July 2013 - Devel - libvirt List Archives

[libvirt] [PATCH RESEND 0/7] LXC: enhance user namespace support for adding the disk and host devices for libvirt lxc
by Gao feng 16 Jul '13

16 Jul '13

Libvirt lxc has supported user namespace since commit c34107dfd3a25232255e6d6f559b1306ef99bb3b,but for the disk devices and hostdevs, libvirt lxc still creates these device nodes in container. this will fail when container enables user namespace, since user namespace is disallowed to create device node. In order to reslove this problem, we should create device nodes on host side for container,and change the owner of these nodes to the root user of container. Gao feng (7): LXC: Setup disks for container on host side LXC: controller: change the owner of disk to the root of container LXC: Move virLXCControllerChown to lxc_container.c LXC: Change the owner of live attached disk device LXC: Create host devices for container on host side LXC: Change the owner of host devices to the root of container LXC: Change the owner of live attached host devices src/lxc/lxc_container.c | 382 +++------------------------------------- src/lxc/lxc_container.h | 2 + src/lxc/lxc_controller.c | 450 ++++++++++++++++++++++++++++++++++++++++++++--- src/lxc/lxc_driver.c | 13 ++ 4 files changed, 460 insertions(+), 387 deletions(-) -- 1.8.3.1

3 15

[libvirt] [PATCH] qemu: Fix double free of returned JSON array in qemuAgentGetVCPUs()
by Peter Krempa 16 Jul '13

16 Jul '13

A part of the returned monitor response was freed twice and caused crashes of the daemon when using guest agent cpu count retrieval. # virsh vcpucount dom --guest Introduced in v1.0.6-48-gc6afcb0 --- src/qemu/qemu_agent.c | 1 - 1 file changed, 1 deletion(-) diff --git a/src/qemu/qemu_agent.c b/src/qemu/qemu_agent.c index aca5ff3..72bf211 100644 --- a/src/qemu/qemu_agent.c +++ b/src/qemu/qemu_agent.c @@ -1529,7 +1529,6 @@ qemuAgentGetVCPUs(qemuAgentPtr mon, cleanup: virJSONValueFree(cmd); virJSONValueFree(reply); - virJSONValueFree(data); return ret; } -- 1.8.3.2

2 2

[libvirt] [PATCH] rbd: Do not free the secret if it is not set
by Wido den Hollander 16 Jul '13

16 Jul '13

Not all RBD (Ceph) storage pools have cephx authentication turned on, so "secret" might not be initialized. It could also be that the secret couldn't be located. Only call virSecretFree() if "secret" is initialized earlier. Signed-off-by: Wido den Hollander <wido(a)widodh.nl> --- src/storage/storage_backend_rbd.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/src/storage/storage_backend_rbd.c b/src/storage/storage_backend_rbd.c index 493e33b..1f75481 100644 --- a/src/storage/storage_backend_rbd.c +++ b/src/storage/storage_backend_rbd.c @@ -176,7 +176,11 @@ static int virStorageBackendRBDOpenRADOSConn(virStorageBackendRBDStatePtr *ptr, cleanup: VIR_FREE(secret_value); VIR_FREE(rados_key); - virSecretFree(secret); + + if (secret != NULL) { + virSecretFree(secret); + } + virBufferFreeAndReset(&mon_host); VIR_FREE(mon_buff); return ret; -- 1.7.9.5

2 1

[libvirt] [PATCH v3 00/10] Re-enable memballoon driver statistics reporting
by John Ferlan 16 Jul '13

