Skip to contentSkip to conversation

iio: chemical: scd30: Replace manual locking with RAII locking

PATCH Superseded Maxwell Doose 2 participants 5 messages May 23 – May 27

Subsystems (2)

v4 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

PATCH v4 1/1 iio: chemical: scd30: Replace manual locking with RAII locking

scd30_core.c currently uses manual mutex_lock() and mutex_unlock() calls. Replace them with the newer guard(mutex)() for cleaner RAII patterns and to improve maintainability.

Add new helper function scd30_trigger_handler_helper_locked() containing the critical section for scd30_trigger_handler(). After moving scd30_trigger_handler()'s critical section into the new helper, tune up control logic to return ret early and not memcpy() if it's an error condition.

In addition, small refactor to replace "?:" operator with regular if/else returns.

---
v2:
- Fix callback issue as noted by Jonathan v1.
- Refactor critical section of scd30_trigger_handler() into helper called scd30_trigger_handler_helper_locked().
- Revert removal of goto in scd30_trigger_handler().

v3:
- Tune up helper to return early on failure condition per Jonathan's suggestion.

v4:
- Forgot to commit changes listed in v3, now fixed.

 drivers/iio/chemical/scd30_core.c | 68 ++++++++++++++++++++-----------
 1 file changed, 44 insertions(+), 24 deletions(-)
drivers/iio/chemical/scd30_core.c +44 -24
index a665fcb78806..8dbba9a7c426 100644
@@ -368,11 +368,13 @@ static ssize_t calibration_auto_enable_show(struct device *dev, struct device_at
 	int ret;
 	u16 val;
 
-	mutex_lock(&state->lock);
-	ret = scd30_command_read(state, CMD_ASC, &val);
-	mutex_unlock(&state->lock);
+	guard(mutex)(&state->lock);
 
-	return ret ?: sysfs_emit(buf, "%d\n", val);
+	ret = scd30_command_read(state, CMD_ASC, &val);
+	if (ret)
+		return ret;
+
+	return sysfs_emit(buf, "%d\n", val);
 }
 
 static ssize_t calibration_auto_enable_store(struct device *dev, struct device_attribute *attr,
@@ -387,11 +389,13 @@ static ssize_t calibration_auto_enable_store(struct device *dev, struct device_a
 	if (ret)
 		return ret;
 
-	mutex_lock(&state->lock);
-	ret = scd30_command_write(state, CMD_ASC, val);
-	mutex_unlock(&state->lock);
+	guard(mutex)(&state->lock);
 
-	return ret ?: len;
+	ret = scd30_command_write(state, CMD_ASC, val);
+	if (ret)
+		return ret;
+
+	return len;
 }
 
 static ssize_t calibration_forced_value_show(struct device *dev, struct device_attribute *attr,
@@ -402,11 +406,13 @@ static ssize_t calibration_forced_value_show(struct device *dev, struct device_a
 	int ret;
 	u16 val;
 
-	mutex_lock(&state->lock);
-	ret = scd30_command_read(state, CMD_FRC, &val);
-	mutex_unlock(&state->lock);
+	guard(mutex)(&state->lock);
 
-	return ret ?: sysfs_emit(buf, "%d\n", val);
+	ret = scd30_command_read(state, CMD_FRC, &val);
+	if (ret)
+		return ret;
+
+	return sysfs_emit(buf, "%d\n", val);
 }
 
 static ssize_t calibration_forced_value_store(struct device *dev, struct device_attribute *attr,
@@ -424,11 +430,13 @@ static ssize_t calibration_forced_value_store(struct device *dev, struct device_
 	if (val < SCD30_FRC_MIN_PPM || val > SCD30_FRC_MAX_PPM)
 		return -EINVAL;
 
-	mutex_lock(&state->lock);
-	ret = scd30_command_write(state, CMD_FRC, val);
-	mutex_unlock(&state->lock);
+	guard(mutex)(&state->lock);
 
-	return ret ?: len;
+	ret = scd30_command_write(state, CMD_FRC, val);
+	if (ret)
+		return ret;
+
+	return len;
 }
 
 static IIO_DEVICE_ATTR_RO(sampling_frequency_available, 0);
@@ -579,24 +587,36 @@ static irqreturn_t scd30_irq_thread_handler(int irq, void *priv)
 	return IRQ_HANDLED;
 }
 
