- Devel - libvirt List Archives

[libvirt] Using dlls for Windows provided in http://libvirt.org/sources/win32_experimental/Libvirt-0.8.7-2.exe
by Schnizer, Monika 24 Feb '11

24 Feb '11

Hello libvirt-users, could anybody tell me, whether we could use dlls provided in the experimental package for windows, and distribute it with our application. We would do the following: a) Install the executable from http://libvirt.org/sources/win32_experimental/Libvirt-0.8.7-2.exe on our production machine. b) Add the dlls to the setup of our application c) Install the dlls on the customer's system in some location named /<product root>/lib. Questions: - do we violate the LGPL by doing so? - does anybody have better suggestions on how to provide the functionality to the customer? - who could tell me how to contact the copyright owners, in order to directly discuss this issue with them. Let me thank in advance for any any information. With best regards, -- Monika Schnizer Software Development FTS TSP x86 E SW4 FUJITSU Fujitsu TECHNOLOGY SOLUTIONS GmbH Domagk-Str.28, 80807 Munich, Germany Tel: +49 (89) 3222 2287 Fax: +49 (89) 3222 329 2287 Email: Monika.Schnizer(a)ts.fujitsu.com Web: http://ts.fujitsu.com<http://ts.fujitsu.com/> Company Details: de.ts.fujitsu.com/imprint.html

2 4

[libvirt] [PATCH] Make commandtest more robust wrt its execution environment
by Daniel P. Berrange 24 Feb '11

24 Feb '11

When executed from cron, commandtest would failure to correctly identified daemon processes. Set session ID and process group IDs at startup to ensure we have a consistent environment to run in. * tests/commandtest.c: Call setsid() and setpgid() --- tests/commandtest.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/tests/commandtest.c b/tests/commandtest.c index 7157c51..dc2f8a1 100644 --- a/tests/commandtest.c +++ b/tests/commandtest.c @@ -730,6 +730,9 @@ mymain(int argc, char **argv) if (chdir("/tmp") < 0) return(EXIT_FAILURE); + setpgid(0, 0); + setsid(); + /* Kill off any inherited fds that might interfere with our * testing. */ fd = 3; -- 1.7.4

2 1

[libvirt] [Xen-devel] Re: [PATCH V2] Add libxenlight driver
by Ian Jackson 24 Feb '11

24 Feb '11

Paolo Bonzini writes ("[Xen-devel] Re: [PATCH V2] Add libxenlight driver"): > On 02/23/2011 03:56 AM, Jim Fehlig wrote: > > Add a new xen driver based on libxenlight [1], which is the primary > > toolstack starting with Xen 4.1.0. The driver is stateful, runs > > privileged only, and is accessed with libxl:/// URI. > > Why not let the user keep xen:/// ? Because for now you want to be able to have both drivers. Unless, perhaps, it's somehow possible to select between them at runtime for the same URI scheme ? Ian.

3 2

[libvirt] [PATCH v2 0/7] Switch over qemu command line capabilities to virBitmap
by Jiri Denemark 24 Feb '11

24 Feb '11

There are two main reasons behind this patchset: - we are getting very close to 64 flags which is the maximum we can use with unsigned long long - by using LL constants in enum we already violates C99 constraint that enum values have to fit into int Patches 3/7 and 5/7 are fully and 4/7 mostly mechanical. 6/7 and 7/7 need to be squashed into a single patch but are provided separately for easier review. Changes in version 2 are described in each patch separately. Jiri Denemark (7): util: Use unsigned long as a base type for virBitmap util: Add API for converting virBitmap into printable string qemu: Rename QEMUD_CMD_FLAG_* to QEMU_CAPS_* qemu: Use helper functions for handling cmd line capabilities qemu: Rename qemud\?CmdFlags to qemuCaps qemu: Switch over command line capabilities to virBitmap tests: Use virBitmap for qemu command line caps cfg.mk | 3 +- src/libvirt_private.syms | 1 + src/qemu/qemu_capabilities.c | 238 +++++++++----- src/qemu/qemu_capabilities.h | 152 +++++---- src/qemu/qemu_command.c | 392 +++++++++++----------- src/qemu/qemu_command.h | 30 +- src/qemu/qemu_driver.c | 48 ++-- src/qemu/qemu_hotplug.c | 162 +++++----- src/qemu/qemu_hotplug.h | 32 +- src/qemu/qemu_migration.c | 22 +- src/qemu/qemu_process.c | 29 +- src/util/bitmap.c | 51 +++- src/util/bitmap.h | 3 + tests/qemuargv2xmltest.c | 2 +- tests/qemuhelptest.c | 747 +++++++++++++++++++++--------------------- tests/qemuxml2argvtest.c | 474 +++++++++++++------------- 16 files changed, 1263 insertions(+), 1123 deletions(-) -- 1.7.4.1

2 21

[libvirt] [PATCH 1/2] libvirt/qemu - support updating inactive domains (take 2)
by KAMEZAWA Hiroyuki 24 Feb '11

24 Feb '11

>From d6c65102224c493f67008b6521b0115d798f5a67 Mon Sep 17 00:00:00 2001 From: KAMEZAWA Hiroyuki <kamezawa(a)bluextal.(none)> Date: Thu, 24 Feb 2011 12:14:25 +0900 Subject: [PATCH 1/2] libvirt/qemu - support updating inactive domains. Now, virsh attach-disk/detach-disk has --persistent option and it can update XML definition of inactive domains. But, it's only supported in Xen. This patch adds support for attach-disk/detach-disk for qemu. Changelog v1->v2: - fixed TABs - fixed PCI address assignment for VIRTIO driver. Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu(a)jp.fujitsu.com> --- src/qemu/qemu_driver.c | 168 +++++++++++++++++++++++++++++++++++++++++++++--- 1 files changed, 158 insertions(+), 10 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 8b15a3e..e75478a 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -4085,14 +4085,164 @@ cleanup: return ret; } -static int qemudDomainAttachDeviceFlags(virDomainPtr dom, +static int qemuDomainFindDiskByName(virDomainDefPtr vmdef, const char *name) +{ + virDomainDiskDefPtr vdisk; + int i; + + for (i = 0; i < vmdef->ndisks; i++) { + vdisk = vmdef->disks[i]; + if (!strcmp(vdisk->dst, name)) + return i; + } + return -1; +} +/* + * Attach a device given by XML, the change will be persistent + * and domain XML definition file is updated. + */ +static int qemuDomainAttachDevicePersistent(virDomainDefPtr vmdef, + virDomainDeviceDefPtr newdev) +{ + virDomainDiskDefPtr disk; + + /* At first, check device confliction */ + switch(newdev->type) { + case VIR_DOMAIN_DEVICE_DISK: + disk = newdev->data.disk; + if (qemuDomainFindDiskByName(vmdef, disk->dst) >= 0) { + qemuReportError(VIR_ERR_INVALID_ARG, + _("target %s already exists."), disk->dst); + return -1; + } + + if (virDomainDiskInsert(vmdef, disk)) { + virReportOOMError(); + return -1; + } + if (disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) { + /* + * Address/Drive information is NULL. If virtio, PCI address + * shoule be fixed. Other devices as IDE, SCSI...will get ID + * automatically + */ + qemuDomainAssignPCIAddresses(vmdef); + + } + newdev->data.disk = NULL; + break; + default: + qemuReportError(VIR_ERR_INVALID_ARG, "%s", + _("Sorry, the device is not suppored for now")); + return -1; + } + + return 0; +} + +static int qemuDomainDetachDevicePersistent(virDomainDefPtr vmdef, + virDomainDeviceDefPtr device) +{ + int x; + virDomainDiskDefPtr disk; + + switch(device->type) { + case VIR_DOMAIN_DEVICE_DISK: + disk = device->data.disk; + x = qemuDomainFindDiskByName(vmdef, disk->dst); + if (x < 0) { + qemuReportError(VIR_ERR_INVALID_ARG, + _("target %s doesn't exist."), disk->dst); + return -1; + } + virDomainDiskRemove(vmdef, x); + break; + default: + qemuReportError(VIR_ERR_INVALID_ARG, "%s", + _("Sorry, the device is not suppored for now")); + return -1; + } + return 0; +} + +static int qemuDomainModifyDevicePersistent(virDomainPtr dom, const char *xml, - unsigned int flags) { - if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) { - qemuReportError(VIR_ERR_OPERATION_INVALID, - "%s", _("cannot modify the persistent configuration of a domain")); + unsigned int attach) +{ + struct qemud_driver *driver; + virDomainDeviceDefPtr device; + virDomainDefPtr vmdef; + virDomainObjPtr vm; + int ret = -1; + + if (!dom || !dom->conn || !dom->name || !xml) { + qemuReportError(VIR_ERR_INVALID_ARG, + _("internal error : %s"), __FUNCTION__); + return -1; + } + + if (dom->conn->flags & VIR_CONNECT_RO) return -1; + + driver = dom->conn->privateData; + qemuDriverLock(driver); + vm = virDomainFindByUUID(&driver->domains, dom->uuid); + if (!vm) { + qemuReportError(VIR_ERR_NO_DOMAIN, _("cannot find domain '%s'"), + dom->name); + goto unlock_out; + } + + if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) { + /* + * For now, just allow updating inactive domains. Further development + * will allow updating both active domain and its config file at + * the same time. + */ + qemuReportError(VIR_ERR_INVALID_ARG, + _("cannot update alive domain : %s"), __FUNCTION__); + goto endjob; + } + vmdef = virDomainObjGetPersistentDef(driver->caps, vm); + + if (!vmdef) + goto endjob; + + device = virDomainDeviceDefParse(driver->caps, + vmdef, xml, VIR_DOMAIN_XML_INACTIVE); + if (!device) + goto endjob; + + if (attach) { + ret = qemuDomainAttachDevicePersistent(vmdef, device); + if (ret < 0) + goto out; + } else { + ret = qemuDomainDetachDevicePersistent(vmdef, device); + if (ret < 0) + goto out; } + ret = virDomainSaveConfig(driver->configDir, vmdef); + +out: + virDomainDeviceDefFree(device); +endjob: + if (qemuDomainObjEndJob(vm) == 0) + vm = NULL; + if (vm) + virDomainObjUnlock(vm); + /* Note: insert of newdev is done by copy */ +unlock_out: + qemuDriverUnlock(driver); + return ret; +} + +static int qemudDomainAttachDeviceFlags(virDomainPtr dom, + const char *xml, + unsigned int flags) { + + if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) + return qemuDomainModifyDevicePersistent(dom, xml, 1); return qemudDomainAttachDevice(dom, xml); } @@ -4306,11 +4456,9 @@ cleanup: static int qemudDomainDetachDeviceFlags(virDomainPtr dom, const char *xml, unsigned int flags) { - if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) { - qemuReportError(VIR_ERR_OPERATION_INVALID, - "%s", _("cannot modify the persistent configuration of a domain")); - return -1; - } + + if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) + return qemuDomainModifyDevicePersistent(dom, xml, 0); return qemudDomainDetachDevice(dom, xml); } -- 1.7.1

2 8

[libvirt] [PATCH 2/2] Ignore backing file errors in FS storage pool (v2)
by Philipp Hahn 24 Feb '11

24 Feb '11

Currently a single storage volume with a broken backing file will disable the whole storage pool. This can happen when the backing file is on some unavailable network storage or if the backing volume is deleted, while the storage volumes using it are not. Since the storage pool then can not be re-activated, re-creating the missing or deleting the now useless volumes using libvirt only is impossible. To "fix" this case, all errors detected during storage pool activation are now (silently) ignored. Errors are still logged by the called functions, which have more knowledge on the detailed error condition. To reproduce: dir=$(mktemp -d) virsh pool-create-as tmp dir '' '' '' '' "$dir" virsh vol-create-as --format qcow2 tmp back 1G virsh vol-create-as --format qcow2 --backing-vol-format qcow2 --backing-vol back tmp cow 1G virsh vol-delete --pool tmp back virsh pool-refresh tmp After the last step, the pool will be gone (because it was not persistent). As long as the now broken image stays in the directory, you will not be able to re-create or re-start the pool. Signed-off-by: Philipp Hahn <hahn(a)univention.de> --- src/storage/storage_backend_fs.c | 10 +++++----- 1 files changed, 5 insertions(+), 5 deletions(-)

1 0

[libvirt] [PATCH 2/2] Ignore backing file errors in FS storage pool
by Philipp Hahn 24 Feb '11

24 Feb '11

Currently a single storage volume with a broken backing file will disable the whole storage pool. This can happen when the backing file is on some unavailable network storage or if the backing volume is deleted, while the storage volumes using it are not. Since the storage pool then can not be re-activated, re-creating the missing or deleting the now useless volumes using libvirt only is impossible. To "fix" this case, all errors detected during storage pool activation are now (silently) ignored. Errors are still logged by the called functions, which have more knowledge on the detailed error condition. To reproduce: dir=$(mktemp -d) virsh pool-create-as tmp dir '' '' '' '' "$dir" virsh vol-create-as --format qcow2 tmp back 1G virsh vol-create-as --format qcow2 --backing-vol-format qcow2 --backing-vol back tmp cow 1G virsh vol-delete --pool tmp back virsh pool-refresh tmp After the last step, the pool will be gone (because it was not persistent). As long as the now broken image stays in the directory, you will not be able to re-create or re-start the pool. Signed-off-by: Philipp Hahn <hahn(a)univention.de> --- src/storage/storage_backend_fs.c | 10 +++++----- 1 files changed, 5 insertions(+), 5 deletions(-)

1 0

[libvirt] Invitation to connect on LinkedIn
by Lyre Li 24 Feb '11

24 Feb '11

LinkedIn ------------ I'd like to add you to my professional network on LinkedIn. - Lyre Lyre Li Engineer at Skybility China Confirm that you know Lyre Li https://www.linkedin.com/e/k4zv0-gkj412ae-3z/isd/2389916274/QzDJLaNZ/ -- (c) 2011, LinkedIn Corporation

1 0

[libvirt] [PATCH] Add APIs for killing off processes inside a cgroup
by Daniel P. Berrange 24 Feb '11

24 Feb '11

The virCgroupKill method kills all PIDs found in a cgroup The virCgroupKillRecursively method does this recursively for child cgroups. The virCgroupKillPainfully method does a recursive kill several times in a row until everything has really died --- src/libvirt_private.syms | 3 + src/util/cgroup.c | 229 ++++++++++++++++++++++++++++++++++++++++++++++ src/util/cgroup.h | 4 + 3 files changed, 236 insertions(+), 0 deletions(-) diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms index 797a670..989b3eb 100644 --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -74,6 +74,9 @@ virCgroupGetMemoryHardLimit; virCgroupGetMemorySoftLimit; virCgroupGetMemoryUsage; virCgroupGetSwapHardLimit; +virCgroupKill; +virCgroupKillRecursive; +virCgroupKillPainfully; virCgroupMounted; virCgroupRemove; virCgroupSetBlkioWeight; diff --git a/src/util/cgroup.c b/src/util/cgroup.c index b71eef9..2c7efb4 100644 --- a/src/util/cgroup.c +++ b/src/util/cgroup.c @@ -23,6 +23,7 @@ #include <stdbool.h> #include <sys/stat.h> #include <sys/types.h> +#include <signal.h> #include <libgen.h> #include <dirent.h> @@ -32,6 +33,7 @@ #include "cgroup.h" #include "logging.h" #include "files.h" +#include "hash.h" #define CGROUP_MAX_VAL 512 @@ -258,6 +260,19 @@ static int virCgroupPathOfController(virCgroupPtr group, const char *key, char **path) { + if (controller == -1) { + int i; + for (i = 0 ; i < VIR_CGROUP_CONTROLLER_LAST ; i++) { + if (group->controllers[i].mountPoint && + group->controllers[i].placement) { + controller = i; + break; + } + } + } + if (controller == -1) + return -ENOSYS; + if (group->controllers[controller].mountPoint == NULL) return -ENOENT; @@ -1291,3 +1306,217 @@ int virCgroupGetFreezerState(virCgroupPtr group, char **state) VIR_CGROUP_CONTROLLER_CPU, "freezer.state", state); } + +static int virCgroupKillInternal(virCgroupPtr group, int signum, virHashTablePtr pids) +{ + int rc; + int killedAny = 0; + char *keypath = NULL; + bool done = false; + VIR_DEBUG("group=%p path=%s signum=%d pids=%p", group, group->path, signum, pids); + + rc = virCgroupPathOfController(group, -1, "tasks", &keypath); + if (rc != 0) { + VIR_DEBUG("No path of %s, tasks", group->path); + return rc; + } + + /* PIDs may be forking as we kill them, so loop + * until there are no new PIDs found + */ + while (!done) { + done = true; + FILE *fp; + if (!(fp = fopen(keypath, "r"))) { + rc = -errno; + VIR_DEBUG("Failed to read %s: %m\n", keypath); + goto cleanup; + } else { + while (!feof(fp)) { + unsigned long pid; + if (fscanf(fp, "%lu", &pid) != 1) { + if (feof(fp)) + break; + rc = -errno; + break; + } + if (virHashLookup(pids, (void*)pid)) + continue; + + VIR_DEBUG("pid=%lu", pid); + if (kill((pid_t)pid, signum) < 0) { + if (errno != ESRCH) { + rc = -errno; + goto cleanup; + } + /* Leave RC == 0 since we didn't kill one */ + } else { + killedAny = 1; + done = false; + } + + virHashAddEntry(pids, (void*)pid, (void*)1); + } + fclose(fp); + } + } + + rc = killedAny ? 1 : 0; + +cleanup: + VIR_FREE(keypath); + + return rc; +} + + +static unsigned long virCgroupPidCode(const void *name) +{ + return (unsigned long)name; +} +static bool virCgroupPidEqual(const void *namea, const void *nameb) +{ + return namea == nameb; +} +static void *virCgroupPidCopy(const void *name) +{ + return (void*)name; +} + +/* + * Returns + * < 0 : errno that occurred + * 0 : no PIDs killed + * 1 : at least one PID killed + */ +int virCgroupKill(virCgroupPtr group, int signum) +{ + VIR_DEBUG("group=%p path=%s signum=%d", group, group->path, signum); + int rc; + /* The 'tasks' file in cgroups can contain duplicated + * pids, so we use a hash to track which we've already + * killed. + */ + virHashTablePtr pids = virHashCreateFull(100, + NULL, + virCgroupPidCode, + virCgroupPidEqual, + virCgroupPidCopy, + NULL); + + rc = virCgroupKillInternal(group, signum, pids); + + virHashFree(pids); + + return rc; +} + + +static int virCgroupKillRecursiveInternal(virCgroupPtr group, int signum, virHashTablePtr pids, bool dormdir) +{ + int rc; + int killedAny = 0; + char *keypath = NULL; + DIR *dp; + virCgroupPtr subgroup = NULL; + struct dirent *ent; + VIR_DEBUG("group=%p path=%s signum=%d pids=%p", group, group->path, signum, pids); + + rc = virCgroupPathOfController(group, -1, "", &keypath); + if (rc != 0) { + VIR_DEBUG("No path of %s, tasks", group->path); + return rc; + } + + if ((rc = virCgroupKillInternal(group, signum, pids)) != 0) + return rc; + + VIR_DEBUG("Iterate over children of %s", keypath); + if (!(dp = opendir(keypath))) { + rc = -errno; + return rc; + } + + while ((ent = readdir(dp))) { + char *subpath; + + if (STREQ(ent->d_name, ".")) + continue; + if (STREQ(ent->d_name, "..")) + continue; + if (ent->d_type != DT_DIR) + continue; + + VIR_DEBUG("Process subdir %s", ent->d_name); + if (virAsprintf(&subpath, "%s/%s", group->path, ent->d_name) < 0) { + rc = -ENOMEM; + goto cleanup; + } + + if ((rc = virCgroupNew(subpath, &subgroup)) != 0) + goto cleanup; + + if ((rc = virCgroupKillRecursiveInternal(subgroup, signum, pids, true)) < 0) + goto cleanup; + if (rc == 1) + killedAny = 1; + + if (dormdir) + virCgroupRemove(subgroup); + + virCgroupFree(&subgroup); + } + + rc = killedAny; + +cleanup: + virCgroupFree(&subgroup); + closedir(dp); + + return rc; +} + +int virCgroupKillRecursive(virCgroupPtr group, int signum) +{ + int rc; + VIR_DEBUG("group=%p path=%s signum=%d", group, group->path, signum); + virHashTablePtr pids = virHashCreateFull(100, + NULL, + virCgroupPidCode, + virCgroupPidEqual, + virCgroupPidCopy, + NULL); + + rc = virCgroupKillRecursiveInternal(group, signum, pids, false); + + virHashFree(pids); + + return rc; +} + + +int virCgroupKillPainfully(virCgroupPtr group) +{ + int i; + int rc; + VIR_DEBUG("cgroup=%p path=%s", group, group->path); + for (i = 0 ; i < 15 ; i++) { + int signum; + if (i == 0) + signum = SIGTERM; + else if (i == 8) + signum = SIGKILL; + else + signum = 0; /* Just check for existance */ + + rc = virCgroupKillRecursive(group, signum); + VIR_DEBUG("Iteration %d rc=%d", i, rc); + /* If rc == -1 we hit error, if 0 we ran out of PIDs */ + if (rc <= 0) + break; + + usleep(200 * 1000); + } + VIR_DEBUG("Complete %d", rc); + return rc; +} diff --git a/src/util/cgroup.h b/src/util/cgroup.h index f1bdd0f..d468cb3 100644 --- a/src/util/cgroup.h +++ b/src/util/cgroup.h @@ -90,4 +90,8 @@ int virCgroupRemove(virCgroupPtr group); void virCgroupFree(virCgroupPtr *group); bool virCgroupMounted(virCgroupPtr cgroup, int controller); +int virCgroupKill(virCgroupPtr group, int signum); +int virCgroupKillRecursive(virCgroupPtr group, int signum); +int virCgroupKillPainfully(virCgroupPtr group); + #endif /* CGROUP_H */ -- 1.7.4

3 7

[libvirt] NIC Bonding in libvirt
by Shi Jin 24 Feb '11

24 Feb '11

Hi there, >From http://wiki.libvirt.org/page/FAQ, it says: Network interface management: Any host running the libvirt daemon can be used to manage physical and logical network interfaces. Enumerate existing interfaces, as well as configure (and create) interfaces, bridges, vlans, and bond devices. This is with the help of netcf.<br/><br/>For more info, see: https://fedorahosted.org/netcf/ It suggests that NIC bonding (teaming) is supported in libvirt but all I can find is this page http://wiki.libvirt.org/page/TodoNICBonding which has nothing in it. Can anyone advise on which stage this work is at and where to get more information on it. My simplistic thought is to have a bridge on a bonded device but that seems to have ARP problems according to https://bugzilla.redhat.com/show_bug.cgi?id=584872 and https://lists.linux-foundation.org/pipermail/bridge/2007-April/005376.html. Thank you very much. Shi -- Shi Jin, PhD

2 1