I upgraded from 2.6.32, using make oldconfig, but upon rebooting 2/3 of my NICs are not detected and the one that was detected is not working. I've tried fiddling a bit with my kernel config but have had no luck. Reproducible: Always Steps to Reproduce: 1. Copy a working kernel .config from 2.6.32 to a 2.6.33 directory 2. `make oldconfig` 3. `make && make modules_install`, cp the image to /boot, update grub and kexec 4. reboot Actual Results: NICs not detected or not working Expected Results: NICs detected and working
Please attach your config file and the lspci output
Created attachment 221579 [details] My kernel config file This is my .33 config file, after running make oldconfig and doing some other small tweaking.
Created attachment 221581 [details] lspci output The lspci output you asked for.
If you don't mind can I see the out of lspci -n
Created attachment 221729 [details] Output from running `lspci -n` The requested output from running `lspci -n`
Which NICs fail to appear when running ifconfig -a? Could this be related to openrc ?
(In reply to comment #6) > Which NICs fail to appear when running ifconfig -a? Could this be related to > openrc ? > I expect to see the Intel e100 and the Realtek 8168 cards, the D-Link is not being used right now, but I guess it should be detected as well. It detects only the Intel NIC, but still fails to setup networking.
It is really weird that none of the cards are working. It is highly unlikely all the 3 drivers to be broken on recent kernels. Which openrc version are you using? Could you please attach a full log from the boot phase?
Created attachment 221937 [details] Boot log from openrc I'm using openrc 0.60-r1, the latest in portage.
dmesg after boot, please
Created attachment 222135 [details] dmesg output right after boot The requested dmesg output.
Well it doesnt look like an IRQ conflict. r8169 driver clearly fails to operate ( error -22 ). e100 seems to work properly. I am almost out of ideas :/
(In reply to comment #12) > Well it doesnt look like an IRQ conflict. r8169 driver clearly fails to operate > ( error -22 ). e100 seems to work properly. > I am almost out of ideas :/ > Well, I'm going to play around with it a bit more, trying to use just the e100. I'll report back what happens later on. Does the error -22 mean that the driver is borked for this version of the kernel?
(In reply to comment #12) > Well it doesnt look like an IRQ conflict. r8169 driver clearly fails to operate > ( error -22 ). e100 seems to work properly. > I am almost out of ideas :/ > If the issue with the rtl driver didn't appear from the 2.6.32->2.6.33 transition I would have said that it's connected to these two bugs: http://bugs.gentoo.org/show_bug.cgi?id=298741 http://bugs.gentoo.org/show_bug.cgi?id=237089
I disabled the rtl card and now the e100 is working fine, so I can at least use my .33 kernel, but in a limited way.
(In reply to comment #15) > I disabled the rtl card and now the e100 is working fine, so I can at least use > my .33 kernel, but in a limited way. > Could this be an IRQ conflict issue? which is the last working kernel?
(In reply to comment #16) > (In reply to comment #15) > > I disabled the rtl card and now the e100 is working fine, so I can at least use > > my .33 kernel, but in a limited way. > > > > Could this be an IRQ conflict issue? which is the last working kernel? > .32-r7 worked fine
I had a similar problem, after upgrading to a 2.6.33 kernel, the r8169 nic would not come up, dmesg showed: r8169: eth0: link down ADDRCONF(NETDEV_UP): eth0: link is not ready tried to go back to 2.6.32, same thing. After much googling: The ultimate fix (permanent or not) was to remove power from the machine (thus resetting the card, something to do with Wake on Lan) AND rm /etc/udev/rules.d/70-persistent-net.rules And now eth0 comes up on 2.6.33 Hope this information is useful.
(In reply to comment #18) > I had a similar problem, after upgrading to a 2.6.33 kernel, the r8169 nic > would not come up, dmesg showed: > > r8169: eth0: link down > ADDRCONF(NETDEV_UP): eth0: link is not ready > > tried to go back to 2.6.32, same thing. I was able to switch to my .32 kernel and have things working again > After much googling: > The ultimate fix (permanent or not) was to remove power from the machine (thus > resetting the card, something to do with Wake on Lan) AND rm > /etc/udev/rules.d/70-persistent-net.rules Gave this a shot anyways...it didn't work :(
I went and did a bisection on the kernel sources and was told that http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=ac1aa47b131416a6ff37eb1005a0a1d2541aad6c is the offending patch. I tried blindly removing the patch in a rebase, but it doesn't come out cleanly. Can I get some help here, please?
The obvious thing to try was to turn turn on quirk checking...IT WORKED! So it seems that the patch just deems my hardware as quirky, but everything is good now.