Skip to contentSkip to conversation

gso: fix GSO_DODGY bit handling for related protocols

PATCH In review Yan Zhai 4 participants 7 messages Jul 13, 2023

Subsystems (4)
lore
Share & tools
lore

Yan Zhai

author
#0 lore raw

PATCH 1/1 gso: fix GSO_DODGY bit handling for related protocols

SKB_GSO_DODGY bit indicates a GSO packet comes from an untrusted source. The canonical way is to recompute the gso_segs to avoid device driver issues. Afterwards, the DODGY bit can be removed to avoid re-check at the egress of later devices, e.g. packets can egress to a vlan device backed by a real NIC.

Commit 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.") checks DODGY bit for UDP, but for packets that can be fed directly to the device after gso_segs reset, it actually falls through to fragmentation [1].

Commit 90017accff61 ("sctp: Add GSO support") and commit 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy") both didn't remove the DODGY bit after recomputing gso_segs.

This change fixes the GSO_UDP_L4 handling case, and remove the DODGY bit at other places.

Fixes 90017accff61 ("sctp: Add GSO support")
Fixes 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy")
Fixes 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.")

---
[1]:
https://lore.kernel.org/all/CAJPywTKDdjtwkLVUW6LRA2FU912qcDmQOQGt2WaDo28KzYDg+A@mail.gmail.com/

---
 net/ipv4/tcp_offload.c |  1 +
 net/ipv4/udp_offload.c | 19 +++++++++++++++----
 net/ipv6/udp_offload.c | 19 +++++++++++++++----
 net/sctp/offload.c     |  2 ++
 4 files changed, 33 insertions(+), 8 deletions(-)
net/ipv4/tcp_offload.c +1 -0
index 8311c38267b5..f9b93708c22e 100644
@@ -87,6 +87,7 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
 		/* Packet is from an untrusted source, reset gso_segs. */
 
 		skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
+		skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
 
 		segs = NULL;
 		goto out;
net/ipv4/udp_offload.c +15 -4
index 75aa4de5b731..bd29cf19bb6b 100644
@@ -388,11 +388,22 @@ static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
 	if (!pskb_may_pull(skb, sizeof(struct udphdr)))
 		goto out;
 
-	if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-	    !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-		return __udp_gso_segment(skb, features, false);
-
 	mss = skb_shinfo(skb)->gso_size;
+
+	if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+		if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+			/* Packet is from an untrusted source, reset actual gso_segs */
+			skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+								 mss);
+			skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+			segs = NULL;
+			goto out;
+		} else {
+			return __udp_gso_segment(skb, features, false);
+		}
+	}
+
 	if (unlikely(skb->len <= mss))
 		goto out;
 
net/ipv6/udp_offload.c +15 -4
index ad3b8726873e..6857d9f7bd06 100644
@@ -43,11 +43,22 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
 		if (!pskb_may_pull(skb, sizeof(struct udphdr)))
 			goto out;
 
-		if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-		    !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-			return __udp_gso_segment(skb, features, true);
-
 		mss = skb_shinfo(skb)->gso_size;
+
+		if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+			if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+				/* Packet is from an untrusted source, reset actual gso_segs */
+				skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+									 mss);
+				skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+				segs = NULL;
+				goto out;
+			} else {
+				return __udp_gso_segment(skb, features, true);
+			}
+		}
+
 		if (unlikely(skb->len <= mss))
 			goto out;
 
net/sctp/offload.c +2 -0
index 502095173d88..3d2b44db0d42 100644
@@ -65,6 +65,8 @@ static struct sk_buff *sctp_gso_segment(struct sk_buff *skb,
 		skb_walk_frags(skb, frag_iter)
 			pinfo->gso_segs++;
 
+		pinfo->gso_type &= ~SKB_GSO_DODGY;
+
 		segs = NULL;
 		goto out;
 	}
Signature · 1 line
2.30.2
#1 lore raw
Quoted 12 lines · Yan Zhai

SKB_GSO_DODGY bit indicates a GSO packet comes from an untrusted source. The canonical way is to recompute the gso_segs to avoid device driver issues. Afterwards, the DODGY bit can be removed to avoid re-check at the egress of later devices, e.g. packets can egress to a vlan device backed by a real NIC.

Commit 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.") checks DODGY bit for UDP, but for packets that can be fed directly to the device after gso_segs reset, it actually falls through to fragmentation [1].

Quoted 6 lines

Commit 90017accff61 ("sctp: Add GSO support") and commit 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy") both didn't remove the DODGY bit after recomputing gso_segs.

This change fixes the GSO_UDP_L4 handling case, and remove the DODGY bit at other places.

These two things should not be conflated.

Only the USO fix is strictly needed to fix the reported issue.

Quoted 3 lines
Fixes 90017accff61 ("sctp: Add GSO support")
Fixes 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy")
Fixes 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.")
Quoted 30 lines

---
[1]:
https://lore.kernel.org/all/CAJPywTKDdjtwkLVUW6LRA2FU912qcDmQOQGt2WaDo28KzYDg+A@mail.gmail.com/

---
 net/ipv4/tcp_offload.c |  1 +
 net/ipv4/udp_offload.c | 19 +++++++++++++++----
 net/ipv6/udp_offload.c | 19 +++++++++++++++----
 net/sctp/offload.c     |  2 ++
 4 files changed, 33 insertions(+), 8 deletions(-)
net/ipv4/tcp_offload.c +1 -0
index 8311c38267b5..f9b93708c22e 100644
@@ -87,6 +87,7 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
                /* Packet is from an untrusted source, reset gso_segs. */
 
                skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
+               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
 
                segs = NULL;
                goto out;
net/ipv4/udp_offload.c +0 -0
index 75aa4de5b731..bd29cf19bb6b 100644

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 391

Quoted 8 lines
net/ipv4/udp_offload.c +0 -4
@@ -388,11 +388,22 @@ static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
        if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                goto out;
 
-       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-           !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-               return __udp_gso_segment(skb, features, false);
-
        mss = skb_shinfo(skb)->gso_size;

Why move the block below this line?

Quoted 5 lines
net/ipv4/udp_offload.c +5 -0
+
+       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+               if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                       /* Packet is from an untrusted source, reset actual gso_segs */
+                       skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 406

Quoted 10 lines
net/ipv4/udp_offload.c +10 -0
+                                                                mss);
+                       skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                       segs = NULL;
+                       goto out;
+               } else {
+                       return __udp_gso_segment(skb, features, false);
+               }
+       }
+

The validation should take place inside __udp_gso_segment.

Revert the previous patch to always enter that function for USO packets:

       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
                return __udp_gso_segment(skb, features, false);

And in that function decide to return NULL after validation.

Quoted 50 lines
net/ipv4/udp_offload.c +0 -0
        if (unlikely(skb->len <= mss))
                goto out;
 
net/ipv6/udp_offload.c +15 -4
index ad3b8726873e..6857d9f7bd06 100644
@@ -43,11 +43,22 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
                if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                        goto out;
 
-               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-                   !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-                       return __udp_gso_segment(skb, features, true);
-
                mss = skb_shinfo(skb)->gso_size;
+
+               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+                       if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                               /* Packet is from an untrusted source, reset actual gso_segs */
+                               skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                        mss);
+                               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                               segs = NULL;
+                               goto out;
+                       } else {
+                               return __udp_gso_segment(skb, features, true);
+                       }
+               }
+
                if (unlikely(skb->len <= mss))
                        goto out;
 
net/sctp/offload.c +2 -1
index 502095173d88..3d2b44db0d42 100644
@@ -65,6 +65,8 @@ static struct sk_buff *sctp_gso_segment(struct sk_buff *skb,
                skb_walk_frags(skb, frag_iter)
                        pinfo->gso_segs++;
 
+               pinfo->gso_type &= ~SKB_GSO_DODGY;
+
                segs = NULL;
                goto out;
        }
--

2.30.2

#2 lore raw
Quoted 90 lines · Willem de Bruijn
Quoted 12 lines · Yan Zhai

SKB_GSO_DODGY bit indicates a GSO packet comes from an untrusted source. The canonical way is to recompute the gso_segs to avoid device driver issues. Afterwards, the DODGY bit can be removed to avoid re-check at the egress of later devices, e.g. packets can egress to a vlan device backed by a real NIC.

Commit 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.") checks DODGY bit for UDP, but for packets that can be fed directly to the device after gso_segs reset, it actually falls through to fragmentation [1].

Quoted 6 lines

Commit 90017accff61 ("sctp: Add GSO support") and commit 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy") both didn't remove the DODGY bit after recomputing gso_segs.

This change fixes the GSO_UDP_L4 handling case, and remove the DODGY bit at other places.

These two things should not be conflated.

Only the USO fix is strictly needed to fix the reported issue.

Quoted 3 lines
Fixes 90017accff61 ("sctp: Add GSO support")
Fixes 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy")
Fixes 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.")
Quoted 30 lines

---
[1]:
https://lore.kernel.org/all/CAJPywTKDdjtwkLVUW6LRA2FU912qcDmQOQGt2WaDo28KzYDg+A@mail.gmail.com/

---
 net/ipv4/tcp_offload.c |  1 +
 net/ipv4/udp_offload.c | 19 +++++++++++++++----
 net/ipv6/udp_offload.c | 19 +++++++++++++++----
 net/sctp/offload.c     |  2 ++
 4 files changed, 33 insertions(+), 8 deletions(-)
net/ipv4/tcp_offload.c +1 -0
index 8311c38267b5..f9b93708c22e 100644
@@ -87,6 +87,7 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
                /* Packet is from an untrusted source, reset gso_segs. */
 
                skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
+               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
 
                segs = NULL;
                goto out;
net/ipv4/udp_offload.c +0 -0
index 75aa4de5b731..bd29cf19bb6b 100644

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 391

Quoted 8 lines
net/ipv4/udp_offload.c +0 -4
@@ -388,11 +388,22 @@ static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
        if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                goto out;
 
-       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-           !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-               return __udp_gso_segment(skb, features, false);
-
        mss = skb_shinfo(skb)->gso_size;

Why move the block below this line?

Quoted 5 lines
net/ipv4/udp_offload.c +5 -0
+
+       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+               if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                       /* Packet is from an untrusted source, reset actual gso_segs */
+                       skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 406

Quoted 10 lines
net/ipv4/udp_offload.c +10 -0
+                                                                mss);
+                       skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                       segs = NULL;
+                       goto out;
+               } else {
+                       return __udp_gso_segment(skb, features, false);
+               }
+       }
+

The validation should take place inside __udp_gso_segment.

Quoted 6 lines

Revert the previous patch to always enter that function for USO packets:

       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
                return __udp_gso_segment(skb, features, false);

And in that function decide to return NULL after validation.

+1

Thanks

Quoted 53 lines
Quoted 52 lines
net/ipv4/udp_offload.c +0 -0
        if (unlikely(skb->len <= mss))
                goto out;
 
net/ipv6/udp_offload.c +15 -4
index ad3b8726873e..6857d9f7bd06 100644
@@ -43,11 +43,22 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
                if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                        goto out;
 
-               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-                   !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-                       return __udp_gso_segment(skb, features, true);
-
                mss = skb_shinfo(skb)->gso_size;
+
+               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+                       if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                               /* Packet is from an untrusted source, reset actual gso_segs */
+                               skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                        mss);
+                               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                               segs = NULL;
+                               goto out;
+                       } else {
+                               return __udp_gso_segment(skb, features, true);
+                       }
+               }
+
                if (unlikely(skb->len <= mss))
                        goto out;
 
net/sctp/offload.c +2 -1
index 502095173d88..3d2b44db0d42 100644
@@ -65,6 +65,8 @@ static struct sk_buff *sctp_gso_segment(struct sk_buff *skb,
                skb_walk_frags(skb, frag_iter)
                        pinfo->gso_segs++;
 
+               pinfo->gso_type &= ~SKB_GSO_DODGY;
+
                segs = NULL;
                goto out;
        }
--

2.30.2

author
#3 lore raw
In reply to Willem de Bruijn
Quoted 18 lines · Willem de Bruijn
Quoted 16 lines · Yan Zhai

SKB_GSO_DODGY bit indicates a GSO packet comes from an untrusted source. The canonical way is to recompute the gso_segs to avoid device driver issues. Afterwards, the DODGY bit can be removed to avoid re-check at the egress of later devices, e.g. packets can egress to a vlan device backed by a real NIC.

Commit 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.") checks DODGY bit for UDP, but for packets that can be fed directly to the device after gso_segs reset, it actually falls through to fragmentation [1].

Commit 90017accff61 ("sctp: Add GSO support") and commit 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy") both didn't remove the DODGY bit after recomputing gso_segs.

Quoted 7 lines
Quoted 2 lines

This change fixes the GSO_UDP_L4 handling case, and remove the DODGY bit at other places.

These two things should not be conflated.

Only the USO fix is strictly needed to fix the reported issue.

It's my OCD of wanting to avoid a cover letter for two patches... Let's address just this UDP issue then this time. The removal of DODGY is in fact more suitable as RFC for small improvements.

Quoted 39 lines
Quoted 3 lines
Fixes 90017accff61 ("sctp: Add GSO support")
Fixes 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy")
Fixes 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.")
Quoted 33 lines

---
[1]:
https://lore.kernel.org/all/CAJPywTKDdjtwkLVUW6LRA2FU912qcDmQOQGt2WaDo28KzYDg+A@mail.gmail.com/

---
 net/ipv4/tcp_offload.c |  1 +
 net/ipv4/udp_offload.c | 19 +++++++++++++++----
 net/ipv6/udp_offload.c | 19 +++++++++++++++----
 net/sctp/offload.c     |  2 ++
 4 files changed, 33 insertions(+), 8 deletions(-)
net/ipv4/tcp_offload.c +1 -0
index 8311c38267b5..f9b93708c22e 100644
@@ -87,6 +87,7 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
                /* Packet is from an untrusted source, reset gso_segs. */
 
                skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
+               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
 
                segs = NULL;
                goto out;
net/ipv4/udp_offload.c +0 -0
index 75aa4de5b731..bd29cf19bb6b 100644
@@ -388,11 +388,22 @@ static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
        if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                goto out;
 
Quoted 8 lines

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 391

Quoted 5 lines
net/ipv4/udp_offload.c +0 -4
-       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-           !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-               return __udp_gso_segment(skb, features, false);
-
        mss = skb_shinfo(skb)->gso_size;

Why move the block below this line?

if we move the dodgy handling into __udp_gso_segment then it does not need to move below this line.

Quoted 18 lines
Quoted 5 lines
net/ipv4/udp_offload.c +5 -0
+
+       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+               if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                       /* Packet is from an untrusted source, reset actual gso_segs */
+                       skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 406

Quoted 10 lines
net/ipv4/udp_offload.c +10 -0
+                                                                mss);
+                       skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                       segs = NULL;
+                       goto out;
+               } else {
+                       return __udp_gso_segment(skb, features, false);
+               }
+       }
+

The validation should take place inside __udp_gso_segment.

Quoted 7 lines

Revert the previous patch to always enter that function for USO packets:

       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
                return __udp_gso_segment(skb, features, false);

And in that function decide to return NULL after validation.

Good call, that's indeed better. Thanks

Quoted 51 lines
Quoted 51 lines
net/ipv4/udp_offload.c +0 -0
        if (unlikely(skb->len <= mss))
                goto out;
 
net/ipv6/udp_offload.c +15 -4
index ad3b8726873e..6857d9f7bd06 100644
@@ -43,11 +43,22 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
                if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                        goto out;
 
-               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-                   !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-                       return __udp_gso_segment(skb, features, true);
-
                mss = skb_shinfo(skb)->gso_size;
+
+               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+                       if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                               /* Packet is from an untrusted source, reset actual gso_segs */
+                               skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                        mss);
+                               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                               segs = NULL;
+                               goto out;
+                       } else {
+                               return __udp_gso_segment(skb, features, true);
+                       }
+               }
+
                if (unlikely(skb->len <= mss))
                        goto out;
 
net/sctp/offload.c +2 -1
index 502095173d88..3d2b44db0d42 100644
@@ -65,6 +65,8 @@ static struct sk_buff *sctp_gso_segment(struct sk_buff *skb,
                skb_walk_frags(skb, frag_iter)
                        pinfo->gso_segs++;
 
+               pinfo->gso_type &= ~SKB_GSO_DODGY;
+
                segs = NULL;
                goto out;
        }
--

2.30.2

Signature · 1 line
Yan
#4 lore raw
In reply to Yan Zhai
Quoted 7 lines · Yan Zhai

SKB_GSO_DODGY bit indicates a GSO packet comes from an untrusted source. The canonical way is to recompute the gso_segs to avoid device driver issues. Afterwards, the DODGY bit can be removed to avoid re-check at the egress of later devices, e.g. packets can egress to a vlan device backed by a real NIC.

Quoted 8 lines

Commit 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.") checks DODGY bit for UDP, but for packets that can be fed directly to the device after gso_segs reset, it actually falls through to fragmentation [1].

Commit 90017accff61 ("sctp: Add GSO support") and commit 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy") both didn't remove the DODGY bit after recomputing gso_segs.

If we try to fix two issues, we'd better use separate patches.

Quoted 47 lines

This change fixes the GSO_UDP_L4 handling case, and remove the DODGY bit at other places.

Fixes 90017accff61 ("sctp: Add GSO support")
Fixes 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy")
Fixes 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.")

---
[1]:
https://lore.kernel.org/all/CAJPywTKDdjtwkLVUW6LRA2FU912qcDmQOQGt2WaDo28KzYDg+A@mail.gmail.com/

---
 net/ipv4/tcp_offload.c |  1 +
 net/ipv4/udp_offload.c | 19 +++++++++++++++----
 net/ipv6/udp_offload.c | 19 +++++++++++++++----
 net/sctp/offload.c     |  2 ++
 4 files changed, 33 insertions(+), 8 deletions(-)
net/ipv4/tcp_offload.c +1 -0
index 8311c38267b5..f9b93708c22e 100644
@@ -87,6 +87,7 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
                /* Packet is from an untrusted source, reset gso_segs. */
 
                skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
+               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
 
                segs = NULL;
                goto out;
net/ipv4/udp_offload.c +2 -4
index 75aa4de5b731..bd29cf19bb6b 100644
@@ -388,11 +388,22 @@ static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
        if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                goto out;
 
-       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-           !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-               return __udp_gso_segment(skb, features, false);
-
        mss = skb_shinfo(skb)->gso_size;
+
+       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 403

Quoted 10 lines
net/ipv4/udp_offload.c +10 -0
+               if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                       /* Packet is from an untrusted source, reset actual gso_segs */
+                       skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                mss);
+                       skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                       segs = NULL;
+                       goto out;
+               } else {
+                       return __udp_gso_segment(skb, features, false);

I think it's better and cleaner to move those changes in __udp_gso_segment() as Willem suggests.

Quoted 16 lines
net/ipv4/udp_offload.c +3 -0
+               }
+       }
+
        if (unlikely(skb->len <= mss))
                goto out;
 
net/ipv6/udp_offload.c +0 -2
index ad3b8726873e..6857d9f7bd06 100644
@@ -43,11 +43,22 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
                if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                        goto out;
 
-               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-                   !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))

net/ipv6/udp_offload.c static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb, · line 53

Quoted 10 lines
net/ipv6/udp_offload.c +7 -2
-                       return __udp_gso_segment(skb, features, true);
-
                mss = skb_shinfo(skb)->gso_size;
+
+               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+                       if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                               /* Packet is from an untrusted source, reset actual gso_segs */
+                               skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                        mss);
+                               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;

Any reason you want to remove the DODGY here? Is this an optimization? We will lose the chance to recognize/validate it elsewhere.

Thanks

Quoted 27 lines
net/ipv6/udp_offload.c +8 -0
+
+                               segs = NULL;
+                               goto out;
+                       } else {
+                               return __udp_gso_segment(skb, features, true);
+                       }
+               }
+
                if (unlikely(skb->len <= mss))
                        goto out;
 
net/sctp/offload.c +2 -1
index 502095173d88..3d2b44db0d42 100644
@@ -65,6 +65,8 @@ static struct sk_buff *sctp_gso_segment(struct sk_buff *skb,
                skb_walk_frags(skb, frag_iter)
                        pinfo->gso_segs++;
 
+               pinfo->gso_type &= ~SKB_GSO_DODGY;
+
                segs = NULL;
                goto out;
        }
--

2.30.2

author
#5 lore raw
Quoted 95 lines · Jason Wang
Quoted 7 lines · Yan Zhai

SKB_GSO_DODGY bit indicates a GSO packet comes from an untrusted source. The canonical way is to recompute the gso_segs to avoid device driver issues. Afterwards, the DODGY bit can be removed to avoid re-check at the egress of later devices, e.g. packets can egress to a vlan device backed by a real NIC.

Quoted 8 lines

Commit 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.") checks DODGY bit for UDP, but for packets that can be fed directly to the device after gso_segs reset, it actually falls through to fragmentation [1].

Commit 90017accff61 ("sctp: Add GSO support") and commit 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy") both didn't remove the DODGY bit after recomputing gso_segs.

If we try to fix two issues, we'd better use separate patches.

Quoted 47 lines

This change fixes the GSO_UDP_L4 handling case, and remove the DODGY bit at other places.

Fixes 90017accff61 ("sctp: Add GSO support")
Fixes 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy")
Fixes 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.")

---
[1]:
https://lore.kernel.org/all/CAJPywTKDdjtwkLVUW6LRA2FU912qcDmQOQGt2WaDo28KzYDg+A@mail.gmail.com/

---
 net/ipv4/tcp_offload.c |  1 +
 net/ipv4/udp_offload.c | 19 +++++++++++++++----
 net/ipv6/udp_offload.c | 19 +++++++++++++++----
 net/sctp/offload.c     |  2 ++
 4 files changed, 33 insertions(+), 8 deletions(-)
net/ipv4/tcp_offload.c +1 -0
index 8311c38267b5..f9b93708c22e 100644
@@ -87,6 +87,7 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
                /* Packet is from an untrusted source, reset gso_segs. */
 
                skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
+               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
 
                segs = NULL;
                goto out;
net/ipv4/udp_offload.c +2 -4
index 75aa4de5b731..bd29cf19bb6b 100644
@@ -388,11 +388,22 @@ static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
        if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                goto out;
 
-       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-           !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-               return __udp_gso_segment(skb, features, false);
-
        mss = skb_shinfo(skb)->gso_size;
+
+       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 403

Quoted 10 lines
net/ipv4/udp_offload.c +10 -0
+               if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                       /* Packet is from an untrusted source, reset actual gso_segs */
+                       skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                mss);
+                       skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                       segs = NULL;
+                       goto out;
+               } else {
+                       return __udp_gso_segment(skb, features, false);

I think it's better and cleaner to move those changes in __udp_gso_segment() as Willem suggests.

Quoted 14 lines
net/ipv4/udp_offload.c +3 -0
+               }
+       }
+
        if (unlikely(skb->len <= mss))
                goto out;
 
net/ipv6/udp_offload.c +0 -0
index ad3b8726873e..6857d9f7bd06 100644
@@ -43,11 +43,22 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
                if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                        goto out;
 
Quoted 16 lines

net/ipv6/udp_offload.c static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb, · line 53

Quoted 12 lines
net/ipv6/udp_offload.c +7 -4
-               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-                   !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-                       return __udp_gso_segment(skb, features, true);
-
                mss = skb_shinfo(skb)->gso_size;
+
+               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+                       if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                               /* Packet is from an untrusted source, reset actual gso_segs */
+                               skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                        mss);
+                               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;

Any reason you want to remove the DODGY here? Is this an optimization? We will lose the chance to recognize/validate it elsewhere.

It is intended as a small optimization. And this is in fact the piece I am not fully confident about: after validating the gso_segs at a trusted location (i.e. assuming the kernel is the trusted computing base), do we need to validate it somewhere else? For example, in our scenario, we have a tun/tap device in a net namespace, so the packet going out will enter from the tap, get forwarded through an veth, and then a vlan backed by a real ethernet interface. If the bit is carried over, then at each egress of these devices, we need to enter the GSO code, which feels pretty redundant as long as the packet does not leave kernel space. WDYT?

thanks

Quoted 30 lines

Thanks

Quoted 27 lines
net/ipv6/udp_offload.c +8 -0
+
+                               segs = NULL;
+                               goto out;
+                       } else {
+                               return __udp_gso_segment(skb, features, true);
+                       }
+               }
+
                if (unlikely(skb->len <= mss))
                        goto out;
 
net/sctp/offload.c +2 -1
index 502095173d88..3d2b44db0d42 100644
@@ -65,6 +65,8 @@ static struct sk_buff *sctp_gso_segment(struct sk_buff *skb,
                skb_walk_frags(skb, frag_iter)
                        pinfo->gso_segs++;
 
+               pinfo->gso_type &= ~SKB_GSO_DODGY;
+
                segs = NULL;
                goto out;
        }
--

2.30.2

Signature · 1 line
Yan
maintainer
#6 lore raw
Quoted 112 lines · Yan Zhai
Quoted 95 lines · Jason Wang
Quoted 7 lines · Yan Zhai

SKB_GSO_DODGY bit indicates a GSO packet comes from an untrusted source. The canonical way is to recompute the gso_segs to avoid device driver issues. Afterwards, the DODGY bit can be removed to avoid re-check at the egress of later devices, e.g. packets can egress to a vlan device backed by a real NIC.

Quoted 8 lines

Commit 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.") checks DODGY bit for UDP, but for packets that can be fed directly to the device after gso_segs reset, it actually falls through to fragmentation [1].

Commit 90017accff61 ("sctp: Add GSO support") and commit 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy") both didn't remove the DODGY bit after recomputing gso_segs.

If we try to fix two issues, we'd better use separate patches.

Quoted 47 lines

This change fixes the GSO_UDP_L4 handling case, and remove the DODGY bit at other places.

Fixes 90017accff61 ("sctp: Add GSO support")
Fixes 3820c3f3e417 ("[TCP]: Reset gso_segs if packet is dodgy")
Fixes 1fd54773c267 ("udp: allow header check for dodgy GSO_UDP_L4 packets.")

---
[1]:
https://lore.kernel.org/all/CAJPywTKDdjtwkLVUW6LRA2FU912qcDmQOQGt2WaDo28KzYDg+A@mail.gmail.com/

---
 net/ipv4/tcp_offload.c |  1 +
 net/ipv4/udp_offload.c | 19 +++++++++++++++----
 net/ipv6/udp_offload.c | 19 +++++++++++++++----
 net/sctp/offload.c     |  2 ++
 4 files changed, 33 insertions(+), 8 deletions(-)
net/ipv4/tcp_offload.c +1 -0
index 8311c38267b5..f9b93708c22e 100644
@@ -87,6 +87,7 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
                /* Packet is from an untrusted source, reset gso_segs. */
 
                skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
+               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
 
                segs = NULL;
                goto out;
net/ipv4/udp_offload.c +2 -4
index 75aa4de5b731..bd29cf19bb6b 100644
@@ -388,11 +388,22 @@ static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
        if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                goto out;
 
-       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-           !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-               return __udp_gso_segment(skb, features, false);
-
        mss = skb_shinfo(skb)->gso_size;
+
+       if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {

net/ipv4/udp_offload.c static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb, · line 403

Quoted 10 lines
net/ipv4/udp_offload.c +10 -0
+               if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                       /* Packet is from an untrusted source, reset actual gso_segs */
+                       skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                mss);
+                       skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
+
+                       segs = NULL;
+                       goto out;
+               } else {
+                       return __udp_gso_segment(skb, features, false);

I think it's better and cleaner to move those changes in __udp_gso_segment() as Willem suggests.

Quoted 14 lines
net/ipv4/udp_offload.c +3 -0
+               }
+       }
+
        if (unlikely(skb->len <= mss))
                goto out;
 
net/ipv6/udp_offload.c +0 -0
index ad3b8726873e..6857d9f7bd06 100644
@@ -43,11 +43,22 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
                if (!pskb_may_pull(skb, sizeof(struct udphdr)))
                        goto out;
 
Quoted 16 lines

net/ipv6/udp_offload.c static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb, · line 53

Quoted 12 lines
net/ipv6/udp_offload.c +7 -4
-               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4 &&
-                   !skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST))
-                       return __udp_gso_segment(skb, features, true);
-
                mss = skb_shinfo(skb)->gso_size;
+
+               if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+                       if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
+                               /* Packet is from an untrusted source, reset actual gso_segs */
+                               skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len - sizeof(*uh),
+                                                                        mss);
+                               skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;

Any reason you want to remove the DODGY here? Is this an optimization? We will lose the chance to recognize/validate it elsewhere.

Quoted 10 lines

It is intended as a small optimization. And this is in fact the piece I am not fully confident about: after validating the gso_segs at a trusted location (i.e. assuming the kernel is the trusted computing base), do we need to validate it somewhere else? For example, in our scenario, we have a tun/tap device in a net namespace, so the packet going out will enter from the tap, get forwarded through an veth, and then a vlan backed by a real ethernet interface. If the bit is carried over, then at each egress of these devices, we need to enter the GSO code, which feels pretty redundant as long as the packet does not leave kernel space. WDYT?

As an optimization, I think it should land on a different (net-next) patch. Additionally I think it should be possible to get a greater gain adding the ROBUST feature to virtual devices (but I'm not sure if syzkaller will be able to use that in nasty ways).

Cheers,

Paolo