2011-11-29

Cant get a connection with an R8169 NIC and Linux/Windows

And so, earlier today, I had a problem with my MSI Wind U230. Normally, I can connect via ethernet (A realtek 8111/8168B chip that used the r8169 driver)  without any problems, but today... it decided to not connect to anything. dmesg showed that the link was down and I only kept getting a "Cable unplugged" message from my DE (KDE 4.X in this incarnation of my linux installation).

Fortunately, wifi was available, and I checked a few forums and bugs that recommended installing the r8168 driver instead of using the stock r8169 that's shipped with Kubuntu. Then there was this other link (sorry, but I can't recall them anymore, since I hopped between dozens of pages looking for a solution) that mentioned that if you dual-boot (which I do), to check in the windows device manager and set a few wake-on-* settings to either enabled or disabled, depending on the driver version. Neither solution appealed to me for some reason. I've kept to using stock ubuntu kernels (they work just fine for my case, and am currently on 3.0.0-13-generic) and have no problems. As for the windows solution, I couldn't see where or how windows would put my NIC to sleep (more like a coma, was the way it was descrobed), since I jump between Linux and Windows 7, and my NIC never went into a coma before.

What worried me most was that I was playing with some grub options (specifically pcie_aspm=force and acpi_osi=Linux). Although I wasn't scared that those would be permanent errors (well, I was hoping they wouldn't), I wen't back and forth with different combinations in using them (side note: they didn't work for me in regard to power-saving).

What fixed the whole thing eventually was unplugging this laptop from power, and pulling out the battery. With all power sources disconnected, I held down the power button for a good 15-30 seconds (sorry, I wasn't counting, but it was a significant number of seconds though), and plugged in again. After booting, all was fine.

The only thing I could think of that might have triggered this was either the pcie_aspm=force or the acpi_osi=Linux or both, that put my nic to sleep. But after draining all the residual power from the board - which is what disconnecting all power sources and holding down the power button does - it seems whatever was got cached(?) in the nic got cleared and a new surge of juice set it right up again.

Ah well... something new that I never experienced before. Took me all of 30 minutes to figure out through trial and error.

No comments: