On Thu, Jun 30, 2022 at 6:37 PM Mark Brown [off-list ref] wrote:

On Thu, Jun 30, 2022 at 01:39:11PM +0800, Shengjiu Wang wrote:

quoted

+static int fsl_micfil_reparent_rootclk(struct fsl_micfil *micfil,

unsigned int sample_rate)

quoted

+{
+     struct device *dev = &micfil->pdev->dev;
+     u64 ratio = sample_rate;
+     struct clk *clk;
+     int ret;
+
+     /* Reparent clock if required condition is true */
+     if (!micfil->pll8k_clk || !micfil->pll11k_clk)
+             return 0;
+
+     ratio = do_div(ratio, 8000) ? CLK_11K_FREQ : CLK_8K_FREQ;
+
+     /* Get root clock */
+     clk = micfil->mclk;
+     if (IS_ERR_OR_NULL(clk)) {
+             dev_err(dev, "no mclk clock in devicetree\n");
+             return PTR_ERR(clk);
+     }
+
+     /* Disable clock first, for it was enabled by pm_runtime */
+     clk_disable_unprepare(clk);
+     fsl_asoc_reparent_pll_clocks(dev, clk, micfil->pll8k_clk,
+                                  micfil->pll11k_clk, ratio);
+     ret = clk_prepare_enable(clk);
+     if (ret)
+             return ret;
+
+     return 0;
+}

Seems like more of this logic could be factored out into the reparent
function if the target sample rate is passed in?

Yes, let me update.

Best regards
Wang Shengjiu