This is pure code adjustment. The structure is going to be needed
later as it will hold a reference that will be used to talk to
virtlockd. However, so far this is no functional change just code
preparation.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

It's better if we stat() file that we are about to chown() at
first and check if there's something we need to change. Not that
it would make much difference, but for the upcoming patches we
need to be doing stat() anyway. Moreover, if we do things this
way, we can drop @chown_errno variable which will become
redundant.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Correctly mark the places where we need to remember and recall
file ownership. We don't want to mislead any potential developer.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

So we have this mechanism that on SIGUSR1 the virtlockd dumps its
internal state into a JSON file, reexec itself and the reloads
the internal state back. However, there's a bug in our
implementation:

  (gdb) signal SIGUSR1
  Continuing with signal SIGUSR1.
  [Thread 0x7fd094f7b700 (LWP 10602) exited]
  process 10600 is executing new program: /home/zippy/work/libvirt/libvirt.git/src/virtlockd
  warning: Could not load shared library symbols for linux-vdso.so.1.
  Do you need "set solib-search-path" or "set sysroot"?
  [Thread debugging using libthread_db enabled]
  Using host libthread_db library "/lib64/libthread_db.so.1".
  [New Thread 0x7fb28bc3c700 (LWP 14501)]

  Program received signal SIGSEGV, Segmentation fault.
  0x00007fb29133d530 in virExpandN (ptrptr=0x70, size=8, countptr=0x68, add=1, report=true, domcode=7, filename=0x7fb29138aeab "rpc/virnetserver.c", funcname=0x7fb29138b680 <__FUNCTION__.15821> "virNetServerAddProgram", linenr=661) at util/viralloc.c:288
  288         if (*countptr + add < *countptr) {
  (gdb) bt
  #0  0x00007fb29133d530 in virExpandN (ptrptr=0x70, size=8, countptr=0x68, add=1, report=true, domcode=7, filename=0x7fb29138aeab "rpc/virnetserver.c", funcname=0x7fb29138b680 <__FUNCTION__.15821> "virNetServerAddProgram", linenr=661) at util/viralloc.c:288
  #1  0x00007fb29132a267 in virNetServerAddProgram (srv=0x0, prog=0x7fb2915d08b0) at rpc/virnetserver.c:661
  #2  0x00007fb29131f27f in main (argc=1, argv=0x7fff8f771298) at locking/lock_daemon.c:1445

Notice the NULL @srv passed to frame 2? Usually, the @srv
variable is initialized on fresh start. However, in case of
daemon reload, the code path that is responsible for initializing
the value was not triggered and therefore we crashed immediately.
Fix this by always setting the variable.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Tunnelled migration can hang if the destination qemu exits despite all the
ABI checks. This happens whenever the destination qemu exits before the
complete transfer is noticed by source qemu. The savevm state checks at
runtime can fail at destination and cause qemu to error out.
The source qemu cant notice it as the EPIPE is not propogated to it.
The qemuMigrationIOFunc() notices the stream being broken from virStreamSend()
and it cleans up the stream alone. The qemuMigrationWaitForCompletion() would
never get to 100% transfer completion.
The qemuMigrationWaitForCompletion() never breaks out as well since
the ssh connection to destination is healthy, and the source qemu also thinks
the migration is ongoing as the Fd to which it transfers, is never
closed or broken. So, the migration will hang forever. Even Ctrl-C on the
virsh migrate wouldn't be honoured. Close the source side FD when there is
an error in the stream. That way, the source qemu updates itself and
qemuMigrationWaitForCompletion() notices the failure.

Close the FD for all kinds of errors to be sure. The error message is not
copied for EPIPE so that the destination error is copied instead later.

Note:
Reproducible with repeated migrations between Power hosts running in different
subcores-per-core modes.
Signed-off-by: NShivaprasad G Bhat <sbhat@linux.vnet.ibm.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1264008

The existing algorithm assumed that someone was making small, incremental
changes; however, it is possible to change iothreads from 0 (or relatively
small number) to some really large number and the algorithm would possibly
spin its wheels doing unnecessary searches.

