FS#33271 - [linux] 3.6.10-1 Wake on Lan does not work using r8169 kernel module

Attached to Project: Arch Linux
Opened by portenier tiziano (tiziano) - Wednesday, 02 January 2013, 16:39 GMT
Last edited by Tobias Powalowski (tpowa) - Thursday, 31 January 2013, 21:07 GMT
Task Type Bug Report
Category Upstream Bugs
Status Closed
Assigned To Tobias Powalowski (tpowa)
Thomas Bächler (brain0)
Architecture x86_64
Severity Medium
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:
When waking a host by sending a magic packet over lan, a subsequent shutdown will imply some kind of reboot instead of power off the host. If the host was powered on by pressing the power button, a subsequent shutdown works as expected.

The NIC is a RTL8111/8168B. Whenever power off is initiated by typing: "$ sudo shutdown -h now", the system powers off but restarts after about 5 seconds of being powered off. This only happens if the system was previously started via WOL. After this strange kind of reboot as well as after power up the machine using the power button, shutdown works as expected until the next boot via WOL. The Wake-on mode can be either set to u, m, or g, it does not seem to matter.

This is a newly built rig and thus I don't know which version of linux introduced this bug. Archlinux was installed without any further packages: no acpid, no X, just a pure fresh installation. The GPU was removed after installing, thus the host is accessible over ssh only. Since I don't know if the issue is present in other distributions (I don't even know which distribution actually uses such a recent kernel version), I post it here and not upstream. Feel free to close this report if I was wrong.

Two workarounds are known to me:
i) install r8168-all from AUR and blacklist r8169. This is not an option to me, since a forgotten rebuild of the module after a kernel update will leave the machine in a unaccessible state (accessible over ssh only).

ii) install and use linux-lts. This is the option I actually use.

Additional info:
$ lspci|grep 81
03:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 06)

Steps to reproduce:
* Install Archlinux
* Enable Wake on LAN
* Boot by sending a magic packet over LAN
* type: "$ shutdown -h now" as root
This task depends upon

Closed by  Tobias Powalowski (tpowa)
Thursday, 31 January 2013, 21:07 GMT
Reason for closing:  Not a bug
Comment by Gerardo Exequiel Pozzi (djgera) - Wednesday, 02 January 2013, 17:43 GMT
Please test 3.7.1, to see if such issue persist. Otherwise report to upstream. There is nothing to do here with packaging.
Comment by portenier tiziano (tiziano) - Saturday, 05 January 2013, 11:21 GMT
I tested with linux-3.6.11-1, where the problem is still present. 3.7.1 is not available here, isn't that still in testing? If you say "there's nothing to do here with packaging": Does this mean that arch does not make any changes to the kernel but just packs what is delivered from upstream? If this is the case, should I close this task and report it upstream?
Comment by Tobias Powalowski (tpowa) - Wednesday, 23 January 2013, 16:09 GMT
Status on 3.7.x?
Comment by portenier tiziano (tiziano) - Friday, 25 January 2013, 22:14 GMT
Well using 3.7.4-1-ARCH introduces another issue: Whenever I try to shutdown or reboot over ssh, the system seems to hang somewhere during shutdown. It is no more accessible over ssh but does never halt. The issue is not present if I plug in a GPU thus it is hard to track down where the heck it comes from. Therefore I cannot test if the issue concerning WOL persists in 3.7.4-1-ARCH, unfortunately. Is there any way how I could 'see' what is going on during shutdown without using a physical GPU + screen?
Comment by portenier tiziano (tiziano) - Friday, 25 January 2013, 22:35 GMT
I finally had a GPU available to test with 3.7.4-1-ARCH. The issue initially described still persists in 3.7.4-1-ARCH :(
Comment by portenier tiziano (tiziano) - Saturday, 26 January 2013, 12:37 GMT
Today I was able to test another (intel based) NIC using the e1000e kernel module. The symptoms are equal thus it does not concern the r8169 module. I have to test it with a different motherbord and maybe with another distribution to track this down.
Comment by portenier tiziano (tiziano) - Thursday, 31 January 2013, 21:02 GMT
I finally replaced the mainboard (different model, same NIC) and the problem has gone. It seems to be a bug in the appropriate mainboard (ASUS P8H61 pro) thus it is absolutly unrelated to arch linux. Please excuse me for wasting your time. This task can be closed I think.

Loading...