[alsa-devel] [RFC][PATCH] ASoC: soc-core: verify Sound Card normality

[Kuninori Morimoto]

Hi Mark

Thank you for your feedback

> > To avoid Sound Card crash in (2) case, what we can do now is, add dirty
> > flag on Sound Card, and avoid to open Sound Card.
> > This patch solved this issue.
> 
> I think this is a good direction to at least start to mitigate these
> problems (which we really should be doing) and hopefully make it easier
> to do further improvements in future.  There's obviously more places
> where we should be checking the flag (controls for example) but they can
> be added later.  One thing I would like to see is instead of setting the
> flag directly when we see a problem call a function to do it.  That way
> if we want to improve things in the future we can do that without having
> to update the callers again.

I guess you are pointing about snd_soc_dapm_shutdown() in snd_soc_unregister_card() ?
If so, I agree. Actually, I don't like this kind of adhoc handling.
I want to explain about it.

Order
[1] Disable Sound Card         -> Disable CPU/Codec/Platform or do nothing
[2] Disable CPU/Codec/Platform -> Disable Card or do nothing

Operation
1) shutdown all Card connected DAPM
2) cleanup Card resource.

In case of order [1], operation 1) -> 2) is no problem, because all card connected DAPM
exists on Card.
But, in case of order [2], operation 1) will try to call disconnected DAPM.
The DAPM disconnection from Card is done by snd_soc_dapm_free() which is called from
soc_remove_component().
This soc_remove_component() will be called with "remove_order" from operation 2).
One note is that reordering operation to 2) -> 1) will avoid this crash issue,
but it is no meaning, because 1) will do nothing ;)

Here, other solution is calling soc_remove_component() without "remove_order"
or calling snd_soc_dapm_free() when CPU/Codec/Platform are disabled.
But, 1 bad-point is that disabled CPU/Codec/Platform have no chance to
be called about snd_soc_dapm_set_bias_level() (= 1) operation).
Of course, adhoc function can solve this issue too.

If you are OK, I can work for this idea and remove adhoc operation
from snd_soc_unregister_card()

Best regards
---
Kuninori Morimoto