Re: [irda-users] Workaround for a bad stir4200

[andrzej z. <ba...@za...>]

On 15/09/06, Samuel Ortiz <sa...@so...> wrote:
> Hi Andrzej,
>
> On Mon, Sep 11, 2006 at 03:37:32PM +0200, andrzej zaborowski wrote:
> > Hi,
> > I have a stir 4200 dongle which doesn't work with the driver from the
> > kernel. The chip does have a SigmaTel logo on it and looks genuine.
> > Infrared data seems to be received correctly but nothing meaningfull
> > is being emitted. It turns out that the FIFOCTL register always has
> > the RXERR bit set when the driver reads it so the loop in
> > fifo_txwait() exits immediately and no delay between TX bytes occurs,
> > so the data comes out all garbled. I made a small change in the loop
> > to make it keep waiting despite RXERR or TXERR and IrDA frames can now
> > be sent and received correctly. The errors are still reported, they
> > just don't interrupt the loop.
> I was reading the only stir4200 spec file that I have, and it seems that
> bits 0 and 1 from FIFOCTL are reserved, we shouldn't care about them.
> I'm actually considering removing those checks from the driver, but first I'd
> like to hear from the maintainers (Paul, Stephen, Jean). They may explain
> us why this check on some reserved and undocumented bits is done.

Ah, cool. So my dongle complies with the specification. Thanks for
checking this, I was almost sure there wouldn't be any specification
availiable so I didn't bother to look for it :-)  Instead I tried
"spying" on a closed-source driver I had, which didn't help much. The
closed-source driver does more or less the same usb requests except it
sets the "magic" REG_DPLL value to 08 instead of 15.

>
> Cheers,
> Samuel.
>

Regards,
-- 
balrog 2oo6