January 2017 - Devel - libvirt List Archives

[libvirt] [PATCH] util: disallow resetting a pci-bridge device
by Shivaprasad G Bhat 18 Jan '17

18 Jan '17

It is distructive to attempt a reset on pci-bridge, the host can crash. The bridges won't contain any guest data and neither they can be passed through using vfio/stub. So, no point in allowing a reset for them. Signed-off-by: Shivaprasad G Bhat <sbhat(a)linux.vnet.ibm.com> --- src/util/virpci.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/src/util/virpci.c b/src/util/virpci.c index 0601f49..860f7aa 100644 --- a/src/util/virpci.c +++ b/src/util/virpci.c @@ -933,6 +933,16 @@ virPCIDeviceReset(virPCIDevicePtr dev, char *drvName = NULL; int ret = -1; int fd = -1; + int hdrType = -1; + + if (virPCIGetHeaderType(dev, &hdrType) < 0) + return -1; + + if (hdrType == VIR_PCI_HEADER_PCI_BRIDGE) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Resetting a pci-bridge device is not allowed")); + return -1; + } if (activeDevs && virPCIDeviceListFind(activeDevs, dev)) { virReportError(VIR_ERR_INTERNAL_ERROR,

3 2

[libvirt] [PATCH] Fix the incorrect memory freeing which will result in crash
by Wu Zongyong 18 Jan '17

18 Jan '17

The number of elements in new_params is equal to the length of info, instead of nparams, so it's wrong to free new_params using nparams. Signed-off-by: Wu Zongyong <wuzongyo(a)mail.ustc.edu.cn> --- libvirt-override.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/libvirt-override.c b/libvirt-override.c index db14244..dac4250 100644 --- a/libvirt-override.c +++ b/libvirt-override.c @@ -632,7 +632,7 @@ libvirt_virDomainSetSchedulerParametersFlags(PyObject *self ATTRIBUTE_UNUSED, int nparams = 0; Py_ssize_t size = 0; unsigned int flags; - virTypedParameterPtr params, new_params = NULL; + virTypedParameterPtr params = NULL, new_params = NULL; if (!PyArg_ParseTuple(args, (char *)"OOI:virDomainSetScedulerParametersFlags", @@ -694,7 +694,7 @@ libvirt_virDomainSetSchedulerParametersFlags(PyObject *self ATTRIBUTE_UNUSED, cleanup: virTypedParamsFree(params, nparams); - virTypedParamsFree(new_params, nparams); + virTypedParamsFree(new_params, size); return ret; } @@ -7682,7 +7682,7 @@ libvirt_virNodeSetMemoryParameters(PyObject *self ATTRIBUTE_UNUSED, int nparams = 0; Py_ssize_t size = 0; unsigned int flags; - virTypedParameterPtr params, new_params = NULL; + virTypedParameterPtr params = NULL, new_params = NULL; if (!PyArg_ParseTuple(args, (char *)"OOI:virNodeSetMemoryParameters", @@ -7741,7 +7741,7 @@ libvirt_virNodeSetMemoryParameters(PyObject *self ATTRIBUTE_UNUSED, cleanup: virTypedParamsFree(params, nparams); - virTypedParamsFree(new_params, nparams); + virTypedParamsFree(new_params, size); return ret; } -- 2.10.0.windows.1

2 1

[libvirt] Entering freeze for libvirt-3.0.0
by Daniel Veillard 18 Jan '17

18 Jan '17

As planned, I tagged the RC1 in git and pushed signed tarball and rpms to the usual location: ftp://libvirt.org/libvirt/ This seems to work fine in my limited testing but as usual we need community feedback on it and especially testing on different systems and architectures, so please give it a try ! If everything works fine then I will push rc2 on Friday and hopefully 3.0.0 final on Monday, so please give it a try to make sure that 0.0 release is nonetheless a good one :-) thanks ! Daniel -- Daniel Veillard | Open Source and Standards, Red Hat veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

7 16

[libvirt] TSC frequency configuration & invtsc migration (was Re: [PATCH 4/4] kvm: Allow migration with invtsc)
by Eduardo Habkost 18 Jan '17

18 Jan '17

On Wed, Jan 04, 2017 at 11:39:16AM -0200, Eduardo Habkost wrote: > On Wed, Jan 04, 2017 at 09:56:56AM -0200, Marcelo Tosatti wrote: > > On Tue, Dec 27, 2016 at 05:21:20PM -0200, Eduardo Habkost wrote: > > > Instead of blocking migration on the source when invtsc is > > > enabled, rely on the migration destination to ensure there's no > > > TSC frequency mismatch. > > > > > > We can't allow migration unconditionally because we don't know if > > > the destination is a QEMU version that is really going to ensure > > > there's no TSC frequency mismatch. To ensure we are migrating to > > > a destination that won't ignore SET_TSC_KHZ errors, allow invtsc > > > migration only on pc-*-2.9 and newer. > > > > > > Signed-off-by: Eduardo Habkost <ehabkost(a)redhat.com> [...] > > > @@ -2655,12 +2656,14 @@ int kvm_arch_put_registers(CPUState *cpu, int level) > > > } > > > > > > if (level == KVM_PUT_FULL_STATE) { > > > - /* We don't check for kvm_arch_set_tsc_khz() errors here, > > > - * because TSC frequency mismatch shouldn't abort migration, > > > - * unless the user explicitly asked for a more strict TSC > > > - * setting (e.g. using an explicit "tsc-freq" option). > > > + /* Migration TSC frequency mismatch is fatal only if we are > > > + * actually reporting Invariant TSC to the guest. > > > */ > > > - kvm_arch_set_tsc_khz(cpu); > > > + ret = kvm_arch_set_tsc_khz(cpu); > > > + if ((x86_cpu->env.features[FEAT_8000_0007_EDX] & CPUID_APM_INVTSC) && > > > + ret < 0) { > > > + return ret; > > > + } > > > } > > > > Will the guest continue in the source in this case? > > > > I think this is past the point where migration has been declared > > successful. > > > > Otherwise looks good. > > Good point. I will make additional tests and see if there's some > other place where the kvm_arch_set_tsc_khz() call can be moved > to. So, if we solve this and do something on (for example) post_load, we still have a problem: device state is migrated after RAM. This means QEMU will check for TSC scaling and abort migration very late. We could solve that by manually registering a SaveVMHandler that will send the TSC frequency on save_live_setup, so migration is aborted earlier. But: this sounds like just a complex hack to work around the real problems: 1) TSC frequency is guest-visible, and anything that affects guest ABI should depend on the VM configuration, not on host capabilities; 2) Setting TSC frequency depending on the host will make migratability unpredictable for management software: the same VM config could be migratable to host A when started on host B, but not migratable to host A when started on host C. I suggest we allow migration with invtsc if and only if tsc-frequency is set explicitly by management software. In other words, apply only patches 1/4 and 2/4 from this series. After that, we will need libvirt support for configuring tsc-frequency. -- Eduardo

5 21

[libvirt] [PATCH v3 0/2] Allow migration with invtsc if TSC frequency is explicitly set
by Eduardo Habkost 18 Jan '17

18 Jan '17

This series makes QEMU accept migration if tsc-frequency is explicitly set on configuration. As management software is required to keep device configuration the same on migration source or destination, explicit tsc-frequency will ensure that either: * The destination host has a matching TSC frequency; or * The destination host has TSC scaling available. Changelog ========= v2 -> v3: * Fix build failure ((missing closing braces) v1 -> v2: * v1 series subject was: * [PATCH 0/4] Allow migration with invtsc if there's no frequency mismatch * Removed patches 3/4 and 4/4, that allowed migration if no explicit tsc-frequency was set. Implementing the check on post_load or post_init is not enough to make migration abort, so we will need a more complex solution to implement that feature. Plans for future work ===================== 1) Querying host TSC frequency/scaling capability ------------------------------------------------- I plan to include TSC frequency/scaling information on query-cpu-model-expansion model="host" in a future series. Then management software will be able to automatically configure TSC frequency when invtsc is enabled, instead of requiring the user to configure it explicitly. While we don't implement that, invtsc migration will be possible only if the user configures TSC frequency manually. 2) invtsc migration with no explicit TSC frequency -------------------------------------------------- A future series can implement migration when TSC frequency is not specified explicitly. It will be a bit more complex because it requires either letting the destination abort the migration, or sending TSC frequency/scaling information from destination to source. --- Cc: Marcelo Tosatti <mtosatti(a)redhat.com> Cc: "Daniel P. Berrange" <berrange(a)redhat.com> Cc: Paolo Bonzini <pbonzini(a)redhat.com> Cc: kvm(a)vger.kernel.org Cc: Haozhong Zhang <haozhong.zhang(a)intel.com> Cc: "Michael S. Tsirkin" <mst(a)redhat.com> Cc: Igor Mammedov <imammedo(a)redhat.com> Cc: libvir-list(a)redhat.com Cc: Jiri Denemark <jdenemar(a)redhat.com> Eduardo Habkost (2): kvm: Simplify invtsc check kvm: Allow invtsc migration if tsc-khz is set explicitly target/i386/kvm.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) -- 2.11.0.259.g40922b1

2 1

[libvirt] [PATCH] Post-release version bump to 3.1.0
by Michal Privoznik 18 Jan '17

18 Jan '17

Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Pushed under trivial rule. configure.ac | 2 +- docs/news.xml | 11 +++++++++++ 2 files changed, 12 insertions(+), 1 deletion(-) diff --git a/configure.ac b/configure.ac index 9e41f8983..a217fc165 100644 --- a/configure.ac +++ b/configure.ac @@ -16,7 +16,7 @@ dnl You should have received a copy of the GNU Lesser General Public dnl License along with this library. If not, see dnl <http://www.gnu.org/licenses/>. -AC_INIT([libvirt], [3.0.0], [libvir-list(a)redhat.com] [], [http://libvirt.org] +AC_INIT([libvirt], [3.1.0], [libvir-list(a)redhat.com] [], [http://libvirt.org] AC_CONFIG_SRCDIR([src/libvirt.c]) AC_CONFIG_AUX_DIR([build-aux]) AC_CONFIG_HEADERS([config.h]) diff --git a/docs/news.xml b/docs/news.xml index b912d8a62..4ff0cb638 100644 --- a/docs/news.xml +++ b/docs/news.xml @@ -10,6 +10,17 @@ and optionally a description where it's explained in more detail --> <libvirt> + <release version="v3.1.0" date="unreleased"> + <section title="New features"> + <change/> + </section> + <section title="Improvements"> + <change/> + </section> + <section title="Bug fixes"> + <change/> + </section> + </release> <release version="v3.0.0" date="2017-01-17"> <section title="New features"> <change> -- 2.11.0

1 0

[libvirt] [PATCH] tests: Distribute qemuhotplugtestcpus
by Michal Privoznik 18 Jan '17

18 Jan '17

Starting from a245abce436f4f333 another set of tests for qemuhotplugtest has been introduced. This time for vcpu hotplug. However, the test data (which live in qemuhotplugtestcpus dir) are not being distributed properly. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Pushed under build breaker rule. tests/Makefile.am | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/Makefile.am b/tests/Makefile.am index c7d474817..87197841b 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -121,6 +121,7 @@ EXTRA_DIST = \ qemucapabilitiesdata \ qemucaps2xmldata \ qemuhelpdata \ + qemuhotplugtestcpus \ qemuhotplugtestdevices \ qemuhotplugtestdomains \ qemumonitorjsondata \ -- 2.11.0

1 0

[libvirt] [PATCH 0/4] gluster: cleanup and fix code for pool discovery
by Peter Krempa 18 Jan '17

18 Jan '17

Peter Krempa (4): storage: Fix error reporting when looking up storage pool sources storage: gluster: Report error if no volumes were found in pool lookup storage: gluster: Remove build-time dependency on the 'gluster' cli tool spec: Depend on the gluster command line tool configure.ac | 2 +- libvirt.spec.in | 4 +++ src/storage/storage_backend.c | 48 ++++++++++++++++++++--------------- src/storage/storage_backend.h | 3 ++- src/storage/storage_backend_fs.c | 22 ++++++++++------ src/storage/storage_backend_gluster.c | 14 +++++++--- 6 files changed, 60 insertions(+), 33 deletions(-) -- 2.11.0

2 9

[libvirt] [PATCH] news: Fix xml dump of autogenerated websocket
by Nikolay Shirokovskiy 18 Jan '17

18 Jan '17

--- Sorry for delay and thus not getting this news bit into release notes in time. docs/news.xml | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/docs/news.xml b/docs/news.xml index 95b021a..3ff8f4e 100644 --- a/docs/news.xml +++ b/docs/news.xml @@ -274,6 +274,15 @@ for x86_64 HVM domains. </description> </change> + <change> + <summary> + qemu: Fix xml dump of autogenerated websocket + </summary> + <description> + As a result autogenerated websocket port is regenerated on domain restore, + migration and such as it should be. + </description> + </change> </section> </release> <release version="v2.5.0" date="2016-12-04"> -- 1.8.3.1

2 1

[libvirt] Release of libvirt-3.0.0
by Daniel Veillard 18 Jan '17

18 Jan '17

So I got mixed reports in the last day about the state of the head but one of the big issues seems solved, and I'm not sure keeping the freeze much longer will help, so libvirt-3.0.0 is out. It's tagged in git, signed tarball and rpms are available at thet usual place: ftp://libvirt.org/libvirt/ I also pushed python bindings release too which one can find at ftp://libvirt.org/libvirt/python/ It includes a fair amount of changes despite the slowdown around the end of year break: * New features - Domain events for metadata content changes The domain events framework has a new event ID that can be used to get notifications when domain metadata content changes. - Event notifications for the secret object The secret object now supports event notifications, covering lifcycle changes and secret value changes. - New localPtr attribute for "ip" element in network XML - qemu: Support QEMU group I/O throttling Add the capability to allow group I/O throttling via a new domain <disk> <iotune> subelement "group_name" to allow sharing I/O throttling quota between multiple drives. - nss: Introduce libvirt_guest New libvirt_guest nss module that translates libvirt guest names into IP addresses. - daemon: Add support for runtime logging settings adjustment Logging-related settings like log outputs and filters can now be adjusted during runtime using the admin interface without the necessity of the daemon's restart. - storage: Add virStorageVolInfoFlags API Add the API to support using the VIR_STORAGE_VOL_GET_PHYSICAL flag in order to return the host physical size in bytes of the image container in the allocation field of the _virStorageVolInfo structure. The --physical flag has been added to the virsh vol-info command to access the data. - libxl: Implement virDomainGetMaxVcpus API - storage: Add overwrite flag checking for logical pool Add support for the OVERWRITE flags for the logical storage backend including checking for existing data on the target volumes when building a new logical pool on target volume(s). - qemu: Add support for guest CPU configuration on s390(x) * Improvements - perf: Add more perf statistics Add support to get the count of branch instructions executed, branch misses, bus cycles, stalled frontend cpu cycles, stalled backend cpu cycles, and ref cpu cycles by applications running on the platform. - conf: Display <physical> for volume xml Add a display of the <physical> size of a disk volume in the output of the volume XML. - qemu: Use virtio-pci by default for aarch64 mach-virt guests virtio-pci provides several advantages over virtio-mmio, such as the ability to hotplug devices and improved performance. While opting in to virtio-pci has been possible for a while, newly-defined guests will now use it automatically. - vbox: remove support for VirtualBox 3.x and older Those old VirtualBox versions have been unsupported by upstream for a long time and the API of 4.0 and newer has diverged enough to require code abstractions to handle differences. Removing support for those old versions drops lots of code from the driver and simplifies the logic to ease implementation of new features going forward. - virsh: pool-info: introduce option --bytes Add option --bytes to virsh pool-info in order ti allow display of units in bytes rather than default of human readable output. - scsi: Add parent wwnn/wwpn or fabric capability for createVport Improve the algorithm searching for the parent scsi_host device for vHBA/NPIV scsi_host creation. Rather than supplying the "parent" by name, it's now possible to define the parent by it's wwnn/wwpn or fabric_wwn in the node device create XML or the storage pool XML. - qemu: aggregate pcie-root-ports onto multiple functions of a slot When pcie-root-ports are added to pcie-root in order to provide a place to connect PCI Express endpoint devices, libvirt now aggregates multiple root ports together onto the same slot (up to 8 per slot) in order to conserve slots. Using this method, it's possible to connect more than 200 endpoint devices to a guest that uses PCIe without requiring setup of any PCIe switches. * Bug fixes - lxc: fix accidental killing of containers during libvirtd restart The libvirt_lxc process was previously not moved into the container scope. As a result, if systemd reloads its config after a container is started, when libvirtd is later restarted it will accidentally kill the containers. - qemu: Correct GetBlockInfo values For an active domain, correct the physical value provided for a raw sparse file backed storage and the allocation value provided for a qcow2 file backed storage that hasn't yet been opened on the domain. - qemu: Make virtio console usable on ppc64 guests The chardev detection code has been improved and can now handle this configuration properly. - qemu: Enable mount namespace To avoid funny races with udev relabelling devices under our hands and to enhance security, libvirt now spawns each qemu process with its own /dev. - storage: Fix implementation of no-overwrite for file system backend Fix file system storage backend implementation of the OVERWRITE flags to be consistent between code and documentation. Add checks to ensure that when building a new file system on a target volume that there is not something already on the disk in a format that libvirt can recognize. - qemu: Create hugepage path on per domain basis Historically, all hugepage enabled domains shared the same path under hugetlbfs. This left libvirt unable to correctly set security labels on it. With this release, however, each domain is put into a separate path which is also correctly labeled. - conf: Reject domains with duplicate drive addresses Reject duplicate drive addresses for disks and hostdevs at domain definition. - libxl: reverse defaults on HVM net device attach Fixes network interface attach for HVM domains when no model is specified. Emulated hotplug isn't yet supported and hence we should default to the general working scenario. - libxl: always enable pae for x86_64 HVM By default pae is disabled in libxl. Without an explicit <pae/> setting in the domain <features> configuration, an x86_64 HVM domain would be get an i686 environment. pae should always be enabled for x86_64 HVM domains. Thanks everybody for your contributions to this release, hopefully that won't be a brown paper bag one ! Daniel -- Daniel Veillard | Red Hat Developers Tools http://developer.redhat.com/ veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

3 2