[alsa-devel] Digital Audio Multiplexer of the i.MX27 processor

[Juergen Beisert]

Aditya,

On Friday 30 May 2008 10:57, Aditya Shankar wrote:
> On Tue, May 27, 2008 at 5:13 PM, Juergen Beisert <jbe at pengutronix.de> wrote:
> > Hi,
> >
> > does anybody understand the DAM unit in the i.MX27 processor?
> > This routing device drivers me crazy, as I do not understand how to route
> > clocks, syncs and the data signals of my internal i2s unit to the
> > processor's
> > pins.
>
>  The DAM in the Freescale chip has 3 internal ports and 3 external ports.

Are you sure? It seems the DAM unit in the i.MX27 has two on chip ports and 
four external ports.

This I found yet (comparing mx27ads sources with mx27ads schematics):

  SSI     DAM       ext. pins
---------------------------------------
 SSI1    port_1
 SSI2    port_2
         port_3        ???
         port_4   SSI1_* pin group  <-- connected to PMIC on my board
         port_5   SSI2_* pin group
         port_6        ???

There are still the SSI3_* and SSI4_* pin groups, where I'm not sure about 
their connection to DAM unit's ports.

> > All I can see is, when I enable the internal SSI unit in i2s mode, the
> > DMA unit transfers some data into it, but the SSI unit does not shift it
> > out. So
> > everything stops, the DMA timeouts and I cannot hear anything. I can see
> > the
> > external device generates clock and frame signals, but it seems they do
> > not reach the internal SSI unit.
>
>  When you configure the SSI in a slave mode, the PMIC
> ( or your external device)will drive the FS and the clock
>  required.

SSI1 is configured as a slave, FS and clocks are inputs. Routing through the 
DAM unit seems to work now, as the SSI unit shifts out the data I'm writing 
into its FIFO.

> You say that the clock and frame sync from the ext. device is generated.
>  SO can you check if the SSI is interrupting the DMA.

This is currently my problem. The DMA does not work. If I enable the DMA it 
seems to write something into the SSI FIFO. I can check it by reading the 
FIFO counter. But at this point of time SSI1's dma_req is still disabled 
(TDMAE is 0 in register SIER)! When I enable TDMAE after enabling the DMA 
channel, next the DMA unit outputs an error: Error while transfer (DSESR bit 
of the used channel goes 1).  When I transfer the sound data in software it 
works. I'm confused.

Juergen

-- 
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
 Pengutronix - Linux Solutions for Science and Industry
    Handelsregister: Amtsgericht Hildesheim, HRA 2686
         Vertretung Sued/Muenchen, Germany
   Phone: +49-8766-939 228 |  Fax: +49-5121-206917-9