Help with wm8731 support for the Utilite (imx6q)

[Jonathan Bennett]

Hey, I'm working with a compulab utilite, trying to get everything working
with 3.17, with the eventual goal of getting better support for it
upstream. Currently fighting the sound. It's a imx6q talking to a wm8731
over i2c/i2s.
I am booting Fedora 21 arm Alpha, with my own dtb (
http://pastebin.com/4a4PBKwx) and driver (http://pastebin.com/1CsesuVV)

It shows up in aplay -l, but trying to play anything to it gives me a div
by 0 in a driver: [<c04f27e4>] (Ldiv0) from [<bf50512c>]
(fsl_ssi_hw_params+0x194/0x364 [snd_soc_fsl_ssi]) This div0 happens in
sound/soc/fsl/fsl_ssi.c, in the function fsl_ssi_set_bclk.

My driver file is an almost direct copy of what compulab shipped with a
3.10 kernel. This is my first attempt at kernel hacking, so any help would
be appreciated.

~Jonathan Bennett

[Fabio Estevam]

Hi Jonathan,

Post by Jonathan Bennett
Hey, I'm working with a compulab utilite, trying to get everything working
with 3.17, with the eventual goal of getting better support for it
upstream. Currently fighting the sound. It's a imx6q talking to a wm8731
over i2c/i2s.
I am booting Fedora 21 arm Alpha, with my own dtb (
http://pastebin.com/4a4PBKwx) and driver (http://pastebin.com/1CsesuVV)

I don't have access to a utilite board, but just trying to understand:
is wm8731 audio functional with the 3.10.17 kernel com compulab?

Post by Jonathan Bennett
It shows up in aplay -l, but trying to play anything to it gives me a div
by 0 in a driver: [<c04f27e4>] (Ldiv0) from [<bf50512c>]
(fsl_ssi_hw_params+0x194/0x364 [snd_soc_fsl_ssi]) This div0 happens in
sound/soc/fsl/fsl_ssi.c, in the function fsl_ssi_set_bclk.
My driver file is an almost direct copy of what compulab shipped with a
3.10 kernel. This is my first attempt at kernel hacking, so any help would
be appreciated.

I would suggest as a first step to get the dts for compulab upstreamed.

There was an attempt from Valentin Raevsky (added on Cc) last year:
http://lists.infradead.org/pipermail/linux-arm-kernel/2013-October/204896.html

After the basic dts file is added then you can work on adding audio support.

Some suggestions:

- The dts from 3.10.17 can not be applied 'as-is' into 3.17. Take a
look for example into the audio node you have:

sound {
compatible = "fsl,imx6q-cm-fx6-wm8731",
"fsl,imx-audio-wm8731";
model = "wm8731-audio";
ssi-controller = <&ssi2>;
src-port = <2>;
ext-port = <4>;
audio-codec = <&codec>;
audio-routing = "LOUT", "ROUT", "LLINEIN", "RLINEIN";
};

'src-port' and 'ext-port' are not valid properties.

They should be like this instead:

mux-int-port = <2>;
mux-ext-port = <4>;

Take a look at arch/arm/boot/dts/imx6qdl-sabresd.dtsi for a reference.

Also the codec clocks that are passed into the dts are not correct.
The wm8731 does not acquire any clocks in the mainline driver.

You should also try to use the 'simple-audio-card' binding instead of
creating the imx-wm8731 machine file.

Here is a reference of the arch/arm/boot/dts/vf610-twr.dts where you
can see simple-audio-card in use.

Good luck!

Regards,

Fabio Estevam

[Jonathan Bennett]

Post by Fabio Estevam
Hi Jonathan,

Post by Jonathan Bennett
Hey, I'm working with a compulab utilite, trying to get everything

working

Post by Jonathan Bennett
with 3.17, with the eventual goal of getting better support for it
upstream. Currently fighting the sound. It's a imx6q talking to a wm8731
over i2c/i2s.
I am booting Fedora 21 arm Alpha, with my own dtb (
http://pastebin.com/4a4PBKwx) and driver (http://pastebin.com/1CsesuVV)

is wm8731 audio functional with the 3.10.17 kernel com compulab?
The audio works on their pre-canned ubuntu image, which runs 3.10.

Post by Jonathan Bennett
It shows up in aplay -l, but trying to play anything to it gives me a div
by 0 in a driver: [<c04f27e4>] (Ldiv0) from [<bf50512c>]
(fsl_ssi_hw_params+0x194/0x364 [snd_soc_fsl_ssi]) This div0 happens in
sound/soc/fsl/fsl_ssi.c, in the function fsl_ssi_set_bclk.
My driver file is an almost direct copy of what compulab shipped with a
3.10 kernel. This is my first attempt at kernel hacking, so any help

would

Post by Jonathan Bennett
be appreciated.

I would suggest as a first step to get the dts for compulab upstreamed.
http://lists.infradead.org/pipermail/linux-arm-kernel/2013-October/204896.html
After the basic dts file is added then you can work on adding audio support.
There is already a small dts file upstreamed. It barely works. I've been

get usb and the second ethernet port working, which I would like to get
pushed upstream. I was hoping to have more of it working before then,
though.

