Skip to contentSkip to conversation

iio: adc: stm32-adc: warn if dt uses legacy channel config

PATCH Superseded Sean Nyekjaer 4 participants 7 messages Mar 27, 2023 – Apr 3, 2023

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/3 iio: adc: stm32-adc: warn if dt uses legacy channel config

Since nearly all stm32 dt's are using the legacy adc channel config, we should warn users about using it.

---
 drivers/iio/adc/stm32-adc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/iio/adc/stm32-adc.c +2 -0
index 1aadb2ad2cab..d8e755d0cc52 100644
@@ -1993,6 +1993,8 @@ static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm
 	const struct stm32_adc_info *adc_info = adc->cfg->adc_info;
 	int num_channels = 0, ret;
 
+	dev_warn(&indio_dev->dev, "using legacy channel config\n");
+
 	ret = device_property_count_u32(dev, "st,adc-channels");
 	if (ret > adc_info->max_channels) {
 		dev_err(&indio_dev->dev, "Bad st,adc-channels?\n");
Signature · 1 line
2.39.1
#1 lore raw

PATCH 2/3 iio: adc: stm32-adc: skip adc-diff-channels setup if none is present

If no adc differential channels are defined driver will fail with EINVAL:
stm32-adc: probe of 48003000.adc:adc@0 failed with error -22

Fix this by skipping the initialization if no channels are defined.

This applies only to the legacy way of initializing adc channels.

Fixes d7705f35448a ("iio: adc: stm32-adc: convert to device properties")
---
 drivers/iio/adc/stm32-adc.c | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)
drivers/iio/adc/stm32-adc.c +9 -10
index d8e755d0cc52..a04fcb2dc80a 100644
@@ -2008,16 +2008,15 @@ static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm
 	 * to get the *real* number of channels.
 	 */
 	ret = device_property_count_u32(dev, "st,adc-diff-channels");
-	if (ret < 0)
-		return ret;
-
-	ret /= (int)(sizeof(struct stm32_adc_diff_channel) / sizeof(u32));
-	if (ret > adc_info->max_channels) {
-		dev_err(&indio_dev->dev, "Bad st,adc-diff-channels?\n");
-		return -EINVAL;
-	} else if (ret > 0) {
-		adc->num_diff = ret;
-		num_channels += ret;
+	if (ret > 0) {
+		ret /= (int)(sizeof(struct stm32_adc_diff_channel) / sizeof(u32));
+		if (ret > adc_info->max_channels) {
+			dev_err(&indio_dev->dev, "Bad st,adc-diff-channels?\n");
+			return -EINVAL;
+		} else if (ret > 0) {
+			adc->num_diff = ret;
+			num_channels += ret;
+		}
 	}
 
 	/* Optional sample time is provided either for each, or all channels */
Signature · 1 line
2.39.1
#2 lore raw

PATCH 3/3 iio: adc: stm32-adc: skip adc-channels setup if none is present

In reply to Sean Nyekjaer

If only adc differential channels are defined driver will fail with
stm32-adc: probe of 48003000.adc:adc@0 failed with error -22

Fix this by skipping the initialization if no channels are defined.

This applies only to the legacy way of initializing adc channels.

Fixes d7705f35448a ("iio: adc: stm32-adc: convert to device properties")
---
 drivers/iio/adc/stm32-adc.c | 38 +++++++++++++++++++------------------
 1 file changed, 20 insertions(+), 18 deletions(-)
drivers/iio/adc/stm32-adc.c +20 -18
index a04fcb2dc80a..6d87cfaadb5d 100644
@@ -2065,28 +2065,30 @@ static int stm32_adc_legacy_chan_init(struct iio_dev *indio_dev,
 		}
 	}
 
-	ret = device_property_read_u32_array(dev, "st,adc-channels", chans,
-					     nchans);
-	if (ret)
-		return ret;
-
-	for (c = 0; c < nchans; c++) {
-		if (chans[c] >= adc_info->max_channels) {
-			dev_err(&indio_dev->dev, "Invalid channel %d\n",
-				chans[c]);
-			return -EINVAL;
-		}
+	if (nchans - num_diff > 0) {
+		ret = device_property_read_u32_array(dev, "st,adc-channels", chans,
+				nchans);
+		if (ret)
+			return ret;
 
-		/* Channel can't be configured both as single-ended & diff */
-		for (i = 0; i < num_diff; i++) {
-			if (chans[c] == diff[i].vinp) {
-				dev_err(&indio_dev->dev, "channel %d misconfigured\n",	chans[c]);
+		for (c = 0; c < nchans; c++) {
+			if (chans[c] >= adc_info->max_channels) {
+				dev_err(&indio_dev->dev, "Invalid channel %d\n",
+						chans[c]);
 				return -EINVAL;
 			}
-		}
-		stm32_adc_chan_init_one(indio_dev, &channels[scan_index],
+
+			/* Channel can't be configured both as single-ended & diff */
+			for (i = 0; i < num_diff; i++) {
+				if (chans[c] == diff[i].vinp) {
+					dev_err(&indio_dev->dev, "channel %d misconfigured\n",	chans[c]);
+					return -EINVAL;
+				}
+			}
+			stm32_adc_chan_init_one(indio_dev, &channels[scan_index],
 					chans[c], 0, scan_index, false);
-		scan_index++;
+			scan_index++;
+		}
 	}
 
 	if (adc->nsmps > 0) {
Signature · 1 line
2.39.1
#3 lore raw

Re: [Linux-stm32] [PATCH 3/3] iio: adc: stm32-adc: skip adc-channels setup if none is present

Hi Sean,

Thanks for your patch.
You're right. The DT updates to use the generic bindings are not yet upstreamed, and there are some regressions on legacy bindings support. Please, find here after some comments.

BRs
Olivier

Quoted 18 lines · Sean Nyekjaer

If only adc differential channels are defined driver will fail with
stm32-adc: probe of 48003000.adc:adc@0 failed with error -22

Fix this by skipping the initialization if no channels are defined.

This applies only to the legacy way of initializing adc channels.

Fixes d7705f35448a ("iio: adc: stm32-adc: convert to device properties")
---
  drivers/iio/adc/stm32-adc.c | 38 +++++++++++++++++++------------------
  1 file changed, 20 insertions(+), 18 deletions(-)
drivers/iio/adc/stm32-adc.c +0 -0
index a04fcb2dc80a..6d87cfaadb5d 100644

drivers/iio/adc/stm32-adc.c static int stm32_adc_legacy_chan_init(struct iio_dev *indio_dev, · line 2067

Quoted 3 lines
drivers/iio/adc/stm32-adc.c +0 -0
@@ -2065,28 +2065,30 @@ static int stm32_adc_legacy_chan_init(struct iio_dev *indio_dev,
  		}
  	}
 

In PIO mode an extra channel channel is defined for timestamps. This additional channel must be ignored in channel count when initializing single and diff channels.

This can be handled in stm32_adc_legacy_chan_init() call from stm32_adc_legacy_chan_init() function:

ret = stm32_adc_legacy_chan_init(indio_dev, adc, channels,
timestamping ? num_channels - 1 : num_channels);

Quoted 4 lines
drivers/iio/adc/stm32-adc.c +0 -4
-	ret = device_property_read_u32_array(dev, "st,adc-channels", chans,
-					     nchans);
-	if (ret)
-		return ret;

drivers/iio/adc/stm32-adc.c static int stm32_adc_legacy_chan_init(struct iio_dev *indio_dev, · line 2070

Quoted 10 lines
drivers/iio/adc/stm32-adc.c +3 -7
-
-	for (c = 0; c < nchans; c++) {
-		if (chans[c] >= adc_info->max_channels) {
-			dev_err(&indio_dev->dev, "Invalid channel %d\n",
-				chans[c]);
-			return -EINVAL;
-		}
+	if (nchans - num_diff > 0) {
+		ret = device_property_read_u32_array(dev, "st,adc-channels", chans,
+				nchans);

num_se = nchans - num_diff represents single ended channels number. single ended count has to be used also in device_property_read_u32_array() call:
ret = device_property_read_u32_array(dev, "st,adc-channels", chans, num_se);

drivers/iio/adc/stm32-adc.c static int stm32_adc_legacy_chan_init(struct iio_dev *indio_dev, · line 2074

