diff mbox

[net-next] skbuff: only inherit relevant tx_flags

Message ID 20170608153503.74199-1-willemdebruijn.kernel@gmail.com
State Accepted, archived
Delegated to: David Miller
Headers show

Commit Message

Willem de Bruijn June 8, 2017, 3:35 p.m. UTC 
From: Willem de Bruijn <willemb@google.com>

When inheriting tx_flags from one skbuff to another, always apply a
mask to avoid overwriting unrelated other bits in the field.

The two SKBTX_SHARED_FRAG cases clears all other bits. In practice,
tx_flags are zero at this point now. But this is fragile. Timestamp
flags are set, for instance, if in tcp_gso_segment, after this clear
in skb_segment.

The SKBTX_ANY_TSTAMP mask in __skb_tstamp_tx ensures that new
skbs do not accidentally inherit flags such as SKBTX_SHARED_FRAG.

Signed-off-by: Willem de Bruijn <willemb@google.com>
---
 net/core/skbuff.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

Comments

Soheil Hassas Yeganeh June 8, 2017, 3:59 p.m. UTC | #1 
On Thu, Jun 8, 2017 at 11:35 AM, Willem de Bruijn
<willemdebruijn.kernel@gmail.com> wrote:
>
> From: Willem de Bruijn <willemb@google.com>
>
> When inheriting tx_flags from one skbuff to another, always apply a
> mask to avoid overwriting unrelated other bits in the field.
>
> The two SKBTX_SHARED_FRAG cases clears all other bits. In practice,
> tx_flags are zero at this point now. But this is fragile. Timestamp
> flags are set, for instance, if in tcp_gso_segment, after this clear
> in skb_segment.
>
> The SKBTX_ANY_TSTAMP mask in __skb_tstamp_tx ensures that new
> skbs do not accidentally inherit flags such as SKBTX_SHARED_FRAG.
>
> Signed-off-by: Willem de Bruijn <willemb@google.com>

Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Eric Dumazet June 8, 2017, 4:20 p.m. UTC | #2 
On Thu, 2017-06-08 at 11:35 -0400, Willem de Bruijn wrote:
> From: Willem de Bruijn <willemb@google.com>
> 
> When inheriting tx_flags from one skbuff to another, always apply a
> mask to avoid overwriting unrelated other bits in the field.
> 
> The two SKBTX_SHARED_FRAG cases clears all other bits. In practice,
> tx_flags are zero at this point now. But this is fragile. Timestamp
> flags are set, for instance, if in tcp_gso_segment, after this clear
> in skb_segment.
> 
> The SKBTX_ANY_TSTAMP mask in __skb_tstamp_tx ensures that new
> skbs do not accidentally inherit flags such as SKBTX_SHARED_FRAG.
> 
> Signed-off-by: Willem de Bruijn <willemb@google.com>


Acked-by: Eric Dumazet <edumazet@google.com>
diff mbox

Patch

diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 82cfc9c7a090..e508c1eae67f 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -2646,7 +2646,8 @@  void skb_split(struct sk_buff *skb, struct sk_buff *skb1, const u32 len)
 {
 	int pos = skb_headlen(skb);
 
-	skb_shinfo(skb1)->tx_flags = skb_shinfo(skb)->tx_flags & SKBTX_SHARED_FRAG;
+	skb_shinfo(skb1)->tx_flags |= skb_shinfo(skb)->tx_flags &
+				      SKBTX_SHARED_FRAG;
 	if (len < pos)	/* Split line is inside header. */
 		skb_split_inside_header(skb, skb1, len, pos);
 	else		/* Second chunk has no header, nothing to copy. */
@@ -3261,8 +3262,8 @@  struct sk_buff *skb_segment(struct sk_buff *head_skb,
 		skb_copy_from_linear_data_offset(head_skb, offset,
 						 skb_put(nskb, hsize), hsize);
 
-		skb_shinfo(nskb)->tx_flags = skb_shinfo(head_skb)->tx_flags &
-			SKBTX_SHARED_FRAG;
+		skb_shinfo(nskb)->tx_flags |= skb_shinfo(head_skb)->tx_flags &
+					      SKBTX_SHARED_FRAG;
 
 		while (pos < offset + len) {
 			if (i >= nfrags) {
@@ -3948,7 +3949,8 @@  void __skb_tstamp_tx(struct sk_buff *orig_skb,
 		return;
 
 	if (tsonly) {
-		skb_shinfo(skb)->tx_flags = skb_shinfo(orig_skb)->tx_flags;
+		skb_shinfo(skb)->tx_flags |= skb_shinfo(orig_skb)->tx_flags &
+					     SKBTX_ANY_TSTAMP;
 		skb_shinfo(skb)->tskey = skb_shinfo(orig_skb)->tskey;
 	}