Skip to contentSkip to conversation

DEVICE_UNPLUG_ERROR QAPI event

PATCH In review Daniel Henrique Barboza 4 participants 24 messages Jul 19, 2021 – Aug 23, 2021

lore
Share & tools
lore
#0 lore raw

v6 cover DEVICE_UNPLUG_ERROR QAPI event

Hi,

This version was rebased with QEMU master @ 7457b407edd6e8555e4b4 and has a few simple changes based on Greg's review.

changes from v5:
- patch 1:
* fixed function name in commit msg
-patch 6:
* kept error message in a single line
- all patches:
* added Greg's R-b
- v5 link: https://lists.gnu.org/archive/html/qemu-devel/2021-07/msg03239.html

changes from v4:
- patch 1 (new):
* avoid emitting MEM_UNPLUG_ERROR when dev->id == NULL in
memory_hotplug.c
- patch 2 (new):
* avoid emitting MEM_UNPLUG_ERROR when dev->id == NULL in
spapr.c
- patch 3 (new):
* do not error_report() when drc->dev->id == NULL
- patch 4 (new):
* update DEVICE_DELETED API docs
- patch 5 (former 1):
* added 2 spaces after each sentence
* appended "Some errors cannot be detected." in DEVICE_UNPLUG_ERROR
docs
* changed release from 6.1 to 6.2 in docs
* changed DEVICE_UNPLUG_ERROR API to behave more like device_deleted
- patches 6 and 7:
* changed to use the new DEVICE_UNPLUG_ERROR API
- v4 link: https://lists.gnu.org/archive/html/qemu-devel/2021-07/msg01355.html

changes from v3:
- patch 1:
* fixed format
- all patches:
* rebased with master
* added David's R-b
- v3 link: https://lists.gnu.org/archive/html/qemu-devel/2021-06/msg05842.html

changes from v2:
- patch 1:
* moved DEVICE_UNPLUG_ERROR declaration to qapi/qdev.json
* updated 'device_del' description
* added 'deprecated' notice on MEM_UNPLUG_ERROR
* added MEM_UNPLUG_ERROR 'deprecated' info in docs/system/deprecated.rst
- patch 2:
* send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
- patch 3 (new):
* send DEVICE_UNPLUG_ERROR in acpi/memory_hotplug.c
- v2 link: https://lists.gnu.org/archive/html/qemu-devel/2021-06/msg01304.html

changes from v1:
- former patches 1 and 2: dropped
- patch 1 (former 3): changed the version to '6.1'
- patch 2 (former 4): add a DEVICE_UNPLUG_ERROR event in the device unplug error path of CPUs and DIMMs
- v1 link: https://lists.gnu.org/archive/html/qemu-devel/2021-03/msg04682.html

Daniel Henrique Barboza (7):
hw/acpi/memory_hotplug.c: avoid sending MEM_UNPLUG_ERROR if dev->id is
NULL
spapr.c: avoid sending MEM_UNPLUG_ERROR if dev->id is NULL spapr_drc.c: do not error_report() when drc->dev->id == NULL qapi/qdev.json: fix DEVICE_DELETED parameters doc qapi/qdev.json: add DEVICE_UNPLUG_ERROR QAPI event
spapr: use DEVICE_UNPLUG_ERROR to report unplug errors memory_hotplug.c: send DEVICE_UNPLUG_ERROR in
acpi_memory_hotplug_write()

 docs/about/deprecated.rst | 10 ++++++++++
 hw/acpi/memory_hotplug.c  | 19 +++++++++++++++++--
 hw/ppc/spapr.c            | 17 +++++++++++++----
 hw/ppc/spapr_drc.c        | 21 +++++++++++++++------
 qapi/machine.json         |  6 +++++-
 qapi/qdev.json            | 34 +++++++++++++++++++++++++++++++---
 stubs/qdev.c              |  7 +++++++
 7 files changed, 98 insertions(+), 16 deletions(-)
Signature · 1 line
2.31.1
#1 lore raw

PATCH v6 1/7 hw/acpi/memory_hotplug.c: avoid sending MEM_UNPLUG_ERROR if dev->id is NULL

qapi_event_send_mem_unplug_error() deals with @device being NULL by replacing it with an empty string ("") when emitting the event. Aside from the fact that this is a side effect that can be patched someday, there's also the lack of utility that the event brings to listeners, e.g. "a memory unplug error happened somewhere".

We're better of not emitting the event if dev->id is NULL. Next patches will introduce a new device unplug error event that is better suited to deal with dev->id NULL scenarios. MEM_UNPLUG_ERROR will continue to be emitted to avoid breaking existing APIs, but it'll be deprecated and removed in the future.

---
 hw/acpi/memory_hotplug.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
hw/acpi/memory_hotplug.c +7 -2
index af37889423..e37acb0367 100644
@@ -177,9 +177,14 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
             /* call pc-dimm unplug cb */
             hotplug_handler_unplug(hotplug_ctrl, dev, &local_err);
             if (local_err) {
+                const char *error_pretty = error_get_pretty(local_err);
+
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
-                qapi_event_send_mem_unplug_error(dev->id,
-                                                 error_get_pretty(local_err));
+
+                if (dev->id) {
+                    qapi_event_send_mem_unplug_error(dev->id, error_pretty);
+                }
+
                 error_free(local_err);
                 break;
             }
Signature · 1 line
2.31.1
#2 lore raw

Re: [PATCH for-6.2 v6 1/7] hw/acpi/memory_hotplug.c: avoid sending MEM_UNPLUG_ERROR if dev->id is NULL

I apologize for the not reviewing this promptly.

Quoted 3 lines · Daniel Henrique Barboza

qapi_event_send_mem_unplug_error() deals with @device being NULL by replacing it with an empty string ("") when emitting the event. Aside from the fact that this is a side effect that can be patched someday,

I guess by "side effect" you allude to the output visitor's misfeature to map null pointer to "".

Quoted 2 lines

there's also the lack of utility that the event brings to listeners, e.g. "a memory unplug error happened somewhere".

True.

Quoted 1 line

We're better of not emitting the event if dev->id is NULL.

On a green field, yes. But is it worth an incompatible change now? I doubt it.

Quoted 24 lines

Next patches
will introduce a new device unplug error event that is better suited to deal with dev->id NULL scenarios. MEM_UNPLUG_ERROR will continue to be emitted to avoid breaking existing APIs, but it'll be deprecated and removed in the future.

---
 hw/acpi/memory_hotplug.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
hw/acpi/memory_hotplug.c +2 -0
index af37889423..e37acb0367 100644
@@ -177,9 +177,14 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
             /* call pc-dimm unplug cb */
             hotplug_handler_unplug(hotplug_ctrl, dev, &local_err);
             if (local_err) {
+                const char *error_pretty = error_get_pretty(local_err);
+
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);

hw/acpi/memory_hotplug.c static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data, · line 190

Quoted 10 lines
hw/acpi/memory_hotplug.c +5 -2
-                qapi_event_send_mem_unplug_error(dev->id,
-                                                 error_get_pretty(local_err));
+
+                if (dev->id) {
+                    qapi_event_send_mem_unplug_error(dev->id, error_pretty);
+                }
+
                 error_free(local_err);
                 break;
             }

Three options:

1. Make the incompatible change. Keep this patch. Needs a release note.

2. Continue to rely on the output visitor's misfeature. Drop this patch.

Relying on the misfeature is best avoided, at least in new code.

3. Make the mapping explicit here, and avoid relying on the misfeature. Something like the appended patch.

I don't like 1. I think the choice between 2. and 3. depends on how it fits with the remainder of this series.

hw/acpi/memory_hotplug.c +1 -1
index af37889423..89c411dd5c 100644
@@ -178,7 +178,7 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
             hotplug_handler_unplug(hotplug_ctrl, dev, &local_err);
             if (local_err) {
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
-                qapi_event_send_mem_unplug_error(dev->id,
+                qapi_event_send_mem_unplug_error(dev->id ?: "",
                                                  error_get_pretty(local_err));
                 error_free(local_err);
                 break;
#3 lore raw

Re: [PATCH for-6.2 v6 1/7] hw/acpi/memory_hotplug.c: avoid sending MEM_UNPLUG_ERROR if dev->id is NULL

Quoted 61 lines · Markus Armbruster

I apologize for the not reviewing this promptly.

Quoted 3 lines · Daniel Henrique Barboza

qapi_event_send_mem_unplug_error() deals with @device being NULL by replacing it with an empty string ("") when emitting the event. Aside from the fact that this is a side effect that can be patched someday,

I guess by "side effect" you allude to the output visitor's misfeature to map null pointer to "".

Quoted 2 lines

there's also the lack of utility that the event brings to listeners, e.g. "a memory unplug error happened somewhere".

True.

Quoted 1 line

We're better of not emitting the event if dev->id is NULL.

On a green field, yes. But is it worth an incompatible change now? I doubt it.

Quoted 24 lines

Next patches
will introduce a new device unplug error event that is better suited to deal with dev->id NULL scenarios. MEM_UNPLUG_ERROR will continue to be emitted to avoid breaking existing APIs, but it'll be deprecated and removed in the future.

---
  hw/acpi/memory_hotplug.c | 9 +++++++--
  1 file changed, 7 insertions(+), 2 deletions(-)
hw/acpi/memory_hotplug.c +2 -0
index af37889423..e37acb0367 100644
@@ -177,9 +177,14 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
              /* call pc-dimm unplug cb */
              hotplug_handler_unplug(hotplug_ctrl, dev, &local_err);
              if (local_err) {
+                const char *error_pretty = error_get_pretty(local_err);
+
                  trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);

hw/acpi/memory_hotplug.c static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data, · line 190

Quoted 10 lines
hw/acpi/memory_hotplug.c +5 -2
-                qapi_event_send_mem_unplug_error(dev->id,
-                                                 error_get_pretty(local_err));
+
+                if (dev->id) {
+                    qapi_event_send_mem_unplug_error(dev->id, error_pretty);
+                }
+
                  error_free(local_err);
                  break;
              }

Three options:

1. Make the incompatible change. Keep this patch. Needs a release note.

Quoted 7 lines

2. Continue to rely on the output visitor's misfeature. Drop this patch.

Relying on the misfeature is best avoided, at least in new code.

3. Make the mapping explicit here, and avoid relying on the misfeature. Something like the appended patch.

I like (3) more. If no one opposes I'll respin patches 1 and 2 with this approach.

Daniel

Quoted 19 lines

I don't like 1. I think the choice between 2. and 3. depends on how it fits with the remainder of this series.

hw/acpi/memory_hotplug.c +1 -1
index af37889423..89c411dd5c 100644
@@ -178,7 +178,7 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
              hotplug_handler_unplug(hotplug_ctrl, dev, &local_err);
              if (local_err) {
                  trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
-                qapi_event_send_mem_unplug_error(dev->id,
+                qapi_event_send_mem_unplug_error(dev->id ?: "",
                                                   error_get_pretty(local_err));
                  error_free(local_err);
                  break;
#4 lore raw

PATCH v6 2/7 spapr.c: avoid sending MEM_UNPLUG_ERROR if dev->id is NULL

In reply to Daniel Henrique Barboza

As done in hw/acpi/memory_hotplug.c, avoid sending qapi_event_send_mem_unplug_error() if dev->id is NULL.

---
 hw/ppc/spapr.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
hw/ppc/spapr.c +5 -3
index 81699d4f8b..1611d7ab05 100644
@@ -3688,9 +3688,11 @@ void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev)
      * Tell QAPI that something happened and the memory
      * hotunplug wasn't successful.
      */
-    qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
-                                 "for device %s", dev->id);
-    qapi_event_send_mem_unplug_error(dev->id, qapi_error);
+    if (dev->id) {
+        qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
+                                     "for device %s", dev->id);
+        qapi_event_send_mem_unplug_error(dev->id, qapi_error);
+    }
 }
 
 /* Callback to be called during DRC release. */
Signature · 1 line
2.31.1
#5 lore raw

Re: [PATCH for-6.2 v6 2/7] spapr.c: avoid sending MEM_UNPLUG_ERROR if dev->id is NULL

Quoted 19 lines · Daniel Henrique Barboza

As done in hw/acpi/memory_hotplug.c, avoid sending qapi_event_send_mem_unplug_error() if dev->id is NULL.

---
 hw/ppc/spapr.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
hw/ppc/spapr.c +0 -1
index 81699d4f8b..1611d7ab05 100644
@@ -3688,9 +3688,11 @@ void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev)
      * Tell QAPI that something happened and the memory
      * hotunplug wasn't successful.
      */
-    qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "

hw/ppc/spapr.c void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev) · line 3698

Quoted 10 lines
hw/ppc/spapr.c +5 -2
-                                 "for device %s", dev->id);
-    qapi_event_send_mem_unplug_error(dev->id, qapi_error);
+    if (dev->id) {
+        qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
+                                     "for device %s", dev->id);
+        qapi_event_send_mem_unplug_error(dev->id, qapi_error);
+    }
 }
 
 /* Callback to be called during DRC release. */

My review of PATCH 1 applies.

#6 lore raw

PATCH v6 3/7 spapr_drc.c: do not error_report() when drc->dev->id == NULL

In reply to Daniel Henrique Barboza

The error_report() call in drc_unisolate_logical() is not considering that drc->dev->id can be NULL, and the underlying functions error_report() calls to do its job (vprintf(), g_strdup_printf() ...) has undefined behavior when trying to handle "%s" with NULL arguments.

Besides, there is no utility into reporting that an unknown device was rejected by the guest.

---
 hw/ppc/spapr_drc.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
hw/ppc/spapr_drc.c +5 -2
index a2f2634601..a4d9496f76 100644
@@ -167,8 +167,11 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
             }
 
             drc->unplug_requested = false;
-            error_report("Device hotunplug rejected by the guest "
-                         "for device %s", drc->dev->id);
+
+            if (drc->dev->id) {
+                error_report("Device hotunplug rejected by the guest "
+                             "for device %s", drc->dev->id);
+            }
 
             /*
              * TODO: send a QAPI DEVICE_UNPLUG_ERROR event when
Signature · 1 line
2.31.1
Reviewed-by on v6 3/7
#7 lore raw

Re: [PATCH for-6.2 v6 3/7] spapr_drc.c: do not error_report() when drc->dev->id == NULL

Quoted 22 lines · Daniel Henrique Barboza

The error_report() call in drc_unisolate_logical() is not considering that drc->dev->id can be NULL, and the underlying functions error_report() calls to do its job (vprintf(), g_strdup_printf() ...) has undefined behavior when trying to handle "%s" with NULL arguments.

Besides, there is no utility into reporting that an unknown device was rejected by the guest.

---
 hw/ppc/spapr_drc.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
hw/ppc/spapr_drc.c +0 -0
index a2f2634601..a4d9496f76 100644
@@ -167,8 +167,11 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
             }
 
             drc->unplug_requested = false;

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 177

Quoted 10 lines
hw/ppc/spapr_drc.c +5 -2
-            error_report("Device hotunplug rejected by the guest "
-                         "for device %s", drc->dev->id);
+
+            if (drc->dev->id) {
+                error_report("Device hotunplug rejected by the guest "
+                             "for device %s", drc->dev->id);
+            }
 
             /*
              * TODO: send a QAPI DEVICE_UNPLUG_ERROR event when

This differs from PATCH 1 and 2 in that it actually fixes a crash bug.

The alternative is something like

error_report("Device hotunplug rejected by the guest "
"for device %s", drc->dev->id ?: "");

If the maintainer is okay with dropping the message when the device has no ID, then so am I:

#8 lore raw

Re: [PATCH for-6.2 v6 3/7] spapr_drc.c: do not error_report() when drc->dev->id == NULL

Quoted 37 lines · Markus Armbruster
Quoted 22 lines · Daniel Henrique Barboza

The error_report() call in drc_unisolate_logical() is not considering that drc->dev->id can be NULL, and the underlying functions error_report() calls to do its job (vprintf(), g_strdup_printf() ...) has undefined behavior when trying to handle "%s" with NULL arguments.

Besides, there is no utility into reporting that an unknown device was rejected by the guest.

---
 hw/ppc/spapr_drc.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
hw/ppc/spapr_drc.c +0 -0
index a2f2634601..a4d9496f76 100644
@@ -167,8 +167,11 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
             }
 
             drc->unplug_requested = false;

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 177

Quoted 10 lines
hw/ppc/spapr_drc.c +5 -2
-            error_report("Device hotunplug rejected by the guest "
-                         "for device %s", drc->dev->id);
+
+            if (drc->dev->id) {
+                error_report("Device hotunplug rejected by the guest "
+                             "for device %s", drc->dev->id);
+            }
 
             /*
              * TODO: send a QAPI DEVICE_UNPLUG_ERROR event when

This differs from PATCH 1 and 2 in that it actually fixes a crash bug.

Quoted 7 lines

The alternative is something like

error_report("Device hotunplug rejected by the guest "
"for device %s", drc->dev->id ?: "");

If the maintainer is okay with dropping the message when the device has no ID, then so am I:

I am, given how briefly this message has even existed - and through all that time it would have crashed if you tried.

Quoted 1 line

Folded into my tree.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#9 lore raw

PATCH v6 4/7 qapi/qdev.json: fix DEVICE_DELETED parameters doc

In reply to Daniel Henrique Barboza

Clarify that @device is optional and that 'path' is the device path from QOM.

This change follows Markus' suggestion verbatim, provided in full context here:

https://lists.gnu.org/archive/html/qemu-devel/2021-07/msg01891.html

---
 qapi/qdev.json | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
qapi/qdev.json +2 -2
index b83178220b..d1d3681a50 100644
@@ -108,9 +108,9 @@
 # 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.
 #
-# @device: device name
+# @device: the device's ID if it has one
 #
-# @path: device path
+# @path: the device's path within the object model
 #
 # Since: 1.5
 #
Signature · 1 line
2.31.1
Reviewed-by on v6 4/7
#10 lore raw

Re: [PATCH for-6.2 v6 4/7] qapi/qdev.json: fix DEVICE_DELETED parameters doc

Quoted 20 lines · Daniel Henrique Barboza

Clarify that @device is optional and that 'path' is the device path from QOM.

This change follows Markus' suggestion verbatim, provided in full context here:

https://lists.gnu.org/archive/html/qemu-devel/2021-07/msg01891.html

---
 qapi/qdev.json | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
qapi/qdev.json +0 -0
index b83178220b..d1d3681a50 100644

qapi/qdev.json line 113

Quoted 8 lines
qapi/qdev.json +2 -2
@@ -108,9 +108,9 @@
 # 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.
 #
-# @device: device name
+# @device: the device's ID if it has one
 #
-# @path: device path
+# @path: the device's path within the object model

Recommend "the device's QOM path".

qapi/qdev.json line 116

Quoted 3 lines
qapi/qdev.json +0 -0
 #
 # Since: 1.5
 #
#11 lore raw

PATCH v6 5/7 qapi/qdev.json: add DEVICE_UNPLUG_ERROR QAPI event

In reply to Daniel Henrique Barboza

At this moment we only provide one event to report a hotunplug error, MEM_UNPLUG_ERROR. As of Linux kernel 5.12 and QEMU 6.0.0, the pseries machine is now able to report unplug errors for other device types, such as CPUs.

Instead of creating a (device_type)_UNPLUG_ERROR for each new device, create a generic DEVICE_UNPLUG_ERROR event that can be used by all unplug errors in the future. This event has a similar API as the existing DEVICE_DELETED event, with an extra optional 'msg' parameter that can be used to explain the reason for the error.

With this new generic event, MEM_UNPLUG_ERROR is now marked as deprecated.

---
 docs/about/deprecated.rst | 10 ++++++++++
 qapi/machine.json         |  6 +++++-
 qapi/qdev.json            | 30 +++++++++++++++++++++++++++++-
 stubs/qdev.c              |  7 +++++++
 4 files changed, 51 insertions(+), 2 deletions(-)
docs/about/deprecated.rst +10 -0
index 6d438f1c8d..c0c3431ada 100644
@@ -204,6 +204,16 @@ The ``I7200`` guest CPU relies on the nanoMIPS ISA, which is deprecated
 (the ISA has never been upstreamed to a compiler toolchain). Therefore
 this CPU is also deprecated.
 
