I am having the exact same issue with same driver r8169 with kernel 3.0.0-13-generic
It actually only started for me as of this kernel, and it only happens when I bring my network card close to a max throughput (1Gb), as soon as I start backing up to my desktop over the network, I start getting net_ratelimit errors and my eth0 link up errors repeat a lot and shortly after a full freezup.
Not sure if compiling a new network driver module might do it or not, have not done much troubleshooting yet other than a quick google search which brought me here
]]>It sounds like compiling r8168 from Realtek is an alternative, I may try this as there's already a package in the AUR. Though obviously it would be preferable if r8169 were fixed.
]]>The changes that were introduced between Arch Linux's 2.6.37.3 and 2.6.37.4 kernels seem to correspond to Linux kernel commits:
0d672e9f8ac320c6d1ea9103db6df7f99ea20361
f60ac8e7ab7cbb413a0131d5665b053f9f386526
1519e57fe81c14bb8fa4855579f19264d1ef63b4
b5ba6d12bdac21bc0620a5089e0f24e362645efd
Does that make more sense?
]]>Anyway, I think maybe git bisect on the Arch Linux tree is a better idea than on the Linux kernel itself which is what I initially tried. Let me know if you have any advice for doing this, the Arch Linux commit messages are a bit confusing to me.
]]>git-bisect is a good idea. There aren't many commits, and it won't make you go through all of them.
]]>I am not a programmer but I wonder if upstream commit ba04c7c93bbcb48ce880cf75b6e9dffcd79d4c7b could be the culprit? It claims to fix something that sounds similar to what we're having, but it was also a change introduced at the right time to explain the issue (between Arch 2.6.37.3 and 2.6.37.4):
http://www.kernel.org/pub/linux/kernel/ … -2.6.32.32
Maybe I'll try reverting that in a 2.6.38 kernel and see if I still get the problem.
]]>If anyone has tested with anything newer than 2.6.38.2 please let me know. If not I may try git bisect to see if I can figure out what's gone wrong.
]]>You should reply to the kernel bug report that the problem started between 2.6.37.3 - 2.6.37.4 and any other useful things you can add (this would also generate some traffic so it doesn't get ignored).
]]>Anyone tested this with 2.6.37.5?
Problem remains.
Anyone tested it with 2.6.38.2?
]]>