So, optimize the algorithm using a bitmap to find available iothread_id's
starting at 1 that aren't already defined by a "<thread id='#'>" and
filling in the iothreadids array with those iothread_id values.

https://bugzilla.redhat.com/show_bug.cgi?id=1249981

When qemuDomainPinIOThread was added in commit id 'fb562614', a check
for the IOThread capability was not needed since a check for iothreadpids
covered the condition where the support for IOThreads was not present.
The iothreadpids array was only created if qemuProcessDetectIOThreadPIDs
was able to query the monitor for IOThreads. It would only do that if
the QEMU_CAPS_OBJECT_IOTHREAD capability was set.

However, when iothreadids were added in commit id '8d4614a5' and the
check for iothreadpids was replaced by a search through the iothreadids[]
array for the matching iothread_id that left open the possibility that
an iothreadids[] array was defined, but the entries essentially pointed
to elements with only the 'iothread_id' defined leaving the 'thread_id'
value of 0 and eventually the cpumap entry of NULL.

This was because, the original IOThreads commit id '72edaae7' only
checked if IOThreads were defined and if the emulator had the IOThreads
capability, then IOThread objects were added at startup. The "capability
failure" check was only done when a disk was assigned to an IOThread in
qemuCheckIOThreads. This was because the initial implementation had no way
to dynamically add IOThreads, but it was possible to dynamically add a
disk to the domain. So the decision was if the domain supported it, then
add the IOThread objects. Then if a disk with an IOThread defined was
added, it could check the capability and fail to add if not there. This
just meant the 'iothreads' value was essentially ignored.

Eventually commit id 'a27ed6e7' allowed for the dynamic addition and
deletion of IOThread objects. So it was no longer necessary to generate
IOThread objects to dynamically attach a disk to. However, the startup
and disk check code was not modified to reflect this.

This patch will move the capability failure check to when IOThread
objects are being added to the command line. Thus a domain that has
IOThreads defined will not be started if the emulator doesn't support
the capability. This means when qemuCheckIOThreads is called to add
a disk, it's no longer necessary to check the capability. Instead the
code can use the IOThreadFind call to indicate that the IOThread
doesn't exist.

Finally because it could be possible to have a domain running with the
iothreadids[] defined prior to this change if libvirtd is restarted each
having mostly empty elements, qemuProcessDetectIOThreadPIDs will check
if there are niothreadids when the QEMU_CAPS_OBJECT_IOTHREAD capability
check fails and remove the elements and array if it exists.

With these changes in place, it turns out the cputune-numatune test
was failing because the right bit wasn't set in the test. So used the
opportunity to fix that and create a test that would expect to fail
with some sort of iothreads defined and used, but not having the
correct capability.

If there are no IOThreads defined, no sense making other checks

Although theoretically both should be the same value, the niothreadids
should be used in favor of iothreads when performing comparisons. This
leaves the iothreads as a purely numeric value to be saved in the config
file.  The one exception to the rule is virDomainIOThreadIDDefArrayInit
where the iothreadids are being generated from the iothreads count since
iothreadids were added after initial iothreads support.

Event implementations need to be registered before a connection to the
Hypervisor is opened, otherwise event handling can be impaired (e.g.
delayed messages). This fact is referenced in an e-mail [1], but should
also be noted in the documentation of the registration functions.

[1] https://www.redhat.com/archives/libvirt-users/2014-April/msg00011.html

Don't compare a bool variable against the literal, "true".
Signed-off-by: NWei Jiangang <weijg.fnst@cn.fujitsu.com>
Signed-off-by: NJiri Denemark <jdenemar@redhat.com>

seclets ==> selects
qualfied ==> qualified
Signed-off-by: NWei Jiangang <weijg.fnst@cn.fujitsu.com>
Signed-off-by: NJiri Denemark <jdenemar@redhat.com>

Create a separate local API that will fill in the iothreadid array
entries that were not defined by <iothread id='#'> entries in the XML.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

Rework the code in order to use the "ret = -1;" and goto cleanup;
coding style.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

After a successful creation of a directory, if some other call results
in returning a failure, let's remove the directory we created to
prevent another round trip or confusion in the caller. In particular, this
function can be called during a storage backend buildVol, so in order
to ensure that caller doesn't need to distinguish between failed create
or some other failure after create, just remove the directory we created.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

After a successful creation of a file, if some other call results
in returning a failure, let's unlink the file we created to prevent
another round trip or confusion in the caller. In particular, this
function can be called during a storage backend buildVol, so in order
to ensure that caller doesn't need to distinguish between failed create
or some other failure after create, just remove the volume we created.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1233003

Track when the logical volume was successfully created in order to
properly handle the call to virStorageBackendLogicalDeleteVol. It's
possible that the failure to create was because someone created an
LV in the pool outside of libvirt's knowledge. In this case, we don't
want to delete that LV.  A subsequent or future refresh of the pool
will find the volume and cause an earlier failure
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

Commit id '1b5685da' refactored the code to move buildvoldef inside
the buildVol conditional; however, the VIR_FREE of the memory was
left only when 'buildret' failed, thus we're leaking memory.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

As of commit id '155ca616' a 'refreshVol' is called after a buildVol
succeeds in storageVolCreateXML, thus a volStorageBackendSheepdogRefreshVolInfo
call in virStorageBackendSheepdogBuildVol is no longer necessary.

Additionally, the 'conn' parameter becomes unused.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

As of commit id '155ca616' a 'refreshVol' is called after the buildVol
succeeds in storageVolCreateXML, thus the volStorageBackendRBDRefreshVolInfo
call in virStorageBackendRBDBuildVol is no longer necessary.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>

Without this, building on cygwin fails with:

  CC       libvirt_admin_la-libvirt-admin.lo
libvirt-admin.c:25:21: fatal error: rpc/rpc.h: No such file or directory
 #include <rpc/rpc.h>
                     ^
Reported-by: NYaakov Selkowitz <yselkowi@redhat.com>
Signed-off-by: NEric Blake <eblake@redhat.com>

Our apibuild.py script does not cope with ATTRIBUTE_NONNULL:

Parse Error: parsing function type, ')' expected
Got token  ('name', 'char')
Last token:  ('name', 'char')
Token queue:  [('op', '*'), ('name', 'dconnuri'), ('sep', ')')]
Line 3297 end:
Makefile:2441: recipe for target '../../docs/apibuild.py.stamp' failed

Let's drop it. Moreover, up until e17ae3cc where it was
introduced we did not really care about NULL-ity of dconnuri. And
moreover the ATTRIBUTE_NONNULL merely checks for static calls
over NULL, it won't catch the dynamic ones, where a NULL is
passed by a variable at runtime.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

https://bugzilla.redhat.com/show_bug.cgi?id=1256999

After creating a copy of the 'authdef' in a pool -> disk translation,
unconditionally clear the 'authType' in the resulting disk auth def
structure since that's used for a storage pool and not a disk.  This
ensures virStorageAuthDefFormat will properly format the <auth> XML
for a <disk> (e.g. it won't have a <auth type='%s'.../>).

Check dconnuri is not null or we will catch nullpointer later.
I hope this makes Coverity happy.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

virDomainMigrateUnmanagedParams is not a good candidate for this functionality
as it is used by migrate family functions too and its have its own checks that
are superset of extracted and we don't need to check twice.

Actually name of the function is slightly misleading as there is also a check
for consistensy of flags parameter alone. So it could be refactored further and
reused by all migrate functions but for now let it be a matter of a different
patchset.

