mv643xx_eth.c: Multicast and IPv6 support

Hello,

currently, the driver mv643xx_eth.c doesn't support receiving of multicast and 
sending of IPv6 packets.
I found a piece of code in the u-boot bootloader which supports multicast 
receiving and did some cut and paste. Now receiving multicast packets works 
for me.
To be able to send IPv6 packets I had to remove the NETIF_F_HW_CSUM flag. 
Hardware checksumming for IPv4 should continue to work because of the 
NETIF_F_IP_CSUM flag.

I attach a diff against 2.6.15.

Does anyone know how to get documentation for the marvell mv643xx-family?

Regards
 Wolfram

In article [off-list ref]
pegasos@frokaschwei.de writes:

currently, the driver mv643xx_eth.c doesn't support receiving of multicast and 
sending of IPv6 packets.
I found a piece of code in the u-boot bootloader which supports multicast 
receiving and did some cut and paste. Now receiving multicast packets works 
for me.
To be able to send IPv6 packets I had to remove the NETIF_F_HW_CSUM flag. 
Hardware checksumming for IPv4 should continue to work because of the 
NETIF_F_IP_CSUM flag.

I attach a diff against 2.6.15.

Thanks.  On December 28th, I posted a patch to netdev@vger.kernel.org that
adds multicast support to the mv643xx driver.  I'll ping jgarzik
about it.

The IPv6 changes look good.  Would you please re-submit these without
the multicast changes and with a Signed-off-by line?

-Dale

Hello,

quoted

currently, the driver mv643xx_eth.c doesn't support receiving of

Thanks.  On December 28th, I posted a patch to netdev@vger.kernel.org that
adds multicast support to the mv643xx driver.  I'll ping jgarzik
about it.

Oh, didn't find that patch yesterday.

The IPv6 changes look good.  Would you please re-submit these without
the multicast changes and with a Signed-off-by line?

Here it comes. However, IPv6 uses multicasting for address resolution
(replacement for ARP), so IPv6 doesn't work until the multicast patch is applied, too.

Wolfram.

---
This patch removes the NETIF_F_HW_CSUM flag to be able to use other protocols
than IPv4. Hardware checksums for IPv4 should continue to work.
The sanity-check has been enhanced to check the used protocol and to not
access skb->iph for non-ipv4-packets.

Signed-off-by: Wolfram Joost <redacted>
---
diff -up linux-2.6.15/drivers/net/mv643xx_eth.c linux-2.6.15.patched/drivers/net/mv643xx_eth.c

--- linux-2.6.15/drivers/net/mv643xx_eth.c	2006-01-03 04:21:10.000000000 +0100
+++ linux-2.6.15.patched/drivers/net/mv643xx_eth.c	2006-01-12 21:52:32.000000000 +0100

@@ -1151,25 +1151,28 @@ linear:
 			pkt_info.l4i_chk = 0;
 		} else {
 
-			pkt_info.cmd_sts = ETH_TX_ENABLE_INTERRUPT |
-					   ETH_TX_FIRST_DESC |
-					   ETH_TX_LAST_DESC |
-					   ETH_GEN_TCP_UDP_CHECKSUM |
-					   ETH_GEN_IP_V_4_CHECKSUM |
-					   skb->nh.iph->ihl << ETH_TX_IHL_SHIFT;
 			/* CPU already calculated pseudo header checksum. */
-			if (skb->nh.iph->protocol == IPPROTO_UDP) {
-				pkt_info.cmd_sts |= ETH_UDP_FRAME;
+			if ( (skb->protocol == ETH_P_IP) &&
+			     (skb->nh.iph->protocol == IPPROTO_UDP) ) {
+				pkt_info.cmd_sts = ETH_UDP_FRAME;
 				pkt_info.l4i_chk = skb->h.uh->check;
-			} else if (skb->nh.iph->protocol == IPPROTO_TCP)
+			} else if ( (skb->protocol == ETH_P_IP) &&
+				    (skb->nh.iph->protocol == IPPROTO_TCP) ) {
+				pkt_info.cmd_sts = 0;
 				pkt_info.l4i_chk = skb->h.th->check;
-			else {
+			} else {
 				printk(KERN_ERR
-					"%s: chksum proto != TCP or UDP\n",
+					"%s: chksum proto != IPv4 TCP or UDP\n",
 					dev->name);
 				spin_unlock_irqrestore(&mp->lock, flags);
 				return 1;
 			}
+			pkt_info.cmd_sts |= ETH_TX_ENABLE_INTERRUPT |
+					    ETH_TX_FIRST_DESC |
+					    ETH_TX_LAST_DESC |
+					    ETH_GEN_TCP_UDP_CHECKSUM |
+					    ETH_GEN_IP_V_4_CHECKSUM |
+					    skb->nh.iph->ihl << ETH_TX_IHL_SHIFT;
 		}
 		pkt_info.byte_cnt = skb->len;
 		pkt_info.buf_ptr = dma_map_single(NULL, skb->data, skb->len,

