Previously I had run qemu with no problems.
I was trying to set up audio , ( i just did
export QEMU_AUDIO_DRV=pa
and added -soundhw hda
to the QEMU command).
But when I launched, shut down, and launched the vm again, it just froze, and QEMU completely refused to die. I left it for about 10 minutes and, out of nowhere, the process finally stopped.
I rebooted my host, when I launched the vm, it worked fine, but when I shut it down and launched it again, the same thing happened. It’s always the same, it works the first time and then stops working.
If I try to shut down the host, the monitors go blank but then nothing happens, it keeps running, and I have to use the reboot button to get any reaction.
I undid the changes I had previously made to my QEMU command, and the problem persists.
If I don’t pass through the GPU, it works just fine.
I had been using another bios for the GPU though the romfile option, to get rid of code 43, but now even when I don’t use that bios, it still works, so I’m thinking maybe the GPU bios got corrupted. However, rom-parser outputs the same as it did before I had this problem:
Valid ROM signature found @0h, PCIR offset 234h PCIR: type 0 (x86 PC-AT), vendor: 1002, device: 6939, class: 030000 PCIR: revision 0, vendor revision: f30 Error, ran off the end
My setup:
Arch linux, I get the same problem on 4.14.14 and 4.14.11
Ryzen 1700
B350 board
GT 610 host GPU on the secondary slot
RX 380 guest GPU on the primary slot
My QEMU command:
qemu-system-x86_64
-cpu host,kvm=off -smp 1,sockets=1,cores=6,threads=1,maxcpus=6 -m 6G
-enable-kvm -machine q35,accel=kvm
-drive file=/dev/sda,format=raw
-device ivshmem-plain,memdev=ivshmem
-object memory-backend-file,id=ivshmem,share=on,mem-path=/dev/shm/looking-glass,size=64M
-device vfio-pci,host=25:00.0,x-vga=on,romfile=/home/e2298/vm/rom_gpu.rom
-spice port=5900,addr=127.0.0.1,disable-ticketing
-vga none