It is *not* a pure refactoring patch as it introduces offline check for older
versions. Looks like it must be done that way and no one will be broken too.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

Finally on this step we get what we were aimed for - toURI{1, 2} (and
migration{*} APIs too) now can work thru V3_PARAMS protocol. Execution path
goes thru unchanged virDomainMigrateUnmanaged adapter function which is called
by all target places.

Note that we keep the fact that direct migration never works
thru V3_PARAMS proto. We can't change this aspect without
further investigation.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

Move virDomainMigrateUnmanagedProto* expected params list check into
function itself and use common virTypedParamsCheck for this purpose.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

Extract parameter adaptation and checking which is protocol dependent into
designated functions. Leave only branching and common checks in
virDomainMigrateUnmanagedParams.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

Let's put main functionality into params version of virDomainMigrateUnmanaged
as a preparation step for merging it with virDomainMigratePeer2PeerParams.
virDomainMigrateUnmanaged then does nothing more then just adapting arguments.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

p2p plain and direct function are good candidates for code reuse. Their main
function is same - to branch among different versions of migration protocol and
implementation of this function is also same. Also they have other common
functionality in lesser aspects. So let's merge them.

But as they have different signatures we have to get to convention on how to
pass direct migration 'uri' in 'dconnuri' and 'miguri'. Fortunately we alreay
have such convention in parameters passed to toURI2 function, just let's follow
it. 'uri' is passed in miguri and dconnuri is ignored.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

We use miguri name for this parameter in other places. So
make naming more consitent.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

Direct migration should work if *perform3 is present but *perform
is not. This is situation when driver migration is implemented
after new version of driver function is introduced. We should not
be forced to support old version too as its parameter space is
subspace of newer one.

This is more structured code so it will be easier to add branch for _PARAMS
protocol here. It is not a pure refactoring strictly speaking as we remove
scenarios for broken cases when driver defines V3 feature and implements
perform function. So it is additionally a more solid code.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

Refactor dconnuri local server URI check to common API.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

'useParams' parameter usage is an example of control coupling. Most of the work
inside the function is done differently except for the uri check. Lets split
this function into two, one with extensible parameters set and one with hardcoded
parameter set.
Signed-off-by: NNikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>

The internal representation of a JSON array counts the items in
size_t. However, for some reason, when asking for the count it's
reported as int. Firstly, we need the function to return a signed
type as it's returning -1 on an error. But, not every system has
integer the same size as size_t. Therefore, lets return ssize_t.
Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>

Commit 4e803227 used $(builddir) in the header search
path to fix a build issue; however, $(builddir) is not defined
by old autoconf versions such as the one available in CentOS 5,
resulting in the following error:

  cc1: error: /util: No such file or directory
  make[3]: *** [libvirt_driver_la-fdstream.lo] Error 1

Since $(builddir) is defined to always be '.', just use that
value directly instead.

Since a9fe6203, we are generating virkeymaps.h at build
time; however, we are not including $(builddir)/util in the
header search path, so when doing a VPATH build the compiler
is unable to locate the file.

make[2]: Entering directory
`/home/jenkins/libvirt/systems/libvirt-fedora-20/build/src'
  GEN      util/virkeymaps.h
  ...
  CC       util/libvirt_util_la-virkeycode.lo
  CC       util/libvirt_util_la-virkeyfile.lo
  CC       util/libvirt_util_la-virlockspace.lo
  CC       util/libvirt_util_la-virlog.lo
../../src/util/virkeycode.c:27:24: fatal error: virkeymaps.h: No such file or directory
 #include "virkeymaps.h"
                        ^
compilation terminated.

Remove unused definitions, functions and structure fields.
Signed-off-by: NMaxim Nestratov <mnestratov@virtuozzo.com>

Cleanup calls to virStorageBackendCopyToFD a bit.
Signed-off-by: NJohn Ferlan <jferlan@redhat.com>