Skip to contentSkip to conversation

xfrm: do not set IPv4 DF flag when encapsulating IPv6 frames <= 1280 bytes.

PATCH Applied Maciej Żenczykowski 2 participants 5 messages May 18, 2022 – May 30, 2022

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

PATCH 1/1 xfrm: do not set IPv4 DF flag when encapsulating IPv6 frames <= 1280 bytes.

From: Maciej Żenczykowski <[email protected]>

One may want to have DF set on large packets to support discovering path mtu and limiting the size of generated packets (hence not setting the XFRM_STATE_NOPMTUDISC tunnel flag), while still supporting networks that are incapable of carrying even minimal sized IPv6 frames (post encapsulation).

Having IPv4 Don't Frag bit set on encapsulated IPv6 frames that are not larger than the minimum IPv6 mtu of 1280 isn't useful, because the resulting ICMP Fragmentation Required error isn't actionable (even assuming you receive it) because IPv6 will not drop it's path mtu below 1280 anyway. While the IPv4 stack could prefrag the packets post encap, this requires the ICMP error to be successfully delivered and causes a loss of the original IPv6 frame (thus requiring a retransmit and latency hit). Luckily with IPv4 if we simply don't set the DF flag, we'll just make further fragmenting the packets some other router's problems.

We'll still learn the correct IPv4 path mtu through encapsulation of larger IPv6 frames.

I'm still not convinced this patch is entirely sufficient to make everything happy... but I don't see how it could possibly make things worse.

See also recent:
4ff2980b6bd2 'xfrm: fix tunnel model fragmentation behavior'
and friends

Bug 203183943
Cc Lorenzo Colitti <[email protected]>
Cc Eric Dumazet <[email protected]>
Cc Lina Wang <[email protected]>
Cc Steffen Klassert <[email protected]>
---
 net/xfrm/xfrm_output.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/xfrm/xfrm_output.c +2 -1
index d4935b3b9983..555ab35cd119 100644
@@ -273,6 +273,7 @@ static int xfrm4_beet_encap_add(struct xfrm_state *x, struct sk_buff *skb)
  */
 static int xfrm4_tunnel_encap_add(struct xfrm_state *x, struct sk_buff *skb)
 {
+	bool small_ipv6 = (skb->protocol == htons(ETH_P_IPV6)) && (skb->len <= IPV6_MIN_MTU);
 	struct dst_entry *dst = skb_dst(skb);
 	struct iphdr *top_iph;
 	int flags;
@@ -303,7 +304,7 @@ static int xfrm4_tunnel_encap_add(struct xfrm_state *x, struct sk_buff *skb)
 	if (flags & XFRM_STATE_NOECN)
 		IP_ECN_clear(top_iph);
 
-	top_iph->frag_off = (flags & XFRM_STATE_NOPMTUDISC) ?
+	top_iph->frag_off = (flags & XFRM_STATE_NOPMTUDISC) || small_ipv6 ?
 		0 : (XFRM_MODE_SKB_CB(skb)->frag_off & htons(IP_DF));
 
 	top_iph->ttl = ip4_dst_hoplimit(xfrm_dst_child(dst));
Signature · 1 line
2.36.1.124.g0e6072fb45-goog
maintainer
#1 lore raw
Quoted 27 lines · Maciej Żenczykowski

From: Maciej Żenczykowski <[email protected]>

One may want to have DF set on large packets to support discovering path mtu and limiting the size of generated packets (hence not setting the XFRM_STATE_NOPMTUDISC tunnel flag), while still supporting networks that are incapable of carrying even minimal sized IPv6 frames (post encapsulation).

Having IPv4 Don't Frag bit set on encapsulated IPv6 frames that are not larger than the minimum IPv6 mtu of 1280 isn't useful, because the resulting ICMP Fragmentation Required error isn't actionable (even assuming you receive it) because IPv6 will not drop it's path mtu below 1280 anyway. While the IPv4 stack could prefrag the packets post encap, this requires the ICMP error to be successfully delivered and causes a loss of the original IPv6 frame (thus requiring a retransmit and latency hit). Luckily with IPv4 if we simply don't set the DF flag, we'll just make further fragmenting the packets some other router's problems.

We'll still learn the correct IPv4 path mtu through encapsulation of larger IPv6 frames.

I'm still not convinced this patch is entirely sufficient to make everything happy... but I don't see how it could possibly make things worse.

Quoted 5 lines

See also recent:
4ff2980b6bd2 'xfrm: fix tunnel model fragmentation behavior'
and friends

Bug: 203183943

To what does this bug number refer to? Bugzilla? Please make that clear if you want to have this number in the commit message.

Thanks!

maintainer Applied
#2 lore raw
In reply to Maciej Żenczykowski
Quoted 31 lines · Maciej Żenczykowski

From: Maciej Żenczykowski <[email protected]>

One may want to have DF set on large packets to support discovering path mtu and limiting the size of generated packets (hence not setting the XFRM_STATE_NOPMTUDISC tunnel flag), while still supporting networks that are incapable of carrying even minimal sized IPv6 frames (post encapsulation).

