(This is the last in a series of five posts. See part 1, part 2, part 3, part 4.)

As usual, when you spend three days hunting a bug, it turns out to be an embarrassingly simple fix. In this case, the issue was that the VLAN driver called skb_cow(skb, skb_headroom(skb)), which expands the skb to have at least the amount of headroom specified in ( Read more... )

(This is the fourth in a series of five posts. See part 1, part 2, part 3, part 5.)

The checksum problem is happening in the call to vlan_check_reorder_header() on line 203 of net/8021q/vlan_dev.c. The checksum field is 0xffffffff before the call and 0x001fffff after it [1]. The only difference between the packets I can see so far is that the ( Read more... )

(This is the third in a series of five posts. See part 1, part 2, part 4, part 5.)

The checksum test in nf_conntrack can be disabled with sysctl -w net.netfilter.nf_conntrack_checksum=0 (see, for example, the net->ct.sysctl_checksum test on line 185 of net/ipv4/netfilter/nf_conntrack_proto_icmp.c), but this just moves the error elsewhere:

br0: hw ( Read more... )

(This is the second in a series of five posts. See part 1, part 3, part 4, part 5.)

By instrumenting the VLAN driver code with debug output, I have demonstrated that the checksum computed for the offending packets immediately after the call to skb_pull_rcsum are all 0xffff; i.e., the checksum for only the part of the packet from the IP header ( Read more... )

(This is the first in a series of five posts. See part 2, part 3, part 4, part 5.)

I have been stalking a very elusive kernel bug:

Using Linux kernel 2.6.34 on a Sun Ultra 10 (cronus.persephoneslair.org) with a PCI Quad-Fast Ethernet card and the sunhme driver, there seems to be a subtle problem with VLANs. In my particular config, the second ( Read more... )

It's been nine days since my last post about this. The prom console thing worked, but it exposed another problem: smp_boot_one_cpu() was hanging trying to start the other processors. Well, to be precise, it was calling prom_startcpu(), and that was returning just fine, but the next printk() after that would output one character and then hang.

I ( Read more... )