- Devel - libvirt List Archives

[libvirt] [PATCH v3 1/1] perf: add support to perf event for MBM
by Qiaowei Ren 20 May '16

20 May '16

Some Intel processor families (e.g. the Intel Xeon processor E5 v3 family) introduced some RDT (Resource Director Technology) features to monitor or control shared resource. Among these features, MBM (Memory Bandwidth Monitoring), which is build on the CMT (Cache Monitoring Technology) infrastructure, provides OS/VMM a way to monitor bandwidth from one level of cache to another. With current perf framework, this patch adds support to perf event for MBM. Signed-off-by: Qiaowei Ren <qiaowei.ren(a)intel.com> --- include/libvirt/libvirt-domain.h | 26 ++++++++++++++++- src/libvirt-domain.c | 12 ++++++++ src/qemu/qemu_driver.c | 41 +++++++++++++++++++------- src/util/virperf.c | 63 ++++++++++++++++++++++++---------------- src/util/virperf.h | 2 ++ 5 files changed, 108 insertions(+), 36 deletions(-) diff --git a/include/libvirt/libvirt-domain.h b/include/libvirt/libvirt-domain.h index 160f20f..27ed29a 100644 --- a/include/libvirt/libvirt-domain.h +++ b/include/libvirt/libvirt-domain.h @@ -1900,10 +1900,34 @@ void virDomainStatsRecordListFree(virDomainStatsRecordPtr *stats); /** * VIR_PERF_PARAM_CMT: * - * Macro for typed parameter name that represents CMT perf event. + * Macro for typed parameter name that represents CMT perf event + * which can be used to measure the usage of cache (bytes) by + * applications running on the platform. It corresponds to the + * "perf.cmt" field in the *Stats APIs. */ # define VIR_PERF_PARAM_CMT "cmt" +/** + * VIR_PERF_PARAM_MBMT: + * + * Macro for typed parameter name that represents MBMT perf event + * which can be used to monitor total system bandwidth (bytes/s) + * from one level of cache to another. It corresponds to the + * "perf.mbmt" field in the *Stats APIs. + + */ +# define VIR_PERF_PARAM_MBMT "mbmt" + +/** + * VIR_PERF_PARAM_MBML: + * + * Macro for typed parameter name that represents MBML perf event + * which can be used to monitor the amount of data (bytes/s) sent + * through the memory controller on the socket. It corresponds to + * the "perf.mbml" field in the *Stats APIs. + */ +# define VIR_PERF_PARAM_MBML "mbml" + int virDomainGetPerfEvents(virDomainPtr dom, virTypedParameterPtr *params, int *nparams, diff --git a/src/libvirt-domain.c b/src/libvirt-domain.c index 4f473c9..73ae369 100644 --- a/src/libvirt-domain.c +++ b/src/libvirt-domain.c @@ -11441,6 +11441,18 @@ virConnectGetDomainCapabilities(virConnectPtr conn, * "block.<num>.physical" - physical size in bytes of the container of the * backing image as unsigned long long. * + * VIR_DOMAIN_STATS_PERF: Return perf event statistics. + * The typed parameter keys are in this format: + * "perf.cmt" - the usage of l3 cache (bytes) by applications running on the + * platform as unsigned long long. It is produced by cmt perf + * event. + * "perf.mbmt" - the total system bandwidth (bytes/s) from one level of cache + * to another as unsigned long long. It is produced by mbmt perf + * event. + * "perf.mbml" - the amount of data (bytes/s) sent through the memory controller + * on the socket as unsigned long long. It is produced by mbml + * perf event. + * * Note that entire stats groups or individual stat fields may be missing from * the output in case they are not supported by the given hypervisor, are not * applicable for the current state of the guest domain, or their retrieval diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index c4c4968..670f620 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -10051,6 +10051,8 @@ qemuDomainSetPerfEvents(virDomainPtr dom, if (virTypedParamsValidate(params, nparams, VIR_PERF_PARAM_CMT, VIR_TYPED_PARAM_BOOLEAN, + VIR_PERF_PARAM_MBMT, VIR_TYPED_PARAM_BOOLEAN, + VIR_PERF_PARAM_MBML, VIR_TYPED_PARAM_BOOLEAN, NULL) < 0) return -1; @@ -19494,24 +19496,38 @@ qemuDomainGetStatsBlock(virQEMUDriverPtr driver, #undef QEMU_ADD_COUNT_PARAM +#define QEMU_ADD_PERF_PARAM_ULL(record, maxparams, name, value) \ +do { \ + char param_name[VIR_TYPED_PARAM_FIELD_LENGTH]; \ + snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH, \ + "perf.