[PATCH v3 net 1/2] net/sched: act_vlan: Push skb->data to mac_header prior calling skb_vlan_*() functions

[PATCH v3 net 1/2] net/sched: act_vlan: Push skb->data to mac_header prior calling skb_vlan_*() functions

[Shmulik Ladkani]

Generic skb_vlan_push/skb_vlan_pop functions don't properly handle the
case where the input skb data pointer does not point at the mac header:

- They're doing push/pop, but fail to properly unwind data back to its
  original location.
  For example, in the skb_vlan_push case, any subsequent
  'skb_push(skb, skb->mac_len)' calls make the skb->data point 4 bytes
  BEFORE start of frame, leading to bogus frames that may be transmitted.

- They update rcsum per the added/removed 4 bytes tag.
  Alas if data is originally after the vlan/eth headers, then these
  bytes were already pulled out of the csum.

OTOH calling skb_vlan_push/skb_vlan_pop with skb->data at mac_header
present no issues.

act_vlan is the only caller to skb_vlan_*() that has skb->data pointing
at network header (upon ingress).
Other calles (ovs, bpf) already adjust skb->data at mac_header.

This patch fixes act_vlan to point to the mac_header prior calling
skb_vlan_*() functions, as other callers do.

Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Pravin Shelar <pshelar@ovn.org>
Cc: Jiri Pirko <jiri@mellanox.com>
---
 net/sched/act_vlan.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/net/sched/act_vlan.c b/net/sched/act_vlan.c
index 691409de3e..4ffc6c13a5 100644
--- a/net/sched/act_vlan.c
+++ b/net/sched/act_vlan.c
@@ -36,6 +36,12 @@ static int tcf_vlan(struct sk_buff *skb, const struct tc_action *a,
 	bstats_update(&v->tcf_bstats, skb);
 	action = v->tcf_action;
 
+	/* Ensure 'data' points at mac_header prior calling vlan manipulating
+	 * functions.
+	 */
+	if (skb_at_tc_ingress(skb))
+		skb_push_rcsum(skb, skb->mac_len);
+
 	switch (v->tcfv_action) {
 	case TCA_VLAN_ACT_POP:
 		err = skb_vlan_pop(skb);
@@ -57,6 +63,9 @@ drop:
 	action = TC_ACT_SHOT;
 	v->tcf_qstats.drops++;
 unlock:
+	if (skb_at_tc_ingress(skb))
+		skb_pull_rcsum(skb, skb->mac_len);
+
 	spin_unlock(&v->tcf_lock);
 	return action;
 }
-- 
2.7.4

[PATCH v3 net 2/2] net: skbuff: Limit skb_vlan_pop/push() to expect skb->data at mac header

[Shmulik Ladkani]

skb_vlan_pop/push were too generic, trying to support the cases where
skb->data is at mac header, and cases where skb->data is arbitrarily
elsewhere.

Supporting an arbitrary skb->data was complex and bogus:
 - It failed to unwind skb->data to its original location post actual
   pop/push.
   (Also, semantic is not well defined for unwinding: If data was into
    the eth header, need to use same offset from start; But if data was
    at network header or beyond, need to adjust the original offset
    according to the push/pull)
 - It mangled the rcsum post actual push/pop, without taking into account
   that the eth bytes might already have been pulled out of the csum.

Most callers (ovs, bpf) already had their skb->data at mac_header upon
invoking skb_vlan_pop/push.
Last caller that failed to do so (act_vlan) has been recently fixed.

Therefore, to simplify things, no longer support arbitrary skb->data
inputs for skb_vlan_pop/push().

skb->data is expected to be exactly at mac_header; WARN otherwise.

Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Pravin Shelar <pshelar@ovn.org>
Cc: Jiri Pirko <jiri@mellanox.com>
---
 v3: Instead of correcting unwinding of skb->data in skb_vlan_pop/push,
     just kill the support for arbitraray skb->data inputs, and assume
     given skb->data always points at mac_header.
     Fix act_vlan, the sole user not adehering to this assumption.

 v2: Instead of reducing mac_len by 4 bytes, which was found incorrect,
     fix the problem of wrong unwinding of 'skb->data'

 net/core/skbuff.c | 37 ++++++++++++++++++++++---------------
 1 file changed, 22 insertions(+), 15 deletions(-)

diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 3864b4b68f..8c38263cdf 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -4478,13 +4478,18 @@ EXPORT_SYMBOL(skb_ensure_writable);
 static int __skb_vlan_pop(struct sk_buff *skb, u16 *vlan_tci)
 {
 	struct vlan_hdr *vhdr;
-	unsigned int offset = skb->data - skb_mac_header(skb);
+	int offset = skb->data - skb_mac_header(skb);
 	int err;
 
-	__skb_push(skb, offset);
+	if (WARN_ONCE(offset,
+		      "__skb_vlan_pop got skb with skb->data not at mac header (offset %d)\n",
+		      offset)) {
+		return -EINVAL;
+	}
+
 	err = skb_ensure_writable(skb, VLAN_ETH_HLEN);
 	if (unlikely(err))
-		goto pull;
+		return err;
 
 	skb_postpull_rcsum(skb, skb->data + (2 * ETH_ALEN), VLAN_HLEN);
 
@@ -4501,12 +4506,13 @@ static int __skb_vlan_pop(struct sk_buff *skb, u16 *vlan_tci)
 		skb_set_network_header(skb, ETH_HLEN);
 
 	skb_reset_mac_len(skb);
-pull:
-	__skb_pull(skb, offset);
 
 	return err;
 }
 
+/* Pop a vlan tag either from hwaccel or from payload.
+ * Expects skb->data at mac header.
+ */
 int skb_vlan_pop(struct sk_buff *skb)
 {
 	u16 vlan_tci;
@@ -4541,29 +4547,30 @@ int skb_vlan_pop(struct sk_buff *skb)
 }
 EXPORT_SYMBOL(skb_vlan_pop);
 
+/* Push a vlan tag either into hwaccel or into payload (if hwaccel tag present).
+ * Expects skb->data at mac header.
+ */
 int skb_vlan_push(struct sk_buff *skb, __be16 vlan_proto, u16 vlan_tci)
 {
 	if (skb_vlan_tag_present(skb)) {
-		unsigned int offset = skb->data - skb_mac_header(skb);
+		int offset = skb->data - skb_mac_header(skb);
 		int err;
 
-		/* __vlan_insert_tag expect skb->data pointing to mac header.
-		 * So change skb->data before calling it and change back to
-		 * original position later
-		 */
-		__skb_push(skb, offset);
+		if (WARN_ONCE(offset,
+			      "skb_vlan_push got skb with skb->data not at mac header (offset %d)\n",
+			      offset)) {
+			return -EINVAL;
+		}
+
 		err = __vlan_insert_tag(skb, skb->vlan_proto,
 					skb_vlan_tag_get(skb));
-		if (err) {
-			__skb_pull(skb, offset);
+		if (err)
 			return err;
-		}
 
 		skb->protocol = skb->vlan_proto;
 		skb->mac_len += VLAN_HLEN;
 
 		skb_postpush_rcsum(skb, skb->data + (2 * ETH_ALEN), VLAN_HLEN);
-		__skb_pull(skb, offset);
 	}
 	__vlan_hwaccel_put_tag(skb, vlan_proto, vlan_tci);
 	return 0;
-- 
2.7.4

Re: [PATCH v3 net 1/2] net/sched: act_vlan: Push skb->data to mac_header prior calling skb_vlan_*() functions

[Shmulik Ladkani]

David,

On Thu, 29 Sep 2016 12:10:40 +0300 Shmulik Ladkani <shmulik.ladkani@gmail.com> wrote:
> This patch fixes act_vlan to point to the mac_header prior calling
> skb_vlan_*() functions, as other callers do.
> 

This 1/2 patch fixes the problem detailed in [1] for act_vlan,
last known caller of skb_vlan_*() with skb->data not at mac_header.

I think it's a good candidate for -stable; it fixes the observed bug and
it is rather focused.

Subsequent 2/2 patch hermetically seals the future possibility that one
might call skb_vlan_*() with skb->data not at mac_header.

This might go to stable as well, but not strictly required.

Thanks,
Shmulik

[1] https://patchwork.ozlabs.org/patch/676111/

Re: [PATCH v3 net 1/2] net/sched: act_vlan: Push skb->data to mac_header prior calling skb_vlan_*() functions

[David Miller]

From: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Date: Thu, 29 Sep 2016 12:10:40 +0300

> Generic skb_vlan_push/skb_vlan_pop functions don't properly handle the
> case where the input skb data pointer does not point at the mac header:
> 
> - They're doing push/pop, but fail to properly unwind data back to its
>   original location.
>   For example, in the skb_vlan_push case, any subsequent
>   'skb_push(skb, skb->mac_len)' calls make the skb->data point 4 bytes
>   BEFORE start of frame, leading to bogus frames that may be transmitted.
> 
> - They update rcsum per the added/removed 4 bytes tag.
>   Alas if data is originally after the vlan/eth headers, then these
>   bytes were already pulled out of the csum.
> 
> OTOH calling skb_vlan_push/skb_vlan_pop with skb->data at mac_header
> present no issues.
> 
> act_vlan is the only caller to skb_vlan_*() that has skb->data pointing
> at network header (upon ingress).
> Other calles (ovs, bpf) already adjust skb->data at mac_header.
> 
> This patch fixes act_vlan to point to the mac_header prior calling
> skb_vlan_*() functions, as other callers do.
> 
> Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>

Applied and queued up for -stable.

Re: [PATCH v3 net 2/2] net: skbuff: Limit skb_vlan_pop/push() to expect skb->data at mac header

[David Miller]

From: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Date: Thu, 29 Sep 2016 12:10:41 +0300

> skb_vlan_pop/push were too generic, trying to support the cases where
> skb->data is at mac header, and cases where skb->data is arbitrarily
> elsewhere.
> 
> Supporting an arbitrary skb->data was complex and bogus:
>  - It failed to unwind skb->data to its original location post actual
>    pop/push.
>    (Also, semantic is not well defined for unwinding: If data was into
>     the eth header, need to use same offset from start; But if data was
>     at network header or beyond, need to adjust the original offset
>     according to the push/pull)
>  - It mangled the rcsum post actual push/pop, without taking into account
>    that the eth bytes might already have been pulled out of the csum.
> 
> Most callers (ovs, bpf) already had their skb->data at mac_header upon
> invoking skb_vlan_pop/push.
> Last caller that failed to do so (act_vlan) has been recently fixed.
> 
> Therefore, to simplify things, no longer support arbitrary skb->data
> inputs for skb_vlan_pop/push().
> 
> skb->data is expected to be exactly at mac_header; WARN otherwise.
> 
> Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>

Applied.