On 1/18/21 2:53 PM, Daniel Henrique Barboza wrote:
There is no need to bother with cgroup tearing down for absent PCI devices, given that their entries in the sysfs are already gone.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com> --- src/qemu/qemu_cgroup.c | 10 ++++++++++ 1 file changed, 10 insertions(+)
diff --git a/src/qemu/qemu_cgroup.c b/src/qemu/qemu_cgroup.c index f7146a71c9..050df21d87 100644 --- a/src/qemu/qemu_cgroup.c +++ b/src/qemu/qemu_cgroup.c @@ -467,6 +467,16 @@ qemuTeardownHostdevCgroup(virDomainObjPtr vm, if (!virCgroupHasController(priv->cgroup, VIR_CGROUP_CONTROLLER_DEVICES)) return 0;
+ /* Skip tearing down Cgroup for hostdevs that represents absent + * PCI devices, e.g. SR-IOV virtual functions that were removed from + * the host while the domain was still running. */ + if (virHostdevIsPCIDevice(dev)) { + const virDomainHostdevSubsysPCI *pcisrc = &dev->source.subsys.u.pci; + + if (!virPCIDeviceExists(&pcisrc->addr)) + return 0;
I would have skipped creating the temprorary variable, since it's only used once, but.. eh. Potato, potahtoe. Reviewed-by: Laine Stump <laine@redhat.com>
+ } + if (qemuDomainGetHostdevPath(dev, &path, NULL) < 0) return -1;