Skip to contentSkip to conversation

adm1026: fix setting fan_div (2nd try)

PATCH In review Gabriele Gorla 3 participants 5 messages Dec 1, 2010 – Dec 2, 2010 Hardware monitoring

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 adm1026: fix setting fan_div (2nd try)

Prevent setting fan_div from stomping on other fans that share the same i2c register.

Signed-off-by Gabriele Gorla <gorlik at penguintown.net>

---

drivers/hwmon/adm1026.c +16 -14
diff -r -U 5 linux-source-2.6.26_orig/drivers/hwmon/adm1026.c linux-source-2.6.26/drivers/hwmon/adm1026.c
@@ -918,34 +918,36 @@
 {
 	struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
 	int nr = sensor_attr->index;
 	struct i2c_client *client = to_i2c_client(dev);
 	struct adm1026_data *data = i2c_get_clientdata(client);
-	int val, orig_div, new_div, shift;
+	int val, orig_div, new_div;
 
 	val = simple_strtol(buf, NULL, 10);
 	new_div = DIV_TO_REG(val);
 	if (new_div = 0) {
 		return -EINVAL;
 	}
 	mutex_lock(&data->update_lock);
 	orig_div = data->fan_div[nr];
-	data->fan_div[nr] = DIV_FROM_REG(new_div);
+	if (orig_div != DIV_FROM_REG(new_div)) {
+		data->fan_div[nr] = DIV_FROM_REG(new_div);
 
-	if (nr < 4) { /* 0 <= nr < 4 */
-		shift = 2 * nr;
-		adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
-			((DIV_TO_REG(orig_div) & (~(0x03 << shift))) |
-			(new_div << shift)));
-	} else { /* 3 < nr < 8 */
-		shift = 2 * (nr - 4);
-		adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
-			((DIV_TO_REG(orig_div) & (~(0x03 << (2 * shift)))) |
-			(new_div << shift)));
-	}
+		if (nr < 4) { /* 0 <= nr < 4 */
+			adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
+				    (DIV_TO_REG(data->fan_div[0]) << 0) |
+				    (DIV_TO_REG(data->fan_div[1]) << 2) |
+				    (DIV_TO_REG(data->fan_div[2]) << 4) |
+				    (DIV_TO_REG(data->fan_div[3]) << 6));
+		} else { /* 3 < nr < 8 */
+			adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
+				    (DIV_TO_REG(data->fan_div[4]) << 0) |
+				    (DIV_TO_REG(data->fan_div[5]) << 2) |
+				    (DIV_TO_REG(data->fan_div[6]) << 4) |
+				    (DIV_TO_REG(data->fan_div[7]) << 6));
+		}
 
-	if (data->fan_div[nr] != orig_div) {
 		fixup_fan_min(dev, nr, orig_div);
 	}
 	mutex_unlock(&data->update_lock);
 	return count;
 }

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

#1 lore raw

ACK

Quoted 68 lines · Gabriele Gorla

Prevent setting fan_div from stomping on other fans that share the same i2c register.

Signed-off-by Gabriele Gorla <gorlik at penguintown.net>

---

diff -r -U 5 linux-source-2.6.26_orig/drivers/hwmon/adm1026.c linux-source-2.6.26/drivers/hwmon/adm1026.c
--- linux-source-2.6.26_orig/drivers/hwmon/adm1026.c 2008-07-13
14:51:29.000000000 -0700
+++ linux-source-2.6.26/drivers/hwmon/adm1026.c 2010-11-30 16:35:27.000000000 -0800

drivers/hwmon/adm1026.c +3 -11
@@ -918,34 +918,36 @@
 {
       struct sensor_device_attribute *sensor_attr =
 to_sensor_dev_attr(attr);
       int nr = sensor_attr->index;
       struct i2c_client *client = to_i2c_client(dev);
       struct adm1026_data *data = i2c_get_clientdata(client);
-       int val, orig_div, new_div, shift;
+       int val, orig_div, new_div;
 
       val = simple_strtol(buf, NULL, 10);
       new_div = DIV_TO_REG(val);
       if (new_div == 0) {
               return -EINVAL;
       }
       mutex_lock(&data->update_lock);
       orig_div = data->fan_div[nr];
-       data->fan_div[nr] = DIV_FROM_REG(new_div);
+       if (orig_div != DIV_FROM_REG(new_div)) {
+               data->fan_div[nr] = DIV_FROM_REG(new_div);
 
-       if (nr < 4) { /* 0 <= nr < 4 */
-               shift = 2 * nr;
-               adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
-                       ((DIV_TO_REG(orig_div) & (~(0x03 << shift))) |
-                       (new_div << shift)));
-       } else { /* 3 < nr < 8 */
-               shift = 2 * (nr - 4);
-               adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
-                       ((DIV_TO_REG(orig_div) & (~(0x03 << (2 * shift))))

|

drivers/hwmon/adm1026.c +2 -2
-                       (new_div << shift)));
-       }
+               if (nr < 4) { /* 0 <= nr < 4 */
+                       adm1026_write_value(client,

ADM1026_REG_FAN_DIV_0_3,

drivers/hwmon/adm1026.c +6 -0
+                                   (DIV_TO_REG(data->fan_div[0]) << 0) |
+                                   (DIV_TO_REG(data->fan_div[1]) << 2) |
+                                   (DIV_TO_REG(data->fan_div[2]) << 4) |
+                                   (DIV_TO_REG(data->fan_div[3]) << 6));
+               } else { /* 3 < nr < 8 */
+                       adm1026_write_value(client,

ADM1026_REG_FAN_DIV_4_7,

drivers/hwmon/adm1026.c +5 -1
+                                   (DIV_TO_REG(data->fan_div[4]) << 0) |
+                                   (DIV_TO_REG(data->fan_div[5]) << 2) |
+                                   (DIV_TO_REG(data->fan_div[6]) << 4) |
+                                   (DIV_TO_REG(data->fan_div[7]) << 6));
+               }
 
-       if (data->fan_div[nr] != orig_div) {
               fixup_fan_min(dev, nr, orig_div);
       }
       mutex_unlock(&data->update_lock);
       return count;
 }
Quoted 5 lines

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

#2 lore raw
In reply to Gabriele Gorla

Hi Gabriele,

Quoted 49 lines · Gabriele Gorla

Prevent setting fan_div from stomping on other fans that share the same i2c register.

Signed-off-by Gabriele Gorla <gorlik at penguintown.net>

---

diff -r -U 5 linux-source-2.6.26_orig/drivers/hwmon/adm1026.c linux-source-2.6.26/drivers/hwmon/adm1026.c

drivers/hwmon/adm1026.c +13 -13
@@ -918,34 +918,36 @@
 {
 	struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
 	int nr = sensor_attr->index;
 	struct i2c_client *client = to_i2c_client(dev);
 	struct adm1026_data *data = i2c_get_clientdata(client);
-	int val, orig_div, new_div, shift;
+	int val, orig_div, new_div;
 
 	val = simple_strtol(buf, NULL, 10);
 	new_div = DIV_TO_REG(val);
 	if (new_div = 0) {
 		return -EINVAL;
 	}
 	mutex_lock(&data->update_lock);
 	orig_div = data->fan_div[nr];
-	data->fan_div[nr] = DIV_FROM_REG(new_div);
+	if (orig_div != DIV_FROM_REG(new_div)) {
+		data->fan_div[nr] = DIV_FROM_REG(new_div);
 
-	if (nr < 4) { /* 0 <= nr < 4 */
-		shift = 2 * nr;
-		adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
-			((DIV_TO_REG(orig_div) & (~(0x03 << shift))) |
-			(new_div << shift)));
-	} else { /* 3 < nr < 8 */
-		shift = 2 * (nr - 4);
-		adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
-			((DIV_TO_REG(orig_div) & (~(0x03 << (2 * shift)))) |
-			(new_div << shift)));
-	}
+		if (nr < 4) { /* 0 <= nr < 4 */
+			adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
+				    (DIV_TO_REG(data->fan_div[0]) << 0) |
+				    (DIV_TO_REG(data->fan_div[1]) << 2) |
+				    (DIV_TO_REG(data->fan_div[2]) << 4) |
+				    (DIV_TO_REG(data->fan_div[3]) << 6));
+		} else { /* 3 < nr < 8 */
+			adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
+				    (DIV_TO_REG(data->fan_div[4]) << 0) |
+				    (DIV_TO_REG(data->fan_div[5]) << 2) |

drivers/hwmon/adm1026.c line 953

Quoted 10 lines
drivers/hwmon/adm1026.c +3 -1
+				    (DIV_TO_REG(data->fan_div[6]) << 4) |
+				    (DIV_TO_REG(data->fan_div[7]) << 6));
+		}
 
-	if (data->fan_div[nr] != orig_div) {
 		fixup_fan_min(dev, nr, orig_div);
 	}
 	mutex_unlock(&data->update_lock);
 	return count;
 }

Your patch is correct. However, in order to get the fix in stable kernel series, I would prefer an even more minimalist fix, only addressing the actual bug. The two minor optimizations (skipping the register write if the value is the same and skipping the call to fixup_fan_min if the value is different but resulted in the same register encoding) would rather be left out at this time.

