diff options
Diffstat (limited to 'results/classifier/accel-gemma3:12b/vmm/1645')
| -rw-r--r-- | results/classifier/accel-gemma3:12b/vmm/1645 | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/results/classifier/accel-gemma3:12b/vmm/1645 b/results/classifier/accel-gemma3:12b/vmm/1645 new file mode 100644 index 000000000..9a347a06d --- /dev/null +++ b/results/classifier/accel-gemma3:12b/vmm/1645 @@ -0,0 +1,9 @@ + +qemu error `hotplug memory" error="QMP command failed: a used vhost backend has no free memory slots left"` +Description of problem: +When I create a Qemu VM with 8 Gpus and hot-plugging memory, this will return the error QMP command failed: a used vhost backend has no free memory slots left. I read some source file https://gitlab.com/qemu-project/qemu/-/blob/master/hw/virtio/vhost-user.c#L2077, and debug show u->user->memory_slots is 32, but this https://gitlab.com/qemu-project/qemu/-/blob/master/hw/virtio/vhost.c#L62 used_memslots is bigger than u->user->memory_slots. `u->user->memory_slots` is defined 32 by https://gitlab.com/qemu-project/qemu/-/blob/master/subprojects/libvhost-user/libvhost-user.h#L37, but I also see VHOST_USER_MAX_RAM_SLOTS defined 512 under x86 architecture. Can I improve `u->user->memory_slots` by any way? +Steps to reproduce: +1.crate kata containers with 8 Gpus +2.kata containers return error +Additional information: + |