Takashi and Kailang,

I did a few additional experiments. Here is what I found:

1. If I connect headphones and ALWAYS keep them connected, then sound does not disappear even after a warm reboot.

2. The moment I disconnect headphones and do a warm reboot, sound is not going to come back. At that point, connecting or disconnecting headphones, rebooting is not going to change anything (i.e., recover the sound) until I fully power off the computer.

3. After reverting the commit, the above problems disappear for speakers, and no additional issues with headphones are observed.

4. Whether I revert or do not revert the above-mentioned commit, the headphones are sometime flaky if I already have an application running that uses sound. I have to close that application, disconnect and connect headphones again. Otherwise, the speaker does not get disconnected. The headphones seem to be even more flaky when NOT reverting the above commit - unless I booted the system with the headphones connected, the speaker will never get disconnected.

Sincerely, Ruslan

On Monday, May 12, 2025 at 11:07:02 AM EDT, Takashi Iwai tiwai@suse.de wrote:

On Mon, 12 May 2025 16:56:17 +0200,

Ruslan Nikolaev wrote:

Hi,

Forwarding my regression bug report for HP 15-fc000 / ALC236 (ALC256 code path).

#regzbot introduced: v6.12.2 https://bugzilla.kernel.org/show_bug.cgi?id=220109

Sincerely, Ruslan

Forwarded Message (see logs in the bug report above):

There was a similar bug in the past (Bug 217440), which was fixed for this laptop. (I believe the code path is for ALC256, see below.) The same issue is occurring again as of kernel v.6.12.2. The symptoms are very similar - initially audio works but after a warm reboot, the audio completely disappears until the computer is powered off (there is no audio output at all).

The issue is caused by a different change now. By bisecting different kernel versions, I found that reverting cc3d0b5dd989d3238d456f9fd385946379a9c13d in patch_realtek.c ( see https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/soun... ) restores the sound and it works fine after the reboot.

It's the commit cc3d0b5dd989d3238d456f9fd385946379a9c13d Author: Kailang Yang kailang@realtek.com Date:  Thu Nov 14 15:21:09 2024 +0800

ALSA: hda/realtek: Update ALC256 depop procedure         Old procedure has a chance to meet Headphone no output.         Fixes: 4a219ef8f370 ("ALSA: hda/realtek - Add ALC256 HP depop function")     Signed-off-by: Kailang Yang kailang@realtek.com     Link: https://lore.kernel.org/463c5f93715d4714967041a0a8cec28e@realtek.com     Signed-off-by: Takashi Iwai tiwai@suse.de

and it modified alc256_init() sequence.

Kailang, could you check this regression?

And, Ruslan, it's the case without the headphone, right? Is the sound still lost when you connect the headphone and apply the same procedure?

thanks,

Takashi

On Thu, Nov 06, 2025 at 10:34:44AM +0100, Thorsten Leemhuis wrote:

Hi, top-posting for once, to make this easily accessible to everyone.

@Kailang, @Takashi: The fix for the mainline/6.12.y regression this thread is about was afaics merged for v6.17-rc3 as f4b3cef55f5f96 ("ALSA: hda/realtek: Audio disappears on HP 15-fc000 after warm boot again").

Greg's scripts despite the Fixes-tag for some reason did not backport it to various series that contain the culprit. Is there a reason for that? Otherwise we likely should ask Greg to pick it up.

We didn't pick it up because it doesn't apply to 6.12 or prior, nor is there a stable tag on that commit :)

I'll backport it to older trees and queue it up.