libvirt/src/conf/object_event.c

1047 lines
30 KiB
C
Raw Normal View History

/*
* object_event.c: object event queue processing helpers
*
* Copyright (C) 2010-2014 Red Hat, Inc.
* Copyright (C) 2008 VirtualIron
* Copyright (C) 2013 SUSE LINUX Products GmbH, Nuernberg, Germany.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library. If not, see
* <http://www.gnu.org/licenses/>.
*
* Author: Ben Guthro
*/
#include <config.h>
#include "domain_event.h"
#include "network_event.h"
#include "object_event.h"
#include "object_event_private.h"
#include "virlog.h"
#include "datatypes.h"
#include "viralloc.h"
#include "virerror.h"
#include "virstring.h"
#define VIR_FROM_THIS VIR_FROM_NONE
struct _virObjectEventCallbackList {
unsigned int nextID;
size_t count;
virObjectEventCallbackPtr *callbacks;
};
struct _virObjectEventQueue {
size_t count;
virObjectEventPtr *events;
};
typedef struct _virObjectEventQueue virObjectEventQueue;
typedef virObjectEventQueue *virObjectEventQueuePtr;
struct _virObjectEventState {
/* The list of domain event callbacks */
virObjectEventCallbackListPtr callbacks;
/* The queue of object events */
virObjectEventQueuePtr queue;
/* Timer for flushing events queue */
int timer;
/* Flag if we're in process of dispatching */
bool isDispatching;
virMutex lock;
};
struct _virObjectEventCallback {
int callbackID;
virClassPtr klass;
int eventID;
virConnectPtr conn;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
int remoteID;
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
bool uuid_filter;
unsigned char uuid[VIR_UUID_BUFLEN];
virConnectObjectEventGenericCallback cb;
void *opaque;
virFreeCallback freecb;
bool deleted;
event: don't allow mix of old- and new-style registration Consider these two calls, in either order: id1 = virConnectDomainEventRegisterAny(conn, NULL, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); Right now, the second call fails, because under the hood, the old-style function registration is tightly coupled to the new style lifecycle eventID, and the two calls both try to register the same global eventID callback representation. We've alreay documented that users should avoid old-style registration and deregistration, so anyone heeding the advice won't run into this situation. But it would be even nicer if we pretend the two interfaces are completely separate, and disallow any cross-linking. That is, a call to old-style deregister should never remove a new-style callback even if it is the same function pointer, and a call to new-style callback using only callbackIDs obtained legitimately should never remove an old-style callback (of course, since our callback IDs are sequential, and there is still coupling under the hood, you can easily guess the callbackID of an old style registration and use new-style deregistration to nuke it - but that starts to be blatantly bad coding on your part rather than a surprising result on what looks like reasonable stand-alone API). With this patch, you can now register a global lifecycle event handler twice, by using both old and new APIs; if such an event occurs, your callback will be entered twice. But that is not a problem in practice, since it is already possible to use the new API to register both a global and per-domain event handler using the same function, which will likewise fire your callback twice for that domain. Duplicates are still prevented when using the same API with same parameters twice (old-style twice, new-style global twice, or new-style per-domain with same domain twice), and things are still bounded (it is not possible to register a single function pointer more than N+2 times per event id, where N is the number of domains available on the connection). Besides, it has always been possible to register as many separate function pointers on the same event id as desired, through either old or new style API, where the bound there is the physical limitation of writing a program with enough distinct function pointers. Adding another event registration in the testsuite is sufficient to cover this, where the test fails without the rest of the patch. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventCallbackLookup): Add argument. (virObjectEventCallbackListAddID, virObjectEventStateCallbackID): Adjust callers. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 12:55:55 +00:00
bool legacy; /* true if end user does not know callbackID */
};
static virClassPtr virObjectEventClass;
static void virObjectEventDispose(void *obj);
2013-12-11 10:38:03 +00:00
static int
virObjectEventOnceInit(void)
{
if (!(virObjectEventClass =
virClassNew(virClassForObject(),
"virObjectEvent",
sizeof(virObjectEvent),
virObjectEventDispose)))
return -1;
return 0;
}
VIR_ONCE_GLOBAL_INIT(virObjectEvent)
/**
* virClassForObjectEvent:
*
* Return the class object to be used as a parent when creating an
* event subclass.
*/
2013-12-11 10:38:03 +00:00
virClassPtr
virClassForObjectEvent(void)
{
if (virObjectEventInitialize() < 0)
return NULL;
return virObjectEventClass;
}
2013-12-11 10:38:03 +00:00
static void
virObjectEventDispose(void *obj)
{
virObjectEventPtr event = obj;
VIR_DEBUG("obj=%p", event);
VIR_FREE(event->meta.name);
}
/**
* virObjectEventCallbackListFree:
* @list: event callback list head
*
* Free the memory in the domain event callback list
*/
static void
virObjectEventCallbackListFree(virObjectEventCallbackListPtr list)
{
size_t i;
if (!list)
return;
for (i = 0; i < list->count; i++) {
virFreeCallback freecb = list->callbacks[i]->freecb;
if (freecb)
(*freecb)(list->callbacks[i]->opaque);
VIR_FREE(list->callbacks[i]);
}
VIR_FREE(list->callbacks);
VIR_FREE(list);
}
event: properly filter count of remaining events On the surface, this sequence of API calls should succeed: id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); id2 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_RTC_CHANGE,...); virConnectDomainEventDeregisterAny(id1); id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); And for test:///default, it does. But for qemu:///system, it fails: libvirt: XML-RPC error : internal error: domain event 0 already registered Looking closer, the bug is caused by miscommunication between the object event engine and the client side of the remote driver. In our implementation, we set up a single server-side event per eventID, then the client side replicates that one event to all callbacks that have been registered client side. To know when to turn the server side eventID on or off, the client side must track how many events for the same eventID have been registered. But while our code was filtering by eventID on event registration, it did not filter on event deregistration. So the above API calls resulted in the deregister returning 1 instead of 0, so no RPC deregister was issued, and the final register detects on the server side that the server is already handling eventID 0. Unfortunately, since the problem is only observable on remote connections, it's not possible to enhance objecteventtest to expose the semantics using only public API entry points. * src/conf/object_event.c (virObjectEventCallbackListCount): New function. (virObjectEventCallbackListAddID) (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Use it. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 21:50:15 +00:00
/**
* virObjectEventCallbackListCount:
* @conn: pointer to the connection
* @cbList: the list
* @klass: the base event class
* @eventID: the event ID
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* @uuid: optional uuid of per-object filtering
* @serverFilter: true if server supports object filtering
event: properly filter count of remaining events On the surface, this sequence of API calls should succeed: id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); id2 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_RTC_CHANGE,...); virConnectDomainEventDeregisterAny(id1); id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); And for test:///default, it does. But for qemu:///system, it fails: libvirt: XML-RPC error : internal error: domain event 0 already registered Looking closer, the bug is caused by miscommunication between the object event engine and the client side of the remote driver. In our implementation, we set up a single server-side event per eventID, then the client side replicates that one event to all callbacks that have been registered client side. To know when to turn the server side eventID on or off, the client side must track how many events for the same eventID have been registered. But while our code was filtering by eventID on event registration, it did not filter on event deregistration. So the above API calls resulted in the deregister returning 1 instead of 0, so no RPC deregister was issued, and the final register detects on the server side that the server is already handling eventID 0. Unfortunately, since the problem is only observable on remote connections, it's not possible to enhance objecteventtest to expose the semantics using only public API entry points. * src/conf/object_event.c (virObjectEventCallbackListCount): New function. (virObjectEventCallbackListAddID) (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Use it. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 21:50:15 +00:00
*
* Internal function to count how many callbacks remain registered for
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* the given @eventID and @uuid; knowing this allows the client side
* of the remote driver know when it must send an RPC to adjust the
* callbacks on the server. When @serverFilter is false, this function
* returns a count that includes both global and per-object callbacks,
* since the remote side will use a single global event to feed both.
