Bug#1004154: xserver-xorg-video-qxl: XOrg frequently crashes when using qxl driver: qxl(0): error doing QXL_ALLOC

[Felix Leimbach <felix@leimis.de>]

Thanks for looking into this Bernhard!

I found the problem vanishes if I downgrade my kernel from 5.15.0-2 to 5.10.0-10.
So it seems to be a regression on the kernel side.

Yes, it is a virtual qemu/kvm machine. I increased the VRAM years ago, when I experimented with spice's h264 acceleration.

Relevant command line:
-vga none -device qxl-vga,ram_size_mb=128,vgamem_mb=32,vram_size_mb=128,vram64_size_mb=256,max_outputs=1

Full command line:
/usr/bin/qemu-system-x86_64 -name fllin -M q35 -spice port=5906,addr=10.84.1.72,disable-ticketing -cpu host -smp 4 -drive file=fllin.raw,if=virtio,format=raw,index=0,cache=none,aio=native -m 16384 -net nic,model=virtio,macaddr=52:54:00:74:01:06 -net tap,ifname=vm-fllin,script=no,downscript=no -k de -monitor tcp:127.0.0.1:5956,server,nowait -rtc base=utc,clock=host -vga none -device qxl-vga,ram_size_mb=128,vgamem_mb=32,vram_size_mb=128,vram64_size_mb=256,max_outputs=1 -device virtio-serial,id=virtio-serial1 -chardev spicevmc,id=vdagent,debug=0,name=vdagent -device virtserialport,bus=virtio-serial1.0,chardev=vdagent,name=com.redhat.spice.0 -rtc base=localtime -daemonize --device virtio-balloon -boot c --enable-kvm -nodefaults -object rng-random,id=rng0,filename=/dev/urandom -device virtio-rng-pci,rng=rng0 -device intel-hda -device hda-duplex

Let me know if I can further assist in debugging this.

FYI, Redhat has the same bug reported: https://bugzilla.redhat.com/show_bug.cgi?id=2020633