Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

failed to setup container for group 51: memory listener initialization failed #9553

Open
wizardyjh opened this issue Apr 25, 2024 · 10 comments
Labels
area/gpu Issues specific to GPU/PCIe bug Incorrect behaviour needs-review Needs to be assessed by the team.

Comments

@wizardyjh
Copy link

wizardyjh commented Apr 25, 2024

I got an error when i run a kata container with gpu passthrough:

command:

nerdctl run --runtime "io.containerd.kata.v2" --mount type=bind,src=/dev,dest=/dev --privileged=true --security-opt privileged-without-host-devices --device /dev/vfio/51 --name=test -d ubuntu:latest

error:
failed to create shim task:QMP command failed:vfio 0000:ca:00.0: failed to setup container for group 51: memory listener initialization failed:Region dimm1:vfio_dma_map(0x55555774cdc0,0x100000000,0x1380000000,0x7f87b3ffe000)=-14(Bad address):unknown

containerd:1.7.2
kata:2.5.2

the host has 128G memory,if i set 'default_memory' > 62G , it report error

@wizardyjh wizardyjh added bug Incorrect behaviour needs-review Needs to be assessed by the team. labels Apr 25, 2024
@zvonkok zvonkok added the area/gpu Issues specific to GPU/PCIe label Apr 25, 2024
@zvonkok
Copy link
Contributor

zvonkok commented Apr 25, 2024

Are you saying that it works with default_memory < 62G ?

@zvonkok
Copy link
Contributor

zvonkok commented Apr 25, 2024

Any reason you're doing these flags? --mount type=bind,src=/dev,dest=/dev --privileged=true

@wizardyjh
Copy link
Author

Are you saying that it works with default_memory < 62G ?

yes

@wizardyjh
Copy link
Author

wizardyjh commented Apr 25, 2024

Any reason you're doing these flags? --mount type=bind,src=/dev,dest=/dev --privileged=true

This is not necessary, but remove these don't change the result.
thanks!

@zvonkok
Copy link
Contributor

zvonkok commented Apr 25, 2024

Are you using OVMF or the default empty setting? Can you share your config.toml ?

@wizardyjh
Copy link
Author

wizardyjh commented Apr 25, 2024

@zvonkok
image

@zvonkok
Copy link
Contributor

zvonkok commented Apr 25, 2024

Any chance you can use a newer Kata release? We've done a lot of fixes for Kata 3.xx

@wizardyjh
Copy link
Author

Any chance you can use a newer Kata release? We've done a lot of fixes for Kata 3.xx

not work on kata 3.4

@zvonkok
Copy link
Contributor

zvonkok commented Apr 30, 2024

What are your bios settings, ACS, Iommu and Virt enabled? What is your kernel commandline?

@wizardyjh
Copy link
Author

image
@zvonkok

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
area/gpu Issues specific to GPU/PCIe bug Incorrect behaviour needs-review Needs to be assessed by the team.
Projects
Issue backlog
  
To do
Development

No branches or pull requests

2 participants