On 10/10/19 8:19 PM, Kuninori Morimoto wrote:
Hi Pierre-Louis
- for_each_card_prelinks(card, i, dai_link) {
ret = soc_init_dai_link(card, dai_link);if (ret) {dev_err(card->dev, "ASoC: failed to init link %s: %d\n",dai_link->name, ret);mutex_unlock(&client_mutex);return ret;}- }
This part is difficult to understand.
There were two calls to soc_init_dai_link(), here and [2] below. Your patch removes the first call and the for loop, is this intentional and needed?
soc_init_dai_link() is just sanity_check now. In my understanding, it is needed before soc_bind_dai_link().
Current code is like below. (1) and (2) are for care prelink:ed dai_link. (A) and (B) are for topology added new dai_link. and (1) is for (2) (A) is for (B)
int snd_soc_instantiate_card() { for_each_card_prelinks(...) { (1) ret = soc_init_dai_link(...); ... } ... for_each_card_prelinks(...) { (2) ret = soc_bind_dai_link(...); ... } ... for_each_card_links(...) { ... (A) ret = soc_init_dai_link(...); ... (B) ret = soc_bind_dai_link(...); } }
It is very confusing/verbose code for me. It can be more simple if we can call soc_init_dai_link() from soc_bind_dai_link().
ok, the explanations help, maye you can add them to the commit message to help explain the intent, e.g.
Current code is like below. (1) and (2) are for care prelink:ed dai_link. (A) and (B) are for topology added new dai_link. and (1) is for (2) (A) is for (B)
for_each_card_prelinks(...) { (2) int snd_soc_instantiate_card() { for_each_card_prelinks(...) { (1) ret = soc_init_dai_link(...); ... } ... ret = soc_bind_dai_link(...); ... } ... for_each_card_links(...) { ... (A) ret = soc_init_dai_link(...); ... (B) ret = soc_bind_dai_link(...); }
and the new code keeps the same flow/steps but collapses the two calls into one
for_each_card_prelinks(...) { (2) int snd_soc_instantiate_card() { for_each_card_prelinks(...) { (1) ret = soc_bind_dai_link(...); ... } ... for_each_card_links(...) { (A) (B) ret = soc_bind_dai_link(...); }