[vfio-users] Passing through eGPU over thunderbolt 3 partly works, how to debug further?

[Hagbard Celine]

2017-10-22 8:56 GMT+02:00, Roman Babenko <oknebab.namor at gmail.com>:
>The system freezes at random times, sometimes in
>2 hours,  sometimes in 5 mins, and  I need to power down both the host
>and the eGPU enclosure to be able to start the VM with working
>displays again.
>
>There are no errors that I can report either in dmesg output or in top
>output before the lockup.  The only weird thing I noticed is that
> immediately after the VM locks up the mouse cursor on the host still
> can be moved 1-2 seconds, then the host locks up and I need to reset
> the host and the enclosure.
>

I've had an issue with the exact same symptoms, only there was no
Thunderbolt involved. In my case it was a bad PCIe-sata controller
that I tried passing-trough. It was a $13 ASM1062 that I bought from
China, when replaced with a $39 ASM1062 the problem went away. The
failing setup worked flawlessly in both Linux and Windows as long as I
did not do virtualisation. The only visible difference between the two
ASM1062 was the PCB-layout, and that the failing one had "classical"
capacitors while the working one had only SMD-components in addition
to the controller-chip and sata-sockets.

Hope this info helps.

-Hagbard Celine