Skip to contentSkip to conversation

kobject: don't oops on null kobject.name

PATCH In review Chuck Ebbert 3 participants 5 messages Jan 13, 2006 – Jan 14, 2006

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

PATCH 1/1 kobject: don't oops on null kobject.name

kobject_get_path() will oops if one of the component names is NULL. Fix that by returning NULL instead of oopsing.

---

Helge, this fixes your "2.6.15 OOPS while trying to mount cdrom".

Probably not the best fix, but It Works For Me (TM).

lib/kobject.c +4 -0
@@ -72,6 +72,8 @@ static int get_kobj_path_length(struct k
 	 * Add 1 to strlen for leading '/' of each level.
 	 */
 	do {
+		if (kobject_name(parent) == NULL)
+			return 0;
 		length += strlen(kobject_name(parent)) + 1;
 		parent = parent->parent;
 	} while (parent);
@@ -107,6 +109,8 @@ char *kobject_get_path(struct kobject *k
 	int len;
 
 	len = get_kobj_path_length(kobj);
+	if (len == 0)
+		return NULL;
 	path = kmalloc(len, gfp_mask);
 	if (!path)
 		return NULL;
Signature · 2 lines
Chuck
Currently reading: _Olympos_ by Dan Simmons
#1 lore raw
Quoted 23 lines · Chuck Ebbert

kobject_get_path() will oops if one of the component names is NULL. Fix that by returning NULL instead of oopsing.

---

Helge, this fixes your "2.6.15 OOPS while trying to mount cdrom".

Probably not the best fix, but It Works For Me (TM).

lib/kobject.c +2 -0
@@ -72,6 +72,8 @@ static int get_kobj_path_length(struct k
 	 * Add 1 to strlen for leading '/' of each level.
 	 */
 	do {
+		if (kobject_name(parent) == NULL)
+			return 0;
 		length += strlen(kobject_name(parent)) + 1;
 		parent = parent->parent;
 	} while (parent);

lib/kobject.c char *kobject_get_path(struct kobject *k · line 116

Quoted 8 lines
lib/kobject.c +2 -0
@@ -107,6 +109,8 @@ char *kobject_get_path(struct kobject *k
 	int len;
 
 	len = get_kobj_path_length(kobj);
+	if (len == 0)
+		return NULL;
 	path = kmalloc(len, gfp_mask);
 	if (!path)
 		return NULL;

I'd have thought that we'd want the test right at the start of kobject_add() - fail it if ->name is zero. I don't know if that'd work for all callers, but kobject_add() does play around with the ->name field and will go oops if ->name==NULL and debugging is enabled.

Why did you choose kobject_get_path()?

#2 lore raw
Quoted 26 lines · Andrew Morton
Quoted 25 lines · Chuck Ebbert

kobject_get_path() will oops if one of the component names is NULL. Fix that by returning NULL instead of oopsing.

---

Helge, this fixes your "2.6.15 OOPS while trying to mount cdrom".

Probably not the best fix, but It Works For Me (TM).

lib/kobject.c +2 -0
@@ -72,6 +72,8 @@ static int get_kobj_path_length(struct k
 	 * Add 1 to strlen for leading '/' of each level.
 	 */
 	do {
+		if (kobject_name(parent) == NULL)
+			return 0;
 		length += strlen(kobject_name(parent)) + 1;
 		parent = parent->parent;
 	} while (parent);
@@ -107,6 +109,8 @@ char *kobject_get_path(struct kobject *k
 	int len;
 
Quoted 11 lines

lib/kobject.c char *kobject_get_path(struct kobject *k · line 116

Quoted 6 lines
lib/kobject.c +2 -0
 	len = get_kobj_path_length(kobj);
+	if (len == 0)
+		return NULL;
 	path = kmalloc(len, gfp_mask);
 	if (!path)
 		return NULL;

I'd have thought that we'd want the test right at the start of kobject_add() - fail it if ->name is zero. I don't know if that'd work for all callers, but kobject_add() does play around with the ->name field and will go oops if ->name==NULL and debugging is enabled.

Something like this instead? (warning, untested...)

I'll try it out in a reboot cycle...

thanks,

greg k-h

lib/kobject.c +5 -0
@@ -164,6 +164,11 @@ int kobject_add(struct kobject * kobj)
 		return -ENOENT;
 	if (!kobj->k_name)
 		kobj->k_name = kobj->name;
+	if (!kobj->k_name) {
+		pr_debug("kobject attempted to be registered with no name!\n");
+		WARN_ON(1);
+		return -EINVAL;
+	}
 	parent = kobject_get(kobj->parent);
 
 	pr_debug("kobject %s: registering. parent: %s, set: %s\n",
#3 lore raw
Quoted 8 lines · Greg KH
Quoted 6 lines

I'd have thought that we'd want the test right at the start of kobject_add() - fail it if ->name is zero. I don't know if that'd work for all callers, but kobject_add() does play around with the ->name field and will go oops if ->name==NULL and debugging is enabled.

Something like this instead?

I think so.

Quoted 1 line

(warning, untested...)

Ship it!

Quoted 6 lines

I'll try it out in a reboot cycle...

lib/kobject.c +0 -0
@@ -164,6 +164,11 @@ int kobject_add(struct kobject * kobj)
 		return -ENOENT;

lib/kobject.c int kobject_add(struct kobject * kobj) · line 174

Quoted 10 lines
lib/kobject.c +5 -0
 	if (!kobj->k_name)
 		kobj->k_name = kobj->name;
+	if (!kobj->k_name) {
+		pr_debug("kobject attempted to be registered with no name!\n");
+		WARN_ON(1);
+		return -EINVAL;
+	}
 	parent = kobject_get(kobj->parent);
 
 	pr_debug("kobject %s: registering. parent: %s, set: %s\n",

It might be worth emitting the warning and then proceeding rather than failing - minimise potential disruption. I guess we'll see...

#4 lore raw
Quoted 8 lines · Andrew Morton
Quoted 7 lines · Greg KH
Quoted 6 lines

I'd have thought that we'd want the test right at the start of kobject_add() - fail it if ->name is zero. I don't know if that'd work for all callers, but kobject_add() does play around with the ->name field and will go oops if ->name==NULL and debugging is enabled.

Quoted 7 lines
Quoted 1 line

Something like this instead?

I think so.

Quoted 1 line

(warning, untested...)

Ship it!

Heh, it works for me, I'm running with it right now :)

Quoted 16 lines
Quoted 16 lines

I'll try it out in a reboot cycle...

lib/kobject.c +5 -0
@@ -164,6 +164,11 @@ int kobject_add(struct kobject * kobj)
 		return -ENOENT;
 	if (!kobj->k_name)
 		kobj->k_name = kobj->name;
+	if (!kobj->k_name) {
+		pr_debug("kobject attempted to be registered with no name!\n");
+		WARN_ON(1);
+		return -EINVAL;
+	}
 	parent = kobject_get(kobj->parent);
 
Quoted 4 lines

lib/kobject.c int kobject_add(struct kobject * kobj) · line 174

Quoted 1 line
lib/kobject.c +0 -0
 	pr_debug("kobject %s: registering. parent: %s, set: %s\n",

It might be worth emitting the warning and then proceeding rather than failing - minimise potential disruption. I guess we'll see...

Hm, I looked at the only user of kobjects in the kernel that I know of that doesn't use sysfs (the cdev code) and even it sets the kobject name to something sane, so I think we should be safe with this.

I'll add it to my tree and let's see what the next -mm causes to pop up :)

thanks,

greg k-h