Skip to contentSkip to conversation

iMac G3 IPv6 issue

DISCUSSION Aurélien GÉRÔME 3 participants 8 messages Sep 12, 2004

lore
Share & tools
lore
#0 lore raw

Hi there,

Since it is kernel related, I think it is the best place to ask.

I installed Debian Sid on a BlueBerry iMac G3.

[20:51:54|3|root@arwen:~]# cat /proc/cpuinfo
processor       : 0
cpu             : 740/750
temperature     : 52 C (uncalibrated)
clock           : 350MHz
revision        : 2.2 (pvr 0008 0202)
bogomips        : 696.32
machine         : PowerMac2,1
motherboard     : PowerMac2,1 MacRISC Power Macintosh
detected as     : 66 (iMac FireWire)
pmac flags      : 00000005
L2 cache        : 512K unified
memory          : 128MB
pmac-generation : NewWorld
[20:54:17|3|root@arwen:~]# uname -a
Linux arwen 2.6.8.1 #1 Sat Aug 21 15:46:55 CEST 2004 ppc GNU/Linux

I put IPv6 support, but the console is flooded by a lot of:

printk: 16 messages suppressed. hw tcp v6 csum failed
printk: 11 messages suppressed. hw tcp v6 csum failed
printk: 5 messages suppressed. hw tcp v6 csum failed
printk: 5 messages suppressed. hw tcp v6 csum failed
printk: 5 messages suppressed. hw tcp v6 csum failed
printk: 5 messages suppressed. hw tcp v6 csum failed
printk: 7 messages suppressed. hw tcp v6 csum failed

However, IPv6 works, and IPv4 packets do not have this kind of issue.

The network card is a Sun Gem. It is kind of weird to see bad packets.

Attached is my /proc/config.gz.

Is anyone aware of this issue?

Bests.

Signature · 19 lines
        _,met$$$$$gg.        Aurélien GÉRÔME
     ,g$$$$$$$$$$$$$$$P.     Free Software Developer
   ,g$$P""       """Y$$.".   Unix Sys & Net Admin
  ,$$P'              `$$$.
',$$P       ,ggs.     `$$b:  Web: https://www.roxor.be/
`d$$'     ,$P"'   .    $$$   Mail: [email protected]
 $$P      d$'     ,    $$P   GPG Key ID: 03E1A172
 $$:      $$.   -    ,d$$'
 $$;      Y$b._   _,d$P'    ))  GNU  ((   /   Linux
 Y$$.    `.`"Y$$$$P"'      //         \\
 `$$b      "-.__          ((__,-"""-,__))     .---.
  `Y$$b                    `--)~   ~(--`     /     \
   `Y$$.      Debian      .-'(       )`-.    \.@-@./
     `$$b.                `~~`@)   (@`~~`    /`\_/`\
       `Y$$b.                 |     |       //  _  \\
         `"Y$b._              |     |      | \     )|_
             `""""            (8___8)     /`\_`>  <_/ \
                               `---`      \__/'---'\__/
BOFH excuse #423: It's not RFC-822 compliant.
#2 lore raw
In reply to Aurélien GÉRÔME
Quoted 4 lines · Aurélien GÉRÔME

I put IPv6 support, but the console is flooded by a lot of:
hw tcp v6 csum failed
However, IPv6 works, and IPv4 packets do not have this kind of issue. The network card is a Sun Gem. It is kind of weird to see bad packets.

I use the sungem in a similar machine with IPv6, and haven't seen any problems. Can you tcpdump from a different machine on the network and confirm whether these reported bad checksums really are happening or if it's the fault of the hardware/driver?

Signature · 1 line
dwmw2
#3 lore raw
Quoted 10 lines · David Woodhouse
Quoted 4 lines · Aurélien GÉRÔME

I put IPv6 support, but the console is flooded by a lot of:
hw tcp v6 csum failed
However, IPv6 works, and IPv4 packets do not have this kind of issue. The network card is a Sun Gem. It is kind of weird to see bad packets.

I use the sungem in a similar machine with IPv6, and haven't seen any problems. Can you tcpdump from a different machine on the network and confirm whether these reported bad checksums really are happening or if it's the fault of the hardware/driver?

I have 2 machines on the same switch and an IPv6 ssh connection between them.

"tcpdump -a -i eth0 -v" on both machines shows [tcp sum ok].

The ssh link is an IPv6 one, arwen is the faulty machine, and aragorn a sane one.

arwen has also a ppp0 interface, no infamous [bad tcp cksum ....!] appears, on lo interface too. All is reported [tcp sum ok].

But during this time, the kernel keeps flooding the console with these "hw tcp v6 csum failed"...

Moreover, like I said, the IPv6 networking works fine.

In that case, is it the driver fault?

I wonder if it is possible to print the checksum and the interface with the LIMIT_NETDEBUG(printk(KERN_DEBUG "hw tcp v6 csum failed\n"));.

Bests.

Signature · 19 lines
        _,met$$$$$gg.        Aurélien GÉRÔME
     ,g$$$$$$$$$$$$$$$P.     Free Software Developer
   ,g$$P""       """Y$$.".   Unix Sys & Net Admin
  ,$$P'              `$$$.
',$$P       ,ggs.     `$$b:  Web: https://www.roxor.be/
`d$$'     ,$P"'   .    $$$   Mail: [email protected]
 $$P      d$'     ,    $$P   GPG Key ID: 03E1A172
 $$:      $$.   -    ,d$$'
 $$;      Y$b._   _,d$P'    ))  GNU  ((   /   Linux
 Y$$.    `.`"Y$$$$P"'      //         \\
 `$$b      "-.__          ((__,-"""-,__))     .---.
  `Y$$b                    `--)~   ~(--`     /     \
   `Y$$.      Debian      .-'(       )`-.    \.@-@./
     `$$b.                `~~`@)   (@`~~`    /`\_/`\
       `Y$$b.                 |     |       //  _  \\
         `"Y$b._              |     |      | \     )|_
             `""""            (8___8)     /`\_`>  <_/ \
                               `---`      \__/'---'\__/
BOFH excuse #88: Boss' kid fucked up the machine
#4 lore raw
In reply to David Woodhouse
Quoted 10 lines · On Sun, 12 Sep 2004 18:57:31 +0100 David Woodhouse <[email protected]> wrote:
Quoted 4 lines · Aurélien GÉRÔME

I put IPv6 support, but the console is flooded by a lot of:
hw tcp v6 csum failed
However, IPv6 works, and IPv4 packets do not have this kind of issue. The network card is a Sun Gem. It is kind of weird to see bad packets.

I use the sungem in a similar machine with IPv6, and haven't seen any problems. Can you tcpdump from a different machine on the network and confirm whether these reported bad checksums really are happening or if it's the fault of the hardware/driver?

One thing that's interesting is that the sungem and sunhme drivers are the only two in the whole tree that support IPV6 checksum offload. Like David, I have never seen this problem on Sparc systems using the GEM chip and this driver.

#5 lore raw
Quoted 14 lines · David S. Miller
Quoted 10 lines · On Sun, 12 Sep 2004 18:57:31 +0100 David Woodhouse <[email protected]> wrote:
Quoted 4 lines · Aurélien GÉRÔME

I put IPv6 support, but the console is flooded by a lot of:
hw tcp v6 csum failed
However, IPv6 works, and IPv4 packets do not have this kind of issue. The network card is a Sun Gem. It is kind of weird to see bad packets.

I use the sungem in a similar machine with IPv6, and haven't seen any problems. Can you tcpdump from a different machine on the network and confirm whether these reported bad checksums really are happening or if it's the fault of the hardware/driver?

Quoted 5 lines

One thing that's interesting is that the sungem and sunhme drivers are the only two in the whole tree that support IPV6 checksum offload. Like David, I have never seen this problem on Sparc systems using the GEM chip and this driver.

The board is this:
0002:02:0f.0 Ethernet controller: Apple Computer Inc. UniNorth GMAC (Sun GEM)

So, it is Apple stuff wrapping a Sun GEM chip. The issue may come from that. Anyway, I am going to dig in the source code to find out.

Cheers.

Signature · 19 lines
        _,met$$$$$gg.        Aurélien GÉRÔME
     ,g$$$$$$$$$$$$$$$P.     Free Software Developer
   ,g$$P""       """Y$$.".   Unix Sys & Net Admin
  ,$$P'              `$$$.
',$$P       ,ggs.     `$$b:  Web: https://www.roxor.be/
`d$$'     ,$P"'   .    $$$   Mail: [email protected]
 $$P      d$'     ,    $$P   GPG Key ID: 03E1A172
 $$:      $$.   -    ,d$$'
 $$;      Y$b._   _,d$P'    ))  GNU  ((   /   Linux
 Y$$.    `.`"Y$$$$P"'      //         \\
 `$$b      "-.__          ((__,-"""-,__))     .---.
  `Y$$b                    `--)~   ~(--`     /     \
   `Y$$.      Debian      .-'(       )`-.    \.@-@./
     `$$b.                `~~`@)   (@`~~`    /`\_/`\
       `Y$$b.                 |     |       //  _  \\
         `"Y$b._              |     |      | \     )|_
             `""""            (8___8)     /`\_`>  <_/ \
                               `---`      \__/'---'\__/
BOFH excuse #113: Root nameservers are out of sync
#6 lore raw
Quoted 6 lines · On Mon, 13 Sep 2004 01:17:31 +0200 Aurélien GÉRÔME <[email protected]> wrote:

The board is this:
0002:02:0f.0 Ethernet controller: Apple Computer Inc. UniNorth GMAC (Sun GEM)

So, it is Apple stuff wrapping a Sun GEM chip. The issue may come from that. Anyway, I am going to dig in the source code to find out.

It would be interesting if some other iMac users can reproduce this too.

BTW, your signature is quite huge, could you trim it down or (alternatively) not attach it when posting to this list? Thanks a lot.

#7 lore raw
Quoted 7 lines · David S. Miller
Quoted 4 lines · On Mon, 13 Sep 2004 01:17:31 +0200 Aurélien GÉRÔME <[email protected]> wrote:

The board is this:
0002:02:0f.0 Ethernet controller: Apple Computer Inc. UniNorth GMAC (Sun GEM)

Quoted 5 lines
Quoted 2 lines

So, it is Apple stuff wrapping a Sun GEM chip. The issue may come from that. Anyway, I am going to dig in the source code to find out.

It would be interesting if some other iMac users can reproduce this too.

If it is reproducible on this board (actually, it is built-in on this machine), then it is really specific to this one. PowerBooks and iBooks have also Sun GEM, and it does not occur.

Quoted 3 lines

BTW, your signature is quite huge, could you trim it down or (alternatively) not attach it when posting to this list? Thanks a lot.

I am sorry, please accept my apologies if I have bothered you or others with it. ;)

Bests.