linux-kernel - Re: [PATCH 10/11] ASoC: qdsp6-dai: add gapless support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <04a7f696-e23d-5563-7cc3-aedfaf2c7636@linaro.org>
Date:   Wed, 8 Jul 2020 10:44:36 +0100
From:   Srinivas Kandagatla <srinivas.kandagatla@...aro.org>
To:     Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>,
        broonie@...nel.org
Cc:     alsa-devel@...a-project.org, ckeepax@...nsource.cirrus.com,
        tiwai@...e.com, lgirdwood@...il.com, linux-kernel@...r.kernel.org,
        vkoul@...nel.org
Subject: Re: [PATCH 10/11] ASoC: qdsp6-dai: add gapless support

Thanks Pierre for review,

On 07/07/2020 18:07, Pierre-Louis Bossart wrote:
> 
> 
> On 7/7/20 11:36 AM, Srinivas Kandagatla wrote:
>> Add support to gapless playback by implementing metadata,
>> next_track, drain and partial drain support.
>>
>> Gapless on Q6ASM is implemented by opening 2 streams in a single asm 
>> stream
> 
> What does 'in a single asm stream' means?

So in QDSP6 ASM (Audio Stream Manager) terminology we have something 
called "asm session" for each ASoC FE DAI, Each asm session can be 
connected with multiple streams (upto 8 I think). However there will be 
only one active stream at anytime. Also there only single data buffer 
associated with each asm session.

For Gapless usecase, we can keep two streams open for one asm-session, 
allowing us to fill in data on second stream while first stream is playing.

> 
>> and toggling them on next track.
> 
> It really seems to me that you have two streams at the lowest level, 
> along with the knowledge of how many samples to remove/insert and hence 
> could do a much better job - including gapless support between unrelated 
> profiles and cross-fading - without the partial drain and next_track 
> mechanism that was defined assuming a single stream/profile.
At the end of the day its a single session with one data buffer but with 
multiple streams.

Achieving cross fade should be easy with this design.

We need those hooks for partial drain and next track to allow us to 
switch between streams and pass silence information to respective stream 
ids.

--srini
>