mbox series

[RFC,v4,0/4] design a way to change audio Jack state by software

Message ID 20210111130557.90208-1-hui.wang@canonical.com
Headers show
Series design a way to change audio Jack state by software | expand

Message

Hui Wang Jan. 11, 2021, 1:05 p.m. UTC
the changes in the v4:
 - change the sound-core to sound and change the sound_core_debugfs_root
   to sound_debugfs_root in the 0001-xxx.patch
 - change kzalloc/kfree to kvzalloc/kvfree in the debugfs fops for
   0001-xxx.patch and 0003-xxx.patch
 - And if needed, I could squash 4 patches into 1 patch before merging.

the changes in the v3 (for easy to review, divide change into 4 patches):
 - address the comment about the snd_jack_report() and _snd_jack_report(),
   the v2 design is hard to understand and is hard to review, in the v3,
   separate the jack_report to snd_jack_report() and snd_jack_inject_report(),
   hw jack events call snd_jack_report() as before, if a jack contains multi
   jack_kctl and the jack_kctl's sw_inject is enabled, the status and the
   related input-dev's events will not be reproted. The injection events call
   snd_jack_inject_report(). This change is squashed to 0001-xxx.patch,  the
   rest part of 0001-xxx.patch is same as v2.

 - address the comment about folders'name in the 0002-xxx.patch, so far, drop
   the '/', ',', '=' and ' ' from the folders' name.

 - address the comment about adding more debugfs nodes in the 0003-xxx.patch,
   it adds kctl_id, mask_bits, status and type.

 - address the comment about save-n-restore jack's hw status in the
   0004-xxx.patch, adding a hw_status_cache and save the last reported jack
   hw event, once the sw_inject is disabled, will restore all jack_kctl's
   state under the same snd_jack with hw_status_cache.
[snip]


the changes in the V2:
 - using debugfs instead of sysfs
 - using jack_ctrl to create a folder instead of snd_jack, since ASoC drivers
   could create multi jack_ctrls within a snd_jack
 - create a folder for each jack_ctrl instead for all jack_ctrls
[ snip ]


Hui Wang (4):
  alsa: jack: implement software jack injection via debugfs
  alsa: jack: adjust jack_kctl debugfs folder's name
  alsa: jack: add more jack_kctl debugfs nodes
  alsa: jack: implement save-and-restore for jack's hw status

 include/sound/core.h |   2 +
 include/sound/jack.h |   1 +
 sound/core/init.c    |   7 +
 sound/core/jack.c    | 352 ++++++++++++++++++++++++++++++++++++++++++-
 sound/core/sound.c   |   8 +
 5 files changed, 366 insertions(+), 4 deletions(-)

Comments

Takashi Iwai Jan. 20, 2021, 2:47 p.m. UTC | #1
On Wed, 20 Jan 2021 09:58:25 +0100,
Hui Wang wrote:
> 
> Hi Takashi,
> 
> Looks like no more comment from others, what is your opinion on the
> patchset, is it ok to merge or need to do some change?

I think the basic concept is OK.  Maybe we want to have a Kconfig to
enable/disable this feature for users who want a slim kernel.

And, it'd be better to reorganize the series.  It's not necessarily
incremental form; e.g. patch 2 doesn't have to be an individual
patch.

I'll take a deeper look for each patch, too.


