Loading...

Linking RHIVOS CVEs to...

Migration: Automation ...

SWIFT: POC Conversion

Sync from "Extern...

XML

Word

Printable

Type: Story
Resolution: Not a Bug
Priority: Undefined
Fix Version/s: None
Affects Version/s: None
Component/s: libvirt
Labels:
None

Severity:
None

AssignedTeam:
rhel-virt-core-libvirt-1
Sub-System Group:

ssg_virtualization

Story Points:
None
Blocked:
False
Ready:
False
Blocked Reason:

Hide

None

Show
None
Product Documentation Required:
None
Sprint:
None

Preliminary Testing:
None
Test Coverage:
None

Experience:

SFDC Cases Counter:
SFDC Cases Open:
SFDC Cases Links:

Planning:
None

Goal

Enabling passthrough of high vram (24+GB) GPUs
- With `X-PciMmio64Mb` left to its default value, the guest starts but the NVIDIA driver fails to probe the device
- See https://github.com/kubevirt/kubevirt/issues/11093 for more details

Acceptance Criteria

OVMF option PciMmio64Mb is not experimental anymore (no "X-" prefix)
Libvirt includes an API to set `PciMmio64Mb` to (at least) 65536
The resulting qemu process cmdline includes "-fw_cfg name=opt/ovmf/X-PciMmio64Mb,string=65536"

is related to

OCPBUGS-55974 Large or multiple vGPUs need bigger MMIO space

ASSIGNED

Assignee:: Andrea Bolognani

Reporter:: Jed Lejosne

Developer:: virt-maint

QA Contact:: virt-bugs

Votes:: 0 Vote for this issue

Watchers:: 9 Start watching this issue

Created:: 2024/05/03 1:16 PM

Updated:: 2025/08/15 2:14 PM

Resolved:: 2024/05/07 3:38 PM