[alsadevel] [PATCH] ASOC: wm8960: Add multiple MCLK frequency support
From: "Chen.Liu" chen.liu.opensource@gmail.com
Issue: MCLK=24MHZ,SYSCLOCK=12.288MHZ. When the 'wm8960_set_dai_pll' function is called,the driver will prompted "WM8960 PLL: unsupported N = 4" error message. However,the value of PLLN should be 8 based on the table45( PLL Frequency Examples) of the wm8960 chip manuanl.
Reason: The intergrated PLL can be used to generate SYSCLK for the wm8960. The pll_factors function in the wm8974.c file is mainly responsible for setting the divider factor of the PLL divider with reference to MCLK clock frequency to obtain the required SYSCLK clock frequency. The configure diagram for this function can be seen below.
MCLK>[f/2]>[R=f2/f1]>[f/4]>[f/N](1)>SYSCLOCK
The configuration of the PLL divider in above diagram is applicable to MCLK clock frequency is less than or equal to 14.4MHZ. examples: (1)MCLK=12MHz, required clock = 12.288MHz. R should be chosen to ensure 5 < PLLN < 13. There is a fixed divide by 4 in the PLL and a selectable divide by N after the PLL which should be set to divide by 1 to meet this requirement. Enabling the divide by 1 sets the required f2 = 4 x 1 x 12.288MHz = 49.152MHz.
R = 49.152 / (12 / 2) = 8.192 PLLN = int R = 8
(2)MCLK=24MHz, required clock = 12.288MHz. Enabling the divide by 1 sets the required f2 = 4 x 1 x 12.288MHZ = 49.152MHZ.
R = 49.152 / (24 / 2) = 4.096 PLLN = int R = 4
And if the [f/N] reg is set to 2,then the required f2 = 4x2x12.288MHZ = 98.304MHZ.
R = 98.304 / (24 / 2) = 8.192 PLLN = int R = 8
Because the [f/N] reg is fixed divide by 1,that's why driver prompted error message.
Solution: The purpose of this patch is when the conditions(5 < PLLN < 13) is false, set the register[f/N] to be 2, and the recalculate the divide factor of the PLL divider.
MCLK>[f/2]>[R=f2/f1]>[f/4]>[f/N](2)>SYSCLOCK
Signedoffby: Chen.Liu chen.liu.opensource@gmail.com  sound/soc/codecs/wm8960.c  25 +++++++++++++++++++++ 1 file changed, 21 insertions(+), 4 deletions()
diff git a/sound/soc/codecs/wm8960.c b/sound/soc/codecs/wm8960.c index 997c446..49a27b1 100644  a/sound/soc/codecs/wm8960.c +++ b/sound/soc/codecs/wm8960.c @@ 1036,17 +1036,25 @@ static bool is_pll_freq_available(unsigned int source, unsigned int target) * to allow rounding later */ #define FIXED_PLL_SIZE ((1 << 24) * 10)
static int pll_factors(unsigned int source, unsigned int target, +static int pll_factors(struct snd_soc_codec *codec, + unsigned int source, unsigned int target, struct _pll_div *pll_div) { unsigned long long Kpart; unsigned int K, Ndiv, Nmod; + int unsupport = 0; + u16 reg;
pr_debug("WM8960 PLL: setting %dHz>%dHz\n", source, target);
/* Scale up target to PLL operating frequency */ target *= 4;
+retry: + if (unsupport) { + target *= 2; + unsupport += 1; + } Ndiv = target / source; if (Ndiv < 6) { source >>= 1; @@ 1056,8 +1064,12 @@ static int pll_factors(unsigned int source, unsigned int target, pll_div>pre_div = 0;
if ((Ndiv < 6)  (Ndiv > 12)) {  pr_err("WM8960 PLL: Unsupported N=%d\n", Ndiv);  return EINVAL; + if (unsupport == 2) { + pr_err("WM8960 PLL: Unsupported N=%d\n", Ndiv); + return EINVAL; + } + unsupport = 1; + goto retry; }
