Dear Kernel maintainers,
I think i'm encountering (for the first time in years !) a regression with the "6.7.arch3-1" kernel (whereas no issues with "6.6.10.arch1-1", on which i reverted).
I'm running a (up-to-date, and non-LTS) ARCHLINUX desktop, on a ASUS B560-I motherboard, with 3 monitors (attached to a 4-HDMI outputs card), plus an audio S/PDIF optic output at motherboard level.
With the latest kernel, the S/PIDF optic output of the motherboard is NOT detected anymore (and i haven't been able to see / find anything in the logs at quick glance, neither journalctl -xe nor dmesg).
Once reverted to 6.6.10, everything is fine again.
For example, in a working situation (6.6.10), i have :
cat /proc/asound/pcm 00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
Whereas while on the latest 6.7 kernel, i only had the 4 HDMI lines (linked to a NVIDIA T600 card, with 4 HDMI outputs) and not the three first ones (attached to the motherboard).
(of course i did several tests with 6.7, reboot, ... without any changes)
Any idea ?
Best regards
-- Serge.
On Tue, Jan 16, 2024 at 09:49:59PM +0100, Serge SIMON wrote:
Dear Kernel maintainers,
I think i'm encountering (for the first time in years !) a regression with the "6.7.arch3-1" kernel (whereas no issues with "6.6.10.arch1-1", on which i reverted).
I'm running a (up-to-date, and non-LTS) ARCHLINUX desktop, on a ASUS B560-I motherboard, with 3 monitors (attached to a 4-HDMI outputs card), plus an audio S/PDIF optic output at motherboard level.
With the latest kernel, the S/PIDF optic output of the motherboard is NOT detected anymore (and i haven't been able to see / find anything in the logs at quick glance, neither journalctl -xe nor dmesg).
Once reverted to 6.6.10, everything is fine again.
For example, in a working situation (6.6.10), i have :
cat /proc/asound/pcm 00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
Whereas while on the latest 6.7 kernel, i only had the 4 HDMI lines (linked to a NVIDIA T600 card, with 4 HDMI outputs) and not the three first ones (attached to the motherboard).
(of course i did several tests with 6.7, reboot, ... without any changes)
Any idea ?
As this is a sound issue, perhaps send this to the linux-sound@vger.kernel.org mailing list (now added).
Any chance you can do a 'git bisect' between 6.6 and 6.7 to track down the issue? Or maybe the sound developers have some things to ask about as there are loads of debugging knobs in sound...
thanks,
greg k-h
Hello,
Any news on this ? Just to say that i tried the 6.7.3 version and i have the exact same problem as described below ("linux-headers-6.7.3.arch1-2-x86_64.pkg.tar.zst" for the exact ARCH package, of course with a system fully up-to-date and rebooted) : no more S/PDIF device detected after reboot (only the monitors are detected, but not anymore the S/PDIF output at motherboard level- which is what i'm using).
Reverting to 6.6.10 does solve the issue, so per what i'm seeing, something has definitely been broken between 6.6.10 and 6.7.0 on that topic.
Is this tracked by a bug somewhere ? Does i have to open one (in addition to these mails) ?
Regards.
On 05.02.24 08:09, Serge SIMON wrote:
Any news on this ?
Apparently not. I added the sound maintainers just to be sure they are aware of this.
Just to say that i tried the 6.7.3 version and i have the exact same problem as described below ("linux-headers-6.7.3.arch1-2-x86_64.pkg.tar.zst" for the exact ARCH package, of course with a system fully up-to-date and rebooted) : no more S/PDIF device detected after reboot (only the monitors are detected, but not anymore the S/PDIF output at motherboard level- which is what i'm using).
Reverting to 6.6.10 does solve the issue, so per what i'm seeing, something has definitely been broken between 6.6.10 and 6.7.0 on that topic.
Unless the sound maintainers come up with something, we most likely need a bisection from you to resolve this.
In case you want to perform a bisection, this guide I'm currently working on might help:
https://www.leemhuis.info/files/misc/How%20to%20bisect%20a%20Linux%20kernel%...
Is this tracked by a bug somewhere ? Does i have to open one (in addition to these mails) ?
No, this thread (for now) is enough.
Ciao, Thorsten
On Wed, Jan 17, 2024 at 6:39 AM Greg KH gregkh@linuxfoundation.org wrote:
On Tue, Jan 16, 2024 at 09:49:59PM +0100, Serge SIMON wrote:
Dear Kernel maintainers,
I think i'm encountering (for the first time in years !) a regression with the "6.7.arch3-1" kernel (whereas no issues with "6.6.10.arch1-1", on which i reverted).