+/* Meant ONLY for scd30_trigger_handler() */
+static int scd30_trigger_handler_helper_locked(struct iio_dev *indio_dev,
+					       int *scan_data, int arr_size)
+{
+	struct scd30_state *state = iio_priv(indio_dev);
+	int ret;
+
+	guard(mutex)(&state->lock);
+
+	if (!iio_trigger_using_own(indio_dev))
+		ret = scd30_read_poll(state);
+	else
+		ret = scd30_read_meas(state);
+	if (ret)
+		return ret;
+	memcpy(scan_data, state->meas, arr_size);
+	return 0;
+}
+
 static irqreturn_t scd30_trigger_handler(int irq, void *p)
 {
 	struct iio_poll_func *pf = p;
 	struct iio_dev *indio_dev = pf->indio_dev;
-	struct scd30_state *state = iio_priv(indio_dev);
 	struct {
 		int data[SCD30_MEAS_COUNT];
 		aligned_s64 ts;
 	} scan = { };
 	int ret;
 
-	mutex_lock(&state->lock);
-	if (!iio_trigger_using_own(indio_dev))
-		ret = scd30_read_poll(state);
-	else
-		ret = scd30_read_meas(state);
-	memcpy(scan.data, state->meas, sizeof(state->meas));
-	mutex_unlock(&state->lock);
+	ret = scd30_trigger_handler_helper_locked(indio_dev, scan.data, sizeof(scan.data));
 	if (ret)
 		goto out;
 
Signature · 1 line
2.54.0
maintainer
#1 lore raw
Quoted 10 lines · On Sat, 23 May 2026 13:25:32 -0500 Maxwell Doose <[email protected]> wrote:

scd30_core.c currently uses manual mutex_lock() and mutex_unlock() calls. Replace them with the newer guard(mutex)() for cleaner RAII patterns and to improve maintainability.

Add new helper function scd30_trigger_handler_helper_locked() containing the critical section for scd30_trigger_handler(). After moving scd30_trigger_handler()'s critical section into the new helper, tune up control logic to return ret early and not memcpy() if it's an error condition.

Quoted 4 lines

In addition, small refactor to replace "?:" operator with regular if/else returns.

Just some naming things inline.

I thought about just changing them and applying but decided I'd rather you took another look to make sure you agree with the suggested changes (and if you do send me a v5)

Thanks,

Jonathan

Quoted 20 lines

static ssize_t calibration_forced_value_store(struct device *dev, struct device_attribute *attr,

drivers/iio/chemical/scd30_core.c +6 -4
@@ -424,11 +430,13 @@ static ssize_t calibration_forced_value_store(struct device *dev, struct device_
 	if (val < SCD30_FRC_MIN_PPM || val > SCD30_FRC_MAX_PPM)
 		return -EINVAL;
 
-	mutex_lock(&state->lock);
-	ret = scd30_command_write(state, CMD_FRC, val);
-	mutex_unlock(&state->lock);
+	guard(mutex)(&state->lock);
 
-	return ret ?: len;
+	ret = scd30_command_write(state, CMD_FRC, val);
+	if (ret)
+		return ret;
+
+	return len;
 }
 
 static IIO_DEVICE_ATTR_RO(sampling_frequency_available, 0);

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 590

Quoted 4 lines
drivers/iio/chemical/scd30_core.c +1 -0
@@ -579,24 +587,36 @@ static irqreturn_t scd30_irq_thread_handler(int irq, void *priv)
 	return IRQ_HANDLED;
 }
 
+/* Meant ONLY for scd30_trigger_handler() */

No need to say this in a comment. The code naming is clear enough. However...

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 592

Quoted 2 lines
drivers/iio/chemical/scd30_core.c +2 -0
+static int scd30_trigger_handler_helper_locked(struct iio_dev *indio_dev,
+					       int *scan_data, int arr_size)

Avoid using _locked() in naming. It isn't clear to readers if that means it is locked already, or will lock.. Also somewhat unnecessary here. We don't need the function name to say why there is a helper.

arr_size would normally mean array size - i.e. how many ints there are in scan_data. Here it is the size of scan_data - probably also make it a size_t

So this whole thing becomes

static int scd30_trigger_handler_helper(struct iio_dev *indio_dev,
int *scan_data, size_t scan_data_size)

Quoted 38 lines
drivers/iio/chemical/scd30_core.c +17 -8
+{
+	struct scd30_state *state = iio_priv(indio_dev);
+	int ret;
+
+	guard(mutex)(&state->lock);
+
+	if (!iio_trigger_using_own(indio_dev))
+		ret = scd30_read_poll(state);
+	else
+		ret = scd30_read_meas(state);
+	if (ret)
+		return ret;
+	memcpy(scan_data, state->meas, arr_size);
+	return 0;
+}
+
 static irqreturn_t scd30_trigger_handler(int irq, void *p)
 {
 	struct iio_poll_func *pf = p;
 	struct iio_dev *indio_dev = pf->indio_dev;
-	struct scd30_state *state = iio_priv(indio_dev);
 	struct {
 		int data[SCD30_MEAS_COUNT];
 		aligned_s64 ts;
 	} scan = { };
 	int ret;
 
-	mutex_lock(&state->lock);
-	if (!iio_trigger_using_own(indio_dev))
-		ret = scd30_read_poll(state);
-	else
-		ret = scd30_read_meas(state);
-	memcpy(scan.data, state->meas, sizeof(state->meas));
-	mutex_unlock(&state->lock);
+	ret = scd30_trigger_handler_helper_locked(indio_dev, scan.data, sizeof(scan.data));
 	if (ret)
 		goto out;
 
#2 lore raw
Quoted 46 lines · Jonathan Cameron
Quoted 10 lines · On Sat, 23 May 2026 13:25:32 -0500 Maxwell Doose <[email protected]> wrote:

scd30_core.c currently uses manual mutex_lock() and mutex_unlock() calls. Replace them with the newer guard(mutex)() for cleaner RAII patterns and to improve maintainability.

Add new helper function scd30_trigger_handler_helper_locked() containing the critical section for scd30_trigger_handler(). After moving scd30_trigger_handler()'s critical section into the new helper, tune up control logic to return ret early and not memcpy() if it's an error condition.

Quoted 4 lines

In addition, small refactor to replace "?:" operator with regular if/else returns.

Just some naming things inline.

I thought about just changing them and applying but decided I'd rather you took another look to make sure you agree with the suggested changes (and if you do send me a v5)

Thanks,

Jonathan

Quoted 18 lines

static ssize_t calibration_forced_value_store(struct device *dev, struct device_attribute *attr,

drivers/iio/chemical/scd30_core.c +6 -4
@@ -424,11 +430,13 @@ static ssize_t calibration_forced_value_store(struct device *dev, struct device_
      if (val < SCD30_FRC_MIN_PPM || val > SCD30_FRC_MAX_PPM)
              return -EINVAL;
 
-     mutex_lock(&state->lock);
-     ret = scd30_command_write(state, CMD_FRC, val);
-     mutex_unlock(&state->lock);
+     guard(mutex)(&state->lock);
 
-     return ret ?: len;
+     ret = scd30_command_write(state, CMD_FRC, val);
+     if (ret)
+             return ret;
+
+     return len;
 }
 
Quoted 9 lines

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 590

Quoted 6 lines
drivers/iio/chemical/scd30_core.c +1 -0
 static IIO_DEVICE_ATTR_RO(sampling_frequency_available, 0);
@@ -579,24 +587,36 @@ static irqreturn_t scd30_irq_thread_handler(int irq, void *priv)
      return IRQ_HANDLED;
 }
 
+/* Meant ONLY for scd30_trigger_handler() */

No need to say this in a comment. The code naming is clear enough. However...

Agree here with the removal of said comment.

Quoted 7 lines

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 592

Quoted 2 lines
drivers/iio/chemical/scd30_core.c +2 -0
+static int scd30_trigger_handler_helper_locked(struct iio_dev *indio_dev,
+                                            int *scan_data, int arr_size)

Avoid using _locked() in naming. It isn't clear to readers if that means it is locked already, or will lock.. Also somewhat unnecessary here. We don't need the function name to say why there is a helper.

Ah...I'm honestly not sure. I know we've decided that _locked was appropriate in certain cases, I wonder if that may also be the case here. I guess though since I'd have to send a v5 anyways I'll change that.

Quoted 5 lines

arr_size would normally mean array size - i.e. how many ints there are in scan_data. Here it is the size of scan_data - probably also make it a size_t

Hm, perhaps? I definitely agree with size_t, I'm not so sure on changing the name. However like I said since I'd likely have to send a v5 anyways I'll switch up the naming.

Quoted 7 lines

So this whole thing becomes

static int scd30_trigger_handler_helper(struct iio_dev *indio_dev,
int *scan_data, size_t scan_data_size)

Sounds good.

best regards,
max

maintainer
#3 lore raw
Quoted 73 lines · On Tue, 26 May 2026 15:31:10 -0500 Maxwell Doose <[email protected]> wrote:
Quoted 46 lines · Jonathan Cameron
Quoted 10 lines · On Sat, 23 May 2026 13:25:32 -0500 Maxwell Doose <[email protected]> wrote:

scd30_core.c currently uses manual mutex_lock() and mutex_unlock() calls. Replace them with the newer guard(mutex)() for cleaner RAII patterns and to improve maintainability.

Add new helper function scd30_trigger_handler_helper_locked() containing the critical section for scd30_trigger_handler(). After moving scd30_trigger_handler()'s critical section into the new helper, tune up control logic to return ret early and not memcpy() if it's an error condition.

Quoted 4 lines

In addition, small refactor to replace "?:" operator with regular if/else returns.

Just some naming things inline.

I thought about just changing them and applying but decided I'd rather you took another look to make sure you agree with the suggested changes (and if you do send me a v5)

Thanks,

Jonathan

Quoted 18 lines

static ssize_t calibration_forced_value_store(struct device *dev, struct device_attribute *attr,

drivers/iio/chemical/scd30_core.c +6 -4
@@ -424,11 +430,13 @@ static ssize_t calibration_forced_value_store(struct device *dev, struct device_
      if (val < SCD30_FRC_MIN_PPM || val > SCD30_FRC_MAX_PPM)
              return -EINVAL;
 
-     mutex_lock(&state->lock);
-     ret = scd30_command_write(state, CMD_FRC, val);
-     mutex_unlock(&state->lock);
+     guard(mutex)(&state->lock);
 
-     return ret ?: len;
+     ret = scd30_command_write(state, CMD_FRC, val);
+     if (ret)
+             return ret;
+
+     return len;
 }
 
Quoted 9 lines

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 590

Quoted 6 lines
drivers/iio/chemical/scd30_core.c +1 -0
 static IIO_DEVICE_ATTR_RO(sampling_frequency_available, 0);
@@ -579,24 +587,36 @@ static irqreturn_t scd30_irq_thread_handler(int irq, void *priv)
      return IRQ_HANDLED;
 }
 
+/* Meant ONLY for scd30_trigger_handler() */  

No need to say this in a comment. The code naming is clear enough. However...

Agree here with the removal of said comment.

Quoted 7 lines

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 592

Quoted 2 lines
drivers/iio/chemical/scd30_core.c +2 -0
+static int scd30_trigger_handler_helper_locked(struct iio_dev *indio_dev,
+                                            int *scan_data, int arr_size)  

Avoid using _locked() in naming. It isn't clear to readers if that means it is locked already, or will lock.. Also somewhat unnecessary here. We don't need the function name to say why there is a helper.

Ah...I'm honestly not sure. I know we've decided that _locked was appropriate in certain cases, I wonder if that may also be the case here. I guess though since I'd have to send a v5 anyways I'll change that.

Quoted 1 line
Quoted 8 lines
Quoted 4 lines

arr_size would normally mean array size - i.e. how many ints there are in scan_data. Here it is the size of scan_data - probably also make it a size_t

Hm, perhaps? I definitely agree with size_t, I'm not so sure on changing the name. However like I said since I'd likely have to send a v5 anyways I'll switch up the naming.

Given people will associate arr_size with ARRAY_SIZE() which on this array would return the number of ints (so /4), that name definitely needs to change. Exactly what it changes to is a different question!

