January 2015 - Devel - libvirt List Archives

[libvirt] [PATCH V4 0/3] Parser for xen-xl config format
by Jim Fehlig 14 Jan '15

14 Jan '15

It's been a long, twisting road to V4 of the Xen xl parser. V3 [1] was based on a flex-based parser that was copied from the Xen project and proved to be a bit challenging to integrate properly with autotools. But as it turns out, Xen provides an interface to the parser via libxlutil. I hadn't realized this interface was available for external consumption since the corresponding header file was never installed. Patch sent to xen-devel [2] to install the header, but in the meantime need to declare gthe imported libxlutil functions as extern (see patch 1). V4 uses libxlutil, which has simplified the series quite a bit and eliminates the potential of the copied flex parser diverging from the canonical source in xen.git. [1] https://www.redhat.com/archives/libvir-list/2014-December/msg00765.html [2] http://lists.xen.org/archives/html/xen-devel/2015-01/msg00690.html Jim Fehlig (1): Introduce support for parsing/formatting Xen xl config format Kiarie Kahurani (2): tests: Tests for the xen-xl parser libxl: Add support for parsing/formating Xen XL config configure.ac | 3 + po/POTFILES.in | 1 + src/Makefile.am | 11 + src/libvirt_xenxlconfig.syms | 12 + src/libxl/libxl_driver.c | 32 ++- src/xenconfig/xen_common.c | 3 +- src/xenconfig/xen_xl.c | 515 +++++++++++++++++++++++++++++++++++ src/xenconfig/xen_xl.h | 35 +++ tests/Makefile.am | 11 + tests/testutilsxen.c | 50 ++++ tests/testutilsxen.h | 9 +- tests/xlconfigdata/test-new-disk.cfg | 26 ++ tests/xlconfigdata/test-new-disk.xml | 51 ++++ tests/xlconfigdata/test-spice.cfg | 32 +++ tests/xlconfigdata/test-spice.xml | 45 +++ tests/xlconfigtest.c | 225 +++++++++++++++ tools/virsh.pod | 8 +- 17 files changed, 1055 insertions(+), 14 deletions(-) create mode 100644 src/libvirt_xenxlconfig.syms create mode 100644 src/xenconfig/xen_xl.c create mode 100644 src/xenconfig/xen_xl.h create mode 100644 tests/xlconfigdata/test-new-disk.cfg create mode 100644 tests/xlconfigdata/test-new-disk.xml create mode 100644 tests/xlconfigdata/test-spice.cfg create mode 100644 tests/xlconfigdata/test-spice.xml create mode 100644 tests/xlconfigtest.c -- 1.8.4.5

3 12

[libvirt] question about the flow of live migration with non-shared storage in libvirt
by Zhang Haoyu 14 Jan '15

14 Jan '15

Hi, Someone can help detail the the flow of live migration with non-shared storage in libvirt? Especially, the flow of migration of non-shared storage part? Thanks, Zhang Haoyu

2 1

[libvirt] [PATCHv2 00/14] Fix vmdef usage after ExitMonitor
by Ján Tomko 14 Jan '15

14 Jan '15

Patches 3-8 fix possible crash/invalid memory access if QEMU crashes while we're in the monitor. Patches 9-13 change all other callers of qemuDomainObjExitMonitor to exit early in that case, but should not fix any real issues. They are necessary to turn on ATTRIBUTE_RETURN_CHECK for the ExitMonitor call. https://bugzilla.redhat.com/show_bug.cgi?id=1161024 Ján Tomko (14): Check for domain liveness in qemuDomainObjExitMonitor Mark the domain as active in qemuhotplugtest Fix vmdef usage after domain crash in monitor on device removal Fix vmdef usage after domain crash in monitor on device detach Fix vmdef usage after domain crash in monitor on device attach Fix vmdef usage while in monitor in qemuDomainHotplugVcpus Fix vmdef usage while in monitor in BlockStat* APIs Fix vmdef usage while in monitor in qemu process Exit early after domain crash in monitor on device hotplug Exit early after domain crash in monitor on migration Exit early after domain crash in monitor in qemu_process Exit early after domain crash in monitor in qemu_driver Exit early after domain crash in monitor on snapshots Add ATTRIBUTE_RETURN_CHECK to qemuDomainObjExitMonitor src/qemu/THREADS.txt | 5 ++ src/qemu/qemu_domain.c | 27 +++++-- src/qemu/qemu_domain.h | 7 +- src/qemu/qemu_driver.c | 196 +++++++++++++++++++++++++++++----------------- src/qemu/qemu_hotplug.c | 183 ++++++++++++++++++++++++++----------------- src/qemu/qemu_hotplug.h | 6 +- src/qemu/qemu_migration.c | 137 ++++++++++++++++---------------- src/qemu/qemu_process.c | 128 +++++++++++++++++------------- tests/qemuhotplugtest.c | 6 ++ 9 files changed, 410 insertions(+), 285 deletions(-) -- 2.0.4

2 35

[libvirt] [PATCH 0/2] check IOMMU group devices usage during vfio device passthrough
by Shivaprasad G Bhat 14 Jan '15

14 Jan '15

Problem: ==================================================================== If a device in the same iommu group is in use by a different vm, the guest boot fails with the below error during vfio passthrough. bash-4.2$ virsh start demo error: Failed to start domain demo error: internal error: process exited while connecting to monitor: 2014-12-02T13:43:52.020136Z qemu-system-x86_64: -device vfio-pci,host=00:1c.3,id=hostdev0,bus=pci.0,addr=0x5: vfio: error opening /dev/vfio/7: Device or resource busy Solution: ===================================================================== The patch iterates through the iommu group devices and errors out cleanly mentioning the device and the guest which is using the device. With Patch bash-4.2$ virsh start demo error: Failed to start domain demo error: Requested operation is not valid: PCI device 0000:0d:00.0 is in use by driver QEMU, domain vm10 --- Shivaprasad G Bhat (2): Move struct _virPCIDevice definition from virpci.c to virpci.h check IOMMU group devices usage when preparing device for vfio passthrough src/util/virhostdev.c | 87 +++++++++++++++++++++++++++++-------------------- src/util/virpci.c | 29 ---------------- src/util/virpci.h | 30 +++++++++++++++++ 3 files changed, 82 insertions(+), 64 deletions(-) -- Signature

3 6

[libvirt] [PATCH v3 0/2] Fix migration to older libvirt
by Pavel Hrdina 14 Jan '15

14 Jan '15

Recently I've implemented new feature that we can set "vgamem_mb" for QXL device and also I've fixed libvirt to honor the 'vram' attribute and pass that value to QEMU process. But with this change the migration to older libvirt stopped working because QEMU silently updates the video memory size if the value is too low or there are some dependencies (for example QXL device has to have 'ram' size twice as 'vgamem'). To fix the migration we need to load the updated values from QEMU and store them into the status XML. v2: - don't fail if JSON monitor isn't available v1: - removed unnecessary movement of qemuMonitorJSONObjectProperty - completely rewritten the second patch Pavel Hrdina (2): qemu_monitor: introduce new function to get QOM path qemu_process: detect updated video ram size values from QEMU src/qemu/qemu_monitor.c | 206 ++++++++++++++++++++++++++++--------------- src/qemu/qemu_monitor.h | 4 + src/qemu/qemu_monitor_json.c | 69 +++++++++++++++ src/qemu/qemu_monitor_json.h | 3 + src/qemu/qemu_process.c | 67 ++++++++++++++ 5 files changed, 279 insertions(+), 70 deletions(-) -- 2.0.5

2 7

[libvirt] [PATCH] qemu_driver: fix setting vcpus for offline domain
by Pavel Hrdina 14 Jan '15

14 Jan '15

Commit e3435caf fixed hot-plugging of vcpus with strict memory pinning on NUMA hosts, but unfortunately it also broke updating number of vcpus for offline guests using our API. The issue is that we try to create a cpu cgroup for non-running guest which fails as there are no cgroups for that domain. We should create cgroups and update cpuset.mems only if we are hot-plugging. Signed-off-by: Pavel Hrdina <phrdina(a)redhat.com> --- src/qemu/qemu_driver.c | 26 ++++++++++++++------------ 1 file changed, 14 insertions(+), 12 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index cdf4173..3a343d0 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -4576,21 +4576,23 @@ qemuDomainSetVcpusFlags(virDomainPtr dom, unsigned int nvcpus, priv = vm->privateData; - if (virCgroupNewEmulator(priv->cgroup, false, &cgroup_temp) < 0) - goto cleanup; - - if (!(all_nodes = virNumaGetHostNodeset())) - goto cleanup; - - if (!(all_nodes_str = virBitmapFormat(all_nodes))) - goto cleanup; - if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0) goto cleanup; - if (virCgroupGetCpusetMems(cgroup_temp, &mem_mask) < 0 || - virCgroupSetCpusetMems(cgroup_temp, all_nodes_str) < 0) - goto endjob; + if (flags & VIR_DOMAIN_AFFECT_LIVE) { + if (virCgroupNewEmulator(priv->cgroup, false, &cgroup_temp) < 0) + goto cleanup; + + if (!(all_nodes = virNumaGetHostNodeset())) + goto cleanup; + + if (!(all_nodes_str = virBitmapFormat(all_nodes))) + goto cleanup; + + if (virCgroupGetCpusetMems(cgroup_temp, &mem_mask) < 0 || + virCgroupSetCpusetMems(cgroup_temp, all_nodes_str) < 0) + goto endjob; + } maximum = (flags & VIR_DOMAIN_VCPU_MAXIMUM) != 0; flags &= ~VIR_DOMAIN_VCPU_MAXIMUM; -- 2.0.5

2 2

[libvirt] [PATCH v2 0/4] UEFI and NVRAM store file format
by Michal Privoznik 13 Jan '15

13 Jan '15

diff to v1: -Laszlo's review worked in (notably, even UEFI can have a different format than raw) Michal Privoznik (4): conf: Make virDomainLoaderDefParseXML parse <nvram/> too conf: Allow UEFI and NVRAM format selection qemu: Adapt to UEFI and NVRAM store file format change domaincaps: Expose UEFI and NVRAM store file format capabilities docs/formatdomain.html.in | 18 +++++-- docs/formatdomaincaps.html.in | 26 ++++++++++ docs/schemas/domaincaps.rng | 13 +++++ docs/schemas/domaincommon.rng | 16 ++++++ src/conf/domain_capabilities.c | 18 +++++++ src/conf/domain_capabilities.h | 2 + src/conf/domain_conf.c | 57 +++++++++++++++++----- src/conf/domain_conf.h | 11 +++++ src/libvirt_private.syms | 2 + src/qemu/qemu_capabilities.c | 10 +++- src/qemu/qemu_command.c | 12 +++-- tests/domaincapsschemadata/domaincaps-full.xml | 10 ++++ .../domaincaps-qemu_1.6.50-1.xml | 10 ++++ tests/domaincapstest.c | 2 + .../qemuxml2argv-bios-nvram-qcow2.args | 10 ++++ .../qemuxml2argv-bios-nvram-qcow2.xml | 40 +++++++++++++++ tests/qemuxml2argvtest.c | 2 + .../qemuxml2xmlout-bios-nvram.xml | 40 +++++++++++++++ tests/qemuxml2xmltest.c | 2 +- 19 files changed, 279 insertions(+), 22 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-bios-nvram-qcow2.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-bios-nvram-qcow2.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-bios-nvram.xml -- 2.0.5

2 9

[libvirt] [PATCH v2] qemu, lxc: Warn if setting QoS on unsupported vNIC types
by Michal Privoznik 13 Jan '15

13 Jan '15

https://bugzilla.redhat.com/show_bug.cgi?id=1165993 So, there are still plenty of vNIC types that we don't know how to set bandwidth on. Let's warn explicitly in case user has requested it instead of pretending everything was set. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Diff to v1: -switch from virReportError(); goto error; to VIR_WARN(). src/lxc/lxc_driver.c | 18 +++++++++++++----- src/lxc/lxc_process.c | 18 +++++++++++++----- src/qemu/qemu_command.c | 19 +++++++++++++------ src/qemu/qemu_hotplug.c | 18 +++++++++++++----- 4 files changed, 52 insertions(+), 21 deletions(-) diff --git a/src/lxc/lxc_driver.c b/src/lxc/lxc_driver.c index f503fcf..07ddce3 100644 --- a/src/lxc/lxc_driver.c +++ b/src/lxc/lxc_driver.c @@ -4148,6 +4148,7 @@ lxcDomainAttachDeviceNetLive(virConnectPtr conn, virLXCDomainObjPrivatePtr priv = vm->privateData; int ret = -1; int actualType; + virNetDevBandwidthPtr actualBandwidth; char *veth = NULL; if (!priv->initpid) { @@ -4195,11 +4196,18 @@ lxcDomainAttachDeviceNetLive(virConnectPtr conn, _("Network device type is not supported")); goto cleanup; } - /* set network bandwidth */ - if (virNetDevSupportBandwidth(actualType) && - virNetDevBandwidthSet(net->ifname, - virDomainNetGetActualBandwidth(net), false) < 0) - goto cleanup; + /* Set bandwidth or warn if requested and not supported. */ + actualBandwidth = virDomainNetGetActualBandwidth(net); + if (actualBandwidth) { + if (virNetDevSupportBandwidth(actualType)) { + if (virNetDevBandwidthSet(net->ifname, actualBandwidth, false) < 0) + goto cleanup; + } else { + VIR_WARN("setting bandwidth on interfaces of " + "type '%s' is not implemented yet", + virDomainNetTypeToString(actualType)); + } + } if (virNetDevSetNamespace(veth, priv->initpid) < 0) { virDomainAuditNet(vm, NULL, net, "attach", false); diff --git a/src/lxc/lxc_process.c b/src/lxc/lxc_process.c index af08de9..89e9249 100644 --- a/src/lxc/lxc_process.c +++ b/src/lxc/lxc_process.c @@ -369,6 +369,7 @@ static int virLXCProcessSetupInterfaces(virConnectPtr conn, for (i = 0; i < def->nnets; i++) { char *veth = NULL; + virNetDevBandwidthPtr actualBandwidth; /* If appropriate, grab a physical device from the configured * network's pool of devices, or resolve bridge device name * to the one defined in the network definition. @@ -422,11 +423,18 @@ static int virLXCProcessSetupInterfaces(virConnectPtr conn, } - /* set network bandwidth */ - if (virNetDevSupportBandwidth(type) && - virNetDevBandwidthSet(net->ifname, - virDomainNetGetActualBandwidth(net), false) < 0) - goto cleanup; + /* Set bandwidth or warn if requested and not supported. */ + actualBandwidth = virDomainNetGetActualBandwidth(net); + if (actualBandwidth) { + if (virNetDevSupportBandwidth(type)) { + if (virNetDevBandwidthSet(net->ifname, actualBandwidth, false) < 0) + goto cleanup; + } else { + VIR_WARN("setting bandwidth on interfaces of " + "type '%s' is not implemented yet", + virDomainNetTypeToString(type)); + } + } (*veths)[(*nveths)-1] = veth; diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index 3346e95..3cafbfb 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -7365,6 +7365,7 @@ qemuBuildInterfaceCommandLine(virCommandPtr cmd, char **tapfdName = NULL; char **vhostfdName = NULL; int actualType = virDomainNetGetActualType(net); + virNetDevBandwidthPtr actualBandwidth; size_t i; if (actualType == VIR_DOMAIN_NET_TYPE_VHOSTUSER) @@ -7415,12 +7416,18 @@ qemuBuildInterfaceCommandLine(virCommandPtr cmd, goto cleanup; } - /* Set Bandwidth */ - if (virNetDevSupportBandwidth(actualType) && - virNetDevBandwidthSet(net->ifname, - virDomainNetGetActualBandwidth(net), - false) < 0) - goto cleanup; + /* Set bandwidth or warn if requested and not supported. */ + actualBandwidth = virDomainNetGetActualBandwidth(net); + if (actualBandwidth) { + if (virNetDevSupportBandwidth(actualType)) { + if (virNetDevBandwidthSet(net->ifname, actualBandwidth, false) < 0) + goto cleanup; + } else { + VIR_WARN("setting bandwidth on interfaces of " + "type '%s' is not implemented yet", + virDomainNetTypeToString(actualType)); + } + } if ((actualType == VIR_DOMAIN_NET_TYPE_NETWORK || actualType == VIR_DOMAIN_NET_TYPE_BRIDGE || diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 1714341..bf1f69a 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -838,6 +838,7 @@ int qemuDomainAttachNetDevice(virConnectPtr conn, bool releaseaddr = false; bool iface_connected = false; int actualType; + virNetDevBandwidthPtr actualBandwidth; virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver); size_t i; @@ -926,11 +927,18 @@ int qemuDomainAttachNetDevice(virConnectPtr conn, if (qemuInterfaceStartDevice(net) < 0) goto cleanup; - /* Set Bandwidth */ - if (virNetDevSupportBandwidth(actualType) && - virNetDevBandwidthSet(net->ifname, - virDomainNetGetActualBandwidth(net), false) < 0) - goto cleanup; + /* Set bandwidth or warn if requested and not supported. */ + actualBandwidth = virDomainNetGetActualBandwidth(net); + if (actualBandwidth) { + if (virNetDevSupportBandwidth(actualType)) { + if (virNetDevBandwidthSet(net->ifname, actualBandwidth, false) < 0) + goto cleanup; + } else { + VIR_WARN("setting bandwidth on interfaces of " + "type '%s' is not implemented yet", + virDomainNetTypeToString(actualType)); + } + } for (i = 0; i < tapfdSize; i++) { if (virSecurityManagerSetTapFDLabel(driver->securityManager, -- 2.0.5

2 1

[libvirt] [PATCH 0/2] Forbid starting VMs with bandwidth on unsupported vNIC type
by Michal Privoznik 13 Jan '15

13 Jan '15

The patch 1/2 is trivial. The patch 2/2 is controversial and I expect heavy discussion on it :-) Thing is, there might be somebody misusing our XML to store the QoS setting, and do the actual setting themselves in startup hook. Now with this patch, we will forbid such usage scenario. Michal Privoznik (2): conf: Increase virNetDevBandwidthParse intelligence qemu, lxc: Fail explicitly if setting QoS on unsupported vNIC types src/conf/domain_conf.c | 10 ++++++---- src/conf/netdev_bandwidth_conf.c | 38 +++++++++++++++++++++++--------------- src/conf/netdev_bandwidth_conf.h | 7 ++++--- src/conf/network_conf.c | 7 +++---- src/lxc/lxc_driver.c | 20 +++++++++++++++----- src/lxc/lxc_process.c | 20 +++++++++++++++----- src/qemu/qemu_command.c | 21 +++++++++++++++------ src/qemu/qemu_hotplug.c | 20 +++++++++++++++----- tests/virnetdevbandwidthtest.c | 12 ++++++------ 9 files changed, 102 insertions(+), 53 deletions(-) -- 2.0.5

4 7

[libvirt] It does not support hot-plugging VHOST-USER networkcard in libvirt?
by zhang bo 13 Jan '15

13 Jan '15

It works well if we use qmp command to directly interact with qemu 2.1 to hot-add vhost-user network card. However, libvirt seems not support hotplugging well. Steps to use qmp commands to directly(without libvirtd) interact with qemu 2.1 : 1 (qemu) chardev-add backend=socket,id=charnet1,path=/var/run/vhost-user/tap1 2 (qemu) netdev_add vhost-user,id=hostnet1,chardev=charnet1 3 (qemu) device_add virtio-net-pci,netdev=hostnet1,id=net1 In libvirt, we found that, Hot-plug is finished with a success result, but the network card doesn't work afterwards, it could not send/receive IOs. As we further digged into the problem, the 1st step of qmp commands is missing if we use libvirt to do the hot-plug job. 1 (qemu) chardev-add backend=socket,id=charnet1,path=/var/run/vhost-user/tap1 After looking into the function qemuDomainAttachNetDevice(), we found that it does have 1) qemuMonitorAddNetdev() 2) qemuMonitorAddDevice() except for *ChardevAdd*() The question is : why does qemuDomainAttachNetDevice() not do the *ChardevAdd*() work? It seems necessary to add chardev to add a vhost-user network card.

2 1