patch. Rudolf, can you test this one to make sure it works correctly
>> on your board?
>
> Hi,
>
> Your patch seems not so solve KBD and RTC correct?

Correct, kbc/rtc init should handle things their own way. I thought
about doing that when we come to the lpc controller, but IIRC that's
already taken care of in vt8237r_lpc, so I figured I'd leave it alone.

> This could be implemented in superio on other board for example. Also
> the clock gating for Ethernet should be made programmable.

Yep, it should.

>
> index = PCI_FUNC(dev->path.u.pci.devfn);
>
> This is not very user friendly, index is quite confusing, in fact I
> was staring to the code quite long time to get how it works. Perhaps
> two variables would be more handy???
>
> I would suggest some lookup table to make it more elegant???? Plus we
> will need at least for RTC and KBD some? variables, just like
> enable_internal_RTC....
> or something shorter.

Uwe and I discussed this on IRC. I'm working on it (among other things) now.

>
> More over you cant enable just UHCI function 2 and not enable function
> zero and one. Plus you cant disable all UHCIs when EHCI is enabled.
> (you would violate PCI specs not having fn0)

Very good point! I'll keep it in mind. And you need at least one UHCI
controller for EHCI to work.

> The lookup table could be:
>
> u8/u16 enable_bits[dev-15][fn=6] = { { 0xb , 0x0}, { c, d, a, 8, 9} ....
>
> Or alternatively we could store the bitmask which would allow to
> enable usb fn0 when someone wants fn1 for example...
>
> The table would be 4*6*2 bytes so 48B, this is not so much contrary to
> the big if else if code...
>
> Rudolf

-Corey