Subject: Re: split fxp driver
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: Johan Danielsson <joda@pdc.kth.se>
List: tech-kern
Date: 05/30/1999 15:06:16
Jason Thorpe <thorpej@nas.nasa.gov> writes:

> On 26 May 1999 16:35:21 +0200 
>  joda@pdc.kth.se (Johan Danielsson) wrote:
> 
>  > I just realized that the fxp driver is all-in-one PCI. I had hoped
>  > that adding cardbus support would be easier than it turned out to be.
>  > 
>  > Has anyone done anything about this?
> 
> Not yet, but it wouldn't be that hard to do...
> 
> Stay tuned.

Ok, I stood tuned, and while waiting I started hacking on this
myself. I've now got something that almost works. The card is found,
and attached, but access to the CSR seem to stop working after that.

The symptom is `WARNING: SCB timed out!' messages, and I interpret
this as if it's not reading from the CSR, but some other memory. I
believe, since it can read the EEPROM, that it works during attachment
(but fxp_attach doesn't access the SCB, so I can't tell if *that* is
working).

Other interesting feature of the driver: it assumes that the EEPROM is
64 words, but in CardBus applications it is 256 words. I don't know
how to detect this. The datasheet says that the EEPROM pulls the line
low for one bit after the address has been successfully shifted in,
but isn't this just electrically, or can you tell from software?

Does anyone have any good ideas about any of this. I'll try to put the
code in some accessible place later.

/Johan