Broadcom Gigabit 5703 and Bridging

Broadcom Gigabit 5703 and Bridging

[Paul Schulz]

Greetings,

I'm seeing 'TCP Checksum' Errors after packets pass through a host
bridging TCP packets with:

 - Kernel 2.4.20
 - Bridge code 
 - tg3 (Broadcom Gigabit 5703)

   eth0: Tigon3 [partno(BCM95703A30) rev 1002 PHY(5703)] 
    (PCIX:100MHz:64-bit) 10/100/1000BaseT
  (eth1 is similar)
  
Pings (ICMP) pass through the bridge just fine.

The hardware is IBM's xSeries 305.

Paul
--
                Paul Schulz - Software Engineer [codito, ergo sum]
        Foursticks Pty Ltd - 2/259 Glen Osmond Road, Frewville, SA 5063
    Phone: +61 8 8338 5500   Fax: +61 8 8338 5511   Mobile: +61 401 981 301
       Email: pschulz@foursticks.com           Web: www.foursticks.com

Re: Broadcom Gigabit 5703 and Bridging

[sbolderoff]

On Wed, Jan 08, 2003 at 02:52:37PM +1030, Paul Schulz wrote:
> Greetings,
> 
> I'm seeing 'TCP Checksum' Errors after packets pass through a host
> bridging TCP packets with:
> 
>  - Kernel 2.4.20
>  - Bridge code
>  - tg3 (Broadcom Gigabit 5703)
> 
>    eth0: Tigon3 [partno(BCM95703A30) rev 1002 PHY(5703)]
>     (PCIX:100MHz:64-bit) 10/100/1000BaseT
>   (eth1 is similar)


The BCM95703A30 rev 1002 has issues with the hardware checksumming.

The following patch for the tg3 (linux-2.4.20-ac2/drivers/net/tg3.c)
driver, fixes the problem.

Note: I've tested this on the 2.4.20-ac2, but it should work OK with
2.4.20 too.

diff -u linux-2.4.20/drivers/net/tg3.c linux-2.4.20-ac2/drivers/net/tg3.c
--- linux-2.4.20/drivers/net/tg3.c	Fri Nov 29 10:23:14 2002
+++ linux-2.4.20-ac2/drivers/net/tg3.c	Wed Jan  8 14:34:44 2003
@@ -6161,6 +6161,10 @@
 	if (tp->pci_chip_rev_id == CHIPREV_ID_5700_B0)
 		tp->tg3_flags |= TG3_FLAG_BROKEN_CHECKSUMS;
 
+	/* 5703 A2 have issues with checksumming too. (sarah) */
+	if (tp->pci_chip_rev_id == CHIPREV_ID_5703_A2)
+		tp->tg3_flags |= TG3_FLAG_BROKEN_CHECKSUMS;
+
 	/* Regardless of whether checksums work or not, we configure
 	 * the StrongARM chips to not compute the pseudo header checksums
 	 * in either direction.  Because of the way Linux checksum support




Cheers,
Sarah Bolderoff
-- 
Foursticks Pty Ltd, http://www.foursticks.com
33 King William Street, ADELAIDE South Australia 5000
Phone +61 8 841114 309            Fax +61 8 841114 777
This message was brought to you by the numbers 0 and 1.

Re: Broadcom Gigabit 5703 and Bridging

[David S. Miller]

On Tue, 2003-01-07 at 20:41, sbolderoff@foursticks.com wrote:
> On Wed, Jan 08, 2003 at 02:52:37PM +1030, Paul Schulz wrote:
> The BCM95703A30 rev 1002 has issues with the hardware checksumming.

Really?

Can you demonstrate the problem with the 5703 without bridging?
Can you demonstrate the problem with bridging and another checksum
capable card?

Unless you can answer both those questions, I think faulting this
5703 variant is premature.  It smells more like a briding bug to
me, perhaps it's corrupting the hw checksumming state of an SKB
as it passes through the bridging layer.