[PATCH] staging: iio: frequency: ad9832: use guard(mutex)

[PATCH] staging: iio: frequency: ad9832: use guard(mutex)

[Melih Emik]

In ad9832_write(), manual mutex handling via mutex_lock() and
mutex_unlock() necessitates error-exit trampolines (goto error_ret),
intermediate variables, and deferred returns. This pattern is prone
to unlock-omission bugs during refactoring and clutters the control
flow.

Modernize the locking by using guard(mutex) from <linux/cleanup.h>.
Scope-based resource management automatically releases the lock upon
exiting the lexical scope, which enables immediate error propagation
at call sites, removes the error_ret label, and simplifies the control
flow across all switch cases.

Tested:
- Compiled with W=1 on x86_64 with zero warnings.
- Verified with lockdep and debug mutexes enabled
  (CONFIG_PROVE_LOCKING=y, CONFIG_DEBUG_MUTEXES=y).
- Boot-tested full bzImage in QEMU with initcall_debug, confirming
  successful driver initialization (ad9832_driver_init returned 0)
  and zero lockdep warnings.
- Disassembly inspected via objdump to ensure mutex_unlock on all
  exit paths.
- Verified with scripts/checkpatch.pl (0 errors, 0 warnings).

Signed-off-by: Melih Emik <melihemik@noirlang.tr>
---
 drivers/staging/iio/frequency/ad9832.c | 39 ++++++++++++++++++--------
 1 file changed, 27 insertions(+), 12 deletions(-)

diff --git a/drivers/staging/iio/frequency/ad9832.c b/drivers/staging/iio/frequency/ad9832.c
index a78249812b26..3ae5defec615 100644
--- a/drivers/staging/iio/frequency/ad9832.c
+++ b/drivers/staging/iio/frequency/ad9832.c
@@ -8,6 +8,7 @@
 #include <linux/array_size.h>
 #include <linux/bitfield.h>
 #include <linux/bitops.h>
+#include <linux/cleanup.h>
 #include <linux/clk.h>
 #include <linux/dev_printk.h>
 #include <linux/err.h>
@@ -182,19 +183,26 @@ static ssize_t ad9832_write(struct device *dev, struct device_attribute *attr,
 
 	ret = kstrtoul(buf, 10, &val);
 	if (ret)
-		goto error_ret;
+		return ret;
+
+	guard(mutex)(&st->lock);
 
-	mutex_lock(&st->lock);
 	switch ((u32)this_attr->address) {
 	case AD9832_FREQ0HM:
 	case AD9832_FREQ1HM:
 		ret = ad9832_write_frequency(st, this_attr->address, val);
+		if (ret)
+			return ret;
+
 		break;
 	case AD9832_PHASE0H:
 	case AD9832_PHASE1H:
 	case AD9832_PHASE2H:
 	case AD9832_PHASE3H:
 		ret = ad9832_write_phase(st, this_attr->address, val);
+		if (ret)
+			return ret;
+
 		break;
 	case AD9832_PINCTRL_EN:
 		st->ctrl_ss &= ~AD9832_SELSRC;
@@ -203,24 +211,27 @@ static ssize_t ad9832_write(struct device *dev, struct device_attribute *attr,
 		st->data = cpu_to_be16(FIELD_PREP(AD9832_CMD_MSK, AD9832_CMD_SYNCSELSRC) |
 						  st->ctrl_ss);
 		ret = spi_sync(st->spi, &st->msg);
+		if (ret)
+			return ret;
+
 		break;
 	case AD9832_FREQ_SYM:
 		if (val == 1 || val == 0) {
 			st->ctrl_fp &= ~AD9832_FREQ;
 			st->ctrl_fp |= FIELD_PREP(AD9832_FREQ, val ? 1 : 0);
 		} else {
-			ret = -EINVAL;
-			break;
+			return -EINVAL;
 		}
 		st->data = cpu_to_be16(FIELD_PREP(AD9832_CMD_MSK, AD9832_CMD_FPSELECT) |
 						  st->ctrl_fp);
 		ret = spi_sync(st->spi, &st->msg);
+		if (ret)
+			return ret;
+
 		break;
 	case AD9832_PHASE_SYM:
-		if (val > 3) {
-			ret = -EINVAL;
-			break;
-		}
+		if (val > 3)
+			return -EINVAL;
 
