Re: UB Sanitizer issue on wake-up from suspend (sound related)

Re: UB Sanitizer issue on wake-up from suspend (sound related)

[Takashi Iwai]

On Mon, 10 Nov 2025 20:46:00 +0100,
Cristian Verde wrote:
> 
> 
> Good evening (in my country),
> 
> My name is Cristian Verde and this is my first bug report, so I'm sorry for
> the crudity.
> 
> [ 6173.298248] UBSAN: invalid-load in /build/linux-PxlDdq/linux-6.16.8/sound/
> core/pcm_native.c:636:46
> [ 6173.298255] load of value 162 is not a valid value for type '_Bool'
> [ 6173.298281] nvidia 0000:01:00.0: PM: pci_pm_suspend(): nv_pmops_suspend
> [nvidia] returns -5
> [ 6173.298660] nvidia 0000:01:00.0: PM: dpm_run_callback(): pci_pm_suspend
> returns -5
> [ 6173.298673] nvidia 0000:01:00.0: PM: failed to suspend async: error -5
> [ 6173.341921] PM: Some devices failed to suspend, or early wake event
> detected
> [ 6173.818931] nvidia 0000:01:00.0: PM: pci_pm_suspend(): nv_pmops_suspend
> [nvidia] returns -5
> [ 6173.818991] nvidia 0000:01:00.0: PM: dpm_run_callback(): pci_pm_suspend
> returns -5
> [ 6173.819003] nvidia 0000:01:00.0: PM: failed to suspend async: error -5
> [ 6173.857619] PM: Some devices failed to suspend, or early wake event
> detected
> 
> Also found this Reddit post
> 
> I reported it already to Pop_OS maintainers, since I don't know where the
> problem is (the Pop_OS 22.04 that I'm using has Linux 6.16.8, as can be seen
> above, so a bit of an older version)
> 
> The issue might be related to why the volume slider doesn't work (since it's
> sound related)? Only happens on the laptop's speakers
> (LENOVO_MT_82Y7_BU_idea_FM_Yoga Pro 7 14IRH8), not any other audio devices.
> 
> If it helps, I had the same issue when I ran Ubuntu 22.04, and didn't have it
> on another machine (some ASUS).
> 
> I can provide other details, and please let me know if I should use
> other communication channels, thank you very much!
> 
> Best regards,
> Cristian Verde

This must be a bug rather in Nvidia download driver.  It means a wrong
value is stored in substream->runtime->stop_operating field which
should have been a boolean.


Takashi