Support for SAM9407?

Support for SAM9407?

[Ash Willis]

Hi, I've just been digging around for information on the old Dream SAM9407 chip.
What's the current support status in ALSA for this chip? I know some work was
done on producing drivers for cards based on this chip and I know there were
plans to port the efforts to ALSA. Did that ever happen? It doesn't seem so.
Was there a specific reason why it wasn't ported in the end?

In the interest of making ALSA that bit more complete as well as trying to
improve my skills, I've been looking around for some older hardware that I can
write drivers for.

This chip seems like a good candidate considering the cards haven't quite
disappeared into oblivion :) I was looking into the possiblity of porting the
remaining kernel OSS drivers to ALSA...but I can't even find any test hardware
because they're so rare.

Ash Willis

=

Re: Support for SAM9407?

[Rene Herman]

On 05/01/2007 01:18 PM, Ash Willis wrote:

> Hi, I've just been digging around for information on the old Dream
> SAM9407 chip. What's the current support status in ALSA for this chip?

Unsupported.

> I know some work was done on producing drivers for cards based on this
> chip and I know there were plans to port the efforts to ALSA. Did that
> ever happen? It doesn't seem so. Was there a specific reason why it
> wasn't ported in the end?

I don't know about reasons others may have had, but I have been working off 
and on, but significantly more off than on, on my TerraTec EWS64XL which is 
a Crystal CS4236 + SAM9407. Basically just two soundcards on one board. The 
CS4326 part works fine with the existing ALSA driver.

The EWS64XL itself is a very complex card with various functionality, such 
as switching the outputs to the SAM from the CS4236, hanging off one of two 
bitbanged I2C busses (that share one clockline), with having to manually 
setup the memory decoding logic after detecting what sort of SIMM you have 
installed, with the 'F' (front) and 'R' (rear) digital extensions with the 
digital out- and inputs both being available but different, with it having 
both coaxial and optical digital in that you have to switch around, with a 
dedicated clock chip for the SAM that can be programmed (by poking bits) to 
various clock rates, and so on, and so on.

Many (if not most, it's been quite some time since I looked at the driver 
and I'm not completely sure anymore about its current state) of those card 
details I have in fact figured out but I have only done some superficial 
probing around the SAM9407 itself. It's not a very nice chip in fact. It's 
strictly PIO only (over ISA!) and strictly firmware driven, with slightly to 
significantly different firmwares (and hence APIs) depending on the card 
upon which it sits and is poorly documented.

MIDI is the main objective of the SAM, but that sort of has gone the way of 
the dodo in itself and streaming PCM out an I/O port to the ISA bus is just 
not something that makes a great deal of sense when a few centimeters away 
you have a DMA capable CS4236 sitting that's already supported (and sounds 
very, very good -- it's just a well-built board). There's a pair of 20-bit 
DACs behing the SAM, but that's only an advantage for things like adding 
effects and I don't generally add effect to my music.

Basically, the SAM9407 is for (home-) musicians, but musicians have long ago 
gone to more capable hardware. For MIDI I myself have stuck a Yamaha DB-50XG 
on its front panel "waveblaster header" (which is routed to the CS4236 
MPU401 and its CD input) and that probably actually sounds better than the 
SAM ever would...

> In the interest of making ALSA that bit more complete as well as trying
> to improve my skills, I've been looking around for some older hardware
> that I can write drivers for.
> 
> This chip seems like a good candidate considering the cards haven't quite
> disappeared into oblivion :) I was looking into the possiblity of
> porting the remaining kernel OSS drivers to ALSA...but I can't even find
> any test hardware because they're so rare.

I'd think twice about the SAM. Other than the complexity of the cards on 
which it sits the SAM itself is also complex. Due to that PIO thing you'll 
need all sorts of double buffering and will need to feed the card from a 
kernel thread or something like that and will furthermore need to concern 
yourself with firmware loading, and yes, all the card details; on this EWS64 
you'd first need to know how to setup memory decoding for example before you 
can use it for the SAM and ofcourse, need to know to switch the outputs to 
the SAM to actually hear anything in the end. The OSS driver that still 
floats around the net is not something you'll enjoy looking at. It doesn't 
require a port but a rewrite from scratch.

For me this EWS64XL was the first reason I ever looked at ALSA and I _have_ 
thought about it twice and will in fact want to completely support it "some 
day" but there's quite a few things I want to happen "some day" ;-/

How about a Media Vision Pro Audio Spectrum 16? Those can still be found 
relatively easily and it's one the cards for which an OSS/Free driver is 
still kept in the tree. I have a pair of those as well so I'll happily test 
if you want to write a driver for it...

Rene.

Re: Support for SAM9407?

[Pieter Palmers]

Ash Willis wrote:

> In the interest of making ALSA that bit more complete as well as trying to
> improve my skills, I've been looking around for some older hardware that I can
> write drivers for.

Are you interested in joining the FFADO project? (short into: FFADO = 
new FreeBoB = drivers for FireWire Audio Interfaces / www.ffado.org).

We need somebody that wants to write the ALSA support for FreeBoB / 
FFADO. Currently we only support jackd directly. The main reason that 
there is no ALSA support is simply that there is nobody that is doing 
it. We have a library that exposes an API that can be used to write 
backends for audio API's, and we have the jackd backend as an example. 
It's simply a matter of putting it all together, but we need someone to 
do it.

To be frank/direct here, I think that ALSA is better 'completed' with 
FFADO support than with driver support for obsolete cards. But of course 
I'm a little biased here :).

Greets,

Pieter Palmers

Re: Support for SAM9407?

[Ash Willis]

Thanks for your detailed response. That chip definitely does seem like a beast
that I really don't want to tame...

> 
> How about a Media Vision Pro Audio Spectrum 16? Those can still be found
> relatively easily and it's one the cards for which an OSS/Free driver is
> still kept in the tree. I have a pair of those as well so I'll happily test
> if you want to write a driver for it...
> 

Yeah, we have a winner. I just ordered one. I'm busy right now but I will
get this started in 4 weeks. Thanks for the offer of testing. I'll hunt you
down when I have something to show you :)

Ash

=

Re: Support for SAM9407?

[Ash Willis]

Hi Pieter,

> Are you interested in joining the FFADO project? (short into: FFADO =
> new FreeBoB = drivers for FireWire Audio Interfaces / www.ffado.org).
> 

I have absolutely zero experience in firewire audio and firewire in
general to be honest, but the project does sound interesting.

> To be frank/direct here, I think that ALSA is better 'completed' with
> FFADO support than with driver support for obsolete cards. But of course
> I'm a little biased here :).

Be as Frank as you want, I totally agree with you :) The thing is, I have
limited time and experience (I have written one ALSA driver). Although
supporting older hardware doesn't bring a whole lot to the project, it
does provide me with a training ground so I can be of greater use in the
future. I can work at my own pace on an area that nobody else will be
working on. I can improve my skills without stepping on other people's
toes.

Your project sounds interesting but I don't feel I have the knowledge to
keep up on a project like this. This is why I'm messing around with this
stuff, so that I can do some 'real work' some day soon ;)

regards,
Ash

=

Re: Support for SAM9407?

[Pieter Palmers]

Ash Willis wrote:
> Hi Pieter,
> 
>> Are you interested in joining the FFADO project? (short into: FFADO =
>> new FreeBoB = drivers for FireWire Audio Interfaces / www.ffado.org).
>>
> 
> I have absolutely zero experience in firewire audio and firewire in
> general to be honest, but the project does sound interesting.
>
You don't need any FireWire experience, because libffado abstracts 
everything FireWire from the user (in this case user = ALSA plugin).

>> To be frank/direct here, I think that ALSA is better 'completed' with
>> FFADO support than with driver support for obsolete cards. But of course
>> I'm a little biased here :).
> 
> Be as Frank as you want, I totally agree with you :) The thing is, I have
> limited time and experience (I have written one ALSA driver). Although
> supporting older hardware doesn't bring a whole lot to the project, it
> does provide me with a training ground so I can be of greater use in the
> future. I can work at my own pace on an area that nobody else will be
> working on. I can improve my skills without stepping on other people's
> toes.
ALSA support for FFADO is not being done by anyone currenlty, so no toes 
to step on there.

If you already have written one ALSA driver, you already have the 
experience required I guess. The ALSA support take the form of an ALSA 
plugin, which should be easier to write than a driver. In any case it is 
userspace, so no kernel oops'es.

> 
> Your project sounds interesting but I don't feel I have the knowledge to
> keep up on a project like this. This is why I'm messing around with this
> stuff, so that I can do some 'real work' some day soon ;)

I think you do... I think I was at about the same level as you are now 
when Daniel and I started FreeBoB. We do this to learn, and as long as 
nobody else steps up, nobody has the right to criticize you. And over 
the 3 year's we're already working on this, there hasn't been anyone 
stepping up for the ALSA support. So I guess you're clear :).

Again, it would be very nice if you could join us. You would make a 
difference for a lot of people.

Greets,

Pieter