Detected by clang.

* src/qemu/qemu_migration.c (qemuMigrationToFile): Nothing later
uses is_reg.

clang didn't like the last increment to nargs.  But why even
track nargs ourselves, when virCommand does it for us?

* src/storage/storage_backend_iscsi.c
(virStorageBackendISCSIConnection): Switch to virCommand to avoid
a dead-store warning on nargs.

Clang detected a dead store to rc.  It turns out that in fixing this,
I also found a FILE* leak.

This is a subtle change in behavior, although unlikely to hit.  The
pidfile is a kernel file, so we've probably got more serious problems
under foot if we fail to parse one.  However, the previous behavior
was that even if one pid file failed to parse, we tried others,
whereas now we give up on the first failure.  Either way, though,
the function returns -1, so the caller will know that something is
going wrong, and that not all pids were necessarily reaped.  Besides,
there were other instances already in the code where failure in the
inner loop aborted the outer loop.

* src/util/cgroup.c (virCgroupKillInternal): Abort rather than
resuming loop on fscanf failure, and cleanup file on error.

Clang 2.8 wasn't quite able to follow that persistentDef was
assigned earlier if (flags & VIR_DOMAIN_MEM_CONFIG) is true.
Silence this false positive, to make clang analysis easier to use.

* src/qemu/qemu_driver.c (qemudDomainSetMemoryFlags): Add an
annotation to silence clang's claim of a NULL dereference.

Clang detected a null-pointer dereference regression, introduced
in commit 4e8969eb.  Without this patch, a device with
unbind_from_stub set to false would eventually try to call
virFileExists on uncomputed drvdir.

* src/util/pci.c (pciUnbindDeviceFromStub): Ensure drvdir is set
before use.

This code has had problems historically.  As originally
written, in commit 6bcf2501 (Jun 08), it could call unlink
on a random string, nuking an unrelated file.

Then commit 182a80b9 (Sep 09), the code was rewritten to
allocate tmp, with both a use-after-free bug and a chance to
call unlink(NULL).

Commit e206946d (Mar 11) fixed the use-after-free, but not the
NULL dereference.  Thanks to clang for catching this!

* src/qemu/qemu_driver.c (qemudDomainMemoryPeek): Don't call
unlink on NULL.

This reverts commit 0e7f7f85.

From the mailing list:

> So, AFAICT, this patch means we will never reconnect to any LXC
> VMs now.
>
> The correct solution, is to refactor LXC driver startup to work
> the same way as the QEMU driver startup.
>
>   - Load all the live state XML files (to pick up running VMs)
>   - Reconnect to all VMs
>   - Load all the persistent config XML files (to pick up any additional
>     inactive guets)

But that solution is invasive enough to be post-0.9.1.

The patch which moved libpciaccess initialization to one place caused
regression - we were not able to run on system with no PCI bus, like
s390(x).

Otherwise if there are inactive lxc domains, lxcStartup will
try to reconnect to sockets of these domains, which results in
errors in libvirtd log.

This commit fixes
qemu/qemu_driver.c: In function 'qemuDomainModifyDeviceFlags':
qemu/qemu_driver.c:4041:8: warning: 'ret' may be used uninitialized in this
function [-Wuninitialized]
qemu/qemu_driver.c:4013:9: note: 'ret' was declared here

The variable is set to -1 so that the error paths are taken when the code
to set it didn't get a chance to run. Without initializing it, we could
return some an undefined value from this function.

While I was at it, I made a trivial whitespace change in the same function
to improve readability.

Call shutdown functions for all subcomponents in nwfilterDriverShutdown.

Make sure that this shutdown functions can safely be called multiple times
and independent from the actual subcomponents state.

Commit e0d014f2 made binary potentially allocated on the heap.
It was freed in the parent in the error path, but not in the success path
that doesn't goto the cleanup label.

Found by 'make -C tests valgrind'.

Commit 1671d1dc introduced a memory leak in virHashFree, and
wholesale table corruption in virHashRemoveSet (elements not
requested to be freed are lost).

* src/util/hash.c (virHashFree): Free bucket array.
(virHashRemoveSet): Don't lose elements.
* tests/hashtest.c (testHashCheckForEachCount): New method.
(testHashCheckCount): Expose the bug.

Support update of disks by MODIFY_CONFIG

This patch includes changes for qemu's disk to support
virDomainUpdateDeviceFlags() with VIR_DOMAIN_DEVICE_MODIFY_CONFIG.

This patch adds support for CDROM/foppy disk types.
Signed-off-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

* src/qemu/qemu_driver.c
(qemuDomainUpdateDeviceConfig): support cdrom/floppy.

V2: Use virAsprintf instead of snprintf/strdup

The xend driver will generate a virDomainNetDef ifname if one is not
specified in xend sexpr, even if domain is inactive.  The result is
network interface XML containing 'vif-1.Y' on dev attribute of target
element, e.g.

  <interface type='bridge'>
    <target dev='vif-1.0'/>
    ...

This patch changes the behavior to only generate the ifname if not
specified in xend sexpr *and* domain is not inactive (id != -1).

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=664059

Reattaching pci device back to host without destroying guest or
detaching device from guest would cause host to crash. This patch adds
a check before doing device reattach. If the device is being assigned
to guest, libvirt refuses to reattach device to host. The patch only
works for Xen, for it just checks xenstore to get pci device
information.
Signed-off-by: NYufang Zhang <yuzhang@redhat.com>

The lone caller to hostsFileWrite (and the callers for at least 3
levels up the return stack) assume that the return value will be < 0
on failure. However, hostsFileWrite returns 0 on success, and a
positive errno on failure. This patch changes hostsFileWrite to return
-errno on failure.

* src/esx/esx_driver.c: Fix spelling of 'relative'.
* src/util/util.c: Likewise.

We support to initialize the hooks at daemon reload if there is no
hooks script is defined, we should also support initialize the hooks
at daemon shutdown if no hooks is defined.

To address bz: https://bugzilla.redhat.com/show_bug.cgi?id=688859

Support changes of disks by MODIFY_CONFIG for qemu.

This patch includes patches for qemu's disk to support
virDomainAt(De)tachDeviceFlags with VIR_DOMAIN_DEVICE_MODIFY_CONFIG.

Other devices can be added incrementally.
Signed-off-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

* /src/conf/domain_conf.c
(virDomainDiskIndexByName): returns array index of disk in vmdef.
(virDomainDiskRemoveByName): removes a disk which has the name in vmdef.
* src/qemu/qemu_driver.c
(qemuDomainAttachDeviceConfig): add support for Disks.
(qemuDomainDetachDeviceConfig): add support for Disks.

This patch adds functions for modify domain's persistent definition.
To do error recovery in easy way, we use a copy of vmdef and update it.

The whole sequence will be:

  make a copy of domain definition.

  if (flags & MODIFY_CONFIG)
      update copied domain definition
  if (flags & MODIF_LIVE)
      do hotplug.
  if (no error)
      save copied one to the file and update cached definition.
  else
      discard copied definition.

This patch is mixuture of Eric Blake's work and mine.
From: Eric Blake <eblake@redhat.com>
Signed-off-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

(virDomainObjCopyPersistentDef): make a copy of persistent vm definition
(qemuDomainAttach/Detach/UpdateDeviceConfig) : callbacks. now empty
(qemuDomainModifyDeviceFlags): add support for MODIFY_CONFIG and MODIFY_CURRENT

So far first entries for each hash key are stored directly in the hash
table while other entries mapped to the same key are linked through
pointers. As a result of that, the code is cluttered with special
handling for the first items.

This patch makes all entries (even the first ones) linked through
pointers, which significantly simplifies the code and makes it more
maintainable.

This adds several tests for remaining hash APIs (custom
hasher/comparator functions are not covered yet, though).

All tests pass both before and after the "Simplify hash implementation".

Steps to reproduce this bug:
1. # cat net.xml # 00:03.0 has been used
    <interface type='network'>
      <mac address='52:54:00:04:72:f3'/>
      <source network='default'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x03' function='0x0'/>
    </interface>

2. # virsh attach-device vm1 net.xml
   error: Failed to attach device from net.xml
   error: internal error unable to reserve PCI address 0:0:3

3. # virsh attach-device vm1 net.xml
   error: Failed to attach device from net.xml
   error: internal error unable to execute QEMU command 'device_add': Device 'rtl8139' could not be initialized

The reason of this bug is that: we can not reserve PCI address 0:0:3 because it has
been used, but we release PCI address when we reserve it failed.

When buf->error is 1, we do not return buf->content in the function
virBufferContentAndReset(). So we should free buf->content when
vsnprintf() failed.

This was broken by the refactoring in ac1e6586. It resulted in a
segfault for 'virsh vol-dumpxml' and related volume functions.

Before the refactoring all users of the ESX_VI__TEMPLATE__DISPATCH
macro dispatched on the item type, as the item is the input to all those
functions.

