On 7/20/22 03:57, Shengjiu Wang wrote: > > +static int imx8ulp_resume(struct snd_sof_dev *sdev) > > +{ > > + struct imx8ulp_priv *priv = (struct imx8ulp_priv > *)sdev->pdata->hw_pdata; > > + int i; > > + > > + imx8_enable_clocks(sdev, priv->clks); > > + > > + for (i = 0; i < DSP_MU_CHAN_NUM; i++) > > + imx_dsp_request_channel(priv->dsp_ipc, i); > > + > > + return 0; > > is the assymetry between suspend and resume intentional? You are missing > the update_bit for EXECUTE_BIT? > > > Yes, intentional. After resume the firmware is reloaded and EXECUTE_BIT > will be updated at trigger DSP start. That's worthy of a comment to help reviewers, thanks.