Skip to contentSkip to conversation

qdev: DEVICE_DELETED event

PATCH Superseded Michael S. Tsirkin 3 participants 5 messages Mar 6, 2013

v2 of 9 Superseded Latest is v9 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 qdev: DEVICE_DELETED event

libvirt has a long-standing bug: when removing the device, it can request removal but does not know when the removal completes. Add an event so we can fix this in a robust way.

---

Changes from v1:
- move to device_unparent
- address comments by Andreas and Eric

Andreas also suggested a more generic object-deleted event, I'm not sure how useful that is so let's add what we already need, for devices with an id and wait and see what's necessary for non-device objects?

 QMP/qmp-events.txt        | 15 +++++++++++++++
 hw/qdev.c                 |  7 +++++++
 include/monitor/monitor.h |  1 +
 monitor.c                 |  1 +
 qapi-schema.json          |  4 +++-
 5 files changed, 27 insertions(+), 1 deletion(-)
QMP/qmp-events.txt +15 -0
index b2698e4..f2f115a 100644
@@ -136,6 +136,21 @@ Example:
 Note: The "ready to complete" status is always reset by a BLOCK_JOB_ERROR
 event.
 
+DEVICE_DELETED
+-----------------
+
+Emitted whenever the device removal completion is acknowledged
+by the guest. At this point, it's safe to reuse the specified device ID.
+Device removal can be initiated by the guest or by HMP/QMP commands.
+
+Data:
+
+- "device": device name (json-string)
+
+{ "event": "DEVICE_DELETED",
+  "data": { "device": "virtio-net-pci-0" },
+  "timestamp": { "seconds": 1265044230, "microseconds": 450486 } }
+
 DEVICE_TRAY_MOVED
 -----------------
 
hw/qdev.c +7 -0
index 689cd54..d603f4f 100644
@@ -29,6 +29,7 @@
 #include "sysemu/sysemu.h"
 #include "qapi/error.h"
 #include "qapi/visitor.h"
+#include "qapi/qmp/qjson.h"
 
 int qdev_hotplug = 0;
 static bool qdev_hot_added = false;
@@ -761,6 +762,12 @@ static void device_unparent(Object *obj)
     DeviceClass *dc = DEVICE_GET_CLASS(dev);
     BusState *bus;
 
+    if (dev->id) {
+        QObject *data = qobject_from_jsonf("{ 'device': %s }", dev->id);
+        monitor_protocol_event(QEVENT_DEVICE_DELETED, data);
+        qobject_decref(data);
+    }
+
     while (dev->num_child_bus) {
         bus = QLIST_FIRST(&dev->child_bus);
         qbus_free(bus);
include/monitor/monitor.h +1 -0
index 87fb49c..b868760 100644
@@ -39,6 +39,7 @@ typedef enum MonitorEvent {
     QEVENT_BLOCK_JOB_CANCELLED,
     QEVENT_BLOCK_JOB_ERROR,
     QEVENT_BLOCK_JOB_READY,
+    QEVENT_DEVICE_DELETED,
     QEVENT_DEVICE_TRAY_MOVED,
     QEVENT_SUSPEND,
     QEVENT_SUSPEND_DISK,
monitor.c +1 -0
index 32a6e74..2a5e7b6 100644
@@ -457,6 +457,7 @@ static const char *monitor_event_names[] = {
     [QEVENT_BLOCK_JOB_CANCELLED] = "BLOCK_JOB_CANCELLED",
     [QEVENT_BLOCK_JOB_ERROR] = "BLOCK_JOB_ERROR",
     [QEVENT_BLOCK_JOB_READY] = "BLOCK_JOB_READY",
+    [QEVENT_DEVICE_DELETED] = "DEVICE_DELETED",
     [QEVENT_DEVICE_TRAY_MOVED] = "DEVICE_TRAY_MOVED",
     [QEVENT_SUSPEND] = "SUSPEND",
     [QEVENT_SUSPEND_DISK] = "SUSPEND_DISK",
qapi-schema.json +3 -1
index 28b070f..bb361e1 100644
@@ -2354,7 +2354,9 @@
 # Notes: When this command completes, the device may not be removed from the
 #        guest.  Hot removal is an operation that requires guest cooperation.
 #        This command merely requests that the guest begin the hot removal
-#        process.
+#        process.  Completion of the device removal process is signaled with a
+#        DEVICE_DELETED event. Guest reset will automatically complete removal
+#        for all devices.
 #
 # Since: 0.14.0
 ##
Signature · 1 line
MST
#1 lore raw

Il 06/03/2013 17:49, Michael S. Tsirkin ha scritto:

Quoted 62 lines

libvirt has a long-standing bug: when removing the device, it can request removal but does not know when the removal completes. Add an event so we can fix this in a robust way.

---

Changes from v1:
- move to device_unparent
- address comments by Andreas and Eric

Andreas also suggested a more generic object-deleted event, I'm not sure how useful that is so let's add what we already need, for devices with an id and wait and see what's necessary for non-device objects?

 QMP/qmp-events.txt        | 15 +++++++++++++++
 hw/qdev.c                 |  7 +++++++
 include/monitor/monitor.h |  1 +
 monitor.c                 |  1 +
 qapi-schema.json          |  4 +++-
 5 files changed, 27 insertions(+), 1 deletion(-)
QMP/qmp-events.txt +15 -0
index b2698e4..f2f115a 100644
@@ -136,6 +136,21 @@ Example:
 Note: The "ready to complete" status is always reset by a BLOCK_JOB_ERROR
 event.
 
+DEVICE_DELETED
+-----------------
+
+Emitted whenever the device removal completion is acknowledged
+by the guest. At this point, it's safe to reuse the specified device ID.
+Device removal can be initiated by the guest or by HMP/QMP commands.
+
+Data:
+
+- "device": device name (json-string)
+
+{ "event": "DEVICE_DELETED",
+  "data": { "device": "virtio-net-pci-0" },
+  "timestamp": { "seconds": 1265044230, "microseconds": 450486 } }
+
 DEVICE_TRAY_MOVED
 -----------------
 
hw/qdev.c +1 -0
index 689cd54..d603f4f 100644
@@ -29,6 +29,7 @@
 #include "sysemu/sysemu.h"
 #include "qapi/error.h"
 #include "qapi/visitor.h"
+#include "qapi/qmp/qjson.h"
 
 int qdev_hotplug = 0;
 static bool qdev_hot_added = false;

hw/qdev.c static void device_unparent(Object *obj) · line 769

Quoted 8 lines
hw/qdev.c +5 -0
@@ -761,6 +762,12 @@ static void device_unparent(Object *obj)
     DeviceClass *dc = DEVICE_GET_CLASS(dev);
     BusState *bus;
 
+    if (dev->id) {
+        QObject *data = qobject_from_jsonf("{ 'device': %s }", dev->id);
+        monitor_protocol_event(QEVENT_DEVICE_DELETED, data);
+        qobject_decref(data);
+    }

Do this at the end of device_unparent, so that parents are reported after their children.

Paolo

Quoted 43 lines
hw/qdev.c +0 -0
     while (dev->num_child_bus) {
         bus = QLIST_FIRST(&dev->child_bus);
         qbus_free(bus);
include/monitor/monitor.h +1 -0
index 87fb49c..b868760 100644
@@ -39,6 +39,7 @@ typedef enum MonitorEvent {
     QEVENT_BLOCK_JOB_CANCELLED,
     QEVENT_BLOCK_JOB_ERROR,
     QEVENT_BLOCK_JOB_READY,
+    QEVENT_DEVICE_DELETED,
     QEVENT_DEVICE_TRAY_MOVED,
     QEVENT_SUSPEND,
     QEVENT_SUSPEND_DISK,
monitor.c +1 -0
index 32a6e74..2a5e7b6 100644
@@ -457,6 +457,7 @@ static const char *monitor_event_names[] = {
     [QEVENT_BLOCK_JOB_CANCELLED] = "BLOCK_JOB_CANCELLED",
     [QEVENT_BLOCK_JOB_ERROR] = "BLOCK_JOB_ERROR",
     [QEVENT_BLOCK_JOB_READY] = "BLOCK_JOB_READY",
+    [QEVENT_DEVICE_DELETED] = "DEVICE_DELETED",
     [QEVENT_DEVICE_TRAY_MOVED] = "DEVICE_TRAY_MOVED",
     [QEVENT_SUSPEND] = "SUSPEND",
     [QEVENT_SUSPEND_DISK] = "SUSPEND_DISK",
qapi-schema.json +3 -1
index 28b070f..bb361e1 100644
@@ -2354,7 +2354,9 @@
 # Notes: When this command completes, the device may not be removed from the
 #        guest.  Hot removal is an operation that requires guest cooperation.
 #        This command merely requests that the guest begin the hot removal
-#        process.
+#        process.  Completion of the device removal process is signaled with a
+#        DEVICE_DELETED event. Guest reset will automatically complete removal
+#        for all devices.
 #
 # Since: 0.14.0
 ##
#2 lore raw
Quoted 75 lines · Paolo Bonzini

Il 06/03/2013 17:49, Michael S. Tsirkin ha scritto:

Quoted 62 lines

libvirt has a long-standing bug: when removing the device, it can request removal but does not know when the removal completes. Add an event so we can fix this in a robust way.

---

Changes from v1:
- move to device_unparent
- address comments by Andreas and Eric

Andreas also suggested a more generic object-deleted event, I'm not sure how useful that is so let's add what we already need, for devices with an id and wait and see what's necessary for non-device objects?

 QMP/qmp-events.txt        | 15 +++++++++++++++
 hw/qdev.c                 |  7 +++++++
 include/monitor/monitor.h |  1 +
 monitor.c                 |  1 +
 qapi-schema.json          |  4 +++-
 5 files changed, 27 insertions(+), 1 deletion(-)
QMP/qmp-events.txt +15 -0
index b2698e4..f2f115a 100644
@@ -136,6 +136,21 @@ Example:
 Note: The "ready to complete" status is always reset by a BLOCK_JOB_ERROR
 event.
 
+DEVICE_DELETED
+-----------------
+
+Emitted whenever the device removal completion is acknowledged
+by the guest. At this point, it's safe to reuse the specified device ID.
+Device removal can be initiated by the guest or by HMP/QMP commands.
+
+Data:
+
+- "device": device name (json-string)
+
+{ "event": "DEVICE_DELETED",
+  "data": { "device": "virtio-net-pci-0" },
+  "timestamp": { "seconds": 1265044230, "microseconds": 450486 } }
+
 DEVICE_TRAY_MOVED
 -----------------
 
hw/qdev.c +1 -0
index 689cd54..d603f4f 100644
@@ -29,6 +29,7 @@
 #include "sysemu/sysemu.h"
 #include "qapi/error.h"
 #include "qapi/visitor.h"
+#include "qapi/qmp/qjson.h"
 
 int qdev_hotplug = 0;
 static bool qdev_hot_added = false;

hw/qdev.c static void device_unparent(Object *obj) · line 769

Quoted 8 lines
hw/qdev.c +5 -0
@@ -761,6 +762,12 @@ static void device_unparent(Object *obj)
     DeviceClass *dc = DEVICE_GET_CLASS(dev);
     BusState *bus;
 
+    if (dev->id) {
+        QObject *data = qobject_from_jsonf("{ 'device': %s }", dev->id);
+        monitor_protocol_event(QEVENT_DEVICE_DELETED, data);
+        qobject_decref(data);
+    }

Do this at the end of device_unparent, so that parents are reported after their children.

Quoted 1 line

Paolo

Hmm yes it seems cleaner, though we'd need to copy the id as the device can go away.

Doing this after

    while (dev->num_child_bus) {
        bus = QLIST_FIRST(&dev->child_bus);
        qbus_free(bus); 
    }

would be enough, isn't it?

Quoted 43 lines
Quoted 43 lines
hw/qdev.c +0 -0
     while (dev->num_child_bus) {
         bus = QLIST_FIRST(&dev->child_bus);
         qbus_free(bus);
include/monitor/monitor.h +1 -0
index 87fb49c..b868760 100644
@@ -39,6 +39,7 @@ typedef enum MonitorEvent {
     QEVENT_BLOCK_JOB_CANCELLED,
     QEVENT_BLOCK_JOB_ERROR,
     QEVENT_BLOCK_JOB_READY,
+    QEVENT_DEVICE_DELETED,
     QEVENT_DEVICE_TRAY_MOVED,
     QEVENT_SUSPEND,
     QEVENT_SUSPEND_DISK,
monitor.c +1 -0
index 32a6e74..2a5e7b6 100644
@@ -457,6 +457,7 @@ static const char *monitor_event_names[] = {
     [QEVENT_BLOCK_JOB_CANCELLED] = "BLOCK_JOB_CANCELLED",
     [QEVENT_BLOCK_JOB_ERROR] = "BLOCK_JOB_ERROR",
     [QEVENT_BLOCK_JOB_READY] = "BLOCK_JOB_READY",
+    [QEVENT_DEVICE_DELETED] = "DEVICE_DELETED",
     [QEVENT_DEVICE_TRAY_MOVED] = "DEVICE_TRAY_MOVED",
     [QEVENT_SUSPEND] = "SUSPEND",
     [QEVENT_SUSPEND_DISK] = "SUSPEND_DISK",
qapi-schema.json +3 -1
index 28b070f..bb361e1 100644
@@ -2354,7 +2354,9 @@
 # Notes: When this command completes, the device may not be removed from the
 #        guest.  Hot removal is an operation that requires guest cooperation.
 #        This command merely requests that the guest begin the hot removal
-#        process.
+#        process.  Completion of the device removal process is signaled with a
+#        DEVICE_DELETED event. Guest reset will automatically complete removal
+#        for all devices.
 #
 # Since: 0.14.0
 ##
#3 lore raw

Il 06/03/2013 17:55, Michael S. Tsirkin ha scritto:

Quoted 81 lines · Paolo Bonzini
Quoted 75 lines

Il 06/03/2013 17:49, Michael S. Tsirkin ha scritto:

Quoted 62 lines

libvirt has a long-standing bug: when removing the device, it can request removal but does not know when the removal completes. Add an event so we can fix this in a robust way.

---

Changes from v1:
- move to device_unparent
- address comments by Andreas and Eric

Andreas also suggested a more generic object-deleted event, I'm not sure how useful that is so let's add what we already need, for devices with an id and wait and see what's necessary for non-device objects?

 QMP/qmp-events.txt        | 15 +++++++++++++++
 hw/qdev.c                 |  7 +++++++
 include/monitor/monitor.h |  1 +
 monitor.c                 |  1 +
 qapi-schema.json          |  4 +++-
 5 files changed, 27 insertions(+), 1 deletion(-)
QMP/qmp-events.txt +15 -0
index b2698e4..f2f115a 100644
@@ -136,6 +136,21 @@ Example:
 Note: The "ready to complete" status is always reset by a BLOCK_JOB_ERROR
 event.
 
+DEVICE_DELETED
+-----------------
+
+Emitted whenever the device removal completion is acknowledged
+by the guest. At this point, it's safe to reuse the specified device ID.
+Device removal can be initiated by the guest or by HMP/QMP commands.
+
+Data:
+
+- "device": device name (json-string)
+
+{ "event": "DEVICE_DELETED",
+  "data": { "device": "virtio-net-pci-0" },
+  "timestamp": { "seconds": 1265044230, "microseconds": 450486 } }
+
 DEVICE_TRAY_MOVED
 -----------------
 
hw/qdev.c +1 -0
index 689cd54..d603f4f 100644
@@ -29,6 +29,7 @@
 #include "sysemu/sysemu.h"
 #include "qapi/error.h"
 #include "qapi/visitor.h"
+#include "qapi/qmp/qjson.h"
 
 int qdev_hotplug = 0;
 static bool qdev_hot_added = false;

hw/qdev.c static void device_unparent(Object *obj) · line 769

Quoted 8 lines
hw/qdev.c +5 -0
@@ -761,6 +762,12 @@ static void device_unparent(Object *obj)
     DeviceClass *dc = DEVICE_GET_CLASS(dev);
     BusState *bus;
 
+    if (dev->id) {
+        QObject *data = qobject_from_jsonf("{ 'device': %s }", dev->id);
+        monitor_protocol_event(QEVENT_DEVICE_DELETED, data);
+        qobject_decref(data);
+    }

Do this at the end of device_unparent, so that parents are reported after their children.

Quoted 1 line

Paolo

Hmm yes it seems cleaner, though we'd need to copy the id as the device can go away.

Quoted 6 lines

Doing this after

    while (dev->num_child_bus) {
        bus = QLIST_FIRST(&dev->child_bus);
        qbus_free(bus); 
    }

would be enough, isn't it?

Yes, but don't worry - the device cannot go away, it is protected while device_unparent is called (see object_unparent in qom/object.c). So put it wherever you prefer.

Paolo

Quoted 43 lines
Quoted 43 lines
Quoted 43 lines
hw/qdev.c +0 -0
     while (dev->num_child_bus) {
         bus = QLIST_FIRST(&dev->child_bus);
         qbus_free(bus);
include/monitor/monitor.h +1 -0
index 87fb49c..b868760 100644
@@ -39,6 +39,7 @@ typedef enum MonitorEvent {
     QEVENT_BLOCK_JOB_CANCELLED,
     QEVENT_BLOCK_JOB_ERROR,
     QEVENT_BLOCK_JOB_READY,
+    QEVENT_DEVICE_DELETED,
     QEVENT_DEVICE_TRAY_MOVED,
     QEVENT_SUSPEND,
     QEVENT_SUSPEND_DISK,
monitor.c +1 -0
index 32a6e74..2a5e7b6 100644
@@ -457,6 +457,7 @@ static const char *monitor_event_names[] = {
     [QEVENT_BLOCK_JOB_CANCELLED] = "BLOCK_JOB_CANCELLED",
     [QEVENT_BLOCK_JOB_ERROR] = "BLOCK_JOB_ERROR",
     [QEVENT_BLOCK_JOB_READY] = "BLOCK_JOB_READY",
+    [QEVENT_DEVICE_DELETED] = "DEVICE_DELETED",
     [QEVENT_DEVICE_TRAY_MOVED] = "DEVICE_TRAY_MOVED",
     [QEVENT_SUSPEND] = "SUSPEND",
     [QEVENT_SUSPEND_DISK] = "SUSPEND_DISK",
qapi-schema.json +3 -1
index 28b070f..bb361e1 100644
@@ -2354,7 +2354,9 @@
 # Notes: When this command completes, the device may not be removed from the
 #        guest.  Hot removal is an operation that requires guest cooperation.
 #        This command merely requests that the guest begin the hot removal
-#        process.
+#        process.  Completion of the device removal process is signaled with a
+#        DEVICE_DELETED event. Guest reset will automatically complete removal
+#        for all devices.
 #
 # Since: 0.14.0
 ##
#4 lore raw
In reply to Michael S. Tsirkin

Am 06.03.2013 17:55, schrieb Michael S. Tsirkin:

Quoted 9 lines · Paolo Bonzini
Quoted 8 lines

Il 06/03/2013 17:49, Michael S. Tsirkin ha scritto:

Quoted 7 lines

libvirt has a long-standing bug: when removing the device, it can request removal but does not know when the removal completes. Add an event so we can fix this in a robust way.

---

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Changes from v1:
- move to device_unparent
- address comments by Andreas and Eric

Andreas also suggested a more generic object-deleted event, I'm not sure how useful that is so let's add what we already need, for devices with an id and wait and see what's necessary for non-device objects?

Fine with me, just wanted to bring that up for consideration since it impacts not only where we call it but also what data we can provide.

Quoted 30 lines
Quoted 25 lines
Quoted 13 lines
hw/qdev.c +1 -0
index 689cd54..d603f4f 100644
@@ -29,6 +29,7 @@
 #include "sysemu/sysemu.h"
 #include "qapi/error.h"
 #include "qapi/visitor.h"
+#include "qapi/qmp/qjson.h"
 
 int qdev_hotplug = 0;
 static bool qdev_hot_added = false;

hw/qdev.c static void device_unparent(Object *obj) · line 769

Quoted 8 lines
hw/qdev.c +5 -0
@@ -761,6 +762,12 @@ static void device_unparent(Object *obj)
     DeviceClass *dc = DEVICE_GET_CLASS(dev);
     BusState *bus;
 
+    if (dev->id) {
+        QObject *data = qobject_from_jsonf("{ 'device': %s }", dev->id);
+        monitor_protocol_event(QEVENT_DEVICE_DELETED, data);
+        qobject_decref(data);
+    }

Do this at the end of device_unparent, so that parents are reported after their children.

Quoted 1 line

Paolo

Hmm yes it seems cleaner, though we'd need to copy the id as the device can go away.

Quoted 6 lines

Doing this after

    while (dev->num_child_bus) {
        bus = QLIST_FIRST(&dev->child_bus);
        qbus_free(bus); 
    }

would be enough, isn't it?

I had implicitly suggested after dev->realized and before dev->parent_bus. The reason being that what is inside the if (dev->realized) {} block actually still needs to be moved into an unrealize function, and realized = false might still do device-specific cleanups that could consume time. I'll try to cook something up tonight, shouldn't collide with your change.

Andreas

Signature · 2 lines
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg