RE: An alternative...

I assume my PCI id is the slot number that I installed the board into. I had
to have it to map the PCI card to the kernel /dev/comedi0. 

My status is the same as last week. I am not knowledgeable enough to go into
the driver code and change these things outlined by Ivan. I am at the mercy
of you "smart guys" to write the drivers. I will test out anything that is
written and give the reflector a status. I am in dire need of using this D/A
card with linux. 

I recently purchased a A/D card (PCI-DAS6402/16) from Computer Boards and it
should be here this Friday. I made sure that the card was in the linux list
prior to purchasing it.

Thanks for all your support. I could not do this without you guys helping
me.

> >-----Original Message-----
> >From: Ivan Martinez [mailto:imr_at_oersted.dtu.dk]
> >Sent: Monday, May 27, 2002 3:42 AM
> >To: Calin A. Culianu
> >Cc: timothy.curry_at_rdec.redstone.army.mil
> >Subject: Re: An alternative...
> >
> >
> >On Friday 24 May 2002 11:20 pm, Calin A. Culianu wrote:
> >> Hmm.  I am looking at these register specs for the 
> >pcim-dda06/16 and it
> >> looks somewhat different than the pci-dda0x/12 series of 
> >boards.  It isn't
> >> like the difference between the PCI-DDA02/04/08 boards, 
> >which basically
> >> seem to be subsets of one another register-wise.
> >
> >	I see... then you better leave the cb_pcidda driver 
> >untouched, and develop a 
> >new one. But, of course, you can take cb_pcidda as an 
> >starting point. I can't 
> >give you detailed instructions, because I developed the 
> >driver some time ago. 
> >	I think you should start doing the following:
> >- change the defines for the meaningfull addresses and 
> >commands in your card
> >- specify the available ranges in comedi_lrange struct
> >- substitute the entries in cb_pcidda_boards for the one 
> >with the specs of 
> >your card.
> >- the same in cb_pcidda_pci_table, you just have to set the 
> >ID of your 
> >specific card.
> >- in cb_pcidda_attach, I think you should only make 
> >modifications from 
> >"Allocate the subdevice structures". There you set one 
> >subdevice for every 
> >functionallity of your card: analog input, analog output, 
> >digital I/O... Just 
> >set the analog output subdevice, at the moment.
> >- change cb_pcidda_ao_winsn to set one output channel 
> >depending on its 
> >arguments.
> >	Notice that the name of many structures and functions 
> >contain the name 
> >"cb_pcidda" and you should change that depending on the name 
> >you are going to 
> >give to your driver.
> >	If you do all this properly and we don't forget 
> >anything, we would be able 
> >to get a simple driver that can set one output with 
> >"comed_data_write" 
> >function. Do this before starting with commands support or 
> >other features.
> >	More questions?.
> >
> >> I am wondering, how do you think Mr. Curry was able to get 
> >his board to be
> >> recognized by the cb_pcidda driver?  I thought each PCI 
> >device model had a
> >> unique vendor + model id?
> >
> >Yeah, that's a good one. The driver should report "Not a 
> >supported ... card 
> >on requested position" in kernel log. How did you manage, 
> >Timothy?. Do you 
> >know your card PCI id?.
> >
> >> -Calin
> >>
> >> On Fri, 24 May 2002, Ivan Martinez wrote:
> >> > If you find difficult to understand what is in 
> >cb_pcidda.c file, you can
> >> > compare the specs of the PCIM-DDA06/16 with the ones of 
> >the PCI-DDA0X/12
> >> > boards. These are the ones used to initially develop the driver.
> >> > Regards, Ivan.
> >

Received on 2002-05-29Z11:53:53