From: "Maciej S. Szmigiero" <[email protected]>
I've noticed that some of mixer controls on my sound card seem to
be partially broken on the 6.0 kernel - alsactl wasn't able to find them
when restoring the mixer state.
The issue was traced down to the recent addition of hashed controls lookup
in commit c27e1efb61c5 ("ALSA: control: Use xarray for faster lookups").
Since that commit it is *not* enough to just directly update the control
name field (like some of ALSA drivers were doing).
Now the hash entries for the modified control have to be updated too.
This patch set adds a snd_ctl_rename() function that takes care of doing
this operation properly for callers that already have the relevant
struct snd_kcontrol at hand and hold the control write lock (or simply
haven't registered the card yet).
These prerequisites hold true for all the call sites modified.
The core controls change and the emu10k1 patch were runtime tested.
Similar patches for other devices were only compile tested.
include/sound/control.h | 1 +
sound/core/control.c | 23 +++++++++++++++++++++++
sound/pci/ac97/ac97_codec.c | 32 ++++++++++++++++++++++++--------
sound/pci/ca0106/ca0106_mixer.c | 2 +-
sound/pci/emu10k1/emumixer.c | 2 +-
sound/pci/hda/patch_realtek.c | 2 +-
sound/usb/mixer.c | 2 +-
7 files changed, 52 insertions(+), 12 deletions(-)
On Thu, 20 Oct 2022 22:46:20 +0200,
Maciej S. Szmigiero wrote:
>
> From: "Maciej S. Szmigiero" <[email protected]>
>
> I've noticed that some of mixer controls on my sound card seem to
> be partially broken on the 6.0 kernel - alsactl wasn't able to find them
> when restoring the mixer state.
>
> The issue was traced down to the recent addition of hashed controls lookup
> in commit c27e1efb61c5 ("ALSA: control: Use xarray for faster lookups").
>
> Since that commit it is *not* enough to just directly update the control
> name field (like some of ALSA drivers were doing).
> Now the hash entries for the modified control have to be updated too.
>
> This patch set adds a snd_ctl_rename() function that takes care of doing
> this operation properly for callers that already have the relevant
> struct snd_kcontrol at hand and hold the control write lock (or simply
> haven't registered the card yet).
>
> These prerequisites hold true for all the call sites modified.
>
> The core controls change and the emu10k1 patch were runtime tested.
> Similar patches for other devices were only compile tested.
Good catch!
Applied all patches now.
thanks,
Takashi