RE: [ECOS] Driver for the SMC 91c96 eth chip in RedBoot?

["Jeremias, Thilo" <tjeremias at rsasecurity dot com dot au>]

As I am currently doing the same thing for an SA1100,
I would like to see those files too!

cheers Thilo

-----Original Message-----
From: schuehle@t-online.de [mailto:schuehle@t-online.de]
Sent: Thursday, 26 July 2001 23:47
To: Jordi Colomer
Cc: ecos
Subject: Re: [ECOS] Driver for the SMC 91c96 eth chip in RedBoot?


Am Don, 26 Jul 2001 schrieben Sie:
> > 
> > I just looked at the Linux version of the driver which supports also 8
bit
> > i/o access. All they do is to exchange the inw/outw with 8 bit versions
which
> > do 2 inb/outb calls.
> > I suppose that if i:
> > 1) implement put_reg() and get_reg() in such a way that 2 8-bit i/o
> > accesses are done
> > 2) write a wrapper func for HAL_READ_UINT16 and HAL_WRITE_UINT16
> > to do 8 bit accesses
> > 3) hide above changes with some #ifdefs and macros
> > the driver could be made quickly working in 8-bit mode.
> > 
> > What do you think? Is this the right way to do it?
> > 
> 
> Your idea looks fine. However, you should take into account what
> I said about the strange address bus alignment. In my patch I have
> solved this by simply adding a config variable that shifts the
> address before the inw/outw by a fixed number of bits. 
> When in 8-bit mode, then the number of bits must be reduced by one.
> 
> It is important not slow down these functions, for obvious 
> performance reasons.
> 
> BTW, I see that my patch hasn't been applied yet because the
> people at Redhat can't find my copyright assignment, but it 
> will be solved soon probably.
> 
> The patch is needed so that the address shift and PCMCIA mode
> is supported. If it gets delayed, I can send you the files.

Sorry to bother you again, but can you please send me the files anyway? I
think
i'll be able to start this thing quite soon, so ....

Thank you,
	Stefan

> 
> Jordi Colomer