On 11/02/14 13:38, Daniel Glöckner wrote:
> On Wed, Feb 05, 2014 at 01:16:53PM +0000, Robert Longbottom wrote:
>> On 28 Jan 2014, at 02:02 AM, Daniel Glöckner <daniel-gl@gmx.net> wrote:
>>> When we cycle through all combinations in one minute, there are about
>>> a hundred PCI cycles per combination left for the chip to be granted
>>> access to the bus. I expect most of the pins to provide a priority
>>> or weighting value for each BT878A, so there should be many combinations
>>> that do something.
>>
>> How difficult is it for me to do this?  And is it obvious when it works?
>> I have an old pc that I can put the card in that doesn't matter. And given
>> I can't get the card to work in windows or Linux its not much use to me as
>> it is, so if it breaks then so be it.
>>
>> I've not done any Linux driver development, but I'm happy enough compiling
>> stuff for the most part.
>
> Try the attached program. It must be linked with -lrt. It will set all 24
> GPIOs of that one chip to output. The output file contains one nibble
> per GPIO combination with each bit representing one of the BT878.
>
> In my tests with a single BT878 the 10us delay sometimes was not enough
> for the RISC PC to advance. It should be enough to recognize a pattern,
> though.
>
>    Daniel
>
>
> DMA from userspace... I feel dirty...

Neat, thanks, I'm impressed you took the time to implement %-complete 
messages :-)  And glad you did because it took a while to run, about 15 
mins and seemed to go slower in the middle, so slow I thought it had died.

I ran it without anything else running, with no video inputs plugged 
into the card and with the bttv module loaded with no additional options 
- i.e. I just let it use the autodetect on the card.  Let me know if I 
should have plugged something in, or specified some module options and 
I'll try that.

The output is attached - I've gzip'd it because it was 8Mb and I didn't 
want to send something that big to the whole list, but I guess it might 
not get through.

I hope it means something to you because it doesn't mean anything to me :-)

Thanks,
Rob.