Re: Sound issues with the 5.10.x kernel (alsa)

Re: Sound issues with the 5.10.x kernel (alsa)

[Stefan Wahren]

Hi Diederik,

Am 08.02.21 um 12:47 schrieb Diederik de Haas:
> Hi,
>
> This is basically me forwarding the bug I reported on Debian's BTS:
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=978025
>
> TL;DR: I have a RPi 3B+ running pure Debian Bullseye arm64 (~from 
> raspi.debian.net), named rpi-mpd, connected via HDMI cable to my AV Receiver.
can you please confirm that the bcm2835-audio driver causing the issues?
> Playing music worked fine with kernel 5.9, but with 5.10.2-1 the music 
> (quality) became not good to quite horribly, because of (static) noise and 
> distortion.
> With kernel version 5.10.9 (linux-image-5.10.0-2-arm64) it all seemed fixed, 
> but returned with 5.10.12 (linux-image-5.10.0-3-arm64) and is also present 
> with 5.10.13.

I cannot explain why 5.10.9 works for you, but we had multiple
regressions in 5.10. Currently i cannot see any of the fixes by Phil
Elwell in linux-stable. Maybe they won't apply and needs to be backport
manually.

Just for reference here are the revelant patches:

https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/staging/vc04_services?h=next-20210205&id=96ae327678eceabf455b11a88ba14ad540d4b046

https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/staging/vc04_services?h=next-20210205&id=88753cc19f087abe0d39644b844e67a59cfb5a3d

Could you please try?

Best regards
Stefan

>
> I can consistently reproduce this issue, rebooting into 5.10.9 and playing 
> music is fine again, so I *think* this is not a local issue.
> This ML is the closest I can think of as 'upstream', so hopefully more insight  
> can be accomplished here ... and a fix.
>
> Cheers,
>   Diederik
>
> _______________________________________________
> linux-rpi-kernel mailing list
> linux-rpi-kernel@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-rpi-kernel

_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Diederik de Haas]

[-- Attachment #1.1: Type: text/plain, Size: 1617 bytes --]

On maandag 8 februari 2021 13:22:56 CET Stefan Wahren wrote:
> > TL;DR: I have a RPi 3B+ running pure Debian Bullseye arm64 (~from
> > raspi.debian.net), named rpi-mpd, connected via HDMI cable to my AV
> > Receiver.
> can you please confirm that the bcm2835-audio driver causing the issues?

How can I confirm that?

> > Playing music worked fine with kernel 5.9, but with 5.10.2-1 the music
> > (quality) became not good to quite horribly, because of (static) noise and
> > distortion.
> > With kernel version 5.10.9 (linux-image-5.10.0-2-arm64) it all seemed
> > fixed, but returned with 5.10.12 (linux-image-5.10.0-3-arm64) and is also
> > present with 5.10.13.
> 
> I cannot explain why 5.10.9 works for you, but we had multiple
> regressions in 5.10. Currently i cannot see any of the fixes by Phil
> Elwell in linux-stable. Maybe they won't apply and needs to be backport
> manually.

FTR: All the kernel versions I mentioned are the Debian kernel versions (and 
thus also Debian's config).

> Just for reference here are the revelant patches:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/d
> rivers/staging/vc04_services?h=next-20210205&id=96ae327678eceabf455b11a88ba1
> 4ad540d4b046
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/d
> rivers/staging/vc04_services?h=next-20210205&id=88753cc19f087abe0d39644b844e
> 67a59cfb5a3d
> 
> Could you please try?

I'll try. But it may take a while as I don't know how to (properly) do that.
And I'm guessing that building on a RPi3B+ isn't exactly fast. But I assume 
I'm able to 'overcome' all that.

[-- Attachment #1.2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 169 bytes --]

_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Diederik de Haas]

[-- Attachment #1.1: Type: text/plain, Size: 3518 bytes --]

On maandag 8 februari 2021 15:00:26 CET Ryutaroh Matsumoto wrote:
> > With kernel version 5.10.9 (linux-image-5.10.0-2-arm64) it all seemed
> > fixed, but returned with 5.10.12 (linux-image-5.10.0-3-arm64) and is also
> > present with 5.10.13.
> 
> The difference in Debian kernel 5.10.9 was that vc4.ko was disabled by
> mistake. I guess that module_blacklist=vc4 in kernel command line
> suppresses the symptom.
> All other Debian kernel 5.10.x have vc4.ko enabled in arm64.

Indeed, you're correct! Thanks.
I haven't tried blacklisting it, but IIUC that means it doesn't work in 5.10 
with vc4 at all (on arm64 at least).

I've now setup another RPi3B+ ~ the same as my rpi-mpd, but / moved to a 
PiDrive and created 52Gi swap partition to build the upstream kernel (+config) 
and subsequently one with the requested patches.

diederik@rpi-mpd:~$ git diff /boot/config-5.10.0-2-arm64 /boot/config-5.10.0-3-
arm64
diff --git a/boot/config-5.10.0-2-arm64 b/boot/config-5.10.0-3-arm64
index 35d6f98..e8eafc9 100644
--- a/boot/config-5.10.0-2-arm64
+++ b/boot/config-5.10.0-3-arm64
@@ -1,6 +1,6 @@
 #
 # Automatically generated file; DO NOT EDIT.
-# Linux/arm64 5.10.9 Kernel Configuration
+# Linux/arm64 5.10.13 Kernel Configuration
 #
 CONFIG_CC_VERSION_TEXT="gcc-10 (Debian 10.2.1-6) 10.2.1 20210110"
 CONFIG_CC_IS_GCC=y
@@ -23,7 +23,7 @@ CONFIG_INIT_ENV_ARG_LIMIT=32
 # CONFIG_COMPILE_TEST is not set
 CONFIG_LOCALVERSION=""
 # CONFIG_LOCALVERSION_AUTO is not set
-CONFIG_BUILD_SALT="5.10.0-2-arm64"
+CONFIG_BUILD_SALT="5.10.0-3-arm64"
 CONFIG_DEFAULT_INIT=""
 CONFIG_DEFAULT_HOSTNAME="(none)"
 CONFIG_SWAP=y
@@ -737,7 +737,7 @@ CONFIG_CRYPTO_AES_ARM64_CE_BLK=m
 CONFIG_CRYPTO_AES_ARM64_NEON_BLK=m
 CONFIG_CRYPTO_CHACHA20_NEON=m
 CONFIG_CRYPTO_POLY1305_NEON=m
-# CONFIG_CRYPTO_NHPOLY1305_NEON is not set
+CONFIG_CRYPTO_NHPOLY1305_NEON=m
 CONFIG_CRYPTO_AES_ARM64_BS=m
 
 #
@@ -3339,7 +3339,7 @@ CONFIG_USB_IPHETH=m
 CONFIG_USB_SIERRA_NET=m
 CONFIG_USB_VL600=m
 CONFIG_USB_NET_CH9200=m
-# CONFIG_USB_NET_AQC111 is not set
+CONFIG_USB_NET_AQC111=m
 CONFIG_WLAN=y
 # CONFIG_WIRELESS_WDS is not set
 CONFIG_WLAN_VENDOR_ADMTEK=y
@@ -5937,7 +5937,6 @@ CONFIG_DRM_AMDGPU_USERPTR=y
 # Display Engine Configuration
 #
 CONFIG_DRM_AMD_DC=y
-CONFIG_DRM_AMD_DC_DCN=y
 CONFIG_DRM_AMD_DC_HDCP=y
 CONFIG_DRM_AMD_DC_SI=y
 # end of Display Engine Configuration
@@ -6097,7 +6096,8 @@ CONFIG_DRM_DW_HDMI_I2S_AUDIO=m
 CONFIG_DRM_DW_MIPI_DSI=m
 # end of Display Interface Bridges
 
-# CONFIG_DRM_VC4 is not set
+CONFIG_DRM_VC4=m
+CONFIG_DRM_VC4_HDMI_CEC=y
 # CONFIG_DRM_ETNAVIV is not set
 # CONFIG_DRM_ARCPGU is not set
 CONFIG_DRM_HISI_HIBMC=m
@@ -9531,8 +9531,6 @@ CONFIG_CRYPTO_WP512=m
 #
 CONFIG_CRYPTO_AES=m
 CONFIG_CRYPTO_AES_TI=m
-CONFIG_CRYPTO_ANUBIS=m
-CONFIG_CRYPTO_ARC4=m
 CONFIG_CRYPTO_BLOWFISH=m
 CONFIG_CRYPTO_BLOWFISH_COMMON=m
 CONFIG_CRYPTO_CAMELLIA=m
@@ -9541,13 +9539,10 @@ CONFIG_CRYPTO_CAST5=m
 CONFIG_CRYPTO_CAST6=m
 CONFIG_CRYPTO_DES=m
 CONFIG_CRYPTO_FCRYPT=m
-CONFIG_CRYPTO_KHAZAD=m
 CONFIG_CRYPTO_SALSA20=m
 CONFIG_CRYPTO_CHACHA20=m
-CONFIG_CRYPTO_SEED=m
 CONFIG_CRYPTO_SERPENT=m
 # CONFIG_CRYPTO_SM4 is not set
-CONFIG_CRYPTO_TEA=m
 CONFIG_CRYPTO_TWOFISH=m
 CONFIG_CRYPTO_TWOFISH_COMMON=m
 
@@ -9577,7 +9572,7 @@ CONFIG_CRYPTO_USER_API_SKCIPHER=m
 CONFIG_CRYPTO_USER_API_RNG=m
 # CONFIG_CRYPTO_USER_API_RNG_CAVP is not set
 CONFIG_CRYPTO_USER_API_AEAD=m
-CONFIG_CRYPTO_USER_API_ENABLE_OBSOLETE=y
+# CONFIG_CRYPTO_USER_API_ENABLE_OBSOLETE is not set
 # CONFIG_CRYPTO_STATS is not set
 CONFIG_CRYPTO_HASH_INFO=y


[-- Attachment #1.2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 169 bytes --]

_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Diederik de Haas]

[-- Attachment #1.1: Type: text/plain, Size: 1215 bytes --]

On maandag 8 februari 2021 13:22:56 CET Stefan Wahren wrote:
> Currently i cannot see any of the fixes by Phil
> Elwell in linux-stable. Maybe they won't apply and needs to be backport
> manually.
> 
> Just for reference here are the revelant patches:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/d
> rivers/staging/vc04_services?h=next-20210205&id=96ae327678eceabf455b11a88ba1
> 4ad540d4b046
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/d
> rivers/staging/vc04_services?h=next-20210205&id=88753cc19f087abe0d39644b844e
> 67a59cfb5a3d
> 
> Could you please try?

I've now successfully build and booted a 5.10.16 kernel 'pristine' and with 
the above mentioned patches.
In the 'pristine' variant, I had the same bad/horrible quality as I 
experienced with Debian's 5.10 kernels.
The 'patched' variant seems to be a tad better, but didn't completely solve 
the audio problem. There's still some noise intermixed with the music and 
other artifacts, but I'm inclined to think it's not as horrible as before.

I'll do some more hearing tests tomorrow when I'm fresh again, but figured I'd 
share these preliminary results already.

Cheers,
  Diederik

[-- Attachment #1.2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 169 bytes --]

_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Diederik de Haas]

[-- Attachment #1.1: Type: text/plain, Size: 2838 bytes --]

On dinsdag 16 februari 2021 01:12:12 CET Diederik de Haas wrote:
> On maandag 8 februari 2021 13:22:56 CET Stefan Wahren wrote:
> > Currently i cannot see any of the fixes by Phil
> > Elwell in linux-stable. Maybe they won't apply and needs to be backport
> > manually.
> > 
> > Just for reference here are the revelant patches:
> > 
> > https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit
> > /d
> > rivers/staging/vc04_services?h=next-20210205&id=96ae327678eceabf455b11a88
> > ba1 4ad540d4b046
> > 
> > https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit
> > /d
> > rivers/staging/vc04_services?h=next-20210205&id=88753cc19f087abe0d39644b8
> > 44e 67a59cfb5a3d
> > 
> > Could you please try?

