February 2014 - Devel - libvirt List Archives

[libvirt] [PATCH] spec: Use correct versions of libgfapi in RHEL builds
by Peter Krempa 26 Feb '14

26 Feb '14

RHEL still uses the 3.4.0 package of libgfapi and the package is built only for x86_64. --- libvirt.spec.in | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/libvirt.spec.in b/libvirt.spec.in index 3d5a69e..67b5557 100644 --- a/libvirt.spec.in +++ b/libvirt.spec.in @@ -98,7 +98,7 @@ %else %define with_storage_sheepdog 0 %endif -%if 0%{?fedora} >= 19 +%if 0%{?fedora} >= 19 || 0%{?rhel >= 6} %define with_storage_gluster 0%{!?_without_storage_gluster:%{server_drivers}} %else %define with_storage_gluster 0 @@ -153,6 +153,13 @@ %define with_numactl 0 %endif +# libgfapi is built only on x86_64 on rhel +%ifnarch x86_64 + %if 0%{?rhel >= 6} + %define with_storage_gluster 0 + %endif +%endif + # RHEL doesn't ship OpenVZ, VBox, UML, PowerHypervisor, # VMWare, libxenserver (xenapi), libxenlight (Xen 4.1 and newer), # or HyperV. @@ -571,9 +578,14 @@ BuildRequires: ceph-devel %endif %endif %if %{with_storage_gluster} +%if 0%{?rhel} >= 6 +BuildRequires: glusterfs-api-devel >= 3.4.0 +BuildRequires: glusterfs-devel >= 3.4.0 +%else BuildRequires: glusterfs-api-devel >= 3.4.1 BuildRequires: glusterfs-devel >= 3.4.1 %endif +%endif %if %{with_numactl} # For QEMU/LXC numa info BuildRequires: numactl-devel -- 1.8.5.5

4 4

[libvirt] [PATCH 0/3] spec file improvements
by Eric Blake 26 Feb '14

26 Feb '14

Our spec file is non-deterministic if you have this in your ~/.rpmmacros: %_without_udev 1 because it then depends on whether you have systemd-devel installed rather than being explicitly stated in the spec file. [At this point, I'm sending the series to get the review started on the two trivial parts; I'm still working on patch 3, including coming up with a formula for proving that the unpatched build was non-deterministic, rather than just my claim above] Eric Blake (2): build: fix spelling of configure --with-systemd-daemon spec: explicitly avoid bhyve on Linux libvirt.spec.in | 2 ++ m4/virt-lib.m4 | 20 +++++++++++++------- 2 files changed, 15 insertions(+), 7 deletions(-) -- 1.8.5.3

3 7

[libvirt] [PATCH] virsh: Honour -q in domblklist, vcpupin and emulatorpin
by Michal Privoznik 26 Feb '14

26 Feb '14

If user wants to grep some info from domain, e.g. disk paths: # virsh -q domblklist win7 | awk '{print $2}' Source /var/lib/libvirt/images/windows.qcow2 /home/zippy/work/tmp/en_windows_7_professional_x64_dvd_X15-65805.iso while with my change: # virsh -q domblklist win7 | awk '{print $2}' /var/lib/libvirt/images/windows.qcow2 /home/zippy/work/tmp/en_windows_7_professional_x64_dvd_X15-65805.iso We don't print table header in other commands, like list. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- tools/virsh-domain-monitor.c | 14 +++++++------- tools/virsh-domain.c | 10 +++++----- 2 files changed, 12 insertions(+), 12 deletions(-) diff --git a/tools/virsh-domain-monitor.c b/tools/virsh-domain-monitor.c index 1192565..4905b7e 100644 --- a/tools/virsh-domain-monitor.c +++ b/tools/virsh-domain-monitor.c @@ -495,12 +495,12 @@ cmdDomblklist(vshControl *ctl, const vshCmd *cmd) goto cleanup; if (details) - vshPrint(ctl, "%-10s %-10s %-10s %s\n", _("Type"), - _("Device"), _("Target"), _("Source")); + vshPrintExtra(ctl, "%-10s %-10s %-10s %s\n", _("Type"), + _("Device"), _("Target"), _("Source")); else - vshPrint(ctl, "%-10s %s\n", _("Target"), _("Source")); + vshPrintExtra(ctl, "%-10s %s\n", _("Target"), _("Source")); - vshPrint(ctl, "------------------------------------------------\n"); + vshPrintExtra(ctl, "------------------------------------------------\n"); for (i = 0; i < ndisks; i++) { char *type = NULL; @@ -609,9 +609,9 @@ cmdDomiflist(vshControl *ctl, const vshCmd *cmd) if (ninterfaces < 0) goto cleanup; - vshPrint(ctl, "%-10s %-10s %-10s %-11s %s\n", _("Interface"), _("Type"), - _("Source"), _("Model"), _("MAC")); - vshPrint(ctl, "-------------------------------------------------------\n"); + vshPrintExtra(ctl, "%-10s %-10s %-10s %-11s %s\n", _("Interface"), + _("Type"), _("Source"), _("Model"), _("MAC")); + vshPrintExtra(ctl, "-------------------------------------------------------\n"); for (i = 0; i < ninterfaces; i++) { char *type = NULL; diff --git a/tools/virsh-domain.c b/tools/virsh-domain.c index 3e73340..2e3f0ed 100644 --- a/tools/virsh-domain.c +++ b/tools/virsh-domain.c @@ -5815,8 +5815,8 @@ cmdVcpuPin(vshControl *ctl, const vshCmd *cmd) if ((ncpus = virDomainGetVcpuPinInfo(dom, info.nrVirtCpu, cpumaps, cpumaplen, flags)) >= 0) { - vshPrint(ctl, "%s %s\n", _("VCPU:"), _("CPU Affinity")); - vshPrint(ctl, "----------------------------------\n"); + vshPrintExtra(ctl, "%s %s\n", _("VCPU:"), _("CPU Affinity")); + vshPrintExtra(ctl, "----------------------------------\n"); for (i = 0; i < ncpus; i++) { if (vcpu != -1 && i != vcpu) continue; @@ -5944,9 +5944,9 @@ cmdEmulatorPin(vshControl *ctl, const vshCmd *cmd) cpumaps = vshMalloc(ctl, cpumaplen); if (virDomainGetEmulatorPinInfo(dom, cpumaps, cpumaplen, flags) >= 0) { - vshPrint(ctl, "%s %s\n", _("emulator:"), _("CPU Affinity")); - vshPrint(ctl, "----------------------------------\n"); - vshPrint(ctl, " *: "); + vshPrintExtra(ctl, "%s %s\n", _("emulator:"), _("CPU Affinity")); + vshPrintExtra(ctl, "----------------------------------\n"); + vshPrintExtra(ctl, " *: "); ret = vshPrintPinInfo(cpumaps, cpumaplen, maxcpu, 0); vshPrint(ctl, "\n"); } -- 1.9.0

2 1

[libvirt] [PATCH] network: don't even call networkRunHook if there is no network
by Laine Stump 26 Feb '14

26 Feb '14

networkAllocateActualDevice() is called for *all* interfaces, not just those with type='network'. In that case, it will jump down to its validate: label immediately, without allocating anything. After validation is done, two counters are potentially updated (one for the network, and one for any particular physical device that is chosen), and then networkRunHook() is called. This patch refactors that code a slight bit so that networkRunHook() doesn't get called if netdef is NULL (i.e. type != network) and to place the conditional increment of dev->connections inside the "if (netdef)" as well - dev can never be non-null if netdef is null (because "dev" is the pointer to a device in a network's pool of devices), so this doesn't have any functional effect, it just makes the code clearer. --- src/network/bridge_driver.c | 47 ++++++++++++++++++++++----------------------- 1 file changed, 23 insertions(+), 24 deletions(-) diff --git a/src/network/bridge_driver.c b/src/network/bridge_driver.c index 152bd06..3fb5ad3 100644 --- a/src/network/bridge_driver.c +++ b/src/network/bridge_driver.c @@ -3662,36 +3662,35 @@ validate: } } - if (dev) { - /* mark the allocation */ - dev->connections++; - if (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV) { - VIR_DEBUG("Using physical device %s, %d connections", - dev->device.dev, dev->connections); - } else { - VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d", - dev->device.pci.domain, dev->device.pci.bus, - dev->device.pci.slot, dev->device.pci.function, - dev->connections); - } - } - if (netdef) { netdef->connections++; VIR_DEBUG("Using network %s, %d connections", netdef->name, netdef->connections); - } - /* finally we can call the 'plugged' hook script if any */ - if (networkRunHook(network, dom, iface, - VIR_HOOK_NETWORK_OP_IFACE_PLUGGED, - VIR_HOOK_SUBOP_BEGIN) < 0) { - /* adjust for failure */ - if (dev) - dev->connections--; - if (netdef) + if (dev) { + /* mark the allocation */ + dev->connections++; + if (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV) { + VIR_DEBUG("Using physical device %s, %d connections", + dev->device.dev, dev->connections); + } else { + VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d", + dev->device.pci.domain, dev->device.pci.bus, + dev->device.pci.slot, dev->device.pci.function, + dev->connections); + } + } + + /* finally we can call the 'plugged' hook script if any */ + if (networkRunHook(network, dom, iface, + VIR_HOOK_NETWORK_OP_IFACE_PLUGGED, + VIR_HOOK_SUBOP_BEGIN) < 0) { + /* adjust for failure */ netdef->connections--; - goto error; + if (dev) + dev->connections--; + goto error; + } } ret = 0; -- 1.8.5.3

3 3

[libvirt] [PATCH] network: unplug bandwidth and call networkRunHook only when appropriate
by Laine Stump 26 Feb '14

26 Feb '14

According to commit b4e0299dm if networkAllocateActualDevice() was successful, it will *always* allocate an iface->data.network.actual, so we can use this during networkReleaseActualDevice() to know if there is really anything to undo. We were properly using this information to only decrement the network connections counter if it had previously been incremented, but we were unconditionally unplugging bandwidth and calling the "unplugged" network hook for *all* interfaces (during qemuProcessStop()) whether they had been previously plugged or not. This caused problems if a domain failed to start at some time prior to all interfaces being allocated. (I encountered this when an interface had a bandwidth floor set but no inbound QoS). This patch changes both the call to networkUnplugBandwidth() and the call to networkRunHook() to only be called if there was a previous call to "plug" for the same interface. --- src/network/bridge_driver.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/src/network/bridge_driver.c b/src/network/bridge_driver.c index 3fb5ad3..c797f8f 100644 --- a/src/network/bridge_driver.c +++ b/src/network/bridge_driver.c @@ -3936,7 +3936,8 @@ networkReleaseActualDevice(virDomainDefPtr dom, } netdef = network->def; - if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE || + if (iface->data.network.actual && + (netdef->forward.type == VIR_NETWORK_FORWARD_NONE || netdef->forward.type == VIR_NETWORK_FORWARD_NAT || netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) && networkUnplugBandwidth(network, iface) < 0) @@ -4029,15 +4030,15 @@ networkReleaseActualDevice(virDomainDefPtr dom, } success: - if (iface->data.network.actual) + if (iface->data.network.actual) { netdef->connections--; + VIR_DEBUG("Releasing network %s, %d connections", + netdef->name, netdef->connections); - /* finally we can call the 'unplugged' hook script if any */ - networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED, - VIR_HOOK_SUBOP_BEGIN); - - VIR_DEBUG("Releasing network %s, %d connections", - netdef->name, netdef->connections); + /* finally we can call the 'unplugged' hook script if any */ + networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED, + VIR_HOOK_SUBOP_BEGIN); + } ret = 0; cleanup: if (network) -- 1.8.5.3

3 2

[libvirt] [PATCH] Fix memory leak in virSCSIDeviceListDel()
by Nehal J Wani 26 Feb '14

26 Feb '14

While running virscsitest, it was found that valgrind pointed out the following memory leak: ==320== 5 bytes in 1 blocks are definitely lost in loss record 4 of 37 ==320== at 0x4A069EE: malloc (vg_replace_malloc.c:270) ==320== by 0x3E6CE81171: strdup (strdup.c:43) ==320== by 0x4CB28DF: virStrdup (virstring.c:554) ==320== by 0x4CAC987: virSCSIDeviceSetUsedBy (virscsi.c:289) ==320== by 0x402321: test2 (virscsitest.c:100) ==320== by 0x403231: virtTestRun (testutils.c:199) ==320== by 0x402121: mymain (virscsitest.c:180) ==320== by 0x4039AD: virtTestMain (testutils.c:782) ==320== by 0x3E6CE1ED1C: (below main) (libc-start.c:226) ==320== --- src/util/virscsi.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/src/util/virscsi.c b/src/util/virscsi.c index acc3815..2f469f2 100644 --- a/src/util/virscsi.c +++ b/src/util/virscsi.c @@ -435,6 +435,7 @@ virSCSIDeviceListDel(virSCSIDeviceListPtr list, for (i = 0; i < dev->n_used_by; i++) { if (STREQ_NULLABLE(dev->used_by[i], name)) { if (dev->n_used_by > 1) { + VIR_FREE(dev->used_by[i]); VIR_DELETE_ELEMENT(dev->used_by, i, dev->n_used_by); } else { tmp = virSCSIDeviceListSteal(list, dev); -- 1.7.1

2 1

[libvirt] [question] doubt about cpu model 'qemu64'
by Wangyufei (James) 26 Feb '14

26 Feb '14

Hello, Recently I noticed that if I gave no model to qemu, then I would get 'qemu64' as default cpu model. Model 'qemu64' defined in qemu and defined in libvirt both support feature 'svm'. And if I start a VM with 'qemu64' as the qemu default model, I can get no 'svm' feature in the GuestOS: fpu de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht pni cx16 popcnt hypervisor syscall nx lm lahf_lm My host cpu is intel E5620, and it doesn't support 'svm' feature: 'fpu', 'vme', 'de', 'pse', 'tsc', 'msr', 'pae', 'mce', 'cx8', 'apic', 'sep', 'mtrr', 'pge', 'mca', 'cmov', 'pat', 'pse36', 'clflush', 'ds', 'acpi', 'mmx', 'fxsr', 'sse', 'sse2', 'ss', 'ht', 'tm', 'pbe', 'pni', 'pclmulqdq', 'dtes64', 'monitor', 'ds_cpl', 'vmx', 'smx', 'est', 'tm2', 'ssse3', 'cx16', 'xtpr', 'pdcm', 'pcid', 'dca', 'sse4.1', 'sse4.2', 'popcnt', 'aes', 'syscall', 'nx', 'pdpe1gb', 'rdtscp', 'lm', 'lahf_lm', 'dts', 'arat', 'constant_tsc' I can get the same result with libvirt API compareCPU on the host: model:qemu64; ret:0 model:Conroe; ret:2 model:Penryn; ret:2 model:Nehalem; ret:2 model:Westmere; ret:2 model:SandyBridge; ret:0 model:Haswell; ret:0 Isn't it strange that the host cpu doesn't support 'qemu64' model, but we can still start a VM with 'qemu64' as the default model on it? Best Regards, -WangYufei

1 0

[libvirt] Entering freeze for libvirt-1.2.2
by Daniel Veillard 26 Feb '14

26 Feb '14

As planned I tagged a release candidate 1 version in git and made tarballs and rpms available at the usual place: ftp://libvirt.org/libvirt/ it seems to work well in my limited testing, but let's spend some time this week making more tests and checking it compiles on other systems and architectures. if all goes well I will probably push the release next week-end, with an rc2 on Wednesday based on feedback, 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/

3 4

[libvirt] [PATCH 0/2] fix mingw build
by Eric Blake 25 Feb '14

25 Feb '14

I hit some issues while running ./autobuild.sh on a machine set up for a mingw cross build; pushing this series under the build-breaker rule. Eric Blake (2): build: fix cgroups on non-Linux build: avoid ld_preload tests on mingw src/util/vircgroup.c | 27 +++++++++++++++++++++++++- tests/Makefile.am | 7 ++++--- tests/virportallocatortest.c | 45 +++++++++++++++++++++++++++----------------- 3 files changed, 58 insertions(+), 21 deletions(-) -- 1.8.5.3

1 2

[libvirt] [PATCH] virNetDevVethCreate: Serialize callers
by Michal Privoznik 25 Feb '14

25 Feb '14

Consider dozen of LXC domains, each of them having this type of interface: <interface type='network'> <mac address='52:54:00:a7:05:4b'/> <source network='default'/> </interface> When starting these domain in parallel, all workers may meet in virNetDevVethCreate() where a race starts. Race over allocating veth pairs because allocation requires two steps: 1) find first nonexistent '/sys/class/net/vnet%d/' 2) run 'ip link add ...' command Now consider two threads. Both of them find N as the first unused veth index but only one of them succeeds allocating it. The other one fails. For such cases, we are running the allocation in a loop with 10 rounds. However this is very flaky synchronization. It should be rather used when libvirt is competing with other process than when libvirt threads fight each other. Therefore, internally we should use mutex to serialize callers, and do the allocation in loop (just in case we are competing with a different process). By the way we have something similar already since 1cf97c87. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/util/virnetdevveth.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/src/util/virnetdevveth.c b/src/util/virnetdevveth.c index 25eb282..e698ce2 100644 --- a/src/util/virnetdevveth.c +++ b/src/util/virnetdevveth.c @@ -39,6 +39,19 @@ /* Functions */ +virMutex virNetDevVethCreateMutex; + +static int virNetDevVethCreateMutexOnceInit(void) +{ + if (virMutexInit(&virNetDevVethCreateMutex) < 0) { + virReportSystemError(errno, "%s", _("unable to init mutex")); + return -1; + } + return 0; +} + +VIR_ONCE_GLOBAL_INIT(virNetDevVethCreateMutex); + static int virNetDevVethExists(int devNum) { int ret; @@ -117,6 +130,10 @@ int virNetDevVethCreate(char** veth1, char** veth2) * We might race with other containers, but this is reasonably * unlikely, so don't do too many retries for device creation */ + if (virNetDevVethCreateMutexInitialize() < 0) + return -1; + + virMutexLock(&virNetDevVethCreateMutex); #define MAX_VETH_RETRIES 10 for (i = 0; i < MAX_VETH_RETRIES; i++) { @@ -179,6 +196,7 @@ int virNetDevVethCreate(char** veth1, char** veth2) MAX_VETH_RETRIES); cleanup: + virMutexUnlock(&virNetDevVethCreateMutex); virCommandFree(cmd); VIR_FREE(veth1auto); VIR_FREE(veth2auto); -- 1.9.0

2 1