Am 07.10.22 um 00:42 schrieb Jason Gunthorpe:
On Thu, Oct 06, 2022 at 01:53:15PM -0600, Alex Williamson wrote:
On Thu, 6 Oct 2022 09:40:35 -0300
Jason Gunthorpe <jgg@xxxxxxxxxx> wrote:
Testing has shown that virtnodedevd will leave the group FD open for long
periods, even after all the cdevs have been destroyed. This blocks
destruction of the VFIO device and is undesirable.
That approach was selected to accomodate SPAPR which has an broken
lifecyle model for the iommu_group. However, we can accomodate SPAPR by
realizing that it doesn't use the iommu core at all, so rules about
iommu_group lifetime do not apply to it.
Giving the KVM code its own kref on the iommu_group allows the VFIO core
code to release its iommu_group reference earlier and we can remove the
sleep that only existed for SPAPR.
Jason Gunthorpe (3):
vfio: Add vfio_file_is_group()
vfio: Hold a reference to the iommu_group in kvm for SPAPR
vfio: Make the group FD disassociate from the iommu_group
drivers/vfio/pci/vfio_pci_core.c | 2 +-
drivers/vfio/vfio.h | 1 -
drivers/vfio/vfio_main.c | 90 +++++++++++++++++++++-----------
include/linux/vfio.h | 1 +
virt/kvm/vfio.c | 45 +++++++++++-----
5 files changed, 94 insertions(+), 45 deletions(-)
Containers aren't getting cleaned up with this series, starting and
shutting down a libvirt managed VM with vfio-pci devices, an mtty mdev
device, and making use of hugepages, /proc/meminfo shows the hugepages
are not released on VM shutdown and I'm unable to subsequently restart
the VM. Thanks,
Oh, I'm surprised the s390 testing didn't hit this!!
I guess its because that most of our CI testcases create a new ephemeral
guest for each testcase. We do test reboot but not shutdown and restart.
Will have a look if that can be improved.
