Please don't do > 80 columns line again. I am not a tty.

Pascal GREGIS [off-list ref] :

Francois Romieu a écrit, le Tue 27 Feb 2007 à 12:48:31AM :

[...]

Where is this init sequence in the driver? I searched in the code and
didn't find something that could satisfy, then I took the 2.6.12.3 r8169.c,

See hw_start.

put it in the 2.6.11.11 source tree, applied my patch, changed
synchronize_sched(); to synchronize_kernel(); because the symbol was
unknown and then I could turn eth1 on and assign it an ip address but the
rx and tx count never grew and when I turned eth0 down (a 8139 card) eth1
didn't reply to ping anymore.

Linux kernel 2.6.12 does not support the 8168 (it includes some 8169 fixes
though).

[...]

quoted

There are 59 r8169 related patches between 2.6.12 and current. Only a few
of those break the API. I'll give it a try tomorrow evening.

Hmm... you wrote this mail at 00 h 48 this morning, what did you mean by
tomorrow evening?

1. I mean now. See:
   http://www.fr.zoreil.com/people/francois/backport/r8169/20070227-00
   (big patch or serie of 54 pieces).

2. Compiled, untested. You know what you have to do.

3. Due to the changes in the driver, one could hope that the link will be
   autocorrectly set. Please give it a try before using ethtool/mii-tool.

4. The patchkit does not include the latest changes/bugfixes. They are
   still experimental but some users have a poor 8168 experience without
   them. YMMV. Please send a complete dmesg, lspci -vvx and the brand of
   the motherboard if the driver stops working randomly.

-- 
Ueimor