Skip to contentSkip to conversation

mac80211: ignore junk at end of frame when parsing IEs

PATCH New John W. Linville 3 participants 6 messages Oct 12, 2007 – Oct 18, 2007 mac80211

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mac80211: ignore junk at end of frame when parsing IEs

Some APs send management frames with junk padding after the last IE. We already account for a similar problem with some Apple Airport devices, but at least one device is known to send more than a single extra byte. The device in question is the Draytek Vigor2900:

	http://www.draytek.com.au/products/Vigor2900.php

The junk in question looks like an IE that runs off the end of the frame. This cause us to return ParseFailed. Since the frame in question is an association response, this causes us to fail to associate with this AP.

The proposed "fix" is to ignore an IE length that overruns the end of the frame if the IE type is unknown. This won't help if the junk type value matches a known IE type, but at least it won't fail association otherwise.

---
Thoughts? I'd love a cleaner solution.

For reference:

	https://bugzilla.redhat.com/show_bug.cgi?id=324191
 net/mac80211/ieee80211_sta.c |   31 +++++++++++++++++++------------
 1 files changed, 19 insertions(+), 12 deletions(-)
net/mac80211/ieee80211_sta.c +19 -12
index e804173..3bbc06d 100644
@@ -122,20 +122,12 @@ static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len,
 
 	while (left >= 2) {
 		u8 id, elen;
+		bool elem_unknown;
 
 		id = *pos++;
 		elen = *pos++;
 		left -= 2;
-
-		if (elen > left) {
-#if 0
-			if (net_ratelimit())
-				printk(KERN_DEBUG "IEEE 802.11 element parse "
-				       "failed (id=%d elen=%d left=%d)\n",
-				       id, elen, left);
-#endif
-			return ParseFailed;
-		}
+		elem_unknown = false;
 
 		switch (id) {
 		case WLAN_EID_SSID:
@@ -208,11 +200,26 @@ static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len,
 				      id, elen);
 #endif
 			unknown++;
+			elem_unknown = true;
 			break;
 		}
 
-		left -= elen;
-		pos += elen;
+		if (elen > left)
+			if (elem_unknown)
+				left = 0;
+			else {
+#if 0
+				if (net_ratelimit())
+					printk(KERN_DEBUG "IEEE 802.11 element parse "
+					       "failed (id=%d elen=%d left=%d)\n",
+					       id, elen, left);
+#endif
+				return ParseFailed;
+			}
+		else {
+			left -= elen;
+			pos += elen;
+		}
 	}
 
 	/* Do not trigger error if left == 1 as Apple Airport base stations
Signature · 1 line
1.5.2.4
#1 lore raw

PATCH 1/1 mac80211: make ieee802_11_parse_elems return void

Some APs send management frames with junk padding after the last IE. We already account for a similar problem with some Apple Airport devices, but at least one device is known to send more than a single extra byte. The device in question is the Draytek Vigor2900:

	http://www.draytek.com.au/products/Vigor2900.php

The junk in question looks like an IE that runs off the end of the frame. This cause us to return ParseFailed. Since the frame in question is an association response, this causes us to fail to associate with this AP.

The return code from ieee802_11_parse_elems is superfluous. All callers still check for the presence of the specific IEs that interest them anyway. So, remove the return code so the parse never "fails".

---
 net/mac80211/ieee80211_sta.c |   53 ++++++++----------------------------------
 1 files changed, 10 insertions(+), 43 deletions(-)
net/mac80211/ieee80211_sta.c +10 -43
index db81aef..7f873b3 100644
@@ -108,14 +108,11 @@ struct ieee802_11_elems {
 	u8 wmm_param_len;
 };
 
-enum ParseRes { ParseOK = 0, ParseUnknown = 1, ParseFailed = -1 };
-
-static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len,
-					    struct ieee802_11_elems *elems)
+static void ieee802_11_parse_elems(u8 *start, size_t len,
+				   struct ieee802_11_elems *elems)
 {
 	size_t left = len;
 	u8 *pos = start;
-	int unknown = 0;
 
 	memset(elems, 0, sizeof(*elems));
 
@@ -126,15 +123,8 @@ static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len,
 		elen = *pos++;
 		left -= 2;
 
-		if (elen > left) {
-#if 0
-			if (net_ratelimit())
-				printk(KERN_DEBUG "IEEE 802.11 element parse "
-				       "failed (id=%d elen=%d left=%d)\n",
-				       id, elen, left);
-#endif
-			return ParseFailed;
-		}
+		if (elen > left)
+			return;
 
 		switch (id) {
 		case WLAN_EID_SSID:
@@ -201,12 +191,6 @@ static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len,
 			elems->ext_supp_rates_len = elen;
 			break;
 		default:
-#if 0
-			printk(KERN_DEBUG "IEEE 802.11 element parse ignored "
-				      "unknown element (id=%d elen=%d)\n",
-				      id, elen);
-#endif
-			unknown++;
 			break;
 		}
 
@@ -214,10 +198,7 @@ static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len,
 		pos += elen;
 	}
 
-	/* Do not trigger error if left == 1 as Apple Airport base stations
-	 * send AssocResps that are one spurious byte too long. */
-
-	return unknown ? ParseUnknown : ParseOK;
+	return;
 }
 
 
@@ -931,12 +912,7 @@ static void ieee80211_auth_challenge(struct net_device *dev,
 
 	printk(KERN_DEBUG "%s: replying to auth challenge\n", dev->name);
 	pos = mgmt->u.auth.variable;
-	if (ieee802_11_parse_elems(pos, len - (pos - (u8 *) mgmt), &elems)
-	    == ParseFailed) {
-		printk(KERN_DEBUG "%s: failed to parse Auth(challenge)\n",
-		       dev->name);
-		return;
-	}
+	ieee802_11_parse_elems(pos, len - (pos - (u8 *) mgmt), &elems);
 	if (!elems.challenge) {
 		printk(KERN_DEBUG "%s: no challenge IE in shared key auth "
 		       "frame\n", dev->name);
@@ -1230,12 +1206,7 @@ static void ieee80211_rx_mgmt_assoc_resp(struct net_device *dev,
 	aid &= ~(BIT(15) | BIT(14));
 
 	pos = mgmt->u.assoc_resp.variable;
-	if (ieee802_11_parse_elems(pos, len - (pos - (u8 *) mgmt), &elems)
-	    == ParseFailed) {
-		printk(KERN_DEBUG "%s: failed to parse AssocResp\n",
-		       dev->name);
-		return;
-	}
+	ieee802_11_parse_elems(pos, len - (pos - (u8 *) mgmt), &elems);
 
 	if (!elems.supp_rates) {
 		printk(KERN_DEBUG "%s: no SuppRates element in AssocResp\n",
@@ -1459,7 +1430,7 @@ static void ieee80211_rx_bss_info(struct net_device *dev,
 	struct ieee80211_local *local = wdev_priv(dev->ieee80211_ptr);
 	struct ieee802_11_elems elems;
 	size_t baselen;
-	int channel, invalid = 0, clen;
+	int channel, clen;
 	struct ieee80211_sta_bss *bss;
 	struct sta_info *sta;
 	struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
@@ -1505,9 +1476,7 @@ static void ieee80211_rx_bss_info(struct net_device *dev,
 #endif /* CONFIG_MAC80211_IBSS_DEBUG */
 	}
 
