On Tue, Dec 19, 2017 at 02:19:48PM +0800, chen liu wrote:
2017-12-18 19:55 GMT+08:00 Charles Keepax ckeepax@opensource.cirrus.com:
On Mon, Dec 18, 2017 at 07:32:41PM +0800, chen liu wrote:
2017-12-18 17:31 GMT+08:00 Charles Keepax <ckeepax@opensource.cirrus.com
On Fri, Dec 15, 2017 at 09:07:15PM +0800, chen liu wrote:
2017-12-15 0:19 GMT+08:00 Charles Keepax <
ckeepax@opensource.cirrus.com
On Wed, Dec 13, 2017 at 08:37:30PM +0800, Chen.Liu wrote:
// Then this branch will be executed. // If the sample rate is 44100HZ, the value of freq_out will be11.2896MHZ freq_out = wm8960_configure_pll(codec, freq_in, &i, &j, &k); if (freq_out < 0) { dev_err(codec->dev, "failed to configure clock via PLL\n"); return freq_out; } // The 'wm8960_set_pll' function will be executed,this function is very important. // But now, the freq_in is 24MHZ and the freq_out is 11.2896MHZ. wm8960_set_pll(codec, freq_in, freq_out); ...
Because the 'wm8960_configure_pll' function has been pre-scaled by 2 for freq_out, but the value of freq_out is not multiplied by 2 in the 'pll_factors' function, it prints the "WM8960 PLL: Unsupported N =" error message.
Ah ok so I hadn't realised you were using the SYSCLK_AUTO stuff and finding an issue with it. Thank you very much for the explanation.
One thing I don't understand though is it looks like the freq_out returned from wm8960_configure_pll should already be taking the SYSCLK_DIV into account:
for (j = 0; j < ARRAY_SIZE(dac_divs); ++j) { sysclk = lrclk * dac_divs[j]; freq_out = sysclk * sysclk_divs[i];
So if lrclk=44100, j=0, i=2 which I believe is the case in question we should get:
sysclk = 44100 * 256 = 11.2896MHz freq_out = 11.2896MHz * 2 = 22.5792MHz
So when wm8960_configure_pll returns i=2 then it should also be setting freq_out as 22.5792MHz. So you final call there to wm8960_set_pll should be:
wm8960_set_pll(codec, 24MHz, 22.5792MHz);
Can you see why that isn't happening?
Thanks, Charles