February 2017 - Devel - libvirt List Archives

[libvirt] Schedule for next release
by Daniel Veillard 01 Mar '17

01 Mar '17

Hi everybody, it's about time to think about 3.1.0, I think to be on the safe side for a March 1st release we should enter freeze at the end of the week, most likely on Friday, so if you have patches that need to get in, please make sure they are approved and pushed in time. Hope this works for everybody, thanks, Daniel -- Daniel Veillard | Red Hat Developers Tools http://developer.redhat.com/ veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

1 1

[libvirt] [PATCH RFC 0/9] Introduce virPoolObj[Table]Ptr data/API's
by John Ferlan 01 Mar '17

01 Mar '17

Based of current top of git commit id '5620c6095'... I imagine things will change while this is on-list, so if there is the desire to create your own branch that's your starting point! This is perhaps more than an RFC considering I've added comments already (at least one reason why the insertions count is high). Still it's not really complete and ready for prime time, but I figured I'd put it out there for feedback. Essentially I've tried to follow the domainobj 'model' by creating an API to manage the "objects" various drivers have the need to manage. Currently there are 4 drivers using linked lists of some sort and 4 using hash tables in some manner. This implementation modifies the linked list abusers to use a common hash table. The common hash table will have uuid/name accessors. The preference remains uuid for uniqueness, but there are a couple of tables (node device, nwfilter) that don't have a uuid and thus name is unique enough for them. The genesis for this was recent patches to the storage driver that were essentially copying many lines of code and just changing the names to do the same thing. As I looked deeper at that - I found a mish-mash of object management throughout the drivers, so this is my attempt to unify that. The "goal" is to have a common look and feel to the API's so there's less thought over how a specific implementation does things. Each trip into the driver and exit can be handled in a common manner especially with respect to locks/refcnt's. Since I was splitting things apart - I also split out the object code/data *_conf.{c,h} into separate vir*obj.{c,h} files. I took some liberties with adjusting some names, modifying some other things along the way since the type A personality took over to try and follow more current coding conventions while I was in the middle of changing things. While the ultimate goal is to finish with the domain code - that's just too busy in order to make adjustments right now. Still at least that code does use a similar hash table model - so while it will be painful - it can be done later if/when this is accepted (and perhaps when we know there's a slow period for posting changes). John Ferlan (9): conf: Modify gendispatch.pl to make ACL argument opaque conf: Introduce virPoolObj and virPoolTableObj and PoolObj API's nodedev: Convert virNodeDevObj[List] to use virPoolObj[Table] interface: Convert virInterfaceObj[List] to use virPoolObj[Table] nwfilter: Convert virNWFilterObj[List] to use virPoolObj[Table] secret: Convert virSecretObj[List] to use virPoolObj[Table] volume: Convert virStorageVolObj[List] to use virPoolObj[Table] storage: Convert virStoragePoolObj[List] to use virPoolObj[Table] network: Convert virNetworkObj[List] to use virPoolObj[Table] include/libvirt/virterror.h | 1 + po/POTFILES.in | 5 + src/Makefile.am | 18 +- src/conf/domain_conf.h | 3 +- src/conf/interface_conf.c | 170 +-- src/conf/interface_conf.h | 47 +- src/conf/network_conf.c | 1308 +----------------- src/conf/network_conf.h | 133 +- src/conf/node_device_conf.c | 497 +------ src/conf/node_device_conf.h | 86 +- src/conf/nwfilter_conf.c | 381 +----- src/conf/nwfilter_conf.h | 69 +- src/conf/secret_conf.c | 3 +- src/conf/secret_conf.h | 2 +- src/conf/storage_conf.c | 925 +------------ src/conf/storage_conf.h | 128 +- src/conf/virinterfaceobj.c | 205 +++ src/conf/virinterfaceobj.h | 46 + src/conf/virnetworkobj.c | 1166 ++++++++++++++++ src/conf/virnetworkobj.h | 121 ++ src/conf/virnodedeviceobj.c | 570 ++++++++ src/conf/virnodedeviceobj.h | 86 ++ src/conf/virnwfilterobj.c | 373 +++++ src/conf/virnwfilterobj.h | 64 + src/conf/virpoolobj.c | 1184 ++++++++++++++++ src/conf/virpoolobj.h | 222 +++ src/conf/virsecretobj.c | 782 ++++------- src/conf/virsecretobj.h | 92 +- src/conf/virstorageobj.c | 1206 +++++++++++++++++ src/conf/virstorageobj.h | 143 ++ src/interface/interface_backend_netcf.c | 6 +- src/interface/interface_backend_udev.c | 5 +- src/libvirt_private.syms | 217 +-- src/network/bridge_driver.c | 1812 +++++++++++++------------ src/network/bridge_driver.h | 11 +- src/network/bridge_driver_platform.h | 3 +- src/node_device/node_device_driver.c | 388 +++--- src/node_device/node_device_driver.h | 2 +- src/node_device/node_device_hal.c | 82 +- src/node_device/node_device_udev.c | 76 +- src/nwfilter/nwfilter_driver.c | 225 ++- src/nwfilter/nwfilter_gentech_driver.c | 111 +- src/nwfilter/nwfilter_gentech_driver.h | 2 +- src/nwfilter/nwfilter_tech_driver.h | 2 +- src/rpc/gendispatch.pl | 5 +- src/secret/secret_driver.c | 187 +-- src/storage/storage_backend.h | 32 +- src/storage/storage_backend_disk.c | 309 +++-- src/storage/storage_backend_fs.c | 145 +- src/storage/storage_backend_gluster.c | 52 +- src/storage/storage_backend_iscsi.c | 65 +- src/storage/storage_backend_logical.c | 177 +-- src/storage/storage_backend_mpath.c | 37 +- src/storage/storage_backend_rbd.c | 115 +- src/storage/storage_backend_scsi.c | 57 +- src/storage/storage_backend_sheepdog.c | 89 +- src/storage/storage_backend_vstorage.c | 47 +- src/storage/storage_backend_zfs.c | 109 +- src/storage/storage_driver.c | 1952 +++++++++++++------------- src/storage/storage_driver.h | 6 +- src/storage/storage_util.c | 234 ++-- src/storage/storage_util.h | 31 +- src/test/test_driver.c | 2257 ++++++++++++++----------------- src/util/virerror.c | 1 + tests/networkxml2conftest.c | 18 +- tests/storagevolxml2argvtest.c | 16 +- 66 files changed, 10145 insertions(+), 8774 deletions(-) create mode 100644 src/conf/virinterfaceobj.c create mode 100644 src/conf/virinterfaceobj.h create mode 100644 src/conf/virnetworkobj.c create mode 100644 src/conf/virnetworkobj.h create mode 100644 src/conf/virnodedeviceobj.c create mode 100644 src/conf/virnodedeviceobj.h create mode 100644 src/conf/virnwfilterobj.c create mode 100644 src/conf/virnwfilterobj.h create mode 100644 src/conf/virpoolobj.c create mode 100644 src/conf/virpoolobj.h create mode 100644 src/conf/virstorageobj.c create mode 100644 src/conf/virstorageobj.h -- 2.7.4

6 28

[libvirt] [PATCH] testNodeDeviceMockCreateVport: Don't call public APIs
by Michal Privoznik 01 Mar '17

01 Mar '17

This function is calling public APIs (virNodeDeviceLookupByName etc.). That requires the driver lock to be unlocked and locked again. If we, however, replace the public APIs calls with the internal calls (that public APIs call anyway), we can drop the lock/unlock exercise. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/test/test_driver.c | 59 ++++++++++++++++++++++++++++---------------------- 1 file changed, 33 insertions(+), 26 deletions(-) diff --git a/src/test/test_driver.c b/src/test/test_driver.c index 5fef3f10b..8495443db 100644 --- a/src/test/test_driver.c +++ b/src/test/test_driver.c @@ -5626,17 +5626,16 @@ testNodeDeviceListCaps(virNodeDevicePtr dev, char **const names, int maxnames) } -static virNodeDeviceDefPtr +static virNodeDeviceObjPtr testNodeDeviceMockCreateVport(virConnectPtr conn, const char *wwnn, const char *wwpn) { testDriverPtr driver = conn->privateData; - virNodeDevicePtr devcpy = NULL; char *xml = NULL; virNodeDeviceDefPtr def = NULL; virNodeDevCapsDefPtr caps; - virNodeDeviceObjPtr obj = NULL; + virNodeDeviceObjPtr obj = NULL, objcopy = NULL; virObjectEventPtr event = NULL; /* In the real code, we'd call virVHBAManageVport which would take the @@ -5648,9 +5647,15 @@ testNodeDeviceMockCreateVport(virConnectPtr conn, * using the scsi_host11 definition, changing the name and the * scsi_host capability fields before calling virNodeDeviceAssignDef * to add the def to the node device objects list. */ - if (!(devcpy = virNodeDeviceLookupByName(conn, "scsi_host11")) || - !(xml = virNodeDeviceGetXMLDesc(devcpy, 0)) || - !(def = virNodeDeviceDefParseString(xml, EXISTING_DEVICE, NULL))) + if (!(objcopy = virNodeDeviceFindByName(&driver->devs, "scsi_host11"))) + goto cleanup; + + xml = virNodeDeviceDefFormat(objcopy->def); + virNodeDeviceObjUnlock(objcopy); + if (!xml) + goto cleanup; + + if (!(def = virNodeDeviceDefParseString(xml, EXISTING_DEVICE, NULL))) goto cleanup; VIR_FREE(def->name); @@ -5684,23 +5689,17 @@ testNodeDeviceMockCreateVport(virConnectPtr conn, if (!(obj = virNodeDeviceAssignDef(&driver->devs, def))) goto cleanup; - virNodeDeviceObjUnlock(obj); + def = NULL; - event = virNodeDeviceEventLifecycleNew(def->name, + event = virNodeDeviceEventLifecycleNew(obj->def->name, VIR_NODE_DEVICE_EVENT_CREATED, 0); testObjectEventQueue(driver, event); cleanup: VIR_FREE(xml); - if (!obj) { - virNodeDeviceDefFree(def); - def = NULL; - } - if (devcpy) - virNodeDeviceFree(devcpy); - - return def; + virNodeDeviceDefFree(def); + return obj; } @@ -5712,8 +5711,8 @@ testNodeDeviceCreateXML(virConnectPtr conn, testDriverPtr driver = conn->privateData; virNodeDeviceDefPtr def = NULL; char *wwnn = NULL, *wwpn = NULL; - virNodeDevicePtr dev = NULL; - virNodeDeviceDefPtr newdef = NULL; + virNodeDevicePtr dev = NULL, ret = NULL; + virNodeDeviceObjPtr obj = NULL; virCheckFlags(0, NULL); @@ -5739,20 +5738,28 @@ testNodeDeviceCreateXML(virConnectPtr conn, * mocking udev. The mock routine will copy an existing vHBA and * rename a few fields to mock that. So in order to allow that to * work properly, we need to drop our lock */ - testDriverUnlock(driver); - if ((newdef = testNodeDeviceMockCreateVport(conn, wwnn, wwpn))) { - if ((dev = virNodeDeviceLookupByName(conn, newdef->name))) - ignore_value(VIR_STRDUP(dev->parent, def->parent)); - } - testDriverLock(driver); - newdef = NULL; + if (!(obj = testNodeDeviceMockCreateVport(conn, wwnn, wwpn))) + goto cleanup; + + if (!(dev = virGetNodeDevice(conn, obj->def->name))) + goto cleanup; + + VIR_FREE(dev->parent); + if (VIR_STRDUP(dev->parent, def->parent) < 0) + goto cleanup; + + ret = dev; + dev = NULL; cleanup: + if (obj) + virNodeDeviceObjUnlock(obj); testDriverUnlock(driver); virNodeDeviceDefFree(def); + virObjectUnref(dev); VIR_FREE(wwnn); VIR_FREE(wwpn); - return dev; + return ret; } static int -- 2.11.0

2 2

[libvirt] A question for external snapshots with flag“VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE”
by lizhengui 28 Feb '17

28 Feb '17

Hello, I got a question here. When we create consistent active external snapshots with flag “VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE” , the qemuDomainSnapshotFSFreeze will be called firstly to freeze all filesystems in vm, and then create snapshots. For windows vm, freezing filesystems used by vss service. If IO pressure in vm is too big or internal error happened in vss , freezing filesytems will failed , and qemuDomainSnapshotFSFreeze returns 0 which meaning that no filesystems are frozen. In the function qemuDomainSnapshotCreateActiveExternal, libvirt creates external snapshots all the same in such a situation that qemuDomainSnapshotFSFreeze returns 0, but the created snapshots are not consistent snapshots in such a situation . So shouldn't we abandon creating snapshots and goto cleanup in the situation that qemuDomainSnapshotFSFreeze returns 0? The code below: static int qemuDomainSnapshotCreateActiveExternal(virConnectPtr conn, virQEMUDriverPtr driver, virDomainObjPtr vm, virDomainSnapshotObjPtr snap, unsigned int flags) { virObjectEventPtr event; bool resume = false; int ret = -1; qemuDomainObjPrivatePtr priv = vm->privateData; char *xml = NULL; bool memory = snap->def->memory == VIR_DOMAIN_SNAPSHOT_LOCATION_EXTERNAL; bool memory_unlink = false; bool atomic = !!(flags & VIR_DOMAIN_SNAPSHOT_CREATE_ATOMIC); bool transaction = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_TRANSACTION); int thaw = 0; /* 1 if freeze succeeded, -1 if freeze failed */ bool pmsuspended = false; virQEMUDriverConfigPtr cfg = NULL; int compressed; char *compressedpath = NULL; /* If quiesce was requested, then issue a freeze command, and a * counterpart thaw command when it is actually sent to agent. * The command will fail if the guest is paused or the guest agent * is not running, or is already quiesced. */ if (flags & VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE) { int freeze = qemuDomainSnapshotFSFreeze(driver, vm, NULL, 0); if (freeze < 0) { /* the helper reported the error */ if (freeze == -2) thaw = -1; /* the command is sent but agent failed */ goto cleanup; } thaw = 1; } ...... ......

3 2

[libvirt] ANNOUNCE: libguestfs 1.36 released
by Richard W.M. Jones 28 Feb '17

28 Feb '17

I'm please to announce libguestfs 1.36, a library and a set of tools for accessing and modifying virtual machine disk images. This release represents about 7 months of work by many contributors. I'd like to call out in particular substantial code contributions from: Pino Toscano, Tomáš Golembiovský, Cédric Bosdonnat, Roman Kagan, Dawid Zamirski and Matteo Cafasso. Also our testing team: Tingting Zheng, Xiang Hua Chen, Ming Xie, Junquin Zhou and Xiaodai Wang. Amongst the new features are substantial enhancements to virt-v2v including support for Debian and Ubuntu guest conversions, and conversions from AWS; a new "virt-tail" command for following log files in guests; even more support for encrypted guests; robustness when handling Windows guests with corrupted registries; ExFAT support; GCC 7 support; and considerable internal code refactoring to make things more maintainable. Of course dozens of other features have been added, and many bugs fixed. See the release notes below for full details. You can get libguestfs 1.36 from here: Main website: http://libguestfs.org/ Source: http://libguestfs.org/download/1.36-stable/ Fedora: https://koji.fedoraproject.org/koji/packageinfo?packageID=8391 Debian/experimental: https://packages.debian.org/libguestfs0 Note for distro packagers: libmagic (part of the 'file' command line tool) is now a required build dependency. It was previously optional. Rich. ---------------------------------------------------------------------- Release notes for libguestfs 1.36 These are also available online at: http://libguestfs.org/guestfs-release-notes.1.html New features New tools Virt-tail is a new tool for following (tailing) log files within a guest, similar to the regular "tail -f" command. New features in existing tools Virt-customize, virt-get-kernel, virt-sparsify and virt-sysprep can now handle encrypted guests (Pino Toscano). Virt-builder and virt-customize now support --append-line, which is useful for adding lines to the end of configuration files. Virt-resize can now shrink and expand swap partitions (Pino Toscano). Virt-resize can now output to non-local disks (Pino Toscano). Virt-sysprep has a new operation called "backup-files" for removing editor backups, and "passwd-backups" for removing /etc/passwd- and similar. Virt-dib can now create checksums using the new --checksum option (Pino Toscano). Virt-dib can now create tgz files. In addition, extended attributes and SELinux labels are preserved in the output tar file (Pino Toscano). Virt-dib can now create squashfs files (Pino Toscano). Bash tab-completion is now available in every command line tool that is shipped. virt-v2v and virt-p2v Conversions of Debian (6+) and Ubuntu (10.04+) guests are now supported (Tomáš Golembiovský, Pino Toscano). Conversions from SUSE Xen hypervisor have now been tested (Cédric Bosdonnat). Windows conversions may now install a program "pnp_wait.exe" which prevents conflicts between the Windows Plug-and-Play Manager and our own scripts that install virtio device drivers. Enabling this requires the "pnp_wait.exe" program to be built separately (Roman Kagan). OVA files exported from AWS can now be converted (Shahar Havivi). When converting OVA files, in some circumstances virt-v2v can now read disk images directly from the OVA input file instead of needing to unpack the OVA file to a temporary directory. This can save large amounts of disk space, and is quicker (Tomáš Golembiovský). Virt-p2v can now be built on RHEL 5 (2007-era) Linux and RHEL 6, which means that 32 bit and hardware with ancient fakeraid disks can be virtualized. Binaries of virt-p2v based on these old versions of RHEL can be found at http://oirase.annexia.org/virt-p2v/ Virt-p2v can now use nbdkit (an NBD server) as an alternative to qemu- nbd. In addition, virt-p2v can use "socket activation" which is a more robust method for opening the NBD listening socket. Socket activation works with qemu-nbd or nbdkit, but requires the most up to date versions. To prevent timeouts during P2V conversions, virt-p2v inhibits power saving on the physical machine, and also sends ping packets over the ssh control and data connections. Virt-v2v no longer removes the "Processor" and "Intelppm" nodes from the Windows Registry. Xen and vCenter conversions can now be done using the libvirt backend, provided libvirt ≥ 2.1.0 is used. /dev/srX (SCSI CD-ROM) devices are ignored (Tomáš Golembiovský). When converting SUSE Linux guests, use /etc/modprobe.conf.local if it exists. Removing VMware tools from Linux guests should now work reliably (Pino Toscano). When converting OVA files, virt-v2v now checks the disk image hash contained in the manifest file, whereas previously it was ignored because of a mistake in the code. In addition, SHA256 hashes are now supported (Tomáš Golembiovský). When converting OVA files that contain compressed disk images, we now trust the "ovf:compression" attribute from the metadata instead of using content sniffing on the disk image (Tomáš Golembiovský). A new flag --vdsm-compat has been added to -o vdsm mode, allowing more efficient qcow2 images to be generated. Since the RHEV product [commercially supported version of oVirt] has now been renamed to RHV, "RHEV" was changed to "RHV" throughout the tooling. In particular, -o rhev becomes -o rhv, although the old name can be used for backwards compatibility. The "RHEV-APT" utility is only installed when the output hypervisor is oVirt or RHV. A problem with virt-v2v running out of memory when doing an SELinux relabel of the guest filesystem should now be resolved. Language bindings Setting "EXTRA_JAVAC_FLAGS" before ./configure allows you to add arbitrary flags to "javac" when compiling the Java bindings. Use of any libguestfs API which returned a single struct, from Perl or Java bindings, would have leaked memory. This has now been fixed (Pino Toscano). Inspection Inspection of Windows guests with mildly corrupted Windows Registry hives should now work instead of failing with an error. This feature requires hivex ≥ 1.3.14 (Dawid Zamirski). For Debian guests, we can now read the URL, source name, summary and full description of installed packages. For RPM-based guests we can read the URL, summary and description (Pino Toscano). Inspection of guest operating systems that use a separate /usr partition should now work more reliably (Pino Toscano). When parsing guest /etc/fstab, paths are reduced to a canonical form (eg. "///usr//local//" → "/usr/local"). Inspection of btrfs subvolumes now works where the /etc/fstab options field contains commas. /dev/cdN devices in guest /etc/fstab are ignored (Pino Toscano). Architectures and platforms Stable releases are now tested on aarch64, ppc64 and ppc64le architectures. The RISC-V architecture is now handled as a guest, and libguestfs will now at least compile on RISC-V (but probably not work as there is no usable qemu at the time of writing). Detection of S/390 and S/390x binaries and guests is now supported. Other ExFAT filesystems are now supported (Miles Wolbe). Security See also guestfs-security(1). There were no CVEs reported in this development cycle. However some security-related hardening was carried out as described below. Temporary filenames are now always generated using randomness from /dev/urandom (previously the C function random(3) was used in one case). The $TERM environment variable is now validated before passing it through to the appliance kernel command line. API New APIs "guestfs_aug_transform" Exposes the Augeas "aug_transform" API (Pino Toscano). "guestfs_find_inode" Find files by inode number (Matteo Cafasso). "guestfs_inspect_get_windows_software_hive" "guestfs_inspect_get_windows_system_hive" Return the path to the Windows "HKLM\SYSTEM" and "HKLM\SOFTWARE" hives computed during inspection. "guestfs_mksquashfs" Create a squashfs filesystem from a path (Pino Toscano). Other API changes "guestfs_add_domain" This call now handles libvirt file-based volumes correctly, where previously these would have been ignored. Also handled are disks which require libvirt authentication secrets to open (Pino Toscano). "guestfs_canonical_device_name" This call will no longer incorrectly modify Linux software RAID device names (like /dev/mdX). "guestfs_file_architecture" Previously the "file_architecture" API could return either of the strings "i386" or "i486" for 32 bit x86 binaries. It now only returns "i386" (as documented). This API can now return the following new values: "riscv32", "riscv64", "riscv128", "s390", "s390x". "guestfs_hivex_open" This now has an optional "GUESTFS_HIVEX_OPEN_UNSAFE" flag which allows certain corrupted Windows Registry hives to be opened. This feature requires hivex ≥ 1.3.14 (Dawid Zamirski). "guestfs_list_partitions" This call now returns Linux software RAID partitions. "guestfs_part_to_dev" This call now correctly handles partition names which include "p<N>" (Pino Toscano). "guestfs_set_label" This call can now change the labels of swap partitions (Pino Toscano). Build changes libmagic, the library part of the "file" command, is now required at build time (previously optional). GCC 7 is now supported. "Silent rules" are now used for OCaml programs, Java bindings. To show the full command line executed, add "V=1" on the make command line (Pino Toscano). Slow testing ("make check-slow") now covers: firstboot scripts in Linux guests; v2v conversion of a selection of real Linux guests; the virt- customize --hostname and --timezone settings; the --root-password parameter; that the serial console works in virt-builder guests. Large generated C source files, eg. the list of commands found in fish/cmds.c (and many more), have been split into smaller files to speed parallel compilation. "make maintainer-check-extra-dist" now checks that all generated files are included in the tarball. The tests no longer assume that "." is in Perl's @INC, as it is going to be removed soon (Pino Toscano). Debian hosts using UsrMerge are now supported (Pino Toscano). Header files and C structs can now have internal documentation using the special "/** ... */" comments. "@VAR@" subtitutions in "./run" are now fully quoted. This is necessary so that (eg) "./configure PYTHON=/some/path" works robustly if "/some/path" contains characters that need to be quoted (Hilko Bengen). gperf ≥ 3.1 is now supported. Kraxel's old edk2 builds can no longer be used for UEFI support. UEFI code is now fully free software, so use the versions bundled with your Linux distro instead. Virt-p2v can now be compiled on RHEL 5 (2007-era) Linux with Gtk 2.10. Internals The generator and mllib "Common_utils" modules are now shared from the same source file. A considerable amount of common code has been moved into the common directory in the source and is now compiled only once. The mini- libraries located under here are: common/edit, common/errnostring, common/miniexpect, common/options, common/parallel, common/progress, common/protocol, common/utils, common/visit, common/windows. The directory containing the main library code has moved from src → lib. All tests written in shell script now use a common file of utility functions (tests/test-functions.sh). There are several new utility functions, mainly for skipping tests. Also these test scripts can now use autoconf-like path variables like $abs_top_srcdir. UEFI paths are now stored in the generator (generator/uefi.ml). The way the generator handles actions and procedure numbers was changed quite substantially. See generator/actions_*.ml and generator/proc_nr.ml. The gnulib "getprogname" module is now used everywhere when needing/printing the program name (Pino Toscano). perl/Guestfs.c is not translatable (Nikos Skalkotos). Virt-builder templates moved from builder/website to builder/templates and there is now a single unified program which can build any template. All Windows registry utilities used by virt-customize and virt-v2v have been moved to a common module called "Registry" under mllib. All POSIX bindings have been moved to a new module called "Unix_utils" under mllib. Inspection, virt-customize and virt-v2v no longer recompute the Windows "%systemroot%", "CurrentControlSet" or paths to the "HKLM\SYSTEM" and "HKLM\SOFTWARE" hives in multiple places. Instead these are all computed once (during inspection) and passed to the other tools through various "guestfs_inspect_get_windows_*" APIs. "/dev/pts" is now available inside the appliance, so any tools we run which require a pty will now work (Pino Toscano). Most OCaml warnings have been fixed. There is now a single common function for creating temporary files ("guestfs_int_make_temp_path") (Matteo Cafasso). The $TERM environment variable is now validated before passing it through to the appliance kernel command line. Useless USB and memballoon devices are no longer created in the appliance (Laine Stump). On aarch64 we now use virtio-pci for the appliance. This is somewhat faster than virtio-mmio. Use of srandom(3) and random(3) has been minimized. In particular, temporary filenames are no longer created based on randomness returned by random(3), but /dev/urandom is used instead. Bugs fixed https://bugzilla.redhat.com/1425306 typo error in virt-tail man page https://bugzilla.redhat.com/1418283 virt-v2v: appliance runs out of memory running setfiles command https://bugzilla.redhat.com/1417549 /usr/bin/x86_64-linux-gnu-ld.bfd.real: ../common/progress/.libs/libprogress.a(libprogress_la-progress.o): undefined reference to symbol 'UP@@NCURSES_TINFO_5.0.19991023' https://bugzilla.redhat.com/1417444 *** No rule to make target '../perl/lib/Sys/Guestfs.c', needed by 'libguestfs.pot' https://bugzilla.redhat.com/1416941 compile of 1.34.3 fails with gperf 3.1 https://bugzilla.redhat.com/1414682 guestfs_canonical_device_name incorrectly returns /dev/sd0 for MD devices (/dev/md0) https://bugzilla.redhat.com/1414510 guestfs_list_filesystems does not recognize ddf partitions https://bugzilla.redhat.com/1409023 [Debian] ldmtool not installed in the appliance https://bugzilla.redhat.com/1404287 qemu-kvm cannot boot RHEL 7 kernel with TCG, hangs at "Probing EDD (edd=off to disable)..." https://bugzilla.redhat.com/1404182 RFE: virt-resize should support a URL as the outdisk https://bugzilla.redhat.com/1401474 Importing VMs from VMware is failing with error "Inspection field 'i_arch' was 'unknown'" https://bugzilla.redhat.com/1401320 RFE: Increate virt-sysprep coverage a bit https://bugzilla.redhat.com/1400205 Add --vdsm-compat=1.1 flag for VDSM https://bugzilla.redhat.com/1398070 typo error in man page https://bugzilla.redhat.com/1392798 secrets from libvirt domains are not read https://bugzilla.redhat.com/1390876 "--machine-readable" info should be updated in virt-v2v manual page https://bugzilla.redhat.com/1379289 RFE: virt-p2v should support mnemonic operations https://bugzilla.redhat.com/1378022 There is virt-v2v warning about <listen type='none'> during converting a guest which has listen type='none' in XML https://bugzilla.redhat.com/1377081 virt-p2v manual should update the new dialog information https://bugzilla.redhat.com/1375157 virt-v2v: -i ova: Permission denied when using libvirt and running as root https://bugzilla.redhat.com/1374651 Can't install qxl driver for display device in win7 guest after converting to glance by virt-v2v https://bugzilla.redhat.com/1374405 There is HTTP 404 error info when convert guest to glance by virt-v2v https://bugzilla.redhat.com/1374232 selinux relabel fails on RHEL 6.2 guests with "libguestfs error: selinux_relabel: : Success" https://bugzilla.redhat.com/1372668 Process status is not normal in windows guest after converted from kvm to rhev by virt-v2v https://bugzilla.redhat.com/1372269 Builder does not set hostname properly for Debian 8 (Jessie) https://bugzilla.redhat.com/1371843 Improve OVA import compatibility https://bugzilla.redhat.com/1370424 virt-manager coredump when vm with gluster image exists https://bugzilla.redhat.com/1367839 Cannot import VMs from Xen and VMware when using RHEL7.3 host. https://bugzilla.redhat.com/1367738 Missing bash completion scripts for: virt-diff guestunmount virt- copy-in virt-copy-out virt-customize virt-get-kernel virt-p2v-make-disk virt-p2v-make-kickstart virt-tar-in virt-tar-out virt-v2v-copy-to-local virt-win-reg https://bugzilla.redhat.com/1367615 OVMF file which is built for rhel7.3 can't be used for virt-v2v uefi conversion https://bugzilla.redhat.com/1366456 Converting rhel7 host installed on RAID:warning: fstrim: fstrim: /sysroot/: the discard operation is not supported https://bugzilla.redhat.com/1366049 RFE: libvirt backend: support handling disks stored as volume name in a pool https://bugzilla.redhat.com/1365005 Guest name is incorrect if convert guest from disk image by virt-v2v https://bugzilla.redhat.com/1362649 RFE: virt-sysprep does not utilize libguestfs encryption support https://bugzilla.redhat.com/1354507 virt-v2v conversions from vCenter do not consistently obey the proxy environment variables https://bugzilla.redhat.com/1168144 warning: fstrim: fstrim: /sysroot/: FITRIM ioctl failed: Operation not supported (ignored) when convert win2003 guest from xen server https://bugzilla.redhat.com/1161019 RFE: Only install RHEV-APT if virt-v2v -o rhev/-o vdsm option is used https://bugzilla.redhat.com/1152369 virt-v2v failed to convert RHEL 6.7 UEFI guest: no grub1/grub-legacy or grub2 configuration file was found https://bugzilla.redhat.com/1141631 [RFE] virt-v2v should support convert a guest to a dir-pool with using pool's uuid https://bugzilla.redhat.com/1134878 libvirt reports json "backing file" is missing https://bugzilla.redhat.com/1019388 firstboot scripts (virt-builder, virt-sysprep) don't work for Debian 6 & 7 guests https://bugzilla.redhat.com/737600 virt-v2v windows xp - machine dies BSOD - processr,sys - workaround provided -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com Fedora Windows cross-compiler. Compile Windows programs, test, and build Windows installers. Over 100 libraries supported. http://fedoraproject.org/wiki/MinGW

1 0

[libvirt] [PATCH] qemuDomainAttachNetDevice: Support attach of type="user"
by Michal Privoznik 28 Feb '17

28 Feb '17

https://bugzilla.redhat.com/show_bug.cgi?id=1420668 This has worked in previous releases. My commit c266b60440 broke it. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/qemu/qemu_hotplug.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 97fb272f6..0c0885695 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -1104,6 +1104,9 @@ qemuDomainAttachNetDevice(virQEMUDriverPtr driver, break; case VIR_DOMAIN_NET_TYPE_USER: + /* No preparation needed. */ + break; + case VIR_DOMAIN_NET_TYPE_SERVER: case VIR_DOMAIN_NET_TYPE_CLIENT: case VIR_DOMAIN_NET_TYPE_MCAST: -- 2.11.0

2 1

[libvirt] A question for external snapshots with flag“VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE”
by WangJie (Captain) 28 Feb '17

28 Feb '17

Hello, I got a question here. When we create consistent active external snapshots with flag “VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE” , the qemuDomainSnapshotFSFreeze will be called firstly to freeze all filesystems in vm, and then create snapshots. For windows vm, freezing filesystems used by vss service. If IO pressure in vm is too big or internal error happened in vss , freezing filesytems will failed , and qemuDomainSnapshotFSFreeze returns 0 which meaning that no filesystems are frozen. In the function qemuDomainSnapshotCreateActiveExternal, libvirt creates external snapshots all the same in such a situation that qemuDomainSnapshotFSFreeze returns 0, but the created snapshots are not consistent snapshots in such a situation . So shouldn't we abandon creating snapshots and goto cleanup in the situation that qemuDomainSnapshotFSFreeze returns 0? The code below: static int qemuDomainSnapshotCreateActiveExternal(virConnectPtr conn, virQEMUDriverPtr driver, virDomainObjPtr vm, virDomainSnapshotObjPtr snap, unsigned int flags) { virObjectEventPtr event; bool resume = false; int ret = -1; qemuDomainObjPrivatePtr priv = vm->privateData; char *xml = NULL; bool memory = snap->def->memory == VIR_DOMAIN_SNAPSHOT_LOCATION_EXTERNAL; bool memory_unlink = false; bool atomic = !!(flags & VIR_DOMAIN_SNAPSHOT_CREATE_ATOMIC); bool transaction = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_TRANSACTION); int thaw = 0; /* 1 if freeze succeeded, -1 if freeze failed */ bool pmsuspended = false; virQEMUDriverConfigPtr cfg = NULL; int compressed; char *compressedpath = NULL; /* If quiesce was requested, then issue a freeze command, and a * counterpart thaw command when it is actually sent to agent. * The command will fail if the guest is paused or the guest agent * is not running, or is already quiesced. */ if (flags & VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE) { int freeze = qemuDomainSnapshotFSFreeze(driver, vm, NULL, 0); if (freeze < 0) { /* the helper reported the error */ if (freeze == -2) thaw = -1; /* the command is sent but agent failed */ goto cleanup; } thaw = 1; } ...... ......

2 2

[libvirt] [PATCH 0/4] Cache systemd presence
by Ján Tomko 28 Feb '17

28 Feb '17

Reduce the number of DBus calls required, and possibly the number of timeouts. Ján Tomko (4): Unify checking for machine1 systemd service Use macros for testing virSystemd APIs Cache the presence of machine1 service Test virSystemd APIs twice to check the cache effects src/libvirt_private.syms | 1 + src/util/virsystemd.c | 55 +++++++++++++++++++++++++++++++++++------------ src/util/virsystemdpriv.h | 33 ++++++++++++++++++++++++++++ tests/virsystemdtest.c | 40 ++++++++++++++++++---------------- 4 files changed, 96 insertions(+), 33 deletions(-) create mode 100644 src/util/virsystemdpriv.h -- 2.10.2

3 6

[libvirt] [PATCH 0/8] Converge Storage Pool vHBA logic with Node Device
by John Ferlan 28 Feb '17

28 Feb '17

Similar to the recent node device focused vHBA patches, but these focus more on adjustments to the Storage Pool vHBA logic and finally the mechanism to define a vHBA for a domain. The first patch fixes a leak found by coverity that showed up because node_device_conf had enough changes so that coverity looked harder... The second patch creates a mechanism to mock creation of the vHBA in order to test the ability for the storage pool to create a vHBA. As much as I dislike forward refs for testNodeDeviceMockCreateVport, it was better than moving all the code.... The third patch extracts out storage device mgmt into it's own set of src/util API's - similar to the existing virstoragefile, but for devices. The fourth patch was uncovered while moving code from storage_backend_scsi into node_device_conf (the fifth patch)... The fifth patch moves the createVport/deleteVport guts into the node_device_conf (although they could have moved to virvhba)... The sixth patch alters the logic to use the node_device API's as the "preferred" mechanism to create/delete the vport... The seventh patch tests the storage pool vHBA creation algorithms. The eigth patch is the reason for all this stirring of the pot. Alter the domain <controller> XML in order to allow definition of a vHBA which more or less sits between a "scsi_hostX" host device and a controller. This is in preparation for https://bugzilla.redhat.com/show_bug.cgi?id=1404962 which can take that created vHBA and automagically add the LUNs from the vHBA to the domain although that requires a bit more magic for which there are already onlist patches to let qemu driver know when a node device has been added/removed. Once all that's in place - the next step will be to converge the two sets of patches. It's a chicken/egg type problem - one has to exist before the other can truly work. John Ferlan (8): conf: Fix leak in virNodeDeviceDefParseXML tests: Add createVHBAByStoragePool-by-parent to fchosttest util: Convert virStoragePoolSourceAdapter to virStorageAdapter util: Rename virFileWaitForDevices storage: Move/rename createVport and deleteVport util: Alter virNodeDevice{Create|Delete}Vport to use nodedev APIs tests: Add more storage pool vHBA tests conf: Add vHBA controller definition to domain docs/schemas/basictypes.rng | 66 ++-- docs/schemas/domaincommon.rng | 12 +- po/POTFILES.in | 1 + src/Makefile.am | 1 + src/conf/domain_audit.c | 32 ++ src/conf/domain_conf.c | 180 ++++++++++- src/conf/domain_conf.h | 2 + src/conf/node_device_conf.c | 342 ++++++++++++++++++++- src/conf/node_device_conf.h | 9 + src/conf/storage_conf.c | 338 +++++--------------- src/conf/storage_conf.h | 35 +-- src/libvirt_private.syms | 20 +- src/libxl/libxl_conf.c | 1 + src/node_device/node_device_driver.c | 2 +- src/phyp/phyp_driver.c | 3 +- src/qemu/qemu_alias.c | 5 + src/qemu/qemu_command.c | 4 + src/qemu/qemu_hotplug.c | 16 + src/storage/storage_backend_disk.c | 6 +- src/storage/storage_backend_iscsi.c | 2 +- src/storage/storage_backend_logical.c | 4 +- src/storage/storage_backend_mpath.c | 2 +- src/storage/storage_backend_scsi.c | 267 +++------------- src/storage/storage_util.c | 2 +- src/test/test_driver.c | 101 +++++- src/util/virfile.h | 2 - src/util/virscsihost.c | 28 +- src/util/virscsihost.h | 8 +- src/util/virstoragedevice.c | 292 ++++++++++++++++++ src/util/virstoragedevice.h | 89 ++++++ src/util/virutil.c | 4 +- src/util/virutil.h | 2 + tests/fchosttest.c | 111 +++++++ .../qemuxml2argv-vhba-no-parent.xml | 38 +++ .../qemuxml2argv-vhba-parent-fabric.xml | 38 +++ .../qemuxml2argv-vhba-parent-name.xml | 38 +++ .../qemuxml2argv-vhba-parent-wwns.xml | 38 +++ .../qemuxml2xmlout-vhba-no-parent.xml | 47 +++ .../qemuxml2xmlout-vhba-parent-fabric.xml | 47 +++ .../qemuxml2xmlout-vhba-parent-name.xml | 47 +++ .../qemuxml2xmlout-vhba-parent-wwns.xml | 47 +++ tests/qemuxml2xmltest.c | 9 + 42 files changed, 1727 insertions(+), 611 deletions(-) create mode 100644 src/util/virstoragedevice.c create mode 100644 src/util/virstoragedevice.h create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-vhba-no-parent.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-vhba-parent-fabric.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-vhba-parent-name.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-vhba-parent-wwns.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-vhba-no-parent.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-vhba-parent-fabric.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-vhba-parent-name.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-vhba-parent-wwns.xml -- 2.9.3

3 24

[libvirt] [PATCH go v2] domain.go: wrong calculation of bitmask in PinVcpuFlags()
by Leonid Podolny 28 Feb '17

28 Feb '17

--- domain.go | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/domain.go b/domain.go index 2bd9852..7276ef8 100644 --- a/domain.go +++ b/domain.go @@ -1768,7 +1768,7 @@ func (d *Domain) PinVcpuFlags(vcpu uint, cpuMap []bool, flags DomainModification ccpumap := make([]C.uchar, maplen) for i := 0; i < len(cpuMap); i++ { if cpuMap[i] { - byte := (i + 7) / 8 + byte := i / 8 bit := i % 8 ccpumap[byte] |= (1 << uint(bit)) } -- 2.11.1

2 1