Re: [PATCH 0/7] topology: Add support for vendor tuples

[Takashi Iwai <tiwai@suse.de>]

On Mon, 28 Mar 2016 09:49:36 +0200,
Mengdong Lin wrote:
> 
> Hi Takashi and Mark,
> 
> Despite the ABI update, could you help to review if this is a doable 
> solution to define vendor-specific data by tuples?

I'll check it tomorrow, as now it's still in Easter holidays.

> Previously there was a patch for Skylake topology conf file, but this 
> file try to import vendor specific config data as a binary blob from 
> other files. The patch was refused for on license of the binary blob and 
> we were suggested to expose the layout of the data. This patch series is 
> to meet this request, providing a way for vendors to define their 
> structured private data.
> 
> Here is the link of the refused patch:
> [alsa-devel] [PATCH] conf: topology: Add topolgy for skylake i2s 
> configuration
> 
> http://mailman.alsa-project.org/pipermail/alsa-devel/2016-February/104099.html

The changes are good if they are mandatory.  But too late for 4.6, in
anyway.  And, the missing piece is the "grand plan".  When this
API/ABI is supposed to be stabilized and released?


thanks,

Takashi

> Thanks
> Mengdong
> 
> 
> On 03/25/2016 01:39 PM, Mengdong Lin wrote:
> > http://mailman.alsa-project.org/pipermail/alsa-devel/2016-February/104099.html
> > On 03/24/2016 07:20 PM, Mark Brown wrote:
> >> On Thu, Mar 24, 2016 at 11:05:29AM +0800, mengdong.lin@linux.intel.com
> >> wrote:
> >>> From: Mengdong Lin <mengdong.lin@linux.intel.com>
> >>>
> >>> This series addes support for vendor tuples to topology, to avoid
> >>> importing binary data blob from other files.
> >>>
> >>> Backward compatibility of ABI is not impacted. A kernel patch is also
> >>> submitted "ASoC: topology: ABI - Define types for vendor tuples".
> >>
> >> It's now been three kernel releases since the original code was merged
> >> and we've still not seen any effort to make the ABI actually available.
> >> Where are we with that?  I've mentioned it a few times previously but
> >> don't really recall any response.
> >>
> >
> > We've been working on this. I'm sorry we're still not fast enough.
> >
> > This ABI update is introduced by previous failure to upstream the
> > topology configuration file for Skylake, which imports vendor-specific
> > binary data from another file. It was refused because of the license of
> > the binary data is not obvious. So we turn to use vendor tuples, because
> > the vendor tuples can show the layout and meaning of binary data, which
> > can help to check if the data is safe to open source.
> > This ABI change has no impact on backward compatibility.
> >
> > Also, the audio device driver for new platforms are under development
> > now, which uses topology for DAI and DAI links. We're checking if
> > current ABI can meet all requirements for the new platforms. This may
> > still need a few weeks.
> >
> > Thanks
> > Mengdong
>