Hi,Xen-devel:
We are trying to implement the igpu sharing function for domU on the AMD (v2000 series) development board environment, and the software architecture is xen + ubuntu 22.04 (dom0) + Ubuntu
22.04 (Dom U). We found that AMD engineers like Mr. Huang
@ray.huang@xxxxxxx have posted information about virtio-gpu for Xen virtualization to the open source community and uploaded
the relevant code.
And We refer to the relevant information linked below to deploy on my AMD hardware platform:
[RFC PATCH 0/5] Add Xen PVH dom0 support for GPU - Huang Rui (kernel.org)
However, when we deploy virtio-GPU related software, we encounter the following issues:
1.
My code of xen /qemu /Virglrenderer checkout to the upstream-for-xen or upstream-for-xen-v2 branch, the xl configuration file is set as follows, see<
ubuntuhvm_virtIO-gpu.cfg>:
device_model_args_hvm= ["-display", "sdl,gl=on", "-device", "virtio-vga-gl,context_init=true,blob=true,hostmem=4G"]
2.
After the domU is started with the xl command, qemu and xen hypervisor will report the following errors:
xen hypervisor report <
xl_dmesg_upstream-for-xen-v2.txt>:
(XEN) d0v8 Over-allocation for d1: 1048833 > 1048832
(XEN) common/memory.c:277:d0v8 Could not allocate order=0 extent: id=1 memflags=0xc0 (192 of 2048)
qemu report in log file <
qemu-dm-ubuntuU_b.hvm.log>:
qemu-system-i386: -device virtio-vga-gl,context_init=true,blob=true,hostmem=4G: xen: failed to populate ram at 110080000
3.
I tried to start pvh dom0 but it didn't work, it looks the same regardless of whether I'm using upstream-for-xen-v2 or upstream-for-xen branch linux
:
Logs in attachment:<
serial_xen_hypervisor-linux-upstream-for-xen-v2.txt>
So I have some questions for you :
1.
Does Dom 0 need to be configured with PVH for VirtIO-GPU?
2.
What may cause the above issues?
We look forward to hearing from you!
Thanks & Best regards!