On 28.01.26 17:12, Jakub Kicinski wrote:

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.

For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
can: gw: use can_gw_hops instead of sk_buff::csum_start

This patch moves the hop counter storage from skb->csum_start to the CAN
skb extension field can_skb_ext::can_gw_hops.

quoted

diff --git a/net/can/gw.c b/net/can/gw.c
index ad89a1913b34..2c134c9e5159 100644
--- a/net/can/gw.c
+++ b/net/can/gw.c

[ ... ]

quoted

@@ -480,19 +480,8 @@ static void can_can_gw_rcv(struct sk_buff *skb, void *data)
  	/* Do not handle CAN frames routed more than 'max_hops' times.
  	 * In general we should never catch this delimiter which is intended
  	 * to cover a misconfiguration protection (e.g. circular CAN routes).
-	 *
-	 * The Controller Area Network controllers only accept CAN frames with
-	 * correct CRCs - which are not visible in the controller registers.
-	 * According to skbuff.h documentation the csum_start element for IP
-	 * checksums is undefined/unused when ip_summed == CHECKSUM_UNNECESSARY.
-	 * Only CAN skbs can be processed here which already have this property.
  	 */
-
-#define cgw_hops(skb) ((skb)->csum_start)
-
-	BUG_ON(skb->ip_summed != CHECKSUM_UNNECESSARY);
-
-	if (cgw_hops(skb) >= max_hops) {
+	if (csx->can_gw_hops >= max_hops) {

             ^^^^^^^^^^^^^^^^
Is csx->can_gw_hops initialized to zero for fresh CAN frames?

The old code used skb->csum_start which was zeroed by skbuff_clear() during
skb allocation. The new code reads from can_skb_ext::can_gw_hops, but looking
at alloc_can_skb() in drivers/net/can/dev/skb.c, only can_iif is initialized:

     csx = can_skb_ext_add(skb);
     ...
     csx->can_iif = dev->ifindex;

The skb extension memory comes from kmem_cache_alloc() without __GFP_ZERO in
__skb_ext_alloc(), so can_gw_hops will contain garbage from the slab.

Correct. It is not only a leftover of using csum_start but also of my 
first implementation attempt using another space in struct sk_buff which 
was also zero-initialized.

I'll add some initialization in can_skb_ext_add() for the next respin.

Excellent feedback!

Many thanks,
Oliver

With default max_hops=1, if the garbage value is >= 1 (probability ~99.6%),
the frame will be dropped at this check before any routing occurs.

Should alloc_can_skb(), alloc_canfd_skb(), and alloc_canxl_skb() initialize
csx->can_gw_hops = 0?

quoted

  		/* indicate deleted frames due to misconfiguration */
  		gwj->deleted_frames++;
  		return;

[ ... ]