Some tests use the same VM state multiple times in a row. But if we
failed loading the VM XML, subsequent tests crash on the NULL def
pointer
Acked-by: NMichal Privoznik <mprivozn@redhat.com>
Signed-off-by: NCole Robinson <crobinso@redhat.com>

- Convert to 'cleanup' label naming
- Use more than one 'tmp' string and do all freeing at the end
- Make the code easier to follow
Acked-by: NMichal Privoznik <mprivozn@redhat.com>
Signed-off-by: NCole Robinson <crobinso@redhat.com>

Handles parse virtType, os.type, bootloader bits, arch, machine,
emulator
Acked-by: NMichal Privoznik <mprivozn@redhat.com>
Signed-off-by: NCole Robinson <crobinso@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1607825

Introduced by commit d48813e8.
Signed-off-by: NHan Han <hhan@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1599973

Commit id fca9afa0 changed the @req->ifname to use
@req->binding->portdevname fillingin the @req->binding
in a similar way that @req->ifname would have been
filled in during virNWFilterDHCPSnoopReq processing.

However, in doing so it did not take into account some
code paths where the @req->binding should be checked
instead of @req->binding->portdevname. These checks
led to SEGVs in some cases during libvirtd reload
processing in virNWFilterSnoopRemAllReqIter (for
stop during nwfilterStateCleanup processing) and
virNWFilterSnoopReqLeaseDel (for start during
nwfilterStateInitialize processing).

In particular, when reading the nwfilter.leases file
a new @req is created, but the @req->binding is not
filled in. That's left to virNWFilterDHCPSnoopReq
processing which checks if the @req already exists
in the @virNWFilterSnoopState.snoopReqs hash table
after adding a virNWFilterSnoopState.ifnameToKey
entry for the @req->binding->portdevname by a
@ref->ikey value.

NB: virNWFilterSnoopIPLeaseInstallRule and
    virNWFilterDHCPSnoopThread do not need the
    req->binding check since they can only be called
    after the filter->binding is created/assigned.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>
ACKed-by: NMichal Privoznik <mprivozn@redhat.com>

The config object is refed but unrefed only on error which leaves
refcount unbalanced on successful return.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

The individual strings are freed, but the array is never freed.

 8 bytes in 1 blocks are definitely lost in loss record 28 of 1,098
    at 0x4C2CE3F: malloc (vg_replace_malloc.c:298)
    by 0x4C2F1BF: realloc (vg_replace_malloc.c:785)
    by 0x52C9C92: virReallocN (viralloc.c:245)
    by 0x52C9D88: virExpandN (viralloc.c:294)
    by 0x23414D99: virLXCProcessSetupInterfaces (lxc_process.c:552)
    by 0x23417457: virLXCProcessStart (lxc_process.c:1356)
    by 0x2341F71C: lxcDomainCreateWithFiles (lxc_driver.c:1088)
    by 0x2341F805: lxcDomainCreate (lxc_driver.c:1123)
    by 0x55917EB: virDomainCreate (libvirt-domain.c:6534)
    by 0x1367D1: remoteDispatchDomainCreate (remote_daemon_dispatch_stubs.h:4434)
    by 0x1366EA: remoteDispatchDomainCreateHelper (remote_daemon_dispatch_stubs.h:4410)
    by 0x546FDF1: virNetServerProgramDispatchCall (virnetserverprogram.c:437)
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

So we originally disabled LXC driver when libvirtd is running
under valgrind back in 05436ab7 (which dates to beginning of
2009) as it was causing valgrind to crash. It's not the case
anymore. Valgrind works with LXC happily.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

There are two places where we report supported sizes of huge pages:

  /capabilities/host/cpu/pages
  /capabilities/host/topology/cells/cell/pages

The former aggregates sizes over all NUMA nodes while the latter
reports supported sizes only for given node. While we are
reporting per NUMA node sizes we are not reporting the aggregated
sizes. I've noticed this when wondering why doesn't allocpages
completer work.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

While not as critical as in qemu driver, there are still some
runtime information we report in capabilities XML that might
change throughout time. For instance, onlined CPUs (which affects
reported L3 cache sizes).
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1600468

If we are unable to read leases file (no matter what the reason
is), we return 0 - just like if there were no leases. However,
because we use virFileReadAll() an error is printed into the log.
Note that not all networks have leases file - only those for
which we start dnsmasq.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

Consider creating a listener socket from a hostname that resolves to
multiple addresses. It might be the case that the hostname resolves to
both an IPv4 and IPv6 address because it is reachable over both
protocols, but the IPv6 connectivity is provided off-host. In such a
case no local NIC will have IPv6 and so bind() would fail with the
EADDRNOTAVAIL errno. Thus it should be treated as non-fatal as long as
at least one socket was succesfully bound.
Reviewed-by: NJiri Denemark <jdenemar@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

When reporting socket/bind failures we want to ensure any fatal error
reported is as accurate as possible. We'll prefer reporting a bind()
errno over a socket() errno, because if socket() works but bind() fails
that is a more significant event.
Reviewed-by: NJiri Denemark <jdenemar@redhat.com>
Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1559867

When attaching a device to the domain we need to be sure
to use the correct domain definition (vm->def or vm->newDef)
when calling virDomainDeviceDefParse because the post parse
processing algorithms that may assign an address for the
device will use whatever domain definition was passed in.

Additionally, some devices (SCSI hostdev and SCSI disk) use
algorithms that rely on knowing what already exists of the
other type when generating the new device's address. Using
the wrong VM definition could result in duplicated addresses.

In the case of the bz, two hostdev's with no domain address
provided were added to the running domain's config only.
However, the parsing algorithm used the live domain in
order to figure out the host device address resulting in
the same address being used and a subsequent start failing
due to duplicate address.

Fix this by separating the checks/code into CONFIG and LIVE
processing using the correct definition for each block and
performing cleanup for both options as necessary.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>
ACKed-by: NMichal Privoznik <mprivozn@redhat.com>

With 'switch' we can utilize the compile time enum checks which we can't
rely on with plain 'if' conditions.
Signed-off-by: NShi Lei <shilei.massclouds@gmx.com>
Reviewed-by: NErik Skultety <eskultet@redhat.com>

Commit 77298458 changed the esx storage
adapter from busLogic to lsilogic, introducing a typo. Changing it back
to lsiLogic (with capital L) solves the issue. With this change, libvirt can now
create volumes in ESX again.

Thanks to Jaroslav Suchanek who figured out what was the issue in the
first place.

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1571759Signed-off-by: NMarcos Paulo de Souza <marcos.souza.org@gmail.com>

Despite being standardized in POSIX.1-2008, the 'm'
sscanf() modifier is currently not available on FreeBSD.

Reimplement parsing without sscanf() to work around the
issue.
Signed-off-by: NAndrea Bolognani <abologna@redhat.com>

Signed-off-by: NHan Han <hhan@redhat.com>
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1589115

When doing a memory snapshot qemuOpenFile() is used. This means
that the file where memory is saved is firstly attempted to be
created under root:root (because that's what libvirtd is running
under) and if this fails the second attempt is done under
domain's uid:gid. This does not make much sense - qemu is given
opened FD so it does not need to access the file. Moreover, if
dynamicOwnership is set in qemu.conf and the file lives on a
squashed NFS this is deadly combination and very likely to fail.

The fix consists of using:

  qemuOpenFileAs(fallback_uid = cfg->user,
                 fallback_gid = cfg->group,
                 dynamicOwnership = false)

In other words, dynamicOwnership is turned off for memory
snapshot (chown() will still be attempted if the file does not
live on NFS) and instead of using domain DAC label, configured
user:group is set as fallback.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>

Extend this existing test so that a case when IQN is provided is
tested too. Since a special iSCSI interface is created and its
name is randomly generated at runtime we need to link with
virrandommock to have predictable names.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

Some tests will want to pass their own callback data into the
testIscsiadmCbData callback. Introduce testIscsiadmCbData struct
to give this some form and order.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

This struct has nothing to do with testIscsiadmCb() rather than
testISCSIGetSession(). Move it closer to the latter.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

Introduce one basic test that tests the simplest case:
logging into portal without any IQN.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

The documentation to virCommandWait() function states that if
@exitstatus is NULL and command finished with error -1 is
returned. In other words, if @dryRunCallback is set and returns
an error (by setting its @status argument to a nonzero value) we
must propagate this error properly honouring the documentation
(and also regular run).
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

After a new iSCSI interface is successfully set up, we issue a
sendtargets command. However, after 56057900 we don't
update the host config which in turn makes login fail because
iscsiadm is unable to find any matching record for the interface.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

When scanning for targets, iSCSI might give different results
depending on the interface used. This is basically just name of
config file under /etc/iscsi/ifaces to use. The file contains
initiator IQN thus different results claim.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

Firstly, we can utilize virCommandSetOutputBuffer() API which
will collect the command output for us. Secondly, sscanf()-ing
through each line is easier to understand (and more robust) than
jumping over a string with strchr().
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

This is in fact 'cleanup' label and it should be named as such.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

Perform some method clean-up to follow more accepted coding standards:

 * Initialize @ret to error value and prove otherwise.
 * Initialize *ifacename to NULL
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

It wasn't updated when libvirtd was moved from daemon/ to src/
Reviewed-by: NJiri Denemark <jdenemar@redhat.com>
Signed-off-by: NCole Robinson <crobinso@redhat.com>

After my change to the original patch that resulted in commit
8ed874b3 it was brought to my attention that all three defines
are the same: FICLONE = BTRFS_IOC_CLONE = XFS_IOC_CLONE (as
documented in ioctl_ficlone(2)). Therefore we should prefer
generic FICLONE over 'specific' defines for btrfs/xfs.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
Reviewed-by: NAndrea Bolognani <abologna@redhat.com>

If we'd fail to enter or exit the monitor the saved error would be
leaked. Introduced in 8498a1e2 .

Pointed out by coverity.
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

Signed-off-by: NMarc Hartmayer <mhartmay@linux.ibm.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

Update the length @remote_params_len only if the related
@remote_params_val has also been set.
Signed-off-by: NMarc Hartmayer <mhartmay@linux.ibm.com>
Reviewed-by: NJohn Ferlan <jferlan@redhat.com>

qemuagenttest also depends on JSON object key ordering:
Invalid value of argument 'vcpus' of command 'guest-set-vcpus':
expected '[{"logical-id":1,"online":false}]' got '[{"online":false,"logical-id":1}]'

Skip it as well.
Signed-off-by: NJán Tomko <jtomko@redhat.com>

'disk-network-source-auth' and 'disk-network-tlsx509'
Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>

Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
Reviewed-by: NJán Tomko <jtomko@redhat.com>