[lm-sensors] [PATCH v3 7/8] sensord: Evaluate return value of

[lm-sensors] [PATCH v3 7/8] sensord: Evaluate return value of

[Andre Prendel]

Evaluate return value of idChip().

Some of the functions called from idChip() can fail which means
idChip() can return an error. Callers of idChip() should be aware of
this.
---
 prog/sensord/sense.c |   24 +++++++++++++++++++-----
 1 file changed, 19 insertions(+), 5 deletions(-)

Index: sensors/prog/sensord/sense.c
=================================--- sensors.orig/prog/sensord/sense.c	2009-11-03 20:20:10.000000000 +0100
+++ sensors/prog/sensord/sense.c	2009-11-03 20:28:44.000000000 +0100
@@ -45,12 +45,23 @@
 
 static int idChip(const sensors_chip_name *chip)
 {
-	const char *adapter;
+	const char *name, *adapter;
+
+	name = chipName(chip);
+	if (!name) {
+		sensorLog(LOG_ERR, "Error getting chip name");
+		return -1;
+	}
+
+	sensorLog(LOG_INFO, "Chip: %s", name);
 
-	sensorLog(LOG_INFO, "Chip: %s", chipName(chip));
 	adapter = sensors_get_adapter_name(&chip->bus);
-	if (adapter)
-		sensorLog(LOG_INFO, "Adapter: %s", adapter);
+	if (!adapter) {
+		sensorLog(LOG_ERR, "Error getting adapter name");
+		return -1;
+	}
+
+	sensorLog(LOG_INFO, "Adapter: %s", adapter);
 
 	return 0;
 }
@@ -151,8 +162,11 @@
 	const FeatureDescriptor *features = descriptor->features;
 	int i, ret = 0;
 
-	if (action = DO_READ)
+	if (action = DO_READ) {
 		ret = idChip(chip);
+		if (ret)
+			return ret;
+	}
 
 	for (i = 0; features[i].format; i++) {
 		ret = do_features(chip, features + i, action);

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH v3 7/8] sensord: Evaluate return value of

[Jean Delvare]

Hi Andre,

On Tue, 3 Nov 2009 21:03:43 +0100, Andre Prendel wrote:
> Evaluate return value of idChip().
> 
> Some of the functions called from idChip() can fail which means
> idChip() can return an error. Callers of idChip() should be aware of
> this.
> ---
>  prog/sensord/sense.c |   24 +++++++++++++++++++-----
>  1 file changed, 19 insertions(+), 5 deletions(-)
> 
> Index: sensors/prog/sensord/sense.c
> =================================> --- sensors.orig/prog/sensord/sense.c	2009-11-03 20:20:10.000000000 +0100
> +++ sensors/prog/sensord/sense.c	2009-11-03 20:28:44.000000000 +0100
> @@ -45,12 +45,23 @@
>  
>  static int idChip(const sensors_chip_name *chip)
>  {
> -	const char *adapter;
> +	const char *name, *adapter;
> +
> +	name = chipName(chip);
> +	if (!name) {
> +		sensorLog(LOG_ERR, "Error getting chip name");
> +		return -1;
> +	}
> +
> +	sensorLog(LOG_INFO, "Chip: %s", name);
>  
> -	sensorLog(LOG_INFO, "Chip: %s", chipName(chip));
>  	adapter = sensors_get_adapter_name(&chip->bus);
> -	if (adapter)
> -		sensorLog(LOG_INFO, "Adapter: %s", adapter);
> +	if (!adapter) {
> +		sensorLog(LOG_ERR, "Error getting adapter name");
> +		return -1;
> +	}
> +
> +	sensorLog(LOG_INFO, "Adapter: %s", adapter);

I seem to remember that we made missing adapter name a non-fatal error
on purpose. It has happened in the past (when sysfs layout changed, for
example) and it did not seem worth causing sensord to fail, as this is
not a key information. So I'd revert this change.

>  
>  	return 0;
>  }
> @@ -151,8 +162,11 @@
>  	const FeatureDescriptor *features = descriptor->features;
>  	int i, ret = 0;
>  
> -	if (action = DO_READ)
> +	if (action = DO_READ) {
>  		ret = idChip(chip);
> +		if (ret)
> +			return ret;
> +	}
>  
>  	for (i = 0; features[i].format; i++) {
>  		ret = do_features(chip, features + i, action);


-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors