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]
Date:   Tue, 9 Oct 2018 16:09:05 +0900
From:   Jiada Wang <jiada_wang@...tor.com>
To:     Kuninori Morimoto <kuninori.morimoto.gx@...esas.com>
CC:     <lgirdwood@...il.com>, <broonie@...nel.org>, <perex@...ex.cz>,
        <tiwai@...e.com>, <alsa-devel@...a-project.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [alsa-devel] [PATCH linux-next v2 9/9] ASoC: rsnd: add busif
 property to dai stream

Hi Morimoto-san

Thanks for your comment

On 2018/10/09 9:44, Kuninori Morimoto wrote:
> Hi Jiada
>
>> SSI can work in following modes
>> 1. Basic Mode: (channel 1, 2, 4, 6, 8, 16)
>> 2. TDM Extended Mode: (channel 6, 8)
>> 3. TDM Split Mode: (channel 1, 2)
>> 4. TDM Ex-Split mode: (Channel 2, 4, 6, 8, 10)
> (snip)
>>> So, my opinions for BUSIFn support are
>>> 	- SSI mode should be selected automatically
>> can you give me your idea, how to automatically determine working mode,
>> when user plays 2 channel stream on playback dai-link
> If my understanding was correct, we can do like this
> If DT indicated sound card has dai-link x N, tdm-slots = <M>,
> 	If (N, M) = (1, 2) : Basic mode
> 	If (N, M) = (1, >2): TDM mode
> 	If (N, M) = (2, 4) : TDM Split mode
> 	If (N, M) = (2, >4): TDM Ex-Split mode
> 	If (N, M) = (>2, 8): TDM Split mode
> 	...
>
> Maybe some combination was wrong, but we can do something like this ?
The idea to consider tdm_slot when determine SSI mode makes sense to me,
by checking runtime channel and tdm_slots combination,
I think SSI mode can be automatically selected like following:

1ch:  (tdm_slots < 4) Basic mode, (tdm_slots >= 4) TDM Split mode
2ch: (2 <= tdm_slots < 8) Basic mode, (tdm_slots >= 8) TDM Ex-Split mode
4ch: (4 <= tdm_slots < 8) Basic mode, (tdm_slots >= 8) TDM Ex-Split mode
6ch: (6 <= tdm_slots < 8) Basic mode, (tdm_slots == 8) TDM Extended 
mode, (8 < tdm_slots) TDM Ex-Split mode
8ch: (6 <= tdm_slots < 8) TDM Extended mode, (8 <= tdm_slots < 16) Basic 
mode, (tdm_slots == 16) TDM Ex-Split
10ch: TDM Ex-Split mode
16ch: Basic Mode

>> for example user asks dai-link0 to playback 2ch audio stream,
>> driver can't determine which mode to work, as it can be Basic mode,
>> Split mode or Ex-Split mode.
> Why do we need to use Basic mode if HW has TDM Split mode connection?
> If user playbacks 2ch audio in such situation,
> we can use TDM Split mode (= only 2ch has sound, other channel has no sound ?)
> user might start to playback for other channels.
> I'm not sure how it works...
>>> 	- BUSIFn connection should be selected on DT
>> since which BUSIFx is used during audio data transfer, is not
>> consideration of user,
>> I think your previous suggestion, (automatically select BUSIFx) makes
>> more sense
> I'm not yet sure detail, but in your idea, does it mean,
> BUSIFx connection might be exchanged runtime ?
no BUSIFx shouldn't be changed during runtime, my idea is BUSIFx can be 
automatically selected
when corresponding dai-link is not active

The reason I added rsnd_ssi_select_busif(io, chan) in rsnd_hw_params() 
in patch ASoC: rsnd: add busif property to dai stream of v2 patch-set,
is because runtime channel is necessary information to determine which 
BUSIFx to select,
(which is mentioned in above)
and at this stage (rsnd_hw_params()), all other control settings 
(register setting, dma address calculation etc)
haven't been done, so corresponding dai-link can be considered to be not 
active at this timing
but maybe you have better suggestion when to automatically select BUSIFx

What is your opinion?

Thanks,
Jiada
> I think BUSIFx connection shouldn't exchanged runtime IMO.
> Otherwise, sound position can't be fixed, and user can't control
> sound, I think...

>
> Best regards
> ---
> Kuninori Morimoto

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