March 2017 - Devel - libvirt List Archives

[libvirt] [PATCH 00/12] qemu: Enforce guest CPU specification
by Jiri Denemark 17 Mar '17

17 Mar '17

When starting a domain with custom guest CPU specification QEMU may add or remove some CPU features. There are several reasons for this, e.g., QEMU/KVM does not support some requested features or the definition of the requested CPU model in libvirt's cpu_map.xml differs from the one QEMU is using. We can't really avoid this because CPU models are allowed to change with machine types and libvirt doesn't know (and probably doesn't even want to know) about such changes. Thus when we want to make sure guest ABI doesn't change when a domain gets migrated to another host, we need to update our live CPU definition according to the CPU QEMU created. Once updated, we will change CPU checking to VIR_CPU_CHECK_FULL to make sure the virtual CPU created after migration exactly matches the one on the source. https://bugzilla.redhat.com/show_bug.cgi?id=822148 https://bugzilla.redhat.com/show_bug.cgi?id=824989 Jiri Denemark (12): tests: Switch to sparse initialization of virCPUDef docs: Clarify /domain/cpu/@match description Introduce /domain/cpu/@check XML attribute qemu: Set default values for CPU check attribute qemu: Refactor Hyper-V features check qemu: Refactor KVM features check qemu: Refactor CPU features check qemu: Refactor qemuProcessVerifyGuestCPU qemu: Use ARCH_IS_X86 in qemuMonitorJSONGetGuestCPU qemu: Ask QEMU for filtered CPU features qemu: Update CPU definition according to QEMU qemu: Enforce guest CPU specification docs/formatdomain.html.in | 52 ++++- docs/schemas/cputypes.rng | 10 + docs/schemas/domaincommon.rng | 3 + src/conf/cpu_conf.c | 30 +++ src/conf/cpu_conf.h | 12 ++ src/conf/domain_conf.c | 21 ++ src/cpu/cpu.c | 45 +++++ src/cpu/cpu.h | 12 ++ src/cpu/cpu_x86.c | 105 ++++++++++ src/libvirt_private.syms | 1 + src/qemu/qemu_domain.c | 42 ++++ src/qemu/qemu_monitor.c | 11 +- src/qemu/qemu_monitor.h | 3 +- src/qemu/qemu_monitor_json.c | 38 +++- src/qemu/qemu_monitor_json.h | 3 +- src/qemu/qemu_process.c | 225 +++++++++++++-------- tests/domaincapstest.c | 38 ++-- tests/qemumonitorjsontest.c | 4 +- .../qemuxml2argv-aarch64-gic-host.xml | 2 +- .../qemuxml2argv-aarch64-gic-v2.xml | 2 +- .../qemuxml2argv-aarch64-gic-v3.xml | 2 +- .../qemuxml2argv-cpu-check-default-none.args | 21 ++ .../qemuxml2argv-cpu-check-default-none.xml | 19 ++ .../qemuxml2argv-cpu-check-default-none2.args | 21 ++ .../qemuxml2argv-cpu-check-default-none2.xml | 21 ++ .../qemuxml2argv-cpu-check-default-partial.args | 22 ++ .../qemuxml2argv-cpu-check-default-partial.xml | 19 ++ .../qemuxml2argv-cpu-check-default-partial2.args | 21 ++ .../qemuxml2argv-cpu-check-default-partial2.xml | 21 ++ .../qemuxml2argv-cpu-check-full.args | 1 + .../qemuxml2argv-cpu-check-full.xml | 21 ++ .../qemuxml2argv-cpu-check-none.args | 21 ++ .../qemuxml2argv-cpu-check-none.xml | 21 ++ .../qemuxml2argv-cpu-check-partial.args | 1 + .../qemuxml2argv-cpu-check-partial.xml | 21 ++ tests/qemuxml2argvtest.c | 8 + .../qemuxml2xmlout-aarch64-aavmf-virtio-mmio.xml | 2 +- ...qemuxml2xmlout-aarch64-video-virtio-gpu-pci.xml | 2 +- .../qemuxml2xmlout-aarch64-virtio-pci-default.xml | 2 +- ...2xmlout-aarch64-virtio-pci-manual-addresses.xml | 2 +- .../qemuxml2xmlout-cpu-check-default-none.xml | 28 +++ .../qemuxml2xmlout-cpu-check-default-none2.xml | 30 +++ .../qemuxml2xmlout-cpu-check-default-partial.xml | 30 +++ .../qemuxml2xmlout-cpu-check-default-partial2.xml | 30 +++ .../qemuxml2xmlout-cpu-check-full.xml | 30 +++ .../qemuxml2xmlout-cpu-check-none.xml | 30 +++ .../qemuxml2xmlout-cpu-check-partial.xml | 30 +++ .../qemuxml2xmlout-cpu-eoi-disabled.xml | 2 +- .../qemuxml2xmlout-cpu-eoi-enabled.xml | 2 +- .../qemuxml2xmlout-cpu-host-kvmclock.xml | 2 +- .../qemuxml2xmlout-cpu-host-model-features.xml | 2 +- ...emuxml2xmlout-cpu-host-passthrough-features.xml | 2 +- .../qemuxml2xmlout-cpu-kvmclock.xml | 2 +- .../qemuxml2xmlout-graphics-spice-timeout.xml | 2 +- tests/qemuxml2xmltest.c | 8 + tests/testutilsqemu.c | 48 ++--- 56 files changed, 1039 insertions(+), 167 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-none.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-none.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-none2.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-none2.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-partial.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-partial.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-partial2.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-default-partial2.xml create mode 120000 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-full.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-full.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-none.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-none.xml create mode 120000 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-partial.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-check-partial.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-cpu-check-default-none.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-cpu-check-default-none2.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-cpu-check-default-partial.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-cpu-check-default-partial2.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-cpu-check-full.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-cpu-check-none.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-cpu-check-partial.xml -- 2.12.0

2 16

[libvirt] [PATCH 0/4] qemu: fix handling of 'group_name' disk io tuning parameter
by Peter Krempa 17 Mar '17

17 Mar '17

Fix a crasher and an invalid configuration. Peter Krempa (4): qemu: Don't steal pointers from 'persistentDef' in qemuDomainGetBlockIoTune qemu: command: Extract blkdeviotune checks into a separate function qemu: command: Extract tests for subsets of blkdeviotune settings qemu: command: Don't allow setting 'group_name' alone src/qemu/qemu_command.c | 181 ++++++++++++++++++++++++++++++------------------ src/qemu/qemu_driver.c | 22 ++++-- 2 files changed, 127 insertions(+), 76 deletions(-) -- 2.12.0

2 5

[libvirt] Failure to probe cpu definitions for e.g. qemu-system-mips
by Guido Günther 17 Mar '17

17 Mar '17

Hi, while looking into a regression failing to start any mips qemu systems (http://bugs.debian.org/854125) I noticed that querying cpu definition does not work for lots of non intel architectures like mips due to lack of support for the query-cpu-definition monitor command: 2017-03-15 04:23:55.336+0000: 4238: warning : virQEMUCapsLogProbeFailure:4838 : Failed to probe capabilities for /usr/bin/qemu-system-mips: internal error: unable to execute QEMU command 'query-cpu-definitions': The command query-cpu-definitions has not been found 2017-03-15 04:23:55.374+0000: 4238: error : qemuMonitorJSONCheckError:387 : internal error: unable to execute QEMU command 'query-cpu-definitions': The command query-cpu-definitions has not been found Is support for query-cpu-definition a hard requirement or should we fail gracefully in this case? If not I can look into restoring that behaviour but would be happy about pointers what would be the best way forward. Cheers, -- Guido

3 6

[libvirt] [PATCH] perf: remote: Compare perf nparams against the correct constant
by Nitesh Konkar 17 Mar '17

17 Mar '17

Currently 'virsh perf domain' errors out as the perf nparams is incorrectly compared against REMOTE_DOMAIN_MEMORY_PARAMETERS_MAX instead of REMOTE_DOMAIN_PERF_EVENTS_MAX. Signed-off-by: Nitesh Konkar <nitkon12(a)linux.vnet.ibm.com> --- daemon/remote.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/daemon/remote.c b/daemon/remote.c index f2b9b9a..1c9708c 100644 --- a/daemon/remote.c +++ b/daemon/remote.c @@ -3105,7 +3105,7 @@ remoteDispatchDomainGetPerfEvents(virNetServerPtr server ATTRIBUTE_UNUSED, if (virDomainGetPerfEvents(dom, &params, &nparams, args->flags) < 0) goto cleanup; - if (nparams > REMOTE_DOMAIN_MEMORY_PARAMETERS_MAX) { + if (nparams > REMOTE_DOMAIN_PERF_EVENTS_MAX) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("nparams too large")); goto cleanup; } -- 1.9.3

2 1

[libvirt] [PATCH] vz: Remove unused variable
by Mikhail Feoktistov 17 Mar '17

17 Mar '17

emulatedType is not used in prlsdkAddDomainHardDisksInfo() --- src/vz/vz_sdk.c | 5 ----- 1 file changed, 5 deletions(-) diff --git a/src/vz/vz_sdk.c b/src/vz/vz_sdk.c index 3fd17db..8e6e89d 100644 --- a/src/vz/vz_sdk.c +++ b/src/vz/vz_sdk.c @@ -817,14 +817,9 @@ prlsdkAddDomainHardDisksInfo(vzDriverPtr driver, PRL_HANDLE sdkdom, virDomainDef for (i = 0; i < hddCount; ++i) { - PRL_UINT32 emulatedType; - pret = PrlVmCfg_GetHardDisk(sdkdom, i, &hdd); prlsdkCheckRetGoto(pret, error); - pret = PrlVmDev_GetEmulatedType(hdd, &emulatedType); - prlsdkCheckRetGoto(pret, error); - if (IS_CT(def) && prlsdkInBootList(sdkdom, hdd)) { -- 1.8.3.1

1 0

[libvirt] [PATCH] apparmor: allow /usr/lib/qemu/qemu-bridge-helper
by Guido Günther 17 Mar '17

17 Mar '17

This is where e.g. Debian puts it. --- examples/apparmor/usr.sbin.libvirtd | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/apparmor/usr.sbin.libvirtd b/examples/apparmor/usr.sbin.libvirtd index 8893e75fe..03a80b7e6 100644 --- a/examples/apparmor/usr.sbin.libvirtd +++ b/examples/apparmor/usr.sbin.libvirtd @@ -67,7 +67,7 @@ # allow changing to our UUID-based named profiles change_profile -> @{LIBVIRT}-[0-9a-f]*-[0-9a-f]*-[0-9a-f]*-[0-9a-f]*-[0-9a-f]*, - /usr/{lib,libexec}/qemu-bridge-helper Cx -> qemu_bridge_helper, + /usr/{lib,lib/qemu,libexec}/qemu-bridge-helper Cx -> qemu_bridge_helper, # child profile for bridge helper process profile qemu_bridge_helper { #include <abstractions/base> -- 2.11.0

2 4

[libvirt] [PATCH 1/1] perf: list all supported events when get domain stats
by Qiaowei Ren 16 Mar '17

16 Mar '17

Currently when virDomainListGetStats is called, the stats for those disabled perf events won't be showed in result. This will produce some problems for applications based on libvirt sometime, e.g. the OpenStack bug https://bugs.launchpad.net/ceilometer/+bug/1670948 This patch just show '0' in result for disabled events for virDomainListGetStats API. I guess this should be also rational to show all stats even though the events are not enabled. Signed-off-by: Qiaowei Ren <qiaowei.ren(a)intel.com> --- src/qemu/qemu_driver.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 2032fac..237bf57 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -19355,8 +19355,10 @@ qemuDomainGetStatsPerfOneEvent(virPerfPtr perf, char param_name[VIR_TYPED_PARAM_FIELD_LENGTH]; uint64_t value = 0; - if (virPerfReadEvent(perf, type, &value) < 0) - return -1; + if (virPerfEventIsEnabled(perf, type)) { + if (virPerfReadEvent(perf, type, &value) < 0) + return -1; + } snprintf(param_name, VIR_TYPED_PARAM_FIELD_LENGTH, "perf.%s", virPerfEventTypeToString(type)); @@ -19383,9 +19385,6 @@ qemuDomainGetStatsPerf(virQEMUDriverPtr driver ATTRIBUTE_UNUSED, int ret = -1; for (i = 0; i < VIR_PERF_EVENT_LAST; i++) { - if (!virPerfEventIsEnabled(priv->perf, i)) - continue; - if (qemuDomainGetStatsPerfOneEvent(priv->perf, i, record, maxparams) < 0) goto cleanup; -- 1.9.1

2 1

[libvirt] libvirtaio: libvirt-python asyncio event impl
by Wojtek Porczyk 16 Mar '17

16 Mar '17

Hello, libvirt-list, I wrote an event implementation wrapping an asyncio event loop [1]. I would like to contribute it back to libvirt-python, to be offered alongside the Default Impl in C. [1] https://github.com/woju/libvirtaio Also it contains a workaround for a memory leak around the custom loop implementation: https://github.com/woju/libvirtaio/blob/8685cfc/libvirtaio.py#L66-L92 How should I submit this? As a patch against libvirt-python.git, appending to libvirt-override.py? -- pozdrawiam / best regards _.-._ Wojtek Porczyk .-^' '^-. Invisible Things Lab |'-.-^-.-'| | | | | I do not fear computers, | '-.-' | I fear lack of them. '-._ : ,-' -- Isaac Asimov `^-^-_>

2 3

[libvirt] limit cpu speed in domain xml
by Vasiliy Tolstov 16 Mar '17

16 Mar '17

Hi. Does it possible to limit vm cpu speed for example to 1Ghz or 500Mhz ? I need to simulate some hardware with specific cpu speed and test my application inside this vm , i want to measure results from each test and need constant cpu speed for testing (i need to run test not only my notebook but on server too and want identical cpu speed inside vm) -- Vasiliy Tolstov, e-mail: v.tolstov(a)selfip.ru

3 4

[libvirt] libvirt-python bug: custom event loop impl calls ff callback from *Remove(Handle|Timeout)Func
by Wojtek Porczyk 16 Mar '17

16 Mar '17

Hello libvirt-list, As of current libvirt-python.git, according to libvirt-override.c, if implementing custom event loop in Python, ff callback is called from libvirt_virEventRemoveHandleFunc, which is a C glue between virEventRegisterImpl and actual removeHandle function written in Python: > result = PyEval_CallObject(removeHandleObj, pyobj_args); > if (!result) { > PyErr_Print(); > PyErr_Clear(); > } else if (!PyTuple_Check(result) || PyTuple_Size(result) != 3) { > DEBUG("%s: %s must return opaque obj registered with %s" > "to avoid leaking libvirt memory\n", > __FUNCTION__, NAME(removeHandle), NAME(addHandle)); > } else { > opaque = PyTuple_GetItem(result, 1); > ff = PyTuple_GetItem(result, 2); > cff = PyvirFreeCallback_Get(ff); > if (cff) > (*cff)(PyvirVoidPtr_Get(opaque)); > retval = 0; > } This is exactly what one shoud not be doing according to documentation [1]: > If the opaque user data requires free'ing when the handle is unregistered, > then a 2nd callback can be supplied for this purpose. This callback needs to > be invoked from a clean stack. If 'ff' callbacks are invoked directly from the > virEventRemoveHandleFunc they will likely deadlock in libvirt. [1] https://libvirt.org/html/libvirt-libvirt-event.html#virEventAddHandleFunc This is true, the deadlock occurs. When the "result" tuple is mangled to have None as third item ("ff"), then cff = PyvirFreeCallback_Get(ff) is NULL and the deadlock does not happen. That's also why script examples/event-test.py does not deadlock, because it does not return anything (that is, returns None) from Python, so the second if block happens and the ff callback, if any, is not executed (and probably something leaks, but I didn't check for that). Everything also applies to to timeouts (libvirt_virEventRemoteTimeoutFunc). -- pozdrawiam / best regards _.-._ Wojtek Porczyk .-^' '^-. Invisible Things Lab |'-.-^-.-'| | | | | I do not fear computers, | '-.-' | I fear lack of them. '-._ : ,-' -- Isaac Asimov `^-^-_>

3 4