* When true, the count is limited to the callbacks with the same
* @uuid, and where a remoteID has already been set on the callback
* with virObjectEventStateSetRemote(). Note that this function
* intentionally ignores the legacy field, since RPC calls use only a
* single callback on the server to manage both legacy and modern
* global domain lifecycle events.
event: properly filter count of remaining events On the surface, this sequence of API calls should succeed: id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); id2 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_RTC_CHANGE,...); virConnectDomainEventDeregisterAny(id1); id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); And for test:///default, it does. But for qemu:///system, it fails: libvirt: XML-RPC error : internal error: domain event 0 already registered Looking closer, the bug is caused by miscommunication between the object event engine and the client side of the remote driver. In our implementation, we set up a single server-side event per eventID, then the client side replicates that one event to all callbacks that have been registered client side. To know when to turn the server side eventID on or off, the client side must track how many events for the same eventID have been registered. But while our code was filtering by eventID on event registration, it did not filter on event deregistration. So the above API calls resulted in the deregister returning 1 instead of 0, so no RPC deregister was issued, and the final register detects on the server side that the server is already handling eventID 0. Unfortunately, since the problem is only observable on remote connections, it's not possible to enhance objecteventtest to expose the semantics using only public API entry points. * src/conf/object_event.c (virObjectEventCallbackListCount): New function. (virObjectEventCallbackListAddID) (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Use it. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 21:50:15 +00:00
*/
static int
virObjectEventCallbackListCount(virConnectPtr conn,
virObjectEventCallbackListPtr cbList,
virClassPtr klass,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
int eventID,
unsigned char uuid[VIR_UUID_BUFLEN],
bool serverFilter)
event: properly filter count of remaining events On the surface, this sequence of API calls should succeed: id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); id2 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_RTC_CHANGE,...); virConnectDomainEventDeregisterAny(id1); id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); And for test:///default, it does. But for qemu:///system, it fails: libvirt: XML-RPC error : internal error: domain event 0 already registered Looking closer, the bug is caused by miscommunication between the object event engine and the client side of the remote driver. In our implementation, we set up a single server-side event per eventID, then the client side replicates that one event to all callbacks that have been registered client side. To know when to turn the server side eventID on or off, the client side must track how many events for the same eventID have been registered. But while our code was filtering by eventID on event registration, it did not filter on event deregistration. So the above API calls resulted in the deregister returning 1 instead of 0, so no RPC deregister was issued, and the final register detects on the server side that the server is already handling eventID 0. Unfortunately, since the problem is only observable on remote connections, it's not possible to enhance objecteventtest to expose the semantics using only public API entry points. * src/conf/object_event.c (virObjectEventCallbackListCount): New function. (virObjectEventCallbackListAddID) (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Use it. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 21:50:15 +00:00
{
size_t i;
int ret = 0;
for (i = 0; i < cbList->count; i++) {
virObjectEventCallbackPtr cb = cbList->callbacks[i];
if (cb->klass == klass &&
cb->eventID == eventID &&
cb->conn == conn &&
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
!cb->deleted &&
(!serverFilter ||
(cb->remoteID >= 0 &&
((uuid && cb->uuid_filter &&
memcmp(cb->uuid, uuid, VIR_UUID_BUFLEN) == 0) ||
(!uuid && !cb->uuid_filter)))))
event: properly filter count of remaining events On the surface, this sequence of API calls should succeed: id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); id2 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_RTC_CHANGE,...); virConnectDomainEventDeregisterAny(id1); id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); And for test:///default, it does. But for qemu:///system, it fails: libvirt: XML-RPC error : internal error: domain event 0 already registered Looking closer, the bug is caused by miscommunication between the object event engine and the client side of the remote driver. In our implementation, we set up a single server-side event per eventID, then the client side replicates that one event to all callbacks that have been registered client side. To know when to turn the server side eventID on or off, the client side must track how many events for the same eventID have been registered. But while our code was filtering by eventID on event registration, it did not filter on event deregistration. So the above API calls resulted in the deregister returning 1 instead of 0, so no RPC deregister was issued, and the final register detects on the server side that the server is already handling eventID 0. Unfortunately, since the problem is only observable on remote connections, it's not possible to enhance objecteventtest to expose the semantics using only public API entry points. * src/conf/object_event.c (virObjectEventCallbackListCount): New function. (virObjectEventCallbackListAddID) (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Use it. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 21:50:15 +00:00
ret++;
}
return ret;
}
/**
* virObjectEventCallbackListRemoveID:
* @conn: pointer to the connection
* @cbList: the list
* @callback: the callback to remove
*
* Internal function to remove a callback from a virObjectEventCallbackListPtr
*/
static int
virObjectEventCallbackListRemoveID(virConnectPtr conn,
virObjectEventCallbackListPtr cbList,
int callbackID)
{
size_t i;
for (i = 0; i < cbList->count; i++) {
virObjectEventCallbackPtr cb = cbList->callbacks[i];
if (cb->callbackID == callbackID && cb->conn == conn) {
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
int ret;
ret = virObjectEventCallbackListCount(conn, cbList, cb->klass,
cb->eventID,
cb->uuid_filter ? cb->uuid : NULL,
cb->remoteID >= 0) - 1;
event: properly filter count of remaining events On the surface, this sequence of API calls should succeed: id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); id2 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_RTC_CHANGE,...); virConnectDomainEventDeregisterAny(id1); id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); And for test:///default, it does. But for qemu:///system, it fails: libvirt: XML-RPC error : internal error: domain event 0 already registered Looking closer, the bug is caused by miscommunication between the object event engine and the client side of the remote driver. In our implementation, we set up a single server-side event per eventID, then the client side replicates that one event to all callbacks that have been registered client side. To know when to turn the server side eventID on or off, the client side must track how many events for the same eventID have been registered. But while our code was filtering by eventID on event registration, it did not filter on event deregistration. So the above API calls resulted in the deregister returning 1 instead of 0, so no RPC deregister was issued, and the final register detects on the server side that the server is already handling eventID 0. Unfortunately, since the problem is only observable on remote connections, it's not possible to enhance objecteventtest to expose the semantics using only public API entry points. * src/conf/object_event.c (virObjectEventCallbackListCount): New function. (virObjectEventCallbackListAddID) (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Use it. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 21:50:15 +00:00
if (cb->freecb)
(*cb->freecb)(cb->opaque);
virObjectUnref(cb->conn);
VIR_FREE(cb);
VIR_DELETE_ELEMENT(cbList->callbacks, i, cbList->count);
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
return ret;
}
}
virReportError(VIR_ERR_INVALID_ARG,
_("could not find event callback %d for deletion"),
callbackID);
return -1;
}
static int
virObjectEventCallbackListMarkDeleteID(virConnectPtr conn,
virObjectEventCallbackListPtr cbList,
int callbackID)
{
size_t i;
for (i = 0; i < cbList->count; i++) {
event: properly filter count of remaining events On the surface, this sequence of API calls should succeed: id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); id2 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_RTC_CHANGE,...); virConnectDomainEventDeregisterAny(id1); id1 = virConnectDomainEventRegisterAny(..., VIR_DOMAIN_EVENT_ID_LIFECYCLE,...); And for test:///default, it does. But for qemu:///system, it fails: libvirt: XML-RPC error : internal error: domain event 0 already registered Looking closer, the bug is caused by miscommunication between the object event engine and the client side of the remote driver. In our implementation, we set up a single server-side event per eventID, then the client side replicates that one event to all callbacks that have been registered client side. To know when to turn the server side eventID on or off, the client side must track how many events for the same eventID have been registered. But while our code was filtering by eventID on event registration, it did not filter on event deregistration. So the above API calls resulted in the deregister returning 1 instead of 0, so no RPC deregister was issued, and the final register detects on the server side that the server is already handling eventID 0. Unfortunately, since the problem is only observable on remote connections, it's not possible to enhance objecteventtest to expose the semantics using only public API entry points. * src/conf/object_event.c (virObjectEventCallbackListCount): New function. (virObjectEventCallbackListAddID) (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Use it. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 21:50:15 +00:00
virObjectEventCallbackPtr cb = cbList->callbacks[i];
if (cb->callbackID == callbackID && cb->conn == conn) {
cb->deleted = true;
return virObjectEventCallbackListCount(conn, cbList, cb->klass,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
cb->eventID,
cb->uuid_filter ? cb->uuid : NULL,
cb->remoteID >= 0);
}
}
virReportError(VIR_ERR_INVALID_ARG,
_("could not find event callback %d for deletion"),
callbackID);
return -1;
}
static int
virObjectEventCallbackListPurgeMarked(virObjectEventCallbackListPtr cbList)
{
int old_count = cbList->count;
int n;
for (n = 0; n < cbList->count; n++) {
if (cbList->callbacks[n]->deleted) {
virFreeCallback freecb = cbList->callbacks[n]->freecb;
if (freecb)
(*freecb)(cbList->callbacks[n]->opaque);
virObjectUnref(cbList->callbacks[n]->conn);
VIR_FREE(cbList->callbacks[n]);
if (n < (cbList->count - 1))
memmove(cbList->callbacks + n,
cbList->callbacks + n + 1,
sizeof(*(cbList->callbacks)) *
(cbList->count - (n + 1)));
cbList->count--;
n--;
}
}
if (cbList->count < old_count &&
VIR_REALLOC_N(cbList->callbacks, cbList->count) < 0) {
; /* Failure to reduce memory allocation isn't fatal */
}
return 0;
}
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
/**
* virObjectEventCallbackLookup:
* @conn: pointer to the connection
* @cbList: the list
* @uuid: the uuid of the object to filter on
* @klass: the base event class
* @eventID: the event ID
* @callback: the callback to locate
event: don't allow mix of old- and new-style registration Consider these two calls, in either order: id1 = virConnectDomainEventRegisterAny(conn, NULL, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); Right now, the second call fails, because under the hood, the old-style function registration is tightly coupled to the new style lifecycle eventID, and the two calls both try to register the same global eventID callback representation. We've alreay documented that users should avoid old-style registration and deregistration, so anyone heeding the advice won't run into this situation. But it would be even nicer if we pretend the two interfaces are completely separate, and disallow any cross-linking. That is, a call to old-style deregister should never remove a new-style callback even if it is the same function pointer, and a call to new-style callback using only callbackIDs obtained legitimately should never remove an old-style callback (of course, since our callback IDs are sequential, and there is still coupling under the hood, you can easily guess the callbackID of an old style registration and use new-style deregistration to nuke it - but that starts to be blatantly bad coding on your part rather than a surprising result on what looks like reasonable stand-alone API). With this patch, you can now register a global lifecycle event handler twice, by using both old and new APIs; if such an event occurs, your callback will be entered twice. But that is not a problem in practice, since it is already possible to use the new API to register both a global and per-domain event handler using the same function, which will likewise fire your callback twice for that domain. Duplicates are still prevented when using the same API with same parameters twice (old-style twice, new-style global twice, or new-style per-domain with same domain twice), and things are still bounded (it is not possible to register a single function pointer more than N+2 times per event id, where N is the number of domains available on the connection). Besides, it has always been possible to register as many separate function pointers on the same event id as desired, through either old or new style API, where the bound there is the physical limitation of writing a program with enough distinct function pointers. Adding another event registration in the testsuite is sufficient to cover this, where the test fails without the rest of the patch. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventCallbackLookup): Add argument. (virObjectEventCallbackListAddID, virObjectEventStateCallbackID): Adjust callers. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 12:55:55 +00:00
* @legacy: true if callback is tracked by function instead of callbackID
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* @remoteID: optionally return a known remoteID
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
*
* Internal function to determine if @callback already has a
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* callbackID in @cbList for the given @conn and other filters. If
* @remoteID is non-NULL, and another callback exists that can be
* serviced by the same remote event, then set it to that remote ID.
*
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
* Return the id if found, or -1 with no error issued if not present.
*/
static int ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2)
virObjectEventCallbackLookup(virConnectPtr conn,
virObjectEventCallbackListPtr cbList,
unsigned char uuid[VIR_UUID_BUFLEN],
virClassPtr klass,
int eventID,
event: don't allow mix of old- and new-style registration Consider these two calls, in either order: id1 = virConnectDomainEventRegisterAny(conn, NULL, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); Right now, the second call fails, because under the hood, the old-style function registration is tightly coupled to the new style lifecycle eventID, and the two calls both try to register the same global eventID callback representation. We've alreay documented that users should avoid old-style registration and deregistration, so anyone heeding the advice won't run into this situation. But it would be even nicer if we pretend the two interfaces are completely separate, and disallow any cross-linking. That is, a call to old-style deregister should never remove a new-style callback even if it is the same function pointer, and a call to new-style callback using only callbackIDs obtained legitimately should never remove an old-style callback (of course, since our callback IDs are sequential, and there is still coupling under the hood, you can easily guess the callbackID of an old style registration and use new-style deregistration to nuke it - but that starts to be blatantly bad coding on your part rather than a surprising result on what looks like reasonable stand-alone API). With this patch, you can now register a global lifecycle event handler twice, by using both old and new APIs; if such an event occurs, your callback will be entered twice. But that is not a problem in practice, since it is already possible to use the new API to register both a global and per-domain event handler using the same function, which will likewise fire your callback twice for that domain. Duplicates are still prevented when using the same API with same parameters twice (old-style twice, new-style global twice, or new-style per-domain with same domain twice), and things are still bounded (it is not possible to register a single function pointer more than N+2 times per event id, where N is the number of domains available on the connection). Besides, it has always been possible to register as many separate function pointers on the same event id as desired, through either old or new style API, where the bound there is the physical limitation of writing a program with enough distinct function pointers. Adding another event registration in the testsuite is sufficient to cover this, where the test fails without the rest of the patch. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventCallbackLookup): Add argument. (virObjectEventCallbackListAddID, virObjectEventStateCallbackID): Adjust callers. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 12:55:55 +00:00
virConnectObjectEventGenericCallback callback,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
bool legacy,
int *remoteID)
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
{
size_t i;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
if (remoteID)
*remoteID = -1;
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
for (i = 0; i < cbList->count; i++) {
virObjectEventCallbackPtr cb = cbList->callbacks[i];
if (cb->deleted)
continue;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
if (cb->klass == klass &&
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
cb->eventID == eventID &&
cb->conn == conn &&
event: don't allow mix of old- and new-style registration Consider these two calls, in either order: id1 = virConnectDomainEventRegisterAny(conn, NULL, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); Right now, the second call fails, because under the hood, the old-style function registration is tightly coupled to the new style lifecycle eventID, and the two calls both try to register the same global eventID callback representation. We've alreay documented that users should avoid old-style registration and deregistration, so anyone heeding the advice won't run into this situation. But it would be even nicer if we pretend the two interfaces are completely separate, and disallow any cross-linking. That is, a call to old-style deregister should never remove a new-style callback even if it is the same function pointer, and a call to new-style callback using only callbackIDs obtained legitimately should never remove an old-style callback (of course, since our callback IDs are sequential, and there is still coupling under the hood, you can easily guess the callbackID of an old style registration and use new-style deregistration to nuke it - but that starts to be blatantly bad coding on your part rather than a surprising result on what looks like reasonable stand-alone API). With this patch, you can now register a global lifecycle event handler twice, by using both old and new APIs; if such an event occurs, your callback will be entered twice. But that is not a problem in practice, since it is already possible to use the new API to register both a global and per-domain event handler using the same function, which will likewise fire your callback twice for that domain. Duplicates are still prevented when using the same API with same parameters twice (old-style twice, new-style global twice, or new-style per-domain with same domain twice), and things are still bounded (it is not possible to register a single function pointer more than N+2 times per event id, where N is the number of domains available on the connection). Besides, it has always been possible to register as many separate function pointers on the same event id as desired, through either old or new style API, where the bound there is the physical limitation of writing a program with enough distinct function pointers. Adding another event registration in the testsuite is sufficient to cover this, where the test fails without the rest of the patch. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventCallbackLookup): Add argument. (virObjectEventCallbackListAddID, virObjectEventStateCallbackID): Adjust callers. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 12:55:55 +00:00
cb->legacy == legacy &&
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
((uuid && cb->uuid_filter &&
memcmp(cb->uuid, uuid, VIR_UUID_BUFLEN) == 0) ||
(!uuid && !cb->uuid_filter))) {
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
if (remoteID)
*remoteID = cb->remoteID;
if (cb->cb == callback)
return cb->callbackID;
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
}
}
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
return -1;
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
}
/**
* virObjectEventCallbackListAddID:
* @conn: pointer to the connection
* @cbList: the list
* @uuid: the uuid of the object to filter on
* @klass: the base event class
* @eventID: the event ID
* @callback: the callback to add
* @opaque: opaque data to pass to @callback
* @freecb: callback to free @opaque
* @callbackID: filled with callback ID
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* @serverFilter: true if server supports object filtering
*
* Internal function to add a callback from a virObjectEventCallbackListPtr
*/
static int
virObjectEventCallbackListAddID(virConnectPtr conn,
virObjectEventCallbackListPtr cbList,
unsigned char uuid[VIR_UUID_BUFLEN],
virClassPtr klass,
int eventID,
virConnectObjectEventGenericCallback callback,
void *opaque,
virFreeCallback freecb,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
int *callbackID,
bool serverFilter)
{
virObjectEventCallbackPtr event;
int ret = -1;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
int remoteID = -1;
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
VIR_DEBUG("conn=%p cblist=%p uuid=%p "
"klass=%p eventID=%d callback=%p opaque=%p",
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
conn, cbList, uuid, klass, eventID, callback, opaque);
/* Check incoming */
if (!cbList) {
return -1;
}
/* check if we already have this callback on our list */
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
if (virObjectEventCallbackLookup(conn, cbList, uuid,
event: don't allow mix of old- and new-style registration Consider these two calls, in either order: id1 = virConnectDomainEventRegisterAny(conn, NULL, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); Right now, the second call fails, because under the hood, the old-style function registration is tightly coupled to the new style lifecycle eventID, and the two calls both try to register the same global eventID callback representation. We've alreay documented that users should avoid old-style registration and deregistration, so anyone heeding the advice won't run into this situation. But it would be even nicer if we pretend the two interfaces are completely separate, and disallow any cross-linking. That is, a call to old-style deregister should never remove a new-style callback even if it is the same function pointer, and a call to new-style callback using only callbackIDs obtained legitimately should never remove an old-style callback (of course, since our callback IDs are sequential, and there is still coupling under the hood, you can easily guess the callbackID of an old style registration and use new-style deregistration to nuke it - but that starts to be blatantly bad coding on your part rather than a surprising result on what looks like reasonable stand-alone API). With this patch, you can now register a global lifecycle event handler twice, by using both old and new APIs; if such an event occurs, your callback will be entered twice. But that is not a problem in practice, since it is already possible to use the new API to register both a global and per-domain event handler using the same function, which will likewise fire your callback twice for that domain. Duplicates are still prevented when using the same API with same parameters twice (old-style twice, new-style global twice, or new-style per-domain with same domain twice), and things are still bounded (it is not possible to register a single function pointer more than N+2 times per event id, where N is the number of domains available on the connection). Besides, it has always been possible to register as many separate function pointers on the same event id as desired, through either old or new style API, where the bound there is the physical limitation of writing a program with enough distinct function pointers. Adding another event registration in the testsuite is sufficient to cover this, where the test fails without the rest of the patch. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventCallbackLookup): Add argument. (virObjectEventCallbackListAddID, virObjectEventStateCallbackID): Adjust callers. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 12:55:55 +00:00
klass, eventID, callback,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
!callbackID,
serverFilter ? &remoteID : NULL) != -1) {
virReportError(VIR_ERR_INVALID_ARG, "%s",
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
_("event callback already tracked"));
return -1;
}
/* Allocate new event */
if (VIR_ALLOC(event) < 0)
goto cleanup;
event->conn = virObjectRef(conn);
event->callbackID = cbList->nextID++;
event->cb = callback;
event->klass = klass;
event->eventID = eventID;
event->opaque = opaque;
event->freecb = freecb;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
event->remoteID = remoteID;
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
/* Only need 'uuid' for matching; 'id' can change as domain
* switches between running and shutoff, and 'name' can change in
* Xen migration. */
if (uuid) {
event->uuid_filter = true;
memcpy(event->uuid, uuid, VIR_UUID_BUFLEN);
}
if (callbackID)
*callbackID = event->callbackID;
event: don't allow mix of old- and new-style registration Consider these two calls, in either order: id1 = virConnectDomainEventRegisterAny(conn, NULL, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); Right now, the second call fails, because under the hood, the old-style function registration is tightly coupled to the new style lifecycle eventID, and the two calls both try to register the same global eventID callback representation. We've alreay documented that users should avoid old-style registration and deregistration, so anyone heeding the advice won't run into this situation. But it would be even nicer if we pretend the two interfaces are completely separate, and disallow any cross-linking. That is, a call to old-style deregister should never remove a new-style callback even if it is the same function pointer, and a call to new-style callback using only callbackIDs obtained legitimately should never remove an old-style callback (of course, since our callback IDs are sequential, and there is still coupling under the hood, you can easily guess the callbackID of an old style registration and use new-style deregistration to nuke it - but that starts to be blatantly bad coding on your part rather than a surprising result on what looks like reasonable stand-alone API). With this patch, you can now register a global lifecycle event handler twice, by using both old and new APIs; if such an event occurs, your callback will be entered twice. But that is not a problem in practice, since it is already possible to use the new API to register both a global and per-domain event handler using the same function, which will likewise fire your callback twice for that domain. Duplicates are still prevented when using the same API with same parameters twice (old-style twice, new-style global twice, or new-style per-domain with same domain twice), and things are still bounded (it is not possible to register a single function pointer more than N+2 times per event id, where N is the number of domains available on the connection). Besides, it has always been possible to register as many separate function pointers on the same event id as desired, through either old or new style API, where the bound there is the physical limitation of writing a program with enough distinct function pointers. Adding another event registration in the testsuite is sufficient to cover this, where the test fails without the rest of the patch. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventCallbackLookup): Add argument. (virObjectEventCallbackListAddID, virObjectEventStateCallbackID): Adjust callers. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 12:55:55 +00:00
else
event->legacy = true;
if (VIR_APPEND_ELEMENT(cbList->callbacks, cbList->count, event) < 0)
goto cleanup;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
ret = virObjectEventCallbackListCount(conn, cbList, klass, eventID,
uuid, serverFilter);
if (serverFilter && remoteID < 0)
ret++;
cleanup:
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
if (event)
virObjectUnref(event->conn);
VIR_FREE(event);
return ret;
}
/**
* virObjectEventQueueClear:
* @queue: pointer to the queue
*
* Removes all elements from the queue
*/
static void
virObjectEventQueueClear(virObjectEventQueuePtr queue)
{
size_t i;
if (!queue)
return;
for (i = 0; i < queue->count; i++) {
virObjectUnref(queue->events[i]);
}
VIR_FREE(queue->events);
queue->count = 0;
}
/**
* virObjectEventQueueFree:
* @queue: pointer to the queue
*
* Free the memory in the queue. We process this like a list here
*/
static void
virObjectEventQueueFree(virObjectEventQueuePtr queue)
{
if (!queue)
return;
virObjectEventQueueClear(queue);
VIR_FREE(queue);
}
static virObjectEventQueuePtr
virObjectEventQueueNew(void)
{
virObjectEventQueuePtr ret;
ignore_value(VIR_ALLOC(ret));
return ret;
}
/**
* virObjectEventStateLock:
* @state: the event state object
*
* Lock event state before calling functions from object_event_private.h.
*/
static void
virObjectEventStateLock(virObjectEventStatePtr state)
{
virMutexLock(&state->lock);
}
/**
* virObjectEventStateUnlock:
* @state: the event state object
*
* Unlock event state after calling functions from object_event_private.h.
*/
static void
virObjectEventStateUnlock(virObjectEventStatePtr state)
{
virMutexUnlock(&state->lock);
}
/**
* virObjectEventStateFree:
* @list: virObjectEventStatePtr to free
*
* Free a virObjectEventStatePtr and its members, and unregister the timer.
*/
void
virObjectEventStateFree(virObjectEventStatePtr state)
{
if (!state)
return;
virObjectEventCallbackListFree(state->callbacks);
virObjectEventQueueFree(state->queue);
if (state->timer != -1)
virEventRemoveTimeout(state->timer);
virMutexDestroy(&state->lock);
VIR_FREE(state);
}
static void virObjectEventStateFlush(virObjectEventStatePtr state);
/**
* virObjectEventTimer:
* @timer: id of the event loop timer
* @opaque: the event state object
*
* Register this function with the event state as its opaque data as
* the callback of a periodic timer on the event loop, in order to
* flush the callback queue.
*/
static void
virObjectEventTimer(int timer ATTRIBUTE_UNUSED, void *opaque)
{
virObjectEventStatePtr state = opaque;
virObjectEventStateFlush(state);
}
/**
* virObjectEventStateNew:
*
* Allocate a new event state object.
*/
virObjectEventStatePtr
virObjectEventStateNew(void)
{
virObjectEventStatePtr state = NULL;
if (VIR_ALLOC(state) < 0)
goto error;
if (virMutexInit(&state->lock) < 0) {
virReportSystemError(errno, "%s",
_("unable to initialize state mutex"));
VIR_FREE(state);
goto error;
}
if (VIR_ALLOC(state->callbacks) < 0)
goto error;
if (!(state->queue = virObjectEventQueueNew()))
goto error;
state->timer = -1;
return state;
error:
virObjectEventStateFree(state);
return NULL;
}
/**
* virObjectEventNew:
* @klass: subclass of event to be created
* @dispatcher: callback for dispatching the particular subclass of event
* @eventID: id of the event
* @id: id of the object the event describes, or 0
* @name: name of the object the event describes
* @uuid: uuid of the object the event describes
*
* Create a new event, with the information common to all events.
