- Devel - libvirt List Archives

[libvirt] [RFC] Add support for QEMU's l3-cache CPU property
by Jiri Denemark 20 Apr '17

20 Apr '17

Hi, Apparently, reporting a level 3 cache on a virtual CPU can dramatically increase performance in some use cases [1]. The interesting part is that l3-cache=on does not provide the real CPU cache data, it's just making it up. Anyway, we should be able to enable this via libvirt. And since there is another property which enables real CPU cache data to be passed to a guest, I suggest the following /domain/cpu/cache element equivalent to l3-cache=on: <cache level='3' mode='emulate'/> If we need to add support for passing the real CPU cache data, we can do that with <cache level='3' mode='passthrough'/> or we can even make the level attribute optional and support <cache mode='passthrough'/> Missing cache element means default behaviour of the hypervisor and we can eventually add <cache mode='disable'/> to turn off passing any CPU cache info to the guest. But I think we should now focus only on <cache level='3' mode='emulate'/> and leaving the rest for the future when we actually need it. This is how a more complete example would look like: <cpu mode='custom' match='exact'> <model>Broadwell</model> <cache level='3' mode='emulate'/> </cpu> And libvirt would translate it into -cpu Broadwell,l3-cache=on. Do you have any thoughts about the XML schema or naming? Thanks, Jirka [1] https://patchwork.kernel.org/patch/9308401/

2 5

[libvirt] [PATCH] virsh: report errors in virshInit()
by Daniel P. Berrange 20 Apr '17

20 Apr '17

There are several functions in virshInit which can fail, especially when running win32 builds under WINE. Currently virsh just exits without reporting what error happened. Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- tools/virsh.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/tools/virsh.c b/tools/virsh.c index 31e23bd..90f8125 100644 --- a/tools/virsh.c +++ b/tools/virsh.c @@ -358,16 +358,22 @@ virshInit(vshControl *ctl) /* set up the library error handler */ virSetErrorFunc(NULL, vshErrorHandler); - if (virEventRegisterDefaultImpl() < 0) + if (virEventRegisterDefaultImpl() < 0) { + vshReportError(ctl); return false; + } - if (virThreadCreate(&ctl->eventLoop, true, vshEventLoop, ctl) < 0) + if (virThreadCreate(&ctl->eventLoop, true, vshEventLoop, ctl) < 0) { + vshReportError(ctl); return false; + } ctl->eventLoopStarted = true; if ((ctl->eventTimerId = virEventAddTimeout(-1, vshEventTimeout, ctl, - NULL)) < 0) + NULL)) < 0) { + vshReportError(ctl); return false; + } if (ctl->connname) { /* Connecting to a named connection must succeed, but we delay -- 2.9.3

2 2

[libvirt] [PATCH] Increase default task limit for libvirtd
by Jim Fehlig 20 Apr '17

20 Apr '17

libvirtd can spawn threads/tasks when creating new domains for some hypervisors such as Xen's libxl driver, quickly reaching the cgroups pids controller default TasksMax setting of 512. When the limit is reached, attempting to create additional domains results in an error from the cgroups pids controller, e.g. kernel: [71282.213347] cgroup: fork rejected by pids controller in /system.slice/libvirtd.service Depending on domain type and configuration, anywhere from 4-7 threads/tasks may be created by libxl when starting a domain. In order to support 4096 domains, similar to commit 27cd763500, increase the TasksMax setting in libvirtd.service to 4096 * 8 = 32768 tasks. Signed-off-by: Jim Fehlig <jfehlig(a)suse.com> --- daemon/libvirtd.service.in | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/daemon/libvirtd.service.in b/daemon/libvirtd.service.in index 899abdf37..fbaf02f3b 100644 --- a/daemon/libvirtd.service.in +++ b/daemon/libvirtd.service.in @@ -29,6 +29,11 @@ Restart=on-failure # If changing this, also consider virtlogd.service & virtlockd.service # limits which are also related to number of guests LimitNOFILE=8192 +# The cgroups pids controller can limit the number of tasks started by +# the daemon, which can limit the number of domains for some hypervisors. +# A conservative default of 8 tasks per guest results in a TasksMax of +# 32k to support 4096 guests. +TasksMax=32768 [Install] WantedBy=multi-user.target -- 2.11.0

2 1

[libvirt] increase daemon task limit
by Jim Fehlig 20 Apr '17

20 Apr '17

I recently received a report of a libvirt+Xen installation reaching the pids cgroup controller TasksMax limit kernel: [71282.213347] cgroup: fork rejected by pids controller in /system.slice/libvirtd.service The default setting of TaskMax is 512 on this system # cat /sys/fs/cgroup/pids/system.slice/libvirtd.service/pids.max 512 Depending on domain type and configuration, I've noticed between 5-7 tasks (IO threads, qemu process, etc.) are created when starting a domain. It doesn't take too many domains before the 512 limit is reached. LimitNOFILE was recently changed by commit 27cd763500 to support 4096 domains. Following similar logic, would it be ok to increase TasksMax to 32768? That would accommodate 4096 domains with 8 tasks each. TasksMax also supports the special value of "infinity", but that seems a bit aggressive to me. Regards, Jim

2 3

[libvirt] [PATCH] Fix error reporting when poll returns POLLHUP/POLLERR
by Daniel P. Berrange 20 Apr '17

20 Apr '17

In the RPC client event loop code, if poll() returns only a POLLHUP or POLLERR status, then we end up reporting a bogus error message: error: failed to connect to the hypervisor error: An error occurred, but the cause is unknown We do actually report an error, but we virNetClientMarkClose method has already captured the error status before we report it, so the real error gets thrown away. The key fix is to report the error before calling virNetClientMarkClose(). In changing this, we also split out reporting of POLLHUP vs POLLERR to make any future bugs easier to diagnose. Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/rpc/virnetclient.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/src/rpc/virnetclient.c b/src/rpc/virnetclient.c index 5174614..c959747 100644 --- a/src/rpc/virnetclient.c +++ b/src/rpc/virnetclient.c @@ -1743,10 +1743,16 @@ static int virNetClientIOEventLoop(virNetClientPtr client, if (error) goto error; - if (fds[0].revents & (POLLHUP | POLLERR)) { + if (fds[0].revents & POLLHUP) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("received hangup event on socket")); virNetClientMarkClose(client, VIR_CONNECT_CLOSE_REASON_EOF); + goto error; + } + if (fds[0].revents & POLLERR) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("received hangup / error event on socket")); + _("received error event on socket")); + virNetClientMarkClose(client, VIR_CONNECT_CLOSE_REASON_ERROR); goto error; } } -- 2.9.3

2 1

[libvirt] [PATCH v2] daemon: Fix domain name leak in error path
by Wang King 20 Apr '17

20 Apr '17

Domain name duplicated in make_nonnull_domain, but not freed when virTypedParamsSerialize return negative. Remove useless error label incidentally. --- daemon/remote.c | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) diff --git a/daemon/remote.c b/daemon/remote.c index 1610fea..d8a55c7 100644 --- a/daemon/remote.c +++ b/daemon/remote.c @@ -648,7 +648,7 @@ remoteRelayDomainEventBlockJob(virConnectPtr conn, /* build return data */ memset(&data, 0, sizeof(data)); if (VIR_STRDUP(data.path, path) < 0) - goto error; + return -1; data.type = type; data.status = status; make_nonnull_domain(&data.dom, dom); @@ -667,9 +667,6 @@ remoteRelayDomainEventBlockJob(virConnectPtr conn, } return 0; - error: - VIR_FREE(data.path); - return -1; } @@ -1025,7 +1022,7 @@ remoteRelayDomainEventBlockJob2(virConnectPtr conn, memset(&data, 0, sizeof(data)); data.callbackID = callback->callbackID; if (VIR_STRDUP(data.dst, dst) < 0) - goto error; + return -1; data.type = type; data.status = status; make_nonnull_domain(&data.dom, dom); @@ -1035,9 +1032,6 @@ remoteRelayDomainEventBlockJob2(virConnectPtr conn, (xdrproc_t)xdr_remote_domain_event_block_job_2_msg, &data); return 0; - error: - VIR_FREE(data.dst); - return -1; } @@ -1067,7 +1061,7 @@ remoteRelayDomainEventTunable(virConnectPtr conn, (virTypedParameterRemotePtr *) &data.params.params_val, &data.params.params_len, VIR_TYPED_PARAM_STRING_OKAY) < 0) - return -1; + goto error; remoteDispatchObjectEventSend(callback->client, remoteProgram, REMOTE_PROC_DOMAIN_EVENT_CALLBACK_TUNABLE, @@ -1075,6 +1069,10 @@ remoteRelayDomainEventTunable(virConnectPtr conn, &data); return 0; + error: + VIR_FREE(data.dom.name); + return -1; + } @@ -1207,13 +1205,16 @@ remoteRelayDomainEventJobCompleted(virConnectPtr conn, (virTypedParameterRemotePtr *) &data.params.params_val, &data.params.params_len, VIR_TYPED_PARAM_STRING_OKAY) < 0) - return -1; + goto error; remoteDispatchObjectEventSend(callback->client, remoteProgram, REMOTE_PROC_DOMAIN_EVENT_CALLBACK_JOB_COMPLETED, (xdrproc_t)xdr_remote_domain_event_callback_job_completed_msg, &data); return 0; + error: + VIR_FREE(data.dom.name); + return -1; } -- 2.8.3

