- Devel - libvirt List Archives

[libvirt] Bug 1058149 �C libvirt crashed when executing qemu-agent-command during shutting down virutal machine
by hzguanqiang＠corp.netease.com 27 Jan '14

27 Jan '14

Hi experts, When I executing qemu-agent-command during shutting down the guest machine, libvirtd caught a segmentation violation and exited abnormally. Details can be seen from the url: https://bugzilla.redhat.com/show_bug.cgi?id=1058149 Is there anyone who would like to see it. Thanks~ ------------------ Best regards! GuanQiang 2014-01-27

1 0

[libvirt] Libvirt and QBG with VEB
by Padmanabhan Krishnan 27 Jan '14

27 Jan '14

Hello, I have a setup of virtual machines and wherein my virtual switch (OVS) takes care of switching locally and it uses the network for talking to VM's outside the server (like a VEB scenario I guess). I want Netlink messages to be sent to LLDPAD. I understand that i had to modify the XML files and start the VM. But, i see from the XML that it supports only VEPA, not VEB. But, irrespective of VEB or VEPA, those VM information should be known to the upstream switch through VDP. 802.1QBG seem to support this (both VEPA and VEB) http://www.ieee802.org/1/files/private/bg-drafts/d2/802-1qbg-d2-2.pdf But, for VEB, no net link messages are sent by libvirt. Why is that? From the XML example, I see that: <interface type=’direct’> <mac address=’d0:0f:d0:0f:02:01’/> <source dev=’eth2.20’ mode=’vepa’/> <source network=’default’/> <virtualport type=’802.1Qbg’> <parameters managerid=’12’ typeid=’0x123456’ typeidversion=’1’ instanceid=’09b00c53-8b5c-4eeb-8f00-d847aa05191b’ /> </virtualport> I tried different combinations by changing the interface type to bridge and and trying Qbg as in below, I find that it doesn't send a Netlink message to LLDPAD. There's no "mode=veb". </controller> <interface type='bridge'> <mac address='00:54:00:4e:46:04'/> <source bridge='xml_br'/> <virtualport type='802.1Qbg'> <parameters managerid='12' typeid='1193046' typeidversion='1' instanceid='09b00c53-8b5c-4eeb-8f00-d847aa05191b'/> </virtualport> <target dev='tap_xml'/> <model type='virtio'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x03' function='0x0'/> </interface> How do i get this to work? Is this a limitation? Thanks,KP

1 0

[libvirt] [PATCH] Fix segmentation fault when accessing default qemu machine type
by Yudai Yamagishi 26 Jan '14

26 Jan '14

From: Yudai Yamagish <yummy(a)sfc.wide.ad.jp> This patch fixes a segmentation fault when creating new virtual machines using QEMU. The segmentation fault is caused by commit f41830680e40d3ec845cefd25419bd87414b9ccf and commit cbb6ec42e2447d7920b30d66923b2a2b2670133b. In virQEMUCapsProbeQMPMachineTypes, when copying machines to qemuCaps, "none" is skipped. Therefore, the value of i and "qemuCaps->nmachineTypes - 1" do not always match. However, defIdx value (used to call virQEMUCapsSetDefaultMachine) is set using the value in i when the array elements are in qemuCaps->nmachineTypes - 1. So, when libvirt tries to create virtual machines using the default machine type, qemuCaps->machineTypes[defIdx] is accessed and since the defIdx is NULL, it results in segmentation fault. Signed-off-by: Yudai Yamagishi <yummy(a)sfc.wide.ad.jp> --- src/qemu/qemu_capabilities.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 5e9c65e..5def55c 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -2150,7 +2150,7 @@ virQEMUCapsProbeQMPMachineTypes(virQEMUCapsPtr qemuCaps, machines[i]->name) < 0) goto cleanup; if (machines[i]->isDefault) - defIdx = i; + defIdx = qemuCaps->nmachineTypes - 1; qemuCaps->machineMaxCpus[qemuCaps->nmachineTypes - 1] = machines[i]->maxCpus; } -- 1.7.2.5

2 5

[libvirt] [PATCH] qemu: Enable 'host-passthrough' cpu mode for aarch64
by Oleg Strikov 25 Jan '14

25 Jan '14

This patch allows libvirt user to specify 'host-passthrough' cpu mode while using qemu/kvm backend on aarch64. It uses 'host' as a CPU model name instead of some other stub (correct CPU detection is not implemented yet) to allow libvirt user to specify 'host-model' cpu mode as well. Signed-off-by: Oleg Strikov <oleg.strikov(a)canonical.com> --- src/cpu/cpu_aarch64.c | 30 ++++++++++++++++++++++++++---- 1 file changed, 26 insertions(+), 4 deletions(-) diff --git a/src/cpu/cpu_aarch64.c b/src/cpu/cpu_aarch64.c index 3959deb..4afe9db 100644 --- a/src/cpu/cpu_aarch64.c +++ b/src/cpu/cpu_aarch64.c @@ -25,6 +25,7 @@ #include "viralloc.h" #include "cpu.h" +#include "virstring.h" #define VIR_FROM_THIS VIR_FROM_CPU @@ -44,16 +45,19 @@ AArch64NodeData(virArch arch) } static int -AArch64Decode(virCPUDefPtr cpu ATTRIBUTE_UNUSED, +AArch64Decode(virCPUDefPtr cpu, const virCPUData *data ATTRIBUTE_UNUSED, const char **models ATTRIBUTE_UNUSED, unsigned int nmodels ATTRIBUTE_UNUSED, const char *preferred ATTRIBUTE_UNUSED, unsigned int flags) { - virCheckFlags(VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES, -1); + if (cpu->model == NULL && + VIR_STRDUP(cpu->model, "host") < 0) + return -1; + return 0; } @@ -63,6 +67,24 @@ AArch64DataFree(virCPUDataPtr data) VIR_FREE(data); } +static int +AArch64Update(virCPUDefPtr guest, + const virCPUDef *host) +{ + guest->match = VIR_CPU_MATCH_EXACT; + virCPUDefFreeModel(guest); + return virCPUDefCopyModel(guest, host, true); +} + +static virCPUCompareResult +AArch64GuestData(virCPUDefPtr host ATTRIBUTE_UNUSED, + virCPUDefPtr guest ATTRIBUTE_UNUSED, + virCPUDataPtr *data ATTRIBUTE_UNUSED, + char **message ATTRIBUTE_UNUSED) +{ + return VIR_CPU_COMPARE_IDENTICAL; +} + struct cpuArchDriver cpuDriverAARCH64 = { .name = "aarch64", .arch = archs, @@ -72,8 +94,8 @@ struct cpuArchDriver cpuDriverAARCH64 = { .encode = NULL, .free = AArch64DataFree, .nodeData = AArch64NodeData, - .guestData = NULL, + .guestData = AArch64GuestData, .baseline = NULL, - .update = NULL, + .update = AArch64Update, .hasFeature = NULL, }; -- 1.7.9.5

3 2

[libvirt] [PATCH] tests: Fix PCI test data filenames for Windows
by Matthias Bolte 25 Jan '14

25 Jan '14

Windows doesn't allow : in filenames. Commit 21685c955e546676a5b2a01f7b788d222e0ee0f5 added files with a : in their names. This broke git operations on Windows as git is not able to create those files on clone or pull. Replace : with - in the offending filenames and adapt the test case. --- tests/virpcimock.c | 19 ++++++++++++++++++- .../{0000:00:01.0.config => 0000-00-01.0.config} | Bin .../{0000:00:02.0.config => 0000-00-02.0.config} | Bin .../{0000:00:03.0.config => 0000-00-03.0.config} | Bin .../{0001:00:00.0.config => 0001-00-00.0.config} | Bin .../{0001:01:00.0.config => 0001-01-00.0.config} | Bin .../{0001:01:00.1.config => 0001-01-00.1.config} | Bin .../{0005:80:00.0.config => 0005-80-00.0.config} | Bin .../{0005:90:01.0.config => 0005-90-01.0.config} | Bin .../{0005:90:01.1.config => 0005-90-01.1.config} | Bin .../{0005:90:01.2.config => 0005-90-01.2.config} | Bin 11 files changed, 18 insertions(+), 1 deletion(-) rename tests/virpcitestdata/{0000:00:01.0.config => 0000-00-01.0.config} (100%) rename tests/virpcitestdata/{0000:00:02.0.config => 0000-00-02.0.config} (100%) rename tests/virpcitestdata/{0000:00:03.0.config => 0000-00-03.0.config} (100%) rename tests/virpcitestdata/{0001:00:00.0.config => 0001-00-00.0.config} (100%) rename tests/virpcitestdata/{0001:01:00.0.config => 0001-01-00.0.config} (100%) rename tests/virpcitestdata/{0001:01:00.1.config => 0001-01-00.1.config} (100%) rename tests/virpcitestdata/{0005:80:00.0.config => 0005-80-00.0.config} (100%) rename tests/virpcitestdata/{0005:90:01.0.config => 0005-90-01.0.config} (100%) rename tests/virpcitestdata/{0005:90:01.1.config => 0005-90-01.1.config} (100%) rename tests/virpcitestdata/{0005:90:01.2.config => 0005-90-01.2.config} (100%) diff --git a/tests/virpcimock.c b/tests/virpcimock.c index 1efcd33..b42e8bf 100644 --- a/tests/virpcimock.c +++ b/tests/virpcimock.c @@ -320,16 +320,33 @@ pci_device_new_from_stub(const struct pciDevice *data) { struct pciDevice *dev; char *devpath; + char *id; + char *c; char *configSrc; char tmp[32]; struct stat sb; + if (VIR_STRDUP_QUIET(id, data->id) < 0) + ABORT_OOM(); + + /* Replace ':' with '-' to create the config filename from the + * device ID. The device ID cannot be used directly as filename + * because it contains ':' and Windows does not allow ':' in + * filenames. */ + c = strchr(id, ':'); + + while (c) { + *c = '-'; + c = strchr(c, ':'); + } + if (VIR_ALLOC_QUIET(dev) < 0 || virAsprintfQuiet(&configSrc, "%s/virpcitestdata/%s.config", - abs_srcdir, data->id) < 0 || + abs_srcdir, id) < 0 || virAsprintfQuiet(&devpath, "%s/devices/%s", fakesysfsdir, data->id) < 0) ABORT_OOM(); + VIR_FREE(id); memcpy(dev, data, sizeof(*dev)); if (virFileMakePath(devpath) < 0) diff --git a/tests/virpcitestdata/0000:00:01.0.config b/tests/virpcitestdata/0000-00-01.0.config similarity index 100% rename from tests/virpcitestdata/0000:00:01.0.config rename to tests/virpcitestdata/0000-00-01.0.config diff --git a/tests/virpcitestdata/0000:00:02.0.config b/tests/virpcitestdata/0000-00-02.0.config similarity index 100% rename from tests/virpcitestdata/0000:00:02.0.config rename to tests/virpcitestdata/0000-00-02.0.config diff --git a/tests/virpcitestdata/0000:00:03.0.config b/tests/virpcitestdata/0000-00-03.0.config similarity index 100% rename from tests/virpcitestdata/0000:00:03.0.config rename to tests/virpcitestdata/0000-00-03.0.config diff --git a/tests/virpcitestdata/0001:00:00.0.config b/tests/virpcitestdata/0001-00-00.0.config similarity index 100% rename from tests/virpcitestdata/0001:00:00.0.config rename to tests/virpcitestdata/0001-00-00.0.config diff --git a/tests/virpcitestdata/0001:01:00.0.config b/tests/virpcitestdata/0001-01-00.0.config similarity index 100% rename from tests/virpcitestdata/0001:01:00.0.config rename to tests/virpcitestdata/0001-01-00.0.config diff --git a/tests/virpcitestdata/0001:01:00.1.config b/tests/virpcitestdata/0001-01-00.1.config similarity index 100% rename from tests/virpcitestdata/0001:01:00.1.config rename to tests/virpcitestdata/0001-01-00.1.config diff --git a/tests/virpcitestdata/0005:80:00.0.config b/tests/virpcitestdata/0005-80-00.0.config similarity index 100% rename from tests/virpcitestdata/0005:80:00.0.config rename to tests/virpcitestdata/0005-80-00.0.config diff --git a/tests/virpcitestdata/0005:90:01.0.config b/tests/virpcitestdata/0005-90-01.0.config similarity index 100% rename from tests/virpcitestdata/0005:90:01.0.config rename to tests/virpcitestdata/0005-90-01.0.config diff --git a/tests/virpcitestdata/0005:90:01.1.config b/tests/virpcitestdata/0005-90-01.1.config similarity index 100% rename from tests/virpcitestdata/0005:90:01.1.config rename to tests/virpcitestdata/0005-90-01.1.config diff --git a/tests/virpcitestdata/0005:90:01.2.config b/tests/virpcitestdata/0005-90-01.2.config similarity index 100% rename from tests/virpcitestdata/0005:90:01.2.config rename to tests/virpcitestdata/0005-90-01.2.config -- 1.8.1.2

2 2

[libvirt] [RFC PATCH] Cannot set persistent config mem/vcpu above live maxmem/maxvcpu
by Geoff Franks 25 Jan '14

25 Jan '14

I submitted bug https://bugzilla.redhat.com/show_bug.cgi?id=1038363 for being unable to raise the persistent mem/vcpu values above a live domain’s maxmem/maxvcpu values (rather than the persistent maxmem/maxvcpu values). I was asked to submit my patch here for a wider review. For memory, check the newmem value against mem.max_baloon of the live def for non-persistent changes, or mem.max_baloon of the persistent def for persistent changes. Same theory for setting vcpus/maxvcpus. --- a/libvirt0/libvirt-1.1.1/src/qemu/qemu_driver.c +++ b/libvirt0/libvirt-1.1.1/src/qemu/qemu_driver.c @@ -2236,13 +2236,13 @@ static int qemuDomainSetMemoryFlags(virDomainPtr dom, unsigned long newmem, } else { /* resize the current memory */ - if (newmem > vm->def->mem.max_balloon) { - virReportError(VIR_ERR_INVALID_ARG, "%s", - _("cannot set memory higher than max memory")); - goto endjob; - } - if (flags & VIR_DOMAIN_AFFECT_LIVE) { + if (newmem > vm->def->mem.max_balloon) { + virReportError(VIR_ERR_INVALID_ARG, "%s", + _("cannot set memory higher than max memory")); + goto endjob; + } + priv = vm->privateData; qemuDomainObjEnterMonitor(driver, vm); r = qemuMonitorSetBalloon(priv->mon, newmem); @@ -2262,6 +2262,12 @@ static int qemuDomainSetMemoryFlags(virDomainPtr dom, unsigned long newmem, } if (flags & VIR_DOMAIN_AFFECT_CONFIG) { + if (newmem > persistentDef->mem.max_balloon) { + virReportError(VIR_ERR_INVALID_ARG, "%s", + _("cannot set memory higher than max memory")); + goto endjob; + } + sa_assert(persistentDef); persistentDef->mem.cur_balloon = newmem; ret = virDomainSaveConfig(cfg->configDir, persistentDef); @@ -4160,6 +4166,7 @@ qemuDomainSetVcpusFlags(virDomainPtr dom, unsigned int nvcpus, virDomainDefPtr persistentDef; int ret = -1; bool maximum; + unsigned int maxvcpus; virQEMUDriverConfigPtr cfg = NULL; virCapsPtr caps = NULL; qemuAgentCPUInfoPtr cpuinfo = NULL; @@ -4207,11 +4214,12 @@ qemuDomainSetVcpusFlags(virDomainPtr dom, unsigned int nvcpus, goto endjob; } - if (!maximum && nvcpus > vm->def->maxvcpus) { + maxvcpus = (flags & VIR_DOMAIN_AFFECT_LIVE) ? vm->def->maxvcpus : persistentDef->maxvcpus; + if (!maximum && nvcpus > maxvcpus) { virReportError(VIR_ERR_INVALID_ARG, _("requested vcpus is greater than max allowable" " vcpus for the domain: %d > %d"), - nvcpus, vm->def->maxvcpus); + nvcpus, maxvcpus); goto endjob; }

2 2

[libvirt] [PATCH] Using qemu with sheepdog pool
by joel SIMOES 25 Jan '14

25 Jan '14

From: Joel SIMOES <joel.simoes(a)laposte.net> --- src/qemu/qemu_conf.c | 71 +++++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 65 insertions(+), 6 deletions(-) diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c index ac53f6d..dfafcdc 100644 --- a/src/qemu/qemu_conf.c +++ b/src/qemu/qemu_conf.c @@ -1185,6 +1185,56 @@ int qemuDriverAllocateID(virQEMUDriverPtr driver) return virAtomicIntInc(&driver->nextvmid); } + +static int +qemuAddSheepPoolSourceHost(virDomainDiskDefPtr def, + virStoragePoolDefPtr pooldef) +{ + int ret = -1; + char **tokens = NULL; + + /* Only support one host */ + if (pooldef->source.nhost != 1) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("Expected exactly 1 host for the storage pool")); + goto cleanup; + } + + /* iscsi pool only supports one host */ + def->nhosts = 1; + + if (VIR_ALLOC_N(def->hosts, def->nhosts) < 0) + goto cleanup; + + if (VIR_STRDUP(def->hosts[0].name, pooldef->source.hosts[0].name) < 0) + goto cleanup; + + if (virAsprintf(&def->hosts[0].port, "%d", + pooldef->source.hosts[0].port ? + pooldef->source.hosts[0].port : + 7000) < 0) + goto cleanup; + + + + + + /* Storage pool have not supported these 2 attributes yet, + * use the defaults. + */ + def->hosts[0].transport = VIR_DOMAIN_DISK_PROTO_TRANS_TCP; + def->hosts[0].socket = NULL; + + + def->protocol = VIR_DOMAIN_DISK_PROTOCOL_SHEEPDOG; + + ret = 0; + +cleanup: + virStringFreeList(tokens); + return ret; +} + static int qemuAddISCSIPoolSourceHost(virDomainDiskDefPtr def, virStoragePoolDefPtr pooldef) @@ -1352,14 +1402,15 @@ qemuTranslateDiskSourcePool(virConnectPtr conn, def->srcpool->pooltype = pooldef->type; def->srcpool->voltype = info.type; - - if (def->srcpool->mode && pooldef->type != VIR_STORAGE_POOL_ISCSI) { + + if (def->srcpool->mode && pooldef->type != VIR_STORAGE_POOL_ISCSI && !(def->srcpool->mode == VIR_DOMAIN_DISK_SOURCE_POOL_MODE_DIRECT && pooldef->type == VIR_STORAGE_POOL_SHEEPDOG) ) { + virReportError(VIR_ERR_XML_ERROR, "%s", _("disk source mode is only valid when " - "storage pool is of iscsi type")); + "storage pool is of iscsi type or only direct for sheepdog ")); goto cleanup; } - + VIR_FREE(def->src); virDomainDiskHostDefFree(def->nhosts, def->hosts); virDomainDiskAuthClear(def); @@ -1439,9 +1490,17 @@ qemuTranslateDiskSourcePool(virConnectPtr conn, } break; - case VIR_STORAGE_POOL_MPATH: - case VIR_STORAGE_POOL_RBD: case VIR_STORAGE_POOL_SHEEPDOG: + def->srcpool->actualtype = VIR_DOMAIN_DISK_TYPE_NETWORK; + // force direct mode + def->srcpool->mode = VIR_DOMAIN_DISK_SOURCE_POOL_MODE_DIRECT; + def->protocol = VIR_DOMAIN_DISK_PROTOCOL_SHEEPDOG; + def->src = virStorageVolGetPath(vol); + + qemuAddSheepPoolSourceHost(def, pooldef); + break; + case VIR_STORAGE_POOL_MPATH: + case VIR_STORAGE_POOL_RBD: case VIR_STORAGE_POOL_GLUSTER: case VIR_STORAGE_POOL_LAST: virReportError(VIR_ERR_CONFIG_UNSUPPORTED, -- 1.8.3.2

2 1

[libvirt] [PATCH v2] Block info query: Add check for transient domain
by John Ferlan 24 Jan '14

24 Jan '14

Currently the qemuDomainGetBlockInfo will return allocation == physical for most backing stores. For a qcow2 block backed device it's possible to return the highest lv extent allocated from qemu for an active guest. That is a value where allocation != physical and one would hope be less. However, if the guest is not running, then the code falls back to returning allocation == physical. This turns out to be problematic for rhev which monitors the size of the backing store. During a migration, before the VM has been started on the target and while it is deemed inactive on the source, there's a small window of time where the allocation is returned as physical triggering the code to extend the file unnecessarily. Since rhev uses transient domains and this is edge condition for a transient domain, rather than returning good status and allocation == physical when this "window of opportunity" exists, this patch will check for a transient (or non persistent) domain and return a failure to the caller rather than returning the defaults. For a persistent domain, the defaults will be returned. The description for the virDomainGetBlockInfo has been updated to describe the phenomena. Signed-off-by: John Ferlan <jferlan(a)redhat.com> --- The v1 of this patch is at: https://www.redhat.com/archives/libvir-list/2013-December/msg00984.html There is a bz associated as well: https://bugzilla.redhat.com/show_bug.cgi?id=1040507 I've read, reread, changed the text for the function so many times. I'm open to any suggestions for adjustments! It's not easy to describe the issues without getting into too much detail. include/libvirt/libvirt.h.in | 8 ++++--- src/libvirt.c | 53 ++++++++++++++++++++++++++++++++++++++++++++ src/qemu/qemu_driver.c | 29 +++++++++++++++++++----- 3 files changed, 82 insertions(+), 8 deletions(-) diff --git a/include/libvirt/libvirt.h.in b/include/libvirt/libvirt.h.in index 47a896b..b3ce000 100644 --- a/include/libvirt/libvirt.h.in +++ b/include/libvirt/libvirt.h.in @@ -2140,7 +2140,8 @@ int virDomainBlockResize (virDomainPtr dom, /** virDomainBlockInfo: * - * This struct provides information about the size of a block device backing store + * This struct provides information about the size of a block device + * backing store * * Examples: * @@ -2153,11 +2154,12 @@ int virDomainBlockResize (virDomainPtr dom, * * - qcow2 file in filesystem * * capacity: logical size from qcow2 header - * * allocation, physical: logical size of the file / highest qcow extent (identical) + * * allocation, physical: logical size of the file / + * highest qcow extent (identical) * * - qcow2 file in a block device * * capacity: logical size from qcow2 header - * * allocation: highest qcow extent written + * * allocation: highest qcow extent written for an active domain * * physical: size of the block device container */ typedef struct _virDomainBlockInfo virDomainBlockInfo; diff --git a/src/libvirt.c b/src/libvirt.c index c15e29a..9cc5b1c 100644 --- a/src/libvirt.c +++ b/src/libvirt.c @@ -8403,6 +8403,59 @@ error: * can be found by calling virDomainGetXMLDesc() and inspecting * elements within //domain/devices/disk. * + * For QEMU domains, the allocation and physical virDomainBlockInfo + * values returned will generally be the same, except when using a + * non raw, block backing device, such as qcow2 for an active domain. + * When the persistent domain is not active, QEMU will return the + * default which is the same value for allocation and physical. + * + * Active QEMU domains can return an allocation value which is more + * representative of the currently used blocks by the device compared + * to the physical size of the device. Applications can use/monitor + * the allocation value with the understanding that if the domain + * becomes inactive during an attempt to get the value, the default + * values will be returned. Thus, the application should check + * after the call for the domain being inactive if the values are + * the same. Optionally, the application could be watching for a + * shutdown event and then ignore any values received afterwards. + * This can be an issue when a domain is being migrated and the + * exact timing of the domain being made inactive and check of + * the allocation value results the default being returned. For + * a transient domain in the similar situation, this call will return + * -1 and an error message indicating the "domain is not running". + * + * The following is some pseudo code illustrating the call sequence: + * + * ... + * virDomainPtr dom; + * virDomainBlockInfo info; + * char *device; + * ... + * // Either get a list of all domains or a specific domain + * // via a virDomainLookupBy*() call. + * // + * // It's also required to fill in the device pointer, but that's + * // specific to the implementation. For the purposes of this example + * // a qcow2 backed device name string would need to be provided. + * ... + * // If the following call is made on a persistent domain with a + * // qcow2 block backed block device, then it's possible the returned + * // allocation equals the physical value. In that case, the domain + * // that may have been active prior to calling has become inactive, + * // such as is the case during a domain migration. Thus once we + * // get data returned, check for active domain when the values are + * // the same. + * if (virDomainGetBlockInfo(dom, device, &info, 0) < 0) + * goto failure; + * if (info.allocation == info.physical) { + * // If the domain is no longer active, + * // then the defaults are being returned. + * if (!virDomainIsActive()) + * goto ignore_return; + * } + * // Do something with the allocation and physical values + * ... + * * Returns 0 in case of success and -1 in case of failure. */ int diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 6e21267..699cbf5 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -10166,6 +10166,7 @@ qemuDomainGetBlockInfo(virDomainPtr dom, struct stat sb; int idx; int format; + int activeFail = false; virQEMUDriverConfigPtr cfg = NULL; char *alias = NULL; @@ -10265,15 +10266,23 @@ qemuDomainGetBlockInfo(virDomainPtr dom, /* Set default value .. */ info->allocation = info->physical; - /* ..but if guest is running & not using raw - disk format and on a block device, then query - highest allocated extent from QEMU */ + /* ..but if guest is not using raw disk format and on a block device, + * then query highest allocated extent from QEMU + */ if (disk->type == VIR_DOMAIN_DISK_TYPE_BLOCK && format != VIR_STORAGE_FILE_RAW && - S_ISBLK(sb.st_mode) && - virDomainObjIsActive(vm)) { + S_ISBLK(sb.st_mode)) { qemuDomainObjPrivatePtr priv = vm->privateData; + /* If the guest is not running, then success/failure return + * depends on whether domain is persistent + */ + if (!virDomainObjIsActive(vm)) { + activeFail = true; + ret = 0; + goto cleanup; + } + if (VIR_STRDUP(alias, disk->info.alias) < 0) goto cleanup; @@ -10287,6 +10296,7 @@ qemuDomainGetBlockInfo(virDomainPtr dom, &info->allocation); qemuDomainObjExitMonitor(driver, vm); } else { + activeFail = true; ret = 0; } @@ -10300,6 +10310,15 @@ cleanup: VIR_FREE(alias); virStorageFileFreeMetadata(meta); VIR_FORCE_CLOSE(fd); + + /* If we failed to get data from a domain because it's inactive and + * it's not a persisent domain, then force failure. + */ + if (activeFail && vm && !vm->persistent) { + virReportError(VIR_ERR_OPERATION_INVALID, "%s", + _("domain is not running")); + ret = -1; + } if (vm) virObjectUnlock(vm); virObjectUnref(cfg); -- 1.8.4.2

2 2

[libvirt] [PATCH] [PATCH]Correct syntax-check Using qemu with sheepdog pool
by joel SIMOES 24 Jan '14

24 Jan '14

From: Joel SIMOES <joel.simoes(a)laposte.net> --- src/qemu/qemu_conf.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c index dfafcdc..954542a 100644 --- a/src/qemu/qemu_conf.c +++ b/src/qemu/qemu_conf.c @@ -1215,17 +1215,13 @@ qemuAddSheepPoolSourceHost(virDomainDiskDefPtr def, 7000) < 0) goto cleanup; - - - - /* Storage pool have not supported these 2 attributes yet, * use the defaults. */ def->hosts[0].transport = VIR_DOMAIN_DISK_PROTO_TRANS_TCP; def->hosts[0].socket = NULL; - + def->protocol = VIR_DOMAIN_DISK_PROTOCOL_SHEEPDOG; ret = 0; @@ -1402,15 +1398,14 @@ qemuTranslateDiskSourcePool(virConnectPtr conn, def->srcpool->pooltype = pooldef->type; def->srcpool->voltype = info.type; - - if (def->srcpool->mode && pooldef->type != VIR_STORAGE_POOL_ISCSI && !(def->srcpool->mode == VIR_DOMAIN_DISK_SOURCE_POOL_MODE_DIRECT && pooldef->type == VIR_STORAGE_POOL_SHEEPDOG) ) { - + + if (def->srcpool->mode && pooldef->type != VIR_STORAGE_POOL_ISCSI && !(def->srcpool->mode == VIR_DOMAIN_DISK_SOURCE_POOL_MODE_DIRECT && pooldef->type == VIR_STORAGE_POOL_SHEEPDOG)) { virReportError(VIR_ERR_XML_ERROR, "%s", _("disk source mode is only valid when " "storage pool is of iscsi type or only direct for sheepdog ")); goto cleanup; } - + VIR_FREE(def->src); virDomainDiskHostDefFree(def->nhosts, def->hosts); virDomainDiskAuthClear(def); @@ -1496,11 +1491,10 @@ qemuTranslateDiskSourcePool(virConnectPtr conn, def->srcpool->mode = VIR_DOMAIN_DISK_SOURCE_POOL_MODE_DIRECT; def->protocol = VIR_DOMAIN_DISK_PROTOCOL_SHEEPDOG; def->src = virStorageVolGetPath(vol); - qemuAddSheepPoolSourceHost(def, pooldef); break; case VIR_STORAGE_POOL_MPATH: - case VIR_STORAGE_POOL_RBD: + case VIR_STORAGE_POOL_RBD: case VIR_STORAGE_POOL_GLUSTER: case VIR_STORAGE_POOL_LAST: virReportError(VIR_ERR_CONFIG_UNSUPPORTED, -- 1.8.3.2

2 1

[libvirt] [question]Is there a good way to get the mac of guest interface after I attached it to a guest
by Sheldon 24 Jan '14

24 Jan '14

Now I working on a new KVM management tool base on html5 and libvirt we will support a REST API for detach and attach guest interface. we want to use mac as the identifier of the network interface. That meas we should get the mac after the user create a interface. also other people may be doing the exact same thing at the exact same time to create a interface. Here are the demo codes: dom = self._get_vm(vm) xml = """ <interface type='network'> <source network='default'/> </interface> """ dom.attachDeviceFlags(xml, libvirt.VIR_DOMAIN_AFFECT_CURRENT) # now I want to get the mac, how to get it? I have check libvirt use this code to generate a mac address. does libvirt can gurantee that the MAC address generated is unique? void virMacAddrGenerate(const unsigned char prefix[VIR_MAC_PREFIX_BUFLEN], virMacAddrPtr addr) { addr->addr[0] = prefix[0]; addr->addr[1] = prefix[1]; addr->addr[2] = prefix[2]; addr->addr[3] = virRandomBits(8); addr->addr[4] = virRandomBits(8); addr->addr[5] = virRandomBits(8); } libvirt also use the KVM prefix as default as it's in the privately administered range: 52:54:00:XX:XX:XX -- Thanks and best regards! Sheldon Feng(冯少合)<shaohef(a)linux.vnet.ibm.com> IBM Linux Technology Center

3 7