http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c4 --- Comment #4 from Kriton Kyrimis --- Created attachment 780889 --> http://bugzilla.opensuse.org/attachment.cgi?id=780889&action=edit dmesg and hwinfo output with cable attached to the US Robotics NIC Yes, enp5s0 is the US Robotics controller. The reason that it never gets connected, even with 4.17, is that the network cable is on the other controller! I am attaching new versions of the files, with the network cable attached to the US Robotics controller. This time you should see a difference. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c6 --- Comment #6 from Takashi Iwai --- Hm, both 4.17 and 4.18 messages show that enp5s0 link up, and that's all. So if it's really a kernel regression, it must be in a deeper level. Could you report rather this to upstream? I have currently little time for checking this, and it's not my expertise area, either. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c8 --- Comment #8 from david chang --- There are 43 commits change the r8169 driver between v4.17..v4.18. And I find the following suspect commits. 4fd48c4ac0a0 r8169: move common initializations to tp->hw_start (v4.18-rc1) 82d3ff6dd199 r8169: remove calls to rtl_set_rx_mode (v4.18-rc1) 3559d81e76bf r8169: simplify rtl_hw_start_8169 (v4.18-rc1) 0ae0974eb3d0 r8169: improve CPlusCmd handling (v4.18-rc1) 61cb532d5224 r8169: change hw_start argument type (v4.18-rc1) 1d0254dd9b44 r8169: replace rx_buf_sz with a constant (v4.18-rc1) I have just created a testing kmp based on one of these, which is before merge patch "simplify rtl_hw_start_8169", could you please give it a try? https://download.opensuse.org/repositories/home:/david_chang:/bsc1105573/ope... Thanks! -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c10 --- Comment #10 from david chang --- (In reply to Kriton Kyrimis from comment #9)

Do I just install the rpm and reboot (and remember to uninstall the rpm when the time comes to erase the current kernel)?

Yes, just install and reboot. The driver will be installed at "/lib/modules/4.18.5-1-default/updates/r8169.ko". Thanks! -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c12 --- Comment #12 from david chang --- (In reply to Kriton Kyrimis from comment #11)

This kmp fixed the problem, so I guess the culprit is one of the subsequent four commits.

It's good to know.

I'll gladly test more kmps if you want to identify which commit caused the problem.

That would be great! I'll prepare the others kmp later. Thank you! -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c14 --- Comment #14 from Kriton Kyrimis --- The problem reappeared with the first of these three rpms: r8169-kmp-3559d81-kmp-default-2.3_k4.18.5_1-1.1.x86_64.rpm Per your instructions, I didn't test the other two rpms. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c16 --- Comment #16 from david chang --- Since I did not have the same chip netcard handy. And I created another temporary kmp with the possible fix. Could you please give it a try? Thank you! https://build.opensuse.org/package/binary/download/home:david_chang:bsc11055... -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c18 --- Comment #18 from david chang --- (In reply to Kriton Kyrimis from comment #17)

(In reply to david chang from comment #16)

I'm afraid the problem persists with this rpm.

Well, my guess is not right. This time I didn't change the logic of the patch. But the patch also removed a function 'rtl_init_rxcfg()', I add it back. Would you mind to give it a try? https://build.opensuse.org/package/binary/download/home:david_chang:bsc11055... Thank you! -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c20 --- Comment #20 from david chang --- (In reply to Kriton Kyrimis from comment #19)

(In reply to david chang from comment #18)

...

This time I didn't change the logic of the patch. But the patch also removed a function 'rtl_init_rxcfg()', I add it back. Would you mind to give it a try?

This one doesn't work, either.

However, searching for "rtl_init_rxcfg", I found this message: https://www.spinics.net/lists/netdev/msg520023.html which appears to be a copy of this message: https://www.mail-archive.com/netdev@vger.kernel.org/msg246859.html

It provides a patch along the lines that you are investigating, saying that it fixes the problem.

Cool.

The second message seems to be from a kernel.org mailing list, so could it be that they've fixed the problem upstream, and all that we need to do is wait (and perhaps backport the fix in the current kernel)?

I've just backported the patch, could you please test it? https://build.opensuse.org/package/binary/download/home:david_chang:bsc11055... Thank you! -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c22 --- Comment #22 from david chang --- (In reply to Kriton Kyrimis from comment #21)

(In reply to david chang from comment #20)

Yes! This one works!

Thanks for your conformation. I think you could use the kmp for now until the next kernel update.

I suppose I should add a comment in the upstream bug that I opened. Would something like "fixed with commit 05212ba" be appropriate there?

Yeah, that's a good idea. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c24 --- Comment #24 from Kriton Kyrimis --- (In reply to david chang from comment #23)

You could get the update via the next kernel update.

Kernel 4.18.6-1.5, that came out today, does not fix the problem, requiring manual installation of the patched r8169.ko for the network adapter to work. I am not reopening the bug, as I assume that this kernel had already been in the pipeline without the patch, but please confirm. Thanks. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1105573 http://bugzilla.opensuse.org/show_bug.cgi?id=1105573#c26 --- Comment #26 from Kriton Kyrimis --- Upstream added the fix in kernel 4.18.8, as well. Confirming that the problem is fixed in Tumbleweed kernel 4.18.8-1.3. -- You are receiving this mail because: You are on the CC list for the bug.