%s", name); \ + if (virTypedParamsAddULLong(&(record)->params, \ + &(record)->nparams, \ + maxparams, \ + param_name, \ + value) < 0) \ + goto cleanup; \ +} while (0) + static int -qemuDomainGetStatsPerfCmt(virPerfPtr perf, +qemuDomainGetStatsPerfRdt(virPerfPtr perf, + virPerfEventType type, virDomainStatsRecordPtr record, int *maxparams) { - uint64_t cache = 0; + uint64_t value = 0; - if (virPerfReadEvent(perf, VIR_PERF_EVENT_CMT, &cache) < 0) + if (virPerfReadEvent(perf, type, &value) < 0) return -1; - if (virTypedParamsAddULLong(&record->params, - &record->nparams, - maxparams, - "perf.cache", - cache) < 0) - return -1; + QEMU_ADD_PERF_PARAM_ULL(record, maxparams, + virPerfEventTypeToString(type), + value); return 0; + + cleanup: + return -1; } static int @@ -19531,7 +19547,10 @@ qemuDomainGetStatsPerf(virQEMUDriverPtr driver ATTRIBUTE_UNUSED, switch (i) { case VIR_PERF_EVENT_CMT: - if (qemuDomainGetStatsPerfCmt(priv->perf, record, maxparams) < 0) + case VIR_PERF_EVENT_MBMT: + case VIR_PERF_EVENT_MBML: + if (qemuDomainGetStatsPerfRdt(priv->perf, i, + record, maxparams) < 0) goto cleanup; break; } @@ -19543,6 +19562,8 @@ qemuDomainGetStatsPerf(virQEMUDriverPtr driver ATTRIBUTE_UNUSED, return ret; } +#undef QEMU_ADD_PERF_PARAM_ULL + typedef int (*qemuDomainGetStatsFunc)(virQEMUDriverPtr driver, virDomainObjPtr dom, diff --git a/src/util/virperf.c b/src/util/virperf.c index bd65587..450b3ba 100644 --- a/src/util/virperf.c +++ b/src/util/virperf.c @@ -38,7 +38,7 @@ VIR_LOG_INIT("util.perf"); #define VIR_FROM_THIS VIR_FROM_PERF VIR_ENUM_IMPL(virPerfEvent, VIR_PERF_EVENT_LAST, - "cmt"); + "cmt", "mbmt", "mbml"); struct virPerfEvent { int type; @@ -110,10 +110,10 @@ virPerfGetEvent(virPerfPtr perf, } static int -virPerfCmtEnable(virPerfEventPtr event, +virPerfRdtEnable(virPerfEventPtr event, pid_t pid) { - struct perf_event_attr cmt_attr; + struct perf_event_attr rdt_attr; char *buf = NULL; char *tmp = NULL; unsigned int event_type, scale; @@ -127,32 +127,42 @@ virPerfCmtEnable(virPerfEventPtr event, if (virStrToLong_ui(buf, NULL, 10, &event_type) < 0) { virReportSystemError(errno, "%s", - _("failed to get cmt event type")); + _("failed to get rdt event type")); goto error; } VIR_FREE(buf); - if (virFileReadAll("/sys/devices/intel_cqm/events/llc_occupancy.scale", - 10, &buf) < 0) - goto error; + memset(&rdt_attr, 0, sizeof(rdt_attr)); + rdt_attr.size = sizeof(rdt_attr); + rdt_attr.type = event_type; + rdt_attr.inherit = 1; + rdt_attr.disabled = 1; + rdt_attr.enable_on_exec = 0; - if (virStrToLong_ui(buf, NULL, 10, &scale) < 0) { - virReportSystemError(errno, "%s", - _("failed to get cmt scaling factor")); - goto error; + switch (event->type) { + case VIR_PERF_EVENT_CMT: + if (virFileReadAll("/sys/devices/intel_cqm/events/llc_occupancy.scale", + 10, &buf) < 0) + goto error; + + if (virStrToLong_ui(buf, NULL, 10, &scale) < 0) { + virReportSystemError(errno, "%s", + _("failed to get cmt scaling factor")); + goto error; + } + event->efields.cmt.scale = scale; + + rdt_attr.config = 1; + break; + case VIR_PERF_EVENT_MBMT: + rdt_attr.config = 2; + break; + case VIR_PERF_EVENT_MBML: + rdt_attr.config = 3; + break; } - event->efields.cmt.scale = scale; - - memset(&cmt_attr, 0, sizeof(cmt_attr)); - cmt_attr.size = sizeof(cmt_attr); - cmt_attr.type = event_type; - cmt_attr.config = 1; - cmt_attr.inherit = 1; - cmt_attr.disabled = 1; - cmt_attr.enable_on_exec = 0; - - event->fd = syscall(__NR_perf_event_open, &cmt_attr, pid, -1, -1, 0); + event->fd = syscall(__NR_perf_event_open, &rdt_attr, pid, -1, -1, 0); if (event->fd < 0) { virReportSystemError(errno, _("Unable to open perf type=%d for pid=%d"), @@ -161,8 +171,9 @@ virPerfCmtEnable(virPerfEventPtr event, } if (ioctl(event->fd, PERF_EVENT_IOC_ENABLE) < 0) { - virReportSystemError(errno, "%s", - _("Unable to enable perf event for CMT")); + virReportSystemError(errno, + _("Unable to enable perf event for %s"), + virPerfEventTypeToString(event->type)); goto error; } @@ -186,7 +197,9 @@ virPerfEventEnable(virPerfPtr perf, switch (type) { case VIR_PERF_EVENT_CMT: - if (virPerfCmtEnable(event, pid) < 0) + case VIR_PERF_EVENT_MBMT: + case VIR_PERF_EVENT_MBML: + if (virPerfRdtEnable(event, pid) < 0) return -1; break; case VIR_PERF_EVENT_LAST: diff --git a/src/util/virperf.h b/src/util/virperf.h index 8ec8753..769e85a 100644 --- a/src/util/virperf.h +++ b/src/util/virperf.h @@ -26,6 +26,8 @@ typedef enum { VIR_PERF_EVENT_CMT, + VIR_PERF_EVENT_MBMT, + VIR_PERF_EVENT_MBML, VIR_PERF_EVENT_LAST } virPerfEventType; -- 1.9.1

3 2

[libvirt] [PATCH] tests: nodeinfotest: Remove virSaveLastError() usage
by Cole Robinson 19 May '16

19 May '16

It's overkill here, we can use virGetLast* instead --- tests/nodeinfotest.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/tests/nodeinfotest.c b/tests/nodeinfotest.c index d8eace5..03500fb 100644 --- a/tests/nodeinfotest.c +++ b/tests/nodeinfotest.c @@ -44,10 +44,8 @@ linuxTestCompareFiles(char *sysfs_prefix, memset(&nodeinfo, 0, sizeof(nodeinfo)); if (linuxNodeInfoCPUPopulate(sysfs_prefix, cpuinfo, arch, &nodeinfo) < 0) { if (virTestGetDebug()) { - virErrorPtr error = virSaveLastError(); - if (error && error->code != VIR_ERR_OK) - VIR_TEST_DEBUG("\n%s\n", error->message); - virFreeError(error); + if (virGetLastError()) + VIR_TEST_DEBUG("\n%s\n", virGetLastErrorMessage()); } VIR_FORCE_FCLOSE(cpuinfo); goto fail; -- 2.7.4

2 1

[libvirt] [PATCH v3 0/3] More usage of virGetLastErrorMessage()
by Jovanka Gulicoska 19 May '16

19 May '16

Use virGetLastErrorMessage() instead of virGetLastError() Link to task: http://wiki.libvirt.org/page/BiteSizedTasks#More_usage_of_virGetLastErrorMe… Jovanka Gulicoska (3): tests: nodeinfotest: Convert to virGetLastErrorMessage() tests: More usage of virGetLastErrorMessage() More usage of virGetLastErrorMessage daemon/libvirtd.c | 8 +--- examples/object-events/event-test.c | 9 ++--- src/bhyve/bhyve_driver.c | 3 +- src/conf/virsecretobj.c | 5 +-- src/libvirt.c | 3 +- src/libxl/libxl_domain.c | 3 +- src/libxl/libxl_driver.c | 4 +- src/locking/lock_daemon.c | 8 +--- src/logging/log_daemon.c | 8 +--- src/lxc/lxc_container.c | 9 ++--- src/lxc/lxc_controller.c | 9 ++--- src/lxc/lxc_domain.c | 4 +- src/lxc/lxc_process.c | 6 +-- src/rpc/virnettlscontext.c | 3 +- src/storage/storage_driver.c | 16 ++------ src/uml/uml_driver.c | 3 +- src/util/iohelper.c | 10 +---- src/util/virhook.c | 3 +- src/util/virhostdev.c | 20 ++++----- tests/commandtest.c | 81 +++++++++++++------------------------ tests/libvirtdconftest.c | 26 ++++++------ tests/nodeinfotest.c | 8 ++-- tests/openvzutilstest.c | 7 +--- tests/qemucapsprobe.c | 6 +-- tests/securityselinuxtest.c | 6 +-- 25 files changed, 91 insertions(+), 177 deletions(-) -- 2.5.5

2 4

[libvirt] [PATCH 0/6] conf: qemu: Introduce relaxed config validation
by Peter Krempa 19 May '16

19 May '16

Recently I NACKed quite a few patches attempting to add checks to the post parse callback infrastructure that would inhbit configs from being loaded after daemon restart. To solve this introduce a new infrastructure that will get called only on codepaths that define and start the config leaving the code paths that load configs untouched. Peter Krempa (6): conf: disk: Rename virDomainDiskDefValidate to virDomainDiskDefParseValidate qemu: driver: Fix function header alignment of some functions conf: Introduce infrastructure to add config validation to define time qemu: Move check that validates 'min_guarantee' to qemuDomainDefValidate conf: Move check that validates disk info to virDomainDefValidate conf: Add validation infrastructure for device hot/cold plug src/conf/domain_conf.c | 128 +++++++++++++++++++++++++++++++++++++++-------- src/conf/domain_conf.h | 11 ++-- src/libvirt_private.syms | 3 +- src/qemu/qemu_conf.h | 2 + src/qemu/qemu_domain.c | 38 ++++++++++++++ src/qemu/qemu_domain.h | 3 ++ src/qemu/qemu_driver.c | 26 ++++++++-- src/qemu/qemu_process.c | 19 ++++--- 8 files changed, 190 insertions(+), 40 deletions(-) -- 2.8.2

5 15

[libvirt] [PATCH 0/3] More usage of virGetLastErrorMessage()
by Jovanka Gulicoska 19 May '16

19 May '16

v2 of patches Use virGetLastErrorMessage() insted of virGetLastError() Link to task: http://wiki.libvirt.org/page/BiteSizedTasks#More_usage_of_virGetLastErrorMe… Jovanka Gulicoska (3): virGetLastErrorMessage() in tests More usage of virGetLastErrorMessage() virGetLastErrorMessage in nodeinfotest daemon/libvirtd.c | 8 +--- examples/object-events/event-test.c | 9 ++--- src/bhyve/bhyve_driver.c | 3 +- src/conf/virsecretobj.c | 5 +-- src/libvirt.c | 3 +- src/libxl/libxl_domain.c | 3 +- src/libxl/libxl_driver.c | 4 +- src/locking/lock_daemon.c | 8 +--- src/logging/log_daemon.c | 8 +--- src/lxc/lxc_container.c | 9 ++--- src/lxc/lxc_controller.c | 9 ++--- src/lxc/lxc_domain.c | 4 +- src/lxc/lxc_process.c | 6 +-- src/rpc/virnettlscontext.c | 3 +- src/storage/storage_driver.c | 16 ++------ src/uml/uml_driver.c | 3 +- src/util/iohelper.c | 10 +---- src/util/virhook.c | 3 +- src/util/virhostdev.c | 20 ++++----- tests/commandtest.c | 81 +++++++++++++------------------------ tests/libvirtdconftest.c | 26 ++++++------ tests/nodeinfotest.c | 8 ++-- tests/openvzutilstest.c | 7 +--- tests/qemucapsprobe.c | 6 +-- tests/securityselinuxtest.c | 6 +-- 25 files changed, 91 insertions(+), 177 deletions(-) -- 2.5.5

2 7

[libvirt] [PATCH] tests: Add forgotten backslash
by Michal Privoznik 19 May '16

19 May '16

While introducing virtestmock.la, I've forgotten to add '\' at the end of one line leaving our Makefile.am mangled. Fortunately, the only thing that comes after is '$(NULL)' so nothing is terribly broken. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Pushed under trivial rule. tests/Makefile.am | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/Makefile.am b/tests/Makefile.am index 238f6da..c7c9a03 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -447,7 +447,7 @@ endif WITH_DBUS if WITH_LINUX test_libraries += virusbmock.la \ virnetdevbandwidthmock.la \ - virtestmock.la + virtestmock.la \ $(NULL) endif WITH_LINUX -- 2.8.1

1 0

[libvirt] [PATCH 0/4] Remove need for STATIC_ANALYSIS in viralloc.h
by John Ferlan 19 May '16

19 May '16

Not sure which release exactly, but the coverity analysis bug I filed that was the reason for commit id 'c9a85af31' has been fixed, so it's time to remove that... In doing so uncovered another issue - it seems the fix for Coverity addressed the primary problem we've seen, but when there's functions returning allocated strings by reference that can also VIR_FREE(*string), whatever was added to address the main issue doesn't seem to recognize this other usage, resulinting in a false positive resource_leak. Luckily there's only two instances in our code for that. One is addressed by patch 3 and the other in virPCIDeviceGetDriverPathAndName which would require quite a few more changes to address (essentially split up the function - it'll be on my "to do" list). Patch 1: Addresses an issue seen once patches 2-4 were applied - that there's a real problem with the error path. It's a simple fix. Patches 2 & 3: Address a false positive resource leak even with patch 4 applied. Patch 4: Remove the rather ugly !STATIC_ANALYSIS within the VIR_FREE and VIR_DISPOSE* macros. As I found with a build after the fact, VIR_DISPOSE_STRING was missing an argument anyway (it should have been ": 0, 1, NULL" instead of ": 1, NULL"). NB: Patch 4 has been run through the private coverity server... John Ferlan (4): util: Fix error path for virPCIGetVirtualFunctions util: Remove need for ret in virPCIGetPhysicalFunction util: Adjust return for virPCIGetDeviceAddressFromSysfsLink util: Remove need for STATIC_ANALYSIS check src/util/viralloc.h | 34 ++++++---------------------------- src/util/virpci.c | 39 +++++++++++++++++++-------------------- 2 files changed, 25 insertions(+), 48 deletions(-) -- 2.5.5

2 5

[libvirt] [PATCH 0/3] Patches to set up the secret encryption
by John Ferlan 19 May '16

19 May '16

Based on v4 of the AES secret object series: http://www.redhat.com/archives/libvir-list/2016-May/msg01292.html Could have been an RFC too... Precursor for the AES secret object patches - figured I'd make sure this was the "right path" and hopefully to get some help with the mock environment since the virRandomBytes would move out of qemuxml2argvtest. Patch 3 is new and will work independent of 1 & 2, although it would be nice to alther the vircryptotest to use a mocked virRandomBytes rather than seeing the encalg and iv with what the mocked code would return. John Ferlan (3): tests: Add mock for virRandomBytes tests: Need help... Trying to add virRandomBytes mock util: Introduce encryption APIs configure.ac | 1 + src/libvirt_private.syms | 2 + src/util/vircrypto.c | 189 ++++++++++++++++++++++++++++++++++++++++++++++- src/util/vircrypto.h | 20 ++++- tests/Makefile.am | 16 ++++ tests/commandhelper.c | 1 + tests/vircryptotest.c | 86 +++++++++++++++++++++ tests/virrandommock.c | 56 ++++++++++++++ tests/virrandomtest.c | 86 +++++++++++++++++++++ 9 files changed, 455 insertions(+), 2 deletions(-) create mode 100644 tests/virrandommock.c create mode 100644 tests/virrandomtest.c -- 2.5.5

2 9

[libvirt] hotplug support for "filesystem" devices?
by Snyder, Emile 19 May '16

19 May '16

Hello all, I'm a newbie to libvirt, coming to it because I'm trying to understand how to provide the ability to attach/detach network filesystem shares to openstack guest instances in something analogous to the way block devices can be now. While playing with the existing virtfs/9p support for qemu, I found that I can't use virsh to attach a "filesystem" device while the VM is running. For example, I create a snippet of XML like <filesystem type='mount' accessmode='mapped'> <source dir='/tmp/share'/> <target dir='share_tag'/> </filesystem> and then try: stack@devstack-8366:~/devstack$ virsh list Id Name State ---------------------------------------------------- 3 instance-00000002 running 5 instance-00000004 running 8 instance-00000007 running stack@devstack-8366:~/devstack$ virsh attach-device 8 share.xml error: Failed to attach device from share.xml error: Operation not supported: live attach of device 'filesystem' is not supported I see that error coming from src/qemu/qemu_driver.c in the qemuDomainAttachDeviceLive function. Can anyone tell me if this is something that is going to be hard or easy to make possible? Pointers to where I should look in the source, or in qemu? Is anyone working it already? Any advice at all welcome, Thanks -Emile Snyder

2 1

[libvirt] [PATCH 0/3] qemu: Don't get stuck in bulk-stats API on broken NFS
by Peter Krempa 19 May '16

19 May '16

Accessing VM disk files while we weren't able to get the stats from qemu directly as it was most probably stuck on missing storage will make libvrtd's thread handling the stats get stuck. Avoid touching files in such case. Peter Krempa (3): qemu: driver: Remove unnecessary flag in qemuDomainGetStatsBlock qemu: driver: Separate bulk stats worker for block devices qemu: bulk stats: Don't access possibly blocked storage src/qemu/qemu_driver.c | 78 +++++++++++++++++++++++++++++++------------------- 1 file changed, 49 insertions(+), 29 deletions(-) -- 2.8.2

2 6