Dreamcast AICA driver

Dreamcast AICA driver

[Adrian McMenamin]

After a mere three years in development I now have a working (mono only
for now) ALSA driver for the Dreamcast driver to test.

Sources are available for browsing at:

http://newgolddream.dyndns.info/cgi-bin/cvsweb


If there are any testers out there please let me know!

Adrian



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Lee Revell]

On Sun, 2006-03-12 at 00:52 +0000, Adrian McMenamin wrote:
> After a mere three years in development I now have a working (mono only
> for now) ALSA driver for the Dreamcast driver to test.
> 
> Sources are available for browsing at:
> 
> http://newgolddream.dyndns.info/cgi-bin/cvsweb
> 
> 
> If there are any testers out there please let me know!

Cool, what did you have to do to get it working?

Lee



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Adrian McMenamin]

On Sat, 2006-03-11 at 20:11 -0500, Lee Revell wrote:
> On Sun, 2006-03-12 at 00:52 +0000, Adrian McMenamin wrote:
> > After a mere three years in development I now have a working (mono only
> > for now) ALSA driver for the Dreamcast driver to test.
> > 
> > Sources are available for browsing at:
> > 
> > http://newgolddream.dyndns.info/cgi-bin/cvsweb
> > 
> > 
> > If there are any testers out there please let me know!
> 
> Cool, what did you have to do to get it working?


Key point was to switch to timer polling. It turned out there were
issues with the ARM7 firmware and interrupts that meant it behaved in
odd ways and essentially didn't work :)

So I still use DMA to transfer the samples, but I use timer polling to
test how much has been played



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Lee Revell]

On Sun, 2006-03-12 at 01:20 +0000, Adrian McMenamin wrote:
> On Sat, 2006-03-11 at 20:11 -0500, Lee Revell wrote:
> > On Sun, 2006-03-12 at 00:52 +0000, Adrian McMenamin wrote:
> > > After a mere three years in development I now have a working (mono only
> > > for now) ALSA driver for the Dreamcast driver to test.
> > > 
> > > Sources are available for browsing at:
> > > 
> > > http://newgolddream.dyndns.info/cgi-bin/cvsweb
> > > 
> > > 
> > > If there are any testers out there please let me know!
> > 
> > Cool, what did you have to do to get it working?
> 
> 
> Key point was to switch to timer polling. It turned out there were
> issues with the ARM7 firmware and interrupts that meant it behaved in
> odd ways and essentially didn't work :)
> 
> So I still use DMA to transfer the samples, but I use timer polling to
> test how much has been played
> 

Great!!!  I am really glad you finally got this to work - your plight
reminded me of trying to develop the emu10k1 multichannel support, when
I knew nothing about ALSA drivers, and only had existing driver and
reverse engineered hardware docs...

You should submit this to ALSA now, as a patch against CVS...

Lee



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Lee Revell]

On Sun, 2006-03-12 at 01:14 -0500, Lee Revell wrote:
> You should submit this to ALSA now, as a patch against CVS...


static int snd_aicapcm_free(snd_card_aica_t *dreamcastcard)
{
	release_mem_region(0xa0702c00, 0x4);
	release_mem_region(0xa0800000, 0x200000);
	return 0;
}

Actually this driver still has way too many magic numbers - if you're
not exactly sure what a constant does then document it with comments in
the header file.

It looks like many of these will be trivial to document:

	if (request_mem_region(0xa0702c00, 4, "AICA ARM control")== NULL) return -ENOMEM;
	if (request_mem_region(0xa0800000, 0x200000, "AICA Sound RAM") == NULL)

Lee



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Lee Revell]

On Sun, 2006-03-12 at 01:20 +0000, Adrian McMenamin wrote: 
>  http://newgolddream.dyndns.info/cgi-bin/cvsweb

You should split this into a .c and .h file

/* AICA channels stuff */

/* Command values */
#define AICA_CMD_KICK 0x80000000
#define AICA_CMD_NONE 0
#define AICA_CMD_START 1
#define AICA_CMD_STOP 2
#define AICA_CMD_VOL 3

/* Sound modes */
#define SM_8BIT		1
#define SM_16BIT	0
#define SM_ADPCM	2

etc

Lee



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Lee Revell]

On Sun, 2006-03-12 at 01:20 +0000, Adrian McMenamin wrote:
> > > http://newgolddream.dyndns.info/cgi-bin/cvsweb
> > > 
> 

> 	dreamcastcard->clicks++;
> 	if (dreamcastcard->clicks == 8) dreamcastcard->clicks = 0;

dreamcastcard->clicks %= 8

;-)

> 	/* reschedule the timer */
> 	dreamcastcard->timer.expires = jiffies + 2;

What exactly do you mean to do here - I suspect there's an unintended
dependency on HZ.  Do you intend to delay for $FOO ms?

