Public bug reported:
On an Acer Aspire A515-57 (Realtek ALC256, subsystem 0x10251616), if a
device is already plugged into the 3.5 mm combo jack when the machine
boots, the output has constant, loud noise (much louder than the audio,
independent of volume) for as long as the audio path is active.
Physically unplugging and replugging the jack immediately fixes it until
the next boot with the plug inserted. The same hardware does not have
the problem on Windows (installed Windows and Windows To Go).
The only difference I could find between the bad and the good state is
vendor coefficient 0x46 of node 0x20: the bits 0x0c00 are clear at boot
and set after a physical replug. These bits are read-only (hardware-
set), so I cannot set them from software.
Hardware / software
Laptop: Acer Aspire A515-57, board Ganymede_ADU, BIOS V1.33 (06/09/2026)
Codec: Realtek ALC256, vendor id 0x10ec0256, subsystem id 0x10251616
Controller: Intel Alder Lake PCH-P HD Audio [8086:51c8]
Distro: Linux Mint 22.3 (Ubuntu 24.04 base)
Kernels tested: 7.0.0-42-generic and 6.14.0-37-generic (both affected)
Drivers tested: SOF (sof-hda-dsp, firmware 2.2.0-57864, ABI 3:22:1 with kernel
ABI 3:23:1) and legacy snd_hda_intel (snd-intel-dspcfg.dsp_driver=1): both
affected
Audio server: PipeWire 1.0.5 / WirePlumber 0.4.17
Device on the jack: a powered speaker (3-pole TRS cable)
Steps to reproduce
Plug a device into the 3.5 mm jack.
Boot the laptop (cold boot or reboot).
Log in and play any audio (or any system sound).
Actual: constant noise while the audio path is open. Expected: clean audio.
Workaround: unplug and replug the jack. A plain log-out/log-in does not trigger
the problem. Plugging after login works normally.
What I measured
Reading all 128 coefficients of node 0x20 in the bad state (plugged at boot,
noisy) and in the good state (after replug), with the same driver and a stream
playing in both:
coef[0x46]: bad = 0x0004 good = 0x0c04
All other 127 coefficients are identical. In another boot the bad state read
0x0404. The /proc/asound/card0/codec#0 dump (pin controls, amps, power states,
unsolicited tags) shows no difference apart from stream state. Pin sense of
node 0x21 reads 0x80000000 (present) in the good state.
Writing coef 0x46 through hda-verb shows bits 0x0c00 are read-only:
write 0x0000 -> reads 0xc00 write 0x0004 -> reads 0xc04
write 0x0804 -> reads 0xc04 write 0x0400 -> reads 0xc00
write 0x0c04 -> reads 0xc04
So the bit 0x0004 is writable and 0x0c00 is forced by hardware, apparently only
after a physical plug event.
Things that did NOT help
options snd-hda-intel power_save=0 power_save_controller=N
Legacy HDA driver instead of SOF
Kernel 6.14 instead of 7.0
Lowering the codec Headphone amp volume to -45 dB (noise unchanged, so it is
not through the gain stage)
echo 1 > /sys/class/sound/hwC0D0/reconfig (with WirePlumber stopped)
With hda-verb, in the bad state: AFG power cycle D3->D0, toggling pin 0x21
pin-widget-control, re-arming unsolicited events on 0x21, pin 0x21 power cycle,
toggling EAPD on 0x21. None changed coef 0x46 or the noise.
Related
Ubuntu bug 1944575 describes the same family of behaviour (headset plugged at
boot is not detected correctly, replug fixes it) on an Acer Aspire A515-56 with
ALC255. The kernel quirk table has an entry for A515-56 (0x1025:0x1466) but
none for A515-57 (0x1025:0x1616).
Attachments
alsa-info output (alsa-info --no-upload)
coefficient dumps of node 0x20, bad vs good state
codec proc dump before and after replug
I can test patches or other kernels on this machine.
** Affects: linux (Ubuntu)
Importance: Undecided
Status: New
** Attachment added: "anexos-audio.tar.gz"
https://bugs.launchpad.net/bugs/2172565/+attachment/6013576/+files/anexos-audio.tar.gz
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2172565
Title:
ALC256 (Acer Aspire A515-57): constant loud noise on the 3.5 mm jack
when a device is plugged in at boot; replugging fixes it. coef 0x46
jack-detect bits not set at boot
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2172565/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
