Comment # 12 on bug 1048348 from Joerg Roedel

Hey Bruno,

(In reply to Bruno Friedmann from comment #10)
> Created attachment 732286 [details]
> dmesg with 4.4.76 iommu enabled
> 
> So I've installed 4.4.76 patched proposed kernel.
> Since then I've rebooted there's no evidence of crash, which is good.
> I'm doing heavy testing with vm and heavy load on io to make sure iommu can
> resist.
> I'm still seeing now AMD-Vi: Command buffer timeout (even before the vm was
> loaded)
> I don't know if they are important or not. The host and the guests seems to
> work as expected.

Thanks a lot for testing this. The messages about the "Command buffer timeout"
are still worrisome to me. They come from a back-ported upstream patch that
went into 4.13. Can you please test v4.13-rc1 on the box when it comes out
(probably this Sunday) and report whether you see the same messages there?

I'd like to isolate if the patch has an issue in general or if this is another
problem with my back-port.

Thanks a lot, Joerg