I've cherry-picked both patches on top of 5.10.16 and they apply cleanly.
 
> I've now successfully build and booted a 5.10.16 kernel 'pristine' and with
> the above mentioned patches.
> In the 'pristine' variant, I had the same bad/horrible quality as I
> experienced with Debian's 5.10 kernels.
> The 'patched' variant seems to be a tad better, but didn't completely solve
> the audio problem. There's still some noise intermixed with the music and
> other artifacts, but I'm inclined to think it's not as horrible as before.
> 
> I'll do some more hearing tests tomorrow when I'm fresh again, but figured
> I'd share these preliminary results already.

It may have been wishful thinking. 
If there is a difference then isn't big like night and day that would make 
sound play properly. 
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=935456#33 also mentions
that digital audio is not working on arm64, but does work with armhf.

What is bloody annoying is that the sequence of the sound cards keep 
switching, which in turn means I have to check the config and enable or 
disable the settings in /etc/asound.conf to make ALSA use the ALSA card.
The net result is that instead of being able to compare the sounds within 
(say) 1 minute, it takes 3 minutes, 'just' too long to (easily) compare.

diederik@rpi-mpd:~$ aplay -l
**** List of PLAYBACK Hardware Devices ****
card 0: vc4hdmi [vc4-hdmi], device 0: MAI PCM vc4-hdmi-hifi-0 [MAI PCM vc4-hdmi-hifi-0]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: ALSA [bcm2835 ALSA], device 0: bcm2835 ALSA [bcm2835 ALSA]
  Subdevices: 6/7
  Subdevice #0: subdevice #0
  Subdevice #1: subdevice #1
  Subdevice #2: subdevice #2
  Subdevice #3: subdevice #3
  Subdevice #4: subdevice #4
  Subdevice #5: subdevice #5
  Subdevice #6: subdevice #6
card 1: ALSA [bcm2835 ALSA], device 1: bcm2835 IEC958/HDMI [bcm2835 IEC958/HDMI]
  Subdevices: 1/1
  Subdevice #0: subdevice #0

My rpi-mpd device is connected to my multichannel audio capable AV receiver. 
Is one of these cards capable of just passing through the bitstream? 
If so, which. And how?


[-- Attachment #1.2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 169 bytes --]

_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Ryutaroh Matsumoto]

Hi all,

From: Stefan Wahren <stefan.wahren@i2se.com>
Subject: Re: Sound issues with the 5.10.x kernel (alsa)
Date: Mon, 8 Feb 2021 13:22:56 +0100
>> This is basically me forwarding the bug I reported on Debian's BTS:
>> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=978025
>> TL;DR: I have a RPi 3B+ running pure Debian Bullseye arm64 (~from 
>> raspi.debian.net), named rpi-mpd, connected via HDMI cable to my AV Receiver.
> can you please confirm that the bcm2835-audio driver causing the issues?

I think the root cause of this issue is that both vc4.ko and snd_bcm2835.ko
try to provide ALSA sinks to HDMI audio outputs from RPi.
Why do the two drivers provide the same functionality for the same device?
It seems nonsense.
There must be some coordination between vc4.ko and snd_bcm2835.ko
on who provides ALSA sinks of RPi HDMI output.

The non-coordination that both drivers provide different ALSA sinks of the same
device already causes another symptom as
http://lists.infradead.org/pipermail/linux-rpi-kernel/2021-March/008077.html

Best regards, Ryutaroh Matsumoto
_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Ryutaroh Matsumoto]

> I think the root cause of this issue is that both vc4.ko and snd_bcm2835.ko
> try to provide ALSA sinks to HDMI audio outputs from RPi.
> Why do the two drivers provide the same functionality for the same device?
> It seems nonsense.
> There must be some coordination between vc4.ko and snd_bcm2835.ko
> on who provides ALSA sinks of RPi HDMI output.

Giving "enable_hdmi=0" to snd_bcm2835.ko does not improve the situation,
while module_blacklisting snd_bcm2835 suppresses all the symptoms reported.
snd_bcm2835.ko in 5.10.24 seems having a bug ignoring "enable_hdmi=0".

Best regards, Ryutaroh
_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Stefan Wahren]

Hi,

Am 28.03.21 um 07:58 schrieb Ryutaroh Matsumoto:
>> I think the root cause of this issue is that both vc4.ko and snd_bcm2835.ko
>> try to provide ALSA sinks to HDMI audio outputs from RPi.
>> Why do the two drivers provide the same functionality for the same device?
>> It seems nonsense.
>> There must be some coordination between vc4.ko and snd_bcm2835.ko
>> on who provides ALSA sinks of RPi HDMI output.
> Giving "enable_hdmi=0" to snd_bcm2835.ko does not improve the situation,
> while module_blacklisting snd_bcm2835 suppresses all the symptoms reported.
> snd_bcm2835.ko in 5.10.24 seems having a bug ignoring "enable_hdmi=0".
i assume you set this option in the config.txt? This shouldn't have any
affect for the mainline kernel / DT.
>
> Best regards, Ryutaroh
_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Ryutaroh Matsumoto]

Hi Stefan, thank you for your response.

> i assume you set this option in the config.txt? This shouldn't have any
> affect for the mainline kernel / DT.

I am aware of that...
I did "snd_bcm2835 enable_hdmi=0" in /etc/modules.
"modinfo snd_bcm2835" shows as below. Doesn't it indicate snd_bcm2835 having
an option enable_hdmi???

It seems that the contention of grabbing HDMI audio output between vc4.ko
and snd_bcm2835.ko can be addressed if snd_bcm2835.ko honors the
option enable_hdmi with its default =0, but currently enable_hdmi seems
having no effect in behavior of snd_bcm2835...

# modinfo snd_bcm2835
filename:       /lib/modules/5.10.0-5-rt-arm64/kernel/drivers/staging/vc04_services/bcm2835-audio/snd-bcm2835.ko
alias:          platform:bcm2835_audio
license:        GPL
description:    Alsa driver for BCM2835 chip
author:         Dom Cobley
depends:        snd-pcm,vchiq,snd
staging:        Y
intree:         Y
name:           snd_bcm2835
vermagic:       5.10.0-5-rt-arm64 SMP preempt_rt mod_unload modversions aarch64
sig_id:         PKCS#7
signer:         Debian Secure Boot CA
sig_key:        4B:6E:F5:AB:CA:66:98:25:17:8E:05:2C:84:66:7C:CB:C0:53:1F:8C
sig_hashalgo:   sha256
signature:      omitted, too long.
parm:           force_bulk:Force use of vchiq bulk for audio (bool)
parm:           enable_hdmi:Enables HDMI virtual audio device (bool)
parm:           enable_headphones:Enables Headphones virtual audio device (bool)
parm:           enable_compat_alsa:Enables ALSA compatibility virtual audio device (bool)
parm:           num_channels:Number of audio channels (default: 8) (int)

Best regards, Ryutaroh
_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Re: Sound issues with the 5.10.x kernel (alsa)

[Ryutaroh Matsumoto]

>> i assume you set this option in the config.txt? This shouldn't have any
>> affect for the mainline kernel / DT.
> I am aware of that...

I include my config.txt on RPi4B for reference...
arm_64bit=1
enable_uart=1
upstream_kernel=1
kernel=vmlinuz-5.10.0-5-rt-arm64
initramfs initrd.img-5.10.0-5-rt-arm64
disable_fw_kms_setup=1
disable_overscan=1
hdmi_enable_4kp60=1

Best regards, Ryutaroh Matsumoto
_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel