Skip to contentSkip to conversation

net: xdp: pull ethernet header off packet after computing skb->protocol

PATCH In review Jason A. Donenfeld 3 participants 5 messages Aug 15, 2020 – Aug 17, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net: xdp: pull ethernet header off packet after computing skb->protocol

When an XDP program changes the ethernet header protocol field, eth_type_trans is used to recalculate skb->protocol. In order for eth_type_trans to work correctly, the ethernet header must actually be part of the skb data segment, so the code first pushes that onto the head of the skb. However, it subsequently forgets to pull it back off, making the behavior of the passed-on packet inconsistent between the protocol modifying case and the static protocol case. This patch fixes the issue by simply pulling the ethernet header back off of the skb head.

Fixes 297249569932 ("net: fix generic XDP to handle if eth header was mangled")
Cc Jesper Dangaard Brouer <[email protected]>
Cc David S. Miller <[email protected]>
---
 net/core/dev.c | 1 +
 1 file changed, 1 insertion(+)
net/core/dev.c +1 -0
index 7df6c9617321..151f1651439f 100644
@@ -4676,6 +4676,7 @@ static u32 netif_receive_generic_xdp(struct sk_buff *skb,
 	    (orig_bcast != is_multicast_ether_addr_64bits(eth->h_dest))) {
 		__skb_push(skb, ETH_HLEN);
 		skb->protocol = eth_type_trans(skb, skb->dev);
+		__skb_pull(skb, ETH_HLEN);
 	}
 
 	switch (act) {
Signature · 1 line
2.28.0
maintainer
#1 lore raw

From: "Jason A. Donenfeld" <[email protected]>
Date: Sat, 15 Aug 2020 09:29:30 +0200

Quoted 10 lines

When an XDP program changes the ethernet header protocol field, eth_type_trans is used to recalculate skb->protocol. In order for eth_type_trans to work correctly, the ethernet header must actually be part of the skb data segment, so the code first pushes that onto the head of the skb. However, it subsequently forgets to pull it back off, making the behavior of the passed-on packet inconsistent between the protocol modifying case and the static protocol case. This patch fixes the issue by simply pulling the ethernet header back off of the skb head.

Quoted 4 lines
Fixes 297249569932 ("net: fix generic XDP to handle if eth header was mangled")
Cc Jesper Dangaard Brouer <[email protected]>
Cc David S. Miller <[email protected]>

Applied and queued up for -stable, thanks.

Jesper, I wonder how your original patch was tested because it pushes a packet with skb->data pointing at the ethernet header into the stack. That should be popped at this point as per this fix here.

Thanks.

#2 lore raw
Quoted 18 lines · On Sun, 16 Aug 2020 15:29:37 -0700 (PDT) David Miller <[email protected]>…

From: "Jason A. Donenfeld" <[email protected]>
Date: Sat, 15 Aug 2020 09:29:30 +0200

Quoted 10 lines

When an XDP program changes the ethernet header protocol field, eth_type_trans is used to recalculate skb->protocol. In order for eth_type_trans to work correctly, the ethernet header must actually be part of the skb data segment, so the code first pushes that onto the head of the skb. However, it subsequently forgets to pull it back off, making the behavior of the passed-on packet inconsistent between the protocol modifying case and the static protocol case. This patch fixes the issue by simply pulling the ethernet header back off of the skb head.

Quoted 4 lines
Fixes 297249569932 ("net: fix generic XDP to handle if eth header was mangled")
Cc Jesper Dangaard Brouer <[email protected]>
Cc David S. Miller <[email protected]>
Quoted 5 lines

Applied and queued up for -stable, thanks.

Jesper, I wonder how your original patch was tested because it pushes a packet with skb->data pointing at the ethernet header into the stack. That should be popped at this point as per this fix here.

I think this patch is wrong, because eth_type_trans() also does a skb_pull_inline(skb, ETH_HLEN).

To Jason, are you sure about this fix?
How did you test this?

I usually test with:

 $ cd tools/testing/selftests/bpf/
 $ sudo ./test_xdp_vlan_mode_generic.sh 

But currently I get a build error on net-next in:
net-next/tools/testing/selftests/bpf/tools/build/bpftool/pid_iter.bpf.o
So, I could not run this test.

- -
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer

Quoted 12 lines · On Sat, 15 Aug 2020 09:29:30 +0200 "Jason A. Donenfeld" <[email protected]> wrote:
net/core/dev.c +1 -0
index 7df6c9617321..151f1651439f 100644
@@ -4676,6 +4676,7 @@ static u32 netif_receive_generic_xdp(struct sk_buff *skb,
 	    (orig_bcast != is_multicast_ether_addr_64bits(eth->h_dest))) {
 		__skb_push(skb, ETH_HLEN);
 		skb->protocol = eth_type_trans(skb, skb->dev);
+		__skb_pull(skb, ETH_HLEN);
 	}
 
 	switch (act) {
#3 lore raw
Quoted 24 lines · Jesper Dangaard Brouer
Quoted 21 lines · On Sun, 16 Aug 2020 15:29:37 -0700 (PDT) David Miller <[email protected]>…

From: "Jason A. Donenfeld" <[email protected]>
Date: Sat, 15 Aug 2020 09:29:30 +0200

Quoted 10 lines

When an XDP program changes the ethernet header protocol field, eth_type_trans is used to recalculate skb->protocol. In order for eth_type_trans to work correctly, the ethernet header must actually be part of the skb data segment, so the code first pushes that onto the head of the skb. However, it subsequently forgets to pull it back off, making the behavior of the passed-on packet inconsistent between the protocol modifying case and the static protocol case. This patch fixes the issue by simply pulling the ethernet header back off of the skb head.

Quoted 5 lines
Fixes 297249569932 ("net: fix generic XDP to handle if eth header was

mangled")

Cc Jesper Dangaard Brouer <[email protected]>
Cc David S. Miller <[email protected]>

Applied and queued up for -stable, thanks.

Quoted 8 lines
Quoted 5 lines

Jesper, I wonder how your original patch was tested because it pushes a packet
with skb->data pointing at the ethernet header into the stack. That should be
popped at this point as per this fix here.

I think this patch is wrong, because eth_type_trans() also does a skb_pull_inline(skb, ETH_HLEN).

Huh, wow. That's one unusual and confusing function. But indeed it seems like I'm the one who needs to reevaluate testing methodology here. I'm very sorry for the noise and hassle.

Jason

maintainer
#4 lore raw

From: "Jason A. Donenfeld" <[email protected]>
Date: Mon, 17 Aug 2020 09:48:10 +0200

Quoted 31 lines · Jesper Dangaard Brouer
Quoted 30 lines
Quoted 26 lines · On Sun, 16 Aug 2020 15:29:37 -0700 (PDT) David Miller <[email protected]>…

From: "Jason A. Donenfeld" <[email protected]>
Date: Sat, 15 Aug 2020 09:29:30 +0200

Quoted 10 lines

When an XDP program changes the ethernet header protocol field, eth_type_trans is used to recalculate skb->protocol. In order for eth_type_trans to work correctly, the ethernet header must actually be part of the skb data segment, so the code first pushes that onto the head of the skb. However, it subsequently forgets to pull it back off, making the behavior of the passed-on packet inconsistent between the protocol modifying case and the static protocol case. This patch fixes the issue by simply pulling the ethernet header back off of the skb head.

Quoted 5 lines
Fixes 297249569932 ("net: fix generic XDP to handle if eth header was

mangled")

Cc Jesper Dangaard Brouer <[email protected]>
Cc David S. Miller <[email protected]>

Applied and queued up for -stable, thanks.

Jesper, I wonder how your original patch was tested because it pushes a packet
with skb->data pointing at the ethernet header into the stack. That should be
popped at this point as per this fix here.

Quoted 6 lines
Quoted 2 lines

I think this patch is wrong, because eth_type_trans() also does a skb_pull_inline(skb, ETH_HLEN).

Huh, wow. That's one unusual and confusing function. But indeed it seems like I'm the one who needs to reevaluate testing methodology here. I'm very sorry for the noise and hassle.

I've reverted this change from my tree.