Having IPv4 Don't Frag bit set on encapsulated IPv6 frames that are not larger than the minimum IPv6 mtu of 1280 isn't useful, because the resulting ICMP Fragmentation Required error isn't actionable (even assuming you receive it) because IPv6 will not drop it's path mtu below 1280 anyway. While the IPv4 stack could prefrag the packets post encap, this requires the ICMP error to be successfully delivered and causes a loss of the original IPv6 frame (thus requiring a retransmit and latency hit). Luckily with IPv4 if we simply don't set the DF flag, we'll just make further fragmenting the packets some other router's problems.

We'll still learn the correct IPv4 path mtu through encapsulation of larger IPv6 frames.

I'm still not convinced this patch is entirely sufficient to make everything happy... but I don't see how it could possibly make things worse.

See also recent:
4ff2980b6bd2 'xfrm: fix tunnel model fragmentation behavior'
and friends

Quoted 6 lines
Bug 203183943
Cc Lorenzo Colitti <[email protected]>
Cc Eric Dumazet <[email protected]>
Cc Lina Wang <[email protected]>
Cc Steffen Klassert <[email protected]>

Applied, thanks a lot!

#3 lore raw
Quoted 33 lines · Steffen Klassert
Quoted 31 lines · Maciej Żenczykowski

From: Maciej Żenczykowski <[email protected]>

One may want to have DF set on large packets to support discovering path mtu and limiting the size of generated packets (hence not setting the XFRM_STATE_NOPMTUDISC tunnel flag), while still supporting networks that are incapable of carrying even minimal sized IPv6 frames (post encapsulation).

Having IPv4 Don't Frag bit set on encapsulated IPv6 frames that are not larger than the minimum IPv6 mtu of 1280 isn't useful, because the resulting ICMP Fragmentation Required error isn't actionable (even assuming you receive it) because IPv6 will not drop it's path mtu below 1280 anyway. While the IPv4 stack could prefrag the packets post encap, this requires the ICMP error to be successfully delivered and causes a loss of the original IPv6 frame (thus requiring a retransmit and latency hit). Luckily with IPv4 if we simply don't set the DF flag, we'll just make further fragmenting the packets some other router's problems.

We'll still learn the correct IPv4 path mtu through encapsulation of larger IPv6 frames.

I'm still not convinced this patch is entirely sufficient to make everything happy... but I don't see how it could possibly make things worse.

See also recent:
4ff2980b6bd2 'xfrm: fix tunnel model fragmentation behavior'
and friends

Quoted 8 lines
Quoted 6 lines
Bug 203183943
Cc Lorenzo Colitti <[email protected]>
Cc Eric Dumazet <[email protected]>
Cc Lina Wang <[email protected]>
Cc Steffen Klassert <[email protected]>

Applied, thanks a lot!

Thanks.

Is this published somewhere, since I'd lack to backport it to Android Common Kernel 5.10+, but can't find a sha1 (yet?)

maintainer
#4 lore raw
Quoted 42 lines · Maciej Żenczykowski
Quoted 40 lines · Steffen Klassert
Quoted 37 lines · Maciej Żenczykowski

From: Maciej Żenczykowski <[email protected]>

One may want to have DF set on large packets to support discovering path mtu and limiting the size of generated packets (hence not setting the XFRM_STATE_NOPMTUDISC tunnel flag), while still supporting networks that are incapable of carrying even minimal sized IPv6 frames (post encapsulation).

Having IPv4 Don't Frag bit set on encapsulated IPv6 frames that are not larger than the minimum IPv6 mtu of 1280 isn't useful, because the resulting ICMP Fragmentation Required error isn't actionable (even assuming you receive it) because IPv6 will not drop it's path mtu below 1280 anyway. While the IPv4 stack could prefrag the packets post encap, this requires the ICMP error to be successfully delivered and causes a loss of the original IPv6 frame (thus requiring a retransmit and latency hit). Luckily with IPv4 if we simply don't set the DF flag, we'll just make further fragmenting the packets some other router's problems.

We'll still learn the correct IPv4 path mtu through encapsulation of larger IPv6 frames.

I'm still not convinced this patch is entirely sufficient to make everything happy... but I don't see how it could possibly make things worse.

See also recent:
4ff2980b6bd2 'xfrm: fix tunnel model fragmentation behavior'
and friends

Bug 203183943
Cc Lorenzo Colitti <[email protected]>
Cc Eric Dumazet <[email protected]>
Cc Lina Wang <[email protected]>
Cc Steffen Klassert <[email protected]>
Quoted 6 lines
Quoted 1 line

Applied, thanks a lot!

Thanks.

Is this published somewhere, since I'd lack to backport it to Android Common Kernel 5.10+, but can't find a sha1 (yet?)

Actually I applied it, but forgot to push it out.
It is now in the ipsec tree:

git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec.git