It is possible to use block devices with domain save/restore. Upon
failure QEMU unlinks the path being saved to. This isn't good when
it is a block device !

* src/qemu/qemu_driver.c: Don't unlink block devices if save fails

If a transient QEMU crashes during save attempt, then the virDomainPtr
object may be freed. If a persistent QEMU crashes during save, then
the 'priv->mon' field is no longer valid since it will be inactive.

* src/qemu/qemu_driver.c: Fix two crashes when QEMU exits
  during a save attempt

In particular I was forgetting to take the qemuMonitorPrivatePtr
lock (via qemuDomainObjBeginJob), which would cause problems
if two users tried to access the same domain at the same time.
This patch also fixes a problem where I was forgetting to remove
a transient domain from the list of domains.

Thanks to Stephen Shaw for pointing out the problem and testing
out the initial patch.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

With JSON qemu monitor, we get a STOP event from qemu whenever qemu
stops guests CPUs. The downside of it is that vm->state is changed to
PAUSED and a new generic paused event is send to applications. However,
when we ask qemu to stop the CPUs we are not really interested in qemu
event and we usually want to issue a more specific event.

By setting vm->status to PAUSED before actually sending the request to
qemu (and resetting it back if the request fails) we can ignore the
event since the event handler does nothing when the guest is already
paused. This solution is quite hacky but unfortunately it's the best
solution which I was able to come up with and it doesn't introduce a
race condition.

While doing some testing of the snapshot code I noticed that
if qemuDomainSnapshotLoad failed, it would print a NULL as
part of the error.  That's not desirable, so leave the
full_path variable around until after we are done printing
errors.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

The hang fix in d376b7d6 was incomplete
since it left quite a few {Enter,Exit}Monitor calls which require driver
to be unlocked. Since the driver is locked throughout the whole
function, {Enter,Exit}MonitorWithDriver need to be used instead to
ensure driver is not locked when issuing monitor commands.

The comment in qemuDomainWaitForMigrationComplete says we are polling
every 50ms but the code sleeps only for 50us. This was already discussed
during review but apparently forgotten when the series was pushed.

First, inital VCPU pinning is set correctly but then it is reset by
assigning qemu process to a new cgroup (which contains all CPUs). It's
easily fixed by swapping these two actions.

The initial boot of VMs uses -device for NICs where available. The
corresponding monitor command is device_add, but the network hotplug
code was still using device_del by mistake.

* src/qemu/qemu_driver.c: Use device_add for NIC hotplug where
  available

Disk devices in QEMU have two parts, the guest device and the host
backend driver. Historically these two parts have had the same
"unique" name. With the switch to using -device though, they now
have separate names. Thus when changing CDROM media, for guests
using -device syntax, we need to prepend the QEMU_DRIVE_HOST_PREFIX
constant

* src/qemu/qemu_conf.c, src/qemu/qemu_conf.h: Add helper function
  qemuDeviceDriveHostAlias() for building a host backend alias
* src/qemu/qemu_driver.c: Use qemuDeviceDriveHostAlias() to determine
  the host backend alias for performing eject/change commands in the
  monitor

The QEMU driver is mistakenly calling directly into the text
mode monitor for the domain memory stats query.

* src/qemu/qemu_driver.c: Replace qemuMonitorTextGetMemoryStats with
  qemuMonitorGetMemoryStats
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h: Add the new
  wrapper for qemuMonitorGetMemoryStats
* src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_json.h: Add
  qemuMonitorJSONGetMemoryStats implementation

To avoid race-conditions, the tear down of a filter has to happen before
the tap interface disappears and another tap interface with the same
name can re-appear. This patch tries to fix this. In one place, where
communication with the qemu monitor may fail, I am only tearing the
filters down after knowing that the function did not fail.

I am also moving the tear down functions into an include file for other
drivers to reuse.

