diff --git a/include/sound/simple_card_utils.h b/include/sound/simple_card_utils.h index 86a1e95..9825308 100644 --- a/include/sound/simple_card_utils.h +++ b/include/sound/simple_card_utils.h @@ -56,6 +56,10 @@ struct asoc_simple_priv { struct asoc_simple_dai *dais; struct snd_soc_codec_conf *codec_conf; struct gpio_desc *pa_gpio;
const struct snd_soc_ops *ops;unsigned int force_dpcm:1;uintptr_t dpcm_selectable;void *data;I have opinions about these.
About dpcm_selectable, indeed current audio-graph is using it as "uintptr_t", but as you know, it checks whether it was non-NULL or not only. This means we can use it as bit-field.
Yes that is true. Something like this would work?
graph_probe(...) { ...
if (of_device_get_match_data(dev)) priv->dpcm_selectable = 1;
... }
BTW, do we need to have dpcm_selectable at priv ?
Tegra audio graph driver does not require this because already it is populating 'force_dpcm' flag and having 'selectable' does not make much sense for it.
One note is that, -scu- user is only me (locally), and it will be removed when audio-graph2 was created. (My plan is keep code for you, but remove compatible)
Right now I am just keeping it to allow current code work. Later you can remove this during graph2.
About *data, I think we can avoid *data if driver side priv includes asoc_simple_priv ?
struct my_priv { struct asoc_simple_priv *simple; ... }; #define simple_to_priv(_simple) container_of((_simple), struct my_priv, simple)
This seems like a better plan, will do this.