Post by Fabio Estevam
- The dts from 3.10.17 can not be applied 'as-is' into 3.17. Take a
sound {
compatible = "fsl,imx6q-cm-fx6-wm8731",
"fsl,imx-audio-wm8731";
model = "wm8731-audio";
ssi-controller = <&ssi2>;
src-port = <2>;
ext-port = <4>;
audio-codec = <&codec>;
audio-routing = "LOUT", "ROUT", "LLINEIN", "RLINEIN";
};
'src-port' and 'ext-port' are not valid properties.
mux-int-port = <2>;
mux-ext-port = <4>;
Take a look at arch/arm/boot/dts/imx6qdl-sabresd.dtsi for a reference.
Also the codec clocks that are passed into the dts are not correct.
The wm8731 does not acquire any clocks in the mainline driver.
You should also try to use the 'simple-audio-card' binding instead of
creating the imx-wm8731 machine file.
Here is a reference of the arch/arm/boot/dts/vf610-twr.dts where you
can see simple-audio-card in use.
I am trying to use simple-audio-card, without any success. I've been

hammering at this for a couple days now. My current dts looks like
this(just the audio related bits):

sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
mux-int-port = <2>;
mux-ext-port = <4>;
simple-audio-card,format = "i2s";
simple-audio-card,mclk-fs = <256>;
simple-audio-card,cpu {
sound-dai = <&ssi2>;
};
simple-audio-card,codec {
sound-dai = <&codec>;
};
};
....
/* i2c3 */
&i2c3 {
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_i2c3>;
status = "okay";

***@50 {
compatible = "at24,24c02";
reg = <0x50>;
pagesize = <16>;
};

codec: ***@1a {
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
};
};
...
&ssi2 {
fsl,mode = "i2s-master";
status = "okay";
#sound-dai-cells = <0>;
};


It registers:
# aplay -l
**** List of PLAYBACK Hardware Devices ****
card 0: D202c000ssiwm87 [202c000.ssi-wm8731-hifi], device 0:
202c000.ssi-wm8731-hifi wm8731-hifi-0 []
Subdevices: 1/1
Subdevice #0: subdevice #0


However:
# speaker-test

speaker-test 1.0.28

Playback device is default
Stream parameters are 48000Hz, S16_LE, 1 channels
Using 16 octaves of pink noise
Rate set to 48000Hz (requested 48[ 57.443583] asoc-simple-card sound:
ASoC: machine hw_params failed: -22
000Hz)
Buffer size range from 128 to 32768
Period size range from 64 to 16384
Using max buffer size 32768
Periods = 4
Unable to set hw params for playback: Invalid argument
Setting of hwparams failed: Invalid argument




It's a wm8731 hanging off an i2c bus from an i.mx6q chip. Seems like
exactly what simple-audio-card is designed for. I imagine that there is
something simple that I am overlooking, but I'm at a loss to find it.

Post by Fabio Estevam
Good luck!
Regards,
Fabio Estevam

Thanks for the help!
~Jonathan Bennett

[Jonathan Bennett]

And a different dts that is closer to the example given:

sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
mux-int-port = <2>;
mux-ext-port = <4>;
simple-audio-card,format = "i2s";
simple-audio-card,cpu {
sound-dai = <&ssi2>;
master-clkdir-out;
frame-master;
bitclock-master;
};
simple-audio-card,codec {
frame-master;
bitclock-master;
sound-dai = <&codec>;
};
};

/* i2c3 */
&i2c3 {
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_i2c3>;
status = "okay";

***@50 {
compatible = "at24,24c02";
reg = <0x50>;
pagesize = <16>;
};

codec: ***@1a {
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
clocks = <&clks IMX6QDL_CLK_SSI2>;
};
};



&ssi2 {
fsl,mode = "i2s-master";
status = "okay";
#sound-dai-cells = <0>;
};

Gives me this output:
# aplay -l
**** List of PLAYBACK Hardware Devices ****
card 0: imxspdif [imx-spdif], device 0: S/PDIF PCM snd-soc-dummy-dai-0 []
Subdevices: 1/1
Subdevice #0: subdevice #0
card 1: D202c000ssiwm87 [202c000.ssi-wm8731-hifi], device 0:
202c000.ssi-wm8731-hifi wm8731-hifi-0 []
Subdevices: 1/1
Subdevice #0: subdevice #0
[***@localhost ~]# speaker-test -D hw:1

speaker-test 1.0.28

Playback device is hw:1
Stream parameters are 48000Hz, S16_LE, 1 channels
Using 16 octaves of pink noise
Rate set to 48000Hz (requested 48000Hz)
Buffer size range from 128 to 32768
Period size range from 64 to 16384
Using max buffer size 32768
Periods = 4
was set period_size = 8192
was set buffer_size = 32768
0 - Front Left
Write error: -5,Input/output error
xrun_recovery failed: -5,Input/output error
Transfer failed: Input/output error

Post by Jonathan Bennett

Post by Fabio Estevam
Hi Jonathan,

Post by Jonathan Bennett
Hey, I'm working with a compulab utilite, trying to get everything

working

Post by Jonathan Bennett
with 3.17, with the eventual goal of getting better support for it
upstream. Currently fighting the sound. It's a imx6q talking to a wm8731
over i2c/i2s.
I am booting Fedora 21 arm Alpha, with my own dtb (
http://pastebin.com/4a4PBKwx) and driver (http://pastebin.com/1CsesuVV)

is wm8731 audio functional with the 3.10.17 kernel com compulab?
The audio works on their pre-canned ubuntu image, which runs 3.10.

Post by Jonathan Bennett
It shows up in aplay -l, but trying to play anything to it gives me a

div

Post by Jonathan Bennett
by 0 in a driver: [<c04f27e4>] (Ldiv0) from [<bf50512c>]
(fsl_ssi_hw_params+0x194/0x364 [snd_soc_fsl_ssi]) This div0 happens in
sound/soc/fsl/fsl_ssi.c, in the function fsl_ssi_set_bclk.
My driver file is an almost direct copy of what compulab shipped with a
3.10 kernel. This is my first attempt at kernel hacking, so any help

would

Post by Jonathan Bennett
be appreciated.

I would suggest as a first step to get the dts for compulab upstreamed.
http://lists.infradead.org/pipermail/linux-arm-kernel/2013-October/204896.html
After the basic dts file is added then you can work on adding audio support.
There is already a small dts file upstreamed. It barely works. I've been

get usb and the second ethernet port working, which I would like to get
pushed upstream. I was hoping to have more of it working before then,
though.

Post by Fabio Estevam
- The dts from 3.10.17 can not be applied 'as-is' into 3.17. Take a
sound {
compatible = "fsl,imx6q-cm-fx6-wm8731",
"fsl,imx-audio-wm8731";
model = "wm8731-audio";
ssi-controller = <&ssi2>;
src-port = <2>;
ext-port = <4>;
audio-codec = <&codec>;
audio-routing = "LOUT", "ROUT", "LLINEIN", "RLINEIN";
};
'src-port' and 'ext-port' are not valid properties.
mux-int-port = <2>;
mux-ext-port = <4>;
Take a look at arch/arm/boot/dts/imx6qdl-sabresd.dtsi for a reference.
Also the codec clocks that are passed into the dts are not correct.
The wm8731 does not acquire any clocks in the mainline driver.
You should also try to use the 'simple-audio-card' binding instead of
creating the imx-wm8731 machine file.
Here is a reference of the arch/arm/boot/dts/vf610-twr.dts where you
can see simple-audio-card in use.
I am trying to use simple-audio-card, without any success. I've been

hammering at this for a couple days now. My current dts looks like
sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
mux-int-port = <2>;
mux-ext-port = <4>;
simple-audio-card,format = "i2s";
simple-audio-card,mclk-fs = <256>;
simple-audio-card,cpu {
sound-dai = <&ssi2>;
};
simple-audio-card,codec {
sound-dai = <&codec>;
};
};
....
/* i2c3 */
&i2c3 {
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_i2c3>;
status = "okay";
compatible = "at24,24c02";
reg = <0x50>;
pagesize = <16>;
};
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
};
};
...
&ssi2 {
fsl,mode = "i2s-master";
status = "okay";
#sound-dai-cells = <0>;
};
# aplay -l
**** List of PLAYBACK Hardware Devices ****
202c000.ssi-wm8731-hifi wm8731-hifi-0 []
Subdevices: 1/1
Subdevice #0: subdevice #0
# speaker-test
speaker-test 1.0.28
Playback device is default
Stream parameters are 48000Hz, S16_LE, 1 channels
Using 16 octaves of pink noise
ASoC: machine hw_params failed: -22
000Hz)
Buffer size range from 128 to 32768
Period size range from 64 to 16384
Using max buffer size 32768
Periods = 4
Unable to set hw params for playback: Invalid argument
Setting of hwparams failed: Invalid argument
It's a wm8731 hanging off an i2c bus from an i.mx6q chip. Seems like
exactly what simple-audio-card is designed for. I imagine that there is
something simple that I am overlooking, but I'm at a loss to find it.

Post by Fabio Estevam
Good luck!
Regards,
Fabio Estevam

Thanks for the help!
~Jonathan Bennett

[Fabio Estevam]

Post by Fabio Estevam
sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
mux-int-port = <2>;
mux-ext-port = <4>;
simple-audio-card,format = "i2s";
simple-audio-card,cpu {
sound-dai = <&ssi2>;
master-clkdir-out;
frame-master;
bitclock-master;
};
simple-audio-card,codec {
frame-master;
bitclock-master;
sound-dai = <&codec>;
};
};
/* i2c3 */
&i2c3 {
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_i2c3>;
status = "okay";
compatible = "at24,24c02";
reg = <0x50>;
pagesize = <16>;
};
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
clocks = <&clks IMX6QDL_CLK_SSI2>;

This 'clocks' entry does not seem correct.

IMX6QDL_CLK_SSI2 is the clock from SSI2 module, which is turned on by
the ssi driver.

Who provides the MCLK for wm8731 on utilite board? Isn't it mx6 CLKO pin?

Are you able to monitor this clock with a scope and make sure that it
is actually being turned on?

We use CLKO output to drive MCLK pins on mx6qdl-sabresd board with
wm8762 codec. You can use it as a reference.

Adding the Compulab's folks on Cc.

[Fabio Estevam]

Post by Fabio Estevam

Post by Jonathan Bennett
clocks = <&clks IMX6QDL_CLK_SSI2>;

This 'clocks' entry does not seem correct.
IMX6QDL_CLK_SSI2 is the clock from SSI2 module, which is turned on by
the ssi driver.
Who provides the MCLK for wm8731 on utilite board? Isn't it mx6 CLKO pin?
Are you able to monitor this clock with a scope and make sure that it
is actually being turned on?
We use CLKO output to drive MCLK pins on mx6qdl-sabresd board with
wm8762 codec. You can use it as a reference.
Adding the Compulab's folks on Cc.

Do the changes below help? (using IMX6QDL_CLK_CKO)

codec: ***@1a {
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
clocks = <&clks IMX6QDL_CLK_CKO>;

with this one applied:

--- a/sound/soc/codecs/wm8731.c
+++ b/sound/soc/codecs/wm8731.c
@@ -13,6 +13,7 @@
* published by the Free Software Foundation.
*/

+#include <linux/clk.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/init.h>
@@ -46,6 +47,7 @@ struct wm8731_priv {
struct regmap *regmap;
struct regulator_bulk_data supplies[WM8731_NUM_SUPPLIES];
const struct snd_pcm_hw_constraint_list *constraints;
+ struct clk *mclk;
unsigned int sysclk;
int sysclk_type;
int playback_fs;
@@ -742,6 +744,21 @@ static int wm8731_i2c_probe(struct i2c_client *i2c,
return ret;
}

+ wm8731->mclk = devm_clk_get(&i2c->dev, NULL);
+ if (IS_ERR(wm8731->mclk)) {
+ ret = PTR_ERR(wm8731->mclk);
+ dev_err(&i2c->dev, "Failed to get mclock: %d\n", ret);
+ /* Defer the probe to see if the clk will be provided later */
+ if (ret == -ENOENT)
+ return -EPROBE_DEFER;
+ }
+
+ if (!IS_ERR(wm8731->mclk)) {
+ ret = clk_prepare_enable(wm8731->mclk);
+ if (ret)
+ return ret;
+ }
+
i2c_set_clientdata(i2c, wm8731);

ret = snd_soc_register_codec(&i2c->dev,

[Jonathan Bennett]