* src/qemu/qemu_driver.c (qemudDomainAttachSCSIDisk):
Initialize "cont" to NULL, so clang knows it's set.
Add an sa_assert so it knows it's non-NULL when dereferenced.

* src/qemu/qemu_driver.c (qemudDomainGetSecurityLabel): Remove store
and declaration.

* src/qemu/qemu_driver.c (qemuDomainSnapshotLoad): Remove dead store
into "snap", as well as its declaration.

* src/qemu/qemu_driver.c (qemudStartVMDaemon): Initialize "logfile"
to ensure that we don't use it uninitialized -- thus closing an
arbitrary file descriptor -- in the cleanup block.

and domainIsPersistent.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

When starting up qemu VNC autoport guests, we were
only looking through ports 5900 to 6000, meaning we
were limited to 100 total clients.  Increase that
limit to 65535 (the last available port), so we can
have up to 59635 VNC autoport guests.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

Signed-off-by: NChris Lalancette <clalance@redhat.com>

Signed-off-by: NChris Lalancette <clalance@redhat.com>

The images are saved in /var/lib/libvirt/qemu/save/
and named $domainname.save . The directory is created appropriately
at daemon startup. When a domain is started while a saved image is
available, libvirt will try to load this saved image, and start the
domain as usual in case of failure. In any case the saved image is
discarded once the domain is created.

* src/qemu/qemu_conf.h: adds an extra save path to the driver config
* src/qemu/qemu_driver.c: implement the 3 new operations and handling
  of the image directory

virDomainManagedSave() is to be run on a running domain. Once the call
complete, as in virDomainSave() the domain is stopped upon completion,
but there is no restore counterpart as any order to start the domain
from the API would load the state from the managed file, similary if
the domain is autostarted when libvirtd starts.
Once a domain has restarted his managed save image is destroyed,
basically managed save image can only exist for a stopped domain,
for a running domain that would be by definition outdated data.

* include/libvirt/libvirt.h.in src/libvirt.c src/libvirt_public.syms:
  adds the new entry points virDomainManagedSave(),
  virDomainHasManagedSaveImage() and virDomainManagedSaveRemove()
* src/driver.h src/esx/esx_driver.c src/lxc/lxc_driver.c
  src/opennebula/one_driver.c  src/openvz/openvz_driver.c
  src/phyp/phyp_driver.c src/qemu/qemu_driver.c src/vbox/vbox_tmpl.c
  src/remote/remote_driver.c src/test/test_driver.c src/uml/uml_driver.c
  src/xen/xen_driver.c: add corresponding new internal drivers entry
  points

virFileReadLimFD is a poor fit for reading the header
of the restore file.  The problem is that virFileReadLimFD
returns an error when there is more data after the amount
you ask to read, but that is *expected* in this case.

This patch is essentially a revert of
1a4d5c95, but I don't think
that commit does what it says anyway.  It purports to prevent
an unwarranted OOM error, but since virFileReadLimFD will
allocate memory up to the maximum anyway, the upper limit
on the total amount of memory allocated is the same for either
the old version or the new version.  Since the old saferead
actually works and virFileReadLimFD does not, revert to
using saferead.
Signed-off-by: NChris Lalancette <clalance@redhat.com>

When a watchdog/IO error occurs, one of the possible actions that
QEMU might take is to pause the guest. In this scenario libvirt
needs to update its internal state for the VM, and emit a
lifecycle event:

  VIR_DOMAIN_EVENT_SUSPENDED

with a detail being one of:

  VIR_DOMAIN_EVENT_SUSPENDED_IOERROR
  VIR_DOMAIN_EVENT_SUSPENDED_WATCHDOG

To future proof against possible QEMU support for multiple monitor
consoles, this patch also hooks into the 'STOPPED' event in QEMU
and emits a generic VIR_DOMAIN_EVENT_SUSPENDED_PAUSED event

