Skip to contentSkip to conversation

cfg80211: allow wext to remove keys that don't exist

PATCH In review Johannes Berg 2 participants 4 messages May 18, 2009 802.11 (including cfg80211/NL80211)

lore
Share & tools
lore

Johannes Berg

author maintainer
#0 lore raw

PATCH 1/1 cfg80211: allow wext to remove keys that don't exist

Some applications using wireless extensions expect to be able to remove a key that doesn't exist. One example is wpa_supplicant which doesn't actually change behaviour when running into an error while trying to do that, but it prints an error message which users interpret as wpa_supplicant having problems.

The safe thing to do is not change the behaviour of wireless extensions any more, so when the driver reports -ENOENT let the wext bridge code return success to userspace. To guarantee this, also document that drivers should return -ENOENT when the key doesn't exist.

---
 include/net/cfg80211.h     |    5 +++--
 net/wireless/wext-compat.c |    7 +++++++
 2 files changed, 10 insertions(+), 2 deletions(-)
include/net/cfg80211.h +3 -2
@@ -778,10 +778,11 @@ enum wiphy_params_flags {
  * @get_key: get information about the key with the given parameters.
  *	@mac_addr will be %NULL when requesting information for a group
  *	key. All pointers given to the @callback function need not be valid
- *	after it returns.
+ *	after it returns. This function should return an error if it is
+ *	not possible to retrieve the key, -ENOENT if it doesn't exist.
  *
  * @del_key: remove a key given the @mac_addr (%NULL for a group key)
- *	and @key_index
+ *	and @key_index, return -ENOENT if the key doesn't exist.
  *
  * @set_default_key: set the default key on an interface
  *
net/wireless/wext-compat.c +7 -0
@@ -504,6 +504,13 @@ static int cfg80211_set_encryption(struc
 			else if (idx == wdev->wext.default_mgmt_key)
 				wdev->wext.default_mgmt_key = -1;
 		}
+		/*
+		 * Applications using wireless extensions expect to be
+		 * able to delete keys that don't exist, so allow that.
+		 */
+		if (err == -ENOENT)
+			return 0;
+
 		return err;
 	} else {
 		if (addr)
#1 lore raw
Quoted 5 lines · Johannes Berg

Some applications using wireless extensions expect to be able to remove a key that doesn't exist. One example is wpa_supplicant which doesn't actually change behaviour when running into an error while trying to do that, but it prints an error message which users interpret as wpa_supplicant having problems.

It sounds like you are working around a userspace problem in the kernel.

Quoted 5 lines

The safe thing to do is not change the behaviour of wireless extensions any more, so when the driver reports -ENOENT let the wext bridge code return success to userspace. To guarantee this, also document that drivers should return -ENOENT when the key doesn't exist.

You patch is changing the behavior or wireless extensions. It would be much more reasonable for wpa_supplicant not to remove non-existent keys or (if it's unsafe or non-practical for some reason) not to report -ENOENT to the user.

Signature · 2 lines
Regards,
Pavel Roskin
author maintainer
#2 lore raw
Quoted 8 lines · Pavel Roskin
Quoted 5 lines · Johannes Berg

Some applications using wireless extensions expect to be able to remove a key that doesn't exist. One example is wpa_supplicant which doesn't actually change behaviour when running into an error while trying to do that, but it prints an error message which users interpret as wpa_supplicant having problems.

It sounds like you are working around a userspace problem in the kernel.

More a "user problem" rather than "userspace problem"...

Quoted 10 lines
Quoted 5 lines

The safe thing to do is not change the behaviour of wireless extensions any more, so when the driver reports -ENOENT let the wext bridge code return success to userspace. To guarantee this, also document that drivers should return -ENOENT when the key doesn't exist.

You patch is changing the behavior or wireless extensions. It would be much more reasonable for wpa_supplicant not to remove non-existent keys or (if it's unsafe or non-practical for some reason) not to report -ENOENT to the user.

Umm, no... my previous patch changed the behaviour and this restores it.

johannes

#3 lore raw
Quoted 11 lines · Johannes Berg
Quoted 8 lines · Pavel Roskin
Quoted 5 lines · Johannes Berg

Some applications using wireless extensions expect to be able to remove a key that doesn't exist. One example is wpa_supplicant which doesn't actually change behaviour when running into an error while trying to do that, but it prints an error message which users interpret as wpa_supplicant having problems.

It sounds like you are working around a userspace problem in the kernel.

More a "user problem" rather than "userspace problem"...

Let's give users a break. They cannot know which error messages are important and which are not. Users' attention should not be attracted to trivial things such as failure to remove something that never existed.

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines

The safe thing to do is not change the behaviour of wireless extensions any more, so when the driver reports -ENOENT let the wext bridge code return success to userspace. To guarantee this, also document that drivers should return -ENOENT when the key doesn't exist.

Quoted 6 lines
Quoted 4 lines

You patch is changing the behavior or wireless extensions. It would be much more reasonable for wpa_supplicant not to remove non-existent keys or (if it's unsafe or non-practical for some reason) not to report -ENOENT to the user.

Umm, no... my previous patch changed the behaviour and this restores it.

OK, that makes me more sympathetic to the patch, at least if the change did not get exposed in any released kernel.

Signature · 2 lines
Regards,
Pavel Roskin