[libvirt] [PATCH] qemu: Create or remove cgroup when doing vcpu (un)hotpluging
Various APIs use cgroup to either set or get the statistics of host or guest. Hotplug or unhotplug new vcpus without creating or removing the cgroup for the vcpus could cause problems for those APIs. E.g. % virsh vcpucount dom maximum config 10 maximum live 10 current config 1 current live 1 % virsh setvcpu dom 2 % virsh schedinfo dom --set vcpu_quota=1000 Scheduler : posix error: Unable to find vcpu cgroup for rhel6.2(vcpu: 1): No such file or directory This patch fixes the problem by creating cgroups for each of the onlined vcpus, and destroying cgroups for each of the offlined vcpus. --- src/qemu/qemu_driver.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 53 insertions(+), 0 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 97ad23e..b8ceca1 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -3601,6 +3601,8 @@ static int qemudDomainHotplugVcpus(struct qemud_driver *driver, int vcpus = oldvcpus; pid_t *cpupids = NULL; int ncpupids; + virCgroupPtr cgroup = NULL; + virCgroupPtr cgroup_vcpu = NULL; qemuDomainObjEnterMonitor(driver, vm); @@ -3654,6 +3656,53 @@ static int qemudDomainHotplugVcpus(struct qemud_driver *driver, goto cleanup; } + if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) == 0) { + int rv = -1; + + if (nvcpus > oldvcpus) { + for (i = oldvcpus; i < nvcpus; i++) { + /* Create cgroup for the onlined vcpu */ + rv = virCgroupForVcpu(cgroup, i, &cgroup_vcpu, 1); + if (rv < 0) { + virReportSystemError(-rv, + _("Unable to create vcpu cgroup for %s(vcpu:" + " %d)"), + vm->def->name, i); + goto cleanup; + } + + /* Add vcpu thread to the cgroup */ + rv = virCgroupAddTask(cgroup_vcpu, cpupids[i]); + if (rv < 0) { + virReportSystemError(-rv, + _("unable to add vcpu %d task %d to cgroup"), + i, cpupids[i]); + virCgroupRemove(cgroup_vcpu); + goto cleanup; + } + + virCgroupFree(&cgroup_vcpu); + } + } else { + for (i = oldvcpus - 1; i >= nvcpus; i--) { + rv = virCgroupForVcpu(cgroup, i, &cgroup_vcpu, 0); + if (rv < 0) { + virReportSystemError(-rv, + _("Unable to access vcpu cgroup for %s(vcpu:" + " %d)"), + vm->def->name, i); + goto cleanup; + } + + /* Remove cgroup for the offlined vcpu */ + virCgroupRemove(cgroup_vcpu); + virCgroupFree(&cgroup_vcpu); + } + } + + virCgroupFree(&cgroup); + } + priv->nvcpupids = ncpupids; VIR_FREE(priv->vcpupids); priv->vcpupids = cpupids; @@ -3664,6 +3713,10 @@ cleanup: vm->def->vcpus = vcpus; VIR_FREE(cpupids); virDomainAuditVcpu(vm, oldvcpus, nvcpus, "update", rc == 1); + if (cgroup) + virCgroupFree(&cgroup); + if (cgroup_vcpu) + virCgroupFree(&cgroup_vcpu); return ret; unsupported: -- 1.7.7.6
On 10/09/2012 06:08 PM, Osier Yang wrote:
Various APIs use cgroup to either set or get the statistics of host or guest. Hotplug or unhotplug new vcpus without creating or removing the cgroup for the vcpus could cause problems for those APIs. E.g.
% virsh vcpucount dom maximum config 10 maximum live 10 current config 1 current live 1
% virsh setvcpu dom 2
% virsh schedinfo dom --set vcpu_quota=1000 Scheduler : posix error: Unable to find vcpu cgroup for rhel6.2(vcpu: 1): No such file or directory
This patch fixes the problem by creating cgroups for each of the onlined vcpus, and destroying cgroups for each of the offlined vcpus. --- src/qemu/qemu_driver.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 53 insertions(+), 0 deletions(-)
ACK, Martin
On 10/09/12 18:08, Osier Yang wrote:
Various APIs use cgroup to either set or get the statistics of host or guest. Hotplug or unhotplug new vcpus without creating or removing the cgroup for the vcpus could cause problems for those APIs. E.g.
% virsh vcpucount dom maximum config 10 maximum live 10 current config 1 current live 1
% virsh setvcpu dom 2
% virsh schedinfo dom --set vcpu_quota=1000 Scheduler : posix error: Unable to find vcpu cgroup for rhel6.2(vcpu: 1): No such file or directory
This patch fixes the problem by creating cgroups for each of the onlined vcpus, and destroying cgroups for each of the offlined vcpus. --- src/qemu/qemu_driver.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 53 insertions(+), 0 deletions(-)
s/unhotplug/hotunplug/g in your patch The code looks fine, so Martin's ACK stands. Peter
On 2012年10月10日 17:11, Peter Krempa wrote:
On 10/09/12 18:08, Osier Yang wrote:
Various APIs use cgroup to either set or get the statistics of host or guest. Hotplug or unhotplug new vcpus without creating or removing the cgroup for the vcpus could cause problems for those APIs. E.g.
% virsh vcpucount dom maximum config 10 maximum live 10 current config 1 current live 1
% virsh setvcpu dom 2
% virsh schedinfo dom --set vcpu_quota=1000 Scheduler : posix error: Unable to find vcpu cgroup for rhel6.2(vcpu: 1): No such file or directory
This patch fixes the problem by creating cgroups for each of the onlined vcpus, and destroying cgroups for each of the offlined vcpus. --- src/qemu/qemu_driver.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 53 insertions(+), 0 deletions(-)
s/unhotplug/hotunplug/g in your patch
The code looks fine, so Martin's ACK stands.
Thanks, but can you review more? :-) More patches to improve vcpupin related stuffs are posted, since they are related with each other tightly, better to push together after a thorough reviewing. https://www.redhat.com/archives/libvir-list/2012-October/msg00387.html Regards, Osier
participants (3)
-
Martin Kletzander -
Osier Yang -
Peter Krempa