@@ -1216,23 +1219,26 @@ linear:
 			pkt_info.cmd_sts = ETH_TX_FIRST_DESC |
 					   5 << ETH_TX_IHL_SHIFT;
 		else {
-			pkt_info.cmd_sts = ETH_TX_FIRST_DESC |
-					   ETH_GEN_TCP_UDP_CHECKSUM |
-					   ETH_GEN_IP_V_4_CHECKSUM |
-					   skb->nh.iph->ihl << ETH_TX_IHL_SHIFT;
 			/* CPU already calculated pseudo header checksum. */
-			if (skb->nh.iph->protocol == IPPROTO_UDP) {
-				pkt_info.cmd_sts |= ETH_UDP_FRAME;
+			if ( (skb->protocol == ETH_P_IP) &&
+			     (skb->nh.iph->protocol == IPPROTO_UDP) ) {
+				pkt_info.cmd_sts = ETH_UDP_FRAME;
 				pkt_info.l4i_chk = skb->h.uh->check;
-			} else if (skb->nh.iph->protocol == IPPROTO_TCP)
+			} else if ( (skb->protocol == ETH_P_IP) &&
+				    (skb->nh.iph->protocol == IPPROTO_TCP) ) {
+				pkt_info.cmd_sts = 0;
 				pkt_info.l4i_chk = skb->h.th->check;
-			else {
+			} else {
 				printk(KERN_ERR
-					"%s: chksum proto != TCP or UDP\n",
+					"%s: chksum proto != IPv4 TCP or UDP\n",
 					dev->name);
 				spin_unlock_irqrestore(&mp->lock, flags);
 				return 1;
 			}
+			pkt_info.cmd_sts |= ETH_TX_FIRST_DESC |
+					    ETH_GEN_TCP_UDP_CHECKSUM |
+					    ETH_GEN_IP_V_4_CHECKSUM |
+					    skb->nh.iph->ihl << ETH_TX_IHL_SHIFT;
 		}
 
 		status = eth_port_send(mp, &pkt_info);

@@ -1441,7 +1447,7 @@ static int mv643xx_eth_probe(struct plat
 	 * Zero copy can only work if we use Discovery II memory. Else, we will
 	 * have to map the buffers to ISA memory which is only 16 MB
 	 */
-	dev->features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_HW_CSUM;
+	dev->features = NETIF_F_SG | NETIF_F_IP_CSUM;
 #endif
 #endif
 

On Thu, Jan 12, 2006 at 10:16:18PM +0100, Wolfram Joost wrote:

quoted

quoted

currently, the driver mv643xx_eth.c doesn't support receiving of

quoted

Thanks.  On December 28th, I posted a patch to netdev@vger.kernel.org that
adds multicast support to the mv643xx driver.  I'll ping jgarzik
about it.

Oh, didn't find that patch yesterday.

quoted

The IPv6 changes look good.  Would you please re-submit these without
the multicast changes and with a Signed-off-by line?

Here it comes. However, IPv6 uses multicasting for address resolution
(replacement for ARP), so IPv6 doesn't work until the multicast patch is applied, too.

Thanks Wolfram.  I'll put it in my queue for testing and if all goes well,
I'll forward it to Jeff Garzik this weekend.

-Dale