-	if (ieee802_11_parse_elems(mgmt->u.beacon.variable, len - baselen,
-				   &elems) == ParseFailed)
-		invalid = 1;
+	ieee802_11_parse_elems(mgmt->u.beacon.variable, len - baselen, &elems);
 
 	if (sdata->type == IEEE80211_IF_TYPE_IBSS && elems.supp_rates &&
 	    memcmp(mgmt->bssid, sdata->u.sta.bssid, ETH_ALEN) == 0 &&
@@ -1724,9 +1693,7 @@ static void ieee80211_rx_mgmt_beacon(struct net_device *dev,
 	if (baselen > len)
 		return;
 
-	if (ieee802_11_parse_elems(mgmt->u.beacon.variable, len - baselen,
-				   &elems) == ParseFailed)
-		return;
+	ieee802_11_parse_elems(mgmt->u.beacon.variable, len - baselen, &elems);
 
 	if (elems.erp_info && elems.erp_info_len >= 1)
 		ieee80211_handle_erp_ie(dev, elems.erp_info[0]);
Signature · 1 line
1.5.2.4
Acked-by on v1 1/1
#2 lore raw

Re: [PATCH] mac80211: make ieee802_11_parse_elems return void

Quoted 12 lines · On Tuesday 16 October 2007 22:38:43 John W. Linville wrote:

Some APs send management frames with junk padding after the last IE. We already account for a similar problem with some Apple Airport devices, but at least one device is known to send more than a single extra byte. The device in question is the Draytek Vigor2900:

	http://www.draytek.com.au/products/Vigor2900.php

The junk in question looks like an IE that runs off the end of the frame. This cause us to return ParseFailed. Since the frame in question is an association response, this causes us to fail to associate with this AP.

Quoted 6 lines

The return code from ieee802_11_parse_elems is superfluous. All callers still check for the presence of the specific IEs that interest them anyway. So, remove the return code so the parse never "fails".

-Michael Wu

maintainer
#3 lore raw

Re: [PATCH] mac80211: make ieee802_11_parse_elems return void

In reply to John W. Linville

On Tue, 2007-10-16 at 22:38 -0400, John W. Linville wrote:

Looks good to me, but

net/mac80211/ieee80211_sta.c static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len, · line 201

Quoted 5 lines
net/mac80211/ieee80211_sta.c +1 -4
-	/* Do not trigger error if left == 1 as Apple Airport base stations
-	 * send AssocResps that are one spurious byte too long. */
-
-	return unknown ? ParseUnknown : ParseOK;
+	return;

huh?

johannes

#4 lore raw

Re: [PATCH] mac80211: make ieee802_11_parse_elems return void

Quoted 11 lines · Johannes Berg

On Tue, 2007-10-16 at 22:38 -0400, John W. Linville wrote:

Looks good to me, but

net/mac80211/ieee80211_sta.c static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len, · line 201

Quoted 5 lines
net/mac80211/ieee80211_sta.c +1 -4
-	/* Do not trigger error if left == 1 as Apple Airport base stations
-	 * send AssocResps that are one spurious byte too long. */
-
-	return unknown ? ParseUnknown : ParseOK;
+	return;

huh?

You don't like the bare return at the end? Or are you questioning something else?

John

Signature · 2 lines
John W. Linville
[email protected]
maintainer
#5 lore raw

Re: [PATCH] mac80211: make ieee802_11_parse_elems return void

Quoted 13 lines · John W. Linville
Quoted 11 lines · Johannes Berg

On Tue, 2007-10-16 at 22:38 -0400, John W. Linville wrote:

Looks good to me, but

net/mac80211/ieee80211_sta.c static enum ParseRes ieee802_11_parse_elems(u8 *start, size_t len, · line 201

Quoted 5 lines
net/mac80211/ieee80211_sta.c +1 -4
-	/* Do not trigger error if left == 1 as Apple Airport base stations
-	 * send AssocResps that are one spurious byte too long. */
-
-	return unknown ? ParseUnknown : ParseOK;
+	return;

huh?

Quoted 2 lines

You don't like the bare return at the end? Or are you questioning something else?

Bare return at the end of a void functions seems.. pointless.

johannes