Re: [PATCH v3 4/5] net: dsa: add support for Atheros AR9331 TAG format

On Mon, Oct 21, 2019 at 05:49:00PM +0200, Andrew Lunn wrote:

quoted

+static struct sk_buff *ar9331_tag_rcv(struct sk_buff *skb,
+				      struct net_device *ndev,
+				      struct packet_type *pt)
+{
+	u8 ver, port;
+	u16 hdr;
+
+	if (unlikely(!pskb_may_pull(skb, AR9331_HDR_LEN)))
+		return NULL;
+
+	hdr = le16_to_cpu(*(__le16 *)skb_mac_header(skb));
+
+	ver = FIELD_GET(AR9331_HDR_VERSION_MASK, hdr);
+	if (unlikely(ver != AR9331_HDR_VERSION)) {
+		netdev_warn_once(ndev, "%s:%i wrong header version 0x%2x\n",
+				 __func__, __LINE__, hdr);
+		return NULL;
+	}
+
+	if (unlikely(hdr & AR9331_HDR_FROM_CPU)) {
+		netdev_warn_once(ndev, "%s:%i packet should not be from cpu 0x%2x\n",
+				 __func__, __LINE__, hdr);
+		return NULL;
+	}
+
+	skb_pull(skb, AR9331_HDR_LEN);
+	skb_set_mac_header(skb, -ETH_HLEN);

No other tag driver calls skb_set_mac_header().  Also, the -ETH_HLEN
looks odd, give you have just pulled off AR9331_HDR_LEN?

Hm.. is it corrected somewhere else? Any way, B.T.M.A.N need a proper
value ant it seems to work correctly. So I remove it.

What other tag drivers use is skb_pull_rcsum().

It is build in switch and internal Ethernet controller do not set csum. There is
nothing to recalculate... on other hand, it adds no overhead. So, I have
nothing against it. Are there other arguments?

Regards,
Oleksij
-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |