It was fairly trivial to return snapshot listing based on a
point in the hierarchy, rather than starting at all roots.

* src/esx/esx_driver.c (esxDomainSnapshotNumChildren)
(esxDomainSnapshotListChildrenNames): New functions.

Not too hard to wire up.  The trickiest part is realizing that
listing children of a snapshot cannot use SNAPSHOT_LIST_ROOTS,
and that we overloaded that bit to also mean SNAPSHOT_LIST_DESCENDANTS;
we use that bit to decide which iteration to use, but don't want
the existing counting/listing functions to see that bit.

* src/conf/domain_conf.h (virDomainSnapshotObjListNumFrom)
(virDomainSnapshotObjListGetNamesFrom): New prototypes.
* src/conf/domain_conf.c (virDomainSnapshotObjListNumFrom)
(virDomainSnapshotObjListGetNamesFrom): New functions.
* src/libvirt_private.syms (domain_conf.h): Export them.
* src/qemu/qemu_driver.c (qemuDomainSnapshotNumChildren)
(qemuDomainSnapshotListChildrenNames): New functions.

Very mechanical.  I'm so glad we've automated the generation of things,
compared to what it was in 0.8.x days, where this would be much longer.

* src/remote/remote_protocol.x
(REMOTE_PROC_DOMAIN_SNAPSHOT_NUM_CHILDREN)
(REMOTE_PROC_DOMAIN_SNAPSHOT_LIST_CHILDREN_NAMES): New rpcs.
(remote_domain_snapshot_num_children_args)
(remote_domain_snapshot_num_children_ret)
(remote_domain_snapshot_list_children_names_args)
(remote_domain_snapshot_list_children_names_ret): New structs.
* src/remote/remote_driver.c (remote_driver): Use it.
* src/remote_protocol-structs: Update.

Given a list of snapshots and their parents, finding all descendants
requires a hairy traversal.  This code is O(n^3); it could maybe be
made to scale O(n^2) with the use of a hash table, but that costs more
memory.  Hopefully there aren't too many people with a hierarchy
so large as to approach REMOTE_DOMAIN_SNAPSHOT_LIST_NAMES_MAX (1024).

* tools/virsh.c (cmdSnapshotList): Add final fallback.

Iterating over one level of children requires parsing all snapshots
and their parents; a bit of code shuffling makes it pretty easy
to do this as well.

* tools/virsh.c (cmdSnapshotList): Add another fallback.

Emulating --from requires grabbing the entire list of snapshots
and their parents, and recursively iterating over the list from
the point of interest - but we already do that for --tree.  This
turns on emulation for that situation.

* tools/virsh.c (__vshControl): Rename member.
(vshReconnect, cmdConnect, vshGetSnapshotParent): Update clients.
(cmdSnapshotList): Add fallback.

Sometimes, we only care about one branch of the snapshot hierarchy.
Make it easier to list a single branch, by using the new APIs.

Technically, I could emulate these new virsh options on old servers
by doing a complete dump, then scraping xml to filter out just the
snapshots that I care about, but I didn't want to do that in this patch.

* tools/virsh.c (cmdSnapshotList): Add --from, --descendants.
* tools/virsh.pod (snapshot-list): Document them.

The previous API addition allowed traversal up the hierarchy;
this one makes it easier to traverse down the hierarchy.

In the python bindings, virDomainSnapshotNumChildren can be
generated, but virDomainSnapshotListChildrenNames had to copy
from the hand-written example of virDomainSnapshotListNames.

* include/libvirt/libvirt.h.in (virDomainSnapshotNumChildren)
(virDomainSnapshotListChildrenNames): New prototypes.
(VIR_DOMAIN_SNAPSHOT_LIST_DESCENDANTS): New flag alias.
* src/libvirt.c (virDomainSnapshotNumChildren)
(virDomainSnapshotListChildrenNames): New functions.
* src/libvirt_public.syms: Export them.
* src/driver.h (virDrvDomainSnapshotNumChildren)
(virDrvDomainSnapshotListChildrenNames): New callbacks.
* python/generator.py (skip_impl, nameFixup): Update lists.
* python/libvirt-override-api.xml: Likewise.
* python/libvirt-override.c
(libvirt_virDomainSnapshotListChildrenNames): New wrapper function.

</space> doesn't exist.  Introduced in commit 4bb4109f.

* docs/formatdomain.html.in: Use correct end tag.

On xen 4.1 I observed configurations that look like:

(image
    (hvm
        (kernel '')
        (loader '/foo/bar')
))

The kernel element is there but unset. This leads to an empty <kernel/>
element in the XML and even worse makes us skip the boot order parsing
and therefore not emit a <boot device='$dev>'/> element which breaks CD
booting.

otherwise a missing UUID in a domain config just shows:

error: An error occurred, but the cause is unknown

Now we have:

error: configuration file syntax error: config value uuid was missing

I noticed a couple typos in recent commits, and fixed the remaining
instances of them.

* docs/internals/command.html.in: Fix spelling errors.
* include/libvirt/libvirt.h.in (virConnectDomainEventCallback):
Likewise.
* python/libvirt-override.py (virEventAddHandle): Likewise.
* src/lxc/lxc_container.c (lxcContainerChild): Likewise.
* src/util/hash.c (virHashCreateFull): Likewise.
* src/storage/storage_backend_logical.c
(virStorageBackendLogicalMakeVol): Likewise.
* src/esx/esx_driver.c (esxFormatVMXFileName): Likewise.
* src/vbox/vbox_tmpl.c (vboxIIDIsEqual_v3_x): Likewise.

I was a bit surprised that 'virsh snapshot-edit dom name' silently
allowed me to clone things, while still telling me the old name,
especially since other commands like 'virsh edit dom' reject rename
attempts (*).  This fixes things to be more explicit (**).

(*) Technically, 'virsh edit dom' relies on virDomainDefineXML
behavior, which rejects attempts to mix a new name with existing
uuid or new uuid with existing name, but you can create a new
domain by changing both uuid and name.  On the other hand, while
snapshot-edit --clone is a true clone, creating a new domain
would also have to decide whether to clone snapshot metadata,
managed save, and any other secondary data related to the domain.
Domain renames are not trivial either.

(**) Renaming or creating a clone is still a risky proposition -
for offline snapshots and system checkpoints, if the new name
does not match an actual name recorded in the qcow2 internal
snapshots, then you cannot revert to the new checkpoint.  But it
is assumed that anyone using the new virsh flags knows what they
are doing, and can deal with the fallout caused by a rename/clone;
that is, we can't completely prevent a user from shooting
themselves in the foot, so much as we are making the default
action less risky.

* tools/virsh.c (cmdSnapshotEdit): Add --rename, --clone.
* tools/virsh.pod (snapshot-edit): Document them.

Otherwise, the results are not repeatable.

* tools/virsh.c (cmdSnapshotList): Print tree in predictable order.

* src/storage/storage_backend_logical.c:

If a logical vol is created as striped. (e.g. --stripes 3),
the "device" field of lvs output will have multiple fileds which are
seperated by comma. Thus the RE we write in the codes will not
work well anymore. E.g. (lvs output for a stripped vol, uses "#" as
seperator here):

test_stripes##fSLSZH-zAS2-yAIb-n4mV-Al9u-HA3V-oo9K1B#\
/dev/sdc1(10240),/dev/sdd1(0)#42949672960#4194304

The RE we use:

    const char *regexes[] = {
        "^\\s*(\\S+),(\\S*),(\\S+),(\\S+)\\((\\S+)\\),(\\S+),([0-9]+),?\\s*$"
    };

Also the RE doesn't match the "devices" field of striped vol properly,
it contains multiple "device path" and "offset".

This patch mainly does:
    1) Change the seperator into "#"
    2) Change the RE for "devices" field from "(\\S+)\\((\\S+)\\)"
       into "(\\S+)".
    3) Add two new options for lvs command, (segtype, stripes)
    4) Extend the RE to match the value for the two new fields.
    5) Parse the "devices" field seperately in virStorageBackendLogicalMakeVol,
       multiple "extents" info are generated if the vol is striped. The
       number of "extents" is equal to the stripes number of the striped vol.

A incidental fix: (virStorageBackendLogicalMakeVol)
    Free "vol" if it's new created and there is error.

Demo on striped vol with the patch applied:

% virsh vol-dumpxml /dev/test_vg/vol_striped2
<volume>
  <name>vol_striped2</name>
  <key>QuWqmn-kIkZ-IATt-67rc-OWEP-1PHX-Cl2ICs</key>
  <source>
    <device path='/dev/sda5'>
      <extent start='79691776' end='88080384'/>
    </device>
    <device path='/dev/sda6'>
      <extent start='62914560' end='71303168'/>
    </device>
  </source>
  <capacity>8388608</capacity>
  <allocation>8388608</allocation>
  <target>
    <path>/dev/test_vg/vol_striped2</path>
    <permissions>
      <mode>0660</mode>
      <owner>0</owner>
      <group>6</group>
      <label>system_u:object_r:fixed_disk_device_t:s0</label>
    </permissions>
  </target>
</volume>

RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=727474

If parsing qemu command line fails (e.g. because of non-existing
process number supplied), we jump to cleanup label where we free
pidfile. Therefore it needs to be initialized. Otherwise we free
random pointer.

Coverity complained that 4 out of 5 callers to virJSONValueObjectGetBoolean
checked for errors.  But we documented that we don't care in this case.

* src/qemu/qemu_monitor_json.c (qemuMonitorJSONGetBlockInfo): Use
ignore_value.

Detected by Coverity.  We want to increment the size_t counter,
not the pointer to the counter.  Bug present since 5f5c6fde (0.9.5).

* src/lxc/lxc_controller.c (lxcSetupLoopDevices): Use correct
precedence.

virsh undefine command can now undefine an active guest, but the help information is still the old.
This patch modifies it and make it coincident to the manpage of virsh.
Signed-off-by: Ntangchen <tangchen@cn.fujitsu.com>

If we send back an unknown program error for async messages,
we will confuse the client because they only expect replies
for method calls. Just log & drop any invalid async messages

* src/rpc/virnetserver.c: Don't send error for async messages

Commit 597fe3ce accidentally
introduced a deadlock when reporting an unknown RPC program.
The virNetServerDispatchNewMessage method is called with
the client locked, and must therefore not attempt to send
any RPC messages back to the client. Only once the incoming
message is passed off to the virNetServerHandleJob worker
is it safe to start sending messages back

* src/rpc/virnetserver.c: Delay checking for unknown RPC
  program until in worker thread

Redefining disk-only snapshot xml should work even if the user
did not explicitly pass VIR_DOMAIN_SNAPSHOT_CREATE_DISK_ONLY;
the flag is only required for conditions where the <state>
subelement is not already present in parsing (that is, defining
a new snapshot).

Also, fix the error code of some user-visible errors (the remaining
VIR_ERR_INTERNAL_ERROR should not be user-visible, since parsing
of <active> is only done from internal code).

* src/conf/domain_conf.c (virDomainSnapshotDefParseString): Allow
disks during redefinition of disk snapshot.

It was impossible for 'virsh snapshot-current dom name' to set name
as the current snapshot, if name is a disk-only snapshot.

Using strstr rather than full-blown xml parsing is safe, since the
xml is assumed to be well-formed coming from libvirtd rather than
arbitrary text coming from the user.

* tools/virsh.c (cmdSnapshotCurrent, cmdSnapshotEdit): Pass
disk_only flag when redefining a disk snapshot.

* tools/virsh.c (cmdSnapshotList): Spell exclusive correctly.

Don't xdr_free event data as they are freed by our caller
virNetClientProgramDispatch.

I am getting this failure with 'make distcheck':

  GEN    ../../src/remote_protocol-structs
/bin/sh: ../../src/remote_protocol-structs-t: Permission denied
make[4]: *** [../../src/remote_protocol-structs] Error 1

since it attempts a sub-run of a VPATH 'make check' where $(srcdir)
is intentionally read-only.  I'm not sure which commit introduced
the problem, although I suspect it was around 62dee6fa when I
refactored protocol struct checking to be more powerful.

$(@F) is required by POSIX, and although it is not yet portable
to all make implementations, we already require GNU make.

* src/Makefile.am (PDWTAGS): Generate temp file into current
directory, since $(srcdir) is read-only during distcheck.

Previously libvirt's disk device XML only had a single attribute,
error_policy, to control both read and write error policy, but qemu
has separate options for controlling read and write. In one case
(enospc) a policy is allowed for write errors but not read errors.

This patch adds a separate attribute that sets only the read error
policy. If just error_policy is set, it will apply to both read and
write error policy (previous behavior), but if the new rerror_policy
attribute is set, it will override error_policy for read errors only.
Possible values for rerror_policy are "stop", "report", and "ignore"
("report" is the qemu-controlled default for rerror_policy when
error_policy isn't specified).

For consistency, the value "report" has been added to the possible
values for error_policy as well.

commit 12062abb set rerror=ignore when error_policy="enospace" was
selected (since the rerror option in qemu doesn't accept "enospc", as
the werror option does).

After that patch was already pushed, Paolo Bonzini noticed it and
commented that leaving rerror at the default ("report") would be a
better choice. This patch corrects the problem - if error_policy =
"enospace" is given, rerror is left off the qemu commandline,
effectively setting it to "report". For other values, rerror is still
set to match werror.

Additionally, the parsing of error_policy was changed to no longer
erroneously allow "default" as a choice - as with most other
attributes, if you want the default setting, just don't specify an
error_policy.

Finally, two ommissions in the first patch were corrected - a
long-dormant qemuxml2argv test for enospace was enabled, and fixed to
pass, and the argv2xml parser in qemu_command.c was updated to
recognize the different spelling on the qemu commandline.

Now that RHEL 6.2 Beta is out, it would be nice to test multifunction
devices on that platform.  This changes things so that the multifunction
cap bit can be set in two different ways: by version comparison (needed
for qemu 0.13 which lacked a -device query), and by -device query
(provided by qemu.git and backported to the RHEL beta build of
qemu-kvm which still claims to be a modified 0.12, and therefore needed
for RHEL).

* src/qemu/qemu_capabilities.c (qemuCapsParseDeviceStr): Allow
second method of setting multifunction cap bit.
* tests/qemuhelptest.c (mymain): Test it.
* tests/qemuhelpdata/qemu-kvm-0.12.1.2-rhel62-beta: New file.
* tests/qemuhelpdata/qemu-kvm-0.12.1.2-rhel62-beta-device: Likewise.

If using one of the new non-NAT/routed virtual network
configurations, the LXC driver would not know how to
setup the VETH devices. Adding in calls to setup the
"actual" network configuration at VM startup and cleanup
when shutting down fixes this.

* src/lxc/lxc_driver.c: Setup/cleanup actual net devs

https://bugzilla.redhat.com/show_bug.cgi?id=740899 documents that
if qemu uses aio=native for its disks, then it consumes 128 aio
requests per disk.  On a host with multiple guests, this can quickly
run out of kernel aio requests with the default aio-max-nr of
65536.  Kernel developers have confirmed that there is no up-front
cost to raising this limit (a larger limit merely implies that more
aio requests can be issued in parallel, which in turn will result
in more kernel memory allocation, only if the system really does use
that many requests).  Since the system default limit prevents 256
disks, which is well within libvirt's current scalability, this
patch installs a file to raise the limit and document it in case a
system administrator has further cause to tune the limit.  The
install only works on platforms new enough to source /etc/sysctl.d/*
alongside /etc/sysctl.conf (F14 and RHEL 6).

* daemon/libvirtd.sysctl: New file.
* daemon/Makefile.am (EXTRA_DIST): Ship it.
(install-init, uninstall-init): Install it.
* libvirt.spec.in (%files): Include it in rpm.

Consistent use of tabs, fewer long lines, and a typo fix.

* include/libvirt/virterror.h: Fix typos, layout.

Implements the documentation for snapshot revert vs. force.

Part of the patch tightens existing behavior (previously, reverting
to an old snapshot without <domain> was blindly attempted, now it
requires force), while part of it relaxes behavior (previously, it
was not possible to revert an active domain to an ABI-incompatible
active snapshot, now force allows this transition).

* src/qemu/qemu_driver.c (qemuDomainRevertToSnapshot): Check for
risky situations, and allow force to get past them.

Once we know which set of disks belong to a snapshot, reverting or
deleting that snapshot should visit just those disks, rather than
also visiting disks that were hot-plugged in the meantime or
skipping disks that were hot-unplugged in the meantime.

* src/qemu/qemu_domain.c (qemuDomainSnapshotForEachQcow2): Use
snapshot domain details when available.  Avoid NULL deref.

Although reverting to a snapshot is a form of data loss, this is
normally expected.  However, there are two cases where additional
surprises (failure to run the reverted state, or a break in
connectivity to the domain) can come into play.  Requiring extra
acknowledgment in these cases will make it less likely that
someone can get into an unrecoverable state due to a default revert.

Also create a new error code, so users can distinguish when forcing
would make a difference, rather than having to blindly request force.

* include/libvirt/libvirt.h.in (VIR_DOMAIN_SNAPSHOT_REVERT_FORCE):
New flag.
* src/libvirt.c (virDomainRevertToSnapshot): Document it.
* include/libvirt/virterror.h (VIR_ERR_SNAPSHOT_REVERT_RISKY): New
error value.
* src/util/virterror.c (virErrorMsg): Implement it.
* tools/virsh.c (cmdDomainSnapshotRevert): Add --force to virsh.
* tools/virsh.pod (snapshot-revert): Document it.

Commit 9f5e53e2 introduced the ability to filter snapshots to
just roots, but it was never implemented for VBox until now.

The VBox implementation prohibits deletion of a snapshot with
multiple children.  Hence, there can only be at most one root,
which is found by searching for the snapshot with a NULL uuid.

Prior to 4.0, snapshotGet looked up by UUID, and snapshotFind
looked up by name; after that point, snapshotGet disappeared
and snapshotFind handles uuid or name.

* src/vbox/vbox_tmpl.c (vboxDomainSnapshotNum)
(vboxDomainSnapshotListNames): Implement limiting list to root.

Qemu driver tries to update balloon data in virDomainGetInfo and if it
can't do so because there is another monitor job running, it just
reports what's known in domain def. However, if there was no job running
but getting the data from qemu fails, we would fail the whole API. This
doesn't make sense. Let's make the failure nonfatal.

No need to request the parent of a snapshot if we aren't going to use it.

* src/esx/esx_vi.c (esxVI_GetSnapshotTreeByName): Make parent
optional.
* src/esx/esx_driver.c (esxDomainSnapshotCreateXML)
(esxDomainSnapshotLookupByName, esxDomainRevertToSnapshot)
(esxDomainSnapshotDelete): Simplify accordingly.