Lee



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Adrian McMenamin]

On Sun, 2006-03-12 at 01:38 -0500, Lee Revell wrote:
> On Sun, 2006-03-12 at 01:20 +0000, Adrian McMenamin wrote:
> > > > http://newgolddream.dyndns.info/cgi-bin/cvsweb
> > > > 
> > 
> 
> > 	dreamcastcard->clicks++;
> > 	if (dreamcastcard->clicks == 8) dreamcastcard->clicks = 0;
> 
> dreamcastcard->clicks %= 8
> 
> ;-)
> 
> > 	/* reschedule the timer */
> > 	dreamcastcard->timer.expires = jiffies + 2;
> 
> What exactly do you mean to do here - I suspect there's an unintended
> dependency on HZ.  Do you intend to delay for $FOO ms?

Just copied that off the dummy driver in the kernel but yes it's just an
arbitrary delay



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Adrian McMenamin]

On Sun, 2006-03-12 at 01:28 -0500, Lee Revell wrote:
> On Sun, 2006-03-12 at 01:14 -0500, Lee Revell wrote:
> > You should submit this to ALSA now, as a patch against CVS...
> 
> 
> static int snd_aicapcm_free(snd_card_aica_t *dreamcastcard)
> {
> 	release_mem_region(0xa0702c00, 0x4);
> 	release_mem_region(0xa0800000, 0x200000);
> 	return 0;
> }
> 
> Actually this driver still has way too many magic numbers - if you're
> not exactly sure what a constant does then document it with comments in
> the header file.
> 

I know what these are - they can be turned into constants - they are
either references to the ARM7 register bank or the ARM7 memory space.


> It looks like many of these will be trivial to document:
> 
> 	if (request_mem_region(0xa0702c00, 4, "AICA ARM control")== NULL) return -ENOMEM;
> 	if (request_mem_region(0xa0800000, 0x200000, "AICA Sound RAM") == NULL)
> 
> Lee
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking scripting language
> that extends applications into web and mobile media. Attend the live webcast
> and join the prime developer group breaking into this new coding territory!
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/alsa-devel



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Adrian McMenamin]

On Sun, 2006-03-12 at 01:14 -0500, Lee Revell wrote:

> 
> You should submit this to ALSA now, as a patch against CVS...
> 
> Lee

A bit more tidying to be done as you say



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[James Courtier-Dutton]

Adrian McMenamin wrote:
> I know what these are - they can be turned into constants - they are
> either references to the ARM7 register bank or the ARM7 memory space.
>
>
>   
>> It looks like many of these will be trivial to document:
>>
>> 	if (request_mem_region(0xa0702c00, 4, "AICA ARM control")== NULL) return -ENOMEM;
>> 	if (request_mem_region(0xa0800000, 0x200000, "AICA Sound RAM") == NULL)
>>
>> Lee
>>
>>     

Adrian,

It was good to talk to you on IRC. Now, as you have the source code for 
the ARM7 firmware as well as the SH4 driver code, it might be sensible 
to highlight the dependencies between them, by maybe using the same 
human readable constant name in both bits of code. It also makes things 
easier to understand if the hex addresses are given as offsets from some 
sort of base.

if (request_mem_region(0xa0702c00, 4, "AICA ARM control")== NULL) return -ENOMEM;

then becomes in a .h file:
#define ARM7_BASE_MEM  0xa0700000
#define ARM7_AICA_CONTROL  0x2c00

During driver init:
base_mem = ARM7_BASE_MEM

if (request_mem_region(base_mem + ARM7_AICA_CONTROL, 4, "AICA ARM control")== NULL) return -ENOMEM;

This would then allow movement of the base_mem without much modification to the driver code, if for example the ARM7 firmware required the change.