 		st->ctrl_fp &= ~AD9832_PHASE_MASK;
 		st->ctrl_fp |= FIELD_PREP(AD9832_PHASE_MASK, val);
@@ -228,6 +239,9 @@ static ssize_t ad9832_write(struct device *dev, struct device_attribute *attr,
 		st->data = cpu_to_be16(FIELD_PREP(AD9832_CMD_MSK, AD9832_CMD_FPSELECT) |
 						  st->ctrl_fp);
 		ret = spi_sync(st->spi, &st->msg);
+		if (ret)
+			return ret;
+
 		break;
 	case AD9832_OUTPUT_EN:
 		if (val)
@@ -238,14 +252,15 @@ static ssize_t ad9832_write(struct device *dev, struct device_attribute *attr,
 		st->data = cpu_to_be16(FIELD_PREP(AD9832_CMD_MSK, AD9832_CMD_SLEEPRESCLR) |
 						  st->ctrl_src);
 		ret = spi_sync(st->spi, &st->msg);
+		if (ret)
+			return ret;
+
 		break;
 	default:
-		ret = -ENODEV;
+		return -ENODEV;
 	}
-	mutex_unlock(&st->lock);
 
-error_ret:
-	return ret ? ret : len;
+	return len;
 }
 
 /*
-- 
2.56.0

Re: [PATCH] staging: iio: frequency: ad9832: use guard(mutex)

[Joshua Crofts]

On Sun, 18 Oct 2026 13:26:04 +0300
Melih Emik <melihemik@noirlang.tr> wrote:

> In ad9832_write(), manual mutex handling via mutex_lock() and
> mutex_unlock() necessitates error-exit trampolines (goto error_ret),
> intermediate variables, and deferred returns. This pattern is prone
> to unlock-omission bugs during refactoring and clutters the control
> flow.
> 
> Modernize the locking by using guard(mutex) from <linux/cleanup.h>.
> Scope-based resource management automatically releases the lock upon
> exiting the lexical scope, which enables immediate error propagation
> at call sites, removes the error_ret label, and simplifies the control
> flow across all switch cases.
> 
> Tested:
> - Compiled with W=1 on x86_64 with zero warnings.
> - Verified with lockdep and debug mutexes enabled
>   (CONFIG_PROVE_LOCKING=y, CONFIG_DEBUG_MUTEXES=y).
> - Boot-tested full bzImage in QEMU with initcall_debug, confirming
>   successful driver initialization (ad9832_driver_init returned 0)
>   and zero lockdep warnings.
> - Disassembly inspected via objdump to ensure mutex_unlock on all
>   exit paths.
> - Verified with scripts/checkpatch.pl (0 errors, 0 warnings).

The Tested: part goes under the ---, no need to preserve this info
in the git log.

-- 
Kind regards,
Joshua Crofts

Re: [PATCH] staging: iio: frequency: ad9832: use guard(mutex)

[Melih Emik]

Hi Joshua,

Thanks for the review. I sent a v2 earlier with the test notes placed below
the "---" line.

Best regards,
Melih

Re: [PATCH] staging: iio: frequency: ad9832: use guard(mutex)

[Andy Shevchenko]

On Sun, Oct 18, 2026 at 05:55:58PM +0200, Joshua Crofts wrote:
> On Sun, 18 Oct 2026 13:26:04 +0300
> Melih Emik <melihemik@noirlang.tr> wrote:
> 
> > In ad9832_write(), manual mutex handling via mutex_lock() and
> > mutex_unlock() necessitates error-exit trampolines (goto error_ret),
> > intermediate variables, and deferred returns. This pattern is prone
> > to unlock-omission bugs during refactoring and clutters the control
> > flow.
> > 
> > Modernize the locking by using guard(mutex) from <linux/cleanup.h>.
> > Scope-based resource management automatically releases the lock upon
> > exiting the lexical scope, which enables immediate error propagation
> > at call sites, removes the error_ret label, and simplifies the control
> > flow across all switch cases.
> > 
> > Tested:
> > - Compiled with W=1 on x86_64 with zero warnings.
> > - Verified with lockdep and debug mutexes enabled
> >   (CONFIG_PROVE_LOCKING=y, CONFIG_DEBUG_MUTEXES=y).
> > - Boot-tested full bzImage in QEMU with initcall_debug, confirming
> >   successful driver initialization (ad9832_driver_init returned 0)
> >   and zero lockdep warnings.
> > - Disassembly inspected via objdump to ensure mutex_unlock on all
> >   exit paths.
> > - Verified with scripts/checkpatch.pl (0 errors, 0 warnings).
> 
> The Tested: part goes under the ---, no need to preserve this info
> in the git log.

If it was tested on a real HW, the summary (like a couple of words only)
is still good to have in the commit message to leave a trail of the actual
HW that makes the change appear.

-- 
With Best Regards,
Andy Shevchenko