pll_div>n = Ndiv; @@ 1077,6 +1089,11 @@ static int pll_factors(unsigned int source, unsigned int target,
pll_div>k = K;
+ if (unsupport) { + reg = snd_soc_read(codec, WM8960_CLOCK1)  WM8960_SYSCLK_DIV_2; + snd_soc_write(codec, WM8960_CLOCK1, reg); + } + pr_debug("WM8960 PLL: N=%x K=%x pre_div=%d\n", pll_div>n, pll_div>k, pll_div>pre_div);
@@ 1091,7 +1108,7 @@ static int wm8960_set_pll(struct snd_soc_codec *codec, int ret;
if (freq_in && freq_out) {  ret = pll_factors(freq_in, freq_out, &pll_div); + ret = pll_factors(codec, freq_in, freq_out, &pll_div); if (ret != 0) return ret; }
On Tue, Dec 12, 2017 at 07:58:56PM +0800, Chen.Liu wrote:
From: "Chen.Liu" chen.liu.opensource@gmail.com
Please make sure to CC patches@opensource.cirrus.com on patches to the Wolfson CODEC drivers.
Issue: MCLK=24MHZ,SYSCLOCK=12.288MHZ. When the 'wm8960_set_dai_pll' function is called,the driver will
wm8960_set_pll, not wm8960_set_dai_pll?
prompted "WM8960 PLL: unsupported N = 4" error message. However,the value of PLLN should be 8 based on the table45( PLL Frequency Examples) of the wm8960 chip manuanl.
Reason: The intergrated PLL can be used to generate SYSCLK for the wm8960. The pll_factors function in the wm8974.c file is mainly responsible
Assuming that should be wm8960.c?
static int pll_factors(unsigned int source, unsigned int target, +static int pll_factors(struct snd_soc_codec *codec,
unsigned int source, unsigned int target, struct _pll_div *pll_div)
{ unsigned long long Kpart; unsigned int K, Ndiv, Nmod;
int unsupport = 0;
u16 reg;
pr_debug("WM8960 PLL: setting %dHz>%dHz\n", source, target);
/* Scale up target to PLL operating frequency */ target *= 4;
+retry:
 if (unsupport) {
target *= 2;
unsupport += 1;
 } Ndiv = target / source; if (Ndiv < 6) { source >>= 1;
@@ 1056,8 +1064,12 @@ static int pll_factors(unsigned int source, unsigned int target, pll_div>pre_div = 0;
Feels like it would be nice to wrap this Ndiv calculation in a loop rather than adding the goto into the below if statement.
if ((Ndiv < 6)  (Ndiv > 12)) {
pr_err("WM8960 PLL: Unsupported N=%d\n", Ndiv);
return EINVAL;
if (unsupport == 2) {
pr_err("WM8960 PLL: Unsupported N=%d\n", Ndiv);
return EINVAL;
}
unsupport = 1;
goto retry;
}
pll_div>n = Ndiv;
@@ 1077,6 +1089,11 @@ static int pll_factors(unsigned int source, unsigned int target,
pll_div>k = K;
 if (unsupport) {
reg = snd_soc_read(codec, WM8960_CLOCK1)  WM8960_SYSCLK_DIV_2;
snd_soc_write(codec, WM8960_CLOCK1, reg);
snd_soc_update_bits?
Also there is no code that seems to reset this bit should we later ask for a configuration that no longer needs it.
Thanks, Charles
On Tue, Dec 12, 2017 at 01:21:04PM +0000, Charles Keepax wrote:
On Tue, Dec 12, 2017 at 07:58:56PM +0800, Chen.Liu wrote:
From: "Chen.Liu" chen.liu.opensource@gmail.com
Please make sure to CC patches@opensource.cirrus.com on patches to the Wolfson CODEC drivers.
Apologies that one is my bad you had the wolfsonmicro.com address on there that is just as good.
Thanks, Charles
participants (3)

Charles Keepax

Chen.Liu

Mark Brown