On 7/8/21 5:11 PM, Mukunda,Vijendar wrote:
On 7/7/21 10:04 PM, Pierre-Louis Bossart wrote:
+static int snd_acp5x_suspend(struct device *dev) +{
- int ret;
- struct acp5x_dev_data *adata;
- adata = dev_get_drvdata(dev);
- ret = acp5x_deinit(adata->acp5x_base);
- if (ret)
dev_err(dev, "ACP de-init failed\n");- else
dev_dbg(dev, "ACP de-initialized\n");- return ret;
+}
+static int snd_acp5x_resume(struct device *dev) +{
- int ret;
- struct acp5x_dev_data *adata;
- adata = dev_get_drvdata(dev);
- ret = acp5x_init(adata->acp5x_base);
- if (ret) {
dev_err(dev, "ACP init failed\n");return ret;- }
- return 0;
+}
+static const struct dev_pm_ops acp5x_pm = {
- .runtime_suspend = snd_acp5x_suspend,
- .runtime_resume = snd_acp5x_resume,
- .resume = snd_acp5x_resume,
use SET_SYSTEM_SLEEP_PM_OPS and SET_RUNTIME_PM_OPS?
suspend and resume callbacks implementation is same for runtime pm ops and system level pm ops in ACP PCI driver i.e in suspend callback acp de-init sequence will be invoked and in resume callback acp init sequence will be invoked. As per our understanding if we safeguard code with CONFIG_PM_SLEEP macro, then runtime pm ops won't work.
Do we need to duplicate the same code as mentioned below?
static const struct dev_pm_ops acp5x_pm = { SET_RUNTIME_PM_OPS(snd_acp5x_runtime_suspend, snd_acp5x_runtime_resume, NULL) SET_SYSTEM_SLEEP_PM_OPS(snd_acp5x_suspend, snd_acp5x_resume) };
where snd_acp5x_runtime_suspend() & snd_acp5x_suspend() API implementation is same. Similarly snd_acp5x_runtime_resume() & snd_acp5x_resume() implementation is same.
We will modify the code.
also not clear why you don't have a .suspend here
It was a miss. we will add .suspend callback which invokes same callback "snd_acp5x_suspend".
And to avoid warnings use __maybe_unused for those callbacks when PM is disabled?
Agreed. We will modify the code and post the new version.