The VM-to-VM link with driver domain confirmed working:
SR-IOV still beats it though.
I’m also suspecting PV PCI passthrough might be borked:
[ 349.614026] pciback 0000:17:01.0: xen-pciback: Driver tried to write to a read-only configuration space field at offset 0x110, size 4. This may be harmless, but if you have problems with your device:
1) see permissive attribute in sysfs
2) report problems to the xen-devel mailing list along with details of your device obtained from lspci.
[ 349.742430] pciback 0000:17:01.4: xen_pciback: vpci: assign to virtual slot 0
[ 349.742535] pciback 0000:17:01.4: registering for 3
[ 350.318006] pciback 0000:17:01.0: enabling device (0100 -> 0102)
[ 356.414446] xen-blkback: backend/vbd/3/5632: using 2 queues, protocol 1 (x86_64-abi) persistent grants
[ 361.991788] pciback 0000:17:01.4: xen-pciback: Driver tried to write to a read-only configuration space field at offset 0x110, size 4. This may be harmless, but if you have problems with your device:
1) see permissive attribute in sysfs
2) report problems to the xen-devel mailing list along with details of your device obtained from lspci.
[ 362.458767] pciback 0000:17:01.4: enabling device (0100 -> 0102)
This is for VFs. And I did notice some issues with the passed-through devices, like broken connections between VMs, failed pacman
pulls and the like.