*/
2013-12-11 10:38:03 +00:00
void *
virObjectEventNew(virClassPtr klass,
virObjectEventDispatchFunc dispatcher,
2013-12-11 10:38:03 +00:00
int eventID,
int id,
const char *name,
const unsigned char *uuid)
{
virObjectEventPtr event;
if (virObjectEventInitialize() < 0)
return NULL;
if (!virClassIsDerivedFrom(klass, virObjectEventClass)) {
virReportInvalidArg(klass,
_("Class %s must derive from virObjectEvent"),
virClassName(klass));
return NULL;
}
if (!(event = virObjectNew(klass)))
return NULL;
event->dispatch = dispatcher;
event->eventID = eventID;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
event->remoteID = -1;
if (VIR_STRDUP(event->meta.name, name) < 0) {
VIR_FREE(event);
return NULL;
}
event->meta.id = id;
memcpy(event->meta.uuid, uuid, VIR_UUID_BUFLEN);
VIR_DEBUG("obj=%p", event);
return event;
}
/**
* virObjectEventQueuePush:
* @evtQueue: the object event queue
* @event: the event to add
*
* Internal function to push to the back of a virObjectEventQueue
*
* Returns: 0 on success, -1 on failure
*/
static int
virObjectEventQueuePush(virObjectEventQueuePtr evtQueue,
virObjectEventPtr event)
{
if (!evtQueue)
return -1;
if (VIR_APPEND_ELEMENT(evtQueue->events, evtQueue->count, event) < 0)
return -1;
return 0;
}
static bool
2013-12-11 10:38:03 +00:00
virObjectEventDispatchMatchCallback(virObjectEventPtr event,
virObjectEventCallbackPtr cb)
{
if (!cb)
return false;
if (cb->deleted)
return false;
if (!virObjectIsClass(event, cb->klass))
return false;
if (cb->eventID != event->eventID)
return false;
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
if (cb->remoteID != event->remoteID)
return false;
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
if (cb->uuid_filter) {
/* Deliberately ignoring 'id' for matching, since that
* will cause problems when a domain switches between
* running & shutoff states & ignoring 'name' since
* Xen sometimes renames guests during migration, thus
* leaving 'uuid' as the only truly reliable ID we can use. */
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
return memcmp(event->meta.uuid, cb->uuid, VIR_UUID_BUFLEN) == 0;
}
return true;
}
static void
virObjectEventStateDispatchCallbacks(virObjectEventStatePtr state,
virObjectEventPtr event,
virObjectEventCallbackListPtr callbacks)
{
size_t i;
/* Cache this now, since we may be dropping the lock,
and have more callbacks added. We're guaranteed not
to have any removed */
size_t cbCount = callbacks->count;
for (i = 0; i < cbCount; i++) {
virObjectEventCallbackPtr cb = callbacks->callbacks[i];
if (!virObjectEventDispatchMatchCallback(event, cb))
continue;
/* Drop the lock whle dispatching, for sake of re-entrancy */
virObjectEventStateUnlock(state);
event->dispatch(cb->conn, event, cb->cb, cb->opaque);
virObjectEventStateLock(state);
}
}
static void
virObjectEventStateQueueDispatch(virObjectEventStatePtr state,
virObjectEventQueuePtr queue,
virObjectEventCallbackListPtr callbacks)
{
size_t i;
for (i = 0; i < queue->count; i++) {
virObjectEventStateDispatchCallbacks(state, queue->events[i],
callbacks);
virObjectUnref(queue->events[i]);
}
VIR_FREE(queue->events);
queue->count = 0;
}
/**
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* virObjectEventStateQueueRemote:
* @state: the event state object
* @event: event to add to the queue
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* @remoteID: limit dispatch to callbacks with the same remote id
*
* Adds @event to the queue of events to be dispatched at the next
* safe moment. The caller should no longer use @event after this
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* call. If @remoteID is non-negative, the event will only be sent to
* callbacks where virObjectEventStateSetRemote() registered a remote
* id.
*/
void
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
virObjectEventStateQueueRemote(virObjectEventStatePtr state,
virObjectEventPtr event,
int remoteID)
{
if (state->timer < 0) {
virObjectUnref(event);
return;
}
virObjectEventStateLock(state);
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
event->remoteID = remoteID;
if (virObjectEventQueuePush(state->queue, event) < 0) {
VIR_DEBUG("Error adding event to queue");
virObjectUnref(event);
}
if (state->queue->count == 1)
virEventUpdateTimeout(state->timer, 0);
virObjectEventStateUnlock(state);
}
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
/**
* virObjectEventStateQueue:
* @state: the event state object
* @event: event to add to the queue
*
* Adds @event to the queue of events to be dispatched at the next
* safe moment. The caller should no longer use @event after this
* call.
*/
void
virObjectEventStateQueue(virObjectEventStatePtr state,
virObjectEventPtr event)
{
virObjectEventStateQueueRemote(state, event, -1);
}
static void
virObjectEventStateFlush(virObjectEventStatePtr state)
{
virObjectEventQueue tempQueue;
virObjectEventStateLock(state);
state->isDispatching = true;
/* Copy the queue, so we're reentrant safe when dispatchFunc drops the
* driver lock */
tempQueue.count = state->queue->count;
tempQueue.events = state->queue->events;
state->queue->count = 0;
state->queue->events = NULL;
virEventUpdateTimeout(state->timer, -1);
virObjectEventStateQueueDispatch(state,
&tempQueue,
state->callbacks);
/* Purge any deleted callbacks */
virObjectEventCallbackListPurgeMarked(state->callbacks);
state->isDispatching = false;
virObjectEventStateUnlock(state);
}
/**
* virObjectEventStateRegisterID:
* @conn: connection to associate with callback
* @state: domain event state
* @uuid: uuid of the object for event filtering
* @klass: the base event class
* @eventID: ID of the event type to register for
* @cb: function to invoke when event occurs
* @opaque: data blob to pass to @callback
* @freecb: callback to free @opaque
* @callbackID: filled with callback ID
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* @serverFilter: true if server supports object filtering
*
* Register the function @cb with connection @conn, from @state, for
* events of type @eventID, and return the registration handle in
* @callbackID.
*
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* The return value is only important when registering client-side
* mirroring of remote events (since the public API is documented to
* return the callbackID rather than a count). A return of 1 means
* that this is the first use of this type of event, so a remote event
* must be enabled; a return larger than 1 means that an existing
* remote event can already feed this callback. If @serverFilter is
* false, the return count assumes that a single global remote feeds
* both generic and per-object callbacks, and that the event queue
* will be fed with virObjectEventStateQueue(). If it is true, then
* the return count assumes that the remote side is capable of per-
* object filtering; the user must call virObjectEventStateSetRemote()
* to record the remote id, and queue events with
* virObjectEventStateQueueRemote().
*
* Returns: the number of callbacks now registered, or -1 on error.
*/
int
virObjectEventStateRegisterID(virConnectPtr conn,
virObjectEventStatePtr state,
unsigned char *uuid,
virClassPtr klass,
int eventID,
virConnectObjectEventGenericCallback cb,
void *opaque,
virFreeCallback freecb,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
int *callbackID,
bool serverFilter)
{
int ret = -1;
virObjectEventStateLock(state);
if ((state->callbacks->count == 0) &&
(state->timer == -1) &&
(state->timer = virEventAddTimeout(-1,
virObjectEventTimer,
state,
NULL)) < 0) {
virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
_("could not initialize domain event timer"));
goto cleanup;
}
ret = virObjectEventCallbackListAddID(conn, state->callbacks,
event: don't turn offline domain into global event If a user registers for a domain event filtered to a particular domain, but the persistent domain is offline at the time, then the code silently failed to set up the filter. As a result, the event fires for all domains, rather than being filtered. Network events were immune, since they always passed an id 0 argument. The key to this patch is realizing that virObjectEventDispatchMatchCallback() only cared about uuid; so refusing to create a meta for a negative id is pointless, and in fact, malloc'ing meta at all was overkill; instead, just directly store a uuid and a flag of whether to filter. Note that virObjectEventPtr still needs all fields of meta, because this is how we reconstruct a virDomainPtr inside the dispatch handler before calling the end user's callback pointer with the correct object, even though only the uuid portion of meta is used in deciding whether a callback matches the given event. So while uuid is optional for callbacks, it is mandatory for events. The change to testDomainCreateXMLMixed is merely on the setup scenario (as you can't register for a domain unless it is either running or persistent). I actually first wrote that test for this patch, then rebased it to also cover a prior patch (commit 4221d64), but had to adjust it for that patch to use Create instead of Define for setting up the domain long enough to register the event in order to work around this bug. But while the setup is changed, the main body of the test is still about whether creation events fire as expected. * src/conf/object_event_private.h (_virObjectEventCallback): Replace meta with uuid and flag. (virObjectEventCallbackListAddID): Update signature. * src/conf/object_event.h (virObjectEventStateRegisterID): Likewise. * src/conf/object_event_private.h (virObjectEventNew): Document use of name and uuid in events. * src/conf/object_event.c (virObjectEventCallbackListAddID): Drop arguments that don't affect filtering. (virObjectEventCallbackListRemoveID) (virObjectEventDispatchMatchCallback) (virObjectEventStateRegisterID): Update clients. * src/conf/domain_event.c (virDomainEventCallbackListAdd) (virDomainEventStateRegisterID): Likewise. * src/conf/network_event.c (virNetworkEventStateRegisterID): Likewise. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 00:02:17 +00:00
uuid, klass, eventID,
cb, opaque, freecb,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
callbackID, serverFilter);
if (ret == -1 &&
state->callbacks->count == 0 &&
state->timer != -1) {
virEventRemoveTimeout(state->timer);
state->timer = -1;
}
cleanup:
virObjectEventStateUnlock(state);
return ret;
}
/**
* virObjectEventStateDeregisterID:
* @conn: connection to associate with callback
* @state: object event state
* @callbackID: ID of the function to remove from event
*
* Unregister the function @callbackID with connection @conn,
* from @state, for events.
*
* Returns: the number of callbacks still registered, or -1 on error
*/
int
virObjectEventStateDeregisterID(virConnectPtr conn,
virObjectEventStatePtr state,
int callbackID)
{
int ret;
virObjectEventStateLock(state);
if (state->isDispatching)
ret = virObjectEventCallbackListMarkDeleteID(conn,
state->callbacks,
callbackID);
else
ret = virObjectEventCallbackListRemoveID(conn,
state->callbacks, callbackID);
if (state->callbacks->count == 0 &&
state->timer != -1) {
virEventRemoveTimeout(state->timer);
state->timer = -1;
virObjectEventQueueClear(state->queue);
}
virObjectEventStateUnlock(state);
return ret;
}
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
/**
* virObjectEventStateCallbackID:
* @conn: connection associated with callback
* @state: object event state
* @klass: the base event class
* @eventID: the event ID
* @callback: function registered as a callback
*
* Returns the callbackID of @callback, or -1 with an error issued if the
event: don't allow mix of old- and new-style registration Consider these two calls, in either order: id1 = virConnectDomainEventRegisterAny(conn, NULL, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); Right now, the second call fails, because under the hood, the old-style function registration is tightly coupled to the new style lifecycle eventID, and the two calls both try to register the same global eventID callback representation. We've alreay documented that users should avoid old-style registration and deregistration, so anyone heeding the advice won't run into this situation. But it would be even nicer if we pretend the two interfaces are completely separate, and disallow any cross-linking. That is, a call to old-style deregister should never remove a new-style callback even if it is the same function pointer, and a call to new-style callback using only callbackIDs obtained legitimately should never remove an old-style callback (of course, since our callback IDs are sequential, and there is still coupling under the hood, you can easily guess the callbackID of an old style registration and use new-style deregistration to nuke it - but that starts to be blatantly bad coding on your part rather than a surprising result on what looks like reasonable stand-alone API). With this patch, you can now register a global lifecycle event handler twice, by using both old and new APIs; if such an event occurs, your callback will be entered twice. But that is not a problem in practice, since it is already possible to use the new API to register both a global and per-domain event handler using the same function, which will likewise fire your callback twice for that domain. Duplicates are still prevented when using the same API with same parameters twice (old-style twice, new-style global twice, or new-style per-domain with same domain twice), and things are still bounded (it is not possible to register a single function pointer more than N+2 times per event id, where N is the number of domains available on the connection). Besides, it has always been possible to register as many separate function pointers on the same event id as desired, through either old or new style API, where the bound there is the physical limitation of writing a program with enough distinct function pointers. Adding another event registration in the testsuite is sufficient to cover this, where the test fails without the rest of the patch. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventCallbackLookup): Add argument. (virObjectEventCallbackListAddID, virObjectEventStateCallbackID): Adjust callers. * tests/objecteventtest.c (testDomainCreateXMLMixed): Enhance test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-04 12:55:55 +00:00
* function is not currently registered. This only finds functions
* registered via virConnectDomainEventRegister, even if modern style
* virConnectDomainEventRegisterAny also registers the same callback.
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
*/
int
virObjectEventStateCallbackID(virConnectPtr conn,
virObjectEventStatePtr state,
virClassPtr klass,
int eventID,
virConnectObjectEventGenericCallback callback)
{
int ret = -1;
virObjectEventStateLock(state);
ret = virObjectEventCallbackLookup(conn, state->callbacks, NULL,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
klass, eventID, callback, true, NULL);
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
virObjectEventStateUnlock(state);
if (ret < 0)
virReportError(VIR_ERR_INVALID_ARG,
event: don't let old-style events clobber per-domain events Right now, the older virConnectDomainEventRegister (takes a function pointer, returns 0 on success) and the newer virConnectDomainEventRegisterID (takes an eventID, returns a callbackID) share the underlying implementation (the older API ends up consuming a callbackID for eventID 0 under the hood). We implemented that by a lot of copy and pasted code between object_event.c and domain_event.c, according to whether we are dealing with a function pointer or an eventID. However, our copy and paste is not symmetric. Consider this sequence: id1 = virConnectDomainEventRegisterAny(conn, dom, VIR_DOMAIN_EVENT_ID_LIFECYCLE, VIR_DOMAIN_EVENT_CALLBACK(callback), NULL, NULL); virConnectDomainEventRegister(conn, callback, NULL, NULL); virConnectDomainEventDeregister(conn, callback); virConnectDomainEventDeregsiterAny(conn, id1); the first three calls would succeed, but the third call ended up nuking the id1 callbackID (the per-domain new-style handler), then the fourth call failed with an error about an unknown callbackID, leaving us with the global handler (old-style) still live and receiving events. It required another old-style deregister to clean up the mess. Root cause was that virDomainEventCallbackList{Remove,MarkDelete} were only checking for function pointer match, rather than also checking for whether the registration was global. Rather than playing with the guts of object_event ourselves in domain_event, it is nicer to add a mapping function for the internal callback id, then share common code for event removal. For now, the function-to-id mapping is used only internally; I thought about whether a new public API to let a user learn the callback would be useful, but decided exposing this to the user is probably a disservice, since we already publicly document that they should avoid the old style, and since this patch already demonstrates that older libvirt versions have weird behavior when mixing old and new styles. And like all good bug fix patches, I enhanced the testsuite, validating that the changes in tests/ expose the failure without the rest of the patch. * src/conf/object_event.c (virObjectEventCallbackLookup) (virObjectEventStateCallbackID): New functions. (virObjectEventCallbackLookup): Use helper function. * src/conf/object_event_private.h (virObjectEventStateCallbackID): Declare new function. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateDeregister): Let common code handle the complexity. (virDomainEventCallbackListRemove) (virDomainEventCallbackListMarkDelete) (virDomainEventCallbackListAdd): Drop unused functions. * tests/objecteventtest.c (testDomainCreateXMLMixed): New test. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-03 23:50:14 +00:00
_("event callback function %p not registered"),
callback);
return ret;
}
/**
* virObjectEventStateEventID:
* @conn: connection associated with the callback
* @state: object event state
* @callbackID: the callback to query
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* @remoteID: optionally output remote ID of the callback
*
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
* Query what event ID type is associated with the callback
* @callbackID for connection @conn. If @remoteID is non-null, it
* will be set to the remote id previously registered with
* virObjectEventStateSetRemote().
*
* Returns 0 on success, -1 on error
*/
int
virObjectEventStateEventID(virConnectPtr conn,
virObjectEventStatePtr state,
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
int callbackID,
int *remoteID)
{
int ret = -1;
size_t i;
virObjectEventCallbackListPtr cbList = state->callbacks;
virObjectEventStateLock(state);
for (i = 0; i < cbList->count; i++) {
virObjectEventCallbackPtr cb = cbList->callbacks[i];
if (cb->deleted)
continue;
if (cb->callbackID == callbackID && cb->conn == conn) {
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
if (remoteID)
*remoteID = cb->remoteID;
ret = cb->eventID;
break;
}
}
virObjectEventStateUnlock(state);
if (ret < 0)
virReportError(VIR_ERR_INVALID_ARG,
_("event callback id %d not registered"),
callbackID);
return ret;
}
event: add notion of remoteID for filtering client network events In order to mirror a server with per-object filtering, the client needs to track which server callbackID is servicing the client callback. This patch introduces the notion of a serverID, as well as the plumbing to use it for network events, although the actual complexity of using per-object filtering in the remote driver is deferred to a later patch. * src/conf/object_event.h (virObjectEventStateEventID): Add parameter. (virObjectEventStateQueueRemote, virObjectEventStateSetRemote): New prototypes. (virObjectEventStateRegisterID): Move... * src/conf/object_event_private.h: ...here, and add parameter. (_virObjectEvent): Add field. * src/conf/network_event.h (virNetworkEventStateRegisterClient): New prototype. * src/conf/object_event.c (_virObjectEventCallback): Add field. (virObjectEventStateSetRemote): New function. (virObjectEventStateQueue): Make wrapper around... (virObjectEventStateQueueRemote): New function. (virObjectEventCallbackListCount): Tweak return count when remote id matching is used. (virObjectEventCallbackLookup, virObjectEventStateRegisterID): Tweak registration when remote id matching will be used. (virObjectEventNew): Default to no remote id. (virObjectEventCallbackListAddID): Likewise, but set remote id when one is available. (virObjectEventCallbackListRemoveID) (virObjectEventCallbackListMarkDeleteID): Adjust return value when remote id was set. (virObjectEventStateEventID): Query existing id. (virObjectEventDispatchMatchCallback): Require matching event id. (virObjectEventStateCallbackID): Adjust caller. * src/conf/network_event.c (virNetworkEventStateRegisterClient): New function. (virNetworkEventStateRegisterID): Update caller. * src/conf/domain_event.c (virDomainEventStateRegister) (virDomainEventStateRegisterID): Update callers. * src/remote/remote_driver.c (remoteConnectNetworkEventRegisterAny) (remoteConnectNetworkEventDeregisterAny) (remoteConnectDomainEventDeregisterAny): Likewise. (remoteEventQueue): Hoist earlier to avoid forward declaration, and add parameter. Adjust all callers. * src/libvirt_private.syms (conf/object_event.h): Drop function. Signed-off-by: Eric Blake <eblake@redhat.com>
2014-01-06 12:32:55 +00:00
/**
* virObjectEventStateSetRemote:
* @conn: connection associated with the callback
* @state: object event state
* @callbackID: the callback to adjust
* @remoteID: the remote ID to associate with the callback
*
* Update @callbackID for connection @conn to record that it is now
* tied to @remoteID, and will therefore only match events that are
* sent with virObjectEventStateQueueRemote() with the same remote ID.
* Silently does nothing if @callbackID is invalid.
*/
void
virObjectEventStateSetRemote(virConnectPtr conn,
virObjectEventStatePtr state,
int callbackID,
int remoteID)
{
size_t i;
virObjectEventStateLock(state);
for (i = 0; i < state->callbacks->count; i++) {
virObjectEventCallbackPtr cb = state->callbacks->callbacks[i];
if (cb->deleted)
continue;
if (cb->callbackID == callbackID && cb->conn == conn) {
cb->remoteID = remoteID;
break;
}
}
virObjectEventStateUnlock(state);
}