thanks,

Takashi





> 
> Thanks,
> 
> Hui.
> 
> On 1/11/21 9:05 PM, Hui Wang wrote:
> > the changes in the v4:
> >   - change the sound-core to sound and change the sound_core_debugfs_root
> >     to sound_debugfs_root in the 0001-xxx.patch
> >   - change kzalloc/kfree to kvzalloc/kvfree in the debugfs fops for
> >     0001-xxx.patch and 0003-xxx.patch
> >   - And if needed, I could squash 4 patches into 1 patch before merging.
> >
> > the changes in the v3 (for easy to review, divide change into 4 patches):
> >   - address the comment about the snd_jack_report() and _snd_jack_report(),
> >     the v2 design is hard to understand and is hard to review, in the v3,
> >     separate the jack_report to snd_jack_report() and snd_jack_inject_report(),
> >     hw jack events call snd_jack_report() as before, if a jack contains multi
> >     jack_kctl and the jack_kctl's sw_inject is enabled, the status and the
> >     related input-dev's events will not be reproted. The injection events call
> >     snd_jack_inject_report(). This change is squashed to 0001-xxx.patch,  the
> >     rest part of 0001-xxx.patch is same as v2.
> >
> >   - address the comment about folders'name in the 0002-xxx.patch, so far, drop
> >     the '/', ',', '=' and ' ' from the folders' name.
> >
> >   - address the comment about adding more debugfs nodes in the 0003-xxx.patch,
> >     it adds kctl_id, mask_bits, status and type.
> >
> >   - address the comment about save-n-restore jack's hw status in the
> >     0004-xxx.patch, adding a hw_status_cache and save the last reported jack
> >     hw event, once the sw_inject is disabled, will restore all jack_kctl's
> >     state under the same snd_jack with hw_status_cache.
> > [snip]
> >
> >
> > the changes in the V2:
> >   - using debugfs instead of sysfs
> >   - using jack_ctrl to create a folder instead of snd_jack, since ASoC drivers
> >     could create multi jack_ctrls within a snd_jack
> >   - create a folder for each jack_ctrl instead for all jack_ctrls
> > [ snip ]
> >
> >
> > Hui Wang (4):
> >    alsa: jack: implement software jack injection via debugfs
> >    alsa: jack: adjust jack_kctl debugfs folder's name
> >    alsa: jack: add more jack_kctl debugfs nodes
> >    alsa: jack: implement save-and-restore for jack's hw status
> >
> >   include/sound/core.h |   2 +
> >   include/sound/jack.h |   1 +
> >   sound/core/init.c    |   7 +
> >   sound/core/jack.c    | 352 ++++++++++++++++++++++++++++++++++++++++++-
> >   sound/core/sound.c   |   8 +
> >   5 files changed, 366 insertions(+), 4 deletions(-)
> >
>
Hui Wang Jan. 21, 2021, 12:48 p.m. UTC | #2
On 1/20/21 10:47 PM, Takashi Iwai wrote:
> On Wed, 20 Jan 2021 09:58:25 +0100,
> Hui Wang wrote:
>> Hi Takashi,
>>
>> Looks like no more comment from others, what is your opinion on the
>> patchset, is it ok to merge or need to do some change?
> I think the basic concept is OK.  Maybe we want to have a Kconfig to
> enable/disable this feature for users who want a slim kernel.
>
> And, it'd be better to reorganize the series.  It's not necessarily
> incremental form; e.g. patch 2 doesn't have to be an individual
> patch.
>
> I'll take a deeper look for each patch, too.
>
>
> thanks,
>
> Takashi
>
Got it, will address all comments, and will send a v5 to review soon.

Thanks,

Hui.

>
>
>
>> Thanks,
>>
>> Hui.
>>
>> On 1/11/21 9:05 PM, Hui Wang wrote:
>>> the changes in the v4:
>>>    - change the sound-core to sound and change the sound_core_debugfs_root
>>>      to sound_debugfs_root in the 0001-xxx.patch
>>>    - change kzalloc/kfree to kvzalloc/kvfree in the debugfs fops for
>>>      0001-xxx.patch and 0003-xxx.patch
>>>    - And if needed, I could squash 4 patches into 1 patch before merging.
>>>
>>> the changes in the v3 (for easy to review, divide change into 4 patches):
>>>    - address the comment about the snd_jack_report() and _snd_jack_report(),
>>>      the v2 design is hard to understand and is hard to review, in the v3,
>>>      separate the jack_report to snd_jack_report() and snd_jack_inject_report(),
>>>      hw jack events call snd_jack_report() as before, if a jack contains multi
>>>      jack_kctl and the jack_kctl's sw_inject is enabled, the status and the
>>>      related input-dev's events will not be reproted. The injection events call
>>>      snd_jack_inject_report(). This change is squashed to 0001-xxx.patch,  the
>>>      rest part of 0001-xxx.patch is same as v2.
>>>
>>>    - address the comment about folders'name in the 0002-xxx.patch, so far, drop
>>>      the '/', ',', '=' and ' ' from the folders' name.
>>>
>>>    - address the comment about adding more debugfs nodes in the 0003-xxx.patch,
>>>      it adds kctl_id, mask_bits, status and type.
>>>
>>>    - address the comment about save-n-restore jack's hw status in the
>>>      0004-xxx.patch, adding a hw_status_cache and save the last reported jack
>>>      hw event, once the sw_inject is disabled, will restore all jack_kctl's
>>>      state under the same snd_jack with hw_status_cache.
>>> [snip]
>>>
>>>
>>> the changes in the V2:
>>>    - using debugfs instead of sysfs
>>>    - using jack_ctrl to create a folder instead of snd_jack, since ASoC drivers
>>>      could create multi jack_ctrls within a snd_jack
>>>    - create a folder for each jack_ctrl instead for all jack_ctrls
>>> [ snip ]
>>>
>>>
>>> Hui Wang (4):
>>>     alsa: jack: implement software jack injection via debugfs
>>>     alsa: jack: adjust jack_kctl debugfs folder's name
>>>     alsa: jack: add more jack_kctl debugfs nodes
>>>     alsa: jack: implement save-and-restore for jack's hw status
>>>
>>>    include/sound/core.h |   2 +
>>>    include/sound/jack.h |   1 +
>>>    sound/core/init.c    |   7 +
>>>    sound/core/jack.c    | 352 ++++++++++++++++++++++++++++++++++++++++++-
>>>    sound/core/sound.c   |   8 +
>>>    5 files changed, 366 insertions(+), 4 deletions(-)
>>>