Re: [WHL] launch two UOSs in the same time
|
Jianjie Lin
As you suggest, I add each UOS launch script with an UUID value, and launch the file according to the order of uuid, defined in the vm_uuid.h file. And, Yes, only one UUID is sufficient. Thank you.
Actually, I found some documentation error in the version of v2.0 For example. In this tutorial: https://projectacrn.github.io/2.0/getting-started/rt_industry_ubuntu.html for build ACRN Documented as $ make all BOARD_FILE=misc/acrn-config/xmls/board-xmls/whl-ipc-i7.xml SCENARIO_FILE=misc/acrn-config/xmls/config-xmls/whl-ipc-i7/industry.xml RELEASE=0 $ sudo make install $ sudo cp build/hypervisor/acrn.bin /boot/acrn/
è It should be $ make all BOARD_FILE=misc/acrn-config/xmls/board-xmls/whl-ipc-i7.xml SCENARIO_FILE=misc/acrn-config/xmls/config-xmls/whl-ipc-i7/industry.xml RELEASE=0 $ sudo make install BOARD=whl-ipc-i7 SCENARIO=industry $ sudo cp build/hypervisor/acrn.bin /boot/acrn/
Otherwise, the sudo make install will automatically install BOARD as nuc7i7dnb and SCENARIO=sdc.
Another example,
if you did not define the CONFIG_HV_RAM_SIZE in the scenario file, and build the acrn-hypervisor, it will has a problem of .bss. In my case, the old whiskey board is automatilcally defined as 0x1400000, which is not enough for the new whiskey. I manually change the config in the build/hypervisor/.config, file. And rebuild again.
By the way, if I recompile the acrn-hypervisor with different scenario, should I recompile my acrn-kernel again.
Cheers Jianjie Lin
Von: acrn-users@... [mailto:acrn-users@...]
Im Auftrag von Geoffroy Van Cutsem
Thanks for the confirmation Jianjie Lin!
You shouldn’t have to add more UUIDs to that file though, using different ones in each launch script (by adding the “-U” option) should be sufficient, were you not able to do this without adding new ones??
Thanks!
From:
acrn-users@... <acrn-users@...>
On Behalf Of Jianjie Lin
Sent: Friday, September 3, 2021 9:43 am To: acrn-users@... Subject: Re: [acrn-users] [WHL] launch two UOSs in the same time
Hi Geoffory, Thank you very much. The problem is fixed. It really depends on the UUID in my case. I need to add different UUID from the vm_uuid.h .. Thanks again.
Cheers, Jianjie Lin
Von: acrn-users@... [mailto:acrn-users@...]
Im Auftrag von Geoffroy Van Cutsem
That header file contains a fixed (and static) list of UUIDs that can be used to launch VMs. I’m thinking that perhaps what is happening is the acrn-dm uses the same UUID by default for the second VM because it is not specified via the command-line. Adding the -U parameter will force it to use a UUID that we have chosen ourselves, it *must* be one from the list in that header file.
Does that help clarify?
Thanks, Geoffroy
From:
acrn-users@... <acrn-users@...>
On Behalf Of Jianjie Lin
Sorry, you mean the uuid can only assigned to the one which is saved in the vm_uuid.h? Am I missunderstanding?
Mit freundlichen Grüßen From:
acrn-users@... <acrn-users@...> on behalf of Geoffroy Van Cutsem <geoffroy.vancutsem@...>
I’m suspicious of something wrong with the UUIDs.
E.g.: -U 615db82a-e189-4b4f-8dbb-d321343e4ab3 (similar to what you see here: https://github.com/projectacrn/acrn-hypervisor/blob/master/misc/config_tools/data/sample_launch_scripts/nuc/launch_hard_rt_vm.sh#L47, but with a different UUID taken from here: https://github.com/projectacrn/acrn-hypervisor/blob/master/hypervisor/include/common/vm_uuids.h
Geoffroy
From:
acrn-users@... <acrn-users@...>
On Behalf Of Jianjie Lin
Is that happening no matter which order you try to load them ? .--> yes, no matter in the gpu version or console version. The first one is always correct, but the second one fails always.
Von:
acrn-users@... [mailto:acrn-users@...]
Im Auftrag von Geoffroy Van Cutsem
Hi Jianjie Lin,
Do you have more logs to share when the second VM fails to start (acrn-dm console & serial port output)? Is that happening no matter which order you try to load them?
A couple of remarks/questions: * Your industry.xml file only ever assigns pCPU 0 and 1 to all User VMs, but your console_launch_uos3_4.sh tries to use pCPUs 2 and 3 * Is gpu_launch_uos4.sh supposed to launch a Windows User VM? I see the ‘--windows’ parameter in it but it seems to want to use GVTg and picks up an image from a folder called “ubuntu_images”. The “--windows” parameter makes acrn-dm use the Oracle Virtio block device.
Thanks, Geoffroy
From:
acrn-users@... <acrn-users@...>
On Behalf Of Jianjie Lin
Hi ACRN Community,
We faced the problem of launching two UOSs at the same time. We deployed the whiskey lake board: whl-IPC-i7.xml, and we customize the scenario file with 4 Post-STD VM.
We can successfully launch one VM, but we can not create the second VM in the acrn kernel.
I attached the two launch file gpu_launch_uos3.sh console_launch_uos3_4.sh
Besides, I attach my board file and scenario file if it can help to find out the issue.
I add the following the command line in the file in the /etc/grub.d/40.cutsom for Gvt-g “i915.nuclear_pageflip=1 i915.avail_planes_per_pipe=0x010101 i915.domain_plane_owners=0x011100001111 i915.enable_gvt=1 i915.enable_guc=0 i915.enable_conformance_check=0”
For each launch file, we can launch the VM without any problem. But If I launch them at the same time, only the first one is successful, the another one is failed to created.
I go through the issue from GitHub. However, no direct solution is founded based on those comments. Do you have any idea how I can forward further?
Thank you very much firstly for your support.
Mit freundlichen Grüßen / Kind regards Jianjie Lin
E-Mail: Jianjie.Lin@... Mobil: +49-176-8443-6814 |
|
|