Skip to contentSkip to conversation

kobject: kobject_uevent() returns manageable value

PATCH In review Mauricio Lin 3 participants 3 messages Nov 30, 2006 – Dec 1, 2006

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

PATCH 1/1 kobject: kobject_uevent() returns manageable value

[PATCH 2.6.19] kobject: kobject_uevent() returns manageable value

Since kobject_uevent() function does not return an integer value to indicate if its operation was completed with success or not, it is worth changing it in order to report a proper status (success or error) instead of returning void.

Keep kobject_uevent() returning the status as integer provide a easier way for detecting possible failure in the function. Using void returning style may take people to waste more time to figure out if the "send to" or "receive from" an event is a bug in the kernel or user space. Furthermore, the current way to detect where the error is taking place in the kobject_uevent() requires additional inclusion of printk() in each "if" condition that can lead to failure.

include/linux/kobject.h +5 -2
Index: kernel/linux-2.6.19-rc6/include/linux/kobject.h
===================================================================
@@ -263,14 +263,17 @@
 					struct subsys_attribute *);
 
 #if defined(CONFIG_HOTPLUG)
-void kobject_uevent(struct kobject *kobj, enum kobject_action action);
+int kobject_uevent(struct kobject *kobj, enum kobject_action action);
 
 int add_uevent_var(char **envp, int num_envp, int *cur_index,
 			char *buffer, int buffer_size, int *cur_len,
 			const char *format, ...)
 	__attribute__((format (printf, 7, 8)));
 #else
-static inline void kobject_uevent(struct kobject *kobj, enum kobject_action action) { }
+static inline int kobject_uevent(struct kobject *kobj, enum kobject_action action)
+{
+	return 0;
+}
 
 static inline int add_uevent_var(char **envp, int num_envp, int *cur_index,
 				      char *buffer, int buffer_size, int *cur_len, 
lib/kobject_uevent.c +23 -11
Index: kernel/linux-2.6.19-rc6/lib/kobject_uevent.c
===================================================================
@@ -60,8 +60,11 @@
  *
  * @action: action that is happening (usually KOBJ_ADD and KOBJ_REMOVE)
  * @kobj: struct kobject that the action is happening to
+ *
+ * Returns 0 if kobject_uevent() is completed with success or the
+ * corresponding error when it fails.
  */
-void kobject_uevent(struct kobject *kobj, enum kobject_action action)
+int kobject_uevent(struct kobject *kobj, enum kobject_action action)
 {
 	char **envp;
 	char *buffer;
@@ -81,7 +84,7 @@
 
 	action_string = action_to_string(action);
 	if (!action_string)
-		return;
+		return -EINVAL;
 
 	/* search the kset we belong to */
 	top_kobj = kobj;
@@ -91,7 +94,7 @@
 		} while (!top_kobj->kset && top_kobj->parent);
 	}
 	if (!top_kobj->kset)
-		return;
+		return -EINVAL;
 
 	kset = top_kobj->kset;
 	uevent_ops = kset->uevent_ops;
@@ -99,22 +102,27 @@
 	/*  skip the event, if the filter returns zero. */
 	if (uevent_ops && uevent_ops->filter)
 		if (!uevent_ops->filter(kset, kobj))
-			return;
+			return -EINVAL;
 
 	/* environment index */
 	envp = kzalloc(NUM_ENVP * sizeof (char *), GFP_KERNEL);
 	if (!envp)
-		return;
+		return -ENOMEM;
 
 	/* environment values */
 	buffer = kmalloc(BUFFER_SIZE, GFP_KERNEL);
-	if (!buffer)
-		goto exit;
+	if (!buffer) {
+		kfree(envp);
+		return -ENOMEM;
+	}
 
 	/* complete object path */
 	devpath = kobject_get_path(kobj, GFP_KERNEL);
-	if (!devpath)
-		goto exit;
+	if (!devpath) {
+		kfree(envp);
+		kfree(buffer);
+		return -ENOMEM;
+	}
 
 	/* originating subsystem */
 	if (uevent_ops && uevent_ops->name)
@@ -179,7 +187,11 @@
 			}
 
 			NETLINK_CB(skb).dst_group = 1;
-			netlink_broadcast(uevent_sock, skb, 0, 1, GFP_KERNEL);
+			retval = netlink_broadcast(uevent_sock, skb, 0, 1,
+						   GFP_KERNEL);
+			if (retval)
+				pr_debug ("%s - netlink_broadcast() returned "
+					  "%d\n", __FUNCTION__, retval);
 		}
 	}
 #endif
@@ -198,7 +210,7 @@
 	kfree(devpath);
 	kfree(buffer);
 	kfree(envp);
-	return;
+	return 0;
 }
 EXPORT_SYMBOL_GPL(kobject_uevent);
 
#1 lore raw
Quoted 5 lines · On Thu, 30 Nov 2006 16:58:47 -0400 Mauricio Lin <[email protected]>…

Since kobject_uevent() function does not return an integer value to indicate if its operation was completed with success or not, it is worth changing it in order to report a proper status (success or error) instead of returning void.

Quoted 7 lines

Keep kobject_uevent() returning the status as integer provide a easier way for detecting possible failure in the function. Using void returning style may take people to waste more time to figure out if the "send to" or "receive from" an event is a bug in the kernel or user space. Furthermore, the current way to detect where the error is taking place in the kobject_uevent() requires additional inclusion of printk() in each "if" condition that can lead to failure.

Admirable idea, but we have large changes pending against that code and none of this patch applies.

A patch against Greg's driver tree, or against latest -mm would suit, thanks.

#2 lore raw

Hi Andrew,

Quoted 16 lines · Andrew Morton
Quoted 5 lines · On Thu, 30 Nov 2006 16:58:47 -0400 Mauricio Lin <[email protected]>…

Since kobject_uevent() function does not return an integer value to indicate if its operation was completed with success or not, it is worth changing it in order to report a proper status (success or error) instead of returning void.

Quoted 7 lines

Keep kobject_uevent() returning the status as integer provide a easier way for detecting possible failure in the function. Using void returning style may take people to waste more time to figure out if the "send to" or "receive from" an event is a bug in the kernel or user space. Furthermore, the current way to detect where the error is taking place in the kobject_uevent() requires additional inclusion of printk() in each "if" condition that can lead to failure.

Quoted 2 lines

Admirable idea, but we have large changes pending against that code and none of this patch applies.

I have used the kobject_uevent() to send event to user space, but when the event does not happen I have to figure out if this problem is related to kernel or user space code. After mentioning this issue with Greg KH, he recommended to provide a patch to fix it.

Quoted 2 lines

A patch against Greg's driver tree, or against latest -mm would suit, thanks.

OK. That's good. Thanks.

BR,

Mauricio Lin.