AMD-Vi: Event logged [IO_PAGE_FAULT (Ethernet controller)

Originally Posted by kjhambrick Ook -- Does this Ubuntu Thread sound familiar ? What version of the Realtek Module is loaded ? I found it on google like this ... There seem to be a lot of exact hits for that Kernel Message. Since you're not using the adapter maybe it could be blacklisted or ??? HTH. -- kjh

Yeah, I saw that one. And a lot of similar reports. I'm going to disable IOMMU and see what happens. And I never thought about blacklisting the driver, that might be worth trying...

Originally Posted by Ook Yeah, I saw that one. And a lot of similar reports. I'm going to disable IOMMU and see what happens. And I never thought about blacklisting the driver, that might be worth trying...

So the adventure continues. I disabled IOMMU, and now 1) The kernel log gets flooded with nommu errors, and 2) The nic will come up, but it will not get an ip address. Enabled IOMMU, nommu errors stop, nic will get an ip address and we are back to the flood of AMD-Vi: Event logged [IO_PAGE_FAULT in kernel log.

Switched back to the onboard realtek nic, installed latest realtek driver. Flood of AMD-Vi: Event logged [IO_PAGE_FAULT stops. So far system is stable...

And a final report. After installing the latest driver from Realtek, this problem stopped. At this point I think it's fair to say the driver that comes with the 4.4.88 kernel is at fault.

I also believe the driver for the Intel nic is likewise faulty, as it gives the same error, even though it does not disconnect the nic. The Realtec driver actually disconnects the nic.

I know this is an old thread, but I am now getting the exact same problem after updating to the 4.4.118 kernel:

[ 8559.116434] AMD-Vi: Event logged [IO_PAGE_FAULT device=07:00.0 domain=0x000d address=0x0000000000cca000 flags=0x0050]

and device 07:00:0

07:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller (rev 15)

Last time I fixed this by manually installing the nic driver downloaded from the realtek site, and that worked again this time. AFAICT the realtek nic driver in the 4.4.118 kernel still has the same problem. The one on the realtek website, which was last updated about six months ago, fixes this. I poked around into the kernel source and it looks like this kernel uses the RTL8169 2.3LK-NAPI driver, but I have no idea how old it is. The realtek website has the 8168 8.45.8, from last year. The source is very different, and I don't have the time to dive into the source to see if anything interesting pops out.

Anyhow, that is it. 4.4.118 kernel still has the buggy realtek driver, the one from the realtek website fixes the problem.

PS: I'm in the process of filing a bug report, probably won't be able to finish that until tonight. This only happens when I'm transferring data from a Windows 7 box, but I can duplicate it very easily. Need to do some testing and gather logs so my bug report will hopefully be helpful enough for the maintainer to be able to do something about it.

Edit: Done. Looks like this was fixed in newer kernels, but the 4.4.x line is using the older buggy r8169.c driver. Not sure what the maintainer will do with this but I guess time will tell...