With those changes:
[ 8.656319] wm8731 2-001a: simple-card: set_sysclk error
[ 8.661964] asoc-simple-card sound: ASoC: failed to init
202c000.ssi-wm8731-hifi: -22
[ 8.670067] asoc-simple-card sound: ASoC: failed to instantiate card -22
[ 8.678063] Bluetooth: vendor=0x2df, device=0x911b, class=255, fn=3
[ 8.679996] asoc-simple-card: probe of sound failed with error -22


Looking at the 3.10 driver file and digging through some documentation, it
looks like the wm8731 can be used in either slave or master mode. According
to Compulab's driver, the supplying clocks when the wm8731 is in slave mode
are pll4 (IMX6QDL_CLK_PLL4_AUDIO) and ssi.1 (IMX6QDL_CLK_SSI2)

In master mode, it's IMX6QDL_CLK_CKO and IMX6QDL_CLK_CKO2.

I have not found an actual schematic to see how it is physically wired.
Popped open the case, and the codec is on the bottom side of the board with
the cpu on it, 100% unaccessible during operations. Even at that, the cpu
and codec are bga, so no probing of pins.

Looking at their driver file, I am also wondering if we need to tweak the
audmux init. I've played with this some in the dts, but hasn't made much
difference.

Post by Fabio Estevam

Post by Fabio Estevam

Post by Jonathan Bennett
clocks = <&clks IMX6QDL_CLK_SSI2>;

This 'clocks' entry does not seem correct.
IMX6QDL_CLK_SSI2 is the clock from SSI2 module, which is turned on by
the ssi driver.
Who provides the MCLK for wm8731 on utilite board? Isn't it mx6 CLKO pin?
Are you able to monitor this clock with a scope and make sure that it
is actually being turned on?
We use CLKO output to drive MCLK pins on mx6qdl-sabresd board with
wm8762 codec. You can use it as a reference.
Adding the Compulab's folks on Cc.

Do the changes below help? (using IMX6QDL_CLK_CKO)
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
clocks = <&clks IMX6QDL_CLK_CKO>;
--- a/sound/soc/codecs/wm8731.c
+++ b/sound/soc/codecs/wm8731.c
@@ -13,6 +13,7 @@
* published by the Free Software Foundation.
*/
+#include <linux/clk.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/init.h>
@@ -46,6 +47,7 @@ struct wm8731_priv {
struct regmap *regmap;
struct regulator_bulk_data supplies[WM8731_NUM_SUPPLIES];
const struct snd_pcm_hw_constraint_list *constraints;
+ struct clk *mclk;
unsigned int sysclk;
int sysclk_type;
int playback_fs;
@@ -742,6 +744,21 @@ static int wm8731_i2c_probe(struct i2c_client *i2c,
return ret;
}
+ wm8731->mclk = devm_clk_get(&i2c->dev, NULL);
+ if (IS_ERR(wm8731->mclk)) {
+ ret = PTR_ERR(wm8731->mclk);
+ dev_err(&i2c->dev, "Failed to get mclock: %d\n", ret);
+ /* Defer the probe to see if the clk will be provided later */
+ if (ret == -ENOENT)
+ return -EPROBE_DEFER;
+ }
+
+ if (!IS_ERR(wm8731->mclk)) {
+ ret = clk_prepare_enable(wm8731->mclk);
+ if (ret)
+ return ret;
+ }
+
i2c_set_clientdata(i2c, wm8731);
ret = snd_soc_register_codec(&i2c->dev,

[Valentin Raevsky]

Hi Jonathan,

See my comments in line.

Moreover, I'd appreciate it if you could try the CompuLab 3.10.17 kernel
on your board in order to rule out a hw problem.

Take the latest 3.10.17 source from here:
https://github.com/utilite-computer/linux-kernel
tag: v3.10.17-cm-fx6-1-beta1

Build instructions:

# export LOADADDR=0x10008000
# make cm_fx6_defconfig
# make zImage modules -j8 CONFIG_DEBUG_SECTION_MISMATCH=y
# make imx6q-sbc-fx6m.dtb
# make imx6q-sbc-fx6.dtb

imx6q-sbc-fx6m.dtb - Utilite
imx6q-sbc-fx6.dtb - Evaluation Board

Regards,
Valentin.

Post by Jonathan Bennett
[ 8.656319] wm8731 2-001a: simple-card: set_sysclk error
[ 8.661964] asoc-simple-card sound: ASoC: failed to init
202c000.ssi-wm8731-hifi: -22
[ 8.670067] asoc-simple-card sound: ASoC: failed to instantiate card -22
[ 8.678063] Bluetooth: vendor=0x2df, device=0x911b, class=255, fn=3
[ 8.679996] asoc-simple-card: probe of sound failed with error -22
Looking at the 3.10 driver file and digging through some documentation,
it looks like the wm8731 can be used in either slave or master mode.
According to Compulab's driver, the supplying clocks when the wm8731 is
in slave mode are pll4 (IMX6QDL_CLK_PLL4_AUDIO) and ssi.1 (IMX6QDL_CLK_SSI2)

These clocks are correct for slave mode.

Post by Jonathan Bennett
In master mode, it's IMX6QDL_CLK_CKO and IMX6QDL_CLK_CKO2.

These clocks are correct for master mode, but do not try make it work.
Master mode requires a hw change on the module.

Post by Jonathan Bennett
I have not found an actual schematic to see how it is physically wired.
Popped open the case, and the codec is on the bottom side of the board
with the cpu on it, 100% unaccessible during operations. Even at that,
the cpu and codec are bga, so no probing of pins.
Looking at their driver file, I am also wondering if we need to tweak
the audmux init. I've played with this some in the dts, but hasn't made
much difference.

Post by Fabio Estevam

Post by Jonathan Bennett
clocks = <&clks IMX6QDL_CLK_SSI2>;

This 'clocks' entry does not seem correct.
IMX6QDL_CLK_SSI2 is the clock from SSI2 module, which is turned on by
the ssi driver.
Who provides the MCLK for wm8731 on utilite board? Isn't it mx6 CLKO pin?
Are you able to monitor this clock with a scope and make sure that it
is actually being turned on?
We use CLKO output to drive MCLK pins on mx6qdl-sabresd board with
wm8762 codec. You can use it as a reference.
Adding the Compulab's folks on Cc.

Do the changes below help? (using IMX6QDL_CLK_CKO)
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
clocks = <&clks IMX6QDL_CLK_CKO>;
--- a/sound/soc/codecs/wm8731.c
+++ b/sound/soc/codecs/wm8731.c
@@ -13,6 +13,7 @@
* published by the Free Software Foundation.
*/
+#include <linux/clk.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/init.h>
@@ -46,6 +47,7 @@ struct wm8731_priv {
struct regmap *regmap;
struct regulator_bulk_data supplies[WM8731_NUM_SUPPLIES];
const struct snd_pcm_hw_constraint_list *constraints;
+ struct clk *mclk;
unsigned int sysclk;
int sysclk_type;
int playback_fs;
@@ -742,6 +744,21 @@ static int wm8731_i2c_probe(struct i2c_client *i2c,
return ret;
}
+ wm8731->mclk = devm_clk_get(&i2c->dev, NULL);
+ if (IS_ERR(wm8731->mclk)) {
+ ret = PTR_ERR(wm8731->mclk);
+ dev_err(&i2c->dev, "Failed to get mclock: %d\n", ret);
+ /* Defer the probe to see if the clk will be provided later */
+ if (ret == -ENOENT)
+ return -EPROBE_DEFER;
+ }
+
+ if (!IS_ERR(wm8731->mclk)) {
+ ret = clk_prepare_enable(wm8731->mclk);
+ if (ret)
+ return ret;
+ }
+
i2c_set_clientdata(i2c, wm8731);
ret = snd_soc_register_codec(&i2c->dev,

[Jonathan Bennett]

Post by Fabio Estevam
Hi Jonathan,
See my comments in line.
Moreover, I'd appreciate it if you could try the CompuLab 3.10.17 kernel
on your board in order to rule out a hw problem.
https://github.com/utilite-computer/linux-kernel
tag: v3.10.17-cm-fx6-1-beta1
# export LOADADDR=0x10008000
# make cm_fx6_defconfig
# make zImage modules -j8 CONFIG_DEBUG_SECTION_MISMATCH=y
# make imx6q-sbc-fx6m.dtb
# make imx6q-sbc-fx6.dtb
imx6q-sbc-fx6m.dtb - Utilite
imx6q-sbc-fx6.dtb - Evaluation Board
Regards,
Valentin.

Hey. I have pulled the Ubuntu image and booted it up. Sound worked without
a hitch there. I think we can rule out an actual HW problem.

Post by Fabio Estevam

Post by Jonathan Bennett
[ 8.656319] wm8731 2-001a: simple-card: set_sysclk error
[ 8.661964] asoc-simple-card sound: ASoC: failed to init
202c000.ssi-wm8731-hifi: -22
[ 8.670067] asoc-simple-card sound: ASoC: failed to instantiate card -22
[ 8.678063] Bluetooth: vendor=0x2df, device=0x911b, class=255, fn=3
[ 8.679996] asoc-simple-card: probe of sound failed with error -22
Looking at the 3.10 driver file and digging through some documentation,
it looks like the wm8731 can be used in either slave or master mode.
According to Compulab's driver, the supplying clocks when the wm8731 is
in slave mode are pll4 (IMX6QDL_CLK_PLL4_AUDIO) and ssi.1
(IMX6QDL_CLK_SSI2)

These clocks are correct for slave mode.

Post by Jonathan Bennett
I realized something just last night. I was grabbing changes from the

devel branch of github.com/utilite-computer, thinking that those changes
were based off the vanilla 3.10 kernel. Those changes are actually based on
the Freescale 3.10 kernel. I pulled a Freescale patch that is
obviously missing: "ENGR00276249-1 ARM: imx6q: Add missing baud clock for
ssi" I also switched back to imx-wm8731.c for the driver. With this setup,
the driver registers, and speaker-test thinks it plays, however there is no
actual sound output. I call it progress.

My previous ldiv0 was apparently a result of the missing baudclock in
clk-imx6q.c. s

Post by Fabio Estevam
In master mode, it's IMX6QDL_CLK_CKO and IMX6QDL_CLK_CKO2.
These clocks are correct for master mode, but do not try make it work.
Master mode requires a hw change on the module.

Post by Jonathan Bennett
I have not found an actual schematic to see how it is physically wired.
Popped open the case, and the codec is on the bottom side of the board
with the cpu on it, 100% unaccessible during operations. Even at that,
the cpu and codec are bga, so no probing of pins.
Looking at their driver file, I am also wondering if we need to tweak
the audmux init. I've played with this some in the dts, but hasn't made
much difference.

Post by Fabio Estevam

Post by Jonathan Bennett
clocks = <&clks IMX6QDL_CLK_SSI2>;

This 'clocks' entry does not seem correct.
IMX6QDL_CLK_SSI2 is the clock from SSI2 module, which is turned on

by

Post by Fabio Estevam
the ssi driver.
Who provides the MCLK for wm8731 on utilite board? Isn't it mx6

CLKO pin?

Post by Fabio Estevam
Are you able to monitor this clock with a scope and make sure that

it

Post by Fabio Estevam
is actually being turned on?
We use CLKO output to drive MCLK pins on mx6qdl-sabresd board with
wm8762 codec. You can use it as a reference.
Adding the Compulab's folks on Cc.

Do the changes below help? (using IMX6QDL_CLK_CKO)
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
clocks = <&clks IMX6QDL_CLK_CKO>;
--- a/sound/soc/codecs/wm8731.c
+++ b/sound/soc/codecs/wm8731.c
@@ -13,6 +13,7 @@
* published by the Free Software Foundation.
*/
+#include <linux/clk.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/init.h>
@@ -46,6 +47,7 @@ struct wm8731_priv {
struct regmap *regmap;
struct regulator_bulk_data supplies[WM8731_NUM_SUPPLIES];
const struct snd_pcm_hw_constraint_list *constraints;
+ struct clk *mclk;
unsigned int sysclk;
int sysclk_type;
int playback_fs;
@@ -742,6 +744,21 @@ static int wm8731_i2c_probe(struct i2c_client *i2c,
return ret;
}
+ wm8731->mclk = devm_clk_get(&i2c->dev, NULL);
+ if (IS_ERR(wm8731->mclk)) {
+ ret = PTR_ERR(wm8731->mclk);
+ dev_err(&i2c->dev, "Failed to get mclock: %d\n", ret);
+ /* Defer the probe to see if the clk will be provided later */
+ if (ret == -ENOENT)
+ return -EPROBE_DEFER;
+ }
+
+ if (!IS_ERR(wm8731->mclk)) {
+ ret = clk_prepare_enable(wm8731->mclk);
+ if (ret)
+ return ret;
+ }
+
i2c_set_clientdata(i2c, wm8731);
ret = snd_soc_register_codec(&i2c->dev,

[Fabio Estevam]

I realized something just last night. I was grabbing changes from the devel
branch of github.com/utilite-computer, thinking that those changes were
based off the vanilla 3.10 kernel. Those changes are actually based on the
"ENGR00276249-1 ARM: imx6q: Add missing baud clock for ssi" I also switched

The baud clock patch is present in 3.18-rc1:

commit 935632e9938e3e286b9c62a7e54d35c428533cc9
Author: Shengjiu Wang <***@freescale.com>
Date: Tue Sep 9 17:13:26 2014 +0800

ARM: dts: imx6qdl: add baud clock and clock-names for ssi

Baud clock is used for bit clock generation in master mode. Ipg clock
is peripheral clock and peripheral access clock.

back to imx-wm8731.c for the driver. With this setup, the driver registers,
and speaker-test thinks it plays, however there is no actual sound output. I
call it progress.

Great progress: some things to check :

1. audmux settings

2. Can you try to run 'alsamixer' and make sure that the volume is not
at the lowest level?

[Jonathan Bennett]

Post by Jonathan Bennett

Post by Jonathan Bennett
I realized something just last night. I was grabbing changes from the

devel

Post by Jonathan Bennett
branch of github.com/utilite-computer, thinking that those changes were
based off the vanilla 3.10 kernel. Those changes are actually based on

the

Post by Jonathan Bennett
Freescale 3.10 kernel. I pulled a Freescale patch that is obviously
"ENGR00276249-1 ARM: imx6q: Add missing baud clock for ssi" I also

switched
commit 935632e9938e3e286b9c62a7e54d35c428533cc9
Date: Tue Sep 9 17:13:26 2014 +0800
ARM: dts: imx6qdl: add baud clock and clock-names for ssi
Baud clock is used for bit clock generation in master mode. Ipg clock
is peripheral clock and peripheral access clock.

Awesome!
It's great to see this stuff go mainline.

Post by Jonathan Bennett

Post by Jonathan Bennett
back to imx-wm8731.c for the driver. With this setup, the driver

registers,

Post by Jonathan Bennett
and speaker-test thinks it plays, however there is no actual sound

output. I

Post by Jonathan Bennett
call it progress.

1. audmux settings

Well, the imx-wm8731 driver (It's really just glue for loading wm8731) does
have audmux settings in it. These get called during the module probe. Not
sure what might need to change there. My thoughts are that they should be
correct.

Post by Jonathan Bennett
2. Can you try to run 'alsamixer' and make sure that the volume is not
at the lowest level?

Yeah, I've made sure to check alsamixer and turn the volume up. Still no
joy.

[Jonathan Bennett]

Post by Jonathan Bennett

Post by Jonathan Bennett

Post by Jonathan Bennett
I realized something just last night. I was grabbing changes from the

devel

Post by Jonathan Bennett
branch of github.com/utilite-computer, thinking that those changes were
based off the vanilla 3.10 kernel. Those changes are actually based on

the

Post by Jonathan Bennett
Freescale 3.10 kernel. I pulled a Freescale patch that is obviously
"ENGR00276249-1 ARM: imx6q: Add missing baud clock for ssi" I also

switched
commit 935632e9938e3e286b9c62a7e54d35c428533cc9
Date: Tue Sep 9 17:13:26 2014 +0800
ARM: dts: imx6qdl: add baud clock and clock-names for ssi
Baud clock is used for bit clock generation in master mode. Ipg clock
is peripheral clock and peripheral access clock.

Awesome!
It's great to see this stuff go mainline.

I have rebased to 3.18 rc-1. It has simplified the patch set a bit, as more
of the Freescale things have hit mainline. http://pastebin.com/C652GKe3 is
my patchset.
Again, the card registers and thinks it plays, but there is no actual
output.

Post by Jonathan Bennett

Post by Jonathan Bennett

Post by Jonathan Bennett
back to imx-wm8731.c for the driver. With this setup, the driver

registers,

Post by Jonathan Bennett
and speaker-test thinks it plays, however there is no actual sound

output. I

Post by Jonathan Bennett
call it progress.

1. audmux settings

Well, the imx-wm8731 driver (It's really just glue for loading wm8731)
does have audmux settings in it. These get called during the module probe.
Not sure what might need to change there. My thoughts are that they should
be correct.

Post by Jonathan Bennett
2. Can you try to run 'alsamixer' and make sure that the volume is not
at the lowest level?

Yeah, I've made sure to check alsamixer and turn the volume up. Still no
joy.

[Valentin Raevsky]

On Tue, Oct 21, 2014 at 4:04 PM, Jonathan Bennett

I realized something just last night. I was grabbing changes from the devel
branch ofgithub.com/utilite-computer

<http://github.com/utilite-computer>, thinking that those
changes were

based off the vanilla 3.10 kernel. Those changes are actually based on the
"ENGR00276249-1 ARM: imx6q: Add missing baud clock for ssi" I also switched

commit 935632e9938e3e286b9c62a7e54d35c428533cc9
Date: Tue Sep 9 17:13:26 2014 +0800
ARM: dts: imx6qdl: add baud clock and clock-names for ssi
Baud clock is used for bit clock generation in master mode. Ipg clock
is peripheral clock and peripheral access clock.
Awesome!
It's great to see this stuff go mainline.
I have rebased to 3.18 rc-1. It has simplified the patch set a bit, as
more of the Freescale things have hit mainline.
http://pastebin.com/C652GKe3 is my patchset.
Again, the card registers and thinks it plays, but there is no actual
output.

Make sure that "Output Mixer HiFi Playback Switch" is on
Issue:
# amixer cset numid=14,iface=MIXER,name='Output Mixer HiFi Playback
Switch' on
# alsactl store wm8731audio

back to imx-wm8731.c for the driver. With this setup, the driver registers,
and speaker-test thinks it plays, however there is no actual sound output. I
call it progress.

1. audmux settings
Well, the imx-wm8731 driver (It's really just glue for loading
wm8731) does have audmux settings in it. These get called during the
module probe. Not sure what might need to change there. My thoughts
are that they should be correct.
2. Can you try to run 'alsamixer' and make sure that the volume is not
at the lowest level?
Yeah, I've made sure to check alsamixer and turn the volume up.
Still no joy.

[Jonathan Bennett]

Post by Valentin Raevsky

On Tue, Oct 21, 2014 at 4:04 PM, Jonathan Bennett

Post by Jonathan Bennett
I realized something just last night. I was grabbing changes

from the devel

Post by Jonathan Bennett
branch ofgithub.com/utilite-computer

<http://github.com/utilite-computer>, thinking that those
changes were

Post by Jonathan Bennett
based off the vanilla 3.10 kernel. Those changes are actually

based on the

Post by Jonathan Bennett
Freescale 3.10 kernel. I pulled a Freescale patch that is
"ENGR00276249-1 ARM: imx6q: Add missing baud clock for ssi" I

also switched
commit 935632e9938e3e286b9c62a7e54d35c428533cc9
Date: Tue Sep 9 17:13:26 2014 +0800
ARM: dts: imx6qdl: add baud clock and clock-names for ssi
Baud clock is used for bit clock generation in master mode.
Ipg clock
is peripheral clock and peripheral access clock.
Awesome!
It's great to see this stuff go mainline.
I have rebased to 3.18 rc-1. It has simplified the patch set a bit, as
more of the Freescale things have hit mainline.
http://pastebin.com/C652GKe3 is my patchset.
Again, the card registers and thinks it plays, but there is no actual
output.

Make sure that "Output Mixer HiFi Playback Switch" is on
# amixer cset numid=14,iface=MIXER,name='Output Mixer HiFi Playback
Switch' on
# alsactl store wm8731audio

BRILLIANT!!!
We've got sound. Thank you for the answer.
Questions:
First, what is the best way to set that to on by default? I know "alsactl
store wm8731audio" Makes it survive reboots. On a fresh install, though,
that should probably default to on.

Second: I would love to see support for this get mainlined. Should I go
back and try to get simple-audio-card working, or stick with the compulab
glue driver?

Post by Valentin Raevsky

Post by Jonathan Bennett
back to imx-wm8731.c for the driver. With this setup, the

driver registers,

Post by Jonathan Bennett
and speaker-test thinks it plays, however there is no actual

sound output. I

Post by Jonathan Bennett
call it progress.

1. audmux settings
Well, the imx-wm8731 driver (It's really just glue for loading
wm8731) does have audmux settings in it. These get called during the
module probe. Not sure what might need to change there. My thoughts
are that they should be correct.
2. Can you try to run 'alsamixer' and make sure that the volume is not
at the lowest level?
Yeah, I've made sure to check alsamixer and turn the volume up.
Still no joy.

[Fabio Estevam]

Hi Jonathan,

On Wed, Oct 22, 2014 at 10:13 AM, Jonathan Bennett

Post by Jonathan Bennett
BRILLIANT!!!
We've got sound. Thank you for the answer.

Good job :-)

Post by Jonathan Bennett
First, what is the best way to set that to on by default? I know "alsactl
store wm8731audio" Makes it survive reboots. On a fresh install, though,
that should probably default to on.
Second: I would love to see support for this get mainlined. Should I go back
and try to get simple-audio-card working, or stick with the compulab glue
driver?

Yes, please try implementing simple-audio-card support.

[Jonathan Bennett]

Post by Fabio Estevam
Hi Jonathan,
On Wed, Oct 22, 2014 at 10:13 AM, Jonathan Bennett

Post by Jonathan Bennett
BRILLIANT!!!
We've got sound. Thank you for the answer.

Good job :-)

Post by Jonathan Bennett
First, what is the best way to set that to on by default? I know "alsactl
store wm8731audio" Makes it survive reboots. On a fresh install, though,
that should probably default to on.
Second: I would love to see support for this get mainlined. Should I go

back

Post by Jonathan Bennett
and try to get simple-audio-card working, or stick with the compulab glue
driver?

Yes, please try implementing simple-audio-card support.

Applying what I've learned back to the simple-audio-card, I can get a
muffled pop in the headphones, like something is powering up. That's about
it. When playing to the card, things think they play fine, but no real
audio. I've checked volume levels, and also used Valentin's hint.

Sound bits from dts:

sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
simple-audio-card,format = "i2s";
simple-audio-card,audio-routing =
"Line Out", "LOUT",
"Line Out", "ROUT",
"LLINEIN","Line In",
"RLINEIN","Line In";
simple-audio-card,cpu {
sound-dai = <&ssi2>;
bitclock-master;
frame-master;
};
simple-audio-card,codec {
sound-dai = <&codec>;
};
};

/* i2c3 */
&i2c3 {
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_i2c3>;
status = "okay";
***@50 {
compatible = "at24,24c02";
reg = <0x50>;
pagesize = <16>;
};

codec: ***@1a {
#sound-dai-cells = <0>;
compatible = "wlf,wm8731";
reg = <0x1a>;
AVDD-supply = <&reg_3p3v>;
HPVDD-supply = <&reg_3p3v>;
DCVDD-supply = <&reg_3p3v>;
DBVDD-supply = <&reg_3p3v>;
};
};

&ssi2 {
status = "okay";
#sound-dai-cells = <0>;
};


&audmux {
status = "okay";
ssi2 {
fsl,audmux-port = <1>;
fsl,port-config = <
(IMX_AUDMUX_V2_PTCR_SYN |
IMX_AUDMUX_V2_PTCR_TFSDIR |
IMX_AUDMUX_V2_PTCR_TFSEL(3) |
IMX_AUDMUX_V2_PTCR_RCLKDIR |
IMX_AUDMUX_V2_PTCR_RCSEL(3 | 0x8) |
IMX_AUDMUX_V2_PTCR_TCLKDIR |
IMX_AUDMUX_V2_PTCR_TCSEL(3))
IMX_AUDMUX_V2_PDCR_RXDSEL(3)

Post by Fabio Estevam
;

};
pins4 {
fsl,audmux-port = <3>;
fsl,port-config = <
(IMX_AUDMUX_V2_PTCR_SYN |
IMX_AUDMUX_V2_PTCR_RCLKDIR |
IMX_AUDMUX_V2_PTCR_RCSEL(1 | 0x8) |
IMX_AUDMUX_V2_PTCR_TCLKDIR |
IMX_AUDMUX_V2_PTCR_TCSEL(1))
IMX_AUDMUX_V2_PDCR_RXDSEL(1)

Post by Fabio Estevam
;

};

};

[Fabio Estevam]

On Wed, Oct 22, 2014 at 11:32 AM, Jonathan Bennett

Post by Jonathan Bennett
Applying what I've learned back to the simple-audio-card, I can get a
muffled pop in the headphones, like something is powering up. That's about
it. When playing to the card, things think they play fine, but no real
audio. I've checked volume levels, and also used Valentin's hint.
sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
simple-audio-card,format = "i2s";
simple-audio-card,audio-routing =
"Line Out", "LOUT",
"Line Out", "ROUT",
"LLINEIN","Line In",
"RLINEIN","Line In";
simple-audio-card,cpu {
sound-dai = <&ssi2>;
bitclock-master;
frame-master;
};
simple-audio-card,codec {
sound-dai = <&codec>;
};
};

What about changing this part to?

sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
simple-audio-card,format = "i2s";
simple-audio-card,bitclock-master = <&dailink0_master>;
simple-audio-card,frame-master = <&dailink0_master>;
simple-audio-card,audio-routing =
"Line Out", "LOUT",
"Line Out", "ROUT",
"LLINEIN","Line In",
"RLINEIN","Line In";

simple-audio-card,cpu {
sound-dai = <&ssi2>;
};

dailink0_master: simple-audio-card,codec {
sound-dai = <&codec>;
clocks = <&clks IMX6QDL_CLK_CKO>;
};
};

[Fabio Estevam]

Post by Fabio Estevam
What about changing this part to?

Forgot that you use ssi as master and codec as slave, so something
like this maybe:

sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
simple-audio-card,format = "i2s";
simple-audio-card,bitclock-master = <&dailink0_master>;
simple-audio-card,frame-master = <&dailink0_master>;
simple-audio-card,audio-routing =
"Line Out", "LOUT",
"Line Out", "ROUT",
"LLINEIN","Line In",
"RLINEIN","Line In";

dailink0_master: simple-audio-card,cpu {
sound-dai = <&ssi2>;
clocks = <&clks IMX6QDL_CLK_CKO>;
};

simple-audio-card,codec {
sound-dai = <&codec>;

};
};

[Jonathan Bennett]

Post by Fabio Estevam

Post by Fabio Estevam
What about changing this part to?

Forgot that you use ssi as master and codec as slave, so something
sound {
compatible = "simple-audio-card";
simple-audio-card,model = "wm8731-audio";
simple-audio-card,format = "i2s";
simple-audio-card,bitclock-master = <&dailink0_master>;
simple-audio-card,frame-master = <&dailink0_master>;
simple-audio-card,audio-routing =
"Line Out", "LOUT",
"Line Out", "ROUT",
"LLINEIN","Line In",
"RLINEIN","Line In";
dailink0_master: simple-audio-card,cpu {
sound-dai = <&ssi2>;
clocks = <&clks IMX6QDL_CLK_CKO>;

Tried several different clocks here, didn't seem to make a difference. Open
to thoughts here, but I'm beginning to be inclined to stick with the glue
driver that works. I'd eventually like to get this stuff mainlined, but
want to make sure to do it correctly.

Post by Fabio Estevam
};
simple-audio-card,codec {
sound-dai = <&codec>;
};
};