On 13.3.2026 23:48, Florian Fainelli wrote:

On 3/13/26 02:20, Nicolai Buchwitz wrote:

quoted

Add XDP program attachment via ndo_bpf and execute XDP programs in the
RX path. Supported actions:

   - XDP_PASS: build SKB from the (possibly modified) xdp_buff and 
pass
     to the network stack, handling xdp_adjust_head/tail correctly
   - XDP_DROP: return the page to page_pool, no SKB allocated
   - XDP_ABORTED: same as DROP with trace_xdp_exception

XDP_TX and XDP_REDIRECT are not yet supported and will return
XDP_ABORTED.

The XDP hook runs after the HW error checks but before SKB 
construction,
so dropped packets avoid all SKB allocation overhead.

Advertise NETDEV_XDP_ACT_BASIC in xdp_features.

Signed-off-by: Nicolai Buchwitz <redacted>
---

[snip]

quoted

@@ -2345,12 +2397,6 @@ static unsigned int bcmgenet_desc_rx(struct 

bcmgenet_rx_ring *ring,
  		hard_start = page_address(rx_page) + rx_off;
  		status = (struct status_64 *)hard_start;
  		dma_length_status = status->length_status;
-		if (dev->features & NETIF_F_RXCSUM) {
-			rx_csum = (__force __be16)(status->rx_csum & 0xffff);
-			if (rx_csum) {
-				/* defer csum setup to after skb is built */
-			}
-		}

Did you intend for that hunk to be deleted?

Yes, (kinda) intentional. Patch 1 moved the csum setup after SKB 
construction
(since we no longer have an SKB at that point), but left the early read
as a dead no-op - patch 3 then removed it. I'll clean this up in v2 by
dropping the dead block directly in patch 1 instead.

Nicolai