2 1

[libvirt] [PATCH 0/2] qemu: hotplug: Write status XML only during successful vcpu hot(un)plug
by Peter Krempa 20 Apr '17

20 Apr '17

See the second patch. Peter Krempa (2): qemu: hotplug: Unexport qemuDomainHotplugDelVcpu qemu: hotplug: Don't save status XML when monitor is closed src/qemu/qemu_hotplug.c | 31 +++++++++++++++++-------------- src/qemu/qemu_hotplug.h | 3 --- 2 files changed, 17 insertions(+), 17 deletions(-) -- 2.12.2

2 3

[libvirt] [PATCH 0/9] qemu: More CPU enhancements and fixes
by Jiri Denemark 19 Apr '17

19 Apr '17

This is technically a v2 of the not yet pushed part of "qemu: Use migratable host CPU model from QEMU" series, but the patches were significantly rewritten based on the review comments and on the need for additional regression fix. Current libvirt makes a migratable host CPU model by removing features marked in cpu_map.xml with migratable='no'. This is not ideal because cpu_map.xml is a static database while feature migratability may differ depending on a hypervisor or its version. Thus we should preferably use the data we got from QEMU. And to make things even more complicated, we need to use both CPUID data we probed and data from QEMU when checking whether a guest CPU can run on the host to maintain backward compatibility with older libvirt. Thus we need to store three different views of the host CPU model with each QEMU binary. Jiri Denemark (9): qemu: Introduce virQEMUCapsSetHostModel qemu: Move qemuCaps CPU data copying into a separate function qemu: Introduce virQEMUCapsHostCPUDataClear qemu: Move qemuCaps host CPU data in a struct qemu: Prepare qemuCaps for multiple host CPU defs qemu: Pass migratable host CPU model to virCPUUpdate cpu: Drop feature filtering from virCPUUpdate cpu: Introduce virCPUGetHostIsSupported qemu: Use more data for comparing CPUs src/cpu/cpu.c | 20 +++ src/cpu/cpu.h | 3 + src/cpu/cpu_x86.c | 9 +- src/libvirt_private.syms | 1 + src/qemu/qemu_capabilities.c | 288 +++++++++++++++++++++++++++++-------------- src/qemu/qemu_capabilities.h | 16 ++- src/qemu/qemu_command.c | 3 +- src/qemu/qemu_process.c | 6 +- tests/cputest.c | 7 +- 9 files changed, 247 insertions(+), 106 deletions(-) -- 2.12.2

2 19

[libvirt] [PATCH] virt-storage-rbd: fix typo s/RDB/RBD/ in configure check
by Daniel P. Berrange 19 Apr '17

19 Apr '17

Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- Pushed as trivial build fix. m4/virt-storage-rbd.m4 | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/m4/virt-storage-rbd.m4 b/m4/virt-storage-rbd.m4 index 727bd71..17e2115 100644 --- a/m4/virt-storage-rbd.m4 +++ b/m4/virt-storage-rbd.m4 @@ -36,7 +36,7 @@ AC_DEFUN([LIBVIRT_STORAGE_CHECK_RBD], [ LIBS="$old_LIBS" fi - if test "$LIBRDB_FOUND" = "yes"; then + if test "$LIBRBD_FOUND" = "yes"; then with_storage_rbd=yes AC_DEFINE_UNQUOTED([WITH_STORAGE_RBD], [1], [whether RBD backend for storage driver is enabled]) -- 2.9.3

1 0

[libvirt] [PATCH v3 0/2] default USB controller for PPC
by Pavel Hrdina 19 Apr '17

19 Apr '17

Pavel Hrdina (2): conf: add a new parse flag VIR_DOMAIN_DEF_PARSE_ABI_UPDATE_MIGRATION qemu_domain: use correct default USB controller on ppc64 src/conf/domain_conf.c | 3 ++- src/conf/domain_conf.h | 4 ++++ src/qemu/qemu_domain.c | 16 ++++++++++++---- src/qemu/qemu_migration_cookie.c | 2 +- 4 files changed, 19 insertions(+), 6 deletions(-) -- 2.12.2

2 3