[PATCH] ASoC: avoid unused variable warning for rt5659_acpi_match

[PATCH] ASoC: avoid unused variable warning for rt5659_acpi_match

[Arnd Bergmann]

The newly added rt5659 codec driver unconditionally defines an
ACPI device match table but then uses ACPI_PTR() to remove the
only reference to it, so we get a harmless build warning:

sound/soc/codecs/rt5659.c:4200:30: warning: 'rt5659_acpi_match' defined but not used [-Wunused-variable]
 static struct acpi_device_id rt5659_acpi_match[] = {

This changes both the OF match table and the ACPI match table
to follow the same style, using ACPI_PTR/of_match_ptr to
make the reference conditional, and using an #ifdef to hide
the table. This also adds the missing MODULE_DEVICE_TABLE for
the OF case and adapts the formatting to the same style.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
v3: actually builds with CONFIG_OF enabled, testing in a couple of
configurations with and without OF

diff --git a/sound/soc/codecs/rt5659.c b/sound/soc/codecs/rt5659.c
index c166d9394c69..fb8ea05c0de1 100644
--- a/sound/soc/codecs/rt5659.c
+++ b/sound/soc/codecs/rt5659.c
@@ -4183,24 +4183,29 @@ void rt5659_i2c_shutdown(struct i2c_client *client)
 	regmap_write(rt5659->regmap, RT5659_RESET, 0);
 }
 
+#ifdef CONFIG_OF
 static const struct of_device_id rt5659_of_match[] = {
 	{ .compatible = "realtek,rt5658", },
 	{ .compatible = "realtek,rt5659", },
-	{},
+	{ },
 };
+MODULE_DEVICE_TABLE(of, rt5659_of_match);
+#endif
 
+#ifdef CONFIG_ACPI
 static struct acpi_device_id rt5659_acpi_match[] = {
-		{ "10EC5658", 0},
-		{ "10EC5659", 0},
-		{ },
+	{ "10EC5658", 0, },
+	{ "10EC5659", 0, },
+	{ },
 };
 MODULE_DEVICE_TABLE(acpi, rt5659_acpi_match);
+#endif
 
 struct i2c_driver rt5659_i2c_driver = {
 	.driver = {
 		.name = "rt5659",
 		.owner = THIS_MODULE,
-		.of_match_table = rt5659_of_match,
+		.of_match_table = of_match_ptr(rt5659_of_match),
 		.acpi_match_table = ACPI_PTR(rt5659_acpi_match),
 	},
 	.probe = rt5659_i2c_probe,

[PATCH] ASoC: avoid unused variable warning for rt5659_acpi_match

[Arnd Bergmann]

The newly added rt5659 codec driver unconditionally defines an
ACPI device match table but then uses ACPI_PTR() to remove the
only reference to it, so we get a harmless build warning:

sound/soc/codecs/rt5659.c:4200:30: warning: 'rt5659_acpi_match' defined but not used [-Wunused-variable]
 static struct acpi_device_id rt5659_acpi_match[] = {

This removes the ACPI_PTR() to avoid the warning.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
This is a harmless regression against v4.4, found on ARM randconfig builds

diff --git a/sound/soc/codecs/rt5659.c b/sound/soc/codecs/rt5659.c
index c166d9394c69..5a1d789ba58d 100644
--- a/sound/soc/codecs/rt5659.c
+++ b/sound/soc/codecs/rt5659.c
@@ -4201,7 +4201,7 @@ struct i2c_driver rt5659_i2c_driver = {
 		.name = "rt5659",
 		.owner = THIS_MODULE,
 		.of_match_table = rt5659_of_match,
-		.acpi_match_table = ACPI_PTR(rt5659_acpi_match),
+		.acpi_match_table = rt5659_acpi_match,
 	},
 	.probe = rt5659_i2c_probe,
 	.remove = rt5659_i2c_remove,

[PATCH] ASoC: avoid unused variable warning for rt5659_acpi_match

[Mark Brown]

On Wed, Jan 20, 2016 at 11:43:48AM +0100, Arnd Bergmann wrote:
> The newly added rt5659 codec driver unconditionally defines an
> ACPI device match table but then uses ACPI_PTR() to remove the
> only reference to it, so we get a harmless build warning:

> sound/soc/codecs/rt5659.c:4200:30: warning: 'rt5659_acpi_match' defined but not used [-Wunused-variable]
>  static struct acpi_device_id rt5659_acpi_match[] = {

> This removes the ACPI_PTR() to avoid the warning.

Why is this a better fix than conditionally defining the table?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 473 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20160120/ce9df5db/attachment.sig>

[PATCH] ASoC: avoid unused variable warning for rt5659_acpi_match

[Arnd Bergmann]

On Wednesday 20 January 2016 10:45:55 Mark Brown wrote:
> On Wed, Jan 20, 2016 at 11:43:48AM +0100, Arnd Bergmann wrote:
> > The newly added rt5659 codec driver unconditionally defines an
> > ACPI device match table but then uses ACPI_PTR() to remove the
> > only reference to it, so we get a harmless build warning:
> 
> > sound/soc/codecs/rt5659.c:4200:30: warning: 'rt5659_acpi_match' defined but not used [-Wunused-variable]
> >  static struct acpi_device_id rt5659_acpi_match[] = {
> 
> > This removes the ACPI_PTR() to avoid the warning.
> 
> Why is this a better fix than conditionally defining the table?

I'm not overly fond of adding #ifdef if it can be avoided. In this case,
both approaches seemed reasonable (either add an #ifdef or waste a couple
of bytes), and I picked at random. I'll send you the alternative as well,
please apply whichever one you prefer.

	Arnd