* include/libvirt/libvirt.h.in: Add VIR_DOMAIN_EVENT_SUSPENDED_IOERROR
* src/qemu/qemu_driver.c: Update VM state to paused when IO error
  or watchdog events occurrs
* src/qemu/qemu_monitor_json.c: Fix typo in disk IO event name

The QEMU cpu affinity is used in NUMA scenarios to ensure that
guest memory is allocated from a specific node. Normally memory
is allocate on demand in vCPU threads, but when using hugepages
the initial thread leader allocates memory upfront. libvirt was
not setting affinity of the thread leader, or I/O threads. This
patch changes the code to set the process affinity in between
the fork()/exec() of QEMU. This ensures that every single QEMU
thread gets the affinity

* src/qemu/qemu_driver.c: Set affinity on entire QEMU process
  at startup

Right now this implements only 2 basic hooks:
- before the qemu process is being launched
- after the qemu process is terminated
the XML description of the domain is passed to the hook script stdin
/etc/libvirt/hook/qemu

* src/qemu/qemu_driver.c: implement synchronous script hooks for QEmu
  at domain startup and end

This flag is used in migration prepare step to send updated XML
definition of a guest.

Also ``virsh dumpxml --update-cpu [--inactive] guest'' command can be
used to see the updated CPU requirements.

When a domain is defined on host1, migrated to host2 and then migrated
back to host1, its current configuration would overwrite the libvirtd's
in-memory copy of persistent configuration of that domain. This is not
desired as we want to preserve the persistent configuration untouched.

This patch introduces new 'live' parameter to virDomainAssignDef.
Passing 'true' for 'live' means the configuration passed to
virDomainAssignDef describes a configuration of live instance of the
domain. This applies for saved domains which are being restored or for
incoming domains during migration.

All callers have been changed to pass the appropriate value.

Add support for Qemu to have firewall rules applied and removed on VM
startup and shutdown respectively. This  patch also provides support for
the updating of a filter that causes all VMs that reference the filter
to have their ebtables/iptables rules updated.
Signed-off-by: NStefan Berger <stefanb@us.ibm.com>

Use the new virDomainUpdateDeviceFlags API to allow the VNC password
to be changed on the fly

* src/internal.h: Define STREQ_NULLABLE() which is like STREQ()
  but does not crash if either argument is NULL, and treats two
  NULLs as equal.
* src/libvirt_private.syms: Export virDomainGraphicsTypeToString
* src/qemu/qemu_driver.c: Support VNC password change on a live
  machine
* src/qemu/qemu_monitor.c: Disable crazy debugging info. Treat a
  NULL password as "" (empty string), allowing passwords to be
  disabled in the monitor

To allow the new virDomainUpdateDeviceFlags() API to be universally
used with all drivers, this patch adds an impl to all the current
drivers which support CDROM or Floppy disk media change via the
current virDomainAttachDeviceFlags API

* src/qemu/qemu_driver.c, src/vbox/vbox_tmpl.c,
  src/xen/proxy_internal.c, src/xen/xen_driver.c,
  src/xen/xend_internal.c: Implement media change via the
  virDomainUpdateDeviceFlags API
* src/xen/xen_driver.h, src/xen/xen_hypervisor.c,
  src/xen/xen_inotify.c, src/xen/xm_internal.c,
  src/xen/xs_internal.c: Stubs for Xen driver entry points

The current virDomainAttachDevice API can be (ab)used to change
the media of an existing CDROM/Floppy device. Going forward there
will be more devices that can be configured on the fly and overloading
virDomainAttachDevice for this is not too pleasant. This patch adds
a new virDomainUpdateDeviceFlags() explicitly just for modifying
existing devices.

* include/libvirt/libvirt.h.in: Add virDomainUpdateDeviceFlags
* src/driver.h: Internal API for virDomainUpdateDeviceFlags
* src/libvirt.c, src/libvirt_public.syms: Glue public API to
  driver API
* src/esx/esx_driver.c, src/lxc/lxc_driver.c, src/opennebula/one_driver.c,
  src/openvz/openvz_driver.c, src/phyp/phyp_driver.c, src/qemu/qemu_driver.c,
  src/remote/remote_driver.c, src/test/test_driver.c, src/uml/uml_driver.c,
  src/vbox/vbox_tmpl.c, src/xen/xen_driver.c, src/xenapi/xenapi_driver.c: Add
  stubs for new driver entry point

This introduces a new event type

   VIR_DOMAIN_EVENT_ID_GRAPHICS

The same event can be emitted in 3 scenarios

  typedef enum {
      VIR_DOMAIN_EVENT_GRAPHICS_CONNECT = 0,
      VIR_DOMAIN_EVENT_GRAPHICS_INITIALIZE,
      VIR_DOMAIN_EVENT_GRAPHICS_DISCONNECT,
  } virDomainEventGraphicsPhase;

Connect/disconnect are triggered at socket accept/close.
The initialize phase is immediately after the protocol
setup and authentication has completed. ie when the
client is authorized and about to start interacting with
the graphical desktop

This event comes with *a lot* of potential information

 - IP address, port & address family of client
 - IP address, port & address family of server
 - Authentication scheme (arbitrary string)
 - Authenticated subject identity. A subject may have
   multiple identities with some authentication schemes.
   For example, vencrypt+sasl results in a x509dname
   and saslUsername identities.

This results in a very complicated callback :-(

   typedef enum {
      VIR_DOMAIN_EVENT_GRAPHICS_ADDRESS_IPV4,
      VIR_DOMAIN_EVENT_GRAPHICS_ADDRESS_IPV6,
   } virDomainEventGraphicsAddressType;

   struct _virDomainEventGraphicsAddress {
       int family;
       const char *node;
       const char *service;
   };
   typedef struct _virDomainEventGraphicsAddress virDomainEventGraphicsAddress;
   typedef virDomainEventGraphicsAddress *virDomainEventGraphicsAddressPtr;

   struct _virDomainEventGraphicsSubject {
      int nidentity;
      struct {
          const char *type;
          const char *name;
      } *identities;
   };
   typedef struct _virDomainEventGraphicsSubject virDomainEventGraphicsSubject;
   typedef virDomainEventGraphicsSubject *virDomainEventGraphicsSubjectPtr;

   typedef void (*virConnectDomainEventGraphicsCallback)(virConnectPtr conn,
                                                         virDomainPtr dom,
                                                         int phase,
                                                         virDomainEventGraphicsAddressPtr local,
                                                         virDomainEventGraphicsAddressPtr remote,
                                                         const char *authScheme,
                                                         virDomainEventGraphicsSubjectPtr subject,
                                                         void *opaque);

The wire protocol is similarly complex

   struct remote_domain_event_graphics_address {
     int family;
     remote_nonnull_string node;
     remote_nonnull_string service;
   };

   const REMOTE_DOMAIN_EVENT_GRAPHICS_IDENTITY_MAX = 20;

   struct remote_domain_event_graphics_identity {
     remote_nonnull_string type;
     remote_nonnull_string name;
   };

   struct remote_domain_event_graphics_msg {
     remote_nonnull_domain dom;
     int phase;
     remote_domain_event_graphics_address local;
     remote_domain_event_graphics_address remote;
     remote_nonnull_string authScheme;
     remote_domain_event_graphics_identity subject<REMOTE_DOMAIN_EVENT_GRAPHICS_IDENTITY_MAX>;
   };

This is currently implemented in QEMU for the VNC graphics
protocol, but designed to be usable with SPICE graphics in
the future too.

