Re: sky2 doesn't receive any packages after I ssh in via ipv6 and edit a file in vim (reproducable)

Re: sky2 doesn't receive any packages after I ssh in via ipv6 and edit a file in vim (reproducable)

[Stephen Hemminger]

On Mon, 21 Aug 2006 23:22:43 +0200
Thomas Glanzmann <sithglan@stud.uni-erlangen.de> wrote:

> Hello,
> I have the newest sky2 driver (with the newest fix in version 1.5)
> running on my intel mac mini running under legacy bios modus. However if
> I ssh in via IPv6 and start vim in the shell to edit a file the sky2
> network adapter stops to work. This is reproducable. A friend of mine
> thought that maybe the IPv6 code doesn't request the right checksum. But
> I have no clue if it is related to this. I can give access to the
> machine via ssh (including root) if someone is willing to track this
> down. I have a userland programming running which reboots the machine if
> the network connectivity is lost. At least this one is reliable. If
> noone wants a shell account maybe someone can guide me through
> debugging and tracking this down.
> 
> (mini) [/var/log] grep sky2 kern.log
> Aug 20 20:52:10 mini kernel: sky2 v1.5 addr 0x90200000 irq 17 Yukon-EC (0xb6) rev 2
> Aug 20 20:52:10 mini kernel: sky2 eth0: addr 00:16:cb:a3:91:3c
> Aug 20 20:52:12 mini kernel: sky2 eth0: enabling interface
> Aug 20 20:52:12 mini kernel: sky2 eth0: Link is up at 100 Mbps, full duplex, flow control both
> Aug 20 21:01:34 mini kernel: sky2 eth0: hw error interrupt status 0x30
> Aug 20 21:01:34 mini kernel: sky2 eth0: ram data read parity error
> Aug 20 21:01:34 mini kernel: sky2 eth0: ram data write parity error
> Aug 20 21:01:34 mini kernel: sky2 eth0: hw error interrupt status 0x8
> Aug 20 21:01:34 mini kernel: sky2 eth0: MAC parity error

Something is happening long before you start doing ipv6.
The parity error's mean that the PCI transfers are going south.
The driver might be better to just shut it self off after that.

Perhaps the PCI access isn't working properly. What is the config?
Also are you running i386 or x86_64 on this.

Is this fix in your version?
	http://www.mail-archive.com/netdev@vger.kernel.org/msg18603.html


-- 
Stephen Hemminger <shemminger@osdl.org>

All non-trivial abstractions, to some degree, are leaky.

Re: sky2 doesn't receive any packages after I ssh in via ipv6 and edit a file in vim (reproducable)

[Thomas Glanzmann]

Hello,

> Is this fix in your version?
> 	http://www.mail-archive.com/netdev@vger.kernel.org/msg18603.html

I put this one in and recompiled and I am *still* able to reproduce the
problem.

        Thomas

Re: sky2 doesn't receive any packages after I ssh in via ipv6 and edit a file in vim (reproducable)

[Thomas Glanzmann]

[-- Attachment #1: Type: text/plain, Size: 960 bytes --]

Hello,

                 [resent because forget to CC netdev ]

> Something is happening long before you start doing ipv6.  The parity
> error's mean that the PCI transfers are going south.  The driver might
> be better to just shut it self off after that.

I can ssh off the machine for days. But sshing in IPv6 and starting vim
makes the network card hang.

> Perhaps the PCI access isn't working properly. What is the config?
> Also are you running i386 or x86_64 on this.

I use i386, config attached.

> Is this fix in your version?
> 	http://www.mail-archive.com/netdev@vger.kernel.org/msg18603.html

No it isn't. But I am recompiling at the moment with the patch.

And btw. IPv4 is working perfectly. But of course not after the card
stopped responding. I captured the network traffic and tried to replay
it, but it didn't trigger the problem. But I still am able to reproduce
it manually. If you want the capture, I can give it to you.

        Thomas

[-- Attachment #2: config.bz2 --]
[-- Type: application/octet-stream, Size: 8014 bytes --]

Re: sky2 doesn't receive any packages after I ssh in via ipv6 and edit a file in vim (reproducable)

[Daniel Drake]

Thomas Glanzmann wrote:
> I can ssh off the machine for days. But sshing in IPv6 and starting vim
> makes the network card hang.

This is possibly related to an equally strange issue reported on the 
Gentoo bugzilla. As soon as iptables is started, sky2 hangs.

http://bugs.gentoo.org/136508

I'll be sending more details as soon as I have them.