16 Jul '13

This patch set replaces: https://www.redhat.com/archives/libvir-list/2013-July/msg00435.html Changes 01 -> No change 02 & 03 -> Adjust typecaste to switch() and remove "default:" 04 -> Use "unsigned int" as requested 05 -> Add ballooninit, remove get of period before set, remove setting of period during reconnect, other changes per review 06 -> Remove the opaque structure, replace with direct call through "qom-get" to return "guest-stats" and fill in stats[] directly 07 -> 09-> Remove the "flags", "Flags", "FLAGS" (learned something new!) 10 -> Add virsh.pod (hopefully have syntax right). NOTE: The formatdomain.html.in was already updated in 04 to describe the <stats period='#'/> XML syntax. This change describes the virsh command in order to set the period since this is the change where virsh code was adjusted. In 04, all one could do would be virsh edit John Ferlan (10): Add qemuMonitorJSONGetObjectListPaths() method for QMP qom-list command Add qemuMonitorJSONGetObjectProperty() method for QMP qom-get command Add qemuMonitorJSONSetObjectProperty() method for QMP qom-set command Add 'period' for Memballoon statistics gathering capability Determine whether to start balloon memory stats gathering. Add capability to fetch balloon stats Add new public API virDomainSetMemoryStatsPeriodFlags Specify remote protocol for virDomainSetMemoryStatsPeriodFlags Implement the virDomainSetMemoryStatsPeriodFlags for QEMU driver Allow balloon driver collection to be adjusted dynamically docs/formatdomain.html.in | 19 ++ docs/schemas/domaincommon.rng | 7 + include/libvirt/libvirt.h.in | 3 + src/conf/domain_conf.c | 38 +++- src/conf/domain_conf.h | 1 + src/driver.h | 6 + src/libvirt.c | 64 ++++++ src/libvirt_public.syms | 5 + src/qemu/qemu_driver.c | 65 ++++++ src/qemu/qemu_monitor.c | 140 ++++++++++++- src/qemu/qemu_monitor.h | 2 + src/qemu/qemu_monitor_json.c | 460 ++++++++++++++++++++++++++++++++---------- src/qemu/qemu_monitor_json.h | 60 ++++++ src/qemu/qemu_process.c | 14 +- src/remote/remote_driver.c | 1 + src/remote/remote_protocol.x | 15 +- src/remote_protocol-structs | 6 + tests/qemumonitorjsontest.c | 186 +++++++++++++++++ tools/virsh-domain-monitor.c | 66 +++++- tools/virsh.pod | 22 +- 20 files changed, 1060 insertions(+), 120 deletions(-) -- 1.8.1.4

4 38

[libvirt] [PATCH v2] qemu: Prevent crash of libvirtd without guest agent configuration
by Alex Jia 16 Jul '13

16 Jul '13

If users haven't configured guest agent then qemuAgentCommand() will dereference a NULL 'mon' pointer, which causes crash of libvirtd. With the patch, when the qemu-ga service isn't running in the guest, a expected error "error: Guest agent is not responding: Guest agent not available for now" will be raised, and the error "error: argument unsupported: QEMU guest agent is not configured" is raised when the guest hasn't configured guest agent. GDB backtrace: (gdb) bt #0 virNetServerFatalSignal (sig=11, siginfo=<value optimized out>, context=<value optimized out>) at rpc/virnetserver.c:326 #1 <signal handler called> #2 qemuAgentCommand (mon=0x0, cmd=0x7f39300017b0, reply=0x7f394b090910, seconds=-2) at qemu/qemu_agent.c:975 #3 0x00007f39429507f6 in qemuAgentGetVCPUs (mon=0x0, info=0x7f394b0909b8) at qemu/qemu_agent.c:1475 #4 0x00007f39429d9857 in qemuDomainGetVcpusFlags (dom=<value optimized out>, flags=9) at qemu/qemu_driver.c:4849 #5 0x00007f3957dffd8d in virDomainGetVcpusFlags (domain=0x7f39300009c0, flags=8) at libvirt.c:9843 How to reproduce? # To start a guest without guest agent configuration # then run the following cmdline # virsh vcpucount foobar --guest error: End of file while reading data: Input/output error error: One or more references were leaked after disconnect from the hypervisor error: Failed to reconnect to the hypervisor RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=984821 Signed-off-by: Alex Jia <ajia(a)redhat.com> --- src/qemu/qemu_driver.c | 26 ++++++++++++++++++++++++++ 1 files changed, 26 insertions(+), 0 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 495867a..699388c 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -4122,6 +4122,19 @@ qemuDomainSetVcpusFlags(virDomainPtr dom, unsigned int nvcpus, goto endjob; } + if (priv->agentError) { + virReportError(VIR_ERR_AGENT_UNRESPONSIVE, "%s", + _("QEMU guest agent is not " + "available due to an error")); + goto cleanup; + } + + if (!priv->agent) { + virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s", + _("QEMU guest agent is not configured")); + goto cleanup; + } + qemuDomainObjEnterAgent(vm); ncpuinfo = qemuAgentGetVCPUs(priv->agent, &cpuinfo); qemuDomainObjExitAgent(vm); @@ -4839,6 +4852,19 @@ qemuDomainGetVcpusFlags(virDomainPtr dom, unsigned int flags) if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_QUERY) < 0) goto cleanup; + if (priv->agentError) { + virReportError(VIR_ERR_AGENT_UNRESPONSIVE, "%s", + _("QEMU guest agent is not " + "available due to an error")); + goto cleanup; + } + + if (!priv->agent) { + virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s", + _("QEMU guest agent is not configured")); + goto cleanup; + } + if (!virDomainObjIsActive(vm)) { virReportError(VIR_ERR_OPERATION_INVALID, "%s", _("domain is not running")); -- 1.7.1

2 2

[libvirt] [PATCH v4 0/3] Support CHAP authentication for iscsi pool
by John Ferlan 16 Jul '13

16 Jul '13

https://bugzilla.redhat.com/show_bug.cgi?id=957294 Based on reviews of the the v2/v3 changes to the storage backend drivers regarding the call to perform the authentication, see (and followups): https://www.redhat.com/archives/libvir-list/2013-July/msg00910.html I moved the authentication of the pool back into the startPool callback instread of the findPoolSources() as was done for the original patches, see: https://www.redhat.com/archives/libvir-list/2013-May/msg01887.html https://www.redhat.com/archives/libvir-list/2013-May/msg01886.html Although the plain text option was removed. In order to achieve the goal of getting the secret, the startPool path needed a connection to a driver, so like the nwfilter_driver I chose a qemu connection. Deciding whether to use a privileged connection or not was made based on the privileged value set during storage driver state initialization. Maintaining that state allows for the decision further in the storageDriverAutostart() code to make the connection. Adjusted the existing rbd authentication to take advantage of this as well. >From what I see in just reading the code, this path would have sent a NULL 'conn' to the virStorageBackendRBDOpenRADOSConn() via the call from volStorageBackendRBDRefreshVolInfo() (eg, refreshPool()) and would have failed with a "failed to find the secret"). Whether this was a latent issue or not - I'm not quite sure. This code follows Osier's original change to be sure to call he secret driver 'secretGetValue' directly rather than through the virSecretGetValue() API. Using the same connection paradigm for the chap/iscsi authentication path in order to access the secret driver 'secretGetValue' directly rather than through the virSecretGetValue() API. John Ferlan (3): Add a privileged field to storageDriverState Adjust 'ceph' authentication secret usage for rbd pool. storage: Support "chap" authentication for iscsi pool src/conf/storage_conf.h | 1 + src/storage/storage_backend_iscsi.c | 111 +++++++++++++++++++++++++++++++++++- src/storage/storage_backend_rbd.c | 21 ++++++- src/storage/storage_driver.c | 19 ++++-- 4 files changed, 145 insertions(+), 7 deletions(-) -- 1.8.1.4

2 6

[libvirt] [v3 PATCH] storage: skip async-deleted volume while fs storage is being refreshed
by Guannan Ren 16 Jul '13

16 Jul '13

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=977706 v1, v2 by Ján Tomko https://www.redhat.com/archives/libvir-list/2013-July/msg00639.html To get volume file fd in the fist place. The purpose of doing so is to ensure the async deletion of volume doesn't make impact on the following operation for the volume till we recheck the existence of volume file later. If we don't get its fd firstly, it is diffcult to differentiate the non-existent file error caused by system or being deleted asynchronous. --- src/storage/storage_backend.c | 49 +++++++++++++++++----------------------- src/storage/storage_backend_fs.c | 6 +++++ 2 files changed, 27 insertions(+), 28 deletions(-) diff --git a/src/storage/storage_backend.c b/src/storage/storage_backend.c index 8d5880e..aa1635a 100644 --- a/src/storage/storage_backend.c +++ b/src/storage/storage_backend.c @@ -1093,29 +1093,13 @@ virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags) int fd, mode = 0; struct stat sb; char *base = last_component(path); - - if (lstat(path, &sb) < 0) { - virReportSystemError(errno, - _("cannot stat file '%s'"), - path); - return -1; - } - - if (S_ISFIFO(sb.st_mode)) { - VIR_WARN("ignoring FIFO '%s'", path); - return -2; - } else if (S_ISSOCK(sb.st_mode)) { - VIR_WARN("ignoring socket '%s'", path); - return -2; - } + int ret = -1; if ((fd = open(path, O_RDONLY|O_NONBLOCK|O_NOCTTY)) < 0) { - if ((errno == ENOENT || errno == ELOOP) && - S_ISLNK(sb.st_mode)) { - VIR_WARN("ignoring dangling symlink '%s'", path); + if (errno == ENOENT) { + VIR_WARN("volume '%s' does not exist", path); return -2; } - virReportSystemError(errno, _("cannot open volume '%s'"), path); @@ -1126,8 +1110,7 @@ virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags) virReportSystemError(errno, _("cannot stat file '%s'"), path); - VIR_FORCE_CLOSE(fd); - return -1; + goto error; } if (S_ISREG(sb.st_mode)) @@ -1141,26 +1124,36 @@ virStorageBackendVolOpenCheckMode(const char *path, unsigned int flags) if (STREQ(base, ".") || STREQ(base, "..")) { - VIR_FORCE_CLOSE(fd); VIR_INFO("Skipping special dir '%s'", base); - return -2; + goto skipfile; } + } else if (S_ISFIFO(sb.st_mode)) { + VIR_WARN("ignoring FIFO '%s'", path); + goto skipfile; + } else if (S_ISSOCK(sb.st_mode)) { + VIR_WARN("ignoring socket '%s'", path); + goto skipfile; } if (!(mode & flags)) { - VIR_FORCE_CLOSE(fd); - VIR_INFO("Skipping volume '%s'", path); - if (mode & VIR_STORAGE_VOL_OPEN_ERROR) { virReportError(VIR_ERR_INTERNAL_ERROR, _("unexpected storage mode for '%s'"), path); - return -1; + goto error; } - return -2; + VIR_INFO("Skipping volume '%s'", path); + goto skipfile; } return fd; + +skipfile: + ret = -2; + +error: + VIR_FORCE_CLOSE(fd); + return ret; } int virStorageBackendVolOpen(const char *path) diff --git a/src/storage/storage_backend_fs.c b/src/storage/storage_backend_fs.c index d305b06..d0276fc 100644 --- a/src/storage/storage_backend_fs.c +++ b/src/storage/storage_backend_fs.c @@ -889,6 +889,12 @@ virStorageBackendFileSystemRefresh(virConnectPtr conn ATTRIBUTE_UNUSED, } } + /* Recheck the existence of volume again */ + if (access(vol->target.path, F_OK) < 0) { + virStorageVolDefFree(vol); + vol = NULL; + continue; + } if (VIR_REALLOC_N(pool->volumes.objs, pool->volumes.count+1) < 0) -- 1.8.3.1

1 0

[libvirt] [PATCH v4] Make logical pools independent on target path
by Martin Kletzander 16 Jul '13

16 Jul '13

When using logical pools, we had to trust the target->path provided. This parameter, however, can be completely ommited and we can use '/dev/<source.name>' safely and populate it to target.path. Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=952973 Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com> --- Notes: v4: - Don't add anything complicated, just rely on /dev/VG_NAME/LV_NAME v3: - just rebase v2: - don't drop pool's target.path, but fix it instead to /dev/VG_NAME docs/schemas/storagepool.rng | 13 ++++++++++++- src/conf/storage_conf.c | 17 ++++++++++++----- src/storage/storage_backend_logical.c | 22 +++------------------- src/storage/storage_driver.c | 2 +- tests/storagepoolxml2xmlin/pool-logical-create.xml | 2 +- tests/storagepoolxml2xmlin/pool-logical-nopath.xml | 19 +++++++++++++++++++ .../storagepoolxml2xmlout/pool-logical-nopath.xml | 22 ++++++++++++++++++++++ tests/storagepoolxml2xmltest.c | 1 + 8 files changed, 71 insertions(+), 27 deletions(-) create mode 100644 tests/storagepoolxml2xmlin/pool-logical-nopath.xml create mode 100644 tests/storagepoolxml2xmlout/pool-logical-nopath.xml diff --git a/docs/schemas/storagepool.rng b/docs/schemas/storagepool.rng index 3c2158a..1b3f4bc 100644 --- a/docs/schemas/storagepool.rng +++ b/docs/schemas/storagepool.rng @@ -62,7 +62,7 @@ <ref name='commonmetadata'/> <ref name='sizing'/> <ref name='sourcelogical'/> - <ref name='target'/> + <ref name='targetlogical'/> </define> <define name='pooldisk'> @@ -207,6 +207,17 @@ </element> </define> + <define name='targetlogical'> + <element name='target'> + <optional> + <element name='path'> + <ref name='absFilePath'/> + </element> + </optional> + <ref name='permissions'/> + </element> + </define> + <define name='sourceinfohost'> <oneOrMore> <element name='host'> diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c index 524a4d6..a517cbf 100644 --- a/src/conf/storage_conf.c +++ b/src/conf/storage_conf.c @@ -1,7 +1,7 @@ /* * storage_conf.c: config handling for storage driver * - * Copyright (C) 2006-2012 Red Hat, Inc. + * Copyright (C) 2006-2013 Red Hat, Inc. * Copyright (C) 2006-2008 Daniel P. Berrange * * This library is free software; you can redistribute it and/or @@ -953,10 +953,17 @@ virStoragePoolDefParseXML(xmlXPathContextPtr ctxt) /* When we are working with a virtual disk we can skip the target * path and permissions */ if (!(options->flags & VIR_STORAGE_POOL_SOURCE_NETWORK)) { - if (!(target_path = virXPathString("string(./target/path)", ctxt))) { - virReportError(VIR_ERR_XML_ERROR, "%s", - _("missing storage pool target path")); - goto error; + if (ret->type == VIR_STORAGE_POOL_LOGICAL) { + if (virAsprintf(&target_path, "/dev/%s", ret->source.name) < 0) { + goto error; + } + } else { + target_path = virXPathString("string(./target/path)", ctxt); + if (!target_path) { + virReportError(VIR_ERR_XML_ERROR, "%s", + _("missing storage pool target path")); + goto error; + } } ret->target.path = virFileSanitizePath(target_path); if (!ret->target.path) diff --git a/src/storage/storage_backend_logical.c b/src/storage/storage_backend_logical.c index 416a042..8998a11 100644 --- a/src/storage/storage_backend_logical.c +++ b/src/storage/storage_backend_logical.c @@ -454,17 +454,7 @@ virStorageBackendLogicalCheckPool(virConnectPtr conn ATTRIBUTE_UNUSED, virStoragePoolObjPtr pool, bool *isActive) { - char *path; - - *isActive = false; - if (virAsprintf(&path, "/dev/%s", pool->def->source.name) < 0) - return -1; - - if (access(path, F_OK) == 0) - *isActive = true; - - VIR_FREE(path); - + *isActive = (access(pool->def->target.path, F_OK) == 0); return 0; } @@ -794,21 +784,16 @@ virStorageBackendLogicalDeleteVol(virConnectPtr conn ATTRIBUTE_UNUSED, unsigned int flags) { int ret = -1; - char *volpath = NULL; virCommandPtr lvchange_cmd = NULL; virCommandPtr lvremove_cmd = NULL; virCheckFlags(0, -1); - if (virAsprintf(&volpath, "%s/%s", - pool->def->source.name, vol->name) < 0) - goto cleanup; - virFileWaitForDevices(); - lvchange_cmd = virCommandNewArgList(LVCHANGE, "-aln", volpath, NULL); - lvremove_cmd = virCommandNewArgList(LVREMOVE, "-f", volpath, NULL); + lvchange_cmd = virCommandNewArgList(LVCHANGE, "-aln", vol->target.path, NULL); + lvremove_cmd = virCommandNewArgList(LVREMOVE, "-f", vol->target.path, NULL); if (virCommandRun(lvremove_cmd, NULL) < 0) { if (virCommandRun(lvchange_cmd, NULL) < 0) { @@ -821,7 +806,6 @@ virStorageBackendLogicalDeleteVol(virConnectPtr conn ATTRIBUTE_UNUSED, ret = 0; cleanup: - VIR_FREE(volpath); virCommandFree(lvchange_cmd); virCommandFree(lvremove_cmd); return ret; diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c index a8eb731..43bf6de 100644 --- a/src/storage/storage_driver.c +++ b/src/storage/storage_driver.c @@ -1,7 +1,7 @@ /* * storage_driver.c: core driver for storage APIs * - * Copyright (C) 2006-2012 Red Hat, Inc. + * Copyright (C) 2006-2013 Red Hat, Inc. * Copyright (C) 2006-2008 Daniel P. Berrange * * This library is free software; you can redistribute it and/or diff --git a/tests/storagepoolxml2xmlin/pool-logical-create.xml b/tests/storagepoolxml2xmlin/pool-logical-create.xml index 4c67089..fd551e0 100644 --- a/tests/storagepoolxml2xmlin/pool-logical-create.xml +++ b/tests/storagepoolxml2xmlin/pool-logical-create.xml @@ -10,7 +10,7 @@ <device path="/dev/sdb3"/> </source> <target> - <path>/dev/HostVG</path> + <path>/invalid/nonexistent/path</path> <permissions> <mode>0700</mode> <owner>0</owner> diff --git a/tests/storagepoolxml2xmlin/pool-logical-nopath.xml b/tests/storagepoolxml2xmlin/pool-logical-nopath.xml new file mode 100644 index 0000000..e1bb4db --- /dev/null +++ b/tests/storagepoolxml2xmlin/pool-logical-nopath.xml @@ -0,0 +1,19 @@ +<pool type='logical'> + <name>HostVG</name> + <uuid>1c13165a-d0f4-3aee-b447-30fb38789091</uuid> + <capacity>99891544064</capacity> + <allocation>99220455424</allocation> + <available>671088640</available> + <source> + <device path="/dev/sdb1"/> + <device path="/dev/sdb2"/> + <device path="/dev/sdb3"/> + </source> + <target> + <permissions> + <mode>0700</mode> + <owner>0</owner> + <group>0</group> + </permissions> + </target> +</pool> diff --git a/tests/storagepoolxml2xmlout/pool-logical-nopath.xml b/tests/storagepoolxml2xmlout/pool-logical-nopath.xml new file mode 100644 index 0000000..7413f1c --- /dev/null +++ b/tests/storagepoolxml2xmlout/pool-logical-nopath.xml @@ -0,0 +1,22 @@ +<pool type='logical'> + <name>HostVG</name> + <uuid>1c13165a-d0f4-3aee-b447-30fb38789091</uuid> + <capacity unit='bytes'>0</capacity> + <allocation unit='bytes'>0</allocation> + <available unit='bytes'>0</available> + <source> + <device path='/dev/sdb1'/> + <device path='/dev/sdb2'/> + <device path='/dev/sdb3'/> + <name>HostVG</name> + <format type='lvm2'/> + </source> + <target> + <path>/dev/HostVG</path> + <permissions> + <mode>0700</mode> + <owner>0</owner> + <group>0</group> + </permissions> + </target> +</pool> diff --git a/tests/storagepoolxml2xmltest.c b/tests/storagepoolxml2xmltest.c index 53a7f83..d59cff9 100644 --- a/tests/storagepoolxml2xmltest.c +++ b/tests/storagepoolxml2xmltest.c @@ -87,6 +87,7 @@ mymain(void) DO_TEST("pool-dir"); DO_TEST("pool-fs"); DO_TEST("pool-logical"); + DO_TEST("pool-logical-nopath"); DO_TEST("pool-logical-create"); DO_TEST("pool-disk"); DO_TEST("pool-iscsi"); -- 1.8.3.2

2 2

[libvirt] [PATCH v5 0/5] Chardev hotplug
by Michal Privoznik 16 Jul '13

16 Jul '13

I've pushed all ACKed patches from previous rounds. So now just the rest. Michal Privoznik (5): qemuBuildChrDeviceCommandLine: Don't leak devstr domain_conf: Auto fill chardev port qemu: Implement chardev hotplug on config level qemu: Implement chardev hotplug on live level qemuhotplugtest: Introduce test for chardev hotplug src/conf/domain_conf.c | 52 +++++- src/conf/domain_conf.h | 3 + src/libvirt_private.syms | 1 + src/qemu/qemu_command.c | 39 ++++- src/qemu/qemu_driver.c | 42 ++++- src/qemu/qemu_hotplug.c | 175 +++++++++++++++++++ src/qemu/qemu_hotplug.h | 13 ++ tests/qemuhotplugtest.c | 194 +++++++++++++++++---- .../qemuhotplug-console-virtio.xml | 5 + .../qemuxml2argv-console-compat-2.xml | 122 +++++++++++++ 10 files changed, 597 insertions(+), 49 deletions(-) create mode 100644 tests/qemuhotplugtestdata/qemuhotplug-console-virtio.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-console-compat-2.xml -- 1.8.1.5

3 12

[libvirt] [PATCH] qemu: Prevent crash of libvirtd without guest agent configuration
by Alex Jia 16 Jul '13

16 Jul '13

If users haven't configured guest agent then qemuAgentCommand() will dereference a NULL 'mon' pointer, which causes crash of libvirtd. GDB backtrace: (gdb) bt #0 virNetServerFatalSignal (sig=11, siginfo=<value optimized out>, context=<value optimized out>) at rpc/virnetserver.c:326 #1 <signal handler called> #2 qemuAgentCommand (mon=0x0, cmd=0x7f39300017b0, reply=0x7f394b090910, seconds=-2) at qemu/qemu_agent.c:975 #3 0x00007f39429507f6 in qemuAgentGetVCPUs (mon=0x0, info=0x7f394b0909b8) at qemu/qemu_agent.c:1475 #4 0x00007f39429d9857 in qemuDomainGetVcpusFlags (dom=<value optimized out>, flags=9) at qemu/qemu_driver.c:4849 #5 0x00007f3957dffd8d in virDomainGetVcpusFlags (domain=0x7f39300009c0, flags=8) at libvirt.c:9843 How to reproduce? # To start a guest without guest agent configuration # then run the following cmdline # virsh vcpucount foobar --guest error: End of file while reading data: Input/output error error: One or more references were leaked after disconnect from the hypervisor error: Failed to reconnect to the hypervisor RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=984821 Signed-off-by: Alex Jia <ajia(a)redhat.com> --- src/qemu/qemu_driver.c | 12 ++++++++++++ 1 files changed, 12 insertions(+), 0 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 495867a..1448aa7 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -4122,6 +4122,12 @@ qemuDomainSetVcpusFlags(virDomainPtr dom, unsigned int nvcpus, goto endjob; } + if (!priv->agent) { + virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s", + _("QEMU guest agent is not configured")); + goto cleanup; + } + qemuDomainObjEnterAgent(vm); ncpuinfo = qemuAgentGetVCPUs(priv->agent, &cpuinfo); qemuDomainObjExitAgent(vm); @@ -4839,6 +4845,12 @@ qemuDomainGetVcpusFlags(virDomainPtr dom, unsigned int flags) if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_QUERY) < 0) goto cleanup; + if (!priv->agent) { + virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s", + _("QEMU guest agent is not configured")); + goto cleanup; + } + if (!virDomainObjIsActive(vm)) { virReportError(VIR_ERR_OPERATION_INVALID, "%s", _("domain is not running")); -- 1.7.1

2 2