p80211wext.c: possible problem with IW_ENCODE_ALG_WEP

p80211wext.c: possible problem with IW_ENCODE_ALG_WEP

[Julia Lawall]

The file drivers/staging/wlan-ng/p80211wext.c contains the following code:

(starting at line 1490 in a recent linux-next)

		if (!(ext->alg & IW_ENCODE_ALG_WEP)) {
			pr_debug("asked to set a non wep key :(");
			return -EINVAL;
		}

The reference to IW_ENCODE_ALG_WEP seems to refer to the following
definitions in include/linux/wireless.h

/* struct iw_encode_ext ->alg */
#define IW_ENCODE_ALG_NONE	0
#define IW_ENCODE_ALG_WEP	1
#define IW_ENCODE_ALG_TKIP	2
#define IW_ENCODE_ALG_CCMP	3
#define IW_ENCODE_ALG_PMK	4
#define IW_ENCODE_ALG_AES_CMAC	5

But if the alg field returns true for a bit and with IW_ENCODE_ALG_WEP,
then it would also return true for a bit and with IW_ENCODE_ALG_CCMP and
IW_ENCODE_ALG_AES_CMAC.  Is this what is intended, or should the test be
encoded in another way?

julia

Re: p80211wext.c: possible problem with IW_ENCODE_ALG_WEP

[Richard Kennedy]

Julia Lawall wrote:
> The file drivers/staging/wlan-ng/p80211wext.c contains the following code:
> 
> (starting at line 1490 in a recent linux-next)
> 
> 		if (!(ext->alg & IW_ENCODE_ALG_WEP)) {
> 			pr_debug("asked to set a non wep key :(");
> 			return -EINVAL;
> 		}
> 
> The reference to IW_ENCODE_ALG_WEP seems to refer to the following
> definitions in include/linux/wireless.h
> 
> /* struct iw_encode_ext ->alg */
> #define IW_ENCODE_ALG_NONE	0
> #define IW_ENCODE_ALG_WEP	1
> #define IW_ENCODE_ALG_TKIP	2
> #define IW_ENCODE_ALG_CCMP	3
> #define IW_ENCODE_ALG_PMK	4
> #define IW_ENCODE_ALG_AES_CMAC	5
> 
> But if the alg field returns true for a bit and with IW_ENCODE_ALG_WEP,
> then it would also return true for a bit and with IW_ENCODE_ALG_CCMP and
> IW_ENCODE_ALG_AES_CMAC.  Is this what is intended, or should the test be
> encoded in another way?
> 
> julia
Hi Julia,

You're quite right, this is not what was intended.

The card only handles WEP keys so
	if (ext->alg != IW_ENCODE_ALG_WEP)  {...}

would be better.

Do you want to send a patch to correct this? If not, I can do it.

regards
Richard