James



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Adrian McMenamin]

On Sun, 2006-03-12 at 12:02 +0000, James Courtier-Dutton wrote:

> Adrian,
> 
> It was good to talk to you on IRC. Now, as you have the source code for 
> the ARM7 firmware as well as the SH4 driver code, it might be sensible 
> to highlight the dependencies between them, by maybe using the same 
> human readable constant name in both bits of code. It also makes things 
> easier to understand if the hex addresses are given as offsets from some 
> sort of base.
> 
> if (request_mem_region(0xa0702c00, 4, "AICA ARM control")== NULL) return -ENOMEM;
> 
> then becomes in a .h file:
> #define ARM7_BASE_MEM  0xa0700000
> #define ARM7_AICA_CONTROL  0x2c00
> 
> During driver init:
> base_mem = ARM7_BASE_MEM
> 
> if (request_mem_region(base_mem + ARM7_AICA_CONTROL, 4, "AICA ARM control")== NULL) return -ENOMEM;
> 
> This would then allow movement of the base_mem without much modification to the driver code, if for example the ARM7 firmware required the change.
Sure. I understand the point of course



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Adrian McMenamin]

On Sun, 2006-03-12 at 21:35 +0100, Manuel Jander wrote:
> Hi,
> 
> On Sun, 2006-03-12 at 01:38 -0500, Lee Revell wrote:
> > On Sun, 2006-03-12 at 01:20 +0000, Adrian McMenamin wrote:
> > > > > http://newgolddream.dyndns.info/cgi-bin/cvsweb
> > > > > 
> > > 
> > 
> > > 	dreamcastcard->clicks++;
> > > 	if (dreamcastcard->clicks == 8) dreamcastcard->clicks = 0;
> > 
> > dreamcastcard->clicks %= 8;
> 
> Watch out. The ARM7 does not support this in direct assembler. If the
> compiler is not smart enough it would require a division, and that is
> really realy slow on a any ARM processor. 
> Helping the compiler by using
> 
> dreamcastcard->clicks &= 0x00000003;
> 
> ... would be a much faster option. 
> 
This was actually in SH4 code rather than ARM7, though I did have some
fun and games with the ARM7 code - seemingly with division :)



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Manuel Jander]

Hi,

On Sun, 2006-03-12 at 01:38 -0500, Lee Revell wrote:
> On Sun, 2006-03-12 at 01:20 +0000, Adrian McMenamin wrote:
> > > > http://newgolddream.dyndns.info/cgi-bin/cvsweb
> > > > 
> > 
> 
> > 	dreamcastcard->clicks++;
> > 	if (dreamcastcard->clicks == 8) dreamcastcard->clicks = 0;
> 
> dreamcastcard->clicks %= 8;

Watch out. The ARM7 does not support this in direct assembler. If the
compiler is not smart enough it would require a division, and that is
really realy slow on a any ARM processor. 
Helping the compiler by using

dreamcastcard->clicks &= 0x00000003;

... would be a much faster option. 

Best Regards,
Manuel.




-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Re: Dreamcast AICA driver

[Takashi Iwai]

At Sun, 12 Mar 2006 21:35:36 +0100,
Manuel Jander wrote:
> 
> Hi,
> 
> On Sun, 2006-03-12 at 01:38 -0500, Lee Revell wrote:
> > On Sun, 2006-03-12 at 01:20 +0000, Adrian McMenamin wrote:
> > > > > http://newgolddream.dyndns.info/cgi-bin/cvsweb
> > > > > 
> > > 
> > 
> > > 	dreamcastcard->clicks++;
> > > 	if (dreamcastcard->clicks == 8) dreamcastcard->clicks = 0;
> > 
> > dreamcastcard->clicks %= 8;
> 
> Watch out. The ARM7 does not support this in direct assembler. If the
> compiler is not smart enough it would require a division, and that is
> really realy slow on a any ARM processor. 
> Helping the compiler by using
> 
> dreamcastcard->clicks &= 0x00000003;

Should be 7 instead of 3 :)


BTW, you don't have to restrict the buffer size and period size as
long as you have double-buffer, i.e. a buffer on CPU side and transfer
to the h/w buffer.  The only restricted parameter is the period size.
For example, a buffer with 13 periods can be mapped like:

BUF  0 ... 7 8 9 10 11 12 | 0 1 2 3 4 ...
H/W  0 ... 7 0 1 2  3  4  | 5 6 7 0 1 ...


Takashi


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642