Re: [PATCH v11 04/10] virtio-gpu: Support asynchronous fencing

["Alex Bennée" <alex.bennee@linaro.org>]

Dmitry Osipenko <dmitry.osipenko@collabora.com> writes:

> On 4/27/25 14:53, 刘聪 wrote:
>> Hi Dmitry,
>> 
>> The virglrender patch can fix the virgl issue, but the native context still fails to run on my machine.
>> I'm not sure if anyone has successfully run it on an ARM64 machine before.
>
> Thanks for the testing!
>
>> When running with Venus, the virtual machine can successfully run vkcube. However, when using the native context, a KVM error is triggered. Both my guest and host kernels are already updated to version 6.14.
>> 
>> Here are the commands and error messages I encountered:
>> 
>> ```
>> phytium@ubuntu:~/working/virglrenderer$
>> /opt/native-context-v11/bin/qemu-system-aarch64 --machine
>> virt,accel=kvm,memory-backend=mem1 -cpu host -smp 4 -m 4G -drive
>> file=/home/phytium/working/ubuntu24.04-aarch64-native-context,format=raw,if=virtio
>> -bios /usr/share/AAVMF/AAVMF_CODE.ms.fd -netdev user,id=net0 -device
>> virtio-net-pci,netdev=net0 -device
>> virtio-gpu-gl,hostmem=4G,blob=on,venus=on -object
>> memory-backend-memfd,id=mem1,size=4G -display
>> sdl,gl=on,show-cursor=on -device usb-ehci,id=usb -device
>> usb-mouse,bus=usb.0 -device usb-kbd,bus=usb.0
>> phytium@ubuntu:~/working/virglrenderer$ 
>> phytium@ubuntu:~/working/virglrenderer$
>> /opt/native-context-v11/bin/qemu-system-aarch64 --machine
>> virt,accel=kvm,memory-backend=mem1 -cpu host -smp 4 -m 4G -drive
>> file=/home/phytium/working/ubuntu24.04-aarch64-native-context,format=raw,if=virtio
>> -bios /usr/share/AAVMF/AAVMF_CODE.ms.fd -netdev user,id=net0 -device
>> virtio-net-pci,netdev=net0 -device
>> virtio-gpu-gl,hostmem=4G,blob=on,drm_native_context=on -object
>> memory-backend-memfd,id=mem1,size=4G -display
>> sdl,gl=on,show-cursor=on -device usb-ehci,id=usb -device
>> usb-mouse,bus=usb.0 -device usb-kbd,bus=usb.0
>> error: kvm run failed Bad address
>>  PC=0000e2bcbbf31ab0 X00=0000e2bc9c3ae060 X01=0000e2bc7c02af00
>> X02=0000000000000014 X03=0000e2bc9c3ae000 X04=0000e2bc7c02af14
>> X05=0000e2bc9c3ae074 X06=0000000000000200 X07=0000e2bc7c02a8f8
>> X08=00000000000000de X09=0000000000000200 X10=0000000000001000
>> X11=0000000000000004 X12=0000e2bc7c0000b0 X13=0000000000000001
>> X14=0000000000000020 X15=0000e2bc9e465f93 X16=0000e2bcad6a01f0
>> X17=0000e2bcbbf31a80 X18=0000000000000093 X19=0000000000000060
>> X20=0000000000000074 X21=0000e2bc9e46c5f0 X22=0000e2bc9c3ae000
>> X23=0000000000000074 X24=0000c02241da83b0 X25=0000c02241da85a0
>> X26=0000c02241da85a0 X27=0000000000000014 X28=0000e2bc9e46c5f0
>> X29=0000e2bc9e46c610 X30=0000e2bcac809c38  SP=0000e2bc9e46c510
>> PSTATE=20001000 --C- EL0t
>> phytium@ubuntu:~/working/virglrenderer$ uname -a
>> Linux ubuntu 6.14.1-061401-generic #202504071048 SMP PREEMPT_DYNAMIC Mon Apr  7 11:34:37 UTC 2025 aarch64 aarch64 aarch64 GNU/Linux
>> ```
>
> Alex Bennée reported the very same problem with KVM on ARM + native ctx
> AMD dGPU in the past. You may try to add error messages to
> virt/kvm/kvm_main.c of host Linux kernel to find from where KVM error
> originates. Sounds like page refcounting may be not working properly
> on ARM.

Also what hardware is the machine? The AVA (and most things with the
same chipset) have a broken PCI which needs a workaround for unaligned
SIMD access:

  https://github.com/stsquad/linux/tree/testing/altra-tweaks-for-gpu

>
> +CC: Sean Christopherson

-- 
Alex Bennée
Virtualisation Tech Lead @ Linaro