I'm running a (up-to-date, and non-LTS) ARCHLINUX desktop, on a ASUS B560-I motherboard, with 3 monitors (attached to a 4-HDMI outputs card), plus an audio S/PDIF optic output at motherboard level.
With the latest kernel, the S/PIDF optic output of the motherboard is NOT detected anymore (and i haven't been able to see / find anything in the logs at quick glance, neither journalctl -xe nor dmesg).
Once reverted to 6.6.10, everything is fine again.
For example, in a working situation (6.6.10), i have :
cat /proc/asound/pcm 00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
Whereas while on the latest 6.7 kernel, i only had the 4 HDMI lines (linked to a NVIDIA T600 card, with 4 HDMI outputs) and not the three first ones (attached to the motherboard).
(of course i did several tests with 6.7, reboot, ... without any changes)
Any idea ?
As this is a sound issue, perhaps send this to the linux-sound@vger.kernel.org mailing list (now added).
Any chance you can do a 'git bisect' between 6.6 and 6.7 to track down the issue? Or maybe the sound developers have some things to ask about as there are loads of debugging knobs in sound...
thanks,
greg k-h
Hello,
I still encounter this issue (and every month i test the latest kernel, each time with the same results) : - i do have an ASUS B560-I WIFI (ITX) motherboard with a S/PDIF output - everything was working flawlessly until (and including) kernel 6.6.10, and that S/PDIF output was perfectly detected (under GNOME SHELL, etc.) - starting from kernel 6.7.0 (and newest ones, including 6.16.10 tested today) the S/PDIF output it NOT detected anymore at boot time by the kernel (so is not selectable any more under GNOME SHELL or COSMIC, etc.)
With old kernel (example : https://gist.github.com/SR-G/0e86d917716acff0d31cad0365f0b500#file-gistfile1...) :
% cat /proc/asound/pcm
00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
With kernels >= 6.7.0 (example : https://gist.github.com/SR-G/0e86d917716acff0d31cad0365f0b500#file-dmesg-6-1...) :
% cat /proc/asound/pcm 00-03: HDMI 0 : HDMI 0 : playback 1 00-07: HDMI 1 : HDMI 1 : playback 1 00-08: HDMI 2 : HDMI 2 : playback 1 00-09: HDMI 3 : HDMI 3 : playback 1
It seems i'm the only one impacted :(
Who can help me on this topic ?
Thanks in advance and regards.
-- Serge.
On Mon, Feb 5, 2024 at 8:35 AM Linux regression tracking (Thorsten Leemhuis) regressions@leemhuis.info wrote:
On 05.02.24 08:09, Serge SIMON wrote:
Any news on this ?
Apparently not. I added the sound maintainers just to be sure they are aware of this.
Just to say that i tried the 6.7.3 version and i have the exact same problem as described below ("linux-headers-6.7.3.arch1-2-x86_64.pkg.tar.zst" for the exact ARCH package, of course with a system fully up-to-date and rebooted) : no more S/PDIF device detected after reboot (only the monitors are detected, but not anymore the S/PDIF output at motherboard level- which is what i'm using).
Reverting to 6.6.10 does solve the issue, so per what i'm seeing, something has definitely been broken between 6.6.10 and 6.7.0 on that topic.
Unless the sound maintainers come up with something, we most likely need a bisection from you to resolve this.
In case you want to perform a bisection, this guide I'm currently working on might help:
https://www.leemhuis.info/files/misc/How%20to%20bisect%20a%20Linux%20kernel%...
Is this tracked by a bug somewhere ? Does i have to open one (in addition to these mails) ?
No, this thread (for now) is enough.
Ciao, Thorsten
On Wed, Jan 17, 2024 at 6:39 AM Greg KH gregkh@linuxfoundation.org wrote:
On Tue, Jan 16, 2024 at 09:49:59PM +0100, Serge SIMON wrote:
Dear Kernel maintainers,
I think i'm encountering (for the first time in years !) a regression with the "6.7.arch3-1" kernel (whereas no issues with "6.6.10.arch1-1", on which i reverted).
I'm running a (up-to-date, and non-LTS) ARCHLINUX desktop, on a ASUS B560-I motherboard, with 3 monitors (attached to a 4-HDMI outputs card), plus an audio S/PDIF optic output at motherboard level.
With the latest kernel, the S/PIDF optic output of the motherboard is NOT detected anymore (and i haven't been able to see / find anything in the logs at quick glance, neither journalctl -xe nor dmesg).
Once reverted to 6.6.10, everything is fine again.
For example, in a working situation (6.6.10), i have :
cat /proc/asound/pcm 00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
Whereas while on the latest 6.7 kernel, i only had the 4 HDMI lines (linked to a NVIDIA T600 card, with 4 HDMI outputs) and not the three first ones (attached to the motherboard).
