Skip to contentSkip to conversation

Variety of fixes and new features for mr75203 driver

PATCH Superseded Eliav Farber 5 participants 88 messages Aug 30, 2022 – Sep 3, 2022 1 message not in the archive

Subsystems (2)

v3 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

v3 cover Variety of fixes and new features for mr75203 driver

List of fixes:
- Fix "intel,vm-map" property to be optional.
- Fix VM sensor allocation when "intel,vm-map" not defined.
- Fix multi-channel voltage reading.
- Fix voltage equation for negative source input.
- Modify the temperature equation according to series 5 datasheet.
- Fix coding style issue.

List of new features:
- Modify "reset" property to be optional.
- Add optional "moortec,vm-active-channels" property to define the number of active channels per VM.
- Add support for mr76006 pre-scaler to multiply the voltage result by 2.
- Add support for series 6 temperature equation.
- Add coefficient properties to fine tune the temperature equation.
- Add debugfs to read and write temperature coefficients

---------

Changes between v2 and v3:
*) Add "moortec" prefix to all new device-tree properties. *) Change order of patches.
*) Add explanations to better understand the changes.
*) Change "reset" property to be optional and remove the
"reset-control-skip" property.
*) Split the patch for "fix multi-channel voltage reading" to two
patches.
*) Change pre-scaler property format and fix typo (scalar --> scaler). *) Fix voltage equation to support negative values instead of limiting
value to zero.
*) Temperature equation - protect from overflow and add clamping. *) Add new "moortec,ts-series" property to select between temperature
equation of series 5 or series 6.

Changes between v1 and v2:
*) Fix compilation error for patch 08/16:
"warning: ISO C90 forbids variable length array"

---------

Eliav Farber (19):
dt-bindings: hwmon: (mr75203) update "intel,vm-map" property to be
optional
hwmon: (mr75203) fix VM sensor allocation when "intel,vm-map" not
defined
hwmon: (mr75203) update pvt->v_num to the actual number of used
sensors
dt-bindings: hwmon: (mr75203) change "reset" property to be optional
hwmon: (mr75203) skip reset-control deassert for SOCs that don't
support it
hwmon: (mr75203) fix multi-channel voltage reading
hwmon: (mr75203) enable polling for all VM channels
dt-bindings: hwmon: (mr75203) add "moortec,vm-active-channels"
property
hwmon: (mr75203) add VM active channel support
dt-bindings: hwmon: (mr75203) add "moortec,vm-pre-scaler" property
hwmon: (mr75203) add VM pre-scaler support
hwmon: (mr75203) fix voltage equation for negative source input
hwmon: (mr75203) modify the temperature equation according to series 5
datasheet
dt-bindings: hwmon: (mr75203) add "moortec,ts-series" property
hwmon: (mr75203) add support for series 6 temperature equation
dt-bindings: hwmon: (mr75203) add coefficient properties for the
thermal equation
hwmon: (mr75203) parse temperature coefficients from device-tree
hwmon: (mr75203) add debugfs to read and write temperature
coefficients
hwmon: (mr75203) fix coding style space errors

 .../bindings/hwmon/moortec,mr75203.yaml       |  67 ++-
 drivers/hwmon/mr75203.c                       | 568 ++++++++++++++++--
 2 files changed, 569 insertions(+), 66 deletions(-)
Signature · 1 line
2.37.1
#1 lore raw

PATCH v3 1/19 dt-bindings: hwmon: (mr75203) update "intel,vm-map" property to be optional

Change "intel,vm-map" property to be optional instead of required.

The driver implementation indicates it is not mandatory to have "intel,vm-map" in the device tree:
- probe doesn't fail in case it is absent.
- explicit comment in code - "Incase intel,vm-map property is not defined, we assume incremental channel numbers".

---
V3 -> V2:
- Change this patch to be first in the series.
- Add explanation why "intel,vm-map" is not required.

 Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml | 1 -
 1 file changed, 1 deletion(-)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -1
index 6f3e3c01f717..6abde48b746e 100644
@@ -48,7 +48,6 @@ required:
   - compatible
   - reg
   - reg-names
-  - intel,vm-map
   - clocks
   - resets
   - "#thermal-sensor-cells"
Signature · 1 line
2.37.1
maintainer Acked-by on v3 1/19
#2 lore raw

Re: [PATCH v3 01/19] dt-bindings: hwmon: (mr75203) update "intel,vm-map" property to be optional

Quoted 8 lines · Eliav Farber

Change "intel,vm-map" property to be optional instead of required.

The driver implementation indicates it is not mandatory to have "intel,vm-map" in the device tree:
- probe doesn't fail in case it is absent.
- explicit comment in code - "Incase intel,vm-map property is not defined, we assume incremental channel numbers".

Quoted 9 lines

---
V3 -> V2:
- Change this patch to be first in the series.
- Add explanation why "intel,vm-map" is not required.

 Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml | 1 -
 1 file changed, 1 deletion(-)
#3 lore raw

PATCH v3 2/19 hwmon: (mr75203) fix VM sensor allocation when "intel,vm-map" not defined

In reply to Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
 drivers/hwmon/mr75203.c | 28 ++++++++++++----------------
 1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +12 -16
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device *pdev)
 	}
 
 	if (vm_num) {
-		u32 num = vm_num;
-
 		ret = pvt_get_regmap(pdev, "vm", pvt);
 		if (ret)
 			return ret;
@@ -594,30 +592,28 @@ static int mr75203_probe(struct platform_device *pdev)
 		ret = device_property_read_u8_array(dev, "intel,vm-map",
 						    pvt->vm_idx, vm_num);
 		if (ret) {
-			num = 0;
+			/*
+			 * Incase intel,vm-map property is not defined, we
+			 * assume incremental channel numbers.
+			 */
+			for (i = 0; i < vm_num; i++)
+				pvt->vm_idx[i] = i;
 		} else {
 			for (i = 0; i < vm_num; i++)
 				if (pvt->vm_idx[i] >= vm_num ||
-				    pvt->vm_idx[i] == 0xff) {
-					num = i;
+				    pvt->vm_idx[i] == 0xff)
 					break;
-				}
-		}
 
-		/*
-		 * Incase intel,vm-map property is not defined, we assume
-		 * incremental channel numbers.
-		 */
-		for (i = num; i < vm_num; i++)
-			pvt->vm_idx[i] = i;
+			vm_num = i;
+		}
 
-		in_config = devm_kcalloc(dev, num + 1,
+		in_config = devm_kcalloc(dev, vm_num + 1,
 					 sizeof(*in_config), GFP_KERNEL);
 		if (!in_config)
 			return -ENOMEM;
 
-		memset32(in_config, HWMON_I_INPUT, num);
-		in_config[num] = 0;
+		memset32(in_config, HWMON_I_INPUT, vm_num);
+		in_config[vm_num] = 0;
 		pvt_in.config = in_config;
 
 		pvt_info[index++] = &pvt_in;
Signature · 1 line
2.37.1
maintainer
#4 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel,vm-map" not defined

Quoted 41 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------
  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +7 -7
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device *pdev)
  	}
 
  	if (vm_num) {
-		u32 num = vm_num;
-
  		ret = pvt_get_regmap(pdev, "vm", pvt);
  		if (ret)
  			return ret;
@@ -594,30 +592,28 @@ static int mr75203_probe(struct platform_device *pdev)
  		ret = device_property_read_u8_array(dev, "intel,vm-map",
  						    pvt->vm_idx, vm_num);
  		if (ret) {
-			num = 0;
+			/*
+			 * Incase intel,vm-map property is not defined, we
+			 * assume incremental channel numbers.
+			 */
+			for (i = 0; i < vm_num; i++)
+				pvt->vm_idx[i] = i;
  		} else {
  			for (i = 0; i < vm_num; i++)
  				if (pvt->vm_idx[i] >= vm_num ||
-				    pvt->vm_idx[i] == 0xff) {
-					num = i;
+				    pvt->vm_idx[i] == 0xff)
  					break;
-				}
-		}

drivers/hwmon/mr75203.c static int mr75203_probe(struct platform_device *pdev) · line 609

Quoted 10 lines
drivers/hwmon/mr75203.c +2 -6
 
-		/*
-		 * Incase intel,vm-map property is not defined, we assume
-		 * incremental channel numbers.
-		 */
-		for (i = num; i < vm_num; i++)
-			pvt->vm_idx[i] = i;
+			vm_num = i;
+		}
 

So this is actually a functional change: In the old code, unspecified channel numbers (num ... vm_num) were filled with incremental channel numbers. This is no longer the case.

drivers/hwmon/mr75203.c static int mr75203_probe(struct platform_device *pdev) · line 611

Quoted 3 lines
drivers/hwmon/mr75203.c +1 -1
-		in_config = devm_kcalloc(dev, num + 1,
+		in_config = devm_kcalloc(dev, vm_num + 1,
  					 sizeof(*in_config), GFP_KERNEL);

The relevant difference (and maybe bug in the existing code ?) seems to be this change. Have you considered leaving everything else in place and only changing this code as well as the num -> vm_num changes below ?

Thanks,
Guenter

Quoted 10 lines
drivers/hwmon/mr75203.c +2 -2
  		if (!in_config)
  			return -ENOMEM;
 
-		memset32(in_config, HWMON_I_INPUT, num);
-		in_config[num] = 0;
+		memset32(in_config, HWMON_I_INPUT, vm_num);
+		in_config[vm_num] = 0;
  		pvt_in.config = in_config;
 
  		pvt_info[index++] = &pvt_in;
#5 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 45 lines · Guenter Roeck
Quoted 44 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -2
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device 
 *pdev)
       }
 
       if (vm_num) {
-             u32 num = vm_num;
-
               ret = pvt_get_regmap(pdev, "vm", pvt);
               if (ret)

                      return ret;

drivers/hwmon/mr75203.c +6 -1
@@ -594,30 +592,28 @@ static int mr75203_probe(struct platform_device 
 *pdev)
               ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
               if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not defined, we
+                      * assume incremental channel numbers.
+                      */
+                     for (i = 0; i < vm_num; i++)
+                             pvt->vm_idx[i] = i;
               } else {
                       for (i = 0; i < vm_num; i++)

                              if (pvt->vm_idx[i] >= vm_num || -                                 pvt->vm_idx[i] == 0xff) { -                                     num = i;
+                                 pvt->vm_idx[i] == 0xff) break;
-                             }
-             }

Quoted 14 lines
Quoted 9 lines

-             /*
-              * Incase intel,vm-map property is not defined, we assume -              * incremental channel numbers.
-              */
-             for (i = num; i < vm_num; i++)
-                     pvt->vm_idx[i] = i;
+                     vm_num = i;
+             }

So this is actually a functional change: In the old code, unspecified channel numbers (num ... vm_num) were filled with incremental channel numbers. This is no longer the case.

The only place that uses pvt->vm_idx[] besides setting it in the probe function is pvr_read_in().
It is quite clear from pvr_read_in() that unspecified channel numbers (num ... vm_num) are not accessed, therefore there is also no need to set them.

    if (channel >= pvt->v_num)
        return -EINVAL;

    vm_idx = pvt->vm_idx[channel];

Therefore I removed the setting of unspecified channels, and only if “intel,vm-map” property is not defined, I set the specified channels in incremental order.

Quoted 7 lines
Quoted 3 lines

-             in_config = devm_kcalloc(dev, num + 1,
+             in_config = devm_kcalloc(dev, vm_num + 1, sizeof(*in_config), GFP_KERNEL);

The relevant difference (and maybe bug in the existing code ?) seems to be this change. Have you considered leaving everything else in place and only changing this code as well as the num -> vm_num changes below ?

Yes, using vm_num instead of num (which will be 0 if “intel,vm-map” property is not defined) is the actual fix.
Both changes seemed to me to be coupled but if you think it will be more clear to split this patch to two, first that removes unnecessary setting of unspecified channels, and second that changes num to vm_num for in_config, then sure I will do it.

--
Thanks, Eliav

maintainer
#6 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel,vm-map" not defined

In reply to Eliav Farber
Quoted 29 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------
  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +3 -3
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device *pdev)
  	}
 
  	if (vm_num) {
-		u32 num = vm_num;
-
  		ret = pvt_get_regmap(pdev, "vm", pvt);
  		if (ret)
  			return ret;
@@ -594,30 +592,28 @@ static int mr75203_probe(struct platform_device *pdev)
  		ret = device_property_read_u8_array(dev, "intel,vm-map",
  						    pvt->vm_idx, vm_num);
  		if (ret) {
-			num = 0;
+			/*
+			 * Incase intel,vm-map property is not defined, we
+			 * assume incremental channel numbers.

drivers/hwmon/mr75203.c static int mr75203_probe(struct platform_device *pdev) · line 605

Quoted 10 lines
drivers/hwmon/mr75203.c +4 -2
+			 */
+			for (i = 0; i < vm_num; i++)
+				pvt->vm_idx[i] = i;
  		} else {
  			for (i = 0; i < vm_num; i++)
  				if (pvt->vm_idx[i] >= vm_num ||
-				    pvt->vm_idx[i] == 0xff) {
-					num = i;
+				    pvt->vm_idx[i] == 0xff)
  					break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ?
Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ? Is that documented somewhere ?

Thanks,
Guenter

Quoted 25 lines
drivers/hwmon/mr75203.c +5 -11
-				}
-		}
 
-		/*
-		 * Incase intel,vm-map property is not defined, we assume
-		 * incremental channel numbers.
-		 */
-		for (i = num; i < vm_num; i++)
-			pvt->vm_idx[i] = i;
+			vm_num = i;
+		}
 
-		in_config = devm_kcalloc(dev, num + 1,
+		in_config = devm_kcalloc(dev, vm_num + 1,
  					 sizeof(*in_config), GFP_KERNEL);
  		if (!in_config)
  			return -ENOMEM;
 
-		memset32(in_config, HWMON_I_INPUT, num);
-		in_config[num] = 0;
+		memset32(in_config, HWMON_I_INPUT, vm_num);
+		in_config[vm_num] = 0;
  		pvt_in.config = in_config;
 
  		pvt_info[index++] = &pvt_in;
#7 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 38 lines · Guenter Roeck
Quoted 37 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -2
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device 
 *pdev)
       }
 
       if (vm_num) {
-             u32 num = vm_num;
-
               ret = pvt_get_regmap(pdev, "vm", pvt);
               if (ret)

                      return ret;

drivers/hwmon/mr75203.c +6 -1
@@ -594,30 +592,28 @@ static int mr75203_probe(struct platform_device 
 *pdev)
               ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
               if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not defined, we
+                      * assume incremental channel numbers.
+                      */
+                     for (i = 0; i < vm_num; i++)
+                             pvt->vm_idx[i] = i;
               } else {
                       for (i = 0; i < vm_num; i++)

                              if (pvt->vm_idx[i] >= vm_num ||

Quoted 8 lines
Quoted 4 lines

-                                 pvt->vm_idx[i] == 0xff) { -                                     num = i;
+                                 pvt->vm_idx[i] == 0xff) break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ?
Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ? Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

--
Regards, Eliav

#8 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 48 lines · Eliav
Quoted 39 lines · Guenter Roeck
Quoted 38 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -2
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device 
 *pdev)
       }
 
       if (vm_num) {
-             u32 num = vm_num;
-
               ret = pvt_get_regmap(pdev, "vm", pvt);
               if (ret)

                      return ret;

drivers/hwmon/mr75203.c +2 -1
@@ -594,30 +592,28 @@ static int mr75203_probe(struct 
 platform_device *pdev)
               ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
               if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not 

defined, we
+                      * assume incremental channel numbers. +                      */
+                     for (i = 0; i < vm_num; i++) +                             pvt->vm_idx[i] = i;
              } else {
                      for (i = 0; i < vm_num; i++) if (pvt->vm_idx[i] >= vm_num ||

Quoted 8 lines
Quoted 4 lines

-                                 pvt->vm_idx[i] == 0xff) { -                                     num = i;
+                                 pvt->vm_idx[i] == 0xff) break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ?
Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ? Is that documented somewhere ?

Quoted 3 lines

According to the code vm_num is limited to 32 because the mask is only 5 bits:

For 5 bit mask, maximum is of course 31 and not 32.

Quoted 9 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Signature · 1 line
Regards, Eliav
maintainer
#9 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

In reply to Farber, Eliav
Quoted 48 lines · Eliav
Quoted 36 lines · Guenter Roeck
Quoted 35 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +6 -3
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device *pdev)
       }
 
       if (vm_num) {
-             u32 num = vm_num;
-
               ret = pvt_get_regmap(pdev, "vm", pvt);
               if (ret)
                       return ret;
@@ -594,30 +592,28 @@ static int mr75203_probe(struct platform_device *pdev)
               ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
               if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not defined, we
+                      * assume incremental channel numbers.
+                      */
+                     for (i = 0; i < vm_num; i++)
+                             pvt->vm_idx[i] = i;
               } else {
                       for (i = 0; i < vm_num; i++)

                              if (pvt->vm_idx[i] >= vm_num ||

Quoted 8 lines
Quoted 4 lines

-                                 pvt->vm_idx[i] == 0xff) { -                                     num = i;
+                                 pvt->vm_idx[i] == 0xff) break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ?
Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ? Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

Quoted 7 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Sorry, my bad. I misread the patch and thought the first part of the if statement was removed.

Anyway, what is the difference between specifying an vm_idx value of 0xff and not specifying anything ? Or, in other words, taking the dt example, the difference between

	intel,vm-map = [03 01 04 ff ff];
and
	intel,vm-map = [03 01 04];

Thanks,
Guenter

#10 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 62 lines · Guenter Roeck
Quoted 51 lines · Eliav
Quoted 39 lines · Guenter Roeck
Quoted 38 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -2
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct platform_device 
 *pdev)
       }
 
       if (vm_num) {
-             u32 num = vm_num;
-
               ret = pvt_get_regmap(pdev, "vm", pvt);
               if (ret)

                      return ret;

drivers/hwmon/mr75203.c +2 -1
@@ -594,30 +592,28 @@ static int mr75203_probe(struct 
 platform_device *pdev)
               ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
               if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not 

defined, we
+                      * assume incremental channel numbers. +                      */
+                     for (i = 0; i < vm_num; i++) +                             pvt->vm_idx[i] = i;
              } else {
                      for (i = 0; i < vm_num; i++) if (pvt->vm_idx[i] >= vm_num ||

Quoted 8 lines
Quoted 4 lines

-                                 pvt->vm_idx[i] == 0xff) { -                                     num = i;
+                                 pvt->vm_idx[i] == 0xff) break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ?
Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ? Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

Quoted 7 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Sorry, my bad. I misread the patch and thought the first part of the if statement was removed.

Quoted 6 lines

Anyway, what is the difference between specifying an vm_idx value of 0xff and not specifying anything ? Or, in other words, taking the dt example, the difference between
       intel,vm-map = [03 01 04 ff ff];
and
       intel,vm-map = [03 01 04];

The actual number of VMs is read from a HW register:
    ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
    ...
    vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

Also, using:
    ret = device_property_read_u8_array(dev, "intel,vm-map", vm_idx, vm_num);
in the driver will fail if vm_num > sizeof array in device-tree.

So, if for example vm_num = 5, but you will want to map only 3 of them you most set property to be:
    intel,vm-map = [03 01 04 ff ff];
otherwise if you set:
    intel,vm-map = [03 01 04];
it will assume the property doesn't, and will continue the flow in code as if it doesn’t exist (which is not what the user wanted, and before my fix also has a bug).

--
Regards, Eliav

maintainer
#11 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 82 lines · Eliav
Quoted 64 lines · Guenter Roeck
Quoted 53 lines · Eliav
Quoted 41 lines · Guenter Roeck
Quoted 40 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set
to 0, and no voltage channel infos are allocated.

---
  drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

  1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -2
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct
 platform_device *pdev)
       }
 
       if (vm_num) {
-             u32 num = vm_num;
-
               ret = pvt_get_regmap(pdev, "vm", pvt);
               if (ret)

                      return ret;

drivers/hwmon/mr75203.c +2 -1
@@ -594,30 +592,28 @@ static int mr75203_probe(struct
 platform_device *pdev)
               ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
               if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not

defined, we
+                      * assume incremental channel numbers. +                      */
+                     for (i = 0; i < vm_num; i++) +                             pvt->vm_idx[i] = i;
              } else {
                      for (i = 0; i < vm_num; i++) if (pvt->vm_idx[i] >= vm_num || -                                 pvt->vm_idx[i] == 0xff) {

Quoted 8 lines
Quoted 3 lines

-                                     num = i;
+                                 pvt->vm_idx[i] == 0xff) break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ? Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ?
Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

Quoted 7 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Sorry, my bad. I misread the patch and thought the first part of the if statement was removed.

Quoted 6 lines

Anyway, what is the difference between specifying an vm_idx value of 0xff and not specifying anything ? Or, in other words, taking the dt example, the difference between
       intel,vm-map = [03 01 04 ff ff];
and
       intel,vm-map = [03 01 04];

The actual number of VMs is read from a HW register:
    ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
    ...
    vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

Also, using:
    ret = device_property_read_u8_array(dev, "intel,vm-map", vm_idx, vm_num);
in the driver will fail if vm_num > sizeof array in device-tree.

Quoted 8 lines

So, if for example vm_num = 5, but you will want to map only 3 of them you most set property to be:
    intel,vm-map = [03 01 04 ff ff];
otherwise if you set:
    intel,vm-map = [03 01 04];
it will assume the property doesn't, and will continue the flow in code as if it doesn’t exist (which is not what the user wanted, and before my fix also has a bug).

There should be some error handling to catch this case (ie if the number of entries does not match the expected count), or if a value in the array is larger or equal to vm_num. Today the latter is silently handled as end of entries (similar to 0xff), but that should result in an error. This would avoid situations like

	intel,vm-map = [01 02 03 04 05];

ie where the person writing the devicetree file accidentally entered index values starting with 1 instead of 0. A mismatch between vm_num and the number of entries in the array is silently handled as if there was no property at all, which is at the very least misleading and most definitely unexpected and should also result in an error.

Also, what happens if the devicetree content is something like the following ? Would that be valid ?

	intel,vm-map = [00 01 01 01 01 01];

Thanks,
Guenter

#12 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 93 lines · Guenter Roeck
Quoted 83 lines · Eliav
Quoted 20 lines · Guenter Roeck
Quoted 19 lines · Eliav
Quoted 18 lines · Guenter Roeck
Quoted 17 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set
to 0, and no voltage channel infos are allocated.

---
   drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

   1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -0
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct
 platform_device *pdev)
        }
 
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

drivers/hwmon/mr75203.c static int mr75203_probe(struct · line 593

Quoted 9 lines
drivers/hwmon/mr75203.c +0 -2
        if (vm_num) {
-             u32 num = vm_num;
-
                ret = pvt_get_regmap(pdev, "vm", pvt);
                if (ret)

                       return ret;

drivers/hwmon/mr75203.c +0 -0
@@ -594,30 +592,28 @@ static int mr75203_probe(struct
 platform_device *pdev)
                ret = device_property_read_u8_array(dev, 

"intel,vm-map",

Quoted 35 lines
Quoted 25 lines
Quoted 18 lines
Quoted 9 lines
drivers/hwmon/mr75203.c +2 -1
 pvt->vm_idx, vm_num);
                if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not

defined, we
+                      * assume incremental channel numbers. +                      */
+                     for (i = 0; i < vm_num; i++)

Quoted 8 lines

+ pvt->vm_idx[i] = i;
               } else {
                       for (i = 0; i < vm_num; i++) if (pvt->vm_idx[i] >= vm_num ||
- pvt->vm_idx[i] == 0xff) {
-                                     num = i;
+ pvt->vm_idx[i] == 0xff)
                                       break;

Quoted 4 lines

So all vm_idx values from 0x00 to 0xfe would be acceptable ? Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ?
Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

Quoted 7 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Sorry, my bad. I misread the patch and thought the first part of the if statement was removed.

Quoted 6 lines

Anyway, what is the difference between specifying an vm_idx value of 0xff and not specifying anything ? Or, in other words, taking the dt example, the difference between
        intel,vm-map = [03 01 04 ff ff];
and
        intel,vm-map = [03 01 04];

The actual number of VMs is read from a HW register:
    ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
    ...
    vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

Also, using:
    ret = device_property_read_u8_array(dev, "intel,vm-map", vm_idx, vm_num);
in the driver will fail if vm_num > sizeof array in device-tree.

Quoted 8 lines

So, if for example vm_num = 5, but you will want to map only 3 of them you most set property to be:
    intel,vm-map = [03 01 04 ff ff];
otherwise if you set:
    intel,vm-map = [03 01 04];
it will assume the property doesn't, and will continue the flow in code as if it doesn’t exist (which is not what the user wanted, and before my fix also has a bug).

Quoted 11 lines

There should be some error handling to catch this case (ie if the number of entries does not match the expected count), or if a value in the array is larger or equal to vm_num. Today the latter is silently handled as end of entries (similar to 0xff), but that should result in an error. This would avoid situations like
       intel,vm-map = [01 02 03 04 05];
ie where the person writing the devicetree file accidentally entered index values starting with 1 instead of 0. A mismatch between vm_num and the number of entries in the array is silently handled as if there was no property at all, which is at the very least misleading and most definitely unexpected and should also result in an error.

I assume it is possible to tell according to the return value, if property doesn’t exist at all, or if it does exists and size of array in device-tree is smaller than vm_num.
In [PATCH v3 17/19] Andy wrote that “code shouldn't be a YAML validator. Drop this and make sure you have correct DT schema” so I’m a bit confused if code should validate “intel,bm-map” or if it is the user responsibility. As this property was not added by me, I prefer not to fix it as part of this series of patches.

Quoted 3 lines

Also, what happens if the devicetree content is something like the following ? Would that be valid ?
       intel,vm-map = [00 01 01 01 01 01];

If device-tree content would be:
    intel,vm-map = [00 01 01 01 01 01];
and assuming 16 channels for each VM, the hwmon sub-system will expose 90 sysfs to read voltage values.
In practice 16 – 31, 32 – 47, 48 – 63, 64 – 89 will all report the same input signals for VM1.

--
Regards, Eliav

maintainer
#13 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 106 lines · Eliav
Quoted 92 lines · Guenter Roeck
Quoted 82 lines · Eliav
Quoted 64 lines · Guenter Roeck
Quoted 53 lines · Eliav
Quoted 41 lines · Guenter Roeck
Quoted 40 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set
to 0, and no voltage channel infos are allocated.

---
   drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

   1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -2
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct
 platform_device *pdev)
        }
 
        if (vm_num) {
-             u32 num = vm_num;
-
                ret = pvt_get_regmap(pdev, "vm", pvt);
                if (ret)

                       return ret;

drivers/hwmon/mr75203.c +2 -1
@@ -594,30 +592,28 @@ static int mr75203_probe(struct
 platform_device *pdev)
                ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
                if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not

defined, we
+                      * assume incremental channel numbers. +                      */
+                     for (i = 0; i < vm_num; i++)
+ pvt->vm_idx[i] = i;
               } else {
                       for (i = 0; i < vm_num; i++) if (pvt->vm_idx[i] >= vm_num ||
- pvt->vm_idx[i] == 0xff) {

Quoted 8 lines
Quoted 3 lines

-                                     num = i;
+ pvt->vm_idx[i] == 0xff)
                                       break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ? Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ?
Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

Quoted 7 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Sorry, my bad. I misread the patch and thought the first part of the if statement was removed.

Quoted 6 lines

Anyway, what is the difference between specifying an vm_idx value of 0xff and not specifying anything ? Or, in other words, taking the dt example, the difference between
        intel,vm-map = [03 01 04 ff ff];
and
        intel,vm-map = [03 01 04];

The actual number of VMs is read from a HW register:
    ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
    ...
    vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

Also, using:
    ret = device_property_read_u8_array(dev, "intel,vm-map", vm_idx, vm_num);
in the driver will fail if vm_num > sizeof array in device-tree.

Quoted 8 lines

So, if for example vm_num = 5, but you will want to map only 3 of them you most set property to be:
    intel,vm-map = [03 01 04 ff ff];
otherwise if you set:
    intel,vm-map = [03 01 04];
it will assume the property doesn't, and will continue the flow in code as if it doesn’t exist (which is not what the user wanted, and before my fix also has a bug).

Quoted 11 lines

There should be some error handling to catch this case (ie if the number of entries does not match the expected count), or if a value in the array is larger or equal to vm_num. Today the latter is silently handled as end of entries (similar to 0xff), but that should result in an error. This would avoid situations like
       intel,vm-map = [01 02 03 04 05];
ie where the person writing the devicetree file accidentally entered index values starting with 1 instead of 0. A mismatch between vm_num and the number of entries in the array is silently handled as if there was no property at all, which is at the very least misleading and most definitely unexpected and should also result in an error.

Quoted 9 lines

I assume it is possible to tell according to the return value, if property doesn’t exist at all, or if it does exists and size of array in device-tree is smaller than vm_num.
In [PATCH v3 17/19] Andy wrote that “code shouldn't be a YAML validator. Drop this and make sure you have correct DT schema” so I’m a bit confused if code should validate “intel,bm-map” or if it is the user responsibility. As this property was not added by me, I prefer not to fix it as part of this series of patches.

You are changing the driver all over the place with 19 patches, including this code, but you don't want to add code that validates the devicetree data ? That seems odd.

Quoted 5 lines
Quoted 4 lines

Also, what happens if the devicetree content is something like the following ? Would that be valid ?
       intel,vm-map = [00 01 01 01 01 01];

Quoted 7 lines

If device-tree content would be:
    intel,vm-map = [00 01 01 01 01 01];
and assuming 16 channels for each VM, the hwmon sub-system will expose 90 sysfs to read voltage values.
In practice 16 – 31, 32 – 47, 48 – 63, 64 – 89 will all report the same input signals for VM1.

Does that make any sense, and is there a valid reason to have a mapping table like the one in this example ?

Thanks,
Guenter

#14 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 122 lines · Guenter Roeck

CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe.

Quoted 107 lines · Eliav
Quoted 93 lines · Guenter Roeck
Quoted 83 lines · Eliav
Quoted 20 lines · Guenter Roeck
Quoted 19 lines · Eliav
Quoted 18 lines · Guenter Roeck
Quoted 17 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set
to 0, and no voltage channel infos are allocated.

---
   drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

   1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -0
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct
 platform_device *pdev)
        }
 
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

drivers/hwmon/mr75203.c static int mr75203_probe(struct · line 593

Quoted 9 lines
drivers/hwmon/mr75203.c +0 -2
        if (vm_num) {
-             u32 num = vm_num;
-
                ret = pvt_get_regmap(pdev, "vm", pvt);
                if (ret)

                       return ret;

drivers/hwmon/mr75203.c +0 -0
@@ -594,30 +592,28 @@ static int mr75203_probe(struct
 platform_device *pdev)
                ret = device_property_read_u8_array(dev, 

"intel,vm-map",

Quoted 35 lines
Quoted 25 lines
Quoted 18 lines
Quoted 9 lines
drivers/hwmon/mr75203.c +2 -1
 pvt->vm_idx, vm_num);
                if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not

defined, we
+                      * assume incremental channel numbers. +                      */
+                     for (i = 0; i < vm_num; i++)

Quoted 8 lines

+ pvt->vm_idx[i] = i;
               } else {
                       for (i = 0; i < vm_num; i++) if (pvt->vm_idx[i] >= vm_num ||
- pvt->vm_idx[i] == 0xff) {
- num = i;
+ pvt->vm_idx[i] == 0xff)
break;

Quoted 4 lines

So all vm_idx values from 0x00 to 0xfe would be acceptable ? Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ?
Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

Quoted 7 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Sorry, my bad. I misread the patch and thought the first part of the if statement was removed.

Quoted 6 lines

Anyway, what is the difference between specifying an vm_idx value of 0xff and not specifying anything ? Or, in other words, taking the dt example, the difference between
        intel,vm-map = [03 01 04 ff ff];
and
        intel,vm-map = [03 01 04];

The actual number of VMs is read from a HW register:
    ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
    ...
    vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

Also, using:
    ret = device_property_read_u8_array(dev, "intel,vm-map", vm_idx, vm_num);
in the driver will fail if vm_num > sizeof array in device-tree.

Quoted 8 lines

So, if for example vm_num = 5, but you will want to map only 3 of them you most set property to be:
    intel,vm-map = [03 01 04 ff ff];
otherwise if you set:
    intel,vm-map = [03 01 04];
it will assume the property doesn't, and will continue the flow in code as if it doesn’t exist (which is not what the user wanted, and before my fix also has a bug).

Quoted 11 lines

There should be some error handling to catch this case (ie if the number of entries does not match the expected count), or if a value in the array is larger or equal to vm_num. Today the latter is silently handled as end of entries (similar to 0xff), but that should result in an error. This would avoid situations like
       intel,vm-map = [01 02 03 04 05];
ie where the person writing the devicetree file accidentally entered index values starting with 1 instead of 0. A mismatch between vm_num and the number of entries in the array is silently handled as if there was no property at all, which is at the very least misleading and most definitely unexpected and should also result in an error.

Quoted 9 lines

I assume it is possible to tell according to the return value, if property doesn’t exist at all, or if it does exists and size of array in device-tree is smaller than vm_num.
In [PATCH v3 17/19] Andy wrote that “code shouldn't be a YAML validator. Drop this and make sure you have correct DT schema” so I’m a bit confused if code should validate “intel,bm-map” or if it is the user responsibility. As this property was not added by me, I prefer not to fix it as part of this series of patches.

Quoted 4 lines

You are changing the driver all over the place with 19 patches, including this code, but you don't want to add code that validates the devicetree data ? That seems odd.

OK. I have added patch #20 to validate that same VM index doesn't appear more than once in intel,vm-map.

u32 vm_mask = 0;

for (i = 0; i < vm_num; i++) {
    if (vm_idx[i] >= vm_num || vm_idx[i] == 0xff)
        break;

    if (vm_mask & BIT(vm_idx[i])) {
        dev_err(dev, "Same VM appears more than once in intel,vm-map\n", vm_idx[i]);
        return EINVAL;
    }

    vm_mask |= BIT(vm_idx[i]);
}

Quoted 13 lines
Quoted 5 lines
Quoted 4 lines

Also, what happens if the devicetree content is something like the following ? Would that be valid ?
       intel,vm-map = [00 01 01 01 01 01];

Quoted 7 lines

If device-tree content would be:
     intel,vm-map = [00 01 01 01 01 01];
and assuming 16 channels for each VM, the hwmon sub-system will expose 90 sysfs to read voltage values.
In practice 16 – 31, 32 – 47, 48 – 63, 64 – 89 will all report the same input signals for VM1.

Quoted 2 lines

Does that make any sense, and is there a valid reason to have a mapping table like the one in this example ?

I can't find any sense in having such a mapping.
Anyway the new patch will not allow it to happen.

--
Regards, Eliav

maintainer
#15 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 126 lines · Eliav
Quoted 121 lines · Guenter Roeck

CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe.

Quoted 106 lines · Eliav
Quoted 92 lines · Guenter Roeck
Quoted 82 lines · Eliav
Quoted 64 lines · Guenter Roeck
Quoted 53 lines · Eliav
Quoted 41 lines · Guenter Roeck
Quoted 40 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set
to 0, and no voltage channel infos are allocated.

---
   drivers/hwmon/mr75203.c | 28 ++++++++++++----------------

   1 file changed, 12 insertions(+), 16 deletions(-)
drivers/hwmon/mr75203.c +0 -2
index 046523d47c29..0e29877a1a9c 100644
@@ -580,8 +580,6 @@ static int mr75203_probe(struct
 platform_device *pdev)
        }
 
        if (vm_num) {
-             u32 num = vm_num;
-
                ret = pvt_get_regmap(pdev, "vm", pvt);
                if (ret)

                       return ret;

drivers/hwmon/mr75203.c +2 -1
@@ -594,30 +592,28 @@ static int mr75203_probe(struct
 platform_device *pdev)
                ret = device_property_read_u8_array(dev, "intel,vm-map",
 pvt->vm_idx, vm_num);
                if (ret) {
-                     num = 0;
+                     /*
+                      * Incase intel,vm-map property is not

defined, we
+                      * assume incremental channel numbers. +                      */
+                     for (i = 0; i < vm_num; i++)
+ pvt->vm_idx[i] = i;
               } else {
                       for (i = 0; i < vm_num; i++) if (pvt->vm_idx[i] >= vm_num ||
- pvt->vm_idx[i] == 0xff) {

Quoted 8 lines
Quoted 3 lines

- num = i;
+ pvt->vm_idx[i] == 0xff)
break;

So all vm_idx values from 0x00 to 0xfe would be acceptable ? Does the chip really have that many registers (0x200 + 0x40 + 0x200 * 0xfe) ?
Is that documented somewhere ?

According to the code vm_num is limited to 32 because the mask is only 5 bits:

Quoted 7 lines

#define VM_NUM_MSK    GENMASK(20, 16)
#define VM_NUM_SFT    16
vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

In practice according to the data sheet I have:
0 <= VM instances <= 8

Sorry, my bad. I misread the patch and thought the first part of the if statement was removed.

Quoted 6 lines

Anyway, what is the difference between specifying an vm_idx value of 0xff and not specifying anything ? Or, in other words, taking the dt example, the difference between
        intel,vm-map = [03 01 04 ff ff];
and
        intel,vm-map = [03 01 04];

The actual number of VMs is read from a HW register:
    ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
    ...
    vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;

Also, using:
    ret = device_property_read_u8_array(dev, "intel,vm-map", vm_idx, vm_num);
in the driver will fail if vm_num > sizeof array in device-tree.

Quoted 8 lines

So, if for example vm_num = 5, but you will want to map only 3 of them you most set property to be:
    intel,vm-map = [03 01 04 ff ff];
otherwise if you set:
    intel,vm-map = [03 01 04];
it will assume the property doesn't, and will continue the flow in code as if it doesn’t exist (which is not what the user wanted, and before my fix also has a bug).

Quoted 11 lines

There should be some error handling to catch this case (ie if the number of entries does not match the expected count), or if a value in the array is larger or equal to vm_num. Today the latter is silently handled as end of entries (similar to 0xff), but that should result in an error. This would avoid situations like
       intel,vm-map = [01 02 03 04 05];
ie where the person writing the devicetree file accidentally entered index values starting with 1 instead of 0. A mismatch between vm_num and the number of entries in the array is silently handled as if there was no property at all, which is at the very least misleading and most definitely unexpected and should also result in an error.

Quoted 9 lines

I assume it is possible to tell according to the return value, if property doesn’t exist at all, or if it does exists and size of array in device-tree is smaller than vm_num.
In [PATCH v3 17/19] Andy wrote that “code shouldn't be a YAML validator. Drop this and make sure you have correct DT schema” so I’m a bit confused if code should validate “intel,bm-map” or if it is the user responsibility. As this property was not added by me, I prefer not to fix it as part of this series of patches.

Quoted 4 lines

You are changing the driver all over the place with 19 patches, including this code, but you don't want to add code that validates the devicetree data ? That seems odd.

Quoted 7 lines

OK. I have added patch #20 to validate that same VM index doesn't appear more than once in intel,vm-map.

u32 vm_mask = 0;

for (i = 0; i < vm_num; i++) {
    if (vm_idx[i] >= vm_num || vm_idx[i] == 0xff)

I think "vm_idx[i] >= vm_num && vm_idx[i] != 0xff) should also be invalid, ie.

	if (vm_idx[i] == 0xff)
		break;
	if (vm_idx[i] >= vm_num)
		return -EINVAL;

Thanks,
Guenter

Quoted 34 lines

        break;

    if (vm_mask & BIT(vm_idx[i])) {
        dev_err(dev, "Same VM appears more than once in intel,vm-map\n", vm_idx[i]);
        return EINVAL;
    }

    vm_mask |= BIT(vm_idx[i]);
}

Quoted 13 lines
Quoted 5 lines
Quoted 4 lines

Also, what happens if the devicetree content is something like the following ? Would that be valid ?
       intel,vm-map = [00 01 01 01 01 01];

Quoted 7 lines

If device-tree content would be:
     intel,vm-map = [00 01 01 01 01 01];
and assuming 16 channels for each VM, the hwmon sub-system will expose 90 sysfs to read voltage values.
In practice 16 – 31, 32 – 47, 48 – 63, 64 – 89 will all report the same input signals for VM1.

Quoted 2 lines

Does that make any sense, and is there a valid reason to have a mapping table like the one in this example ?

I can't find any sense in having such a mapping.
Anyway the new patch will not allow it to happen.

Signature · 1 line
Regards, Eliav
#16 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

In reply to Farber, Eliav
Quoted 2 lines · Eliav
Quoted 1 line · Guenter Roeck

On Thu, Sep 01, 2022 at 11:39:58AM +0300, Farber, Eliav wrote:

...

Quoted 12 lines
Quoted 11 lines

There should be some error handling to catch this case (ie if the number of entries does not match the expected count), or if a value in the array is larger or equal to vm_num. Today the latter is silently handled as end of entries (similar to 0xff), but that should result in an error. This would avoid situations like
       intel,vm-map = [01 02 03 04 05];
ie where the person writing the devicetree file accidentally entered index values starting with 1 instead of 0. A mismatch between vm_num and the number of entries in the array is silently handled as if there was no property at all, which is at the very least misleading and most definitely unexpected and should also result in an error.

Quoted 8 lines

I assume it is possible to tell according to the return value, if property doesn’t exist at all, or if it does exists and size of array in device-tree is smaller than vm_num.
In [PATCH v3 17/19] Andy wrote that “code shouldn't be a YAML validator. Drop this and make sure you have correct DT schema” so I’m a bit confused if code should validate “intel,bm-map” or if it is the user responsibility. As this property was not added by me, I prefer not to fix it as part of this series of patches.

I'm just a messenger of Rob's words. If I'm mistaken I would like Rob to correct me.

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

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel,vm-map" not defined

In reply to Eliav Farber
Quoted 2 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

Care to provide a Fixes tag for all fixes in your series. Also don't forget to start series with fixes followed by cleanups and new features..

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

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 6 lines · Andy Shevchenko
Quoted 2 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

Care to provide a Fixes tag for all fixes in your series. Also don't forget to start series with fixes followed by cleanups and new features..

For v4 I provided a Fixes tag where it was relevant.
I also changed the order of some patches so that all fixes will be first.

--
Thanks, Eliav

#19 lore raw

Re: [PATCH v3 02/19] hwmon: (mr75203) fix VM sensor allocation when "intel, vm-map" not defined

Quoted 9 lines · Eliav
Quoted 7 lines · Andy Shevchenko
Quoted 2 lines · Eliav Farber

Bug fix - in case "intel,vm-map" is missing in device-tree ,'num' is set to 0, and no voltage channel infos are allocated.

Care to provide a Fixes tag for all fixes in your series. Also don't forget to
start series with fixes followed by cleanups and new features..

For v4 I provided a Fixes tag where it was relevant.

Thanks!

Quoted 1 line

I also changed the order of some patches so that all fixes will be first.

Note, fixes should prepend the other patches in the series.

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

PATCH v3 3/19 hwmon: (mr75203) update pvt->v_num to the actual number of used sensors

In reply to Eliav Farber

This issue is relevant when "intel,vm-map" is set in device-tree, and defines a lower number of VMs than actually supported.

This change is needed for all places that use pvt->v_num later on in the code.

---
 drivers/hwmon/mr75203.c | 1 +
 1 file changed, 1 insertion(+)
drivers/hwmon/mr75203.c +1 -0
index 0e29877a1a9c..f89f7bb5d698 100644
@@ -605,6 +605,7 @@ static int mr75203_probe(struct platform_device *pdev)
 					break;
 
 			vm_num = i;
+			pvt->v_num = i;
 		}
 
 		in_config = devm_kcalloc(dev, vm_num + 1,
Signature · 1 line
2.37.1
maintainer
#21 lore raw

Re: [PATCH v3 03/19] hwmon: (mr75203) update pvt->v_num to the actual number of used sensors

Quoted 16 lines · Eliav Farber

This issue is relevant when "intel,vm-map" is set in device-tree, and defines a lower number of VMs than actually supported.

This change is needed for all places that use pvt->v_num later on in the code.

---
  drivers/hwmon/mr75203.c | 1 +
  1 file changed, 1 insertion(+)
drivers/hwmon/mr75203.c +0 -0
index 0e29877a1a9c..f89f7bb5d698 100644

drivers/hwmon/mr75203.c static int mr75203_probe(struct platform_device *pdev) · line 608

Quoted 4 lines
drivers/hwmon/mr75203.c +1 -0
@@ -605,6 +605,7 @@ static int mr75203_probe(struct platform_device *pdev)
  					break;
 
  			vm_num = i;
+			pvt->v_num = i;

How about the existing assignment in the probe function ?

Quoted 3 lines
drivers/hwmon/mr75203.c +0 -0
  		}
 
  		in_config = devm_kcalloc(dev, vm_num + 1,
#22 lore raw

Re: [PATCH v3 03/19] hwmon: (mr75203) update pvt->v_num to the actual number of used sensors

Quoted 20 lines · Guenter Roeck
Quoted 19 lines · Eliav Farber

This issue is relevant when "intel,vm-map" is set in device-tree, and defines a lower number of VMs than actually supported.

This change is needed for all places that use pvt->v_num later on in the code.

---
  drivers/hwmon/mr75203.c | 1 +

  1 file changed, 1 insertion(+)
drivers/hwmon/mr75203.c +0 -0
index 0e29877a1a9c..f89f7bb5d698 100644
@@ -605,6 +605,7 @@ static int mr75203_probe(struct platform_device 
 *pdev)
                                       break;
 
Quoted 4 lines

drivers/hwmon/mr75203.c static int mr75203_probe(struct platform_device *pdev) · line 608

Quoted 2 lines
drivers/hwmon/mr75203.c +1 -0
                       vm_num = i;
+                     pvt->v_num = i;

How about the existing assignment in the probe function ?

The probe function uses a local variable vm_num which is also updated (just one line earlier in code) in the previous patch.

--
Thanks, Eliav

maintainer
#23 lore raw

Re: [PATCH v3 03/19] hwmon: (mr75203) update pvt->v_num to the actual number of used sensors

Quoted 20 lines · Eliav
Quoted 19 lines · Guenter Roeck
Quoted 18 lines · Eliav Farber

This issue is relevant when "intel,vm-map" is set in device-tree, and defines a lower number of VMs than actually supported.

This change is needed for all places that use pvt->v_num later on in the code.

---
  drivers/hwmon/mr75203.c | 1 +

  1 file changed, 1 insertion(+)
drivers/hwmon/mr75203.c +0 -0
index 0e29877a1a9c..f89f7bb5d698 100644
@@ -605,6 +605,7 @@ static int mr75203_probe(struct platform_device *pdev)
                                       break;
 
Quoted 7 lines
Quoted 4 lines

drivers/hwmon/mr75203.c static int mr75203_probe(struct platform_device *pdev) · line 608

Quoted 2 lines
drivers/hwmon/mr75203.c +1 -0
                       vm_num = i;
+                     pvt->v_num = i;

How about the existing assignment in the probe function ?

The probe function uses a local variable vm_num which is also updated (just one line earlier in code) in the previous patch.

I meant

	pvt->v_num = vm_num;

at line 536 of the current code, but I guess the idea is to overwrite that.

Thanks,
Guenter

#24 lore raw

PATCH v3 4/19 dt-bindings: hwmon: (mr75203) change "reset" property to be optional

In reply to Eliav Farber

Change "reset" property to be optional instead of required, for SOCs that don't support a reset controller.

---
V3 -> v2:
- Change "reset" property to be optional instead of adding new "reset-control-skip" property.

 Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml | 1 -
 1 file changed, 1 deletion(-)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -1
index 6abde48b746e..2ec4b9da4b92 100644
@@ -49,7 +49,6 @@ required:
   - reg
   - reg-names
   - clocks
-  - resets
   - "#thermal-sensor-cells"
 
 additionalProperties: false
Signature · 1 line
2.37.1
#25 lore raw

Re: [PATCH v3 04/19] dt-bindings: hwmon: (mr75203) change "reset" property to be optional

Quoted 17 lines · Eliav Farber

Change "reset" property to be optional instead of required, for SOCs that don't support a reset controller.

---
V3 -> v2:
- Change "reset" property to be optional instead of adding new "reset-control-skip" property.

 Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml | 1 -

 1 file changed, 1 deletion(-)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index 6abde48b746e..2ec4b9da4b92 100644
@@ -49,7 +49,6 @@ required:
Quoted 4 lines

   - reg
   - reg-names
   - clocks
- - resets

Is this just for mr76006? Or does mr75203 work without reset as well?

If it is the former, maybe a new compatible should be added and resets should be kept required

if:
properties:

    compatible:
      contains:
        const: moortec,mr75203

regards
Philipp

#26 lore raw

Re: [PATCH v3 04/19] dt-bindings: hwmon: (mr75203) change "reset" property to be optional

Quoted 25 lines · Philipp Zabel
Quoted 17 lines · Eliav Farber

Change "reset" property to be optional instead of required, for SOCs that don't support a reset controller.

---
V3 -> v2:
- Change "reset" property to be optional instead of adding new "reset-control-skip" property.

 Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml | 1 -

 1 file changed, 1 deletion(-)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index 6abde48b746e..2ec4b9da4b92 100644
@@ -49,7 +49,6 @@ required:
Quoted 4 lines

   - reg
   - reg-names
   - clocks
-  - resets

Is this just for mr76006? Or does mr75203 work without reset as well?

Quoted 2 lines

If it is the former, maybe a new compatible should be added and resets should be kept required

mr75203 also works without a reset.
As I replied in PATCH v3 14/19, series 5/6 is relevant only for the temperature sensor.
The “reset” property is relevant to the controller.

--
Regards, Eliav

Reviewed-by
#27 lore raw

Re: [PATCH v3 04/19] dt-bindings: hwmon: (mr75203) change "reset" property to be optional

Quoted 29 lines · Eliav
Quoted 28 lines · Philipp Zabel
Quoted 20 lines · Eliav Farber

Change "reset" property to be optional instead of required, for SOCs that
don't support a reset controller.

---
V3 -> v2:
- Change "reset" property to be optional instead of adding new "reset-control-skip" property.

 Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml | 1 -

 1 file changed, 1 deletion(-)
(quoted diff) +0 -0

a/Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml b/Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml index 6abde48b746e..2ec4b9da4b92 100644

Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
@@ -49,7 +49,6 @@ required:
Quoted 4 lines

   - reg
   - reg-names
   - clocks
-  - resets

Is this just for mr76006? Or does mr75203 work without reset as well?

Quoted 7 lines
Quoted 2 lines

If it is the former, maybe a new compatible should be added and resets should be kept required

mr75203 also works without a reset.
As I replied in PATCH v3 14/19, series 5/6 is relevant only for the temperature sensor.
The “reset” property is relevant to the controller.

In that case,

regards
Philipp

maintainer
#28 lore raw

Re: [PATCH v3 04/19] dt-bindings: hwmon: (mr75203) change "reset" property to be optional

In reply to Eliav Farber
Quoted 1 line · Eliav Farber

Change "reset" property to be optional instead of required, for SOCs that

'resets'

And subject too.

Quoted 27 lines

don't support a reset controller.

---
V3 -> v2:
- Change "reset" property to be optional instead of adding new "reset-control-skip" property.

 Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml | 1 -
 1 file changed, 1 deletion(-)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -1
index 6abde48b746e..2ec4b9da4b92 100644
@@ -49,7 +49,6 @@ required:
   - reg
   - reg-names
   - clocks
-  - resets
   - "#thermal-sensor-cells"
 
 additionalProperties: false
Signature · 1 line
2.37.1
#29 lore raw

Re: [PATCH v3 04/19] dt-bindings: hwmon: (mr75203) change "reset" property to be optional

Quoted 6 lines · Rob Herring
Quoted 1 line · Eliav Farber

Change "reset" property to be optional instead of required, for SOCs that

'resets'

And subject too.

Fixed in v4.

#30 lore raw

PATCH v3 5/19 hwmon: (mr75203) skip reset-control deassert for SOCs that don't support it

In reply to Eliav Farber

Don't fail the probe function and don't deassert the reset controller if a "reset" property doesn't exist in the device tree.

Change is done for SOCs that don't support a reset controller.

---
V3 -> v2:
- Change "reset" property to be optional instead of skipping it.

 drivers/hwmon/mr75203.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
drivers/hwmon/mr75203.c +7 -4
index f89f7bb5d698..901030125127 100644
@@ -525,14 +525,17 @@ static int mr75203_probe(struct platform_device *pdev)
 		return ret;
 	}
 
-	pvt->rst = devm_reset_control_get_exclusive(dev, NULL);
+	pvt->rst = devm_reset_control_get_optional_exclusive(dev, NULL);
 	if (IS_ERR(pvt->rst))
 		return dev_err_probe(dev, PTR_ERR(pvt->rst),
 				     "failed to get reset control\n");
 
-	ret = pvt_reset_control_deassert(dev, pvt);
-	if (ret)
-		return dev_err_probe(dev, ret, "cannot deassert reset control\n");
+	if (pvt->rst) {
+		ret = pvt_reset_control_deassert(dev, pvt);
+		if (ret)
+			return dev_err_probe(dev, ret,
+					     "cannot deassert reset control\n");
+	}
 
 	ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
 	if(ret < 0)
Signature · 1 line
2.37.1
Reviewed-by on v3 5/19
#31 lore raw

Re: [PATCH v3 05/19] hwmon: (mr75203) skip reset-control deassert for SOCs that don't support it

Hi Eliav,

Quoted 4 lines · Eliav Farber

Don't fail the probe function and don't deassert the reset controller if a "reset" property doesn't exist in the device tree.

Change is done for SOCs that don't support a reset controller.

Not strictly related to this patch, but reading the context I wonder whether it is required that clock/reset are running/deasserted all the time. Would it make sense to implement some kind of runtime PM?

Quoted 1 line

This change is

regards
Philipp

#32 lore raw

PATCH v3 6/19 hwmon: (mr75203) fix multi-channel voltage reading

In reply to Eliav Farber

Fix voltage allocation and reading to support all channels in all VMs. Prior to this change allocation and reading were done only for the first channel in each VM.
This change counts the total number of channels for allocation, and takes into account the channel offset when reading the sample data register.

---
V3 -> V2:
- Remove configuration of ip-polling register to a separate commit.
- Explain the fix.

 drivers/hwmon/mr75203.c | 27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)
drivers/hwmon/mr75203.c +17 -10
index 901030125127..639f5a300170 100644
@@ -68,8 +68,9 @@
 
 /* VM Individual Macro Register */
 #define VM_COM_REG_SIZE	0x200
-#define VM_SDIF_DONE(n)	(VM_COM_REG_SIZE + 0x34 + 0x200 * (n))
-#define VM_SDIF_DATA(n)	(VM_COM_REG_SIZE + 0x40 + 0x200 * (n))
+#define VM_SDIF_DONE(vm)	(VM_COM_REG_SIZE + 0x34 + 0x200 * (vm))
+#define VM_SDIF_DATA(vm, ch)	\
+	(VM_COM_REG_SIZE + 0x40 + 0x200 * (vm) + 0x4 * (ch))
 
 /* SDA Slave Register */
 #define IP_CTRL			0x00
@@ -115,6 +116,7 @@ struct pvt_device {
 	u32			t_num;
 	u32			p_num;
 	u32			v_num;
+	u32			c_num;
 	u32			ip_freq;
 	u8			*vm_idx;
 };
@@ -179,13 +181,14 @@ static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val)
 	struct pvt_device *pvt = dev_get_drvdata(dev);
 	struct regmap *v_map = pvt->v_map;
 	u32 n, stat;
-	u8 vm_idx;
+	u8 vm_idx, ch_idx;
 	int ret;
 
-	if (channel >= pvt->v_num)
+	if (channel >= pvt->v_num * pvt->c_num)
 		return -EINVAL;
 
-	vm_idx = pvt->vm_idx[channel];
+	vm_idx = pvt->vm_idx[channel / pvt->c_num];
+	ch_idx = channel % pvt->c_num;
 
 	switch (attr) {
 	case hwmon_in_input:
@@ -196,7 +199,7 @@ static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val)
 		if (ret)
 			return ret;
 
-		ret = regmap_read(v_map, VM_SDIF_DATA(vm_idx), &n);
+		ret = regmap_read(v_map, VM_SDIF_DATA(vm_idx, ch_idx), &n);
 		if(ret < 0)
 			return ret;
 
@@ -500,7 +503,7 @@ static int pvt_reset_control_deassert(struct device *dev, struct pvt_device *pvt
 static int mr75203_probe(struct platform_device *pdev)
 {
 	const struct hwmon_channel_info **pvt_info;
-	u32 ts_num, vm_num, pd_num, val, index, i;
+	u32 ts_num, vm_num, pd_num, ch_num, val, index, i;
 	struct device *dev = &pdev->dev;
 	u32 *temp_config, *in_config;
 	struct device *hwmon_dev;
@@ -544,9 +547,11 @@ static int mr75203_probe(struct platform_device *pdev)
 	ts_num = (val & TS_NUM_MSK) >> TS_NUM_SFT;
 	pd_num = (val & PD_NUM_MSK) >> PD_NUM_SFT;
 	vm_num = (val & VM_NUM_MSK) >> VM_NUM_SFT;
+	ch_num = (val & CH_NUM_MSK) >> CH_NUM_SFT;
 	pvt->t_num = ts_num;
 	pvt->p_num = pd_num;
 	pvt->v_num = vm_num;
+	pvt->c_num = ch_num;
 	val = 0;
 	if (ts_num)
 		val++;
@@ -583,6 +588,8 @@ static int mr75203_probe(struct platform_device *pdev)
 	}
 
 	if (vm_num) {
+		u32 total_ch = ch_num * vm_num;
+
 		ret = pvt_get_regmap(pdev, "vm", pvt);
 		if (ret)
 			return ret;
@@ -611,13 +618,13 @@ static int mr75203_probe(struct platform_device *pdev)
 			pvt->v_num = i;
 		}
 
-		in_config = devm_kcalloc(dev, vm_num + 1,
+		in_config = devm_kcalloc(dev, total_ch + 1,
 					 sizeof(*in_config), GFP_KERNEL);
 		if (!in_config)
 			return -ENOMEM;
 
-		memset32(in_config, HWMON_I_INPUT, vm_num);
-		in_config[vm_num] = 0;
+		memset32(in_config, HWMON_I_INPUT, total_ch);
+		in_config[total_ch] = 0;
 		pvt_in.config = in_config;
 
 		pvt_info[index++] = &pvt_in;
Signature · 1 line
2.37.1
#33 lore raw

Re: [PATCH v3 06/19] hwmon: (mr75203) fix multi-channel voltage reading

Quoted 5 lines · Eliav Farber

Fix voltage allocation and reading to support all channels in all VMs. Prior to this change allocation and reading were done only for the first channel in each VM.
This change counts the total number of channels for allocation, and takes into account the channel offset when reading the sample data register.

...

drivers/hwmon/mr75203.c static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val) · line 184

Quoted 5 lines
drivers/hwmon/mr75203.c +1 -1
 	struct pvt_device *pvt = dev_get_drvdata(dev);
 	struct regmap *v_map = pvt->v_map;
 	u32 n, stat;
-	u8 vm_idx;
+	u8 vm_idx, ch_idx;

Can you keep it sorted by line length?

Quoted 1 line

int ret;

...

drivers/hwmon/mr75203.c static int pvt_reset_control_deassert(struct device *dev, struct pvt_device *pvt · line 506

Quoted 3 lines
drivers/hwmon/mr75203.c +1 -1
 	const struct hwmon_channel_info **pvt_info;
-	u32 ts_num, vm_num, pd_num, val, index, i;
+	u32 ts_num, vm_num, pd_num, ch_num, val, index, i;

Ditto.

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

Re: [PATCH v3 06/19] hwmon: (mr75203) fix multi-channel voltage reading

Quoted 9 lines · Andy Shevchenko
Quoted 5 lines · Eliav Farber

Fix voltage allocation and reading to support all channels in all VMs. Prior to this change allocation and reading were done only for the first channel in each VM.
This change counts the total number of channels for allocation, and takes into account the channel offset when reading the sample data register.

...

Quoted 7 lines
Quoted 5 lines

      struct pvt_device *pvt = dev_get_drvdata(dev); struct regmap *v_map = pvt->v_map;
      u32 n, stat;
-     u8 vm_idx;
+     u8 vm_idx, ch_idx;

Can you keep it sorted by line length?

Fixed.

Quoted 9 lines
Quoted 1 line

      int ret;

...

Quoted 3 lines

      const struct hwmon_channel_info **pvt_info; -     u32 ts_num, vm_num, pd_num, val, index, i; +     u32 ts_num, vm_num, pd_num, ch_num, val, index, i;

Ditto.

Fixed.

--
Thanks, Eliav

#35 lore raw

PATCH v3 7/19 hwmon: (mr75203) enable polling for all VM channels

In reply to Eliav Farber

Configure ip-polling register to enable polling for all voltage monitor channels.
This enables reading the voltage values for all inputs other than just input 0.

---
V3 -> V2:
- Move configuration of ip-polling register from previous patch to a separate commit.

 drivers/hwmon/mr75203.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
drivers/hwmon/mr75203.c +14 -0
index 639f5a300170..f561806c400d 100644
@@ -388,6 +388,20 @@ static int pvt_init(struct pvt_device *pvt)
 		if (ret)
 			return ret;
 
+		val = GENMASK(pvt->c_num - 1, 0) | VM_CH_INIT |
+		      IP_POLL << SDIF_ADDR_SFT |
+		      SDIF_WRN_W | SDIF_PROG;
+		ret = regmap_write(v_map, SDIF_W, val);
+		if (ret < 0)
+			return ret;
+
+		ret = regmap_read_poll_timeout(v_map, SDIF_STAT,
+					       val, !(val & SDIF_BUSY),
+					       PVT_POLL_DELAY_US,
+					       PVT_POLL_TIMEOUT_US);
+		if (ret)
+			return ret;
+
 		val = CFG1_VOL_MEAS_MODE | CFG1_PARALLEL_OUT |
 		      CFG1_14_BIT | IP_CFG << SDIF_ADDR_SFT |
 		      SDIF_WRN_W | SDIF_PROG;
Signature · 1 line
2.37.1
#36 lore raw

Re: [PATCH v3 07/19] hwmon: (mr75203) enable polling for all VM channels

Quoted 4 lines · Eliav Farber

Configure ip-polling register to enable polling for all voltage monitor channels.
This enables reading the voltage values for all inputs other than just input 0.

...

drivers/hwmon/mr75203.c static int pvt_init(struct pvt_device *pvt) · line 391

Quoted 1 line
drivers/hwmon/mr75203.c +1 -0
+		val = GENMASK(pvt->c_num - 1, 0) | VM_CH_INIT |

I believe in this case (BIT(pvt->c_num) - 1) is better, but not sure if c_num can be 32.

Quoted 2 lines
drivers/hwmon/mr75203.c +2 -0
+		      IP_POLL << SDIF_ADDR_SFT |
+		      SDIF_WRN_W | SDIF_PROG;
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#37 lore raw

Re: [PATCH v3 07/19] hwmon: (mr75203) enable polling for all VM channels

Quoted 6 lines · Andy Shevchenko
Quoted 4 lines · Eliav Farber

Configure ip-polling register to enable polling for all voltage monitor channels.
This enables reading the voltage values for all inputs other than just input 0.

Quoted 6 lines

...

Quoted 1 line

+             val = GENMASK(pvt->c_num - 1, 0) | VM_CH_INIT |

I believe in this case (BIT(pvt->c_num) - 1) is better, but not sure if c_num can be 32.

c_num can't be 32.
I fix it in v4.

--
Thanks, Eliav

#38 lore raw

PATCH v3 8/19 dt-bindings: hwmon: (mr75203) add "moortec,vm-active-channels" property

In reply to Eliav Farber

Add optional "moortec,vm-active-channels" property to define the number of active channels per VM.

This shall be useful to avoid exposing sysfs for reading inputs that are not connected to any voltage source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Add explanation why this change is needed.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++
 1 file changed, 11 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +11 -0
index 2ec4b9da4b92..69cc6caceb2c 100644
@@ -44,6 +44,16 @@ properties:
   "#thermal-sensor-cells":
     const: 1
 
+  moortec,vm-active-channels:
+    description:
+      moortec,vm-active-channels defines the number of channels per VM that are
+      actually used and are connected to some source.
+      Number of values in the array should be the number of VMs.
+      A value of 0 means that the entire VM sensor is nou used.
+    maximum: 16
+    default: 16
+    $ref: /schemas/types.yaml#definitions/uint8-array
+
 required:
   - compatible
   - reg
@@ -65,5 +75,6 @@ examples:
         intel,vm-map = [03 01 04 ff ff];
         clocks = <&osc0>;
         resets = <&rcu0 0x40 7>;
+        moortec,vm-active-channels = <0x10 0x05>;
         #thermal-sensor-cells = <1>;
     };
Signature · 1 line
2.37.1
maintainer
#39 lore raw

Re: [PATCH v3 08/19] dt-bindings: hwmon: (mr75203) add "moortec,vm-active-channels" property

Quoted 6 lines · Eliav Farber

Add optional "moortec,vm-active-channels" property to define the number of active channels per VM.

This shall be useful to avoid exposing sysfs for reading inputs that are not connected to any voltage source.

Quoted 9 lines

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Add explanation why this change is needed.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++
 1 file changed, 11 insertions(+)

My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check' on your patch (DT_CHECKER_FLAGS is new in v5.13):

yamllint warnings/errors:

dtschema/dtc warnings/errors:
./Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml: error checking schema file
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml: ignoring, error in schema: properties: moortec,vm-active-channels
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/hwmon/moortec,mr75203.example.dtb: pvt@e0680000: 'moortec,vm-active-channels' does not match any of the regexes: '^#.*', '^(at25|bm|devbus|dmacap|dsa|exynos|fsi[ab]|gpio-fan|gpio-key|gpio|gpmc|hdmi|i2c-gpio),.*', '^(keypad|m25p|max8952|max8997|max8998|mpmc),.*', '^(pinctrl-single|#pinctrl-single|PowerPC),.*', '^(pl022|pxa-mmc|rcar_sound|rotary-encoder|s5m8767|sdhci),.*', '^(simple-audio-card|st-plgpio|st-spics|ts),.*', '^100ask,.*', '^70mai,.*', '^8dev,.*', '^GEFanuc,.*', '^ORCL,.*', '^SUNW,.*', '^[a-zA-Z0-9#_][a-zA-Z0-9+\\-._@]{0,63}$', '^[a-zA-Z0-9+\\-._]*@[0-9a-zA-Z,]*$', '^abb,.*', '^abilis,.*', '^abracon,.*', '^abt,.*', '^acer,.*', '^acme,.*', '^actions,.*', '^active-semi,.*', '^ad,.*', '^adafruit,.*', '^adapteva,.*', '^adaptrum,.*', '^adh,.*', '^adi,.*', '^advantech,.*', '^aeroflexgaisler,.*', '^aesop,.*', '^airoha,.*', '^al,.*', '^alcatel,.*', '^allegro,.*', '^allo,.*', '^allwinner,.*', '^alphascale,.*', '^al
ps,.*', '^alt,.*', '^altr,.*', '^amarula,.*', '^amazon,.*', '^amcc,.*', '^amd,.*', '^amediatech,.*', '^amlogic,.*', '^ampere,.*', '^ampire,.*', '^ams,.*', '^amstaos,.*', '^analogix,.*', '^andestech,.*', '^anvo,.*', '^apm,.*', '^apple,.*', '^aptina,.*', '^arasan,.*', '^archermind,.*', '^arctic,.*', '^arcx,.*', '^aries,.*', '^arm,.*', '^armadeus,.*', '^arrow,.*', '^artesyn,.*', '^asahi-kasei,.*', '^asc,.*', '^asix,.*', '^aspeed,.*', '^asrock,.*', '^asus,.*', '^atheros,.*', '^atlas,.*', '^atmel,.*', '^auo,.*', '^auvidea,.*', '^avago,.*', '^avia,.*', '^avic,.*', '^avnet,.*', '^awinic,.*', '^axentia,.*', '^axis,.*', '^azoteq,.*', '^azw,.*', '^baikal,.*', '^bananapi,.*', '^beacon,.*', '^beagle,.*', '^bhf,.*', '^bitmain,.*', '^blutek,.*', '^boe,.*', '^bosch,.*', '^boundary,.*', '^brcm,.*', '^broadmobi,.*', '^bsh,.*', '^bticino,.*', '^buffalo,.*', '^bur,.*', '^bytedance,.*', '^calamp,.*', '^calaosystems,.*', '^calxeda,.*', '^canaan,.*', '^caninos,.*', '^capella,.*', '^cascoda,.*', '^catalys t,.*', '^cavium,.*', '^cdns,.*', '^cdtech,.*', '^cellwise,.*', '^ceva,.*', '^checkpoint,.*', '^chefree,.*', '^chipidea,.*', '^chipone,.*', '^chipspark,.*', '^chrontel,.*', '^chrp,.*', '^chunghwa,.*', '^chuwi,.*', '^ciaa,.*', '^cirrus,.*', '^cisco,.*', '^cloudengines,.*', '^cnm,.*', '^cnxt,.*', '^colorfly,.*', '^compulab,.*', '^congatec,.*', '^coreriver,.*', '^corpro,.*', '^cortina,.*', '^cosmic,.*', '^crane,.*', '^creative,.*', '^crystalfontz,.*', '^csky,.*', '^csq,.*', '^ctera,.*', '^ctu,.*', '^cubietech,.*', '^cui,.*', '^cypress,.*', '^cyx,.*', '^cznic,.*', '^dallas,.*', '^dataimage,.*', '^davicom,.*', '^dell,.*', '^delta,.*', '^densitron,.*', '^denx,.*', '^devantech,.*', '^dfi,.*', '^dh,.*', '^difrnce,.*', '^digi,.*', '^digilent,.*', '^dioo,.*', '^dlc,.*', '^dlg,.*', '^dlink,.*', '^dmo,.*', '^domintech,.*', '^dongwoon,.*', '^dptechnics,.*', '^dragino,.*', '^ds,.*', '^dserve,.*', '^dynaimage,.*', '^ea,.*', '^ebang,.*', '^ebbg,.*', '^ebs-systart,.*', '^ebv,.*', '^eckelmann,.*', '^e dimax,.*', '^edt,.*', '^eeti,.*', '^einfochips,.*', '^eink,.*', '^elan,.*', '^element14,.*', '^elgin,.*', '^elida,.*', '^elimo,.*', '^elpida,.*', '^embest,.*', '^emlid,.*', '^emmicro,.*', '^empire-electronix,.*', '^emtrion,.*', '^enclustra,.*', '^endless,.*', '^ene,.*', '^energymicro,.*', '^engicam,.*', '^engleder,.*', '^epcos,.*', '^epfl,.*', '^epson,.*', '^esp,.*', '^est,.*', '^ettus,.*', '^eukrea,.*', '^everest,.*', '^everspin,.*', '^evervision,.*', '^exar,.*', '^excito,.*', '^exegin,.*', '^ezchip,.*', '^facebook,.*', '^fairphone,.*', '^faraday,.*', '^fastrax,.*', '^fcs,.*', '^feixin,.*', '^feiyang,.*', '^fii,.*', '^firefly,.*', '^focaltech,.*', '^forlinx,.*', '^frida,.*', '^friendlyarm,.*', '^fsl,.*', '^fujitsu,.*', '^fxtec,.*', '^gardena,.*', '^gateworks,.*', '^gcw,.*', '^ge,.*', '^geekbuying,.*', '^gef,.*', '^gemei,.*', '^geniatech,.*', '^giantec,.*', '^giantplus,.*', '^globalscale,.*', '^globaltop,.*', '^gmt,.*', '^goodix,.*', '^google,.*', '^grinn,.*', '^grmn,.*', '^gumstix, .*', '^gw,.*', '^hannstar,.*', '^haochuangyi,.*', '^haoyu,.*', '^hardkernel,.*', '^hechuang,.*', '^hideep,.*', '^himax,.*', '^hirschmann,.*', '^hisi,.*', '^hisilicon,.*', '^hit,.*', '^hitex,.*', '^holt,.*', '^holtek,.*', '^honestar,.*', '^honeywell,.*', '^hoperun,.*', '^hp,.*', '^hpe,.*', '^hsg,.*', '^huawei,.*', '^hugsun,.*', '^hwacom,.*', '^hxt,.*', '^hycon,.*', '^hydis,.*', '^hynix,.*', '^hyundai,.*', '^i2se,.*', '^ibm,.*', '^icplus,.*', '^idt,.*', '^ifi,.*', '^ilitek,.*', '^imagis,.*', '^img,.*', '^imi,.*', '^incircuit,.*', '^inet-tek,.*', '^infineon,.*', '^inforce,.*', '^ingenic,.*', '^ingrasys,.*', '^injoinic,.*', '^innolux,.*', '^inside-secure,.*', '^insignal,.*', '^inspur,.*', '^intel,.*', '^intercontrol,.*', '^invensense,.*', '^inventec,.*', '^inversepath,.*', '^iom,.*', '^isee,.*', '^isil,.*', '^issi,.*', '^ite,.*', '^itead,.*', '^itian,.*', '^ivo,.*', '^iwave,.*', '^jdi,.*', '^jedec,.*', '^jesurun,.*', '^jethome,.*', '^jianda,.*', '^joz,.*', '^kam,.*', '^karo,.*', '^keith koep,.*', '^keymile,.*', '^khadas,.*', '^kiebackpeter,.*', '^kinetic,.*', '^kingdisplay,.*', '^kingnovel,.*', '^kionix,.*', '^kobo,.*', '^kobol,.*', '^koe,.*', '^kontron,.*', '^kosagi,.*', '^kvg,.*', '^kyo,.*', '^lacie,.*', '^laird,.*', '^lamobo,.*', '^lantiq,.*', '^lattice,.*', '^leadtek,.*', '^leez,.*', '^lego,.*', '^lemaker,.*', '^lenovo,.*', '^lg,.*', '^lgphilips,.*', '^libretech,.*', '^licheepi,.*', '^linaro,.*', '^linksprite,.*', '^linksys,.*', '^linutronix,.*', '^linux,.*', '^linx,.*', '^liteon,.*', '^litex,.*', '^lltc,.*', '^logicpd,.*', '^logictechno,.*', '^longcheer,.*', '^lontium,.*', '^loongson,.*', '^lsi,.*', '^lwn,.*', '^lxa,.*', '^m5stack,.*', '^macnica,.*', '^mantix,.*', '^mapleboard,.*', '^marvell,.*', '^maxbotix,.*', '^maxim,.*', '^mbvl,.*', '^mcube,.*', '^meas,.*', '^mecer,.*', '^mediatek,.*', '^megachips,.*', '^mele,.*', '^melexis,.*', '^melfas,.*', '^mellanox,.*', '^memsic,.*', '^menlo,.*', '^mentor,.*', '^meraki,.*', '^merrii,.*', '^micrel,.*', '^microchip,.*',
'^microcrystal,.*', '^micron,.*', '^microsoft,.*', '^microsys,.*', '^mikroe,.*', '^mikrotik,.*', '^miniand,.*', '^minix,.*', '^miramems,.*', '^mitsubishi,.*', '^mixel,.*', '^miyoo,.*', '^mntre,.*', '^modtronix,.*', '^mosaixtech,.*', '^motorola,.*', '^moxa,.*', '^mpl,.*', '^mps,.*', '^mqmaker,.*', '^mrvl,.*', '^mscc,.*', '^msi,.*', '^mstar,.*', '^mti,.*', '^multi-inno,.*', '^mundoreader,.*', '^murata,.*', '^mxic,.*', '^mxicy,.*', '^myir,.*', '^national,.*', '^nec,.*', '^neonode,.*', '^netgear,.*', '^netlogic,.*', '^netron-dy,.*', '^netronix,.*', '^netxeon,.*', '^neweast,.*', '^newhaven,.*', '^nexbox,.*', '^nextthing,.*', '^ni,.*', '^nintendo,.*', '^nlt,.*', '^nokia,.*', '^nordic,.*', '^novtech,.*', '^nutsboard,.*', '^nuvoton,.*', '^nvd,.*', '^nvidia,.*', '^nxp,.*', '^oceanic,.*', '^ocs,.*', '^oct,.*', '^okaya,.*', '^oki,.*', '^olimex,.*', '^olpc,.*', '^oneplus,.*', '^onion,.*', '^onnn,.*', '^ontat,.*', '^opalkelly,.*', '^opencores,.*', '^openembed,.*', '^openrisc,.*', '^option,.*',
'^oranth,.*', '^orisetech,.*', '^ortustech,.*', '^osddisplays,.*', '^osmc,.*', '^ouya,.*', '^overkiz,.*', '^ovti,.*', '^oxsemi,.*', '^ozzmaker,.*', '^panasonic,.*', '^parade,.*', '^parallax,.*', '^pda,.*', '^pericom,.*', '^pervasive,.*', '^phicomm,.*', '^phytec,.*', '^picochip,.*', '^pine64,.*', '^pineriver,.*', '^pixcir,.*', '^plantower,.*', '^plathome,.*', '^plda,.*', '^plx,.*', '^ply,.*', '^pni,.*', '^pocketbook,.*', '^polaroid,.*', '^portwell,.*', '^poslab,.*', '^pov,.*', '^powertip,.*', '^powervr,.*', '^primux,.*', '^probox2,.*', '^prt,.*', '^pulsedlight,.*', '^purism,.*', '^qca,.*', '^qcom,.*', '^qemu,.*', '^qi,.*', '^qiaodian,.*', '^qihua,.*', '^qishenglong,.*', '^qnap,.*', '^quanta,.*', '^radxa,.*', '^raidsonic,.*', '^ralink,.*', '^ramtron,.*', '^raspberrypi,.*', '^raydium,.*', '^rda,.*', '^realtek,.*', '^remarkable,.*', '^renesas,.*', '^rervision,.*', '^revotics,.*', '^rex,.*', '^richtek,.*', '^ricoh,.*', '^rikomagic,.*', '^riot,.*', '^riscv,.*', '^rockchip,.*', '^rocktech, .*', '^rohm,.*', '^ronbo,.*', '^roofull,.*', '^roseapplepi,.*', '^samsung,.*', '^samtec,.*', '^sancloud,.*', '^sandisk,.*', '^satoz,.*', '^sbs,.*', '^schindler,.*', '^seagate,.*', '^seeed,.*', '^seirobotics,.*', '^semtech,.*', '^senseair,.*', '^sensirion,.*', '^sensortek,.*', '^sercomm,.*', '^sff,.*', '^sgd,.*', '^sgmicro,.*', '^sgx,.*', '^sharp,.*', '^shift,.*', '^shimafuji,.*', '^shiratech,.*', '^si-en,.*', '^si-linux,.*', '^siemens,.*', '^sifive,.*', '^sigma,.*', '^sii,.*', '^sil,.*', '^silabs,.*', '^silan,.*', '^silead,.*', '^silergy,.*', '^silex-insight,.*', '^siliconfile,.*', '^siliconmitus,.*', '^silvaco,.*', '^simtek,.*', '^sinlinx,.*', '^sinovoip,.*', '^sinowealth,.*', '^sipeed,.*', '^sirf,.*', '^sis,.*', '^sitronix,.*', '^skov,.*', '^skyworks,.*', '^smartlabs,.*', '^smsc,.*', '^snps,.*', '^sochip,.*', '^socionext,.*', '^solidrun,.*', '^solomon,.*', '^sony,.*', '^spansion,.*', '^sparkfun,.*', '^spinalhdl,.*', '^sprd,.*', '^ssi,.*', '^sst,.*', '^sstar,.*', '^st,.*', '^st-eri csson,.*', '^starfive,.*', '^starry,.*', '^startek,.*', '^ste,.*', '^stericsson,.*', '^storlink,.*', '^storm,.*', '^storopack,.*', '^summit,.*', '^sunchip,.*', '^sundance,.*', '^sunplus,.*', '^supermicro,.*', '^swir,.*', '^syna,.*', '^synology,.*', '^synopsys,.*', '^tbs,.*', '^tbs-biometrics,.*', '^tcg,.*', '^tcl,.*', '^tcs,.*', '^tdo,.*', '^team-source-display,.*', '^technexion,.*', '^technologic,.*', '^techstar,.*', '^teltonika,.*', '^tempo,.*', '^terasic,.*', '^tesla,.*', '^tfc,.*', '^thead,.*', '^thine,.*', '^thingyjp,.*', '^thundercomm,.*', '^ti,.*', '^tianma,.*', '^tlm,.*', '^tmt,.*', '^topeet,.*', '^topic,.*', '^toppoly,.*', '^topwise,.*', '^toradex,.*', '^toshiba,.*', '^toumaz,.*', '^tpk,.*', '^tplink,.*', '^tpo,.*', '^tq,.*', '^traverse,.*', '^tronfy,.*', '^tronsmart,.*', '^truly,.*', '^tsd,.*', '^tyan,.*', '^u-blox,.*', '^u-boot,.*', '^ubnt,.*', '^ucrobotics,.*', '^udoo,.*', '^ugoos,.*', '^uniwest,.*', '^upisemi,.*', '^urt,.*', '^usi,.*', '^utoo,.*', '^v3,.*', '^vaisala,.* ', '^vamrs,.*', '^variscite,.*', '^vdl,.*', '^vertexcom,.*', '^via,.*', '^vicor,.*', '^videostrong,.*', '^virtio,.*', '^virtual,.*', '^vishay,.*', '^visionox,.*', '^vitesse,.*', '^vivante,.*', '^vivax,.*', '^vocore,.*', '^voipac,.*', '^vot,.*', '^vxt,.*', '^wanchanglong,.*', '^wand,.*', '^waveshare,.*', '^wd,.*', '^we,.*', '^welltech,.*', '^wetek,.*', '^wexler,.*', '^whwave,.*', '^wi2wi,.*', '^wiligear,.*', '^willsemi,.*', '^winbond,.*', '^wingtech,.*', '^winlink,.*', '^winstar,.*', '^wirelesstag,.*', '^wits,.*', '^wlf,.*', '^wm,.*', '^wobo,.*', '^x-powers,.*', '^xen,.*', '^xes,.*', '^xiaomi,.*', '^xillybus,.*', '^xingbangda,.*', '^xinpeng,.*', '^xiphera,.*', '^xlnx,.*', '^xnano,.*', '^xunlong,.*', '^xylon,.*', '^yadro,.*', '^yamaha,.*', '^yes-optoelectronics,.*', '^yic,.*', '^ylm,.*', '^yna,.*', '^yones-toptech,.*', '^ys,.*', '^ysoft,.*', '^zarlink,.*', '^zealz,.*', '^zeitec,.*', '^zidoo,.*', '^zii,.*', '^zinitix,.*', '^zkmagic,.*', '^zte,.*', '^zyxel,.*', 'pinctrl-[0-9]+'
From schema: /builds/robherring/linux-dt-review/Documentation/devicetree/bindings/vendor-prefixes.yaml
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/hwmon/moortec,mr75203.example.dtb: pvt@e0680000: intel,vm-map: b'\x03\x01\x04\xff\xff' is not of type 'object', 'array', 'boolean', 'null'
From schema: /usr/local/lib/python3.10/dist-packages/dtschema/schemas/dt-core.yaml
Documentation/devicetree/bindings/hwmon/moortec,mr75203.example.dtb:0:0: /example-0/pvt@e0680000: failed to match any schema with compatible: ['moortec,mr75203']

doc reference errors (make refcheckdocs):

See https://patchwork.ozlabs.org/patch/

This check can fail if there are any dependencies. The base for a patch series is generally the most recent rc1.

If you already ran 'make dt_binding_check' and didn't see the above error(s), then make sure 'yamllint' is installed and dt-schema is up to date:

pip3 install dtschema --upgrade

Please check and re-submit.

A message that is not in the archive (sent off-list or not yet received): [email protected]

maintainer
#41 lore raw

Re: [PATCH v3 08/19] dt-bindings: hwmon: (mr75203) add "moortec, vm-active-channels" property

Quoted 21 lines · Eliav

On 8/31/2022 2:39 PM, Rob Herring wrote:

On Tue, 30 Aug 2022 19:22:01 +0000, Eliav Farber wrote:

Add optional "moortec,vm-active-channels" property to define the number of active channels per VM.

This shall be useful to avoid exposing sysfs for reading inputs that are not connected to any voltage source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Add explanation why this change is needed.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++
 1 file changed, 11 insertions(+)
Quoted 5 lines

My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check' on your patch (DT_CHECKER_FLAGS is new in v5.13):

I used dt_binding_check on my changes (I ported it to my kernel). The error is related to "intel-vm-map" which I did not add.

The error is the vendor prefix is not defined in vendor-prefixes.yaml.

Quoted 5 lines

I don't mind fixing it if you wish.
It requires changing:
intel,vm-map = [03 01 04 ff ff];
to:
intel,vm-map = /bits/8 <0x03 0x01 0x04 0xff 0xff>;

That is not the issue. The issue is the type is unknown because your schema fails and we can't get the type from it. Once your schema passes, this should go away.

Rob

#42 lore raw

Re: [PATCH v3 08/19] dt-bindings: hwmon: (mr75203) add "moortec, vm-active-channels" property

Quoted 22 lines · Rob Herring
Quoted 21 lines · Eliav

On 8/31/2022 2:39 PM, Rob Herring wrote:

On Tue, 30 Aug 2022 19:22:01 +0000, Eliav Farber wrote:

Add optional "moortec,vm-active-channels" property to define the number of active channels per VM.

This shall be useful to avoid exposing sysfs for reading inputs that are not connected to any voltage source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Add explanation why this change is needed.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++

 1 file changed, 11 insertions(+)
Quoted 7 lines
Quoted 5 lines

My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check' on your patch (DT_CHECKER_FLAGS is new in v5.13):

I used dt_binding_check on my changes (I ported it to my kernel). The error is related to "intel-vm-map" which I did not add.

The error is the vendor prefix is not defined in vendor-prefixes.yaml.

I fixed the vendor prefix error (will be part of v4).

Quoted 9 lines
Quoted 5 lines

I don't mind fixing it if you wish.
It requires changing:
  intel,vm-map = [03 01 04 ff ff];
to:
  intel,vm-map = /bits/8 <0x03 0x01 0x04 0xff 0xff>;

That is not the issue. The issue is the type is unknown because your schema fails and we can't get the type from it. Once your schema passes, this should go away.

Even after fixing the vendor prefix error I still see this:
moortec,mr75203.yaml: ignoring, error in schema: properties: intel,vm-map
moortec,mr75203.example.dtb: pvt@e0680000: intel,vm-map: b'\x03\x01\x04\xff\xff' is not of type 'object', 'array', 'boolean', 'null'

--
Regards, Eliav

maintainer
#43 lore raw

Re: [PATCH v3 08/19] dt-bindings: hwmon: (mr75203) add "moortec, vm-active-channels" property

Quoted 40 lines · Eliav
Quoted 22 lines · Rob Herring
Quoted 21 lines · Eliav

On 8/31/2022 2:39 PM, Rob Herring wrote:

On Tue, 30 Aug 2022 19:22:01 +0000, Eliav Farber wrote:

Add optional "moortec,vm-active-channels" property to define the number of active channels per VM.

This shall be useful to avoid exposing sysfs for reading inputs that are not connected to any voltage source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Add explanation why this change is needed.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++
 1 file changed, 11 insertions(+)
Quoted 7 lines
Quoted 5 lines

My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check' on your patch (DT_CHECKER_FLAGS is new in v5.13):

I used dt_binding_check on my changes (I ported it to my kernel). The error is related to "intel-vm-map" which I did not add.

The error is the vendor prefix is not defined in vendor-prefixes.yaml.

I fixed the vendor prefix error (will be part of v4).

Quoted 6 lines
Quoted 5 lines

I don't mind fixing it if you wish.
It requires changing:
intel,vm-map = [03 01 04 ff ff];
to:
intel,vm-map = /bits/8 <0x03 0x01 0x04 0xff 0xff>;

Quoted 5 lines
Quoted 3 lines

That is not the issue. The issue is the type is unknown because your schema fails and we can't get the type from it. Once your schema passes, this should go away.

Even after fixing the vendor prefix error I still see this:
moortec,mr75203.yaml: ignoring, error in schema: properties: intel,vm-map

You still have an error in the schema. You should see a more specific reason before this message.

Quoted 5 lines

moortec,mr75203.example.dtb: pvt@e0680000: intel,vm-map:
b'\x03\x01\x04\xff\xff' is not of type 'object', 'array', 'boolean', 'null'

--
Regards, Eliav

#44 lore raw

Re: [PATCH v3 08/19] dt-bindings: hwmon: (mr75203) add "moortec, vm-active-channels" property

Quoted 45 lines · Rob Herring
Quoted 44 lines · Eliav
Quoted 1 line · Rob Herring

On Wed, Aug 31, 2022 at 6:53 AM Farber, Eliav <[email protected]>

Quoted 6 lines · wrote:
Quoted 6 lines · Rob Herring

On Tue, 30 Aug 2022 19:22:01 +0000, Eliav Farber wrote:

Add optional "moortec,vm-active-channels" property to define the

number

Quoted 3 lines
Quoted 3 lines

of active channels per VM.

This shall be useful to avoid exposing sysfs for reading inputs

that are

Quoted 8 lines
Quoted 8 lines

not connected to any voltage source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Add explanation why this change is needed.

Quoted 5 lines
Quoted 5 lines

  .../devicetree/bindings/hwmon/moortec,mr75203.yaml | 11 +++++++++++

  1 file changed, 11 insertions(+)

My bot found errors running 'make DT_CHECKER_FLAGS=-m

dt_binding_check'

Quoted 6 lines
Quoted 4 lines

on your patch (DT_CHECKER_FLAGS is new in v5.13):

I used dt_binding_check on my changes (I ported it to my kernel). The error is related to "intel-vm-map" which I did not add.

The error is the vendor prefix is not defined in vendor-prefixes.yaml.

I fixed the vendor prefix error (will be part of v4).

Quoted 6 lines
Quoted 5 lines

I don't mind fixing it if you wish.
It requires changing:
   intel,vm-map = [03 01 04 ff ff];
to:
   intel,vm-map = /bits/8 <0x03 0x01 0x04 0xff 0xff>;

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

That is not the issue. The issue is the type is unknown because your schema fails and we can't get the type from it. Once your schema passes, this should go away.

Even after fixing the vendor prefix error I still see this:
moortec,mr75203.yaml: ignoring, error in schema: properties: intel,vm-map

You still have an error in the schema. You should see a more specific reason before this message.

Thanks, I found the problem.
I'm using an old version, and I'm missing this commit you did:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml?h=v6.0-rc3&id=d69c6ddd019f31081cc0232fa8ad8ea1cabdf22c

#45 lore raw

PATCH v3 9/19 hwmon: (mr75203) add VM active channel support

In reply to Eliav Farber

Add active channel support per voltage monitor.
The number of active channels is read from the device-tree. When absent in device-tree, all channels are assumed to be used.

This shall be useful to expose sysfs only for inputs that are connected to a voltage source.

Setting number of active channels to 0, means that entire VM sensor is not used.

---
V3 -> V2:
- Refactor the code changes (move code to a new function and group parameters in dedicated structure).

V2 -> V1:
- Fix compilation error for patch 08/16:
"warning: ISO C90 forbids variable length array"

 drivers/hwmon/mr75203.c | 107 ++++++++++++++++++++++++++++++++--------
 1 file changed, 86 insertions(+), 21 deletions(-)
drivers/hwmon/mr75203.c +86 -21
index f561806c400d..6925e8490587 100644
@@ -29,6 +29,8 @@
 #define CH_NUM_MSK	GENMASK(31, 24)
 #define CH_NUM_SFT	24
 
+#define VM_NUM_MAX	(VM_NUM_MSK >> VM_NUM_SFT)
+
 /* Macro Common Register */
 #define CLK_SYNTH		0x00
 #define CLK_SYNTH_LO_SFT	0
@@ -106,6 +108,16 @@
 #define PVT_N_CONST		90
 #define PVT_R_CONST		245805
 
+struct voltage_device {
+	u32 vm_map;	/* Map channel number to VM index */
+	u32 ch_map;	/* Map channel number to channel index */
+};
+
+struct voltage_channels {
+	u32 total;	/* Total number of channels in all VMs */
+	u8 max;		/* Maximum number of channels among all VMs */
+};
+
 struct pvt_device {
 	struct regmap		*c_map;
 	struct regmap		*t_map;
@@ -113,12 +125,12 @@ struct pvt_device {
 	struct regmap		*v_map;
 	struct clk		*clk;
 	struct reset_control	*rst;
+	struct voltage_device	*vd;
+	struct voltage_channels	vm_channels;
 	u32			t_num;
 	u32			p_num;
 	u32			v_num;
-	u32			c_num;
 	u32			ip_freq;
-	u8			*vm_idx;
 };
 
 static umode_t pvt_is_visible(const void *data, enum hwmon_sensor_types type,
@@ -184,11 +196,11 @@ static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val)
 	u8 vm_idx, ch_idx;
 	int ret;
 
-	if (channel >= pvt->v_num * pvt->c_num)
+	if (channel >= pvt->vm_channels.total)
 		return -EINVAL;
 
-	vm_idx = pvt->vm_idx[channel / pvt->c_num];
-	ch_idx = channel % pvt->c_num;
+	vm_idx = pvt->vd[channel].vm_map;
+	ch_idx = pvt->vd[channel].ch_map;
 
 	switch (attr) {
 	case hwmon_in_input:
@@ -388,7 +400,7 @@ static int pvt_init(struct pvt_device *pvt)
 		if (ret)
 			return ret;
 
-		val = GENMASK(pvt->c_num - 1, 0) | VM_CH_INIT |
+		val = GENMASK(pvt->vm_channels.max - 1, 0) | VM_CH_INIT |
 		      IP_POLL << SDIF_ADDR_SFT |
 		      SDIF_WRN_W | SDIF_PROG;
 		ret = regmap_write(v_map, SDIF_W, val);
@@ -514,6 +526,62 @@ static int pvt_reset_control_deassert(struct device *dev, struct pvt_device *pvt
 	return devm_add_action_or_reset(dev, pvt_reset_control_assert, pvt);
 }
 
+static int pvt_get_active_channel(struct device *dev, struct pvt_device *pvt,
+				  u32 vm_num, u32 ch_num, u8 *vm_idx)
+{
+	u8 vm_active_ch[VM_NUM_MAX];
+	int ret, i, j, k;
+
+	ret = device_property_read_u8_array(dev, "moortec,vm-active-channels",
+					    vm_active_ch, vm_num);
+	if (ret) {
+		/*
+		 * Incase vm-active-channels property is not defined,
+		 * we assume each VM sensor has all of its channels
+		 * active.
+		 */
+		for (i = 0; i < vm_num; i++)
+			vm_active_ch[i] = ch_num;
+
+		pvt->vm_channels.max = ch_num;
+		pvt->vm_channels.total = ch_num * vm_num;
+	} else {
+		for (i = 0; i < vm_num; i++) {
+			if (vm_active_ch[i] > ch_num) {
+				dev_err(dev, "invalid active channels: %u\n",
+					vm_active_ch[i]);
+				return -EINVAL;
+			}
+
+			pvt->vm_channels.total += vm_active_ch[i];
+
+			if (vm_active_ch[i] > pvt->vm_channels.max)
+				pvt->vm_channels.max = vm_active_ch[i];
+		}
+	}
+
+	/*
+	 * Map between the channel-number to VM-index and channel-index.
+	 * Example - 3 VMs, vm_active_ch = [05 02 04]:
+	 * vm_map = [0 0 0 0 0 1 1 2 2 2 2]
+	 * ch_map = [0 1 2 3 4 0 1 0 1 2 3]
+	 */
+	pvt->vd = devm_kcalloc(dev, pvt->vm_channels.total, sizeof(*pvt->vd),
+			       GFP_KERNEL);
+	if (!pvt->vd)
+		return -ENOMEM;
+
+	k = 0;
+	for (i = 0; i < vm_num; i++)
+		for (j = 0; j < vm_active_ch[i]; j++) {
+			pvt->vd[k].vm_map = vm_idx[i];
+			pvt->vd[k].ch_map = j;
+			k++;
+		}
+
+	return 0;
+}
+
 static int mr75203_probe(struct platform_device *pdev)
 {
 	const struct hwmon_channel_info **pvt_info;
@@ -565,7 +633,6 @@ static int mr75203_probe(struct platform_device *pdev)
 	pvt->t_num = ts_num;
 	pvt->p_num = pd_num;
 	pvt->v_num = vm_num;
-	pvt->c_num = ch_num;
 	val = 0;
 	if (ts_num)
 		val++;
@@ -602,43 +669,41 @@ static int mr75203_probe(struct platform_device *pdev)
 	}
 
 	if (vm_num) {
-		u32 total_ch = ch_num * vm_num;
+		u8 vm_idx[VM_NUM_MAX];
 
 		ret = pvt_get_regmap(pdev, "vm", pvt);
 		if (ret)
 			return ret;
 
-		pvt->vm_idx = devm_kcalloc(dev, vm_num, sizeof(*pvt->vm_idx),
-					   GFP_KERNEL);
-		if (!pvt->vm_idx)
-			return -ENOMEM;
-
-		ret = device_property_read_u8_array(dev, "intel,vm-map",
-						    pvt->vm_idx, vm_num);
+		ret = device_property_read_u8_array(dev, "intel,vm-map", vm_idx,
+						    vm_num);
 		if (ret) {
 			/*
 			 * Incase intel,vm-map property is not defined, we
 			 * assume incremental channel numbers.
 			 */
 			for (i = 0; i < vm_num; i++)
-				pvt->vm_idx[i] = i;
+				vm_idx[i] = i;
 		} else {
 			for (i = 0; i < vm_num; i++)
-				if (pvt->vm_idx[i] >= vm_num ||
-				    pvt->vm_idx[i] == 0xff)
+				if (vm_idx[i] >= vm_num || vm_idx[i] == 0xff)
 					break;
 
 			vm_num = i;
 			pvt->v_num = i;
 		}
 
-		in_config = devm_kcalloc(dev, total_ch + 1,
+		ret = pvt_get_active_channel(dev, pvt, vm_num, ch_num, vm_idx);
+		if (ret)
+			return ret;
+
+		in_config = devm_kcalloc(dev, pvt->vm_channels.total + 1,
 					 sizeof(*in_config), GFP_KERNEL);
 		if (!in_config)
 			return -ENOMEM;
 
-		memset32(in_config, HWMON_I_INPUT, total_ch);
-		in_config[total_ch] = 0;
+		memset32(in_config, HWMON_I_INPUT, pvt->vm_channels.total);
+		in_config[pvt->vm_channels.total] = 0;
 		pvt_in.config = in_config;
 
 		pvt_info[index++] = &pvt_in;
Signature · 1 line
2.37.1
#46 lore raw

Re: [PATCH v3 09/19] hwmon: (mr75203) add VM active channel support

Quoted 9 lines · Eliav Farber

Add active channel support per voltage monitor.
The number of active channels is read from the device-tree. When absent in device-tree, all channels are assumed to be used.

This shall be useful to expose sysfs only for inputs that are connected to a voltage source.

Setting number of active channels to 0, means that entire VM sensor is not used.

...

drivers/hwmon/mr75203.c line 119

Quoted 9 lines
drivers/hwmon/mr75203.c +9 -0
+struct voltage_device {
+	u32 vm_map;	/* Map channel number to VM index */
+	u32 ch_map;	/* Map channel number to channel index */
+};
+
+struct voltage_channels {
+	u32 total;	/* Total number of channels in all VMs */
+	u8 max;		/* Maximum number of channels among all VMs */
+};

Why not convert them to kernel doc?

...

drivers/hwmon/mr75203.c static int pvt_reset_control_deassert(struct device *dev, struct pvt_device *pvt · line 544

Quoted 10 lines
drivers/hwmon/mr75203.c +10 -0
+	ret = device_property_read_u8_array(dev, "moortec,vm-active-channels",
+					    vm_active_ch, vm_num);
+	if (ret) {
+		/*
+		 * Incase vm-active-channels property is not defined,
+		 * we assume each VM sensor has all of its channels
+		 * active.
+		 */
+		for (i = 0; i < vm_num; i++)
+			vm_active_ch[i] = ch_num;

NIH memset().

Quoted 16 lines
drivers/hwmon/mr75203.c +16 -0
+		pvt->vm_channels.max = ch_num;
+		pvt->vm_channels.total = ch_num * vm_num;
+	} else {
+		for (i = 0; i < vm_num; i++) {
+			if (vm_active_ch[i] > ch_num) {
+				dev_err(dev, "invalid active channels: %u\n",
+					vm_active_ch[i]);
+				return -EINVAL;
+			}
+
+			pvt->vm_channels.total += vm_active_ch[i];
+
+			if (vm_active_ch[i] > pvt->vm_channels.max)
+				pvt->vm_channels.max = vm_active_ch[i];
+		}
+	}

...

drivers/hwmon/mr75203.c static int pvt_reset_control_deassert(struct device *dev, struct pvt_device *pvt · line 579

Quoted 7 lines
drivers/hwmon/mr75203.c +6 -0
+	k = 0;
+	for (i = 0; i < vm_num; i++)
+		for (j = 0; j < vm_active_ch[i]; j++) {
+			pvt->vd[k].vm_map = vm_idx[i];
+			pvt->vd[k].ch_map = j;
 
+			k++;

How is it different from moving this outside the inner loop as

	k += vm_active_ch[i];

?

Review

Quoted 1 line
(quoted diff) +1 -0
+		}

Missed outer {}.

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

Re: [PATCH v3 09/19] hwmon: (mr75203) add VM active channel support

Quoted 13 lines · Andy Shevchenko
Quoted 9 lines · Eliav Farber

Add active channel support per voltage monitor.
The number of active channels is read from the device-tree. When absent in device-tree, all channels are assumed to be used.

This shall be useful to expose sysfs only for inputs that are connected to a voltage source.

Setting number of active channels to 0, means that entire VM sensor is not used.

...

Quoted 12 lines

drivers/hwmon/mr75203.c line 119

Quoted 9 lines
drivers/hwmon/mr75203.c +9 -0
+struct voltage_device {
+     u32 vm_map;     /* Map channel number to VM index */
+     u32 ch_map;     /* Map channel number to channel index */
+};
+
+struct voltage_channels {
+     u32 total;      /* Total number of channels in all VMs */
+     u8 max;         /* Maximum number of channels among all VMs */
+};

Why not convert them to kernel doc?

Fixed in v4.

Quoted 12 lines
Quoted 11 lines

+     ret = device_property_read_u8_array(dev, "moortec,vm-active-channels",
+                                         vm_active_ch, vm_num); +     if (ret) {
+             /*
+              * Incase vm-active-channels property is not defined, +              * we assume each VM sensor has all of its channels +              * active.
+              */
+             for (i = 0; i < vm_num; i++)
+                     vm_active_ch[i] = ch_num;

Quoted 1 line

NIH memset().

Fixed in v4.

Quoted 26 lines
Quoted 17 lines

+             pvt->vm_channels.max = ch_num;
+             pvt->vm_channels.total = ch_num * vm_num;
+     } else {
+             for (i = 0; i < vm_num; i++) {
+                     if (vm_active_ch[i] > ch_num) {
+                             dev_err(dev, "invalid active channels:
%u\n",
+                                     vm_active_ch[i]);
+                             return -EINVAL;
+                     }

drivers/hwmon/mr75203.c +7 -0
+
+                     pvt->vm_channels.total += vm_active_ch[i];
+
+                     if (vm_active_ch[i] > pvt->vm_channels.max)
+                             pvt->vm_channels.max = vm_active_ch[i];
+             }
+     }

...

Quoted 5 lines

+     k = 0;
+     for (i = 0; i < vm_num; i++)
+             for (j = 0; j < vm_active_ch[i]; j++) { +                     pvt->vd[k].vm_map = vm_idx[i]; +                     pvt->vd[k].ch_map = j;

Quoted 7 lines
Quoted 1 line

+                     k++;

How is it different from moving this outside the inner loop as

       k += vm_active_ch[i];

?

k is used inside the inner loop, so increasing it outside the inner loop will result in a different incorrect setting of vm_map and ch_map.

Quoted 3 lines
Quoted 1 line

+             }

Missed outer {}.

Fixed in v4.

--
Regards, Eliav

#48 lore raw

PATCH v3 10/19 dt-bindings: hwmon: (mr75203) add "moortec,vm-pre-scaler" property

In reply to Eliav Farber

Add support for mr76006 pre-scaler which provides divide-by-2 scaling of input voltage, which can then be presented for VM for measurement within its range (the VM input range is limited to -0.1V to 1V).

The new "moortec,vm-pre-scaler" property lists the channels that use a pre-scaler.

The driver will use this list to multiply the voltage result by 2, to present to the user the actual voltage input source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Change property format to be a single u8 array.
- Fix typo: scalar --> scaler.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++
 1 file changed, 11 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +11 -0
index 69cc6caceb2c..4c983d8f8fe7 100644
@@ -54,6 +54,16 @@ properties:
     default: 16
     $ref: /schemas/types.yaml#definitions/uint8-array
 
+  moortec,vm-pre-scaler:
+    description:
+      moortec,vm-pre-scaler property is an array of channels that use a mr76006
+      pre-scaler to divides the input source by 2.
+      The pre-scaler is used for input sources that exceed the VM input range.
+      The driver uses this information to present to the user the actual value
+      of the voltage source.
+    default: 1
+    $ref: /schemas/types.yaml#definitions/uint8-array
+
 required:
   - compatible
   - reg
@@ -76,5 +86,6 @@ examples:
         clocks = <&osc0>;
         resets = <&rcu0 0x40 7>;
         moortec,vm-active-channels = <0x10 0x05>;
+        moortec,vm-pre-scaler = <5 6>;
         #thermal-sensor-cells = <1>;
     };
Signature · 1 line
2.37.1
maintainer
#49 lore raw

Re: [PATCH v3 10/19] dt-bindings: hwmon: (mr75203) add "moortec,vm-pre-scaler" property

Quoted 25 lines · Eliav Farber

Add support for mr76006 pre-scaler which provides divide-by-2 scaling of input voltage, which can then be presented for VM for measurement within its range (the VM input range is limited to -0.1V to 1V).

The new "moortec,vm-pre-scaler" property lists the channels that use a pre-scaler.

The driver will use this list to multiply the voltage result by 2, to present to the user the actual voltage input source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Change property format to be a single u8 array.
- Fix typo: scalar --> scaler.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++
 1 file changed, 11 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index 69cc6caceb2c..4c983d8f8fe7 100644

Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml properties: · line 60

Quoted 7 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +4 -0
@@ -54,6 +54,16 @@ properties:
     default: 16
     $ref: /schemas/types.yaml#definitions/uint8-array
 
+  moortec,vm-pre-scaler:
+    description:
+      moortec,vm-pre-scaler property is an array of channels that use a mr76006
+      pre-scaler to divides the input source by 2.

to divide the

You don't need the property name in the description. The entries are the pre-scaler values for each channel? The array index is the channel? If so, then 'an array of pre-scaler values for each channel ...'.

Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml properties: · line 64

Quoted 4 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +4 -0
+      The pre-scaler is used for input sources that exceed the VM input range.
+      The driver uses this information to present to the user the actual value
+      of the voltage source.
+    default: 1

It's an array, so a scalar default doesn't make sense.

Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml properties: · line 65

Quoted 1 line
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +1 -0
+    $ref: /schemas/types.yaml#definitions/uint8-array

Constraints? I assume there's a finite number of channels to set the array size bounds for example.

Quoted 15 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +2 -0
+
 required:
   - compatible
   - reg
@@ -76,5 +86,6 @@ examples:
         clocks = <&osc0>;
         resets = <&rcu0 0x40 7>;
         moortec,vm-active-channels = <0x10 0x05>;
+        moortec,vm-pre-scaler = <5 6>;
         #thermal-sensor-cells = <1>;
     };
Signature · 1 line
2.37.1
#50 lore raw

Re: [PATCH v3 10/19] dt-bindings: hwmon: (mr75203) add "moortec, vm-pre-scaler" property

Quoted 29 lines · Rob Herring
Quoted 28 lines · Eliav Farber

Add support for mr76006 pre-scaler which provides divide-by-2 scaling of input voltage, which can then be presented for VM for measurement within its range (the VM input range is limited to -0.1V to 1V).

The new "moortec,vm-pre-scaler" property lists the channels that use a pre-scaler.

The driver will use this list to multiply the voltage result by 2, to present to the user the actual voltage input source.

---
V3 -> V2:
- Add "moortec" prefix to property name.
- Change property format to be a single u8 array.
- Fix typo: scalar --> scaler.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml    | 11 +++++++++++

 1 file changed, 11 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index 69cc6caceb2c..4c983d8f8fe7 100644
@@ -54,6 +54,16 @@ properties:

     default: 16
     $ref: /schemas/types.yaml#definitions/uint8-array

Quoted 7 lines
Quoted 5 lines

+  moortec,vm-pre-scaler:
+    description:
+      moortec,vm-pre-scaler property is an array of channels that use a mr76006
+      pre-scaler to divides the input source by 2.

to divide the

Fixed in v4.

Quoted 3 lines

You don't need the property name in the description. The entries are the pre-scaler values for each channel? The array index is the channel? If so, then 'an array of pre-scaler values for each channel ...'.

Removed property name in the description.

The entries are channel numbers that use a pre-scaler.
Assume 2 VMs, with 16 channels each (so 32 channels in total, numbered from 0 to 31) and assume: moortec,vm-pre-scaler = /bits/ 8 <5 6 20>; This means that only channels 5 6 and 20 use a pre-scaler, and the driver will use a factor of 2 only for these channels.

For v4 I renamed the property name to be "moortec,vm-pre-scaler-x2". And I changed binding to:

moortec,vm-pre-scaler-x2:
description:
  Defines the channels that use a mr76006 pre-scaler to divide the input source by 2.
  The pre-scaler is used for input sources that exceed the VM input range. The driver uses this information to present to the user with the actual value of the voltage source.
  For channels that are not listed, no pre-scaler is assumed. Maximum number of items - total number of channels in all VMs. Each channel should not appear more than once.
$ref: /schemas/types.yaml#/definitions/uint8-array

Quoted 8 lines
Quoted 6 lines

+      The pre-scaler is used for input sources that exceed the VM input range.
+      The driver uses this information to present to the user the actual value
+      of the voltage source.
+    default: 1

It's an array, so a scalar default doesn't make sense.

What I meant was the in-case channel is not defined the default pre- scaler value is 1.

Quoted 4 lines
Quoted 1 line

+    $ref: /schemas/types.yaml#definitions/uint8-array

Constraints? I assume there's a finite number of channels to set the array size bounds for example.

Added some constraints in the new description above.

--
Thanks, Eliav

#51 lore raw

PATCH v3 11/19 hwmon: (mr75203) add VM pre-scaler support

In reply to Eliav Farber

Add mr76006 pre-scaler support to normalize the voltage output result for channels that use pre-scaler units to get the measurement to be within the range that the sensor supports.

For channels that are listed in the device-tree to have a pre-scaler, the voltage result is multiplied by a factor of 2, to represent to the user the actual voltage source which is measured.

---
V3 -> V2:
- Modify code according to new property format.

 drivers/hwmon/mr75203.c | 64 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 62 insertions(+), 2 deletions(-)
drivers/hwmon/mr75203.c +62 -2
index 6925e8490587..1cd5ff6eacce 100644
@@ -13,10 +13,12 @@
 #include <linux/module.h>
 #include <linux/mod_devicetable.h>
 #include <linux/mutex.h>
+#include <linux/of.h>
 #include <linux/platform_device.h>
 #include <linux/property.h>
 #include <linux/regmap.h>
 #include <linux/reset.h>
+#include <linux/slab.h>
 
 /* PVT Common register */
 #define PVT_IP_CONFIG	0x04
@@ -108,9 +110,13 @@
 #define PVT_N_CONST		90
 #define PVT_R_CONST		245805
 
+#define PRE_SCALER_X1	1
+#define PRE_SCALER_X2	2
+
 struct voltage_device {
 	u32 vm_map;	/* Map channel number to VM index */
 	u32 ch_map;	/* Map channel number to channel index */
+	u32 pre_scaler;
 };
 
 struct voltage_channels {
@@ -192,7 +198,7 @@ static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val)
 {
 	struct pvt_device *pvt = dev_get_drvdata(dev);
 	struct regmap *v_map = pvt->v_map;
-	u32 n, stat;
+	u32 n, stat, pre_scaler;
 	u8 vm_idx, ch_idx;
 	int ret;
 
@@ -217,7 +223,9 @@ static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val)
 
 		n &= SAMPLE_DATA_MSK;
 		/* Convert the N bitstream count into voltage */
-		*val = (PVT_N_CONST * n - PVT_R_CONST) >> PVT_CONV_BITS;
+		pre_scaler = pvt->vd[channel].pre_scaler;
+		*val = pre_scaler * (PVT_N_CONST * n - PVT_R_CONST) >>
+			PVT_CONV_BITS;
 
 		return 0;
 	default:
@@ -582,6 +590,54 @@ static int pvt_get_active_channel(struct device *dev, struct pvt_device *pvt,
 	return 0;
 }
 
+static int pvt_get_pre_scaler(struct device *dev, struct pvt_device *pvt)
+{
+	const struct device_node *np = dev->of_node;
+	u32 total_channels = pvt->vm_channels.total;
+	u32 channel;
+	u8 *pre_scaler_ch_list;
+	int i, ret, num_ch;
+
+	/* Set default pre-scaler value to be 1. */
+	for (i = 0; i < total_channels; i++)
+		pvt->vd[i].pre_scaler = PRE_SCALER_X1;
+
+	/* Get number of channels configured in "moortec,vm-pre-scaler". */
+	num_ch = of_property_count_u8_elems(np, "moortec,vm-pre-scaler");
+	if (num_ch <= 0)
+		return 0;
+
+	pre_scaler_ch_list = kcalloc(total_channels,
+				     sizeof(*pre_scaler_ch_list), GFP_KERNEL);
+	if (!pre_scaler_ch_list)
+		return -ENOMEM;
+
+	/* Get list of all channels that have pre-scaler of 2. */
+	ret = device_property_read_u8_array(dev, "moortec,vm-pre-scaler",
+					    pre_scaler_ch_list, num_ch);
+	if (ret)
+		goto out;
+
+	for (i = 0; i < num_ch; i++) {
+		channel = pre_scaler_ch_list[i];
+
+		if (channel >= total_channels) {
+			dev_err(dev,
+				"invalid channel (%u) in pre-scaler list\n",
+				channel);
+			ret = -EINVAL;
+			goto out;
+		}
+
+		pvt->vd[channel].pre_scaler = PRE_SCALER_X2;
+	}
+
+out:
+	kfree(pre_scaler_ch_list);
+
+	return ret;
+}
+
 static int mr75203_probe(struct platform_device *pdev)
 {
 	const struct hwmon_channel_info **pvt_info;
@@ -697,6 +753,10 @@ static int mr75203_probe(struct platform_device *pdev)
 		if (ret)
 			return ret;
 
+		ret = pvt_get_pre_scaler(dev, pvt);
+		if (ret)
+			return ret;
+
 		in_config = devm_kcalloc(dev, pvt->vm_channels.total + 1,
 					 sizeof(*in_config), GFP_KERNEL);
 		if (!in_config)
Signature · 1 line
2.37.1
#52 lore raw

Re: [PATCH v3 11/19] hwmon: (mr75203) add VM pre-scaler support

Quoted 7 lines · Eliav Farber

Add mr76006 pre-scaler support to normalize the voltage output result for channels that use pre-scaler units to get the measurement to be within the range that the sensor supports.

For channels that are listed in the device-tree to have a pre-scaler, the voltage result is multiplied by a factor of 2, to represent to the user the actual voltage source which is measured.

...

Quoted 4 lines
drivers/hwmon/mr75203.c +4 -0
+static int pvt_get_pre_scaler(struct device *dev, struct pvt_device *pvt)
+{
+	const struct device_node *np = dev->of_node;
+	u32 total_channels = pvt->vm_channels.total;

drivers/hwmon/mr75203.c static int pvt_get_active_channel(struct device *dev, struct pvt_device *pvt, · line 606

Quoted 10 lines
drivers/hwmon/mr75203.c +10 -0
+	u32 channel;
+	u8 *pre_scaler_ch_list;
+	int i, ret, num_ch;
+
+	/* Set default pre-scaler value to be 1. */
+	for (i = 0; i < total_channels; i++)
+		pvt->vd[i].pre_scaler = PRE_SCALER_X1;
+
+	/* Get number of channels configured in "moortec,vm-pre-scaler". */
+	num_ch = of_property_count_u8_elems(np, "moortec,vm-pre-scaler");

of_ ?!

Quoted 8 lines
drivers/hwmon/mr75203.c +8 -0
+	if (num_ch <= 0)
+		return 0;
+
+	pre_scaler_ch_list = kcalloc(total_channels,
+				     sizeof(*pre_scaler_ch_list), GFP_KERNEL);
+	if (!pre_scaler_ch_list)
+		return -ENOMEM;
+

drivers/hwmon/mr75203.c static int pvt_get_active_channel(struct device *dev, struct pvt_device *pvt, · line 623

Quoted 10 lines
drivers/hwmon/mr75203.c +9 -0
+	/* Get list of all channels that have pre-scaler of 2. */
+	ret = device_property_read_u8_array(dev, "moortec,vm-pre-scaler",
+					    pre_scaler_ch_list, num_ch);
+	if (ret)
+		goto out;
+
+	for (i = 0; i < num_ch; i++) {
+		channel = pre_scaler_ch_list[i];
 
+

Unnecessary blank line.

drivers/hwmon/mr75203.c static int pvt_get_active_channel(struct device *dev, struct pvt_device *pvt, · line 629

Quoted 7 lines
drivers/hwmon/mr75203.c +6 -0
+		if (channel >= total_channels) {
+			dev_err(dev,
+				"invalid channel (%u) in pre-scaler list\n",
+				channel);
+			ret = -EINVAL;
 
+			goto out;

Wouldn't

			break;

suffice? (I understand the point, up to you)

drivers/hwmon/mr75203.c static int pvt_get_active_channel(struct device *dev, struct pvt_device *pvt, · line 635

Quoted 6 lines
drivers/hwmon/mr75203.c +6 -0
+		}
+
+		pvt->vd[channel].pre_scaler = PRE_SCALER_X2;
+	}
+
+out:

out_free:

Quoted 4 lines
drivers/hwmon/mr75203.c +4 -0
+	kfree(pre_scaler_ch_list);
+
+	return ret;
+}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#53 lore raw

Re: [PATCH v3 11/19] hwmon: (mr75203) add VM pre-scaler support

Quoted 11 lines · Andy Shevchenko
Quoted 10 lines · Eliav Farber

+static int pvt_get_pre_scaler(struct device *dev, struct pvt_device *pvt)

drivers/hwmon/mr75203.c +8 -0
+{
+     const struct device_node *np = dev->of_node;
+     u32 total_channels = pvt->vm_channels.total;
+     u32 channel;
+     u8 *pre_scaler_ch_list;
+     int i, ret, num_ch;
+
+     /* Set default pre-scaler value to be 1. */
Quoted 9 lines
Quoted 6 lines

+     for (i = 0; i < total_channels; i++)
+             pvt->vd[i].pre_scaler = PRE_SCALER_X1;
+
+     /* Get number of channels configured in "moortec,vm-pre-scaler". */
+     num_ch = of_property_count_u8_elems(np, "moortec,vm-pre-scaler");

of_ ?!

Replaced of_property_count_u8_elems() with
device_property_count_u8().

Quoted 18 lines
Quoted 17 lines

+     if (num_ch <= 0)
+             return 0;
+
+     pre_scaler_ch_list = kcalloc(total_channels,
+                                  sizeof(*pre_scaler_ch_list), GFP_KERNEL);
+     if (!pre_scaler_ch_list)
+             return -ENOMEM;

drivers/hwmon/mr75203.c +9 -0
+
+     /* Get list of all channels that have pre-scaler of 2. */
+     ret = device_property_read_u8_array(dev, "moortec,vm-pre-scaler",
+                                         pre_scaler_ch_list, num_ch);
+     if (ret)
+             goto out;
+
+     for (i = 0; i < num_ch; i++) {
+             channel = pre_scaler_ch_list[i];
Quoted 3 lines
Quoted 1 line

+

Unnecessary blank line.

Blank line removed.

Quoted 7 lines
Quoted 6 lines

+             if (channel >= total_channels) {
+                     dev_err(dev,
+                             "invalid channel (%u) in pre-scaler list\n",
+                             channel);
+                     ret = -EINVAL;

Quoted 7 lines
Quoted 1 line

+                     goto out;

Wouldn't

                       break;

suffice? (I understand the point, up to you)

I prefer to exit the moment I detect a problem.
For now I can use a break but in the future someone else can add new code in between that will set ret to 0 and instead of failing driver flow will continue with incomplete pre-scaler value.
So I prefer keeping it as it.

--
Best regards, Eliav

#54 lore raw

PATCH v3 12/19 hwmon: (mr75203) fix voltage equation for negative source input

In reply to Eliav Farber

According to Moortec Embedded Voltage Monitor (MEVM) series 3 data sheet, the minimum input signal is -100mv and maximum input signal is +1000mv. When n was small enough, such that PVT_N_CONST * n < PVT_R_CONST it resulted in n overflowing to a very large number (since n is u32 type).

This change fixes the problem by casting n to long and replacing shift right with div operation.

---
V3 -> V2:
- Fix equation to support negative values instead of limiting value to zero.

 drivers/hwmon/mr75203.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/hwmon/mr75203.c +3 -2
index 1cd5ff6eacce..d1f090a9baac 100644
@@ -222,10 +222,11 @@ static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val)
 			return ret;
 
 		n &= SAMPLE_DATA_MSK;
+
 		/* Convert the N bitstream count into voltage */
 		pre_scaler = pvt->vd[channel].pre_scaler;
-		*val = pre_scaler * (PVT_N_CONST * n - PVT_R_CONST) >>
-			PVT_CONV_BITS;
+		*val = pre_scaler * (PVT_N_CONST * (long)n - PVT_R_CONST) /
+			(1 << PVT_CONV_BITS);
 
 		return 0;
 	default:
Signature · 1 line
2.37.1
#55 lore raw

Re: [PATCH v3 12/19] hwmon: (mr75203) fix voltage equation for negative source input

Quoted 7 lines · Eliav Farber

According to Moortec Embedded Voltage Monitor (MEVM) series 3 data sheet, the minimum input signal is -100mv and maximum input signal is +1000mv. When n was small enough, such that PVT_N_CONST * n < PVT_R_CONST it resulted in n overflowing to a very large number (since n is u32 type).

This change fixes the problem by casting n to long and replacing shift right with div operation.

Fixes tag?

...

drivers/hwmon/mr75203.c static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val) · line 225

Quoted 2 lines
drivers/hwmon/mr75203.c +1 -0
 		n &= SAMPLE_DATA_MSK;
+

Unrelated change.

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

Re: [PATCH v3 12/19] hwmon: (mr75203) fix voltage equation for negative source input

Quoted 10 lines · Andy Shevchenko
Quoted 7 lines · Eliav Farber

According to Moortec Embedded Voltage Monitor (MEVM) series 3 data sheet, the minimum input signal is -100mv and maximum input signal is +1000mv. When n was small enough, such that PVT_N_CONST * n < PVT_R_CONST it resulted in n overflowing to a very large number (since n is u32 type).

This change fixes the problem by casting n to long and replacing shift right with div operation.

Fixes tag?

For v4 I modified the commit message to (hopefully) be more understandable:

"
According to Moortec Embedded Voltage Monitor (MEVM) series 3 data sheet, the minimum input signal is -100mv and maximum input signal is +1000mv.

On 64 bit machines sizeof(u32) = 4 and sizeof(long) = 8.
So when measuring a negative input and n is small enough, such that PVT_N_CONST * n < PVT_R_CONST, it results in n overflowing to a very large number which is not negative (because 4 MSB bytes of val are 0).

This change fixes the sign problem and supports negative values by casting n to long and replacing shift right with div operation. "

Quoted 6 lines

...

Quoted 2 lines

              n &= SAMPLE_DATA_MSK;
+

Unrelated change.

Removed.

--
Thanks, Eliav

#57 lore raw

Re: [PATCH v3 12/19] hwmon: (mr75203) fix voltage equation for negative source input

Quoted 21 lines · Eliav
Quoted 11 lines · Andy Shevchenko
Quoted 8 lines · Eliav Farber

According to Moortec Embedded Voltage Monitor (MEVM) series 3 data sheet,
the minimum input signal is -100mv and maximum input signal is +1000mv. When n was small enough, such that PVT_N_CONST * n < PVT_R_CONST it resulted in n overflowing to a very large number (since n is u32 type).

This change fixes the problem by casting n to long and replacing shift right with div operation.

Fixes tag?

For v4 I modified the commit message to (hopefully) be more understandable:

"
According to Moortec Embedded Voltage Monitor (MEVM) series 3 data sheet, the minimum input signal is -100mv and maximum input signal is +1000mv.

Quoted 8 lines

On 64 bit machines sizeof(u32) = 4 and sizeof(long) = 8.
So when measuring a negative input and n is small enough, such that PVT_N_CONST * n < PVT_R_CONST, it results in n overflowing to a very large number which is not negative (because 4 MSB bytes of val are 0).

This change fixes the sign problem and supports negative values by casting n to long and replacing shift right with div operation. "

What I meant is to add the tag of the commit which this one is fixing. We have Fixes tag format for that. You may see how it's done by looking into Git history: git log --grep Fixes:

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

PATCH v3 13/19 hwmon: (mr75203) modify the temperature equation according to series 5 datasheet

In reply to Eliav Farber

Modify the equation and coefficients used to convert the digital output to temperature according to series 5 of the Moortec Embedded Temperature Sensor (METS) datasheet:
T = G + H * (n / cal5 - 0.5) + J * F

Where:
*) G = 60, H = 200, cal5 = 4094, J = -0.1.
*) F = frequency clock in MHz.
*) n is the digital output.

In code, the G, H and J coefficients are multiplied by a factor of 1000 to get the temperature in milli-Celsius.
Final result is clamped in case it exceeds min/max thresholds.

Change is done since it is not clear where the current equation and coefficients came from.

---
V3 -> V2:
- Protect from overflow.
- Add temperature clamping.
- Add better documentation.

 drivers/hwmon/mr75203.c | 29 +++++++++++++++++++++--------
 1 file changed, 21 insertions(+), 8 deletions(-)
drivers/hwmon/mr75203.c +21 -8
index d1f090a9baac..5b8ca1ee4a54 100644
@@ -103,13 +103,17 @@
 
 #define PVT_POLL_DELAY_US	20
 #define PVT_POLL_TIMEOUT_US	20000
-#define PVT_H_CONST		100000
-#define PVT_CAL5_CONST		2047
-#define PVT_G_CONST		40000
+#define PVT_H_CONST		200000
+#define PVT_G_CONST		60000
+#define PVT_J_CONST		-100
+#define PVT_CAL5_CONST		4094
 #define PVT_CONV_BITS		10
 #define PVT_N_CONST		90
 #define PVT_R_CONST		245805
 
+#define PVT_TEMP_MIN		-40000
+#define PVT_TEMP_MAX		125000
+
 #define PRE_SCALER_X1	1
 #define PRE_SCALER_X2	2
 
@@ -157,13 +161,24 @@ static umode_t pvt_is_visible(const void *data, enum hwmon_sensor_types type,
 	return 0;
 }
 
+static long pvt_calc_temp(struct pvt_device *pvt, u32 nbs)
+{
+	/*
+	 * Convert the register value to degrees centigrade temperature:
+	 * T = G + H * (n / cal5 - 0.5) + J * F
+	 */
+	s64 tmp = PVT_G_CONST + PVT_H_CONST * (s64)nbs / PVT_CAL5_CONST -
+		PVT_H_CONST / 2 + PVT_J_CONST * (s64)pvt->ip_freq / HZ_PER_MHZ;
+
+	return clamp_val(tmp, PVT_TEMP_MIN, PVT_TEMP_MAX);
+}
+
 static int pvt_read_temp(struct device *dev, u32 attr, int channel, long *val)
 {
 	struct pvt_device *pvt = dev_get_drvdata(dev);
 	struct regmap *t_map = pvt->t_map;
 	u32 stat, nbs;
 	int ret;
-	u64 tmp;
 
 	switch (attr) {
 	case hwmon_temp_input:
@@ -184,9 +199,7 @@ static int pvt_read_temp(struct device *dev, u32 attr, int channel, long *val)
 		 * Convert the register value to
 		 * degrees centigrade temperature
 		 */
-		tmp = nbs * PVT_H_CONST;
-		do_div(tmp, PVT_CAL5_CONST);
-		*val = tmp - PVT_G_CONST - pvt->ip_freq;
+		*val = pvt_calc_temp(pvt, nbs);
 
 		return 0;
 	default:
@@ -311,7 +324,7 @@ static int pvt_init(struct pvt_device *pvt)
 		    (key >> 1) << CLK_SYNTH_HI_SFT |
 		    (key >> 1) << CLK_SYNTH_HOLD_SFT | CLK_SYNTH_EN;
 
-	pvt->ip_freq = sys_freq * 100 / (key + 2);
+	pvt->ip_freq = clk_get_rate(pvt->clk) / (key + 2);
 
 	if (t_num) {
 		ret = regmap_write(t_map, SDIF_SMPL_CTRL, 0x0);
Signature · 1 line
2.37.1
#59 lore raw

Re: [PATCH v3 13/19] hwmon: (mr75203) modify the temperature equation according to series 5 datasheet

Quoted 10 lines · Eliav Farber

Modify the equation and coefficients used to convert the digital output to temperature according to series 5 of the Moortec Embedded Temperature Sensor (METS) datasheet:
T = G + H * (n / cal5 - 0.5) + J * F

Where:
*) G = 60, H = 200, cal5 = 4094, J = -0.1.
*) F = frequency clock in MHz.
*) n is the digital output.

Quoted 5 lines

In code, the G, H and J coefficients are multiplied by a factor of 1000 to get the temperature in milli-Celsius.
Final result is clamped in case it exceeds min/max thresholds.

Change is done since it is not clear where the current equation and

not clear -> unclear

Quoted 1 line

coefficients came from.

...

drivers/hwmon/mr75203.c line 115

Quoted 2 lines
drivers/hwmon/mr75203.c +2 -0
+#define PVT_TEMP_MIN		-40000
+#define PVT_TEMP_MAX		125000

Unit suffix? _mC perhaps would be enough.

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

Re: [PATCH v3 13/19] hwmon: (mr75203) modify the temperature equation according to series 5 datasheet

Quoted 11 lines · Andy Shevchenko
Quoted 10 lines · Eliav Farber

Modify the equation and coefficients used to convert the digital output to temperature according to series 5 of the Moortec Embedded Temperature Sensor (METS) datasheet:
T = G + H * (n / cal5 - 0.5) + J * F

Where:
*) G = 60, H = 200, cal5 = 4094, J = -0.1.
*) F = frequency clock in MHz.
*) n is the digital output.

Quoted 8 lines
Quoted 5 lines

In code, the G, H and J coefficients are multiplied by a factor of 1000 to get the temperature in milli-Celsius.
Final result is clamped in case it exceeds min/max thresholds.

Change is done since it is not clear where the current equation and

not clear -> unclear

Fixed.

Quoted 8 lines
Quoted 1 line

coefficients came from.

...

drivers/hwmon/mr75203.c line 115

Quoted 2 lines
drivers/hwmon/mr75203.c +2 -0
+#define PVT_TEMP_MIN         -40000
+#define PVT_TEMP_MAX         125000

Unit suffix? _mC perhaps would be enough.

Fixed. Added _mC suffix as you suggested.

--
Thanks, Eliav

#61 lore raw

PATCH v3 14/19 dt-bindings: hwmon: (mr75203) add "moortec,ts-series" property

In reply to Eliav Farber

Add optional "moortec,ts-series" property to define the temperature equation and coefficients that shall be used to convert the digital output to value in milli-Celsius.
Supported series: 5 (default) and 6.

Series 5:
T = G + H * (n / cal5 - 0.5) + J * F
Where: G = 60, H = 200, cal5 = 4094, J = -0.1, F = frequency clock in MHz

Series 6:
T = G + H * (n / cal5 - 0.5)
Where: G = 57.4, H = 249.4, cal5 = 4096

---
V3 -> V2:
- New patch to introduce "moortec,ts-series" property.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml     | 10 ++++++++++
 1 file changed, 10 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +10 -0
index 4c983d8f8fe7..ec2dbe7da9c2 100644
@@ -64,6 +64,16 @@ properties:
     default: 1
     $ref: /schemas/types.yaml#definitions/uint8-array
 
+  moortec,ts-series:
+    description:
+      moortec,ts-series defines the temperature equation and coefficients that
+      shall be used to convert the digital output to value in milli-Celsius.
+      Supported series are 5 and 6.
+    minimum: 5
+    maximum: 6
+    default: 5
+    $ref: /schemas/types.yaml#definitions/uint32
+
 required:
   - compatible
   - reg
Signature · 1 line
2.37.1
#62 lore raw

Re: [PATCH v3 14/19] dt-bindings: hwmon: (mr75203) add "moortec,ts-series" property

Quoted 4 lines · Eliav Farber

Add optional "moortec,ts-series" property to define the temperature equation and coefficients that shall be used to convert the digital output to value in milli-Celsius.
Supported series: 5 (default) and 6.

Is this the difference between mr75xxx and mr76xxx series? If so, should be a compatible "moortec,mr76006" instead?
If the temperature equation could be derived from the compatible, this property would not be necessary.

regards
Philipp

#63 lore raw

Re: [PATCH v3 14/19] dt-bindings: hwmon: (mr75203) add "moortec, ts-series" property

Quoted 10 lines · Philipp Zabel
Quoted 4 lines · Eliav Farber

Add optional "moortec,ts-series" property to define the temperature equation and coefficients that shall be used to convert the digital output to value in milli-Celsius.
Supported series: 5 (default) and 6.

Is this the difference between mr75xxx and mr76xxx series? If so, should be a compatible "moortec,mr76006" instead?
If the temperature equation could be derived from the compatible, this property would not be necessary.

The PVT (Process, Voltage, Temperature) monitoring logic can be constructed from many different sub-blocks:
*) CONTROLLER (mr75203) - controlling TS, PD and VM. *) TS (mr74137) - for measuring temperature in ring. *) PD (mr74139) - for measuring IO based transistors. *) VM (mr74138) - for measuring voltage rails across the SoC. *) Ring oscillators (mr76007/mr76008)
*) Pre-scalers (mr76006)

Besides mr75203 which is digital all other IPs are analog. There is a single mr75203 and there can be several or none of the other units.

The kernel driver is only for the controller (mr75203).
The series 5 or 6 is relevant for the TS (mr74137) and not for the controller (mr75203).
Each of the analog units can have a different series number (for example we use series 3 of the VM).

That is why I didn't change the compatible of mr75203, and instead added a TS-series parameter.

--
Regards, Eliav

#64 lore raw

Re: [PATCH v3 14/19] dt-bindings: hwmon: (mr75203) add "moortec, ts-series" property

Quoted 20 lines · Eliav
Quoted 10 lines · Philipp Zabel
Quoted 4 lines · Eliav Farber

Add optional "moortec,ts-series" property to define the temperature equation and coefficients that shall be used to convert the digital output to value in milli-Celsius.
Supported series: 5 (default) and 6.

Is this the difference between mr75xxx and mr76xxx series? If so, should be a compatible "moortec,mr76006" instead?
If the temperature equation could be derived from the compatible, this property would not be necessary.

The PVT (Process, Voltage, Temperature) monitoring logic can be constructed from many different sub-blocks:
*) CONTROLLER (mr75203) - controlling TS, PD and VM. *) TS (mr74137) - for measuring temperature in ring. *) PD (mr74139) - for measuring IO based transistors. *) VM (mr74138) - for measuring voltage rails across the SoC. *) Ring oscillators (mr76007/mr76008)
*) Pre-scalers (mr76006)

Quoted 3 lines

Besides mr75203 which is digital all other IPs are analog. There is a single mr75203 and there can be several or none of the other units.

Thank you for the explanation, I think this information would be nice to have in a description in moortec,mr75203.yaml.

Quoted 8 lines

The kernel driver is only for the controller (mr75203).
The series 5 or 6 is relevant for the TS (mr74137) and not for the controller (mr75203).
Each of the analog units can have a different series number (for example we use series 3 of the VM).

That is why I didn't change the compatible of mr75203, and instead added a TS-series parameter.

I see, I mistakenly assumed mr76006 referred to a new controller version. This also invalidates my comment on patch 4.

regards
Philipp

#65 lore raw

Re: [PATCH v3 14/19] dt-bindings: hwmon: (mr75203) add "moortec, ts-series" property

Quoted 21 lines · Philipp Zabel
Quoted 20 lines · Eliav
Quoted 10 lines · Philipp Zabel
Quoted 4 lines · Eliav Farber

Add optional "moortec,ts-series" property to define the temperature equation and coefficients that shall be used to convert the digital output to value in milli-Celsius.
Supported series: 5 (default) and 6.

Is this the difference between mr75xxx and mr76xxx series? If so, should be a compatible "moortec,mr76006" instead?
If the temperature equation could be derived from the compatible, this property would not be necessary.

The PVT (Process, Voltage, Temperature) monitoring logic can be constructed from many different sub-blocks:
*) CONTROLLER (mr75203) - controlling TS, PD and VM. *) TS (mr74137) - for measuring temperature in ring. *) PD (mr74139) - for measuring IO based transistors. *) VM (mr74138) - for measuring voltage rails across the SoC. *) Ring oscillators (mr76007/mr76008)
*) Pre-scalers (mr76006)

Quoted 6 lines
Quoted 3 lines

Besides mr75203 which is digital all other IPs are analog. There is a single mr75203 and there can be several or none of the other units.

Thank you for the explanation, I think this information would be nice to have in a description in moortec,mr75203.yaml.

For v4 I added a new patch which adds this description in moortec,mr75203.yaml:

description: |
  A Moortec PVT (Process, Voltage, Temperature) monitoring logic design can include many different units.
  Such a design will usually consists of several Moortec's embedded analog IPs, and a single Moortec controller to configure and control the IPs.

  Some of the Moortec's analog hard IPs that can be used in a design:
  *) Temperature Sensor (TS) - used to monitor core temperature (e.g. mr74137). *) Voltage Monitor (VM) - used to monitor voltage levels (e.g. mr74138). *) Process Detector (PD) - used to assess silicon speed (e.g. mr74139). *) Delay Chain - ring oscillator connected to the PD, used to measure IO based transistors (e.g. mr76008 ring oscillator at 1.1V, mr76007 ring oscillator at 1.8V).
  *) Pre Scaler - provides divide-by-X scaling of input voltage, which can then be presented for VM for measurement within its range (e.g. mr76006 - divide by 2 pre-scaler).

  TS, VM & PD also include a digital interface, which consists of configuration inputs and measurement outputs.
  The mr75203 binding describes configuration for the controller unit, but also for some of the analog IPs.

--
Regards, Eliav

maintainer
#66 lore raw

Re: [PATCH v3 14/19] dt-bindings: hwmon: (mr75203) add "moortec,ts-series" property

In reply to Eliav Farber
Quoted 26 lines · Eliav Farber

Add optional "moortec,ts-series" property to define the temperature equation and coefficients that shall be used to convert the digital output to value in milli-Celsius.
Supported series: 5 (default) and 6.

Series 5:
T = G + H * (n / cal5 - 0.5) + J * F
Where: G = 60, H = 200, cal5 = 4094, J = -0.1, F = frequency clock in MHz

Series 6:

   T = G + H * (n / cal5 - 0.5)

Where: G = 57.4, H = 249.4, cal5 = 4096

---
V3 -> V2:
- New patch to introduce "moortec,ts-series" property.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml     | 10 ++++++++++
 1 file changed, 10 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index 4c983d8f8fe7..ec2dbe7da9c2 100644

Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml properties: · line 71

Quoted 8 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +5 -0
@@ -64,6 +64,16 @@ properties:
     default: 1
     $ref: /schemas/types.yaml#definitions/uint8-array
 
+  moortec,ts-series:
+    description:
+      moortec,ts-series defines the temperature equation and coefficients that
+      shall be used to convert the digital output to value in milli-Celsius.
+      Supported series are 5 and 6.

No need to state constraints in free-form text descriptions.

Quoted 12 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +5 -0
+    minimum: 5
+    maximum: 6
+    default: 5
+    $ref: /schemas/types.yaml#definitions/uint32
+
 required:
   - compatible
   - reg
Signature · 1 line
2.37.1
#67 lore raw

Re: [PATCH v3 14/19] dt-bindings: hwmon: (mr75203) add "moortec, ts-series" property

Quoted 30 lines · Rob Herring
Quoted 29 lines · Eliav Farber

Add optional "moortec,ts-series" property to define the temperature equation and coefficients that shall be used to convert the digital output to value in milli-Celsius.
Supported series: 5 (default) and 6.

Series 5:
  T = G + H * (n / cal5 - 0.5) + J * F
Where: G = 60, H = 200, cal5 = 4094, J = -0.1, F = frequency clock in MHz

Series 6:
   T = G + H * (n / cal5 - 0.5)
Where: G = 57.4, H = 249.4, cal5 = 4096

---
V3 -> V2:
- New patch to introduce "moortec,ts-series" property.

 .../devicetree/bindings/hwmon/moortec,mr75203.yaml     | 10 ++++++++++

 1 file changed, 10 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index 4c983d8f8fe7..ec2dbe7da9c2 100644
@@ -64,6 +64,16 @@ properties:

     default: 1
     $ref: /schemas/types.yaml#definitions/uint8-array

Quoted 9 lines
Quoted 7 lines

+  moortec,ts-series:
+    description:
+      moortec,ts-series defines the temperature equation and coefficients that
+      shall be used to convert the digital output to value in milli-Celsius.
+      Supported series are 5 and 6.

No need to state constraints in free-form text descriptions.

Fixed in v4.

#68 lore raw

PATCH v3 15/19 hwmon: (mr75203) add support for series 6 temperature equation

In reply to Eliav Farber

The current equation used in code is aligned to series 5:
T = G + H * (n / cal5 - 0.5) + J * F
Where:
G = 60, H = 200, cal5 = 4094, J = -0.1, F = frequency clock in MHz

Series 6 has a slightly different equation:
T = G + H * (n / cal5 - 0.5)
and a different set of coefficients:
G = 57.4, H = 249.4, cal5 = 4096

This change supports equation and coefficients for both series. (for series 6, J is set to 0).

The series is determined according to “ts-series” property in device tree.
If absent, series 5 is assumed to be the default.

---
V3 -> V2:
- New patch to support temperature sensor series 6 instead of having to set all 4 coefficients.

 drivers/hwmon/mr75203.c | 69 +++++++++++++++++++++++++++++++++++++----
 1 file changed, 63 insertions(+), 6 deletions(-)
drivers/hwmon/mr75203.c +63 -6
index 5b8ca1ee4a54..6a035fd115ca 100644
@@ -103,10 +103,6 @@
 
 #define PVT_POLL_DELAY_US	20
 #define PVT_POLL_TIMEOUT_US	20000
-#define PVT_H_CONST		200000
-#define PVT_G_CONST		60000
-#define PVT_J_CONST		-100
-#define PVT_CAL5_CONST		4094
 #define PVT_CONV_BITS		10
 #define PVT_N_CONST		90
 #define PVT_R_CONST		245805
@@ -114,9 +110,31 @@
 #define PVT_TEMP_MIN		-40000
 #define PVT_TEMP_MAX		125000
 
+/* Temperature coefficients for series 5 */
+#define PVT_SERIES5_H_CONST	200000
+#define PVT_SERIES5_G_CONST	60000
+#define PVT_SERIES5_J_CONST	-100
+#define PVT_SERIES5_CAL5_CONST	4094
+
+/* Temperature coefficients for series 6 */
+#define PVT_SERIES6_H_CONST	249400
+#define PVT_SERIES6_G_CONST	57400
+#define PVT_SERIES6_J_CONST	0
+#define PVT_SERIES6_CAL5_CONST	4096
+
+#define TEMPERATURE_SENSOR_SERIES_5	5
+#define TEMPERATURE_SENSOR_SERIES_6	6
+
 #define PRE_SCALER_X1	1
 #define PRE_SCALER_X2	2
 
+struct temp_coeff {
+	u32 h;
+	u32 g;
+	u32 cal5;
+	s32 j;
+};
+
 struct voltage_device {
 	u32 vm_map;	/* Map channel number to VM index */
 	u32 ch_map;	/* Map channel number to channel index */
@@ -137,6 +155,7 @@ struct pvt_device {
 	struct reset_control	*rst;
 	struct voltage_device	*vd;
 	struct voltage_channels	vm_channels;
+	struct temp_coeff	ts_coeff;
 	u32			t_num;
 	u32			p_num;
 	u32			v_num;
@@ -167,8 +186,9 @@ static long pvt_calc_temp(struct pvt_device *pvt, u32 nbs)
 	 * Convert the register value to degrees centigrade temperature:
 	 * T = G + H * (n / cal5 - 0.5) + J * F
 	 */
-	s64 tmp = PVT_G_CONST + PVT_H_CONST * (s64)nbs / PVT_CAL5_CONST -
-		PVT_H_CONST / 2 + PVT_J_CONST * (s64)pvt->ip_freq / HZ_PER_MHZ;
+	struct temp_coeff *ts_coeff = &pvt->ts_coeff;
+	s64 tmp = ts_coeff->g + ts_coeff->h * (s64)nbs / ts_coeff->cal5 -
+		ts_coeff->h / 2 + ts_coeff->j * (s64)pvt->ip_freq / HZ_PER_MHZ;
 
 	return clamp_val(tmp, PVT_TEMP_MIN, PVT_TEMP_MAX);
 }
@@ -652,6 +672,39 @@ static int pvt_get_pre_scaler(struct device *dev, struct pvt_device *pvt)
 	return ret;
 }
 
+static int pvt_set_temp_coeff(struct device *dev, struct pvt_device *pvt)
+{
+	struct temp_coeff *ts_coeff = &pvt->ts_coeff;
+	const struct device_node *np = dev->of_node;
+	u32 series;
+	int ret;
+
+	/* Incase ts-series property is not defined, use default 5. */
+	ret = of_property_read_u32(np, "moortec,ts-series", &series);
+	if (ret)
+		series = TEMPERATURE_SENSOR_SERIES_5;
+
+	if (series == TEMPERATURE_SENSOR_SERIES_5) {
+		ts_coeff->h = PVT_SERIES5_H_CONST;
+		ts_coeff->g = PVT_SERIES5_G_CONST;
+		ts_coeff->j = PVT_SERIES5_J_CONST;
+		ts_coeff->cal5 = PVT_SERIES5_CAL5_CONST;
+	} else if (series == TEMPERATURE_SENSOR_SERIES_6) {
+		ts_coeff->h = PVT_SERIES6_H_CONST;
+		ts_coeff->g = PVT_SERIES6_G_CONST;
+		ts_coeff->j = PVT_SERIES6_J_CONST;
+		ts_coeff->cal5 = PVT_SERIES6_CAL5_CONST;
+	} else {
+		dev_err(dev, "invalid temperature sensor series (%u)\n",
+			series);
+		return -EINVAL;
+	}
+
+	dev_dbg(dev, "temperature sensor series = %u\n", series);
+
+	return 0;
+}
+
 static int mr75203_probe(struct platform_device *pdev)
 {
 	const struct hwmon_channel_info **pvt_info;
@@ -722,6 +775,10 @@ static int mr75203_probe(struct platform_device *pdev)
 		if (ret)
 			return ret;
 
+		ret = pvt_set_temp_coeff(dev, pvt);
+		if (ret)
+			return ret;
+
 		temp_config = devm_kcalloc(dev, ts_num + 1,
 					   sizeof(*temp_config), GFP_KERNEL);
 		if (!temp_config)
Signature · 1 line
2.37.1
#69 lore raw

Re: [PATCH v3 15/19] hwmon: (mr75203) add support for series 6 temperature equation

Quoted 16 lines · Eliav Farber

The current equation used in code is aligned to series 5:
T = G + H * (n / cal5 - 0.5) + J * F
Where:
G = 60, H = 200, cal5 = 4094, J = -0.1, F = frequency clock in MHz

Series 6 has a slightly different equation:
T = G + H * (n / cal5 - 0.5)
and a different set of coefficients:
G = 57.4, H = 249.4, cal5 = 4096

This change supports equation and coefficients for both series. (for series 6, J is set to 0).

The series is determined according to “ts-series” property in device tree.
If absent, series 5 is assumed to be the default.

...

drivers/hwmon/mr75203.c line 109

Quoted 4 lines
drivers/hwmon/mr75203.c +0 -4
-#define PVT_H_CONST		200000
-#define PVT_G_CONST		60000
-#define PVT_J_CONST		-100
-#define PVT_CAL5_CONST		4094

You just introduced them patch before. Please, avoid ping-pong style in the same series.

...

drivers/hwmon/mr75203.c static int pvt_get_pre_scaler(struct device *dev, struct pvt_device *pvt) · line 683

Quoted 1 line
drivers/hwmon/mr75203.c +1 -0
+	ret = of_property_read_u32(np, "moortec,ts-series", &series);

of_ ?!

Be consistent. Either you use OF everywhere, or device property APIs.

...

Quoted 8 lines
drivers/hwmon/mr75203.c +8 -0
+	if (ret)
+		series = TEMPERATURE_SENSOR_SERIES_5;
+
+	if (series == TEMPERATURE_SENSOR_SERIES_5) {
+		ts_coeff->h = PVT_SERIES5_H_CONST;
+		ts_coeff->g = PVT_SERIES5_G_CONST;
+		ts_coeff->j = PVT_SERIES5_J_CONST;
+		ts_coeff->cal5 = PVT_SERIES5_CAL5_CONST;

drivers/hwmon/mr75203.c static int pvt_get_pre_scaler(struct device *dev, struct pvt_device *pvt) · line 701

Quoted 10 lines
drivers/hwmon/mr75203.c +10 -0
+	} else if (series == TEMPERATURE_SENSOR_SERIES_6) {
+		ts_coeff->h = PVT_SERIES6_H_CONST;
+		ts_coeff->g = PVT_SERIES6_G_CONST;
+		ts_coeff->j = PVT_SERIES6_J_CONST;
+		ts_coeff->cal5 = PVT_SERIES6_CAL5_CONST;
+	} else {
+		dev_err(dev, "invalid temperature sensor series (%u)\n",
+			series);
+		return -EINVAL;
+	}

switch-case?

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

Re: [PATCH v3 15/19] hwmon: (mr75203) add support for series 6 temperature equation

Quoted 20 lines · Andy Shevchenko
Quoted 16 lines · Eliav Farber

The current equation used in code is aligned to series 5:
T = G + H * (n / cal5 - 0.5) + J * F
Where:
G = 60, H = 200, cal5 = 4094, J = -0.1, F = frequency clock in MHz

Series 6 has a slightly different equation:
T = G + H * (n / cal5 - 0.5)
and a different set of coefficients:
G = 57.4, H = 249.4, cal5 = 4096

This change supports equation and coefficients for both series. (for series 6, J is set to 0).

The series is determined according to “ts-series” property in device tree.
If absent, series 5 is assumed to be the default.

...

Quoted 7 lines

drivers/hwmon/mr75203.c line 109

Quoted 4 lines
drivers/hwmon/mr75203.c +0 -4
-#define PVT_H_CONST          200000
-#define PVT_G_CONST          60000
-#define PVT_J_CONST          -100
-#define PVT_CAL5_CONST               4094

You just introduced them patch before. Please, avoid ping-pong style in the same series.

Fixed for v4.
I now introduce these defines in patch 13 to avoid modifying the above ones and then remove them:
/* Temperature coefficients for series 5 */
#define PVT_SERIES5_H_CONST    200000
#define PVT_SERIES5_G_CONST    60000
#define PVT_SERIES5_J_CONST    -100
#define PVT_SERIES5_CAL5_CONST    4094

Quoted 8 lines

...

Quoted 1 line

+     ret = of_property_read_u32(np, "moortec,ts-series", &series);

of_ ?!

Be consistent. Either you use OF everywhere, or device property APIs.

Fixed for v4.
Using device_property_read_u32() instead.

Quoted 21 lines

...

Quoted 10 lines

+     if (ret)
+             series = TEMPERATURE_SENSOR_SERIES_5; +
+     if (series == TEMPERATURE_SENSOR_SERIES_5) { +             ts_coeff->h = PVT_SERIES5_H_CONST; +             ts_coeff->g = PVT_SERIES5_G_CONST; +             ts_coeff->j = PVT_SERIES5_J_CONST; +             ts_coeff->cal5 = PVT_SERIES5_CAL5_CONST; +     } else if (series == TEMPERATURE_SENSOR_SERIES_6) { +             ts_coeff->h = PVT_SERIES6_H_CONST;

Quoted 8 lines

+             ts_coeff->g = PVT_SERIES6_G_CONST;
+             ts_coeff->j = PVT_SERIES6_J_CONST;
+             ts_coeff->cal5 = PVT_SERIES6_CAL5_CONST;
+     } else {
+             dev_err(dev, "invalid temperature sensor series (%u)\n", +                     series);
+             return -EINVAL;
+     }

Quoted 1 line

switch-case?

Changed to switch-case (will be part of v4).

--
Thanks, Eliav

#71 lore raw

PATCH v3 16/19 dt-bindings: hwmon: (mr75203) add coefficient properties for the thermal equation

In reply to Eliav Farber

Add optional temperature coefficient properties:
*) moortec,ts-coeff-g
*) moortec,ts-coeff-h
*) moortec,ts-coeff-cal5
*) moortec,ts-coeff-j
If defined they shall be used instead of defaults.

The coefficients were added to device tree on top of the series property (which can be used to select between series 5 and series 6), because coefficients can vary between product and product, and code defaults might not be accurate enough.

---
V3 -> V2:
- Add "moortec" prefix to property name.

 .../bindings/hwmon/moortec,mr75203.yaml       | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +33 -0
index ec2dbe7da9c2..a92da6064285 100644
@@ -74,6 +74,37 @@ properties:
     default: 5
     $ref: /schemas/types.yaml#definitions/uint32
 
+  moortec,ts-coeff-g:
+    description:
+      G coefficient for temperature equation.
+      Value should be multiplied by factor 1000.
+      Default for series 5 = 60000
+      Default for series 6 = 57400
+    $ref: /schemas/types.yaml#/definitions/uint32
+
+  moortec,ts-coeff-h:
+    description:
+      H coefficient for temperature equation.
+      Value should be multiplied by factor 1000.
+      Default for series 5 = 200000
+      Default for series 6 = 249400
+    $ref: /schemas/types.yaml#/definitions/uint32
+
+  moortec,ts-coeff-cal5:
+    description:
+      cal5 coefficient for temperature equation (can't be 0).
+      Default for series 5 = 4094
+      Default for series 6 = 4096
+    $ref: /schemas/types.yaml#/definitions/uint32
+
+  moortec,ts-coeff-j:
+    description:
+      J coefficient for temperature equation.
+      Value should be multiplied by factor 1000.
+      Default for series 5 = -100
+      Default for series 6 = 0
+    $ref: /schemas/types.yaml#/definitions/int32
+
 required:
   - compatible
   - reg
@@ -97,5 +128,7 @@ examples:
         resets = <&rcu0 0x40 7>;
         moortec,vm-active-channels = <0x10 0x05>;
         moortec,vm-pre-scaler = <5 6>;
+        moortec,ts-coeff-g = <61400>;
+        moortec,ts-coeff-h = <253700>;
         #thermal-sensor-cells = <1>;
     };
Signature · 1 line
2.37.1
maintainer
#72 lore raw

Re: [PATCH v3 16/19] dt-bindings: hwmon: (mr75203) add coefficient properties for the thermal equation

Quoted 25 lines · Eliav Farber

Add optional temperature coefficient properties:
*) moortec,ts-coeff-g
*) moortec,ts-coeff-h
*) moortec,ts-coeff-cal5
*) moortec,ts-coeff-j
If defined they shall be used instead of defaults.

The coefficients were added to device tree on top of the series property (which can be used to select between series 5 and series 6), because coefficients can vary between product and product, and code defaults might not be accurate enough.

---
V3 -> V2:
- Add "moortec" prefix to property name.

 .../bindings/hwmon/moortec,mr75203.yaml       | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index ec2dbe7da9c2..a92da6064285 100644

Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml properties: · line 80

Quoted 7 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +4 -0
@@ -74,6 +74,37 @@ properties:
     default: 5
     $ref: /schemas/types.yaml#definitions/uint32
 
+  moortec,ts-coeff-g:
+    description:
+      G coefficient for temperature equation.
+      Value should be multiplied by factor 1000.

If you just multiply the values here, you can specify 'multipleOf: 1000'

Either way, some constraints would be nice. Or is 0 - 2^32 valid?

Quoted 5 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +5 -0
+      Default for series 5 = 60000
+      Default for series 6 = 57400
+    $ref: /schemas/types.yaml#/definitions/uint32
+
+  moortec,ts-coeff-h:

Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml properties: · line 95

Quoted 10 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +10 -0
+    description:
+      H coefficient for temperature equation.
+      Value should be multiplied by factor 1000.
+      Default for series 5 = 200000
+      Default for series 6 = 249400
+    $ref: /schemas/types.yaml#/definitions/uint32
+
+  moortec,ts-coeff-cal5:
+    description:
+      cal5 coefficient for temperature equation (can't be 0).

minimum: 1

Quoted 27 lines
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +14 -0
+      Default for series 5 = 4094
+      Default for series 6 = 4096
+    $ref: /schemas/types.yaml#/definitions/uint32
+
+  moortec,ts-coeff-j:
+    description:
+      J coefficient for temperature equation.
+      Value should be multiplied by factor 1000.
+      Default for series 5 = -100
+      Default for series 6 = 0
+    $ref: /schemas/types.yaml#/definitions/int32
+
 required:
   - compatible
   - reg
@@ -97,5 +128,7 @@ examples:
         resets = <&rcu0 0x40 7>;
         moortec,vm-active-channels = <0x10 0x05>;
         moortec,vm-pre-scaler = <5 6>;
+        moortec,ts-coeff-g = <61400>;
+        moortec,ts-coeff-h = <253700>;
         #thermal-sensor-cells = <1>;
     };
Signature · 1 line
2.37.1
#73 lore raw

Re: [PATCH v3 16/19] dt-bindings: hwmon: (mr75203) add coefficient properties for the thermal equation

Quoted 29 lines · Rob Herring
Quoted 28 lines · Eliav Farber

Add optional temperature coefficient properties:
 *) moortec,ts-coeff-g
 *) moortec,ts-coeff-h
 *) moortec,ts-coeff-cal5
 *) moortec,ts-coeff-j
If defined they shall be used instead of defaults.

The coefficients were added to device tree on top of the series property (which can be used to select between series 5 and series 6), because coefficients can vary between product and product, and code defaults might not be accurate enough.

---
V3 -> V2:
- Add "moortec" prefix to property name.

 .../bindings/hwmon/moortec,mr75203.yaml       | 33 +++++++++++++++++++

 1 file changed, 33 insertions(+)
Documentation/devicetree/bindings/hwmon/moortec,mr75203.yaml +0 -0
index ec2dbe7da9c2..a92da6064285 100644
@@ -74,6 +74,37 @@ properties:

     default: 5
     $ref: /schemas/types.yaml#definitions/uint32

Quoted 7 lines
Quoted 4 lines

+  moortec,ts-coeff-g:
+    description:
+      G coefficient for temperature equation.
+      Value should be multiplied by factor 1000.

If you just multiply the values here, you can specify 'multipleOf: 1000'

Fixed in v4.

Quoted 1 line

Either way, some constraints would be nice. Or is 0 - 2^32 valid?

g and h can't be 0, and since they must be multiple of 1000 I added:
minimum: 1000
j is negative or 0, so I added:
maximum: 0

Quoted 10 lines
Quoted 10 lines

+      Default for series 5 = 60000
+      Default for series 6 = 57400
+    $ref: /schemas/types.yaml#/definitions/uint32 +
+  moortec,ts-coeff-h:
+    description:
+      H coefficient for temperature equation. +      Value should be multiplied by factor 1000. +      Default for series 5 = 200000

Quoted 8 lines
Quoted 6 lines

+      Default for series 6 = 249400
+    $ref: /schemas/types.yaml#/definitions/uint32
+
+  moortec,ts-coeff-cal5:
+    description:
+      cal5 coefficient for temperature equation (can't be 0).

minimum: 1

Fixed in v4.

#74 lore raw

PATCH v3 17/19 hwmon: (mr75203) parse temperature coefficients from device-tree

In reply to Eliav Farber

Use thermal coefficients from the device tree if they exist. Otherwise, use default values according to the series (5 or 6). All coefficients can be used or only part of them.

The coefficients shall be used for fine tuning the default values since coefficients can vary between product and product.

---
 drivers/hwmon/mr75203.c | 29 ++++++++++++++++++++++++++++-
 1 file changed, 28 insertions(+), 1 deletion(-)
drivers/hwmon/mr75203.c +28 -1
index 6a035fd115ca..d9fc5d225868 100644
@@ -676,7 +676,8 @@ static int pvt_set_temp_coeff(struct device *dev, struct pvt_device *pvt)
 {
 	struct temp_coeff *ts_coeff = &pvt->ts_coeff;
 	const struct device_node *np = dev->of_node;
-	u32 series;
+	u32 series, coeff_h, coeff_g, coeff_cal5;
+	s32 coeff_j;
 	int ret;
 
 	/* Incase ts-series property is not defined, use default 5. */
@@ -702,6 +703,32 @@ static int pvt_set_temp_coeff(struct device *dev, struct pvt_device *pvt)
 
 	dev_dbg(dev, "temperature sensor series = %u\n", series);
 
+	/* Override ts-coeff-h/g/j/cal5 if they are defined. */
+	ret = of_property_read_u32(np, "moortec,ts-coeff-h", &coeff_h);
+	if (!ret)
+		ts_coeff->h = coeff_h;
+
+	ret = of_property_read_u32(np, "moortec,ts-coeff-g", &coeff_g);
+	if (!ret)
+		ts_coeff->g = coeff_g;
+
+	ret = of_property_read_s32(np, "moortec,ts-coeff-j", &coeff_j);
+	if (!ret)
+		ts_coeff->j = coeff_j;
+
+	ret = of_property_read_u32(np, "moortec,ts-coeff-cal5", &coeff_cal5);
+	if (!ret) {
+		if (coeff_cal5 == 0) {
+			dev_err(dev, "moortec,ts-coeff-cal5 can't be 0\n");
+			return -EINVAL;
+		}
+
+		ts_coeff->cal5 = coeff_cal5;
+	}
+
+	dev_dbg(dev, "ts-coeff: h = %u, g = %u, j = %d, cal5 = %u\n",
+		ts_coeff->h, ts_coeff->g, ts_coeff->j, ts_coeff->cal5);
+
 	return 0;
 }
 
Signature · 1 line
2.37.1
#75 lore raw

Re: [PATCH v3 17/19] hwmon: (mr75203) parse temperature coefficients from device-tree

Quoted 6 lines · Eliav Farber

Use thermal coefficients from the device tree if they exist. Otherwise, use default values according to the series (5 or 6). All coefficients can be used or only part of them.

The coefficients shall be used for fine tuning the default values since coefficients can vary between product and product.

...

drivers/hwmon/mr75203.c static int pvt_set_temp_coeff(struct device *dev, struct pvt_device *pvt) · line 707

Quoted 1 line
drivers/hwmon/mr75203.c +1 -0
+	ret = of_property_read_u32(np, "moortec,ts-coeff-h", &coeff_h);

of_ ?! Ditto for the rest.

Quoted 2 lines
drivers/hwmon/mr75203.c +2 -0
+	if (!ret)
+		ts_coeff->h = coeff_h;

...

drivers/hwmon/mr75203.c static int pvt_set_temp_coeff(struct device *dev, struct pvt_device *pvt) · line 717

Quoted 3 lines
drivers/hwmon/mr75203.c +3 -0
+	ret = of_property_read_s32(np, "moortec,ts-coeff-j", &coeff_j);
+	if (!ret)
+		ts_coeff->j = coeff_j;

You may avoid conditional:

	_property_read_s32(..., "moortec,ts-coeff-j", &ts_coeff->j);

...

drivers/hwmon/mr75203.c static int pvt_set_temp_coeff(struct device *dev, struct pvt_device *pvt) · line 724

Quoted 7 lines
drivers/hwmon/mr75203.c +6 -0
+	ret = of_property_read_u32(np, "moortec,ts-coeff-cal5", &coeff_cal5);
+	if (!ret) {
 
+		if (coeff_cal5 == 0) {
+			dev_err(dev, "moortec,ts-coeff-cal5 can't be 0\n");
+			return -EINVAL;
+		}

Code shouldn't be a YAML validator. Drop this and make sure you have correct DT schema.

drivers/hwmon/mr75203.c static int pvt_set_temp_coeff(struct device *dev, struct pvt_device *pvt) · line 727

Quoted 2 lines
drivers/hwmon/mr75203.c +2 -0
+		ts_coeff->cal5 = coeff_cal5;
+	}

Also see above.

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

Re: [PATCH v3 17/19] hwmon: (mr75203) parse temperature coefficients from device-tree

Quoted 5 lines · Andy Shevchenko
Quoted 4 lines · Eliav Farber

Use thermal coefficients from the device tree if they exist. Otherwise, use default values according to the series (5 or 6). All coefficients can be used or only part of them.

Quoted 8 lines
Quoted 2 lines

The coefficients shall be used for fine tuning the default values since coefficients can vary between product and product.

...

Quoted 1 line

+     ret = of_property_read_u32(np, "moortec,ts-coeff-h", &coeff_h);

of_ ?! Ditto for the rest.

Fixed for v4.
I replaced it with device_property_read_u32().

Quoted 5 lines
Quoted 2 lines

+     if (!ret)
+             ts_coeff->h = coeff_h;

...

Quoted 7 lines
Quoted 3 lines

+     ret = of_property_read_s32(np, "moortec,ts-coeff-j", &coeff_j); +     if (!ret)
+             ts_coeff->j = coeff_j;

You may avoid conditional:

       _property_read_s32(..., "moortec,ts-coeff-j", &ts_coeff->j);

Fixed for v4.
I removed the condition.

Quoted 6 lines

...

Quoted 3 lines

+     ret = of_property_read_u32(np, "moortec,ts-coeff-cal5", &coeff_cal5);
+     if (!ret) {

Quoted 8 lines
Quoted 5 lines

+             if (coeff_cal5 == 0) {
+                     dev_err(dev, "moortec,ts-coeff-cal5 can't be 0\n");
+                     return -EINVAL;
+             }

Code shouldn't be a YAML validator. Drop this and make sure you have correct DT schema.

Fixed for v4.
I dropped the validation check.
The YAML already mentions that it can't be 0.

Quoted 2 lines
Quoted 2 lines

+             ts_coeff->cal5 = coeff_cal5;
+     }

--
Thanks, Eliav

#77 lore raw

PATCH v3 18/19 hwmon: (mr75203) add debugfs to read and write temperature coefficients

In reply to Eliav Farber

This change adds debugfs to read and write temperature sensor coefficients
- g, h, j and cal5.

The coefficients can vary between product and product, so it can be very useful to be able to modify them on the fly during the calibration process.

e.g.:

cat /sys/kernel/debug/940f23d0000.pvt/ts_coeff_cal5 4096

echo 83000 > sys/kernel/debug/940f23d0000.pvt/ts_coeff_g

---
 drivers/hwmon/mr75203.c | 196 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 196 insertions(+)
drivers/hwmon/mr75203.c +196 -0
index d9fc5d225868..8c30f0521452 100644
@@ -9,6 +9,7 @@
  */
 #include <linux/bits.h>
 #include <linux/clk.h>
+#include <linux/debugfs.h>
 #include <linux/hwmon.h>
 #include <linux/module.h>
 #include <linux/mod_devicetable.h>
@@ -153,6 +154,7 @@ struct pvt_device {
 	struct regmap		*v_map;
 	struct clk		*clk;
 	struct reset_control	*rst;
+	struct dentry		*dbgfs_dir;
 	struct voltage_device	*vd;
 	struct voltage_channels	vm_channels;
 	struct temp_coeff	ts_coeff;
@@ -162,6 +164,198 @@ struct pvt_device {
 	u32			ip_freq;
 };
 
+static ssize_t pvt_ts_coeff_h_read(struct file *file,
+				   char __user *user_buf,
+				   size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	char buf[16];
+	unsigned int len;
+
+	len = sprintf(buf, "%u\n", pvt->ts_coeff.h);
+
+	return simple_read_from_buffer(user_buf, count, ppos, buf, len);
+}
+
+static ssize_t pvt_ts_coeff_h_write(struct file *file,
+				    const char __user *user_buf,
+				    size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	int ret;
+	u32 coeff;
+
+	ret = kstrtou32_from_user(user_buf, count, 0, &coeff);
+	if (ret)
+		return ret;
+
+	pvt->ts_coeff.h = coeff;
+
+	return count;
+}
+
+static const struct file_operations pvt_ts_coeff_h_fops = {
+	.read = pvt_ts_coeff_h_read,
+	.write = pvt_ts_coeff_h_write,
+	.open = simple_open,
+	.owner = THIS_MODULE,
+	.llseek = default_llseek,
+};
+
+static ssize_t pvt_ts_coeff_g_read(struct file *file,
+				   char __user *user_buf,
+				   size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	char buf[16];
+	unsigned int len;
+
+	len = sprintf(buf, "%u\n", pvt->ts_coeff.g);
+
+	return simple_read_from_buffer(user_buf, count, ppos, buf, len);
+}
+
+static ssize_t pvt_ts_coeff_g_write(struct file *file,
+				    const char __user *user_buf,
+				    size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	int ret;
+	u32 coeff;
+
+	ret = kstrtou32_from_user(user_buf, count, 0, &coeff);
+	if (ret)
+		return ret;
+
+	pvt->ts_coeff.g = coeff;
+
+	return count;
+}
+
+static const struct file_operations pvt_ts_coeff_g_fops = {
+	.read = pvt_ts_coeff_g_read,
+	.write = pvt_ts_coeff_g_write,
+	.open = simple_open,
+	.owner = THIS_MODULE,
+	.llseek = default_llseek,
+};
+
+static ssize_t pvt_ts_coeff_j_read(struct file *file,
+				   char __user *user_buf,
+				   size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	char buf[16];
+	unsigned int len;
+
+	len = sprintf(buf, "%d\n", pvt->ts_coeff.j);
+
+	return simple_read_from_buffer(user_buf, count, ppos, buf, len);
+}
+
+static ssize_t pvt_ts_coeff_j_write(struct file *file,
+				    const char __user *user_buf,
+				    size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	int ret;
+	s32 coeff;
+
+	ret = kstrtos32_from_user(user_buf, count, 0, &coeff);
+	if (ret)
+		return ret;
+
+	pvt->ts_coeff.j = coeff;
+
+	return count;
+}
+
+static const struct file_operations pvt_ts_coeff_j_fops = {
+	.read = pvt_ts_coeff_j_read,
+	.write = pvt_ts_coeff_j_write,
+	.open = simple_open,
+	.owner = THIS_MODULE,
+	.llseek = default_llseek,
+};
+
+static ssize_t pvt_ts_coeff_cal5_read(struct file *file,
+				      char __user *user_buf,
+				      size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	char buf[16];
+	unsigned int len;
+
+	len = sprintf(buf, "%u\n", pvt->ts_coeff.cal5);
+
+	return simple_read_from_buffer(user_buf, count, ppos, buf, len);
+}
+
+static ssize_t pvt_ts_coeff_cal5_write(struct file *file,
+				       const char __user *user_buf,
+				       size_t count, loff_t *ppos)
+{
+	struct pvt_device *pvt = file->private_data;
+	int ret;
+	u32 coeff;
+
+	ret = kstrtou32_from_user(user_buf, count, 0, &coeff);
+	if (ret)
+		return ret;
+
+	if (coeff == 0)
+		return -EINVAL;
+
+	pvt->ts_coeff.cal5 = coeff;
+
+	return count;
+}
+
+static const struct file_operations pvt_ts_coeff_cal5_fops = {
+	.read = pvt_ts_coeff_cal5_read,
+	.write = pvt_ts_coeff_cal5_write,
+	.open = simple_open,
+	.owner = THIS_MODULE,
+	.llseek = default_llseek,
+};
+
+static void devm_pvt_ts_dbgfs_remove(void *data)
+{
+	struct pvt_device *pvt = (struct pvt_device *)data;
+
+	debugfs_remove_recursive(pvt->dbgfs_dir);
+	pvt->dbgfs_dir = NULL;
+}
+
+static int pvt_ts_dbgfs_create(struct pvt_device *pvt, struct device *dev)
+{
+	int ret;
+
+	pvt->dbgfs_dir = debugfs_create_dir(dev_name(dev), NULL);
+	if (!pvt->dbgfs_dir) {
+		dev_err(dev, "Failed to create dbgfs_dir\n");
+		return -EINVAL;
+	}
+
+	debugfs_create_file("ts_coeff_h", 0644, pvt->dbgfs_dir, pvt,
+			    &pvt_ts_coeff_h_fops);
+	debugfs_create_file("ts_coeff_g", 0644, pvt->dbgfs_dir, pvt,
+			    &pvt_ts_coeff_g_fops);
+	debugfs_create_file("ts_coeff_j", 0644, pvt->dbgfs_dir, pvt,
+			    &pvt_ts_coeff_j_fops);
+	debugfs_create_file("ts_coeff_cal5", 0644, pvt->dbgfs_dir,  pvt,
+			    &pvt_ts_coeff_cal5_fops);
+
+	ret = devm_add_action_or_reset(dev, devm_pvt_ts_dbgfs_remove, pvt);
+	if (ret) {
+		dev_err(dev, "failed to add action to remove pvt dbgfs (%d)\n",
+			ret);
+		return ret;
+	}
+
+	return 0;
+}
+
 static umode_t pvt_is_visible(const void *data, enum hwmon_sensor_types type,
 			      u32 attr, int channel)
 {
@@ -814,6 +1008,8 @@ static int mr75203_probe(struct platform_device *pdev)
 		memset32(temp_config, HWMON_T_INPUT, ts_num);
 		pvt_temp.config = temp_config;
 		pvt_info[index++] = &pvt_temp;
+
+		pvt_ts_dbgfs_create(pvt, dev);
 	}
 
 	if (pd_num) {
Signature · 1 line
2.37.1
#78 lore raw

Re: [PATCH v3 18/19] hwmon: (mr75203) add debugfs to read and write temperature coefficients

Quoted 13 lines · Eliav Farber

This change adds debugfs to read and write temperature sensor coefficients
- g, h, j and cal5.

The coefficients can vary between product and product, so it can be very useful to be able to modify them on the fly during the calibration process.

e.g.:

cat /sys/kernel/debug/940f23d0000.pvt/ts_coeff_cal5 4096

echo 83000 > sys/kernel/debug/940f23d0000.pvt/ts_coeff_g

...

drivers/hwmon/mr75203.c struct pvt_device { · line 338

Quoted 6 lines
drivers/hwmon/mr75203.c +5 -0
+	pvt->dbgfs_dir = debugfs_create_dir(dev_name(dev), NULL);
 
+	if (!pvt->dbgfs_dir) {
+		dev_err(dev, "Failed to create dbgfs_dir\n");
+		return -EINVAL;
+	}

No, just don't check the return value of debugfs API calls.

drivers/hwmon/mr75203.c struct pvt_device { · line 347

Quoted 8 lines
drivers/hwmon/mr75203.c +8 -0
+	debugfs_create_file("ts_coeff_h", 0644, pvt->dbgfs_dir, pvt,
+			    &pvt_ts_coeff_h_fops);
+	debugfs_create_file("ts_coeff_g", 0644, pvt->dbgfs_dir, pvt,
+			    &pvt_ts_coeff_g_fops);
+	debugfs_create_file("ts_coeff_j", 0644, pvt->dbgfs_dir, pvt,
+			    &pvt_ts_coeff_j_fops);
+	debugfs_create_file("ts_coeff_cal5", 0644, pvt->dbgfs_dir,  pvt,
+			    &pvt_ts_coeff_cal5_fops);

debugfs has helpers for POD types, use them and shrink your code by ~80%.

...

drivers/hwmon/mr75203.c struct pvt_device { · line 356

Quoted 8 lines
drivers/hwmon/mr75203.c +8 -0
+	ret = devm_add_action_or_reset(dev, devm_pvt_ts_dbgfs_remove, pvt);
+	if (ret) {
+		dev_err(dev, "failed to add action to remove pvt dbgfs (%d)\n",
+			ret);
+		return ret;
+	}
+
+	return 0;

return devm_add_...

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

Re: [PATCH v3 18/19] hwmon: (mr75203) add debugfs to read and write temperature coefficients

Quoted 17 lines · Andy Shevchenko
Quoted 13 lines · Eliav Farber

This change adds debugfs to read and write temperature sensor coefficients
- g, h, j and cal5.

The coefficients can vary between product and product, so it can be very useful to be able to modify them on the fly during the calibration process.

e.g.:

cat /sys/kernel/debug/940f23d0000.pvt/ts_coeff_cal5 4096

echo 83000 > sys/kernel/debug/940f23d0000.pvt/ts_coeff_g

...

Quoted 9 lines
Quoted 6 lines

+     pvt->dbgfs_dir = debugfs_create_dir(dev_name(dev), NULL);

+     if (!pvt->dbgfs_dir) {
+             dev_err(dev, "Failed to create dbgfs_dir\n"); +             return -EINVAL;
+     }

No, just don't check the return value of debugfs API calls.

Do you mean that I should just do:
debugfs_create_dir(dev_name(dev), NULL);
Can you please explain why it's OK to ignore the return value?

Quoted 11 lines
Quoted 8 lines

+     debugfs_create_file("ts_coeff_h", 0644, pvt->dbgfs_dir, pvt, +                         &pvt_ts_coeff_h_fops);
+     debugfs_create_file("ts_coeff_g", 0644, pvt->dbgfs_dir, pvt, +                         &pvt_ts_coeff_g_fops);
+     debugfs_create_file("ts_coeff_j", 0644, pvt->dbgfs_dir, pvt, +                         &pvt_ts_coeff_j_fops);
+     debugfs_create_file("ts_coeff_cal5", 0644, pvt->dbgfs_dir,  pvt, +                         &pvt_ts_coeff_cal5_fops);

debugfs has helpers for POD types, use them and shrink your code by ~80%.

Fixed for v4.
I used debugfs_create_u32() for ts_coeff_h, ts_coeff_g and ts_coeff_j. For ts_coeff_cal5 I still use debugfs_create_file() because I must make sure it is not set to 0.

Quoted 11 lines
Quoted 10 lines

+     ret = devm_add_action_or_reset(dev, devm_pvt_ts_dbgfs_remove, pvt);
+     if (ret) {
+             dev_err(dev, "failed to add action to remove pvt dbgfs (%d)\n",
+                     ret);
+             return ret;
+     }
+
+     return 0;

Quoted 1 line

return devm_add_...

Fixed for v4.

--
Thanks, Eliav

maintainer
#80 lore raw

Re: [PATCH v3 18/19] hwmon: (mr75203) add debugfs to read and write temperature coefficients

Quoted 26 lines · Eliav
Quoted 25 lines · Andy Shevchenko
Quoted 14 lines · Eliav Farber

This change adds debugfs to read and write temperature sensor coefficients
- g, h, j and cal5.

The coefficients can vary between product and product, so it can be very useful to be able to modify them on the fly during the calibration process.

e.g.:

cat /sys/kernel/debug/940f23d0000.pvt/ts_coeff_cal5 4096

echo 83000 > sys/kernel/debug/940f23d0000.pvt/ts_coeff_g

...

Quoted 6 lines

+     pvt->dbgfs_dir = debugfs_create_dir(dev_name(dev), NULL);

+     if (!pvt->dbgfs_dir) {
+             dev_err(dev, "Failed to create dbgfs_dir\n"); +             return -EINVAL;
+     }

Quoted 6 lines
Quoted 2 lines

No, just don't check the return value of debugfs API calls.

Do you mean that I should just do:
debugfs_create_dir(dev_name(dev), NULL);
Can you please explain why it's OK to ignore the return value?

Because that is how debugfs code is supposed to be implemented. Other debugfs code checks if the dir parameter passed to it is NULL and do nothing if it is. This reduces code size overall by avoiding unnecessary error checks.

Guenter

#81 lore raw

Re: [PATCH v3 18/19] hwmon: (mr75203) add debugfs to read and write temperature coefficients

In reply to Farber, Eliav
Quoted 2 lines · Eliav
Quoted 1 line · Andy Shevchenko

On Tue, Aug 30, 2022 at 07:22:11PM +0000, Eliav Farber wrote:

...

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

+     pvt->dbgfs_dir = debugfs_create_dir(dev_name(dev), NULL);

+     if (!pvt->dbgfs_dir) {
+             dev_err(dev, "Failed to create dbgfs_dir\n"); +             return -EINVAL;
+     }

Quoted 5 lines
Quoted 2 lines

No, just don't check the return value of debugfs API calls.

Do you mean that I should just do:
debugfs_create_dir(dev_name(dev), NULL);
Can you please explain why it's OK to ignore the return value?

Author of the debugfs is speaking:
https://patchwork.kernel.org/project/linux-wireless/patch/[email protected]/

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

PATCH v3 19/19 hwmon: (mr75203) fix coding style space errors

In reply to Eliav Farber

Fix: "ERROR: space required before the open parenthesis '('"

---
 drivers/hwmon/mr75203.c | 40 ++++++++++++++++++++--------------------
 1 file changed, 20 insertions(+), 20 deletions(-)
drivers/hwmon/mr75203.c +20 -20
index 8c30f0521452..6a19f3d2a708 100644
@@ -404,7 +404,7 @@ static int pvt_read_temp(struct device *dev, u32 attr, int channel, long *val)
 			return ret;
 
 		ret = regmap_read(t_map, SDIF_DATA(channel), &nbs);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		nbs &= SAMPLE_DATA_MSK;
@@ -445,7 +445,7 @@ static int pvt_read_in(struct device *dev, u32 attr, int channel, long *val)
 			return ret;
 
 		ret = regmap_read(v_map, VM_SDIF_DATA(vm_idx, ch_idx), &n);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		n &= SAMPLE_DATA_MSK;
@@ -542,19 +542,19 @@ static int pvt_init(struct pvt_device *pvt)
 
 	if (t_num) {
 		ret = regmap_write(t_map, SDIF_SMPL_CTRL, 0x0);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(t_map, SDIF_HALT, 0x0);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(t_map, CLK_SYNTH, clk_synth);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(t_map, SDIF_DISABLE, 0x0);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_read_poll_timeout(t_map, SDIF_STAT,
@@ -567,7 +567,7 @@ static int pvt_init(struct pvt_device *pvt)
 		val = CFG0_MODE_2 | CFG0_PARALLEL_OUT | CFG0_12_BIT |
 		      IP_CFG << SDIF_ADDR_SFT | SDIF_WRN_W | SDIF_PROG;
 		ret = regmap_write(t_map, SDIF_W, val);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_read_poll_timeout(t_map, SDIF_STAT,
@@ -580,7 +580,7 @@ static int pvt_init(struct pvt_device *pvt)
 		val = POWER_DELAY_CYCLE_256 | IP_TMR << SDIF_ADDR_SFT |
 			      SDIF_WRN_W | SDIF_PROG;
 		ret = regmap_write(t_map, SDIF_W, val);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_read_poll_timeout(t_map, SDIF_STAT,
@@ -594,39 +594,39 @@ static int pvt_init(struct pvt_device *pvt)
 		      IP_CTRL << SDIF_ADDR_SFT |
 		      SDIF_WRN_W | SDIF_PROG;
 		ret = regmap_write(t_map, SDIF_W, val);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 	}
 
 	if (p_num) {
 		ret = regmap_write(p_map, SDIF_HALT, 0x0);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(p_map, SDIF_DISABLE, BIT(p_num) - 1);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(p_map, CLK_SYNTH, clk_synth);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 	}
 
 	if (v_num) {
 		ret = regmap_write(v_map, SDIF_SMPL_CTRL, 0x0);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(v_map, SDIF_HALT, 0x0);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(v_map, CLK_SYNTH, clk_synth);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_write(v_map, SDIF_DISABLE, 0x0);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_read_poll_timeout(v_map, SDIF_STAT,
@@ -654,7 +654,7 @@ static int pvt_init(struct pvt_device *pvt)
 		      CFG1_14_BIT | IP_CFG << SDIF_ADDR_SFT |
 		      SDIF_WRN_W | SDIF_PROG;
 		ret = regmap_write(v_map, SDIF_W, val);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_read_poll_timeout(v_map, SDIF_STAT,
@@ -667,7 +667,7 @@ static int pvt_init(struct pvt_device *pvt)
 		val = POWER_DELAY_CYCLE_64 | IP_TMR << SDIF_ADDR_SFT |
 		      SDIF_WRN_W | SDIF_PROG;
 		ret = regmap_write(v_map, SDIF_W, val);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 
 		ret = regmap_read_poll_timeout(v_map, SDIF_STAT,
@@ -681,7 +681,7 @@ static int pvt_init(struct pvt_device *pvt)
 		      IP_CTRL << SDIF_ADDR_SFT |
 		      SDIF_WRN_W | SDIF_PROG;
 		ret = regmap_write(v_map, SDIF_W, val);
-		if(ret < 0)
+		if (ret < 0)
 			return ret;
 	}
 
@@ -967,7 +967,7 @@ static int mr75203_probe(struct platform_device *pdev)
 	}
 
 	ret = regmap_read(pvt->c_map, PVT_IP_CONFIG, &val);
-	if(ret < 0)
+	if (ret < 0)
 		return ret;
 
 	ts_num = (val & TS_NUM_MSK) >> TS_NUM_SFT;
Signature · 1 line
2.37.1
#83 lore raw

Re: [PATCH v3 19/19] hwmon: (mr75203) fix coding style space errors

Quoted 1 line · Eliav Farber

Fix: "ERROR: space required before the open parenthesis '('"

This patch may have other fixes like adding new blank lines (noted in one of the patches in the series), etc.

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

Re: [PATCH v3 19/19] hwmon: (mr75203) fix coding style space errors

Quoted 5 lines · Andy Shevchenko
Quoted 1 line · Eliav Farber

Fix: "ERROR: space required before the open parenthesis '('"

This patch may have other fixes like adding new blank lines (noted in one of the patches in the series), etc.

This patch fixed a specific space error which existed before my changes and repeated many time.
I fixed the blank line I added a previous patch (but is it isn’t an error reported by checkpatch).

--
Regards, Eliav

maintainer
#85 lore raw

Re: [PATCH v3 19/19] hwmon: (mr75203) fix coding style space errors

Quoted 10 lines · Eliav
Quoted 5 lines · Andy Shevchenko
Quoted 1 line · Eliav Farber

Fix: "ERROR: space required before the open parenthesis '('"

This patch may have other fixes like adding new blank lines (noted in one of the patches in the series), etc.

This patch fixed a specific space error which existed before my changes and repeated many time.
I fixed the blank line I added a previous patch (but is it isn’t an error reported by checkpatch).

That should really be fixed where it was introduced, not be introduced and fixed here.

Guenter

#86 lore raw

Re: [PATCH v3 19/19] hwmon: (mr75203) fix coding style space errors

Quoted 7 lines · Guenter Roeck
Quoted 6 lines · Eliav
Quoted 4 lines · Andy Shevchenko
Quoted 1 line · Eliav Farber

Fix: "ERROR: space required before the open parenthesis '('"

This patch may have other fixes like adding new blank lines (noted

in one

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

of the patches in the series), etc.

This patch fixed a specific space error which existed before my changes and repeated many time.
I fixed the blank line I added a previous patch (but is it isn’t an error reported by checkpatch).

That should really be fixed where it was introduced, not be introduced and fixed here.

So what do you suggest?
I can drop the patch from this series and ignore it or move it to be the first patch in the series, or publish it separately later on. I had it because it was annoying seeing existing checkpatch errors when I came to check my change.

--
Thanks, Eliav

maintainer
#87 lore raw

Re: [PATCH v3 19/19] hwmon: (mr75203) fix coding style space errors

Quoted 17 lines · Eliav
Quoted 7 lines · Guenter Roeck
Quoted 6 lines · Eliav
Quoted 5 lines · Andy Shevchenko
Quoted 1 line · Eliav Farber

Fix: "ERROR: space required before the open parenthesis '('"

This patch may have other fixes like adding new blank lines (noted in one of the patches in the series), etc.

Quoted 7 lines
Quoted 4 lines

This patch fixed a specific space error which existed before my changes and repeated many time.
I fixed the blank line I added a previous patch (but is it isn’t an error reported by checkpatch).

That should really be fixed where it was introduced, not be introduced and fixed here.

Quoted 6 lines

So what do you suggest?
I can drop the patch from this series and ignore it or move it to be the first patch in the series, or publish it separately later on. I had it because it was annoying seeing existing checkpatch errors when I came to check my change.

Sorry, you lost me. I referred to "I fixed the blank line I added a previous patch". That should not be fixed in this patch but be dropped from the patch where you introduced it. Did I misunderstand your comment ?

Guenter

#88 lore raw

Re: [PATCH v3 19/19] hwmon: (mr75203) fix coding style space errors

Quoted 26 lines · Guenter Roeck
Quoted 18 lines · Eliav
Quoted 7 lines · Guenter Roeck
Quoted 6 lines · Eliav
Quoted 4 lines · Andy Shevchenko
Quoted 1 line · Eliav Farber

Fix: "ERROR: space required before the open parenthesis '('"

This patch may have other fixes like adding new blank lines

(noted in one

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

of the patches in the series), etc.

This patch fixed a specific space error which existed before my changes and repeated many time.
I fixed the blank line I added a previous patch (but is it isn’t an error reported by checkpatch).

That should really be fixed where it was introduced, not be introduced and fixed here.

Quoted 6 lines

So what do you suggest?
I can drop the patch from this series and ignore it or move it to be the first patch in the series, or publish it separately later on. I had it because it was annoying seeing existing checkpatch errors when I came to check my change.

Quoted 3 lines

Sorry, you lost me. I referred to "I fixed the blank line I added a previous patch". That should not be fixed in this patch but be dropped from the patch where you introduced it. Did I misunderstand your comment ?

Indeed I fixed the blank line in the original patch in which it was introduced.
Patch #19 only addresses previously existing space related checkpatch errors.

--
Thanks, Eliav