Daniel

[PATCH] IPV6 : segmentation offload not set correctly on TCP children

[Stephen Hemminger]

TCP over IPV6 would incorrectly inherit the GSO settings.
This would cause kernel to send Tcp Segmentation Offload packets for
IPV6 data to devices that can't handle it. It caused the sky2 driver
to lock http://bugzilla.kernel.org/show_bug.cgi?id=7050
and the e1000 would generate bogus packets. I can't blame the
hardware for gagging if the upper layers feed it garbage.

This was a new bug in 2.6.18 introduced with GSO support.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>


--- linux-2.6.orig/net/ipv6/tcp_ipv6.c	2006-08-03 09:09:16.000000000 -0700
+++ linux-2.6/net/ipv6/tcp_ipv6.c	2006-08-25 15:30:31.000000000 -0700
@@ -944,7 +944,7 @@
 	 * comment in that function for the gory details. -acme
 	 */
 
-	sk->sk_gso_type = SKB_GSO_TCPV6;
+	newsk->sk_gso_type = SKB_GSO_TCPV6;
 	__ip6_dst_store(newsk, dst, NULL);
 
 	newtcp6sk = (struct tcp6_sock *)newsk;

Re: [PATCH] IPV6 : segmentation offload not set correctly on TCP children

[David Miller]

From: Stephen Hemminger <shemminger@osdl.org>
Date: Fri, 25 Aug 2006 15:43:53 -0700

> TCP over IPV6 would incorrectly inherit the GSO settings.
> This would cause kernel to send Tcp Segmentation Offload packets for
> IPV6 data to devices that can't handle it. It caused the sky2 driver
> to lock http://bugzilla.kernel.org/show_bug.cgi?id=7050
> and the e1000 would generate bogus packets. I can't blame the
> hardware for gagging if the upper layers feed it garbage.
> 
> This was a new bug in 2.6.18 introduced with GSO support.
> 
> Signed-off-by: Stephen Hemminger <shemminger@osdl.org>

Good catch.  Applied, thanks Stephen.

Re: [PATCH] IPV6 : segmentation offload not set correctly on TCP children

[Thomas Glanzmann]

Hello Stephen,
thanks for the fix, it fixes the problem for me. I closed the bug. On
which hardware did you reproduce the bug and how did you found it? Did
you use git bisect?

        Thomas

Re: [PATCH] IPV6 : segmentation offload not set correctly on TCP children

[Stephen Hemminger]

On Sat, 26 Aug 2006 01:06:26 +0200
Thomas Glanzmann <sithglan@stud.uni-erlangen.de> wrote:

> Hello Stephen,
> thanks for the fix, it fixes the problem for me. I closed the bug. On
> which hardware did you reproduce the bug and how did you found it? Did
> you use git bisect?
> 
>         Thomas

Using sky2 on Intel motherboard and git bisect. But Jesse found same problem
on e1000. Bisect was relatively fast since it was either base net code or driver.
Starting with 2.6.17, everything worked, and latest was busted.

	git bisect start net drivers/net/sky2.c
	
Test was ipv6 slogin, then run dmesg. If that worked then run top and cause
lots of other traffic by just doing a firefox open-tabs to blast lots of connections.
Wanted to induce tcp-ipv6 to get congested.

-- 
Stephen Hemminger <shemminger@osdl.org>

All non-trivial abstractions, to some degree, are leaky.

Re: [PATCH] IPV6 : segmentation offload not set correctly on TCP children

[Stephen Hemminger]

On Sat, 26 Aug 2006 01:06:26 +0200
Thomas Glanzmann <sithglan@stud.uni-erlangen.de> wrote:

> Hello Stephen,
> thanks for the fix, it fixes the problem for me. I closed the bug. On
> which hardware did you reproduce the bug and how did you found it? Did
> you use git bisect?
> 
>         Thomas

Finding the line was luck. I spotted similar (but correct) code in
DCCP over IPV6.

-- 
Stephen Hemminger <shemminger@osdl.org>

All non-trivial abstractions, to some degree, are leaky.

Re: [PATCH] IPV6 : segmentation offload not set correctly on TCP children

[Herbert Xu]

On Fri, Aug 25, 2006 at 03:43:53PM -0700, Stephen Hemminger wrote:
> TCP over IPV6 would incorrectly inherit the GSO settings.
> This would cause kernel to send Tcp Segmentation Offload packets for
> IPV6 data to devices that can't handle it. It caused the sky2 driver
> to lock http://bugzilla.kernel.org/show_bug.cgi?id=7050
> and the e1000 would generate bogus packets. I can't blame the
> hardware for gagging if the upper layers feed it garbage.
> 
> This was a new bug in 2.6.18 introduced with GSO support.
> 
> Signed-off-by: Stephen Hemminger <shemminger@osdl.org>

Thanks for catching this Stephen!

Cheers,
-- 
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt