BT + Gnome sound control fail after update to kernel 6.7

Cbhihe · 2024-01-18 17:45:19

Hello,
Upon upgrading to from kernel 6.6.9 to 6.7, I lost BT and the sound input control in Gnome settings GUI. Previously everything worked well.
Concretely:
- Bluetooth does not allow sending files over from host to already coupled devices any longer
- the sound input in the Gnome setting GUI is grayed out but it does come alive when I plug in a pair of wired-earphones.

I have looked at recent posts in the "Newbie" as well as in the "Kernel and Hardware" categories [1, 2, 3, 4, ...] with no success. Because the most recent one (first in list above) looked promising, I actually downgraded libpulse, pulseaudio and pulseaudio-bluetooth from 17.0-3 to 16.1-7:

$ sudo pacman -U https://archive.archlinux.org/packages/l/libpulse/libpulse-16.1-7-x86_64.pkg.tar.zst  \
   https://archive.archlinux.org/packages/p/pulseaudio-16.1-7-x86_64.pkg.tar.zst \
   https://archive.archlinux.org/packages/p/pulseaudio-bluetooth-16.1-7-x86_64.pkg.tar.zst

and rebooted (twice), but:

- for pulseaudio:
I keep getting the same grayed out sound input control palette in the Gnome settings GUI, and
- for BT:

$ systemctl status bluetooth
● bluetooth.service - Bluetooth service
     Loaded: loaded (/usr/lib/systemd/system/bluetooth.service; enabled; preset: disabled)
     Active: active (running) since Thu 2024-01-18 17:55:44 CET; 32min ago
       Docs: man:bluetoothd(8)
   Main PID: 665 (bluetoothd)
     Status: "Running"
      Tasks: 1 (limit: 18786)
     Memory: 4.2M (peak: 4.7M)
        CPU: 63ms
     CGroup: /system.slice/bluetooth.service
             └─665 /usr/lib/bluetooth/bluetoothd

Jan 18 17:55:44 HOST bluetoothd[665]: src/plugin.c:plugin_init() System does not support micp plugin
Jan 18 17:55:44 HOST bluetoothd[665]: src/plugin.c:plugin_init() System does not support vcp plugin
Jan 18 17:55:44 HOST bluetoothd[665]: src/plugin.c:plugin_init() System does not support mcp plugin
Jan 18 17:55:44 HOST bluetoothd[665]: src/plugin.c:plugin_init() System does not support bass plugin
Jan 18 17:55:44 HOST bluetoothd[665]: src/plugin.c:plugin_init() System does not support bap plugin
Jan 18 17:55:44 HOST bluetoothd[665]: Bluetooth management interface 1.22 initialized
Jan 18 17:55:44 HOST systemd[1]: Started Bluetooth service.
Jan 18 17:55:44 HOST bluetoothd[665]: Battery Provider Manager created
Jan 18 17:55:44 HOST bluetoothd[665]: Failed to load LTKs for hci0: Invalid Parameters (0x0d)
Jan 18 17:55:44 HOST bluetoothd[665]: Failed to load IRKs for hci0: Invalid Parameters (0x0d)

$ sudo systemctl restart bluetooth

$ systemctl status bluetooth
● bluetooth.service - Bluetooth service
     Loaded: loaded (/usr/lib/systemd/system/bluetooth.service; enabled; preset: disabled)
     Active: active (running) since Thu 2024-01-18 18:30:15 CET; 4s ago
       Docs: man:bluetoothd(8)
   Main PID: 5345 (bluetoothd)
     Status: "Running"
      Tasks: 1 (limit: 18786)
     Memory: 968.0K (peak: 1.4M)
        CPU: 43ms
     CGroup: /system.slice/bluetooth.service
             └─5345 /usr/lib/bluetooth/bluetoothd

Jan 18 18:30:15 HOST bluetoothd[5345]: src/plugin.c:plugin_init() System does not support bap plugin
Jan 18 18:30:15 HOST bluetoothd[5345]: Bluetooth management interface 1.22 initialized
Jan 18 18:30:15 HOST bluetoothd[5345]: src/adapter.c:reset_adv_monitors_complete() Failed to reset Adv Monitors: Failed (0x03)
Jan 18 18:30:15 HOST bluetoothd[5345]: Battery Provider Manager created
Jan 18 18:30:15 HOST bluetoothd[5345]: Failed to clear UUIDs: Failed (0x03)
Jan 18 18:30:15 HOST bluetoothd[5345]: Failed to add UUID: Failed (0x03)
Jan 18 18:30:15 HOST bluetoothd[5345]: Failed to add UUID: Failed (0x03)
Jan 18 18:30:15 HOST bluetoothd[5345]: Failed to add UUID: Failed (0x03)
Jan 18 18:30:15 HOST bluetoothd[5345]: Failed to load LTKs for hci0: Invalid Parameters (0x0d)
Jan 18 18:30:15 HOST bluetoothd[5345]: Failed to load IRKs for hci0: Invalid Parameters (0x0d)

which is also reflected in the output of `dmesg`.

Here is the output of `journalctl -b1`.

Last edited by Cbhihe (2024-02-02 21:14:49)

cgb_spender · 2024-01-19 01:30:52

Hi, I use timeshift in such cases, and then wait for 1-2 months before updating again.

Cbhihe · 2024-01-22 08:11:10

@cgp_spender: not an option.

This host relies on Bluetooth to close some complicated feedback loop.
As a stop-gap I am currently using another near identical hardware with Ubuntu running on it.
I would have liked the two platforms - Arch and Ubuntu - to run on exactly identical hardware to try to conduct a kind of comparative troubleshooting, but the two machines were bought 1.5 months apart so I cannot be sure they are identical even if `hwinfo` tells me they are.

Back to this host running an up to date Arch system with X11, Xorg and Gnome DM, I tried removing modules btintel and btusb, rebooting, then reinstalling them and rebooting again. But the problem persists.

Last edited by Cbhihe (2024-01-22 08:12:04)

V1del · 2024-01-22 11:49:01

Does it still work on the LTS kernel? if it does you have a regression in 6.7 and would have to bisect. As far as recent bisection kernels goes you might be able to grab some from this unrelated issue: https://bbs.archlinux.org/viewtopic.php?id=292086

Also a potential more relevant candidate for downgrading if it isn't the kernel directly is linux-firmware.

If a kernel downgrade/LTS doesn't work, sometimes the BT state gets corrupted and you should clear out /var/lib/bluetooth and pair the device from scratch again.

Cbhihe · 2024-01-29 18:54:06

I have not tried my luck yet with linux-lts (v6.6.14-1), but I am now on 6.7.2-arch1-1 and from what I see the issue persists from 6.7.0. while the issues were not there with 6.6.9-arch1, and I can't remember whether they were with 6.6.10-arch1-1.

Looking at /var/log/pacman.log though I noticed (maybe unrelated because 30 months old) DKMS build errors on the realtek module r8152, that started with kernel 5.6.15 module building, as shown below:

[2020-05-29T17:43:14+0200] [ALPM-SCRIPTLET] ==> dkms install r8152/2.10.0 -k 5.6.15-arch1-1
[2020-05-29T17:43:16+0200] [ALPM-SCRIPTLET] Error! Bad return status for module build on kernel: 5.6.15-arch1-1 (x86_64)
[2020-05-29T17:43:16+0200] [ALPM-SCRIPTLET] Consult /var/lib/dkms/r8152/2.10.0/build/make.log for more information.
[2020-05-29T17:43:16+0200] [ALPM-SCRIPTLET] ==> Warning, `dkms install r8152/2.10.0 -k 5.6.15-arch1-1' returned 10

That same error has propagated until now in that Archlinux host. The latest log when upgrading from kernel 6.7.1.arch1-1 to 6.7.2-arch1-1 follows:

> cat /var/lib/dkms/r8152/2.16.3.20221209/build/make.log
DKMS make.log for r8152-2.16.3.20221209 for kernel 6.7.2-arch1-1 (x86_64)
Mon 29 Jan 2024 11:23:22 CET
make: Entering directory '/usr/lib/modules/6.7.2-arch1-1/build'
  CC [M]  /var/lib/dkms/r8152/2.16.3.20221209/build/src/r8152.o
/var/lib/dkms/r8152/2.16.3.20221209/build/src/r8152.c: In function ‘r8152_csum_workaround’:
/var/lib/dkms/r8152/2.16.3.20221209/build/src/r8152.c:2074:24: error: implicit declaration of function ‘skb_gso_segment’; did you mean ‘skb_gso_reset’? [-Werror=implicit-function-declaration]
 2074 |                 segs = skb_gso_segment(skb, features);
      |                        ^~~~~~~~~~~~~~~
      |                        skb_gso_reset
/var/lib/dkms/r8152/2.16.3.20221209/build/src/r8152.c:2074:22: warning: assignment to ‘struct sk_buff *’ from ‘int’ makes pointer from integer without a cast [-Wint-conversion]
 2074 |                 segs = skb_gso_segment(skb, features);
      |                      ^
cc1: some warnings being treated as errors
make[2]: *** [scripts/Makefile.build:243: /var/lib/dkms/r8152/2.16.3.20221209/build/src/r8152.o] Error 1
make[1]: *** [/usr/lib/modules/6.7.2-arch1-1/build/Makefile:1911: /var/lib/dkms/r8152/2.16.3.20221209/build/src] Error 2
make: *** [Makefile:234: __sub-make] Error 2
make: Leaving directory '/usr/lib/modules/6.7.2-arch1-1/build'

V1del · 2024-01-29 22:36:22

That seems related to an ethernet adapter? In any case if this is that old, chances are the relevant code got upstreamed and your potentially masking current developments with something. I suggest you remove it and check whether the HW is supported by in tree modules by now (... 5.6 is like... 4 years old or so)

Cbhihe · 2024-02-02 20:05:03

V1del wrote:

That seems related to an ethernet adapter? In any case if this is that old, chances are the relevant code got upstreamed and your potentially masking current developments with something. I suggest you remove it and check whether the HW is supported by in tree modules by now (... 5.6 is like... 4 years old or so)

More like 2.5years counting from its stable `core` release. ;-)

I updated the AUR r8152-dkms module and also went from kernel 6.7.2-arch1-1 to 6.7.3-arch1-1. After rebooting the module installation error disappeared:

$ cat /var/lib/dkms/r8152/2.17.1.20230903/6.7.3-arch1-1/x86_64/log/make.log
DKMS make.log for r8152-2.17.1.20230903 for kernel 6.7.3-arch1-1 (x86_64)
Fri 02 Feb 2024 19:27:15 CET
make: Entering directory '/usr/lib/modules/6.7.3-arch1-1/build'
  CC [M]  /var/lib/dkms/r8152/2.17.1.20230903/build/src/r8152.o
  MODPOST /var/lib/dkms/r8152/2.17.1.20230903/build/src/Module.symvers
  CC [M]  /var/lib/dkms/r8152/2.17.1.20230903/build/src/r8152.mod.o
  LD [M]  /var/lib/dkms/r8152/2.17.1.20230903/build/src/r8152.ko
  BTF [M] /var/lib/dkms/r8152/2.17.1.20230903/build/src/r8152.ko
make: Leaving directory '/usr/lib/modules/6.7.3-arch1-1/build'

I also manually removed the device pairing files at /var/lib/bluetooth/ and rebooted / re-paired HOST to a couple external devices.

Issue still present and with message "The name is not activatable." (yes, "activatable" !!) when trying to transfer a small file from HOST to Android terminal. From Android terminal to HOST all transfers abort quietly (no message).

Last edited by Cbhihe (2024-02-02 20:06:53)

Cbhihe · 2024-02-02 20:25:09

Re-re-rechecking the Bluetooth wiki, I just saw in the revision history that exactly a week ago a small note was added:

Since this commit, bluez-obex and bluez-mesh have been separated from bluez. Therefore, if you plan to transfer files over Bluetooth, bluez-obex needs to be installed.

where THIS in "this commit" links to:

cur prev 20:38, 25 January 2024 Nebulizer9782 talk contribs m 52,734 bytes +318 Added a note stating that the bluez package had two components split off from it. Those components were then packaged as bluez-obex and bluez-mesh and Obex provides file transfer over Bluetooth.

# pacman -S bluez-obex

... and transfers are back to normal. (I am left wondering why the change specified in the commit and its workaround did not make it to the Arch-announce list ...)

Besides I still get:

$ systemctl status bluetooth
● bluetooth.service - Bluetooth service
     Loaded: loaded (/usr/lib/systemd/system/bluetooth.service; enabled; preset: disabled)
     Active: active (running) since Fri 2024-02-02 21:32:53 CET; 7min ago
       Docs: man:bluetoothd(8)
   Main PID: 586 (bluetoothd)
     Status: "Running"
      Tasks: 1 (limit: 18786)
     Memory: 4.2M (peak: 4.7M)
        CPU: 70ms
     CGroup: /system.slice/bluetooth.service
             └─586 /usr/lib/bluetooth/bluetoothd

Feb 02 21:32:53 HOST bluetoothd[586]: src/plugin.c:plugin_init() System does not support micp plugin
Feb 02 21:32:53 HOST bluetoothd[586]: src/plugin.c:plugin_init() System does not support vcp plugin
Feb 02 21:32:53 HOST bluetoothd[586]: src/plugin.c:plugin_init() System does not support mcp plugin
Feb 02 21:32:53 HOST bluetoothd[586]: src/plugin.c:plugin_init() System does not support bass plugin
Feb 02 21:32:53 HOST bluetoothd[586]: src/plugin.c:plugin_init() System does not support bap plugin
Feb 02 21:32:53 HOST bluetoothd[586]: Bluetooth management interface 1.22 initialized
Feb 02 21:32:53 HOST systemd[1]: Started Bluetooth service.
Feb 02 21:32:54 HOST bluetoothd[586]: Battery Provider Manager created
Feb 02 21:32:54 HOST bluetoothd[586]: Failed to load LTKs for hci0: Invalid Parameters (0x0d)
Feb 02 21:32:54 HOST bluetoothd[586]: Failed to load IRKs for hci0: Invalid Parameters (0x0d)

Apparently "plugins" micp, vcp, mcp, bass, bap are related to Bluetooth's Generic Audio Framework (GAF) experimental features and enabling those in /etc/bluetooth/main.conf (following the Wiki) made related messages disappear. The Gnome Sound control window's Input control cursor is still grayed out though, and as before, I need to plug in an audio jack for it to come to life.

Just for reference and for those interested in knowing what the plugins' acronyms stand for:
- Stream Control is implemented by the Basic Audio Profile (BAP)
- Rendering and capture control is implemented by the Volume Control Profile (VCP) and Microphone Input Control Profile (MICP)
- Content control is implemented by the Call Control Profile (CCP) and Media Control Profile (MCP).

At this point, restarting the BT service, I get:

$ sudo systemctl restart bluetooth
$ 
$ systemctl status bluetooth
● bluetooth.service - Bluetooth service
     Loaded: loaded (/usr/lib/systemd/system/bluetooth.service; enabled; preset: disabled)
     Active: active (running) since Fri 2024-02-02 22:21:07 CET; 7s ago
       Docs: man:bluetoothd(8)
   Main PID: 8960 (bluetoothd)
     Status: "Running"
      Tasks: 1 (limit: 18786)
     Memory: 1008.0K (peak: 1.4M)
        CPU: 58ms
     CGroup: /system.slice/bluetooth.service
             └─8960 /usr/lib/bluetooth/bluetoothd

Feb 02 22:21:07 HOST bluetoothd[8960]: Bluetooth management interface 1.22 initialized
Feb 02 22:21:07 HOST bluetoothd[8960]: src/adapter.c:reset_adv_monitors_complete() Failed to reset Adv Monitors: Failed (0x03)
Feb 02 22:21:07 HOST bluetoothd[8960]: Battery Provider Manager created
Feb 02 22:21:07 HOST bluetoothd[8960]: Failed to clear UUIDs: Failed (0x03)
Feb 02 22:21:07 HOST bluetoothd[8960]: Failed to add UUID: Failed (0x03)
Feb 02 22:21:07 HOST bluetoothd[8960]: Failed to add UUID: Failed (0x03)
Feb 02 22:21:07 HOST bluetoothd[8960]: Failed to add UUID: Failed (0x03)
Feb 02 22:21:07 HOST bluetoothd[8960]: Adv Monitor Manager created with supported features:0x00000001, enabled features:0x00000001, max number of supported monitors:32, max number of supported patterns:16 
Feb 02 22:21:07 HOST bluetoothd[8960]: Failed to load LTKs for hci0: Invalid Parameters (0x0d)
Feb 02 22:21:07 HOST bluetoothd[8960]: Failed to load IRKs for hci0: Invalid Parameters (0x0d)

Last edited by Cbhihe (2024-02-04 21:34:14)

Cbhihe · 2024-02-04 12:22:10

BT service daemon is seems to choke on reconnects when the service is restarted. Trying to understand why, I looked at /etc/bluetooth/{input,main}.conf. Both are set to default values:

...
[Policy]
#
# The ReconnectUUIDs defines the set of remote services that should try
# to be reconnected to in case of a link loss (link supervision
# timeout). The policy plugin should contain a sane set of values by
# default, but this list can be overridden here. By setting the list to
# empty the reconnection feature gets disabled.
#ReconnectUUIDs=00001112-0000-1000-8000-00805f9b34fb,0000111f-0000-1000-8000-00805f9b34fb,0000110a-0000-1000-8000-00805f9b34fb,0000110b-0000-1000-8000-00805f9b34fb
...

So I don't know whether I am going down the wrong rabbit hole here, but the "Failed to add UUID:.." part of the status message above seems related...
Found nothing on the plugins related messages though.
Any nudge welcome, perhaps just to put me back on track, in case this is all in my head.

Last edited by Cbhihe (2024-02-07 08:44:46)

Arch Linux

#1 2024-01-18 17:45:19

BT + Gnome sound control fail after update to kernel 6.7

#2 2024-01-19 01:30:52

Re: BT + Gnome sound control fail after update to kernel 6.7

#3 2024-01-22 08:11:10

Re: BT + Gnome sound control fail after update to kernel 6.7

#4 2024-01-22 11:49:01

Re: BT + Gnome sound control fail after update to kernel 6.7

#5 2024-01-29 18:54:06

Re: BT + Gnome sound control fail after update to kernel 6.7

#6 2024-01-29 22:36:22

Re: BT + Gnome sound control fail after update to kernel 6.7

#7 2024-02-02 20:05:03

Re: BT + Gnome sound control fail after update to kernel 6.7

#8 2024-02-02 20:25:09

Re: BT + Gnome sound control fail after update to kernel 6.7

#9 2024-02-04 12:22:10

Re: BT + Gnome sound control fail after update to kernel 6.7

Board footer