Re: [PATCH] NET: SRv6: seg6_local: Fixed SRH processing when segments left is 0

Hi Suprit, 

As you see the in the pseudocode, if Segments Left = 0, the End function will 
stop processing the SRH and will process upper layer protocols for OAM (e.g., ICMP). 

The packet will not be forwarded further. 

What you propose here is a different thing, you want to decap the SRv6 
headers and forward the packet further if Segments Left = 0.

This is not a correct symantic of SRv6 End Behavior.   


But if you need such behavior in your network, it already defined by another 
flavor of the End function called Ultimate Segment Decapsulation (USD) [2]

Please feel free to submit a patch to support this USD flavor of END.  

[2] https://tools.ietf.org/html/draft-ietf-spring-srv6-network-programming-28#section-4.16.3

Thanks
Ahmed
 


On Tue, 2 Feb 2021 16:50:58 +0530
Suprit Japagal [off-list ref] wrote:

Hi,

As per the section 4.1 of [1]:

"When N receives a packet whose IPv6 DA is S and S is a local *End SID*,

   N does:

   S01. When an SRH is processed {
   S02.   If (Segments Left == 0) {
   S03.      Stop processing the SRH, and proceed to process the next
                header in the packet, whose type is identified by
                the Next Header field in the routing header.

   S04. }
"
S is the DA which is a local End SID and when SRH is processed, the
segments left field can be 0 as mentioned above.
Same goes for End.X SID and End.T SID as per section 4.2 and 4.3 of [1]
respectively.

Packets processed by End, End.X and End.T behaviors can have a Segment Left
Value
of zero.

[1]https://tools.ietf.org/html/draft-ietf-spring-srv6-network-programming-28

Thanks,
Suprit J

On Tue, Feb 2, 2021 at 3:22 PM Ahmed Abdelsalam [off-list ref]
wrote:

quoted

The current implementation is correct. This patch is introducing incorrect
symantic to SRv6 End and End.T behaviors.

SRv6 End and End.T behaviors (among other behaviors) are defined in the
SRv6 Network Programming draft (Soon to published as an RFC) [1].

SRv6 End and End.T behaviors are used to implement Traffic Engineering (TE)
use-cases, where a node recieves the packet and send it to the next SID
from
the SRH SIDList.

Packets processed by End and End.T behaviors can not have a Segment Left
Value
of zero.

Please refer to sections 4.1 and 4.3 of [1].


[1]
https://tools.ietf.org/html/draft-ietf-spring-srv6-network-programming-28

Ahmed


On Sun, 31 Jan 2021 10:33:14 -0700
David Ahern [off-list ref] wrote:

quoted

[ cc David Lebrun, author of the original code ]

On 1/31/21 6:08 AM, Suprit Japagal wrote:

quoted

From: "Suprit.Japagal" <redacted>

According to the standard IETF RFC 8754, section 4.3.1.1
(https://tools.ietf.org/html/rfc8754#section-4.3.1.1)
When the segments left in SRH equals to 0, proceed to process the
next header in the packet, whose type is identified by the
Next header field of the routing header.

Signed-off-by: Suprit.Japagal <redacted>
---
 net/ipv6/seg6_local.c | 54

+++++++++++++++++++++++++++++++++++++++++++++------

quoted

quoted

 1 file changed, 48 insertions(+), 6 deletions(-)

diff --git a/net/ipv6/seg6_local.c b/net/ipv6/seg6_local.c
index b07f7c1..b17f9dc 100644
--- a/net/ipv6/seg6_local.c
+++ b/net/ipv6/seg6_local.c

@@ -273,11 +273,25 @@ static int input_action_end(struct sk_buff *skb,

struct seg6_local_lwt *slwt)

quoted

quoted

 {
    struct ipv6_sr_hdr *srh;

-   srh = get_and_validate_srh(skb);
+   srh = get_srh(skb);
    if (!srh)
            goto drop;

-   advance_nextseg(srh, &ipv6_hdr(skb)->daddr);
+#ifdef CONFIG_IPV6_SEG6_HMAC
+   if (srh->segments_left > 0)
+           if (!seg6_hmac_validate_skb(skb))
+                   goto drop;
+#endif
+
+   if (srh->segments_left == 0) {
+           if (!decap_and_validate(skb, srh->nexthdr))
+                   goto drop;
+
+           if (!pskb_may_pull(skb, sizeof(struct ipv6hdr)))
+                   goto drop;
+   } else {
+           advance_nextseg(srh, &ipv6_hdr(skb)->daddr);
+   }

    seg6_lookup_nexthop(skb, NULL, 0);

@@ -293,11 +307,25 @@ static int input_action_end_x(struct sk_buff

*skb, struct seg6_local_lwt *slwt)

quoted

quoted

 {
    struct ipv6_sr_hdr *srh;

-   srh = get_and_validate_srh(skb);
+   srh = get_srh(skb);
    if (!srh)
            goto drop;

-   advance_nextseg(srh, &ipv6_hdr(skb)->daddr);
+#ifdef CONFIG_IPV6_SEG6_HMAC
+   if (srh->segments_left > 0)
+           if (!seg6_hmac_validate_skb(skb))
+                   goto drop;
+#endif
+
+   if (srh->segments_left == 0) {
+           if (!decap_and_validate(skb, srh->nexthdr))
+                   goto drop;
+
+           if (!pskb_may_pull(skb, sizeof(struct ipv6hdr)))
+                   goto drop;
+   } else {
+           advance_nextseg(srh, &ipv6_hdr(skb)->daddr);
+   }

    seg6_lookup_nexthop(skb, &slwt->nh6, 0);

@@ -312,11 +340,25 @@ static int input_action_end_t(struct sk_buff

*skb, struct seg6_local_lwt *slwt)

quoted

quoted

 {
    struct ipv6_sr_hdr *srh;

-   srh = get_and_validate_srh(skb);
+   srh = get_srh(skb);
    if (!srh)
            goto drop;

-   advance_nextseg(srh, &ipv6_hdr(skb)->daddr);
+#ifdef CONFIG_IPV6_SEG6_HMAC
+   if (srh->segments_left > 0)
+           if (!seg6_hmac_validate_skb(skb))
+                   goto drop;
+#endif
+
+   if (srh->segments_left == 0) {
+           if (!decap_and_validate(skb, srh->nexthdr))
+                   goto drop;
+
+           if (!pskb_may_pull(skb, sizeof(struct ipv6hdr)))
+                   goto drop;
+   } else {
+           advance_nextseg(srh, &ipv6_hdr(skb)->daddr);
+   }

    seg6_lookup_nexthop(skb, NULL, slwt->table);

--
Ahmed Abdelsalam [off-list ref]

-- 
Ahmed Abdelsalam [off-list ref]