* daemon/remote.c: Dispatch graphics events to client
* examples/domain-events/events-c/event-test.c: Watch for
  graphics events
* include/libvirt/libvirt.h.in: Define new graphics event ID
  and callback signature
* src/conf/domain_event.c, src/conf/domain_event.h,
  src/libvirt_private.syms: Extend API to handle graphics events
* src/qemu/qemu_driver.c: Connect to the QEMU monitor event
  for VNC events and emit a libvirt graphics event
* src/remote/remote_driver.c: Receive and dispatch graphics
  events to application
* src/remote/remote_protocol.x: Wire protocol definition for
  graphics events
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
  src/qemu/qemu_monitor_json.c: Watch for VNC_CONNECTED,
  VNC_INITIALIZED & VNC_DISCONNETED events from QEMU monitor

This introduces a new event type

   VIR_DOMAIN_EVENT_ID_IO_ERROR

This event includes the action that is about to be taken
as a result of the watchdog triggering

  typedef enum {
     VIR_DOMAIN_EVENT_IO_ERROR_NONE = 0,
     VIR_DOMAIN_EVENT_IO_ERROR_PAUSE,
     VIR_DOMAIN_EVENT_IO_ERROR_REPORT,
  } virDomainEventIOErrorAction;

In addition it has the source path of the disk that had the
error and its unique device alias. It does not include the
target device name (/dev/sda), since this would preclude
triggering IO errors from other file backed devices (eg
serial ports connected to a file)

Thus there is a new callback definition for this event type

typedef void (*virConnectDomainEventIOErrorCallback)(virConnectPtr conn,
                                                     virDomainPtr dom,
                                                     const char *srcPath,
                                                     const char *devAlias,
                                                     int action,
                                                     void *opaque);

This is currently wired up to the QEMU block IO error events

* daemon/remote.c: Dispatch IO error events to client
* examples/domain-events/events-c/event-test.c: Watch for
  IO error events
* include/libvirt/libvirt.h.in: Define new IO error event ID
  and callback signature
* src/conf/domain_event.c, src/conf/domain_event.h,
  src/libvirt_private.syms: Extend API to handle IO error events
* src/qemu/qemu_driver.c: Connect to the QEMU monitor event
  for block IO errors and emit a libvirt IO error event
* src/remote/remote_driver.c: Receive and dispatch IO error
  events to application
* src/remote/remote_protocol.x: Wire protocol definition for
  IO error events
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
  src/qemu/qemu_monitor_json.c: Watch for BLOCK_IO_ERROR event
  from QEMU monitor

This introduces a new event type

   VIR_DOMAIN_EVENT_ID_WATCHDOG

This event includes the action that is about to be taken
as a result of the watchdog triggering

 typedef enum {
     VIR_DOMAIN_EVENT_WATCHDOG_NONE = 0,
     VIR_DOMAIN_EVENT_WATCHDOG_PAUSE,
     VIR_DOMAIN_EVENT_WATCHDOG_RESET,
     VIR_DOMAIN_EVENT_WATCHDOG_POWEROFF,
     VIR_DOMAIN_EVENT_WATCHDOG_SHUTDOWN,
     VIR_DOMAIN_EVENT_WATCHDOG_DEBUG,
 } virDomainEventWatchdogAction;

Thus there is a new callback definition for this event type

 typedef void (*virConnectDomainEventWatchdogCallback)(virConnectPtr conn,
                                                       virDomainPtr dom,
                                                       int action,
                                                       void *opaque);

* daemon/remote.c: Dispatch watchdog events to client
* examples/domain-events/events-c/event-test.c: Watch for
  watchdog events
* include/libvirt/libvirt.h.in: Define new watchdg event ID
  and callback signature
* src/conf/domain_event.c, src/conf/domain_event.h,
  src/libvirt_private.syms: Extend API to handle watchdog events
* src/qemu/qemu_driver.c: Connect to the QEMU monitor event
  for watchdogs and emit a libvirt watchdog event
* src/remote/remote_driver.c: Receive and dispatch watchdog
  events to application
* src/remote/remote_protocol.x: Wire protocol definition for
  watchdog events
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
  src/qemu/qemu_monitor_json.c: Watch for WATCHDOG event
  from QEMU monitor

This introduces a new event type

   VIR_DOMAIN_EVENT_ID_RTC_CHANGE

This event includes the new UTC offset measured in seconds.
Thus there is a new callback definition for this event type

 typedef void (*virConnectDomainEventRTCChangeCallback)(virConnectPtr conn,
                                                        virDomainPtr dom,
                                                        long long utcoffset,
                                                        void *opaque);

If the guest XML configuration for the <clock> is set to
offset='variable', then the XML will automatically be
updated with the new UTC offset value. This ensures that
during migration/save/restore the new offset is preserved.

* daemon/remote.c: Dispatch RTC change events to client
* examples/domain-events/events-c/event-test.c: Watch for
  RTC change events
* include/libvirt/libvirt.h.in: Define new RTC change event ID
  and callback signature
* src/conf/domain_event.c, src/conf/domain_event.h,
  src/libvirt_private.syms: Extend API to handle RTC change events
* src/qemu/qemu_driver.c: Connect to the QEMU monitor event
  for RTC changes and emit a libvirt RTC change event
* src/remote/remote_driver.c: Receive and dispatch RTC change
  events to application
* src/remote/remote_protocol.x: Wire protocol definition for
  RTC change events
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
  src/qemu/qemu_monitor_json.c: Watch for RTC_CHANGE event
  from QEMU monitor

The reboot event is not a normal lifecycle event, since the
virtual machine on the host does not change state. Rather the
guest OS is resetting the virtual CPUs. ie, the QEMU process
does not restart. Thus, this does not belong in the current
lifecycle events callback.

This introduces a new event type

    VIR_DOMAIN_EVENT_ID_REBOOT

It takes no parameters, besides the virDomainPtr, so it can
use the generic callback signature.

* daemon/remote.c: Dispatch reboot events to client
* examples/domain-events/events-c/event-test.c: Watch for
  reboot events
* include/libvirt/libvirt.h.in: Define new reboot event ID
* src/conf/domain_event.c, src/conf/domain_event.h,
  src/libvirt_private.syms: Extend API to handle reboot events
* src/qemu/qemu_driver.c: Connect to the QEMU monitor event
  for reboots and emit a libvirt reboot event
* src/remote/remote_driver.c: Receive and dispatch reboot
  events to application
* src/remote/remote_protocol.x: Wire protocol definition for
  reboot events

The libvirtd daemon impl will need to switch over to using the
new event APIs. To make this simpler, ensure all drivers currently
providing events support both the new APIs and old APIs.

* src/lxc/lxc_driver.c, src/qemu/qemu_driver.c, src/test/test_driver.c,
  src/vbox/vbox_tmpl.c, src/xen/xen_driver.c: Implement the new
  virConnectDomainEvent(Dereg|Reg)isterAny driver entry points

The internal domain events APIs are designed to handle the lifecycle
events. This needs to be refactored to allow arbitrary new event
types to be handled.

 * The signature of virDomainEventDispatchFunc changes to use
   virConnectDomainEventGenericCallback instead of the lifecycle
   event specific virConnectDomainEventCallback
 * Every registered callback gains a unique ID to allow its
   removal based on ID, instead of function pointer
 * Every registered callback gains an 'eventID' to allow callbacks
   for different types of events to be distinguished
 * virDomainEventDispatch is adapted to filter out callbacks
   whose eventID does not match the eventID of the event being
   dispatched
 * virDomainEventDispatch is adapted to filter based on the
   domain name and uuid, if this filter is set for a callback.
 * virDomainEvent type/detail fields are moved into a union to
   allow different data fields for other types of events to be
   added later

* src/conf/domain_event.h, src/conf/domain_event.c: Refactor
  to allow handling of different types of events
* src/lxc/lxc_driver.c, src/qemu/qemu_driver.c,
  src/remote/remote_driver.c, src/test/test_driver.c,
  src/xen/xen_driver.c: Change dispatch function signature
  to use virConnectDomainEventGenericCallback