[alsa-devel] [PATCH v2 6/6] pcm: Update pcm->avail_min with needed_slave_avail_min, after reading unaligned frames

[Channaiah Vanitha (RBEI/ECF3)]

Hello Jaroslav-san, Takashi-san,

Can you please reply your feedback for below mail chain.

Best regards,
Vanitha Channaiah 
RBEI/ECF3

_____________________________________________
From: Channaiah Vanitha (RBEI/ECF3) 
Sent: Tuesday, June 18, 2019 4:45 AM
To: 'Jaroslav Kysela' <perex at perex.cz>; Takashi Iwai <tiwai at suse.de>
Cc: Wischer Timo (ADITG/ESS) <twischer at de.adit-jv.com>; alsa-devel at alsa-project.org
Subject: RE: [alsa-devel] [PATCH v2 6/6] pcm: Update pcm->avail_min with needed_slave_avail_min, after reading unaligned frames


Hello Jaroslav-san, Takashi-san,

> Which hardware exactly? The hw_ptr should be reset when the streaming starts.
> It appears that the problem is specific to the direct plugins only when the period wakeups are a bit different than for the direct hardware access.

Firstly, sorry for late reply.

Issue is seen in RCar Kingfischer H3 es2.0 target.	

The issue was seen during the below commit :
commit 07b7acb51d283d8469696c906b91f1882696a4d4
("ASoC: rsnd: update pointer more accurate")
https://patchwork.kernel.org/patch/9772671/


There could be a non-uniform jitter exists between when interrupt raised [rcar_dmac_isr_channel(), rcar_dmac_isr_channel_thread()]
and the interrupt is processed to read/calculate the DMA position [dma_set_residue()]
This could result in unaligned hw_ptr reported at user-space alsa lib.


Best regards,
Vanitha Channaiah
RBEI/ECF3  

Tel. +91 80 6136-4436 


-----Original Message-----
From: Jaroslav Kysela <mailto:perex at perex.cz> 
Sent: Thursday, May 16, 2019 11:05 PM
To: Channaiah Vanitha (RBEI/ECF3) <mailto:Vanitha.Channaiah at in.bosch.com>; Takashi Iwai <mailto:tiwai at suse.de>
Cc: Wischer Timo (ADITG/ESS) <mailto:twischer at de.adit-jv.com>; mailto:alsa-devel at alsa-project.org
Subject: Re: [alsa-devel] [PATCH v2 6/6] pcm: Update pcm->avail_min with needed_slave_avail_min, after reading unaligned frames

Dne 16. 05. 19 v 19:26 Channaiah Vanitha (RBEI/ECF3) napsal(a):
> Hello  Takashi-san,
> 
>>  Can we avoid adding such a hack in the core code?
>>  Basically the issue is very specific to direct-plugins, so this sort of workaround should be implemented locally there instead.  I guess we can do similar tricks by overriding the calls in the callbacks.
> 
> Issue can be seen without direct plugins too with involvement of only hw which reports unaligned hw ptr. As I have explained in below  detailed description:
> "Also, this issue can be seen without dsnoop plugin, when HW reports unaligned hw_ptr For e.g. "

Which hardware exactly? The hw_ptr should be reset when the streaming starts.
It appears that the problem is specific to the direct plugins only when the period wakeups are a bit different than for the direct hardware access.

						Jaroslav


--
Jaroslav Kysela <mailto:perex at perex.cz>
Linux Sound Maintainer; ALSA Project; Red Hat, Inc.