Skip to contentSkip to conversation

iio: frequency: ad9523: check return value of ad9523_io_update()

PATCH In review Bhargav Joshi 2 participants 4 messages Mar 17 – Mar 18

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 iio: frequency: ad9523: check return value of ad9523_io_update()

Several functions in ad9523 driver calls ad9523_io_update() to apply register changes but fail to check its return value.

If the function fails, driver silently ignore the error and returns success. This results in user-space assuming register changes even if they are failed.

Fix this by properly catching and returning the error code from ad9523_io_update() in ad9523_sync(), ad9523_write_raw(), and ad9523_reg_access()

---
 drivers/iio/frequency/ad9523.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
drivers/iio/frequency/ad9523.c +8 -3
index 63c485e9e44c..66bba37fb199 100644
@@ -495,7 +495,10 @@ static int ad9523_sync(struct iio_dev *indio_dev)
 	if (ret < 0)
 		return ret;
 
-	ad9523_io_update(indio_dev);
+	ret = ad9523_io_update(indio_dev);
+	if (ret < 0)
+		return ret;
+
 	tmp &= ~AD9523_STATUS_SIGNALS_SYNC_MAN_CTRL;
 
 	ret = ad9523_write(indio_dev, AD9523_STATUS_SIGNALS, tmp);
@@ -716,7 +719,7 @@ static int ad9523_write_raw(struct iio_dev *indio_dev,
 	if (ret < 0)
 		goto out;
 
-	ad9523_io_update(indio_dev);
+	ret = ad9523_io_update(indio_dev);
 out:
 	mutex_unlock(&st->lock);
 	return ret;
@@ -732,7 +735,9 @@ static int ad9523_reg_access(struct iio_dev *indio_dev,
 	mutex_lock(&st->lock);
 	if (readval == NULL) {
 		ret = ad9523_write(indio_dev, reg | AD9523_R1B, writeval);
-		ad9523_io_update(indio_dev);
+		if (ret < 0)
+			goto out_unlock;
+		ret = ad9523_io_update(indio_dev);
 	} else {
 		ret = ad9523_read(indio_dev, reg | AD9523_R1B);
 		if (ret < 0)
Signature · 1 line
2.53.0
#1 lore raw
Quoted 10 lines · Bhargav Joshi

Several functions in ad9523 driver calls ad9523_io_update() to apply register changes but fail to check its return value.

If the function fails, driver silently ignore the error and returns success. This results in user-space assuming register changes even if they are failed.

Fix this by properly catching and returning the error code from ad9523_io_update() in ad9523_sync(), ad9523_write_raw(), and ad9523_reg_access()

Missing period at the end.

...

So, now the errors that were non-fatal become fatal. Is it a problem from functional behaviour point of view? How did you test this?

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#2 lore raw

hi,

Quoted 13 lines · Andy Shevchenko
Quoted 10 lines · Bhargav Joshi

Several functions in ad9523 driver calls ad9523_io_update() to apply register changes but fail to check its return value.

If the function fails, driver silently ignore the error and returns success. This results in user-space assuming register changes even if they are failed.

Fix this by properly catching and returning the error code from ad9523_io_update() in ad9523_sync(), ad9523_write_raw(), and ad9523_reg_access()

Quoted 7 lines

Missing period at the end.

...

So, now the errors that were non-fatal become fatal. Is it a problem from functional behaviour point of view? How did you test this?

Regarding errors become fatal, is true for function ad9523_sync() because even if error occurred it clears the sync bit and calls second io_update anyways so returning early here would make it worse, Apologies for this oversight on my part.

However for the ad9523_write_raw() and ad9523_reg_access() ad9523_io_update() is the last step and Returning the error in these cases seems strictly correct to prevent silently ignoring I/O failures and returning false success to user-space regarding the active configuration.

Regarding testing: I do not have access to the physical hardware, so the patch is compile-tested only.

Would you like to send me v2 with dropping ad9523_sync() changes and keeping the other two or just drop the whole patch?

Thanks,
Bhargav Joshi

Quoted 5 lines

--
With Best Regards,
Andy Shevchenko

#3 lore raw
Quoted 3 lines · Bhargav Joshi
Quoted 1 line · Andy Shevchenko

On Wed, Mar 18, 2026 at 04:55:52AM +0530, Bhargav Joshi wrote:

...

Quoted 14 lines
Quoted 2 lines

So, now the errors that were non-fatal become fatal. Is it a problem from functional behaviour point of view? How did you test this?

Regarding errors become fatal, is true for function ad9523_sync() because even if error occurred it clears the sync bit and calls second io_update anyways so returning early here would make it worse, Apologies for this oversight on my part.

However for the ad9523_write_raw() and ad9523_reg_access() ad9523_io_update() is the last step and Returning the error in these cases seems strictly correct to prevent silently ignoring I/O failures and returning false success to user-space regarding the active configuration.

Quoted 5 lines

Regarding testing: I do not have access to the physical hardware, so the patch is compile-tested only.

Would you like to send me v2 with dropping ad9523_sync() changes and keeping the other two or just drop the whole patch?

Don't be in a hurry. Let's wait for others to comment. We have ADI people here, they might chime in and may be even help with testing if they consider the changes valuable.

Signature · 2 lines
With Best Regards,
Andy Shevchenko