Quoted 8 lines
drivers/iio/adc/stm32-adc.c +3 -4
+		if (ret)
+			return ret;
 
-		/* Channel can't be configured both as single-ended & diff */
-		for (i = 0; i < num_diff; i++) {
-			if (chans[c] == diff[i].vinp) {
-				dev_err(&indio_dev->dev, "channel %d misconfigured\n",	chans[c]);
+		for (c = 0; c < nchans; c++) {

and also in this for loop:
for (c = 0; c < num_se; c++) {

Quoted 23 lines
drivers/iio/adc/stm32-adc.c +14 -3
+			if (chans[c] >= adc_info->max_channels) {
+				dev_err(&indio_dev->dev, "Invalid channel %d\n",
+						chans[c]);
  				return -EINVAL;
  			}
-		}
-		stm32_adc_chan_init_one(indio_dev, &channels[scan_index],
+
+			/* Channel can't be configured both as single-ended & diff */
+			for (i = 0; i < num_diff; i++) {
+				if (chans[c] == diff[i].vinp) {
+					dev_err(&indio_dev->dev, "channel %d misconfigured\n",	chans[c]);
+					return -EINVAL;
+				}
+			}
+			stm32_adc_chan_init_one(indio_dev, &channels[scan_index],
  					chans[c], 0, scan_index, false);
-		scan_index++;
+			scan_index++;
+		}
  	}
 
  	if (adc->nsmps > 0) {
#4 lore raw
In reply to Sean Nyekjaer
Quoted 13 lines · Sean Nyekjaer

Since nearly all stm32 dt's are using the legacy adc channel config, we should warn users about using it.

---
 drivers/iio/adc/stm32-adc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/iio/adc/stm32-adc.c +0 -0
index 1aadb2ad2cab..d8e755d0cc52 100644

drivers/iio/adc/stm32-adc.c static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm · line 1997

Quoted 5 lines
drivers/iio/adc/stm32-adc.c +2 -0
@@ -1993,6 +1993,8 @@ static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm
 	const struct stm32_adc_info *adc_info = adc->cfg->adc_info;
 	int num_channels = 0, ret;
 
+	dev_warn(&indio_dev->dev, "using legacy channel config\n");
+

Hi Sean,

I'd recommend to avoid adding a dev_warn() on a per driver basis, for depreacted DT properties. Still I'm not sure if there's some policy in this area.

IMHO, deprecated properties should be checked by using dt tools (dt_binding_check / dtbs_check or other mean?). But I have no idea if this is going to report warnings and when. Purpose would be to avoid introducing no dts files with these. As commented by Olivier on Patch 3, we've some downstream patches to adopt the generic bindings (not upstream 'yet').

Another downside is regarding backward compatibility. In case an old dtb is used to boot the kernel, as long as there's no functionality loss, I'd advise not to use any warning here. That's a valid use of an old dt.

In all case, thanks for pointing issues (e.g. Patch 2 & 3), regarding this patch, I'd nack adding this warning. Please drop this change if you re-submit or turn this into a dev_dbg().

Best Regards,
Fabrice

Quoted 3 lines
drivers/iio/adc/stm32-adc.c +0 -0
 	ret = device_property_count_u32(dev, "st,adc-channels");
 	if (ret > adc_info->max_channels) {
 		dev_err(&indio_dev->dev, "Bad st,adc-channels?\n");
maintainer
#5 lore raw
Quoted 33 lines · On Thu, 30 Mar 2023 17:30:32 +0200 Fabrice Gasnier…
Quoted 13 lines · Sean Nyekjaer

Since nearly all stm32 dt's are using the legacy adc channel config, we should warn users about using it.

---
 drivers/iio/adc/stm32-adc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/iio/adc/stm32-adc.c +0 -0
index 1aadb2ad2cab..d8e755d0cc52 100644

drivers/iio/adc/stm32-adc.c static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm · line 1997

Quoted 5 lines
drivers/iio/adc/stm32-adc.c +2 -0
@@ -1993,6 +1993,8 @@ static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm
 	const struct stm32_adc_info *adc_info = adc->cfg->adc_info;
 	int num_channels = 0, ret;
 
+	dev_warn(&indio_dev->dev, "using legacy channel config\n");
+  

Hi Sean,

I'd recommend to avoid adding a dev_warn() on a per driver basis, for depreacted DT properties. Still I'm not sure if there's some policy in this area.

IMHO, deprecated properties should be checked by using dt tools (dt_binding_check / dtbs_check or other mean?). But I have no idea if this is going to report warnings and when. Purpose would be to avoid introducing no dts files with these. As commented by Olivier on Patch 3, we've some downstream patches to adopt the generic bindings (not upstream 'yet').

Quoted 8 lines

Another downside is regarding backward compatibility. In case an old dtb is used to boot the kernel, as long as there's no functionality loss, I'd advise not to use any warning here. That's a valid use of an old dt.

In all case, thanks for pointing issues (e.g. Patch 2 & 3), regarding this patch, I'd nack adding this warning. Please drop this change if you re-submit or turn this into a dev_dbg().

Agreed. Better to change to dev_dbg().

Other two patches look good to me, but will leave a bit more time for others to comment before I pick them up. As a small side note. They are fixes and this first patch is not, so they should have been before it in the series so I can route them to mainline faster than the non fix.

Jonathan

Quoted 6 lines

Best Regards,
Fabrice

Quoted 3 lines
drivers/iio/adc/stm32-adc.c +0 -0
 	ret = device_property_count_u32(dev, "st,adc-channels");
 	if (ret > adc_info->max_channels) {
 		dev_err(&indio_dev->dev, "Bad st,adc-channels?\n");  
#6 lore raw
Quoted 49 lines · Jonathan Cameron
Quoted 33 lines · On Thu, 30 Mar 2023 17:30:32 +0200 Fabrice Gasnier…
Quoted 13 lines · Sean Nyekjaer

Since nearly all stm32 dt's are using the legacy adc channel config, we should warn users about using it.

---
drivers/iio/adc/stm32-adc.c | 2 ++

1 file changed, 2 insertions(+)
drivers/iio/adc/stm32-adc.c +0 -0
index 1aadb2ad2cab..d8e755d0cc52 100644

drivers/iio/adc/stm32-adc.c static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm · line 1997

Quoted 5 lines
drivers/iio/adc/stm32-adc.c +2 -0
@@ -1993,6 +1993,8 @@ static int stm32_adc_get_legacy_chan_count(struct iio_dev *indio_dev, struct stm
 const struct stm32_adc_info *adc_info = adc->cfg->adc_info;
 int num_channels = 0, ret;
 
+ dev_warn(&indio_dev->dev, "using legacy channel config\n");
+  

Hi Sean,

I'd recommend to avoid adding a dev_warn() on a per driver basis, for depreacted DT properties. Still I'm not sure if there's some policy in this area.

IMHO, deprecated properties should be checked by using dt tools (dt_binding_check / dtbs_check or other mean?). But I have no idea if this is going to report warnings and when. Purpose would be to avoid introducing no dts files with these. As commented by Olivier on Patch 3, we've some downstream patches to adopt the generic bindings (not upstream 'yet').

Quoted 8 lines

Another downside is regarding backward compatibility. In case an old dtb is used to boot the kernel, as long as there's no functionality loss, I'd advise not to use any warning here. That's a valid use of an old dt.

In all case, thanks for pointing issues (e.g. Patch 2 & 3), regarding this patch, I'd nack adding this warning. Please drop this change if you re-submit or turn this into a dev_dbg().

Quoted 8 lines

Agreed. Better to change to dev_dbg().

Other two patches look good to me, but will leave a bit more time for others to comment before I pick them up. As a small side note. They are fixes and this first patch is not, so they should have been before it in the series so I can route them to mainline faster than the non fix.

Jonathan

Hi Jonathan,

I’ll resubmit the first patch with dev_dbg() as a single commit, and then the fixes as a separate series :)

/Sean

Quoted 7 lines
Quoted 7 lines

Best Regards,
Fabrice

Quoted 3 lines
drivers/iio/adc/stm32-adc.c +0 -0
 ret = device_property_count_u32(dev, "st,adc-channels");
 if (ret > adc_info->max_channels) {
 dev_err(&indio_dev->dev, "Bad st,adc-channels?\n");