When a qemu domain is to be rebooted, from outside, at libvirt
level it looks like regular shutdown. To really restart the
domain, libvirt needs to issue reset command on the monitor once
SHUTDOWN event appeared. So, in order to differentiate bare
shutdown and reboot libvirt uses a variable within domain private
data. It's called fakeReboot. When the reboot API is called, the
variable is set, but when the shutdown API is called it must be
cleared out. But it was not for every possible case. So if user
called virDomainReboot(), and there was no ACPI daemon running
inside the guest (so guest didn't initiated shutdown sequence)
and then virDomainShutdown(mode=agent) was called bad thing
happened. We remembered the fakeReboot and instead of shutting
the domain down, we just rebooted it.
Signed-off-by: NZhang Bo <oscar.zhangbo@huawei.com>
Signed-off-by: NWang Yufei <james.wangyufei@huawei.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

This is a simple wrapper around virNetDevBandwidthManipulateFilter() that
will update the desired filter on an interface (usually a network bridge)
with a new MAC address. Although, the MAC address in question usually
refers to some other interface - the one that the filter is constructed
for. Yeah, hard to parse. Thing is, our NATed network has a bridge where
some part of QoS takes place. And vNICs from guests are plugged into
the bridge. However, if a guest decides to change the MAC of its vNIC,
the corresponding qemu process emits an event which we can use to
update the QoS configuration based on the new MAC address.. However,
our QoS hierarchy is currently not notified, therefore it falls apart.
This function (when called in response to the aforementioned event)
will update our QoS hierarchy and duct tape it together again.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Not only this simplifies the code a bit, it prepares the
environment for upcoming patches. The new
virNetDevBandwidthManipulateFilter() function is capable of both
removing a filter and adding a new one. At the same time! Yeah,
this is not currently used anywhere but look at the next commit
where you'll see it.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Currently, when constructing traffic shaping rules, the ingress
filter is created without any priority specified on the command
line. This makes kernel to make up one. While this works, it
simplifies things a bit if we provide the filter priority. In
this case, since it's the root filter lets give it the highest
priority of number 1.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

After a3609121 the formatting of virDomainActualNetDefPtr was
changed a bit. However, during the function rewrite, iface's class_id
is not formatted as frequently as it could be. In fact, after rewrite
it's formatted only for iface of type VIR_DOMAIN_NET_TYPE_DIRECT where
it makes no sense and is unused. While where needed (_TYPE_NETWORK) is
not formatted at all. This makes the daemon forget it upon daemon
restart resulting in bad behaviour.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1211436

This reverts commit b7829f95.

The previous fix was not correct. Like everywhere else, a driver is a
global variable allocated in stateInitialize function (or something
similar for stateless drivers). Later, when a driver API is called,
it's possible that the global variable is accessed and dereferenced.
Now, some drivers require root privileges because they undertake some
actions reserved only for the system admin (e.g. manipulating host
firewall). And here's the trouble, the NWFilter state initializer
exited too early when finding out it's running unprivileged, leaving
the global NWFilter driver variable uninitialized. Any subsequent
API call that tried to lock the driver resulted in dereferencing the
driver and thus crash.

On the other hand, in order to not resurrect the bug the original
commit was fixing, Let's forbid the nwfilter define in session mode.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Conflicts:
	src/nwfilter/nwfilter_driver.c: Context. Code changed a bit
        since 2013.

There is a possibility that we jump onto error label with @lockpath
still initialized to NULL. Here, the @lockpath should be unlink()-ed,
but passing there a NULL is not a good idea. Don't do that. In fact,
we should call unlink() only if we created the lock file successfully.
Reported-by: NJohn Ferlan <jferlan@redhat.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

The 802.11 interfaces can not be moved by themselves, their Phy has to move too.

If there are other interfaces, they have to move too -- hopefully it's not too
confusing. This is a less-invasive alternative to defining a new hostdev type
for PHYs.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

A destroy operation can take considerable time on large memory
domains due to scrubbing the domain's memory.  Unlock the
virDomainObj while libxl_domain_destroy is executing.

