headers
Andreas Mohr | 19 Aug 22:33
Favicon

[PATCH] Try to make our poor little ALS4000 driver worthy of its copious public specs (first step)

[with not-quite-so-unintentional CCs added]

Hi,

there you have it: Linux developers are hypocrites ;)

Never omitting any opportunity to complain about missing vendor
cooperation, but what do they do once people actually do publish ample specs? -
yeah, you got that right: NOTHING!!

Let's try to change that, at least one bit at a time. ;)

ChangeLog:
- use specs-derived register name enums instead of open-coded numeric
values, for better understanding of things
- fix naming confusion ("gcr" was _NOT_ the GCR register stuff, but
  instead the io _base_ which has multiplexed _access_ to GCR config
  space, at _sub_ registers 0x08 and 0x0c)
- add FIXME comments about a race condition and MPU401 features

Oh, and I chose an abbreviated "ALS4K" prefix since I would have gotten
finger cramps otherwise.

Patch created on 2.6.27-rc1-git4 (no als4000.c changes up to -rc3, verified),
pushed through checkpatch.pl, verified on my ALS4000 card.

More work to follow... e.g. would be nice to get audio wakeup things to work,
since this would probably be the first (if admittedly crappy) sound card/driver
combo on Linux to support such features, thus a nice broadening of
Linux capabilities.
(Continue reading)

Permalink | Reply |
headers
John L. Utz III | 20 Aug 00:08
Favicon

Re: [PATCH] Try to make our poor little ALS4000 driver worthy of its copious public specs (first step)

On Tue, 19 Aug 2008 13:34:44 -0700, Andreas Mohr <andi <at> lisas.de> wrote:

> [with not-quite-so-unintentional CCs added]

said cc's now trimmed since they are probably both on alsa-dev.

> Hi,
>
> there you have it: Linux developers are hypocrites ;)
>
> Never omitting any opportunity to complain about missing vendor
> cooperation, but what do they do once people actually do publish ample  
> specs? -
> yeah, you got that right: NOTHING!!

Congratulations on your patch, but i dont think this comment was  
particularly helpful.

Writing audio device drivers is non-trivial, even with the spec.

I (on behalf of my employer) have made minor contributions to ALSA in the  
context of getting our product completed. The *only* reason we have the  
spec is that we committed to buying enuf motherboards that we could  
'incentivize' the motherboard manufacturer to extract the spec from the  
chip vendor.

The spec doc was less than stellar and it took months to get something  
that worked because i had no help from the chip vendor.

I am glad that the ALS4K will now have better support. This is a happy
(Continue reading)

Permalink | Reply |
headers
Rene Herman | 20 Aug 00:24

Re: [PATCH] Try to make our poor little ALS4000 driver worthy of its copious public specs (first step)

On 20-08-08 00:08, John L. Utz III wrote:

>> Never omitting any opportunity to complain about missing vendor 
>> cooperation, but what do they do once people actually do publish
>> ample specs? - yeah, you got that right: NOTHING!!
> 
> Congratulations on your patch, but i dont think this comment was 
> particularly helpful.

Oh, come one. He had his tongue quite firmly and definitely in cheek...

> Had the spec been released concurrent with the release of the chip,
> who knows what would have happened?
> 
> The ALS4k is not currently manufactured, so there is not a good
> reason to invest development resources in supporting it.
> 
> Time is better spent supporting new chips so that new distro releases
> can support new computers so that computer vendors and end users will
> feel less put out by having linux as the default OS for their brand
> new computer!

One of the main reasons for wanting specs over vendor supplied drivers 
in fact is to be able to support _old_ hardware -- hardware that the 
manfacturer feels no need to support anymore, but the users still want 
to use. As such, I believe the above is definitely off the mark.

But anyways, he was definitely being "joke-ish". You took it way too 
seriously.
(Continue reading)

Permalink | Reply |
headers
John L. Utz III | 20 Aug 00:30
Favicon

Re: [PATCH] Try to make our poor little ALS4000 driver worthy of its copious public specs (first step)

On Tue, 19 Aug 2008 15:24:29 -0700, Rene Herman <rene.herman <at> keyaccess.nl>  
wrote:

> On 20-08-08 00:08, John L. Utz III wrote:
>
>>> Never omitting any opportunity to complain about missing vendor  
>>> cooperation, but what do they do once people actually do publish
>>> ample specs? - yeah, you got that right: NOTHING!!
>>  Congratulations on your patch, but i dont think this comment was  
>> particularly helpful.
>
> Oh, come one. He had his tongue quite firmly and definitely in cheek...

Ok Rene, i will take your word for it.

>
>> Had the spec been released concurrent with the release of the chip,
>> who knows what would have happened?
>>  The ALS4k is not currently manufactured, so there is not a good
>> reason to invest development resources in supporting it.
>>  Time is better spent supporting new chips so that new distro releases
>> can support new computers so that computer vendors and end users will
>> feel less put out by having linux as the default OS for their brand
>> new computer!
>
> One of the main reasons for wanting specs over vendor supplied drivers  
> in fact is to be able to support _old_ hardware -- hardware that the  
> manfacturer feels no need to support anymore, but the users still want  
> to use. As such, I believe the above is definitely off the mark.
(Continue reading)

Permalink | Reply |
headers
Takashi Iwai | 20 Aug 10:40
Favicon

Re: [PATCH] Try to make our poor little ALS4000 driver worthy of its copious public specs (first step)

At Tue, 19 Aug 2008 22:34:44 +0200,
Andreas Mohr wrote:
> 
> Hi,
> 
> there you have it: Linux developers are hypocrites ;)
> 
> Never omitting any opportunity to complain about missing vendor
> cooperation, but what do they do once people actually do publish ample specs? -
> yeah, you got that right: NOTHING!!
> 
> Let's try to change that, at least one bit at a time. ;)
> 
> ChangeLog:
> - use specs-derived register name enums instead of open-coded numeric
> values, for better understanding of things
> - fix naming confusion ("gcr" was _NOT_ the GCR register stuff, but
>   instead the io _base_ which has multiplexed _access_ to GCR config
>   space, at _sub_ registers 0x08 and 0x0c)
> - add FIXME comments about a race condition and MPU401 features
> 
> Oh, and I chose an abbreviated "ALS4K" prefix since I would have gotten
> finger cramps otherwise.
> 
> Patch created on 2.6.27-rc1-git4 (no als4000.c changes up to -rc3, verified),
> pushed through checkpatch.pl, verified on my ALS4000 card.

Thanks, applied now.

Takashi
(Continue reading)

Permalink | Reply |

Gmane