ASoC: cs4349: Add support for Cirrus Logic CS4349 breaks on multiple arch

Hi Tim,

This commit:

e40da86a37f64c73b810bc7a63d77c44dc61accb is the first bad commit
commit e40da86a37f64c73b810bc7a63d77c44dc61accb
Author: Tim Howe [off-list ref]
Date:   Thu Jul 16 14:51:40 2015 -0500

    ASoC: cs4349: Add support for Cirrus Logic CS4349

    Signed-off-by: Tim Howe [off-list ref]
    Signed-off-by: Mark Brown [off-list ref]

is showing up in linux-next build failures for mips, parisc and s390.

sound/soc/codecs/cs4349.c:300:2: error: implicit declaration of function
'devm_gpiod_get_optional' [-Werror=implicit-function-declaration]
sound/soc/codecs/cs4349.c:301:12: error: 'GPIOD_OUT_LOW' undeclared
(first use in this function)
sound/soc/codecs/cs4349.c:306:3: error: implicit declaration of function
'gpiod_set_value_cansleep' [-Werror=implicit-function-declaration]

Here are a few examples:

http://kisskb.ellerman.id.au/kisskb/buildresult/12465322/
http://kisskb.ellerman.id.au/kisskb/buildresult/12465445/
http://kisskb.ellerman.id.au/kisskb/buildresult/12465246/

I'm guessing it needs some tighter Kconfig dependencies to block it
from being selected on arch that don't have/enable gpio stuff.

Paul.

Hi Paul,

On Sat, Jul 18, 2015 at 4:28 PM, Paul Gortmaker
[off-list ref] wrote:

Hi Tim,

This commit:

e40da86a37f64c73b810bc7a63d77c44dc61accb is the first bad commit
commit e40da86a37f64c73b810bc7a63d77c44dc61accb
Author: Tim Howe [off-list ref]
Date:   Thu Jul 16 14:51:40 2015 -0500

    ASoC: cs4349: Add support for Cirrus Logic CS4349

    Signed-off-by: Tim Howe [off-list ref]
    Signed-off-by: Mark Brown [off-list ref]

is showing up in linux-next build failures for mips, parisc and s390.

sound/soc/codecs/cs4349.c:300:2: error: implicit declaration of function
'devm_gpiod_get_optional' [-Werror=implicit-function-declaration]
sound/soc/codecs/cs4349.c:301:12: error: 'GPIOD_OUT_LOW' undeclared
(first use in this function)
sound/soc/codecs/cs4349.c:306:3: error: implicit declaration of function
'gpiod_set_value_cansleep' [-Werror=implicit-function-declaration]

Here are a few examples:

http://kisskb.ellerman.id.au/kisskb/buildresult/12465322/
http://kisskb.ellerman.id.au/kisskb/buildresult/12465445/
http://kisskb.ellerman.id.au/kisskb/buildresult/12465246/

I'm guessing it needs some tighter Kconfig dependencies to block it
from being selected on arch that don't have/enable gpio stuff.

Would this fix the problem?

+++ b/sound/soc/codecs/Kconfig

@@ -53,7 +53,7 @@ config SND_SOC_ALL_CODECS
        select SND_SOC_CS4271_I2C if I2C
        select SND_SOC_CS4271_SPI if SPI_MASTER
        select SND_SOC_CS42XX8_I2C if I2C
-       select SND_SOC_CS4349 if I2C
+       select SND_SOC_CS4349 if I2C && GPIOLIB
        select SND_SOC_CX20442 if TTY
        select SND_SOC_DA7210 if SND_SOC_I2C_AND_SPI
        select SND_SOC_DA7213 if I2C

On Sat, Jul 18, 2015 at 5:38 PM, Fabio Estevam [off-list ref] wrote:

Or this one:

--- a/sound/soc/codecs/Kconfig
+++ b/sound/soc/codecs/Kconfig

@@ -53,7 +53,7 @@ config SND_SOC_ALL_CODECS
        select SND_SOC_CS4271_I2C if I2C
        select SND_SOC_CS4271_SPI if SPI_MASTER
        select SND_SOC_CS42XX8_I2C if I2C
-       select SND_SOC_CS4349 if I2C
+       select SND_SOC_CS4349 if I2C && GPIOLIB
        select SND_SOC_CX20442 if TTY
        select SND_SOC_DA7210 if SND_SOC_I2C_AND_SPI
        select SND_SOC_DA7213 if I2C

@@ -410,7 +410,7 @@ config SND_SOC_CS42XX8_I2C
 # Cirrus Logic CS4349 HiFi DAC
 config SND_SOC_CS4349
        tristate "Cirrus Logic CS4349 CODEC"
-       depends on I2C
+       depends on I2C && GPIOLIB

 config SND_SOC_CX20442
        tristate

On Sat, Jul 18, 2015 at 06:04:48PM -0300, Fabio Estevam wrote:

-       select SND_SOC_CS4349 if I2C
+       select SND_SOC_CS4349 if I2C && GPIOLIB

We don't need Kconfig, we need to include the relevant header for the
stubs - gpiolib is like regulator and should stub itself out.  akpm has
a fix in his tree but he hasn't sent it as a patch so I need to go fish
it out of the notification e-mail which will probably happen on Monday
or something.

Though quite why an arch would have I2C and not GPIOs...