network issues after kernel rollback

q23r4r · 02-09-2019, 11:53 AM

In upgrading to the latest kernel patch (4.4.172), I got hit by the NVIDIA bug discussed here: https://www.linuxquestions.org/quest...ia-4175647404/

To workaround, I decided to roll back to 4.4.157 for now. I still had the old package files, so I ran upgradepkg on them and reran lilo as usual.

System boots fine and mostly works. But, now I have a problem with my ethernet network connection. On bootup, it fails to get a carrier. After boot-up, I can successfully get a carrier and connect if I manually run "dhcpcd eth0" But, I only get a 10 MBs connection rather than gigabit.

I didn't make any other changes to my system in between upgrading to 172 then trying to run NVIDIA installer, then rolling back to 157 and rerunning the NVIDIA installer.

Is there any cruft or something that I need to manually remove to fix this? Other ideas?

camorri · 02-09-2019, 01:08 PM

Did you restore the kernel firmware from 4.4.157? There may be something there that is causing the problem.

q23r4r · 02-09-2019, 02:07 PM

I believe so. I ran upgradepkg on all 6 kernel packages.

abga · 02-09-2019, 07:24 PM

I'm a little more radical in upgrading/downgrading packages and using only installpkg and removepkg, always following the log to check if there are leftovers after removepkg - managing them manually. Not that much familiar with upgradepkg, but according to this post:
https://www.linuxquestions.org/quest...5/#post5809664
upgradepkg shouldn't leave "leftovers" and your reported issue might be caused by either a missed step/package in your downgrade, a driver or, as camorri suggested, a firmware file.
Can you please list the "all 6 kernel packages" you were using for the downgrade together with these outputs:

Code:

#the content of /lib/modules:
ls -al /lib/modules
# Ethernet related kernel messages - remove the MAC address from the output - just for privacy concerns
dmesg -T | grep eth

q23r4r · 02-10-2019, 11:20 AM

The six kernel packages that I provided to upgradepkg were:
kernel-firmware-20180913_44d4fca-noarch-1.txz
kernel-generic-4.4.157-x86_64-1.txz
kernel-headers-4.4.157-x86-1.txz
kernel-huge-4.4.157-x86_64-1.txz
kernel-modules-4.4.157-x86_64-1.txz
kernel-source-4.4.157-noarch-1.txz

An ls of the library directory reveals no remnants of 172 (there is no directory named 4.4.172)

Here is the dmesg output (minus MAC address)
There is an error there (but I don't think it is relevant because grep seems to be latching onto the "eth" in Method)

Sat Feb 9 08:25:00 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT1._GTF] (Node ffff88043e9c78c0), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:00 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT0._GTF] (Node ffff88043e9c7848), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:00 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT1._GTF] (Node ffff88043e9c78c0), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:00 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT3._GTF] (Node ffff88043e9c79b0), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:00 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT0._GTF] (Node ffff88043e9c7848), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:00 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT3._GTF] (Node ffff88043e9c79b0), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:04 2019] r8169 0000:04:00.0 eth0: RTL8168e/8111e at 0xffffc9000140e000, **:**:**:**:**:**, XID 0c200000 IRQ 28
[Sat Feb 9 08:25:04 2019] r8169 0000:04:00.0 eth0: jumbo features [frames: 9200 bytes, tx checksumming: ko]
[Sat Feb 9 08:25:07 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT0._GTF] (Node ffff88043e9c7848), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:07 2019] ACPI Error: Method parse/execution failed [\_SB.PCI0.SAT0.SPT0._GTF] (Node ffff88043e9c7848), AE_NOT_FOUND (20150930/psparse-542)
[Sat Feb 9 08:25:12 2019] r8169 0000:04:00.0 eth0: link down
[Sat Feb 9 08:25:12 2019] r8169 0000:04:00.0 eth0: link down
[Sat Feb 9 08:25:14 2019] r8169 0000:04:00.0 eth0: link up
[Sat Feb 9 08:25:15 2019] r8169 0000:04:00.0 eth0: link down
[Sat Feb 9 08:25:18 2019] r8169 0000:04:00.0 eth0: link up
[Sat Feb 9 08:25:19 2019] r8169 0000:04:00.0 eth0: link down
[Sat Feb 9 08:25:22 2019] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[Sat Feb 9 08:25:24 2019] r8169 0000:04:00.0 eth0: link up
[Sat Feb 9 08:25:24 2019] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[Sat Feb 9 08:25:25 2019] r8169 0000:04:00.0 eth0: link down
[Sat Feb 9 08:25:28 2019] r8169 0000:04:00.0 eth0: link up
[Sat Feb 9 08:25:30 2019] r8169 0000:04:00.0 eth0: link down
[Sat Feb 9 08:25:34 2019] r8169 0000:04:00.0 eth0: link up
[Sat Feb 9 08:25:35 2019] r8169 0000:04:00.0 eth0: link down
[Sat Feb 9 08:25:38 2019] r8169 0000:04:00.0 eth0: link up

abga · 02-10-2019, 02:08 PM

Code:

[Sat Feb 9 08:25:04 2019] r8169 0000:04:00.0 eth0: RTL8168e/8111e at 0xffffc9000140e000, **:**:**:**:**:**, XID 0c200000 IRQ 28

I don't have experience with this Ethernet chip but noticed people are building the driver from SlackBuilds, because the default kernel driver doesn't play well. Maybe you did it too in the past and forgot about it.
http://slackbuilds.org/repository/14.2/network/r8168/
https://www.linuxquestions.org/quest...-a-4175647779/