Later patches will implement support for <transient/> disks in libvirt by installing an overlay on top of the configured image. This will require cleanup after the VM will be stopped so that the state is correctly discarded. Since the overlay will be installed only during the startup phase of the VM we need to ensure that qemuProcessStop doesn't delete the original file on some previous failure. This is solved by adding 'inhibitDiskTransientDelete' VM private data member which is set prior to any startup step and will be cleared once transient disk overlays are established. Based on that we can then delete the overlays for any <transient/> disk. Signed-off-by: Peter Krempa <pkrempa(a)redhat.com&gt; --- src/qemu/qemu_domain.c | 2 ++ src/qemu/qemu_domain.h | 4 ++++ src/qemu/qemu_process.c | 22 ++++++++++++++++++++++ 3 files changed, 28 insertions(+) diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c index 279de2997d..dc5949edfa 100644 --- a/src/qemu/qemu_domain.c +++ b/src/qemu/qemu_domain.c @@ -1792,6 +1792,8 @@ qemuDomainObjPrivateDataClear(qemuDomainObjPrivatePtr priv) priv->dbusVMStateIds = NULL; priv->dbusVMState = false; + + priv->inhibitDiskTransientDelete = false; } diff --git a/src/qemu/qemu_domain.h b/src/qemu/qemu_domain.h index c7c3c5c073..ec776ced72 100644 --- a/src/qemu/qemu_domain.h +++ b/src/qemu/qemu_domain.h @@ -263,6 +263,10 @@ struct _qemuDomainObjPrivate { char **dbusVMStateIds; /* true if -object dbus-vmstate was added */ bool dbusVMState; + + /* prevent deletion of <transient> disk overlay files between startup and + * succesful setup of the overlays */ + bool inhibitDiskTransientDelete; }; #define QEMU_DOMAIN_PRIVATE(vm) \ diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index f21b8f1585..ffb3afa9c5 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -5616,6 +5616,9 @@ qemuProcessInit(virQEMUDriverPtr driver, if (virDomainObjSetDefTransient(driver->xmlopt, vm, priv->qemuCaps) < 0) goto cleanup; + /* don't clean up files for <transient> disks until we set them up */ + priv->inhibitDiskTransientDelete = true; + if (flags & VIR_QEMU_PROCESS_START_PRETEND) { if (qemuDomainSetPrivatePaths(driver, vm) < 0) { virDomainObjRemoveTransientDef(vm); @@ -7710,6 +7713,18 @@ void qemuProcessStop(virQEMUDriverPtr driver, } qemuBlockRemoveImageMetadata(driver, vm, disk->dst, disk->src); + + /* for now transient disks are forbidden with migration so they + * can be handled here */ + if (disk->transient && + !priv->inhibitDiskTransientDelete) { + VIR_DEBUG("Removing transient overlay '%s' of disk '%s'", + disk->src->path, disk->dst); + if (qemuDomainStorageFileInit(driver, vm, disk->src, NULL) >= 0) { + virStorageFileUnlink(disk->src); + virStorageFileDeinit(disk->src); + } + } } } @@ -8125,6 +8140,10 @@ qemuProcessReconnect(void *opaque) cfg = virQEMUDriverGetConfig(driver); priv = obj->privateData; + /* expect that libvirt might have crashed during VM start, so prevent + * cleanup of transient disks */ + priv->inhibitDiskTransientDelete = true; + if (qemuDomainObjBeginJob(driver, obj, QEMU_JOB_MODIFY) < 0) goto error; jobStarted = true; @@ -8228,6 +8247,9 @@ qemuProcessReconnect(void *opaque) goto error; } + /* vm startup complete, we can remove transient disks if required */ + priv->inhibitDiskTransientDelete = false; + /* In case the domain shutdown while we were not running, * we need to finish the shutdown process. And we need to do it after * we have virQEMUCaps filled in. --

To implement <transient/> disks we'll need to install an overlay on top of the original disk image which will be discarded after the VM is turned off. This was initially implemented by qemu but libvirt never > picked up this option. With blockdev the qemu feature became unsupported so we need to do this via the snapshot code anyways. The helpers introduced in this patch prepare a fake snapshot disk definition for a disk which is configured as <transient/> and use it to create a snapshot (without actually modifying metada or persistent def). Signed-off-by: Peter Krempa <pkrempa(a)redhat.com&gt; --- src/qemu/qemu_snapshot.c | 91 +++++++++++++++++++++++++++++++++++++++- src/qemu/qemu_snapshot.h | 5 +++ 2 files changed, 95 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c index ca051071aa..d10e7b6b3a 100644 --- a/src/qemu/qemu_snapshot.c +++ b/src/qemu/qemu_snapshot.c @@ -986,6 +986,7 @@ qemuSnapshotDiskPrepareOne(virDomainObjPtr vm, qemuSnapshotDiskDataPtr dd, virHashTablePtr blockNamedNodeData, bool reuse, + bool updateConfig, qemuDomainAsyncJob asyncJob, virJSONValuePtr actions) { @@ -1008,7 +1009,8 @@ qemuSnapshotDiskPrepareOne(virDomainObjPtr vm, return -1; /* modify disk in persistent definition only when the source is the same */ - if (vm->newDef && + if (updateConfig && + vm->newDef && (persistdisk = virDomainDiskByTarget(vm->newDef, dd->disk->dst)) && virStorageSourceIsSameLocation(dd->disk->src, persistdisk->src)) { @@ -1116,6 +1118,54 @@ qemuSnapshotDiskPrepareActiveExternal(virDomainObjPtr vm, snapctxt->dd + snapctxt->ndd++, blockNamedNodeData, reuse, + true, + asyncJob, + snapctxt->actions) < 0) + return NULL; + } + + return g_steal_pointer(&snapctxt); +} + + +static qemuSnapshotDiskContextPtr +qemuSnapshotDiskPrepareDisksTransient(virDomainObjPtr vm, + virQEMUDriverConfigPtr cfg, + virHashTablePtr blockNamedNodeData, + qemuDomainAsyncJob asyncJob) +{ + g_autoptr(qemuSnapshotDiskContext) snapctxt = NULL; + size_t i; + + snapctxt = qemuSnapshotDiskContextNew(vm->def->ndisks, vm, asyncJob); + + for (i = 0; i < vm->def->ndisks; i++) { + virDomainDiskDefPtr domdisk = vm->def->disks[i]; + g_autoptr(virDomainSnapshotDiskDef) snapdisk = g_new0(virDomainSnapshotDiskDef, 1); + + if (!domdisk->transient) + continue; + + /* validation code makes sure that we do this only for local disks + * with a file source */ + snapdisk->name = g_strdup(domdisk->dst); + snapdisk->snapshot = VIR_DOMAIN_SNAPSHOT_LOCATION_EXTERNAL; + snapdisk->src = virStorageSourceNew(); + snapdisk->src->type = VIR_STORAGE_TYPE_FILE; + snapdisk->src->format = VIR_STORAGE_FILE_QCOW2; + snapdisk->src->path = g_strdup_printf("%s.TRANSIENT", domdisk->src->path); + if (virFileExists(snapdisk->src->path)) { + virReportError(VIR_ERR_OPERATION_UNSUPPORTED, + _("Overlay file '%s' for transient disk '%s' already exists"), + snapdisk->src->path, domdisk->dst); + return NULL; + } + + if (qemuSnapshotDiskPrepareOne(vm, cfg, domdisk, snapdisk, + snapctxt->dd + snapctxt->ndd++, + blockNamedNodeData, + false, + false, asyncJob, snapctxt->actions) < 0) return NULL; @@ -1253,6 +1303,45 @@ qemuSnapshotCreateActiveExternalDisks(virDomainObjPtr vm, } +/** + * qemuSnapshotCreateDisksTransient: + * @vm: domain object + * @asyncJob: qemu async job type + * + * Creates overlays on top of disks which are configured as <transient/>. Note + * that the validation code ensures that <transient> disks have appropriate + * configuration. + */ +int +qemuSnapshotCreateDisksTransient(virDomainObjPtr vm, + qemuDomainAsyncJob asyncJob) +{ + qemuDomainObjPrivatePtr priv = vm->privateData; + virQEMUDriverPtr driver = priv->driver; + g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver); + g_autoptr(qemuSnapshotDiskContext) snapctxt = NULL; + g_autoptr(virHashTable) blockNamedNodeData = NULL; + + if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) { + if (!(blockNamedNodeData = qemuBlockGetNamedNodeData(vm, asyncJob))) + return -1; + + if (!(snapctxt = qemuSnapshotDiskPrepareDisksTransient(vm, cfg, + blockNamedNodeData, + asyncJob))) + return -1; + + if (qemuSnapshotDiskCreate(snapctxt, cfg) < 0) + return -1; + } + + /* the overlays are established, so they can be deleted on a shutdown */ + priv->inhibitDiskTransientDelete = false; + + return 0; +} + + static int qemuSnapshotCreateActiveExternal(virQEMUDriverPtr driver, virDomainObjPtr vm, diff --git a/src/qemu/qemu_snapshot.h b/src/qemu/qemu_snapshot.h index 8b3ebe87b1..cf9bfec542 100644 --- a/src/qemu/qemu_snapshot.h +++ b/src/qemu/qemu_snapshot.h @@ -21,6 +21,7 @@ #include "virconftypes.h" #include "datatypes.h" #include "qemu_conf.h" +#include "qemu_domainjob.h" virDomainMomentObjPtr qemuSnapObjFromName(virDomainObjPtr vm, @@ -53,3 +54,7 @@ int qemuSnapshotDelete(virDomainObjPtr vm, virDomainSnapshotPtr snapshot, unsigned int flags); + +int +qemuSnapshotCreateDisksTransient(virDomainObjPtr vm, + qemuDomainAsyncJob asyncJob); --

To implement <transient/> disks we'll need to install an overlay on top of the original disk image which will be discarded after the VM is turned off. This was initially implemented by qemu but libvirt never picked up this option. With blockdev the qemu feature became unsupported so we need to do this via the snapshot code anyways. The helpers introduced in this patch prepare a fake snapshot disk definition for a disk which is configured as <transient/> and use it to create a snapshot (without actually modifying metada or persistent def). Signed-off-by: Peter Krempa <pkrempa(a)redhat.com&gt; --- src/qemu/qemu_snapshot.c | 91 +++++++++++++++++++++++++++++++++++++++- src/qemu/qemu_snapshot.h | 5 +++ 2 files changed, 95 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c index ca051071aa..d10e7b6b3a 100644 --- a/src/qemu/qemu_snapshot.c +++ b/src/qemu/qemu_snapshot.c @@ -986,6 +986,7 @@ qemuSnapshotDiskPrepareOne(virDomainObjPtr vm, qemuSnapshotDiskDataPtr dd, virHashTablePtr blockNamedNodeData, bool reuse, + bool updateConfig, qemuDomainAsyncJob asyncJob, virJSONValuePtr actions) { @@ -1008,7 +1009,8 @@ qemuSnapshotDiskPrepareOne(virDomainObjPtr vm, return -1; /* modify disk in persistent definition only when the source is the same */ - if (vm->newDef && + if (updateConfig && + vm->newDef && (persistdisk = virDomainDiskByTarget(vm->newDef, dd->disk->dst)) && virStorageSourceIsSameLocation(dd->disk->src, persistdisk->src)) { @@ -1116,6 +1118,54 @@ qemuSnapshotDiskPrepareActiveExternal(virDomainObjPtr vm, snapctxt->dd + snapctxt->ndd++, blockNamedNodeData, reuse, + true, + asyncJob, + snapctxt->actions) < 0) + return NULL; + } + + return g_steal_pointer(&snapctxt); +} + + +static qemuSnapshotDiskContextPtr +qemuSnapshotDiskPrepareDisksTransient(virDomainObjPtr vm, + virQEMUDriverConfigPtr cfg, + virHashTablePtr blockNamedNodeData, + qemuDomainAsyncJob asyncJob) +{ + g_autoptr(qemuSnapshotDiskContext) snapctxt = NULL; + size_t i; + + snapctxt = qemuSnapshotDiskContextNew(vm->def->ndisks, vm, asyncJob); + + for (i = 0; i < vm->def->ndisks; i++) { + virDomainDiskDefPtr domdisk = vm->def->disks[i]; + g_autoptr(virDomainSnapshotDiskDef) snapdisk = g_new0(virDomainSnapshotDiskDef, 1); + + if (!domdisk->transient) + continue; + + /* validation code makes sure that we do this only for local disks + * with a file source */ + snapdisk->name = g_strdup(domdisk->dst); + snapdisk->snapshot = VIR_DOMAIN_SNAPSHOT_LOCATION_EXTERNAL; + snapdisk->src = virStorageSourceNew(); + snapdisk->src->type = VIR_STORAGE_TYPE_FILE; + snapdisk->src->format = VIR_STORAGE_FILE_QCOW2; + snapdisk->src->path = g_strdup_printf("%s.TRANSIENT", domdisk->src->path);

+ if (virFileExists(snapdisk->src->path)) { + virReportError(VIR_ERR_OPERATION_UNSUPPORTED, + _("Overlay file '%s' for transient disk '%s' already exists"), + snapdisk->src->path, domdisk->dst); + return NULL; + } + + if (qemuSnapshotDiskPrepareOne(vm, cfg, domdisk, snapdisk, + snapctxt->dd + snapctxt->ndd++, + blockNamedNodeData, + false, + false, asyncJob, snapctxt->actions) < 0) return NULL; @@ -1253,6 +1303,45 @@ qemuSnapshotCreateActiveExternalDisks(virDomainObjPtr vm, } +/** + * qemuSnapshotCreateDisksTransient: + * @vm: domain object + * @asyncJob: qemu async job type + * + * Creates overlays on top of disks which are configured as <transient/>. Note + * that the validation code ensures that <transient> disks have appropriate + * configuration. + */ +int +qemuSnapshotCreateDisksTransient(virDomainObjPtr vm, + qemuDomainAsyncJob asyncJob) +{ + qemuDomainObjPrivatePtr priv = vm->privateData; + virQEMUDriverPtr driver = priv->driver; + g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver); + g_autoptr(qemuSnapshotDiskContext) snapctxt = NULL; + g_autoptr(virHashTable) blockNamedNodeData = NULL; + + if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) { + if (!(blockNamedNodeData = qemuBlockGetNamedNodeData(vm, asyncJob))) + return -1; + + if (!(snapctxt = qemuSnapshotDiskPrepareDisksTransient(vm, cfg, + blockNamedNodeData, + asyncJob))) + return -1; + + if (qemuSnapshotDiskCreate(snapctxt, cfg) < 0) + return -1; + } + + /* the overlays are established, so they can be deleted on a shutdown */

+ priv->inhibitDiskTransientDelete = false; + + return 0; +} + + static int qemuSnapshotCreateActiveExternal(virQEMUDriverPtr driver, virDomainObjPtr vm,

From: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com&gt; Extract the validation of transient disk option. We support transient disks in qemu under the following conditions: - -blockdev is used - the disk source is a local file - the disk type is 'disk' - the disk is not readonly Signed-off-by: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com&gt; Signed-off-by: Peter Krempa <pkrempa(a)redhat.com&gt; --- docs/formatdomain.rst | 5 ++-- src/qemu/qemu_validate.c | 56 +++++++++++++++++++++++++++++++++++----- 2 files changed, 53 insertions(+), 8 deletions(-) diff --git a/docs/formatdomain.rst b/docs/formatdomain.rst index 888db5ea29..cc1467c0e6 100644 --- a/docs/formatdomain.rst +++ b/docs/formatdomain.rst @@ -2974,8 +2974,9 @@ paravirtualized driver is specified via the ``disk`` element. ``transient`` If present, this indicates that changes to the device contents should be reverted automatically when the guest exits. With some hypervisors, marking a - disk transient prevents the domain from participating in migration or - snapshots. Only suppported in vmx hypervisor. :since:`Since 0.9.5` + disk transient prevents the domain from participating in migration, + snapshots, or blockjobs. Only suppported in vmx hypervisor

+ (:since:`Since 0.9.5`) and ``qemu`` hypervisor (:since:`Since 6.9.0`). ``serial`` If present, this specify serial number of virtual hard drive. For example, it may look like ``<serial>WD-WMAP9A966149</serial>``. Not supported for diff --git a/src/qemu/qemu_validate.c b/src/qemu/qemu_validate.c index 3ed4039cdf..3196814aca 100644 --- a/src/qemu/qemu_validate.c +++ b/src/qemu/qemu_validate.c @@ -2186,12 +2186,6 @@ qemuValidateDomainDeviceDefDiskFrontend(const virDomainDiskDef *disk, } } - if (disk->transient) { - virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", - _("transient disks not supported yet")); - return -1; - } - if (disk->iomode == VIR_DOMAIN_DISK_IO_NATIVE && disk->cachemode != VIR_DOMAIN_DISK_CACHE_DIRECTSYNC && disk->cachemode != VIR_DOMAIN_DISK_CACHE_DISABLE) { @@ -2340,6 +2334,53 @@ qemuValidateDomainDeviceDefDiskBlkdeviotune(const virDomainDiskDef *disk, } +static int +qemuValidateDomainDeviceDefDiskTransient(const virDomainDiskDef *disk, + virQEMUCapsPtr qemuCaps) + +{ + virStorageType actualType = virStorageSourceGetActualType(disk->src); + + if (!disk->transient) + return 0; + + if (virStorageSourceIsEmpty(disk->src)) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, + _("transient disk '%s' must not be empty"), disk->dst); + return -1; + } + + if (disk->src->readonly) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, + _("transient disk '%s' must not be read-only"), disk->dst); + return -1; + } + + if (actualType != VIR_STORAGE_TYPE_FILE) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, + _("transient disk supported only with 'file' type (%s)"), + disk->dst); + return -1; + } + + if (disk->device != VIR_DOMAIN_DISK_DEVICE_DISK) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, + _("transient disk supported only with 'disk' device (%s)"), + disk->dst); + return -1; + } + + if (qemuCaps && !virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, + _("transient disk not supported by this qemu binary (%s)"),

+ disk->dst); + return -1; + } + + return 0; +} + + int qemuValidateDomainDeviceDefDisk(const virDomainDiskDef *disk, const virDomainDef *def,