> Yes, I have working code to communicate with SAM9407 chip (microcode
> loading,
> memory management, channel allocation, control interface, etc) and I
> have just start
> working on PCM interface. Unfortunatelly, I haven't got any time to
> implement
> PCM routines, but all the low-level routines are in place. The code is
> written for older
> 0.9.x version of ALSA (as I didn't yet port it to latest CVS).
>
> If anybody is interested in helping with this project, I would gladly
> share my (partially
> functional) code. Code is tested with EWS64XL (PnP detection of SAM9407,
> i2c stuff
> and TEA chips are working OK), has many comments, and basically just
> waits for someone to
> "fill in the blanks" on PCM part.
>
I would surely be interested in this code. I think a lot of code can be shared
between the cards. The transfer mechanism of the ISIS different, but I think
it can fit into a general approach i.e. a generic driver for the SAM970x chip
supported by an 'architecture dependant' driver.
The isis uses the maestro as a PCI to ISA bride and transfers from/to the chip
should be done like this:
(disassembled source at http:/isisalsa.sourceforge.net/assembler.htm)
registers involved:
MAESTRO_BASE + 44h: address of the ISA port (IDMA in ESS terminology) on the
maestro.
MAESTRO_BASE + 46h: data register for the ISA port
so basically transfers come down to this: (disassembled from win drivers)
mov ax, 1 ; address 0x01 is the SAM control port
mov dx, MAESTRO_BASE
add dx, 44h
out dx, ax ; select control port
mov dx, MAESTRO_BASE
add dx, 46h ; switch to data register
in ax, dx ; read value
by the way: does anyone know why they don't do it like this:
mov ax, 1 ; address 0x01 is the SAM control port
mov dx, MAESTRO_BASE
add dx, 44h
out dx, ax ; select control port
add dx, 02h ; switch to data register
in ax, dx ; read value
compiler issue?
to illustrate this even further, consider this piece of code also
disassembled from the win driver:
looppoint1:
mov dx, 1
push dx
mov ax, dx
mov dx, MAESTRO_BASE
add dx, 44h ; Add
out dx, ax ; control port
mov dx, MAESTRO_BASE
add dx, 46h ; Add
in ax, dx ; read port
and ax, 0FFh ; keep lower 8 bits
* push ax
* mov ax, 2
* mov dx, MAESTRO_BASE
* add dx, 44h ; Add
* out dx, ax ; DATA16 port
* pop ax
pop dx
test al, 80h ; is TE=0 ?
jz short read_word ; TE = 1 => data present
loop looppoint1 ; Loop while CX != 0
I marked the strange part of the code with an asterix. Isn't this pure
overhead?
that aside, let's get back to the main issue...
The actual PCM data transfer is done with a 'repsw outw' command as sayd by
Uros. Something like:
mov dx, 2
mov dx, MAESTRO_BASE
add dx, 44h ; Add
out dx, ax ; DATA16 port
mov dx, MAESTRO_BASE
add dx, 46h ; Add
repsw outw ... (forgot syntax)
I wondered if it isn't possible to use the Maestro DMA capability to transfer
the large chunks from main memory to the ISA/IDMA port independant from CPU?
Do we have any information on this 'DSP modem' or IDMA port on the Maestro
family? (I assume that ESS hasn't changed that much between Maestro
generations)
Pieter
PS: I have the sourcecode of the Hoontech ST128 windows drivers (both 9x and
WDM). If you want them, you can download them at
http://isisalsa.sourceforge.net. These cards are based on SAM940x chips.
PPS: What do you think of the statement that this type of transfer isn't
supported in WinXP/NT due to the HAL that OS uses? This is the justification
of Guillemot not to provide WinXP drivers.
-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en
_______________________________________________
Alsa-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/alsa-devel