Commit ac1e6586 made the dynamically dispatched CastFromAnyType
functions use this macro too, but this functions dispatched on the
actual type of the AnyType object. The item is the output of the
CastFromAnyType functions.

This difference was missed in the refactoring, making CastFromAnyType
functions dereferencing the item pointer that is NULL at the time of
the dispatch.

Suggested by Daniel P. Berrange

* .gnulib: Update to latest for passfd fixes.
* bootstrap.conf (gnulib_modules): Add passfd.
* src/util/util.c (virFileOpenAs): Simplify.

They don't make any sense when used together.

otherwise the directory returned by networkDnsmasqLeaseFileName will not
be created if ipdef->nhosts == 0 in networkBuildDnsmasqArgv.

Found by 'make -C tests valgrind'.

xen_xm.c: Dummy allocation via virDomainChrDefNew is directly
overwritten and lost. Free 'script' in success path too.

vmx.c: Free virtualDev_string in success path too.

domain_conf.c: Free compression in success path too.

This will be used in the ESX driver event handling.

We can exploit the fact that gcc warns about int-to-pointer conversion
in ternary cond?(void*):(int) in order to prevent future mistakes of
calling VIR_FREE on a scalar lvalue.  For example, between commits
158ba873 and 802e2df9, we would have had this warning:

cc1: warnings being treated as errors
remote.c: In function 'remoteDispatchListNetworks':
remote.c:3684:70: error: pointer/integer type mismatch in conditional expression

There are still a number of places that malloc into a const char*;
while it would probably be worth scrubbing them to use char*
instead, that is a separate patch, so we have to cast away const
in VIR_FREE for now.

* src/util/memory.h (VIR_FREE): Make gcc warn about integers.
Iteratively developed from a patch by Christophe Fergeau.

mingw lacks the counterpart to PTHREAD_MUTEX_INITIALIZER, so the
best we can do is portably expose once-only runtime initialization.

* src/util/threads.h (virOnceControlPtr): New opaque type.
(virOnceFunc): New callback type.
(virOnce): New prototype.
* src/util/threads-pthread.h (virOnceControl): Declare.
(VIR_ONCE_CONTROL_INITIALIZER): Define.
* src/util/threads-win32.h (virOnceControl)
(VIR_ONCE_CONTROL_INITIALIZER): Likewise.
* src/util/threads-pthread.c (virOnce): Implement in pthreads.
* src/util/threads-win32.c (virOnce): Implement in WIN32.
* src/libvirt_private.syms: Export it.

To be used to share a CURL handle between multiple threads in the
upcoming domain event support.

This patch strips reusable part of qemuDomainUpdateDeviceFlags()
and consolidate it to qemuDomainModifyDeviceFlags().
No functional changes.

* src/qemu/qemu_driver.c
(qemuDomainChangeDiskMediaLive) : pulled out code for updating disks.
(qemuDomainUpdateDeviceLive) : core of UpdateDevice, extracted from
UpdateDeviceFlags()
(qemuDomainModifyDeviceFlags): add support for updating device in live domain.
(qemuDomainUpdateDeviceFlags): reworked as a wrapper function of
qemuDomainModifyDeviceFlags()
Signed-off-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: NEric Blake <eblake@redhat.com>

clean up At(De)tachDeviceFlags() for consolidation.

qemuDomainAttachDeviceFlags()/qemuDomainDetachFlags()/
qemuDomainUpdateDeviceFlags() has similar logics and copied codes.

This patch series tries to unify them to use shared code when it can.
At first, clean up At(De)tachDeviceFlags() and devide it into functions.

By this, this patch pulls out shared components between functions.
Based on patch series by Eric Blake, I added some modification as
switch-case with QEMU_DEVICE_ATTACH, QEMU_DEVICE_DETACH, QEMU_DEVICE_UPDATE

* src/qemu/qemu_driver.c
(qemuDomainAt(De)tachDeviceFlags) : pulled out to qemuDomainModifyDeviceFlags()
(qemuDomainModifyDeviceFlags) : implements generic code for modifying domain.
(qemuDomainAt(De)tachDeviceFlagsLive) : code for at(de)taching devices to
domain in line. no changes in logic from old code.
(qemuDomainAt(De)tachDeviceDiskLive) : for at(de)taching Disks.
(qemuDomainAt(De)tachDeviceControllerLive) : for at(de)taching Controllers
Signed-off-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: NEric Blake <eblake@redhat.com>