netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* TSOv6 broken in atl1e
@ 2009-12-01  3:19 Ben Hutchings
  2009-12-01  5:17 ` Jie Yang
  2009-12-02  2:24 ` Jie Yang
  0 siblings, 2 replies; 3+ messages in thread
From: Ben Hutchings @ 2009-12-01  3:19 UTC (permalink / raw)
  To: Jie Yang; +Cc: 558426, netdev

[-- Attachment #1: Type: text/plain, Size: 1029 bytes --]

I received a bug report <http://bugs.debian.org/558426> that shows atl1e
corrupting IPv6 packets.  I have reproduced this on an Eee PC 901 and
found that it is linked to TSO.  The most obvious thing wrong with the
driver code is that it calculates the super-packet length incorrectly.
However, fixing that:

--- a/drivers/net/atl1e/atl1e_main.c
+++ b/drivers/net/atl1e/atl1e_main.c
@@ -1667,6 +1667,7 @@ static int atl1e_tso_csum(struct atl1e_adapter *adapter,
 
 		if (offload_type & SKB_GSO_TCPV6) {
 			real_len = (((unsigned char *)ipv6_hdr(skb) - skb->data)
+					+ sizeof(struct ipv6hdr)
 					+ ntohs(ipv6_hdr(skb)->payload_len));
 			if (real_len < skb->len)
 				pskb_trim(skb, real_len);
--- END ---

does not solve the problem.  Presumably this function is not
constructing correct DMA descriptors for TSOv6.

Please fix this, or I will submit a patch to remove this feature from
the driver.

Ben.

-- 
Ben Hutchings
Quantity is no substitute for quality, but it's the only one we've got.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

^ permalink raw reply	[flat|nested] 3+ messages in thread

* RE: TSOv6 broken in atl1e
  2009-12-01  3:19 TSOv6 broken in atl1e Ben Hutchings
@ 2009-12-01  5:17 ` Jie Yang
  2009-12-02  2:24 ` Jie Yang
  1 sibling, 0 replies; 3+ messages in thread
From: Jie Yang @ 2009-12-01  5:17 UTC (permalink / raw)
  To: Ben Hutchings; +Cc: 558426@bugs.debian.org, netdev

On Tuesday, December 01, 2009 11:20 AM
Ben Hutchings <ben@decadent.org.uk>
>
> I received a bug report <http://bugs.debian.org/558426> that
> shows atl1e corrupting IPv6 packets.  I have reproduced this
> on an Eee PC 901 and found that it is linked to TSO.  The
> most obvious thing wrong with the driver code is that it
> calculates the super-packet length incorrectly.
> However, fixing that:
>
> --- a/drivers/net/atl1e/atl1e_main.c
> +++ b/drivers/net/atl1e/atl1e_main.c
> @@ -1667,6 +1667,7 @@ static int atl1e_tso_csum(struct
> atl1e_adapter *adapter,
>
>               if (offload_type & SKB_GSO_TCPV6) {
>                       real_len = (((unsigned char
> *)ipv6_hdr(skb) - skb->data)
> +                                     + sizeof(struct ipv6hdr)
>                                       +
> ntohs(ipv6_hdr(skb)->payload_len));
>                       if (real_len < skb->len)
>                               pskb_trim(skb, real_len);
> --- END ---
>
> does not solve the problem.  Presumably this function is not
> constructing correct DMA descriptors for TSOv6.
>
> Please fix this, or I will submit a patch to remove this
> feature from the driver.
>
> Ben.
>
ok, I will try to reproduce it.

Best wishes
jie

^ permalink raw reply	[flat|nested] 3+ messages in thread

* RE: TSOv6 broken in atl1e
  2009-12-01  3:19 TSOv6 broken in atl1e Ben Hutchings
  2009-12-01  5:17 ` Jie Yang
@ 2009-12-02  2:24 ` Jie Yang
  1 sibling, 0 replies; 3+ messages in thread
From: Jie Yang @ 2009-12-02  2:24 UTC (permalink / raw)
  To: Ben Hutchings; +Cc: 558426@bugs.debian.org, netdev

On Tuesday, December 01, 2009 11:20 AM
Ben Hutchings <ben@decadent.org.uk> wrote:


>
> I received a bug report <http://bugs.debian.org/558426> that
> shows atl1e corrupting IPv6 packets.  I have reproduced this
> on an Eee PC 901 and found that it is linked to TSO.  The
> most obvious thing wrong with the driver code is that it
> calculates the super-packet length incorrectly.
> However, fixing that:
>
> --- a/drivers/net/atl1e/atl1e_main.c
> +++ b/drivers/net/atl1e/atl1e_main.c
> @@ -1667,6 +1667,7 @@ static int atl1e_tso_csum(struct
> atl1e_adapter *adapter,
>
>               if (offload_type & SKB_GSO_TCPV6) {
>                       real_len = (((unsigned char
> *)ipv6_hdr(skb) - skb->data)
> +                                     + sizeof(struct ipv6hdr)
>                                       +
> ntohs(ipv6_hdr(skb)->payload_len));
>                       if (real_len < skb->len)
>                               pskb_trim(skb, real_len);
> --- END ---
>
> does not solve the problem.  Presumably this function is not
> constructing correct DMA descriptors for TSOv6.
>
> Please fix this, or I will submit a patch to remove this
> feature from the driver.
>

ou, there is hareware limit to support TSOv6, just disable it.

Best wishes
jie

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2009-12-02  2:24 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-12-01  3:19 TSOv6 broken in atl1e Ben Hutchings
2009-12-01  5:17 ` Jie Yang
2009-12-02  2:24 ` Jie Yang

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).