Quoted 13 lines
Quoted 8 lines

So this whole thing becomes

static int scd30_trigger_handler_helper(struct iio_dev *indio_dev,
int *scan_data, size_t scan_data_size)

Sounds good.

best regards,
max

#4 lore raw
Quoted 82 lines · Jonathan Cameron
Quoted 78 lines · On Tue, 26 May 2026 15:31:10 -0500 Maxwell Doose <[email protected]> wrote:
Quoted 46 lines · Jonathan Cameron
Quoted 10 lines · On Sat, 23 May 2026 13:25:32 -0500 Maxwell Doose <[email protected]> wrote:

scd30_core.c currently uses manual mutex_lock() and mutex_unlock() calls. Replace them with the newer guard(mutex)() for cleaner RAII patterns and to improve maintainability.

Add new helper function scd30_trigger_handler_helper_locked() containing the critical section for scd30_trigger_handler(). After moving scd30_trigger_handler()'s critical section into the new helper, tune up control logic to return ret early and not memcpy() if it's an error condition.

Quoted 4 lines

In addition, small refactor to replace "?:" operator with regular if/else returns.

Just some naming things inline.

I thought about just changing them and applying but decided I'd rather you took another look to make sure you agree with the suggested changes (and if you do send me a v5)

Thanks,

Jonathan

Quoted 18 lines

static ssize_t calibration_forced_value_store(struct device *dev, struct device_attribute *attr,

drivers/iio/chemical/scd30_core.c +6 -4
@@ -424,11 +430,13 @@ static ssize_t calibration_forced_value_store(struct device *dev, struct device_
      if (val < SCD30_FRC_MIN_PPM || val > SCD30_FRC_MAX_PPM)
              return -EINVAL;
 
-     mutex_lock(&state->lock);
-     ret = scd30_command_write(state, CMD_FRC, val);
-     mutex_unlock(&state->lock);
+     guard(mutex)(&state->lock);
 
-     return ret ?: len;
+     ret = scd30_command_write(state, CMD_FRC, val);
+     if (ret)
+             return ret;
+
+     return len;
 }
 
Quoted 9 lines

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 590

Quoted 6 lines
drivers/iio/chemical/scd30_core.c +1 -0
 static IIO_DEVICE_ATTR_RO(sampling_frequency_available, 0);
@@ -579,24 +587,36 @@ static irqreturn_t scd30_irq_thread_handler(int irq, void *priv)
      return IRQ_HANDLED;
 }
 
+/* Meant ONLY for scd30_trigger_handler() */

No need to say this in a comment. The code naming is clear enough. However...

Agree here with the removal of said comment.

Quoted 7 lines

drivers/iio/chemical/scd30_core.c static irqreturn_t scd30_irq_thread_handler(int irq, void *priv) · line 592

Quoted 2 lines
drivers/iio/chemical/scd30_core.c +2 -0
+static int scd30_trigger_handler_helper_locked(struct iio_dev *indio_dev,
+                                            int *scan_data, int arr_size)

Avoid using _locked() in naming. It isn't clear to readers if that means it is locked already, or will lock.. Also somewhat unnecessary here. We don't need the function name to say why there is a helper.

Ah...I'm honestly not sure. I know we've decided that _locked was appropriate in certain cases, I wonder if that may also be the case here. I guess though since I'd have to send a v5 anyways I'll change that.

Quoted 5 lines

arr_size would normally mean array size - i.e. how many ints there are in scan_data. Here it is the size of scan_data - probably also make it a size_t

Quoted 8 lines
Quoted 3 lines

Hm, perhaps? I definitely agree with size_t, I'm not so sure on changing the name. However like I said since I'd likely have to send a v5 anyways I'll switch up the naming.

Given people will associate arr_size with ARRAY_SIZE() which on this array would return the number of ints (so /4), that name definitely needs to change. Exactly what it changes to is a different question!

I wonder, perhaps either size (not as good) or arr_size_bytes? Maybe also scan_data_byte_size? Probably still something better out there, those are just some names I can come up with off the top of my head.

best regards,
max

Quoted 14 lines
Quoted 13 lines
Quoted 8 lines

So this whole thing becomes

static int scd30_trigger_handler_helper(struct iio_dev *indio_dev,
int *scan_data, size_t scan_data_size)

Sounds good.

best regards,
max