- Devel - libvirt List Archives

[libvirt] [PATCH v2] Add 'lxc-enter-namespace' command to virsh
by Daniel P. Berrange 15 Jan '13

15 Jan '13

From: "Daniel P. Berrange" <berrange(a)redhat.com> Add a 'lxc-enter-namespace' command which accepts a domain name and then a command + args to run, attached to the container eg virsh -c lxc:/// lxc-enter-namespace demo -- /bin/ps -auxf Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- tools/Makefile.am | 1 + tools/virsh-domain.c | 94 ++++++++++++++++++++++++++++++++++++++++++++++++++++ tools/virsh.pod | 8 +++++ 3 files changed, 103 insertions(+) diff --git a/tools/Makefile.am b/tools/Makefile.am index 878a114..8a0429c 100644 --- a/tools/Makefile.am +++ b/tools/Makefile.am @@ -133,6 +133,7 @@ virsh_LDADD = \ $(STATIC_BINARIES) \ $(WARN_CFLAGS) \ ../src/libvirt.la \ + ../src/libvirt-lxc.la \ ../src/libvirt-qemu.la \ ../gnulib/lib/libgnu.la \ $(LIBXML_LIBS) \ diff --git a/tools/virsh-domain.c b/tools/virsh-domain.c index e91939c..1fee681 100644 --- a/tools/virsh-domain.c +++ b/tools/virsh-domain.c @@ -43,11 +43,13 @@ #include "conf/domain_conf.h" #include "console.h" #include "viralloc.h" +#include "vircommand.h" #include "virutil.h" #include "virfile.h" #include "virjson.h" #include "virkeycode.h" #include "virmacaddr.h" +#include "virprocess.h" #include "virstring.h" #include "virsh-domain-monitor.h" #include "virerror.h" @@ -6771,6 +6773,97 @@ cleanup: } /* + * "lxc-enter-namespace" namespace + */ +static const vshCmdInfo info_lxc_enter_namespace[] = { + {"help", N_("LXC Guest Enter Namespace")}, + {"desc", N_("Run an arbitrary lxc guest enter namespace; use at your own risk")}, + {NULL, NULL} +}; + +static const vshCmdOptDef opts_lxc_enter_namespace[] = { + {"domain", VSH_OT_DATA, VSH_OFLAG_REQ, N_("domain name, id or uuid")}, + {NULL, 0, 0, NULL} +}; + +static bool +cmdLxcEnterNamespace(vshControl *ctl, const vshCmd *cmd) +{ + virDomainPtr dom = NULL; + bool ret = false; + const vshCmdOpt *opt = NULL; + char **cmdargv = NULL; + size_t ncmdargv = 0; + pid_t pid; + int nfdlist; + int *fdlist; + size_t i; + + dom = vshCommandOptDomain(ctl, cmd, NULL); + if (dom == NULL) + goto cleanup; + + while ((opt = vshCommandOptArgv(cmd, opt))) { + if (VIR_EXPAND_N(cmdargv, ncmdargv, 1) < 0) { + vshError(ctl, _("%s: %d: failed to allocate argv"), + __FILE__, __LINE__); + } + cmdargv[ncmdargv-1] = opt->data; + } + if (VIR_EXPAND_N(cmdargv, ncmdargv, 1) < 0) { + vshError(ctl, _("%s: %d: failed to allocate argv"), + __FILE__, __LINE__); + } + cmdargv[ncmdargv - 1] = NULL; + + if ((nfdlist = virDomainLxcOpenNamespace(dom, &fdlist, 0)) < 0) + goto cleanup; + + /* Fork once because we don't want to affect + * virsh's namespace itself + */ + if (virFork(&pid) < 0) + goto cleanup; + if (pid == 0) { + if (virDomainLxcEnterNamespace(dom, + nfdlist, + fdlist, + NULL, + NULL, + 0) < 0) + _exit(255); + + /* Fork a second time because entering the + * pid namespace only takes effect after fork + */ + if (virFork(&pid) < 0) + _exit(255); + if (pid == 0) { + execv(cmdargv[0], cmdargv); + _exit(255); + } else { + if (virProcessWait(pid, NULL) < 0) + _exit(255); + } + _exit(0); + } else { + for (i = 0 ; i < nfdlist ; i++) + VIR_FORCE_CLOSE(fdlist[i]); + VIR_FREE(fdlist); + if (virProcessWait(pid, NULL) < 0) + goto cleanup; + } + + ret = true; + +cleanup: + if (dom) + virDomainFree(dom); + VIR_FREE(cmdargv); + return ret; +} + +/* * "dumpxml" command */ static const vshCmdInfo info_dumpxml[] = { @@ -8760,6 +8853,7 @@ const vshCmdDef domManagementCmds[] = { {"inject-nmi", cmdInjectNMI, opts_inject_nmi, info_inject_nmi, 0}, {"send-key", cmdSendKey, opts_send_key, info_send_key, 0}, {"send-process-signal", cmdSendProcessSignal, opts_send_process_signal, info_send_process_signal, 0}, + {"lxc-enter-namespace", cmdLxcEnterNamespace, opts_lxc_enter_namespace, info_lxc_enter_namespace, 0}, {"managedsave", cmdManagedSave, opts_managedsave, info_managedsave, 0}, {"managedsave-remove", cmdManagedSaveRemove, opts_managedsaveremove, info_managedsaveremove, 0}, diff --git a/tools/virsh.pod b/tools/virsh.pod index c9ebc8f..e2a2aec 100644 --- a/tools/virsh.pod +++ b/tools/virsh.pod @@ -3056,6 +3056,14 @@ When I<--aysnc> is given, the command waits for timeout whether success or failed. And when I<--block> is given, the command waits forever with blocking timeout. +=item B<lxc-enter-namespace> I<domain> -- /path/to/binary [arg1, [arg2, ...]] + +Enter the namespace of I<domain> and execute the command C</path/to/binary> +passing the requested args. The binary path is relative to the container +root filesystem, not the host root filesystem. The binary will inherit the +environment variables / console visible to virsh. This command only works +when connected to the LXC hypervisor driver. + =back =head1 ENVIRONMENT -- 1.7.11.7

2 1

[libvirt] [PATCH] build: add new file, for lxc_protocol checking
by Eric Blake 15 Jan '13

15 Jan '13

Commit 509eb51 added lxc_protocol.x; but without the initial checkin of src/lxc_protocol-structs, 'make check' would fail for anyone with pdwtags installed: make[3]: *** No rule to make target `lxc_protocol-structs', needed by `check-protocol'. Stop. * src/lxc_protocol-structs: New file. --- Pushing under the build-breaker rule. src/lxc_protocol-structs | 13 +++++++++++++ 1 file changed, 13 insertions(+) create mode 100644 src/lxc_protocol-structs diff --git a/src/lxc_protocol-structs b/src/lxc_protocol-structs new file mode 100644 index 0000000..eb54172 --- /dev/null +++ b/src/lxc_protocol-structs @@ -0,0 +1,13 @@ +/* -*- c -*- */ +struct remote_nonnull_domain { + remote_nonnull_string name; + remote_uuid uuid; + int id; +}; +struct lxc_domain_open_namespace_args { + remote_nonnull_domain dom; + u_int flags; +}; +enum lxc_procedure { + LXC_PROC_DOMAIN_OPEN_NAMESPACE = 1, +}; -- 1.8.0.2

2 1

[libvirt] [PATCH 0/8] *Persistent vHBA support in storage pool
by Osier Yang 15 Jan '13

15 Jan '13

This is the 3rd part to implement NPIV migration support [1]. Part 1: https://www.redhat.com/archives/libvir-list/2013-January/msg00859.html Part 2: (Already ACKed by Michal) https://www.redhat.com/archives/libvir-list/2013-January/msg00859.html The patches are based on Part 2. The new XMLs might be too long, might be deserved to have them as sub-elements of <adapter>. But I'd like to see the feedback first. Osier Yang (8): New XML attributes for storage pool source adapter storage: Make the adapter name be consistent with node device driver storage: Move virStorageBackendSCSIGetHostNumber into iscsi backend phyp: Prohibit fc_host adapter for phyp driver util: Add helper to get the scsi host name by iterating over sysfs util: Fix bug of managing vport storage: Add startPool and stopPool for scsi backend storage: Guess the parent if it's not specified for vHBA docs/formatstorage.html.in | 15 ++- docs/schemas/storagepool.rng | 33 +++++- src/conf/storage_conf.c | 123 ++++++++++++++++-- src/conf/storage_conf.h | 23 +++- src/libvirt_private.syms | 4 + src/phyp/phyp_driver.c | 15 ++- src/storage/storage_backend_iscsi.c | 39 ++++++- src/storage/storage_backend_scsi.c | 190 +++++++++++++++++++-------- src/storage/storage_backend_scsi.h | 3 - src/util/virutil.c | 196 +++++++++++++++++++++++++++- src/util/virutil.h | 7 + tests/storagepoolxml2xmlin/pool-scsi.xml | 2 +- tests/storagepoolxml2xmlin/pool-scsi1.xml | 15 ++ tests/storagepoolxml2xmlout/pool-scsi.xml | 2 +- tests/storagepoolxml2xmlout/pool-scsi1.xml | 18 +++ tests/storagepoolxml2xmltest.c | 1 + 16 files changed, 602 insertions(+), 84 deletions(-) create mode 100644 tests/storagepoolxml2xmlin/pool-scsi1.xml create mode 100644 tests/storagepoolxml2xmlout/pool-scsi1.xml [1] https://www.redhat.com/archives/libvir-list/2012-November/msg00826.html Regards, Osier

1 8

[libvirt] [PATCH 0/2] Add total number of physical cores to the capabilities output
by Peter Krempa 15 Jan '13

15 Jan '13

https://bugzilla.redhat.com/show_bug.cgi?id=888503 The values added by this patch help avoid management applications the dodgy approach of multiplying values in the nodeinfo structure to gather data needed for their management decisions. Peter Krempa (2): nodeinfo: Gather total number of physical cores and fix tests capabilities: Add total number of cores and threads to the XML docs/formatcaps.html.in | 2 +- docs/schemas/capability.rng | 8 +++++++ src/conf/cpu_conf.c | 4 ++++ src/conf/cpu_conf.h | 2 ++ src/libvirt_private.syms | 1 + src/nodeinfo.c | 37 ++++++++++++++++++++++------- src/nodeinfo.h | 2 ++ src/qemu/qemu_capabilities.c | 3 ++- tests/nodeinfodata/linux-x86-test1.expected | 2 +- tests/nodeinfodata/linux-x86-test2.expected | 2 +- tests/nodeinfodata/linux-x86-test3.expected | 2 +- tests/nodeinfodata/linux-x86-test4.expected | 2 +- tests/nodeinfodata/linux-x86-test5.expected | 2 +- tests/nodeinfodata/linux-x86-test6.expected | 2 +- tests/nodeinfodata/linux-x86-test7.expected | 2 +- tests/nodeinfodata/linux-x86-test8.expected | 2 +- tests/nodeinfotest.c | 11 +++++---- tests/testutilsqemu.c | 2 ++ 18 files changed, 66 insertions(+), 22 deletions(-) -- 1.8.1

2 3

[libvirt] Libvirt node is not getting booted through network
by Varun Bhatnagar 15 Jan '13

15 Jan '13

Hi, I am trying to boot my node through network with the following configuration: But I am getting an error message saying: Even my network setting is not getting reflected in the Virtualbox. Can anyone suggest me what is wrong with this configuration. I think to boot a node from network we need to define a boot server but where shall I add that, I am confused.... Any help will be appreciated. Thanks in advance. Regards, Varun =====-----=====-----===== Notice: The information contained in this e-mail message and/or attachments to it may contain confidential or privileged information. If you are not the intended recipient, any dissemination, use, review, distribution, printing or copying of the information contained in this e-mail message and/or attachments to it are strictly prohibited. If you have received this communication in error, please notify us by reply e-mail or telephone and immediately and permanently delete the message and any attachments. Thank you

1 0

[libvirt] [PATCH] .gitignore: Sort alphabetically
by Michal Privoznik 15 Jan '13

15 Jan '13

--- Pushed under trivial rule .gitignore | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.gitignore b/.gitignore index f71c30c..be83e28 100644 --- a/.gitignore +++ b/.gitignore @@ -63,8 +63,8 @@ /docs/devhelp/libvirt.devhelp /docs/hvsupport.html.in /docs/libvirt-api.xml -/docs/libvirt-qemu-*.xml /docs/libvirt-lxc-*.xml +/docs/libvirt-qemu-*.xml /docs/libvirt-refs.xml /docs/search.php /docs/todo.html.in @@ -161,8 +161,8 @@ /tests/reconnect /tests/secaatest /tests/seclabeltest -/tests/securityselinuxtest /tests/securityselinuxlabeltest +/tests/securityselinuxtest /tests/sexpr2xmltest /tests/shunloadtest /tests/sockettest -- 1.8.0.2

1 0

[libvirt] [PATCH 0/4 v3] New API virNodeDeviceLookupSCSIHostByWWN
by Osier Yang 15 Jan '13

15 Jan '13

For easy reviewing, this is splitted from [1]. v2 - v3: * Validate the specified wwnn,wwpn pair before applying it to the new API in virsh-nodedev.c v1 - v2: * Per Daniel's suggestion, change the API name from virNodeDeviceLookupByWWN into virNodeDeviceLookupSCSIHostByWWN. Osier Yang (4): Introduce API virNodeDeviceLookupSCSIHostByWWN remote: Wire up the remote protocol nodedev: Implement virNodeDeviceLookupSCSIHostByWWN virsh: Use virNodeDeviceLookupSCSIHostByWWN include/libvirt/libvirt.h.in | 5 ++ src/driver.h | 6 ++ src/libvirt.c | 46 ++++++++++++++++ src/libvirt_public.syms | 1 + src/node_device/node_device_driver.c | 13 +++-- src/node_device/node_device_driver.h | 4 ++ src/node_device/node_device_hal.c | 1 + src/node_device/node_device_udev.c | 1 + src/remote/remote_driver.c | 1 + src/remote/remote_protocol.x | 13 ++++- src/remote_protocol-structs | 9 +++ src/rpc/gendispatch.pl | 5 ++- tools/virsh-nodedev.c | 97 ++++++++++++++++++++++++++-------- tools/virsh.pod | 15 +++-- 14 files changed, 181 insertions(+), 36 deletions(-) [1] https://www.redhat.com/archives/libvir-list/2013-January/msg00328.html Regards, Osier

1 5

[libvirt] [PATCH 0/4] Resolve UNINIT errors found by Coverity
by John Ferlan 15 Jan '13

15 Jan '13

This set of patches resolves "Error: UNINIT (CWE-457)" errors found by Coverity John Ferlan (4): parallels: Resolve issues with uninitialized 'ret' value openvz: Need to initialize 'ret' for kb_per_pages error path lxc: Initialize dst due to potential cleanup usage before setting interface: Need to initialize 'add_to_list' src/interface/interface_backend_udev.c | 2 +- src/lxc/lxc_driver.c | 2 +- src/openvz/openvz_conf.c | 2 +- src/parallels/parallels_storage.c | 19 ++++++++++++++----- 4 files changed, 17 insertions(+), 8 deletions(-) -- 1.7.11.7

4 13

[libvirt] [PATCH 0/9 v2] Miscs of nodedev
by Osier Yang 15 Jan '13

15 Jan '13

This is the second part of [1]. Some of them are preparations for persistent vHBA support in storage pool, some are fixes, some are improvements. v1 - v2: * One more patch: 9/9 Osier Yang (9): nodedev: Remove the unused enum nodedev: Introduce two new flags for listAll API util: Add one helper virReadFCHost to read the value of fc_host entry nodedev: Use access instead of stat nodedev: Refactor the helpers nodedev: Dump max vports and vports in use for HBA's XML nodedev: Fix the improper logic when enumerating SRIOV VF nodedev: Abstract nodeDeviceVportCreateDelete as util function cleanup libvirt_private.syms docs/formatnode.html.in | 10 +- docs/schemas/nodedev.rng | 6 + include/libvirt/libvirt.h.in | 20 +- src/conf/node_device_conf.c | 41 +- src/conf/node_device_conf.h | 16 +- src/libvirt.c | 2 + src/libvirt_private.syms | 1160 +++++++++++++++-------------- src/node_device/node_device_driver.c | 110 +--- src/node_device/node_device_driver.h | 19 +- src/node_device/node_device_hal.c | 15 +- src/node_device/node_device_linux_sysfs.c | 245 ++---- src/node_device/node_device_udev.c | 14 +- src/util/virpci.c | 36 +- src/util/virutil.c | 224 ++++++ src/util/virutil.h | 19 + tools/virsh-nodedev.c | 6 + tools/virsh.pod | 7 +- 17 files changed, 1034 insertions(+), 916 deletions(-) Regards, Osier

3 28

Re: [libvirt] [RFC qom-cpu v2 1/2] target-i386: Convert CPU definitions into X86CPU subclasses
by Eduardo Habkost 15 Jan '13

15 Jan '13

On Tue, Jan 15, 2013 at 09:41:04AM +0100, Igor Mammedov wrote: > On Mon, 10 Dec 2012 23:59:31 +0100 > Andreas Färber <afaerber(a)suse.de> wrote: > > > TODO: sort classes for -cpu ?, generalize X86CPUListState, more testing > > > > Signed-off-by: Andreas Färber <afaerber(a)suse.de> > > Cc: Eduardo Habkost <ehabkost(a)redhat.com> > > Cc: Igor Mammedov <imammedo(a)redhat.com> [...] > The patch is just renaming of the current builtin_x86_defs > into a bunch of functions and polluting X86CPUClass > with fields from the former x86_def_t. > object_new() still creates a dummy cpu instance whose defaults > are still manually copied from X86CPUClass instead of x86_def_t. > That's a good thing, isn't it? It means the patch is easier to review. :-) No patch alone will do everything we want, because we want to do a lot. We need to do it one step at a time. (BTW, why are you looking at this RFC instead of the more recent one, that I have sent on Jan 4? [that's very similar to this one]) > What's the point in having dummy sub-classes? > How it can help in your CPU re-factoring? It will help us to unify the CPU creation/realization code that's duplicated over all the architectures. It will give libvirt an easy mechanism to list the available CPU models that won't require parsing help output (using "qom-list-types" QMP command). > > > On the other hand converting features to static properties first and > then converting X86CPU to sub-classes, yields already initialized to > defaults sub-class completely removing notion of x86_def_t and > not polluting X86CPUClass with redundant fields. I wouldn't disagree with this approach in principle, but I believe our main problem today is lack of reviewer bandwidth. I learned the hard way that trying to clean up everything before implementing something will make sure the code takes forever to be reviewed. > For example see following patches on > https://github.com/imammedo/qemu/commits/x86-cpu-classes.Jan142013 > > e9fd18f qdev: extend DEFINE_GENERIC_PROP() to support default values > c65eca9 qdev: make qdev_prop_find_bit return non const so prop default value could be modified > 0311952 allow to expolit default value static props for model, family, stepping & vendor props > 8b3080e target-i386: add helpers to change default values of static properties before object is created > ed506d3 target-i386: prepare for subclasses to have its own instance of static properties definitions > a48e252 target-i386: declare subclass for qemu64 cpu model > 9c556c2 target-i386: move cpu_x86_init() & cpu_x86_register() into it and switch to subclasses. PS: implemended only for qemu64 > f5dbfe6 CPU_CLASS_NAME(qemu64) hack > 00e15b8 target-i386: properties list are per subclass: do not set them in superclass to avoid defaults set by subclass be over-written > > -- > Regards, > Igor -- Eduardo

1 0