Implement libxlDomainDestroyInternal wrapper to handle unlocking,
calling destroy, and locking.  Change all callers of
libxl_domain_destroy to use the wrapper.
Signed-off-by: NJim Fehlig <jfehlig@suse.com>

A job should be acquired at the beginning of a domain destroy operation,
not at the end when cleaning up the domain.  Fix two occurrences of this
late job acquisition in the libxl driver.  Doing so renders
libxlDomainCleanupJob unused, so it is removed.
Signed-off-by: NJim Fehlig <jfehlig@suse.com>

Let callers of libxlDomainStart decide when it is appropriate to
acquire a job on the associated virDomainObj.
Signed-off-by: NJim Fehlig <jfehlig@suse.com>

Add support for HVM direct kernel boot in libxl.  Also add a
test to verify domXML <-> native conversions.
Signed-off-by: NChunyan Liu <cyliu@suse.com>
Signed-off-by: NJim Fehlig <jfehlig@suse.com>

In xl config, hvmloader is implied for hvm guests.  It is not
specified with the "kernel" option like xm config.  The "kernel"
option, along with "ramdisk" and "extra", is used for HVM direct
kernel boot.  Instead of using "kernel" option to populate
virDomainDef object's os.loader->path, use hvmloader discovered
when gathering capabilities.

This change required fixing initialization of capabilities in
the test utils and removing 'kernel = "/usr/lib/xen/boot/hvmloader"'
from the test config files.

xl and xm differ a bit in how <os> configuration is represented.
E.g. xl config supports <os><nvram .../></os> via its "bios"
setting.

Move the xenParseOS and xenFormatOS functions from xen_common.c
and copy to xen_xl.c and xen_xm.c so they can be customized for
xm vs xl config.  An unfortunate fallout is reordering of entries
in the test config files.

device_model is parsed in xenParseOS(), then later in
xenParseConfigCommon().  <emulator> is not part of <os>,
so makes sense to remove the parsing in xenParseOS().

Export xenConfigCopyString for use outside of xen_common.c
Signed-off-by: NJim Fehlig <jfehlig@suse.com>

On rhel-6 is broken gcc that reports this warning:

util/virbuffer.c:500: error: logical '&&' with non-zero constant will
    always evaluate as true [-Wlogical-op]

Move the pragma directive before function virBufferEscapeString because
since commit aeb5262e this function uses 'strchr' too.
Signed-off-by: NPavel Hrdina <phrdina@redhat.com>

Invert the logical of retval and clean up code paths, especially goto's
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

ts.tv_nsec was off by a factor of 1000, making timeouts less than a
second in the future often expiring immediately.
Signed-off-by: NMichael Chapman <mike@very.puzzling.org>

Among all the monitor APIs some where checking if mon is NULL and some
were not. Since it's possible to have mon equal to NULL in case a second
call is attempted once entered the monitor. This requires that every
single API checks for the monitor.

This patch adds a macro that helps checking the state of the monitor and
either refactors existing checking code to use the macro or adds it in
case it was missing.

Apparently, files in /usr/lib/sysctl.d are usually prefixed with numbers
for easier ordering. Let's be consistent with this. I chose 60 for
libvirtd so that it goes after 50-default.conf.

https://bugzilla.redhat.com/show_bug.cgi?id=1084876Signed-off-by: NJiri Denemark <jdenemar@redhat.com>

Rather than erroring out make the best attempt to retrieve other data if
disks are inaccessible or missing. The failure will still be logged
though.

Since the bulk stats API is called on multiple domains an error like
this makes the API unusable. This regression was introduced by commit
596a1371

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1209394

The <inbound/> element to <bandwidth/> has several attributes from
which two are mandatory. Well, from two at least one has to be
present: @average or @floor or both. Instead of inventing crazy RNG
schema, let's make all the attributes optional there and rely on our
parsing code to correctly handle the situation.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

The comment is describing arguments passed to the function.
However, there's no @ifmac argument. In 955af4d4 it was replaced
with @ifmac_ptr.  Unfortunately, the comment wasn't updated.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

gcc 4.1.2 (hello RHEL 5) on 32-bit platforms complains:

vircgrouptest.c: In function 'testCgroupGetPercpuStats':
vircgrouptest.c:627: warning: integer constant is too large for 'long' type
vircgrouptest.c:628: warning: this decimal constant is unsigned only in ISO C90
vircgrouptest.c:634: warning: integer constant is too large for 'long' type
vircgrouptest.c:635: warning: this decimal constant is unsigned only in ISO C90
vircgrouptest.c:636: warning: this decimal constant is unsigned only in ISO C90
vircgrouptest.c:644: warning: integer constant is too large for 'long' type

* tests/vircgrouptest.c (testCgroupGetPercpuStats): Use ULL suffix.
Signed-off-by: NEric Blake <eblake@redhat.com>

These cannot be represented in XML.

We have been stripping them, but only if the string had
characters that needed escaping: <>"'&

Extend the strcspn check to include control codes, and strip
them even if we don't do any escaping.

https://bugzilla.redhat.com/show_bug.cgi?id=1184131
https://bugzilla.redhat.com/show_bug.cgi?id=1066564

To prevent generating invalid XML.

https://bugzilla.redhat.com/show_bug.cgi?id=1066564

Including them in the XML makes them unparsable.

https://bugzilla.redhat.com/show_bug.cgi?id=1184131

Add virStringHasControlChars that checks if the string has
any control characters other than \t\r\n,
and virStringStripControlChars that removes them in-place.

For reuse with other Strip* functions.

Only for devices that have an alias.

The counterpart to VIR_DOMAIN_EVENT_ID_DEVICE_REMOVED.

https://bugzilla.redhat.com/show_bug.cgi?id=1206114

Throughout the code, we have several places need to construct a path
somewhere in /sys/class/net/... They are not consistent and nearly
each code piece invents its own way how to do it. So unify this by:

1) use virNetDevSysfsFile() wherever possible

2) At least use common macro SYSFS_NET_DIR declared in virnetdev.h at
   the rest of places which can't go with 1)
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

In one of my previous commits (49ed6cff) I've introduced a test
among with some files stored under virnetdevtestdata folder.
While this works perfectly within a git tree, the folder was not
getting into .tar.gz and therefore the dist-check would fail.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Commit id 'b77ce18a' added a new bitmap, but neglected to virBitmapFree it

Commit a0670aef caused a regression in 'virsh event' and
'virsh qemu-monitor-event' - if a user tries to filter the
command to a specific domain, an error message is printed:

$ virsh event dom --loop
error: internal error: virsh qemu-monitor-event: no domain VSH_OT_DATA option

and then the command continues as though no domain had been
supplied (giving events for ALL domains, instead of the
requested one).  This is because the code was incorrectly
assuming that all "domain" options would be supplied via a
mandatory VSH_OT_DATA, even though "domain" is optional for
these two commands, so we had changed them to VSH_OT_STRING
to quit failing for other reasons (ever since it was decided
that VSH_OT_DATA and VSH_OT_STRING should no longer be
synonyms).

In looking at the situation, though, the code for looking up
a domain was making a pointless check for whether the option
exists prior to finding the option's string value, as
vshCommandOptStringReq does just fine at reporting any errors
when looking up a string whether or not the option was present.

So this is a case of regression fixing by pure code deletion :)

* tools/virsh-domain.c (vshCommandOptDomainBy): Drop useless filter.
* tools/virsh-interface.c (vshCommandOptInterfaceBy): Likewise.
* tools/virsh-network.c (vshCommandOptNetworkBy): Likewise.
* tools/virsh-nwfilter.c (vshCommandOptNWFilterBy): Likewise.
* tools/virsh-secret.c (vshCommandOptSecret): Likewise.
* tools/virsh.h (vshCmdHasOption): Drop unused function.
* tools/virsh.c (vshCmdHasOption): Likewise.
Signed-off-by: NEric Blake <eblake@redhat.com>

All the called functions already report an error.

If a virAsprintf() within the function fails, we call VIR_FREE()
over @rundir variable and jump onto cleanup label, where it is
freed again.  It doesn't hurt, but not make much sense too.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Commit f6563bc3 introduced HMP impl of the function (so that a different
uglier function could be removed). Before the HMP code is called there's
a leftover check that the monitor is JSON which inhibits the code from
working.