- Devel - libvirt List Archives

[libvirt] [PATCH] Fix compiler warning while un-wraping python object to uint in libvirt_virNodeGetFreePages
by Nehal J Wani 26 Jun '14

26 Jun '14

Introduced by commit c8ba859bc78f5e3b0ed2590f4e8a7fa6c26ac08f On compiling libvirt-python against libvirt 1.2.6, we get: libvirt-override.c: In function ‘libvirt_virNodeGetFreePages’: libvirt-override.c:7811:9: warning: pointer targets in passing argument 2 of ‘libvirt_intUnwrap’ differ in signedness [-Wpointer-sign] if (libvirt_intUnwrap(tmp, &pages[i]) < 0) ^ In file included from libvirt-override.c:24:0: typewrappers.h:169:5: note: expected ‘int *’ but argument is of type ‘unsigned int *’ int libvirt_intUnwrap(PyObject *obj, int *val); ^ --- libvirt-override.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libvirt-override.c b/libvirt-override.c index 04fb4cd..40aefcc 100644 --- a/libvirt-override.c +++ b/libvirt-override.c @@ -7808,7 +7808,7 @@ libvirt_virNodeGetFreePages(PyObject *self ATTRIBUTE_UNUSED, for (i = 0; i < pyobj_pagesize_size; i++) { PyObject *tmp = PyList_GetItem(pyobj_pagesize, i); - if (libvirt_intUnwrap(tmp, &pages[i]) < 0) + if (libvirt_uintUnwrap(tmp, &pages[i]) < 0) goto cleanup; } -- 1.9.3

2 1

[libvirt] [PATCHv2 0/3] LXC drop/keep capabilities feature
by Cédric Bosdonnat 25 Jun '14

25 Jun '14

Changes since the previous version are: * add 'b *:* m' and 'c *:* m' to the device white list is CAP_MKNOD is requested instead of setting 'a *:* rwm'. * Add a policy='default|allow|deny' for the features/capabilities element. Cédric Bosdonnat (3): lxc: allow to keep or drop capabilities lxc domain from xml: convert lxc.cap.drop lxc: update doc to mention features/capabilities/* domain configuration docs/drvlxc.html.in | 47 +++++ docs/schemas/domaincommon.rng | 207 +++++++++++++++++++++ src/conf/domain_conf.c | 126 ++++++++++++- src/conf/domain_conf.h | 56 ++++++ src/libvirt_private.syms | 3 + src/lxc/lxc_cgroup.c | 8 + src/lxc/lxc_container.c | 123 ++++++++++-- src/lxc/lxc_native.c | 25 +++ src/util/vircgroup.c | 74 +++++++- src/util/vircgroup.h | 2 + tests/domainschemadata/domain-caps-features.xml | 28 +++ tests/lxcconf2xmldata/lxcconf2xml-blkiotune.xml | 2 + tests/lxcconf2xmldata/lxcconf2xml-cpusettune.xml | 2 + tests/lxcconf2xmldata/lxcconf2xml-cputune.xml | 2 + tests/lxcconf2xmldata/lxcconf2xml-idmap.xml | 2 + .../lxcconf2xmldata/lxcconf2xml-macvlannetwork.xml | 4 + tests/lxcconf2xmldata/lxcconf2xml-memtune.xml | 2 + tests/lxcconf2xmldata/lxcconf2xml-nonenetwork.xml | 4 + tests/lxcconf2xmldata/lxcconf2xml-nonetwork.xml | 2 + tests/lxcconf2xmldata/lxcconf2xml-physnetwork.xml | 4 + tests/lxcconf2xmldata/lxcconf2xml-simple.xml | 8 + tests/lxcconf2xmldata/lxcconf2xml-vlannetwork.xml | 4 + 22 files changed, 710 insertions(+), 25 deletions(-) create mode 100644 tests/domainschemadata/domain-caps-features.xml -- 1.8.4.5

1 3

[libvirt] [PATCH 0/8] storage_scsi: Stable SCSI host addressing
by John Ferlan 25 Jun '14

25 Jun '14

Just over a year ago, Osier Yang submitted some patches to provide stable SCSI host addressing support: http://www.redhat.com/archives/libvir-list/2013-June/msg00396.html Although reviewed - none of the patches were ever pushed and Osier never got back to the work. I eventually inherited the changes and now after languishing in the todo list - I took the time to rework things. There is a bz associated: https://bugzilla.redhat.com/show_bug.cgi?id=963817 Prior to leaving Red Hat, Osier and I spent some time revisiting the patch series and hashing out new/different ideas to come up with the same answer. My focus was to avoid the "generic recursive directory search" (patch 3), to not require having to know whether the scsi_host was using udev or hal, and to simplify as much as possible using existing data/information. The generic directory search I believe was less generic than intended and over complicated things. The old patches essentially used the output of a 'virsh nodedev-list scsi_host' and then selecting a resulting entry ran a 'virsh nodedev-dumpxml' to grab/use the resulting <parent> value. Taking that value and optionally adding a unique_id value was the basis for the design that would generate a PCI address (either in udev or hal format) and recursively search through /sys/devices looking for a matching address in either udev or hal format. These patches replace the bulk of the directory traversal logic with a more direct (and already in use) approach to scan the /sys/class/scsi_host directories looking for a matching PCI address found in the symlink of the files in the directory. The changed logic will add a new XML element 'parentaddr' to describe the scsi_host by it's PCI address. Additionally, the 'unique_id' has become a required attribute. The code reuses the virDevicePCIAddressParseXML() in order parse the required 'address' element. The 'address' will be in the expect PCI Address format like other described host devices. In order to help view the required unique_id value, the nodedev-dumpxml output has been adjusted to provide the <unique_id> value. This value will be an optional value on input. The documentation is updated to describe how to generate the address from the nodedev-dumpxml output. The new scsihosttest creates the expected PCI infrastructure on the fly since adding files with colons (':') is prohibited. There are multiple directories and hosts within each to ensure the search logic worked as expected. Reviewer notes: Patch 1 is new - it's just forcing the specific adapter.type checking Patches 2 & 3 are the former patches 1 & 2 with some edits to adjust for new XML syntax (and changed associated structure) Patch 4 is new - it's just utilizing the 'LINUX_SYSFS_SCSI_HOST_PREFIX' definition rather than using the hardcoded value Patches 5 & 6 are new. They handle the optional unique_id value (including using the new virNodeDevCapsDefParseIntOptional()) Patch 7 is similar to the former patch 5 insomuch as it's where the comparison of the PCI directory path and unique_id file is made. Patch 8 is similar to the former patch 7 insomuch as it's where the link is made between the scsi_host host# and the loading/refreshing of the scsi_host adapter. Former patches 3, 4, 6, and 8-11 are no longer used Please check my XML schema (patch 3) - I think I figured out the right syntax to use regarding using either "<name>" or "<parentaddr>" where "<parentaddr>" is an element. The syntax passes the make check, but it's not an area of expertise for me. Using parentaddr was preferred over overloading the parent attribute. John Ferlan (6): getAdapterName: check for SCSI_HOST scsi_backend: Use existing LINUX_SYSFS_SCSI_HOST_PREFIX definition virutil: Introduce virReadSCSIUniqueId Add unique_id to nodedev output scsi_host: Introduce virFindSCSIHostByPCI getAdapterName: Lookup stable scsi_host Osier Yang (2): virStoragePoolSourceAdapter: Refine the SCSI_HOST adapter name storage: Introduce parentaddr into virStoragePoolSourceAdapter docs/formatnode.html.in | 11 + docs/formatstorage.html.in | 130 +++++++-- docs/schemas/basictypes.rng | 24 +- docs/schemas/nodedev.rng | 6 + src/conf/node_device_conf.c | 23 +- src/conf/node_device_conf.h | 1 + src/conf/storage_conf.c | 111 +++++++- src/conf/storage_conf.h | 8 +- src/libvirt_private.syms | 2 + src/node_device/node_device_linux_sysfs.c | 6 + src/phyp/phyp_driver.c | 8 +- src/storage/storage_backend_scsi.c | 53 +++- src/test/test_driver.c | 5 +- src/util/virutil.c | 154 +++++++++++ src/util/virutil.h | 8 + tests/Makefile.am | 7 + .../pci_8086_27c5_scsi_host_0_unique_id.xml | 8 + tests/nodedevxml2xmltest.c | 1 + tests/scsihosttest.c | 308 +++++++++++++++++++++ .../pool-scsi-type-scsi-host-stable.xml | 19 ++ .../pool-scsi-type-scsi-host-stable.xml | 22 ++ tests/storagepoolxml2xmltest.c | 1 + 22 files changed, 855 insertions(+), 61 deletions(-) create mode 100644 tests/nodedevschemadata/pci_8086_27c5_scsi_host_0_unique_id.xml create mode 100644 tests/scsihosttest.c create mode 100644 tests/storagepoolxml2xmlin/pool-scsi-type-scsi-host-stable.xml create mode 100644 tests/storagepoolxml2xmlout/pool-scsi-type-scsi-host-stable.xml -- 1.9.3

2 15

[libvirt] plan for libvirt-1.2.6 next release
by Daniel Veillard 25 Jun '14

25 Jun '14

In a nutshell, I went on vacations and forgot the freeze, oops sorry ! :-) So I'm suggesting to enter freeze for 1.2.6 tomorrow, then try to release still around the 1st next week, sorry for the late announcement, Daniel -- Daniel Veillard | Open Source and Standards, Red Hat veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

1 0

[libvirt] [PATCH] QEMU: Fix virtio-serial device support on PPC.
by Olivia Yin 25 Jun '14

25 Jun '14

Patch fd0f22ed414e114664f4140306834bedd19f52c9 discard virtio-serial device support on PPC machine. Most PPC boards don't use -chardev with non-virtio serial device except pseries which uses -device spapr-vty with -chardev. Signed-off-by: Olivia Yin <Hong-Hua.Yin(a)freescale.com> --- src/qemu/qemu_capabilities.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 8e0a550..4541f52 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -3493,19 +3493,18 @@ virQEMUCapsSupportsChardev(virDomainDefPtr def, !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE)) return false; - if ((def->os.arch == VIR_ARCH_PPC) || (def->os.arch == VIR_ARCH_PPC64)) { - /* only pseries need -device spapr-vty with -chardev */ - return (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL && - chr->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO); - } - - if ((def->os.arch != VIR_ARCH_ARMV7L) && (def->os.arch != VIR_ARCH_AARCH64)) + if ((def->os.arch != VIR_ARCH_ARMV7L) && (def->os.arch != VIR_ARCH_AARCH64) && + (def->os.arch != VIR_ARCH_PPC) && (def->os.arch != VIR_ARCH_PPC64)) return true; /* This may not be true for all ARM machine types, but at least * the only supported non-virtio serial devices of vexpress and versatile * don't have the -chardev property wired up. */ + /* Currently PPC boards don't use -chardev for non-virtio serial device + * except pseries which uses -device spapr-vty with -chardev. */ return (chr->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_MMIO || (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_CONSOLE && - chr->targetType == VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO)); + chr->targetType == VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO) || + (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL && + chr->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO)); } -- 1.8.5

3 2

[libvirt] [PATCH 0/3] add binding guest numa nodes to host numa nodes support
by Chen Fan 25 Jun '14

25 Jun '14

Since qemu has supported numa option -memdev config: -object memory-backend-ram,size=1024M,policy=membind,host-nodes=0,id=ram-node0 \ -numa node,nodeid=0,cpus=0,memdev=ram-node0 \ for binding guest numa nodes to host numa nodes. So we introduce this capability in libvirt by configuration domain XML like: ... <cpu> <numa> <cell cpus='0-1' memdev='ram0'/> </numa> </cpu> ... <devices> <memdev type='ram' merge='yes' dump='yes' prealloc='yes'> <name>ram0</name> <capacity unit='MiB'>1000</capacity> <source host-nodes='0-1' policy='bind' /> </memdev> </devices> Chen Fan (3): numa: add '-numa memdev=' support conf: add memdev device in <devices> XML tests: add numa -memdev testing and docs support docs/formatdomain.html.in | 71 ++++++- docs/schemas/domaincommon.rng | 76 +++++++- src/conf/cpu_conf.c | 73 ++++++-- src/conf/cpu_conf.h | 13 +- src/conf/domain_conf.c | 203 ++++++++++++++++++++- src/conf/domain_conf.h | 42 +++++ src/libvirt_private.syms | 4 + src/qemu/qemu_capabilities.c | 4 + src/qemu/qemu_capabilities.h | 2 + src/qemu/qemu_command.c | 84 ++++++++- src/qemu/qemu_command.h | 4 + src/qemu/qemu_hotplug.c | 1 + tests/qemuxml2argvdata/qemuxml2argv-cpu-numa3.args | 9 + tests/qemuxml2argvdata/qemuxml2argv-cpu-numa3.xml | 35 ++++ tests/qemuxml2argvdata/qemuxml2argv-cpu-numa4.args | 10 + tests/qemuxml2argvdata/qemuxml2argv-cpu-numa4.xml | 35 ++++ tests/qemuxml2argvtest.c | 2 + 17 files changed, 644 insertions(+), 24 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-numa3.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-numa3.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-numa4.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-cpu-numa4.xml -- 1.9.3

2 4

[libvirt] [libvirt-glib] [PATCH 0/2] Add API to fetch the snapshots of a GVirDomain
by Timm Bäder 25 Jun '14

25 Jun '14

The following two patches add gvir_domain_fetch_snapshots which uses virDomainListAllSnapshots to fetch snapshots from the virDomain (using the given GVirDomainSnapshotListFlags) and gvir_domain_get_snapshots which returns a GList of containing all of the snapshots last fetched. Timm Bäder (2): libvirt-gobject-domain: Add _fetch_snapshots libvirt-gobject-domain: Add _get_snapshots libvirt-gobject/libvirt-gobject-domain.c | 80 ++++++++++++++++++++++++++++++++ libvirt-gobject/libvirt-gobject-domain.h | 40 ++++++++++++++++ libvirt-gobject/libvirt-gobject.sym | 3 ++ 3 files changed, 123 insertions(+) -- 2.0.0

2 4

[libvirt] [PATCH] libxl: prefer qdisk for <driver name='file'>
by Jim Fehlig 25 Jun '14

25 Jun '14

The libxl driver currently sets the disk backend to LIBXL_DISK_BACKEND_TAP when <driver name='file'> is specified in the <disk> config. qdisk should be prefered with this configuration, otherwise existing configuration such as the following, which worked with the old Xen driver, will not work with the libxl driver <disk type='file' device='cdrom'> <driver name='file'/> <source file='/path/to/some/iso'/> <target dev='hdc' bus='ide'/> <readonly/> </disk> In addition, tap performs poorly compared to qdisk. Signed-off-by: Jim Fehlig <jfehlig(a)suse.com> --- src/libxl/libxl_conf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c index fdbb522..4b6b5c0 100644 --- a/src/libxl/libxl_conf.c +++ b/src/libxl/libxl_conf.c @@ -796,7 +796,7 @@ libxlMakeDisk(virDomainDiskDefPtr l_disk, libxl_device_disk *x_disk) return -1; } x_disk->format = LIBXL_DISK_FORMAT_RAW; - x_disk->backend = LIBXL_DISK_BACKEND_TAP; + x_disk->backend = LIBXL_DISK_BACKEND_QDISK; } else if (STREQ(driver, "phy")) { if (format != VIR_STORAGE_FILE_NONE && format != VIR_STORAGE_FILE_RAW) { -- 1.8.4.5

3 5

[libvirt] [PATCH v4] Only set SELinux seclabel if supported by the host.
by Cédric Bosdonnat 25 Jun '14

25 Jun '14

This code depends on new API in libvirt-gconfig to extract the secmodels handled by the host. --- Diff to v3: * Added yet another missing g_object_unref. * Fixed the logic for supportsSelinux libvirt-sandbox/libvirt-sandbox-builder.c | 49 +++++++++++++++++++++++++++---- 1 file changed, 43 insertions(+), 6 deletions(-) diff --git a/libvirt-sandbox/libvirt-sandbox-builder.c b/libvirt-sandbox/libvirt-sandbox-builder.c index 48b3acc..d6b5735 100644 --- a/libvirt-sandbox/libvirt-sandbox-builder.c +++ b/libvirt-sandbox/libvirt-sandbox-builder.c @@ -322,12 +322,10 @@ static gboolean gvir_sandbox_builder_construct_devices(GVirSandboxBuilder *build return TRUE; } - -static gboolean gvir_sandbox_builder_construct_security(GVirSandboxBuilder *builder G_GNUC_UNUSED, - GVirSandboxConfig *config G_GNUC_UNUSED, - const gchar *statedir G_GNUC_UNUSED, - GVirConfigDomain *domain, - GError **error G_GNUC_UNUSED) +static gboolean gvir_sandbox_builder_construct_security_selinux (GVirSandboxBuilder *builder, + GVirSandboxConfig *config, + GVirConfigDomain *domain, + GError **error) { GVirConfigDomainSeclabel *sec = gvir_config_domain_seclabel_new(); const char *label = gvir_sandbox_config_get_security_label(config); @@ -360,6 +358,45 @@ static gboolean gvir_sandbox_builder_construct_security(GVirSandboxBuilder *buil return TRUE; } +static gboolean gvir_sandbox_builder_construct_security(GVirSandboxBuilder *builder, + GVirSandboxConfig *config, + const gchar *statedir G_GNUC_UNUSED, + GVirConfigDomain *domain, + GError **error) +{ + GVirConnection *connection = gvir_sandbox_builder_get_connection(builder); + GVirConfigCapabilities *configCapabilities; + GVirConfigCapabilitiesHost *hostCapabilities; + GList *secmodels, *iter; + gboolean supportsSelinux = FALSE; + + /* What security models are available on the host? */ + if (!(configCapabilities = gvir_connection_get_capabilities(connection, error))) { + return FALSE; + } + + hostCapabilities = gvir_config_capabilities_get_host(configCapabilities); + + secmodels = gvir_config_capabilities_host_get_secmodels(hostCapabilities); + for (iter = secmodels; iter != NULL; iter = iter->next) { + if (g_str_equal(gvir_config_capabilities_host_secmodel_get_model( + GVIR_CONFIG_CAPABILITIES_HOST_SECMODEL(iter->data)), "selinux")) + supportsSelinux = TRUE; + g_object_unref(iter->data); + } + + g_list_free(secmodels); + g_object_unref(hostCapabilities); + g_object_unref(configCapabilities); + g_object_unref(connection); + + if (supportsSelinux) + return gvir_sandbox_builder_construct_security_selinux(builder, config, + domain, error); + + return TRUE; +} + static gboolean gvir_sandbox_builder_clean_post_start_default(GVirSandboxBuilder *builder G_GNUC_UNUSED, GVirSandboxConfig *config G_GNUC_UNUSED, -- 1.8.4.5

3 3

[libvirt] [PATCH] LXC: check whether we get MemSwap[Total|Usage]
by Chen Hanxiao 25 Jun '14

25 Jun '14

Signed-off-by: Chen Hanxiao <chenhanxiao(a)cn.fujitsu.com> --- src/lxc/lxc_cgroup.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/src/lxc/lxc_cgroup.c b/src/lxc/lxc_cgroup.c index 8dfdc60..39e30ad 100644 --- a/src/lxc/lxc_cgroup.c +++ b/src/lxc/lxc_cgroup.c @@ -288,8 +288,11 @@ int virLXCCgroupGetMeminfo(virLXCMeminfoPtr meminfo) if (virLXCCgroupGetMemUsage(cgroup, meminfo) < 0) goto cleanup; - virLXCCgroupGetMemSwapTotal(cgroup, meminfo); - virLXCCgroupGetMemSwapUsage(cgroup, meminfo); + if (virLXCCgroupGetMemSwapTotal(cgroup, meminfo) < 0) + goto cleanup; + + if (virLXCCgroupGetMemSwapUsage(cgroup, meminfo) < 0) + goto cleanup; ret = 0; cleanup: -- 1.9.0

2 1