This would result in the following patch:

---
 drivers/hwmon/adm1026.c |   16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)
drivers/hwmon/adm1026.c +9 -7
@@ -916,7 +916,7 @@ static ssize_t set_fan_div(struct device
 	int nr = sensor_attr->index;
 	struct i2c_client *client = to_i2c_client(dev);
 	struct adm1026_data *data = i2c_get_clientdata(client);
-	int val, orig_div, new_div, shift;
+	int val, orig_div, new_div;
 
 	val = simple_strtol(buf, NULL, 10);
 	new_div = DIV_TO_REG(val);
@@ -928,15 +928,17 @@ static ssize_t set_fan_div(struct device
 	data->fan_div[nr] = DIV_FROM_REG(new_div);
 
 	if (nr < 4) { /* 0 <= nr < 4 */
-		shift = 2 * nr;
 		adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
-			((DIV_TO_REG(orig_div) & (~(0x03 << shift))) |
-			(new_div << shift)));
+				    (DIV_TO_REG(data->fan_div[0]) << 0) |
+				    (DIV_TO_REG(data->fan_div[1]) << 2) |
+				    (DIV_TO_REG(data->fan_div[2]) << 4) |
+				    (DIV_TO_REG(data->fan_div[3]) << 6));
 	} else { /* 3 < nr < 8 */
-		shift = 2 * (nr - 4);
 		adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
-			((DIV_TO_REG(orig_div) & (~(0x03 << (2 * shift)))) |
-			(new_div << shift)));
+				    (DIV_TO_REG(data->fan_div[4]) << 0) |
+				    (DIV_TO_REG(data->fan_div[5]) << 2) |
+				    (DIV_TO_REG(data->fan_div[6]) << 4) |
+				    (DIV_TO_REG(data->fan_div[7]) << 6));
 	}
 
 	if (data->fan_div[nr] != orig_div) {

As you can see, it's more compact and immediately obviously correct. So backporting it to kernels 2.6.36, 2.6.32, 2.6.27 and maybe even 2.6.16 will be trivial. Is this OK with you? You retain patch authorship, of course.

Regarding the optimizations, I'm not even sure if we need them.

The first optimization (skipping the register write if the value is the same) looks nice but OTOH it is not something hwmon drivers are supposed to do. Drivers should do what the user tells them, so if the user says "write divider value 4 to the chip", it should just do so. If you look at all other similar functions in the driver (e.g. set_fan_min) you'll see it doesn't do any test and unconditionally writes the value. As Guenter underlined before, "set" operations are rare enough that there is little point in optimizing them.

As for the second optimization (skipping the call to fixup_fan_min if the value is different but resulted in the same register encoding), if we do strict checking of input values, it will become a no-op because the cases it optimizes will no longer possibly happen.

Thanks,

Signature · 6 lines
Jean Delvare

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
#3 lore raw
In reply to Gabriele Gorla
Quoted 49 lines · Jean Delvare

Your patch is correct. However, in order to get the fix in stable kernel series, I would prefer an even more minimalist fix, only addressing the actual bug. The two minor optimizations (skipping the register write if the value is the same and skipping the call to fixup_fan_min if the value is different but resulted in the same register encoding) would rather be left out at this time.

This would result in the following patch:

---
 drivers/hwmon/adm1026.c |   16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)
drivers/hwmon/adm1026.c +9 -7
@@ -916,7 +916,7 @@ static ssize_t set_fan_div(struct device
 	int nr = sensor_attr->index;
 	struct i2c_client *client = to_i2c_client(dev);
 	struct adm1026_data *data = i2c_get_clientdata(client);
-	int val, orig_div, new_div, shift;
+	int val, orig_div, new_div;
 
 	val = simple_strtol(buf, NULL, 10);
 	new_div = DIV_TO_REG(val);
@@ -928,15 +928,17 @@ static ssize_t set_fan_div(struct device
 	data->fan_div[nr] = DIV_FROM_REG(new_div);
 
 	if (nr < 4) { /* 0 <= nr < 4 */
-		shift = 2 * nr;
 		adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
-			((DIV_TO_REG(orig_div) & (~(0x03 << shift))) |
-			(new_div << shift)));
+				    (DIV_TO_REG(data->fan_div[0]) << 0) |
+				    (DIV_TO_REG(data->fan_div[1]) << 2) |
+				    (DIV_TO_REG(data->fan_div[2]) << 4) |
+				    (DIV_TO_REG(data->fan_div[3]) << 6));
 	} else { /* 3 < nr < 8 */
-		shift = 2 * (nr - 4);
 		adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
-			((DIV_TO_REG(orig_div) & (~(0x03 << (2 * shift)))) |
-			(new_div << shift)));
+				    (DIV_TO_REG(data->fan_div[4]) << 0) |
+				    (DIV_TO_REG(data->fan_div[5]) << 2) |
+				    (DIV_TO_REG(data->fan_div[6]) << 4) |
+				    (DIV_TO_REG(data->fan_div[7]) << 6));
 	}
 
 	if (data->fan_div[nr] != orig_div) {
Quoted 4 lines

As you can see, it's more compact and immediately obviously correct. So backporting it to kernels 2.6.36, 2.6.32, 2.6.27 and maybe even 2.6.16 will be trivial. Is this OK with you? You retain patch authorship, of course.

Sounds good to me. Do I need to submit it again, or you will take care of the re-submission?

Quoted 1 line

Regarding the optimizations, I'm not even sure if we need them.

We could debate this issues for weeks but I am sure we all have better things to do than trying to overoptimize a non critical code path. :-)

thanks,
GG

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

#4 lore raw
In reply to Gabriele Gorla

Hi Gabriele,

Quoted 51 lines · Gabriele Gorla
Quoted 49 lines · Jean Delvare

Your patch is correct. However, in order to get the fix in stable kernel series, I would prefer an even more minimalist fix, only addressing the actual bug. The two minor optimizations (skipping the register write if the value is the same and skipping the call to fixup_fan_min if the value is different but resulted in the same register encoding) would rather be left out at this time.

This would result in the following patch:

---
 drivers/hwmon/adm1026.c |   16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)
drivers/hwmon/adm1026.c +9 -7
@@ -916,7 +916,7 @@ static ssize_t set_fan_div(struct device
 	int nr = sensor_attr->index;
 	struct i2c_client *client = to_i2c_client(dev);
 	struct adm1026_data *data = i2c_get_clientdata(client);
-	int val, orig_div, new_div, shift;
+	int val, orig_div, new_div;
 
 	val = simple_strtol(buf, NULL, 10);
 	new_div = DIV_TO_REG(val);
@@ -928,15 +928,17 @@ static ssize_t set_fan_div(struct device
 	data->fan_div[nr] = DIV_FROM_REG(new_div);
 
 	if (nr < 4) { /* 0 <= nr < 4 */
-		shift = 2 * nr;
 		adm1026_write_value(client, ADM1026_REG_FAN_DIV_0_3,
-			((DIV_TO_REG(orig_div) & (~(0x03 << shift))) |
-			(new_div << shift)));
+				    (DIV_TO_REG(data->fan_div[0]) << 0) |
+				    (DIV_TO_REG(data->fan_div[1]) << 2) |
+				    (DIV_TO_REG(data->fan_div[2]) << 4) |
+				    (DIV_TO_REG(data->fan_div[3]) << 6));
 	} else { /* 3 < nr < 8 */
-		shift = 2 * (nr - 4);
 		adm1026_write_value(client, ADM1026_REG_FAN_DIV_4_7,
-			((DIV_TO_REG(orig_div) & (~(0x03 << (2 * shift)))) |
-			(new_div << shift)));
+				    (DIV_TO_REG(data->fan_div[4]) << 0) |
+				    (DIV_TO_REG(data->fan_div[5]) << 2) |
+				    (DIV_TO_REG(data->fan_div[6]) << 4) |
+				    (DIV_TO_REG(data->fan_div[7]) << 6));
 	}
 
 	if (data->fan_div[nr] != orig_div) {
Quoted 7 lines
Quoted 4 lines

As you can see, it's more compact and immediately obviously correct. So backporting it to kernels 2.6.36, 2.6.32, 2.6.27 and maybe even 2.6.16 will be trivial. Is this OK with you? You retain patch authorship, of course.

Sounds good to me. Do I need to submit it again, or you will take care of the re-submission?

I've done it:
ftp://ftp.kernel.org/pub/linux/kernel/people/jdelvare/linux-2.6/jdelvare-hwmon/hwmon-adm1026-01-fix-setting-fan_div.patch

Quoted 5 lines
Quoted 2 lines

Regarding the optimizations, I'm not even sure if we need them.

We could debate this issues for weeks but I am sure we all have better things to do than trying to overoptimize a non critical code path. :-)

Correct.

Signature · 6 lines
Jean Delvare

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors