May 2014 - Devel - libvirt List Archives

[libvirt] [PATCH v6 0/6] Expose FSFreeze/FSThaw within the guest as API
by Tomoki Sekiyama 01 May '14

01 May '14

Hello, This is patchset v6 to add FSFreeze/FSThaw API for custom disk snapshotting. Changes since v5: * replace disks and ndisks parameter with mountpoints and nmountpoints to specify filesystem to be frozen/thawed, so that we can specify non-disk-backed filesystems * add PATCH 6 to support 'mountpoints' argument for 'guest-fsfreeze-freeze' command of qemu-guest-agent. (This requires a patch for qemu-guest-agent, which isn't merged yet: https://lists.gnu.org/archive/html/qemu-devel/2014-04/msg04434.html ) * drop nested fsfreeze request feature * change API version to 1.2.5 (v5: https://www.redhat.com/archives/libvir-list/2014-April/msg00140.html ) === Description === Currently FSFreeze and FSThaw are supported by qemu guest agent and they are used internally in snapshot-create command with --quiesce option. However, when users want to utilize the native snapshot feature of storage devices (such as LVM over iSCSI, enterprise storage appliances, etc.), they need to issue fsfreeze command separately from libvirt-driven snapshots. (OpenStack cinder provides these storages' snapshot feature, but it cannot quiesce the guest filesystems automatically for now.) Although virDomainQemuGuestAgent() API could be used for this purpose, it is only for debugging and is not supported officially. This patchset adds virDomainFSFreeze()/virDomainFSThaw() APIs and virsh domfsfreeze/domfsthaw commands to enable the users to freeze and thaw domain's filesystems cleanly. <updated> The APIs take disks and ndisks parameters, which is a list of mountpoints of filesystems to be frozen/thawed. If the option is not provided, or guest agent doesn't support the 'mountpoints' argument, every mounted filesystem is frozen/thawed. </updated> The APIs have flags option currently unsupported for future extension. --- Tomoki Sekiyama (6): Introduce virDomainFSFreeze() and virDomainFSThaw() public API remote: Implement virDomainFSFreeze and virDomainFSThaw qemu: track quiesced status in qemuDomainSnapshotFSFreeze qemu: Implement virDomainFSFreeze and virDomainFSThaw virsh: Expose new virDomainFSFreeze and virDomainFSThaw API qemu: Support mountpoints option of guest-fsfreeze-freeze include/libvirt/libvirt.h.in | 10 +++ src/access/viraccessperm.c | 2 - src/access/viraccessperm.h | 6 ++ src/driver.h | 14 ++++ src/libvirt.c | 93 ++++++++++++++++++++++++ src/libvirt_public.syms | 6 ++ src/qemu/qemu_agent.c | 47 +++++++++++- src/qemu/qemu_agent.h | 3 + src/qemu/qemu_domain.c | 5 + src/qemu/qemu_domain.h | 2 + src/qemu/qemu_driver.c | 160 ++++++++++++++++++++++++++++++++++++++---- src/remote/remote_driver.c | 2 + src/remote/remote_protocol.x | 38 +++++++++- src/remote_protocol-structs | 18 +++++ src/rpc/gendispatch.pl | 2 + tests/qemuagenttest.c | 8 +- tools/virsh-domain.c | 130 ++++++++++++++++++++++++++++++++++ tools/virsh.pod | 23 ++++++ 18 files changed, 544 insertions(+), 25 deletions(-)

3 17

[libvirt] [PATCH] qemu: fix crash when removing <filterref> from interface with update-device
by Laine Stump 01 May '14

01 May '14

If a domain network interface that contains a <filterref> is modified "live" using "virsh update-device --live", libvirtd would crash. This was because the code supporting live update of an interface's filterref was assuming that a filterref might be added or modified, but didn't account for removing the filterref, resulting in a null dereference of the filter name. Introduced with commit 258fb278, which was first in libvirt v1.0.1. This addresses https://bugzilla.redhat.com/show_bug.cgi?id=1093301 --- src/qemu/qemu_hotplug.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index ccfb358..cdf9eac 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -1815,7 +1815,8 @@ qemuDomainChangeNetFilter(virConnectPtr conn, virDomainConfNWFilterTeardown(olddev); - if (virDomainConfNWFilterInstantiate(conn, vm->def->uuid, newdev) < 0) { + if (newdev->filter && + virDomainConfNWFilterInstantiate(conn, vm->def->uuid, newdev) < 0) { virErrorPtr errobj; virReportError(VIR_ERR_OPERATION_FAILED, -- 1.9.0

2 1

[libvirt] [PATCH v7 0/4] Handling of undefine and re define snapshots with VirtualBox 4.2 or higher
by Yohan BELLEGUIC 01 May '14

01 May '14

Hello, This is a new series of patches in order to support undefining and redefining snapshots with VirtualBox 4.2 or higher. These patches are based on Manuel Vives' patches, taking into account Daniel P. Berrange's remarks. The VirtualBox API provides only high level operations to manipulate snapshots, so it not possible to support flags like VIR_DOMAIN_SNAPSHOT_CREATE_REDEFINE and VIR_DOMAIN_SNAPSHOT_DELETE_METADATA_ONLY with only API calls. Following an IRC talk with Eric Blake, the decision was made to emulate these behaviours by manipulating directly the .vbox XML files. The first patch adds extra details in the snapshot XML returned by libvirt. We will need those details in order to redefine the snapshots. The second patch adds a new API to manipulate the VirtualBox XML file. It provides several structs describing the VirtualBox XML file nodes and functions which can manipulate these structs. The third patch adds support of the VIR_DOMAIN_SNAPSHOT_CREATE_REDEFINE and VIR_DOMAIN_SNAPSHOT_CREATE_CURRENT flags in virDomainSnapshotCreateXML. The idea is to unregister the machine, add the snapshot in the Virtualbox XML file and re-register the machine. However, VirtualBox forbids a machine to have snapshots but no current snapshot. So, if the real current snapshot has not been redefined yet, we create fake disks, allowing us to have an intermediate state in order to not corrupt the snapshot's read-write disks. These fake disks will be deleted during the next redefine. The fourth and last patch adds support of the VIR_DOMAIN_SNAPSHOT_DELETE_METADATA_ONLY flag in virDomainSnapshotDelete. As in the third patch, we also create fake disks to not corrupt the snapshot's read-write disks. The patches were only tested with VirtualBox 4.3.10 and VirtualBox 4.2.24. Regards Yohan BELLEGUIC v7: * Add vbox_snapshot_conf.{h,c} files to (de)serialize VirtualBox XML files * Update the code to use the API exposed by vbox_snapshot_conf.h * Handle the fact that VirtualBox forbids a machine to have snapshots but no current snapshot v6: * Rebased because of a massive change in vbox_tmpl.c due to changes in the handling of different versions of VirtualBox v5: * The patches are modified according to a first review by Laine Stump: * renamed virSearchUuid to virSearchRegex and moved it from viruuid.{c,h} to virstring.{c,h}. * Various fixes. v4: * The code is compliant with Virtualbox 4.3 API * Some minor modifications in order to satisfy "make syntax-check" v3: * Use of STREQ_NULLABLE instead of STREQ in one case * Fix the method for finding uuids according to Ján Tomko review v2: * Fix a licence problem with the method for string replacement Manuel VIVES (1): vbox_tmpl.c: Better XML description for snapshots Yohan BELLEGUIC (3): Add vbox_snapshot_conf struct vbox_tmpl.c: Patch for redefining snapshots vbox_tmpl.c: Add function for undefining snapshot po/POTFILES.in | 1 + src/Makefile.am | 1 + src/vbox/vbox_snapshot_conf.c | 1383 +++++++++++++++++++++++++ src/vbox/vbox_snapshot_conf.h | 100 ++ src/vbox/vbox_tmpl.c | 2225 +++++++++++++++++++++++++++++++++++++---- 5 files changed, 3538 insertions(+), 172 deletions(-) create mode 100644 src/vbox/vbox_snapshot_conf.c create mode 100644 src/vbox/vbox_snapshot_conf.h -- 1.7.10.4

2 9

[libvirt] [PATCH] conf: handle 'vda[-1]' uniformly
by Eric Blake 01 May '14

01 May '14

Commit f22b7899 stumbled across a difference between 32-bit and 64-bit platforms; on 64-bit, parsing "-1" as a long produces 0xffffffffffffffff, which does not fit in unsigned int; but on 32-bit, it parses as 0xffffffff, which DOES fit. Another patch will tweak virStrToLong_ui to behave the same across platforms, but regardless of which of the two choices it makes, the chain lookup code wants to reject negative numbers rather than treating it as large integers. * src/util/virstoragefile.c (virStorageFileParseChainIndex): Reject negative sign. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Therefore, although this qualifies as a build-breaker fix on 32-bit, I haven't pushed it yet: I'm still working on my patch to virstring.c for uniform behavior; and that turned up the fact that virstringtest.c doesn't have coverage of integer parsing, so of course, I have to expand that test too... src/util/virstoragefile.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/util/virstoragefile.c b/src/util/virstoragefile.c index dcce1ef..5d933a4 100644 --- a/src/util/virstoragefile.c +++ b/src/util/virstoragefile.c @@ -1525,7 +1525,9 @@ virStorageFileParseChainIndex(const char *diskTarget, if (virStringListLength(strings) != 2) goto cleanup; - if (virStrToLong_ui(strings[1], &suffix, 10, &idx) < 0 || + /* Rule out spaces or negative sign */ + if (!c_isdigit(*strings[1]) || + virStrToLong_ui(strings[1], &suffix, 10, &idx) < 0 || STRNEQ(suffix, "]")) goto cleanup; -- 1.9.0

2 2