(of course i did several tests with 6.7, reboot, ... without any changes)
Any idea ?
As this is a sound issue, perhaps send this to the linux-sound@vger.kernel.org mailing list (now added).
Any chance you can do a 'git bisect' between 6.6 and 6.7 to track down the issue? Or maybe the sound developers have some things to ask about as there are loads of debugging knobs in sound...
thanks,
greg k-h
В Пт, 03/10/2025 в 15:20 +0200, Serge SIMON пишет:
Hello,
I still encounter this issue (and every month i test the latest kernel, each time with the same results) :
- i do have an ASUS B560-I WIFI (ITX) motherboard with a S/PDIF
output
- everything was working flawlessly until (and including) kernel
6.6.10, and that S/PDIF output was perfectly detected (under GNOME SHELL, etc.)
- starting from kernel 6.7.0 (and newest ones, including 6.16.10
tested today) the S/PDIF output it NOT detected anymore at boot time by the kernel (so is not selectable any more under GNOME SHELL or COSMIC, etc.)
With old kernel (example : https://gist.github.com/SR-G/0e86d917716acff0d31cad0365f0b500#file-gistfile1... ) :
% cat /proc/asound/pcm
00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
With kernels >= 6.7.0 (example : https://gist.github.com/SR-G/0e86d917716acff0d31cad0365f0b500#file-dmesg-6-1... ) :
Did you disable intel iGPU somehow? Try to add snd_hda_core.gpu_bind=0 to the kernel cmdline.
В Пт, 03/10/2025 в 17:57 +0300, Alexander Tsoy пишет:
В Пт, 03/10/2025 в 15:20 +0200, Serge SIMON пишет:
Hello,
I still encounter this issue (and every month i test the latest kernel, each time with the same results) :
- i do have an ASUS B560-I WIFI (ITX) motherboard with a S/PDIF
output
- everything was working flawlessly until (and including) kernel
6.6.10, and that S/PDIF output was perfectly detected (under GNOME SHELL, etc.)
- starting from kernel 6.7.0 (and newest ones, including 6.16.10
tested today) the S/PDIF output it NOT detected anymore at boot time by the kernel (so is not selectable any more under GNOME SHELL or COSMIC, etc.)
With old kernel (example : https://gist.github.com/SR-G/0e86d917716acff0d31cad0365f0b500#file-gistfile1... ) :
% cat /proc/asound/pcm
00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
With kernels >= 6.7.0 (example : https://gist.github.com/SR-G/0e86d917716acff0d31cad0365f0b500#file-dmesg-6-1... ) :
Did you disable intel iGPU somehow? Try to add snd_hda_core.gpu_bind=0 to the kernel cmdline.
Ah, I see: ... vfio_pci: add [8086:4c8a[ffffffff:ffffffff]] class 0x000000/00000000 ...
You probably passthrough the GPU to VM so the i915 driver is not loaded on the host. Thus you really need snd_hda_core.gpu_bind=0 kernel option. This is a consequence of the following series: https://lore.kernel.org/all/20231009115437.99976-1-maarten.lankhorst@linux.i...
So i've just tested this, and ... you were definitely correct / on the good track.
Once back with 6.10.16 + that "snd_hda_core.gpu_bind=0" kernel parameter : - my S/PDIF output is detected again (and is working as before) - surprisingly, it seems i'm not able to select the HDMI outputs (of the monitors) under GNOME settings (i now only have the S/PDIF entry) ... however a "cat /proc/asound/pcm" is correct, and is showing everything ... but to be honest, i just don't care (the point was to not use the (crappy) monitor sound outputs)
Extra information : - indeed i was bypassing the iGPU (also activated at BIOS level) to some VM - by the way, i also had the "intel_iommu=on" parameter at kernel level - latest dmesg from today (for reference / just in case) : https://gist.github.com/SR-G/ef22e46e96bfb300c3689cf8ea073098#file-dmesg-6-1... - current "cat /proc/asounc/pcm" is back to : 00-00: ALC1220 Analog : ALC1220 Analog : playback 1 : capture 1 00-01: ALC1220 Digital : ALC1220 Digital : playback 1 00-02: ALC1220 Alt Analog : ALC1220 Alt Analog : capture 1 01-03: HDMI 0 : HDMI 0 : playback 1 01-07: HDMI 1 : HDMI 1 : playback 1 01-08: HDMI 2 : HDMI 2 : playback 1 01-09: HDMI 3 : HDMI 3 : playback 1
So a big thank you for your analysis and for having taken the time to investigate into this !
linux-stable-mirror@lists.linaro.org
-
Alexander Tsoy -
Greg KH -
Linux regression tracking (Thorsten Leemhuis) -
Serge SIMON