Skip to contentSkip to conversation

Input: stmpe-ts - mark OF related data as maybe unused

PATCH In review Arnd Bergmann 3 participants 5 messages Feb 25, 2025

Subsystems (18)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 Input: stmpe-ts - mark OF related data as maybe unused

From: Krzysztof Kozlowski <[email protected]>

When compile tested with W=1 on x86_64 with driver as built-in:

stmpe-ts.c:371:34: error: unused variable 'stmpe_ts_ids' [-Werror,-Wunused-const-variable]

Ideally this would be referenced from the platform_driver, but since the compatible string is already matched by the mfd driver for its parent device, that would break probing.

In this case, the of_device_id table just serves as a module alias for loading the driver, while the device itself is probed using the platform device name.

---
 drivers/input/touchscreen/stmpe-ts.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/input/touchscreen/stmpe-ts.c +1 -1
index a94a1997f96b..3900aa2e3a90 100644
@@ -366,7 +366,7 @@ static struct platform_driver stmpe_ts_driver = {
 };
 module_platform_driver(stmpe_ts_driver);
 
-static const struct of_device_id stmpe_ts_ids[] = {
+static const struct of_device_id stmpe_ts_ids[] __maybe_unused = {
 	{ .compatible = "st,stmpe-ts", },
 	{ },
 };
Signature · 1 line
2.39.5
#1 lore raw
Quoted 26 lines · Arnd Bergmann

From: Krzysztof Kozlowski <[email protected]>

When compile tested with W=1 on x86_64 with driver as built-in:

stmpe-ts.c:371:34: error: unused variable 'stmpe_ts_ids' [-Werror,-Wunused-const-variable]

Ideally this would be referenced from the platform_driver, but since the compatible string is already matched by the mfd driver for its parent device, that would break probing.

In this case, the of_device_id table just serves as a module alias for loading the driver, while the device itself is probed using the platform device name.

---
 drivers/input/touchscreen/stmpe-ts.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/input/touchscreen/stmpe-ts.c +0 -0
index a94a1997f96b..3900aa2e3a90 100644

drivers/input/touchscreen/stmpe-ts.c static struct platform_driver stmpe_ts_driver = { · line 372

Quoted 8 lines
drivers/input/touchscreen/stmpe-ts.c +1 -1
@@ -366,7 +366,7 @@ static struct platform_driver stmpe_ts_driver = {
 };
 module_platform_driver(stmpe_ts_driver);
 
-static const struct of_device_id stmpe_ts_ids[] = {
+static const struct of_device_id stmpe_ts_ids[] __maybe_unused = {
 	{ .compatible = "st,stmpe-ts", },
 	{ },
 };

Following this we have

	MODULE_DEVICE_TABLE(of, stmpe_ts_ids);

.

With

include/linux/module.h +2 -1
index 30e5b19bafa9..014f033ef1ba 100644
@@ -250,7 +250,8 @@ extern void cleanup_module(void);
 extern typeof(name) __mod_device_table__##type##__##name		\
   __attribute__ ((unused, alias(__stringify(name))))
 #else  /* !MODULE */
-#define MODULE_DEVICE_TABLE(type, name)
+#define MODULE_DEVICE_TABLE(type, name)					\
+static const typeof(name) *__mod_device_table__##type##__##name##_ptr __attribute__((unused)) = &(name)
 #endif
 
 /* Version of form [<epoch>:]<version>[-<extra-version>].

the warning goes away and stmpe_ts_ids isn't included in the .o file without having to add __maybe_unused to the driver.

I would consider that a superior approach.

Best regards
Uwe

maintainer
#2 lore raw
Quoted 21 lines · Uwe Kleine-König

drivers/input/touchscreen/

Quoted 1 line · Arnd Bergmann
drivers/input/touchscreen/ renamed from drivers/input/touchscreen/stmpe-ts.c +0 -0

With

include/linux/module.h +2 -1
index 30e5b19bafa9..014f033ef1ba 100644
@@ -250,7 +250,8 @@ extern void cleanup_module(void);
 extern typeof(name) __mod_device_table__##type##__##name		\
   __attribute__ ((unused, alias(__stringify(name))))
 #else  /* !MODULE */
-#define MODULE_DEVICE_TABLE(type, name)
+#define MODULE_DEVICE_TABLE(type, name)					\
+static const typeof(name) *__mod_device_table__##type##__##name##_ptr 
 __attribute__((unused)) = &(name)
 #endif
 

/* Version of form [<epoch>:]<version>[-<extra-version>].

Quoted 4 lines

the warning goes away and stmpe_ts_ids isn't included in the .o file without having to add __maybe_unused to the driver.

I would consider that a superior approach.

Not sure, I can see how this avoids some warnings, but this is currently the only remaining instance of this problem (I fixed another two recently), and in most cases a MODULE_DEVICE_TABLE() entry that is completely unused ends up pointing to a real bug, where there is a table but it's not also part of the device_driver definition.

Arnd

#3 lore raw

Hello Arnd,

Quoted 20 lines · Arnd Bergmann
Quoted 19 lines · Uwe Kleine-König

drivers/input/touchscreen/

Quoted 1 line · Arnd Bergmann
drivers/input/touchscreen/ renamed from drivers/input/touchscreen/stmpe-ts.c +0 -0

With

include/linux/module.h +2 -1
index 30e5b19bafa9..014f033ef1ba 100644
@@ -250,7 +250,8 @@ extern void cleanup_module(void);
 extern typeof(name) __mod_device_table__##type##__##name		\
   __attribute__ ((unused, alias(__stringify(name))))
 #else  /* !MODULE */
-#define MODULE_DEVICE_TABLE(type, name)
+#define MODULE_DEVICE_TABLE(type, name)					\
+static const typeof(name) *__mod_device_table__##type##__##name##_ptr 
 __attribute__((unused)) = &(name)
 #endif
 
Quoted 1 line
Quoted 1 line

/* Version of form [<epoch>:]<version>[-<extra-version>].

Hu?

Quoted 11 lines
Quoted 4 lines

the warning goes away and stmpe_ts_ids isn't included in the .o file without having to add __maybe_unused to the driver.

I would consider that a superior approach.

Not sure, I can see how this avoids some warnings, but this is currently the only remaining instance of this problem (I fixed another two recently), and in most cases a MODULE_DEVICE_TABLE() entry that is completely unused ends up pointing to a real bug, where there is a table but it's not also part of the device_driver definition.

It might be the only instance without __maybe_unused and so triggering a warning. But there is also:

$ git grep -E 'of_device_id.*__maybe_unused' | wc -l
231
$ git grep -E 'mdio_device_id.*__maybe_unused' | wc -l
58

Best regards
Uwe

maintainer
#4 lore raw
Quoted 15 lines · Uwe Kleine-König
Quoted 5 lines · Arnd Bergmann
Quoted 4 lines · Uwe Kleine-König

On Tue, Feb 25, 2025 at 03:53:26PM +0100, Arnd Bergmann wrote:
the warning goes away and stmpe_ts_ids isn't included in the .o file without having to add __maybe_unused to the driver.

Quoted 8 lines
Quoted 1 line

I would consider that a superior approach.

Not sure, I can see how this avoids some warnings, but this is currently the only remaining instance of this problem (I fixed another two recently), and in most cases a MODULE_DEVICE_TABLE() entry that is completely unused ends up pointing to a real bug, where there is a table but it's not also part of the device_driver definition.

Quoted 8 lines

It might be the only instance without __maybe_unused and so triggering a warning. But there is also:

$ git grep -E 'of_device_id.*__maybe_unused' | wc -l
231
$ git grep -E 'mdio_device_id.*__maybe_unused' | wc -l
58

I'm not really worried about these at the moment, other than not wanting to pile on to that mess with more __maybe_unused annotations.

My goal here is to get the point of enabling -Wunused-const-variable by default in order to find other bugs before they make it into the kernel.

Andy Shevchenko really wants to remove the of_match_ptr() macro so we can stop adding pointless __maybe_unused annotations for every driver that accidentally uses of_match_ptr(). This is certainly a good idea as well, just not what I'm trying to do this time.

Apparently we have already accumulated a bunch of drivers that ended up with __maybe_unused but no actual reference from of_match_ptr():

$ git grep -l 'of_device_id.*__maybe_unused' |xargs grep -Lw of_match_ptr | wc -l

but only a couple of drivers that don't use of_match_ptr() or of_match_node():

$ git grep -l 'of_device_id.*__maybe_unused' |xargs grep -Lw 'of_match_table\|of_match_node'
drivers/cpufreq/armada-37xx-cpufreq.c
drivers/cpufreq/armada-8k-cpufreq.c
drivers/cpufreq/highbank-cpufreq.c
drivers/cpufreq/sti-cpufreq.c
drivers/hwmon/isl28022.c
drivers/input/touchscreen/stmpe-ts.c
drivers/mfd/twl6030-irq.c
drivers/tty/serial/sc16is7xx.c

I do think it makes sense to change of_match_node() to have a reference to its arguments, as in the patch below. That probably needs a few extra fixups.

Arnd

include/linux/of.h +5 -1
index 9d6b8a61607f..83cfa6c26ee4 100644
@@ -907,7 +907,11 @@ static inline const void *of_device_get_match_data(const struct device *dev)
 }
 
 #define of_match_ptr(_ptr)	NULL
-#define of_match_node(_matches, _node)	NULL
+static inline const struct of_device_id *of_match_node(
+	const struct of_device_id *matches, const struct device_node *node)
+{
+	return NULL;
+}
 #endif /* CONFIG_OF */
 
 /* Default string compare functions, Allow arch asm/prom.h to override */
drivers/crypto/caam/ctrl.c +0 -2
index d4b39184dbdb..bd418dea586d 100644
@@ -80,7 +80,6 @@ static void build_deinstantiation_desc(u32 *desc, int handle)
 	append_jump(desc, JUMP_CLASS_CLASS1 | JUMP_TYPE_HALT);
 }
 
-#ifdef CONFIG_OF
 static const struct of_device_id imx8m_machine_match[] = {
 	{ .compatible = "fsl,imx8mm", },
 	{ .compatible = "fsl,imx8mn", },
@@ -89,7 +88,6 @@ static const struct of_device_id imx8m_machine_match[] = {
 	{ .compatible = "fsl,imx8ulp", },
 	{ }
 };
-#endif
 
 /*
  * run_descriptor_deco0 - runs a descriptor on DECO0, under direct control of
drivers/dma/dw/rzn1-dmamux.c +0 -2
index 4fb8508419db..9dcba3a3ffaa 100644
@@ -104,12 +104,10 @@ static void *rzn1_dmamux_route_allocate(struct of_phandle_args *dma_spec,
 	return ERR_PTR(ret);
 }
 
-#ifdef CONFIG_OF
 static const struct of_device_id rzn1_dmac_match[] = {
 	{ .compatible = "renesas,rzn1-dma" },
 	{}
 };
-#endif
 
 static int rzn1_dmamux_probe(struct platform_device *pdev)
 {
drivers/i2c/busses/i2c-at91-core.c +0 -2
index edc047e3e535..3a9be06dd967 100644
@@ -108,7 +108,6 @@ static const struct platform_device_id at91_twi_devtypes[] = {
 	}
 };
 
-#if defined(CONFIG_OF)
 static struct at91_twi_pdata at91sam9x5_config = {
 	.clk_max_div = 7,
 	.clk_offset = 4,
@@ -178,7 +177,6 @@ static const struct of_device_id atmel_twi_dt_ids[] = {
 	}
 };
 MODULE_DEVICE_TABLE(of, atmel_twi_dt_ids);
-#endif
 
 static struct at91_twi_pdata *at91_twi_get_driver_data(
 					struct platform_device *pdev)
drivers/i2c/busses/i2c-xiic.c +0 -2
index dc1e46d834dc..969e08e4d4f4 100644
@@ -1409,7 +1409,6 @@ static const struct i2c_adapter xiic_adapter = {
 	.algo = &xiic_algorithm,
 };
 
-#if defined(CONFIG_OF)
 static const struct xiic_version_data xiic_2_00 = {
 	.quirks = DYNAMIC_MODE_READ_BROKEN_BIT,
 };
@@ -1420,7 +1419,6 @@ static const struct of_device_id xiic_of_match[] = {
 	{},
 };
 MODULE_DEVICE_TABLE(of, xiic_of_match);
-#endif
 
 static int xiic_i2c_probe(struct platform_device *pdev)
 {
drivers/misc/atmel-ssc.c +0 -2
index 35a196341534..3db592f3b451 100644
@@ -111,7 +111,6 @@ static const struct platform_device_id atmel_ssc_devtypes[] = {
 	}
 };
 
-#ifdef CONFIG_OF
 static const struct of_device_id atmel_ssc_dt_ids[] = {
 	{
 		.compatible = "atmel,at91rm9200-ssc",
@@ -127,7 +126,6 @@ static const struct of_device_id atmel_ssc_dt_ids[] = {
 	}
 };
 MODULE_DEVICE_TABLE(of, atmel_ssc_dt_ids);
-#endif
 
 static inline const struct atmel_ssc_platform_data *
 	atmel_ssc_get_driver_data(struct platform_device *pdev)
drivers/net/can/at91_can.c +0 -2
index 191707d7e3da..9fbbf3587b0c 100644
@@ -1013,7 +1013,6 @@ static const struct attribute_group at91_sysfs_attr_group = {
 	.attrs = at91_sysfs_attrs,
 };
 
-#if defined(CONFIG_OF)
 static const struct of_device_id at91_can_dt_ids[] = {
 	{
 		.compatible = "atmel,at91sam9x5-can",
@@ -1026,7 +1025,6 @@ static const struct of_device_id at91_can_dt_ids[] = {
 	}
 };
 MODULE_DEVICE_TABLE(of, at91_can_dt_ids);
-#endif
 
 static const struct at91_devtype_data *at91_can_get_driver_data(struct platform_device *pdev)
 {
drivers/net/ethernet/cadence/macb_main.c +0 -2
index 6c462de81f20..f942c6e54a1b 100644
@@ -4425,7 +4425,6 @@ static const struct macb_usrio_config macb_default_usrio = {
 	.refclk = MACB_BIT(CLKEN),
 };
 
-#if defined(CONFIG_OF)
 /* 1518 rounded up */
 #define AT91ETHER_MAX_RBUFF_SZ	0x600
 /* max number of receive buffers */
@@ -5144,7 +5143,6 @@ static const struct of_device_id macb_dt_ids[] = {
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, macb_dt_ids);
-#endif /* CONFIG_OF */
 
 static const struct macb_config default_gem_config = {
 	.caps = MACB_CAPS_GIGABIT_MODE_AVAILABLE |