+
+QEMU API (QAPI) events
+----------------------
+
+``MEM_UNPLUG_ERROR`` (since 6.2)
+''''''''''''''''''''''''''''''''''''''''''''''''''''''''
+
+Use the more generic event ``DEVICE_UNPLUG_ERROR`` instead.
+
+
 System emulator machines
 ------------------------
 
qapi/machine.json +5 -1
index c3210ee1fb..a595c753d2 100644
@@ -1271,6 +1271,9 @@
 #
 # @msg: Informative message
 #
+# Features:
+# @deprecated: This event is deprecated. Use @DEVICE_UNPLUG_ERROR instead.
+#
 # Since: 2.4
 #
 # Example:
@@ -1283,7 +1286,8 @@
 #
 ##
 { 'event': 'MEM_UNPLUG_ERROR',
-  'data': { 'device': 'str', 'msg': 'str' } }
+  'data': { 'device': 'str', 'msg': 'str' },
+  'features': ['deprecated'] }
 
 ##
 # @SMPConfiguration:
qapi/qdev.json +29 -1
index d1d3681a50..52c36c7b9c 100644
@@ -84,7 +84,9 @@
 #        This command merely requests that the guest begin the hot removal
 #        process.  Completion of the device removal process is signaled with a
 #        DEVICE_DELETED event. Guest reset will automatically complete removal
-#        for all devices.
+#        for all devices.  If an error in the hot removal process is detected,
+#        the device will not be removed and a DEVICE_UNPLUG_ERROR event is
+#        sent.  Some errors cannot be detected.
 #
 # Since: 0.14
 #
@@ -124,3 +126,29 @@
 ##
 { 'event': 'DEVICE_DELETED',
   'data': { '*device': 'str', 'path': 'str' } }
+
+##
+# @DEVICE_UNPLUG_ERROR:
+#
+# Emitted when a device hot unplug error occurs.
+#
+# @device: the device's ID if it has one
+#
+# @path: the device's path within the object model
+#
+# @msg: optional informative message
+#
+# Since: 6.2
+#
+# Example:
+#
+# <- { "event": "DEVICE_UNPLUG_ERROR"
+#      "data": { "device": "core1",
+#                "msg": "Device hotunplug rejected by the guest for device core1",
+#                "path": "/machine/peripheral/core1" },
+#      },
+#      "timestamp": { "seconds": 1615570772, "microseconds": 202844 } }
+#
+##
+{ 'event': 'DEVICE_UNPLUG_ERROR',
+  'data': { '*device': 'str', 'path': 'str' , '*msg': 'str' } }
stubs/qdev.c +7 -0
index 92e6143134..ffa8f7b59e 100644
@@ -21,3 +21,10 @@ void qapi_event_send_device_deleted(bool has_device,
 {
     /* Nothing to do. */
 }
+
+void qapi_event_send_device_unplug_error(bool has_device, const char *device,
+                                         const char *path,
+                                         bool has_msg, const char *msg)
+{
+    /* Nothing to do. */
+}
Signature · 1 line
2.31.1
Reviewed-by on v6 5/7
#12 lore raw

Re: [PATCH for-6.2 v6 5/7] qapi/qdev.json: add DEVICE_UNPLUG_ERROR QAPI event

Quoted 85 lines · Daniel Henrique Barboza

At this moment we only provide one event to report a hotunplug error, MEM_UNPLUG_ERROR. As of Linux kernel 5.12 and QEMU 6.0.0, the pseries machine is now able to report unplug errors for other device types, such as CPUs.

Instead of creating a (device_type)_UNPLUG_ERROR for each new device, create a generic DEVICE_UNPLUG_ERROR event that can be used by all unplug errors in the future. This event has a similar API as the existing DEVICE_DELETED event, with an extra optional 'msg' parameter that can be used to explain the reason for the error.

With this new generic event, MEM_UNPLUG_ERROR is now marked as deprecated.

---
 docs/about/deprecated.rst | 10 ++++++++++
 qapi/machine.json         |  6 +++++-
 qapi/qdev.json            | 30 +++++++++++++++++++++++++++++-
 stubs/qdev.c              |  7 +++++++
 4 files changed, 51 insertions(+), 2 deletions(-)
docs/about/deprecated.rst +10 -0
index 6d438f1c8d..c0c3431ada 100644
@@ -204,6 +204,16 @@ The ``I7200`` guest CPU relies on the nanoMIPS ISA, which is deprecated
 (the ISA has never been upstreamed to a compiler toolchain). Therefore
 this CPU is also deprecated.
 
+
+QEMU API (QAPI) events
+----------------------
+
+``MEM_UNPLUG_ERROR`` (since 6.2)
+''''''''''''''''''''''''''''''''''''''''''''''''''''''''
+
+Use the more generic event ``DEVICE_UNPLUG_ERROR`` instead.
+
+
 System emulator machines
 ------------------------
 
qapi/machine.json +5 -1
index c3210ee1fb..a595c753d2 100644
@@ -1271,6 +1271,9 @@
 #
 # @msg: Informative message
 #
+# Features:
+# @deprecated: This event is deprecated. Use @DEVICE_UNPLUG_ERROR instead.
+#
 # Since: 2.4
 #
 # Example:
@@ -1283,7 +1286,8 @@
 #
 ##
 { 'event': 'MEM_UNPLUG_ERROR',
-  'data': { 'device': 'str', 'msg': 'str' } }
+  'data': { 'device': 'str', 'msg': 'str' },
+  'features': ['deprecated'] }
 
 ##
 # @SMPConfiguration:
qapi/qdev.json +3 -1
index d1d3681a50..52c36c7b9c 100644
@@ -84,7 +84,9 @@
 #        This command merely requests that the guest begin the hot removal
 #        process.  Completion of the device removal process is signaled with a
 #        DEVICE_DELETED event. Guest reset will automatically complete removal
-#        for all devices.
+#        for all devices.  If an error in the hot removal process is detected,
+#        the device will not be removed and a DEVICE_UNPLUG_ERROR event is
+#        sent.  Some errors cannot be detected.
 #
 # Since: 0.14
 #
@@ -124,3 +126,29 @@
 ##
 { 'event': 'DEVICE_DELETED',

qapi/qdev.json line 137

Quoted 10 lines
qapi/qdev.json +9 -0
   'data': { '*device': 'str', 'path': 'str' } }
+
+##
+# @DEVICE_UNPLUG_ERROR:
+#
+# Emitted when a device hot unplug error occurs.
+#
+# @device: the device's ID if it has one
+#
+# @path: the device's path within the object model

Recommend "the device's QOM path".

qapi/qdev.json line 139

Quoted 2 lines
qapi/qdev.json +2 -0
+#
+# @msg: optional informative message

Is this useful? I guess the remaining patches will tell.

Quoted 20 lines
qapi/qdev.json +15 -0
+#
+# Since: 6.2
+#
+# Example:
+#
+# <- { "event": "DEVICE_UNPLUG_ERROR"
+#      "data": { "device": "core1",
+#                "msg": "Device hotunplug rejected by the guest for device core1",
+#                "path": "/machine/peripheral/core1" },
+#      },
+#      "timestamp": { "seconds": 1615570772, "microseconds": 202844 } }
+#
+##
+{ 'event': 'DEVICE_UNPLUG_ERROR',
+  'data': { '*device': 'str', 'path': 'str' , '*msg': 'str' } }
stubs/qdev.c +0 -0
index 92e6143134..ffa8f7b59e 100644

stubs/qdev.c void qapi_event_send_device_deleted(bool has_device, · line 30

Quoted 10 lines
stubs/qdev.c +7 -0
@@ -21,3 +21,10 @@ void qapi_event_send_device_deleted(bool has_device,
 {
     /* Nothing to do. */
 }
+
+void qapi_event_send_device_unplug_error(bool has_device, const char *device,
+                                         const char *path,
+                                         bool has_msg, const char *msg)
+{
+    /* Nothing to do. */
+}

I'm reserving judgement on member @msg. Other than that:

#13 lore raw

PATCH v6 6/7 spapr: use DEVICE_UNPLUG_ERROR to report unplug errors

In reply to Daniel Henrique Barboza

Linux Kernel 5.12 is now unisolating CPU DRCs in the device_removal error path, signalling that the hotunplug process wasn't successful. This allow us to send a DEVICE_UNPLUG_ERROR in drc_unisolate_logical() to signal this error to the management layer.

We also have another error path in spapr_memory_unplug_rollback() for configured LMB DRCs. Kernels older than 5.13 will not unisolate the LMBs in the hotunplug error path, but it will reconfigure them. Let's send the DEVICE_UNPLUG_ERROR event in that code path as well to cover the case of older kernels.

---
 hw/ppc/spapr.c     |  9 ++++++++-
 hw/ppc/spapr_drc.c | 18 ++++++++++++------
 2 files changed, 20 insertions(+), 7 deletions(-)
hw/ppc/spapr.c +8 -1
index 1611d7ab05..5459f9a7e9 100644
@@ -29,6 +29,7 @@
 #include "qemu/datadir.h"
 #include "qapi/error.h"
 #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
 #include "qapi/visitor.h"
 #include "sysemu/sysemu.h"
 #include "sysemu/hostmem.h"
@@ -3686,13 +3687,19 @@ void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev)
 
     /*
      * Tell QAPI that something happened and the memory
-     * hotunplug wasn't successful.
+     * hotunplug wasn't successful. Keep sending
+     * MEM_UNPLUG_ERROR even while sending DEVICE_UNPLUG_ERROR
+     * until the deprecation MEM_UNPLUG_ERROR is due.
      */
     if (dev->id) {
         qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
                                      "for device %s", dev->id);
         qapi_event_send_mem_unplug_error(dev->id, qapi_error);
     }
+
+    qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                        dev->canonical_path,
+                                        qapi_error != NULL, qapi_error);
 }
 
 /* Callback to be called during DRC release. */
hw/ppc/spapr_drc.c +12 -6
index a4d9496f76..8f0479631f 100644
@@ -17,6 +17,8 @@
 #include "hw/ppc/spapr_drc.h"
 #include "qom/object.h"
 #include "migration/vmstate.h"
+#include "qapi/error.h"
+#include "qapi/qapi-events-qdev.h"
 #include "qapi/visitor.h"
 #include "qemu/error-report.h"
 #include "hw/ppc/spapr.h" /* for RTAS return codes */
@@ -160,6 +162,11 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
          * means that the kernel is refusing the removal.
          */
         if (drc->unplug_requested && drc->dev) {
+            const char qapi_error_fmt[] = \
+"Device hotunplug rejected by the guest for device %s";
+
+            g_autofree char *qapi_error = NULL;
+
             if (spapr_drc_type(drc) == SPAPR_DR_CONNECTOR_TYPE_LMB) {
                 spapr = SPAPR_MACHINE(qdev_get_machine());
 
@@ -169,14 +176,13 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
             drc->unplug_requested = false;
 
             if (drc->dev->id) {
-                error_report("Device hotunplug rejected by the guest "
-                             "for device %s", drc->dev->id);
+                qapi_error = g_strdup_printf(qapi_error_fmt, drc->dev->id);
+                error_report(qapi_error_fmt, drc->dev->id);
             }
 
-            /*
-             * TODO: send a QAPI DEVICE_UNPLUG_ERROR event when
-             * it is implemented.
-             */
+            qapi_event_send_device_unplug_error(!!drc->dev->id, drc->dev->id,
+                                                drc->dev->canonical_path,
+                                                qapi_error != NULL, qapi_error);
         }
 
         return RTAS_OUT_SUCCESS; /* Nothing to do */
Signature · 1 line
2.31.1
#14 lore raw

Re: [PATCH for-6.2 v6 6/7] spapr: use DEVICE_UNPLUG_ERROR to report unplug errors

Quoted 40 lines · Daniel Henrique Barboza

Linux Kernel 5.12 is now unisolating CPU DRCs in the device_removal error path, signalling that the hotunplug process wasn't successful. This allow us to send a DEVICE_UNPLUG_ERROR in drc_unisolate_logical() to signal this error to the management layer.

We also have another error path in spapr_memory_unplug_rollback() for configured LMB DRCs. Kernels older than 5.13 will not unisolate the LMBs in the hotunplug error path, but it will reconfigure them. Let's send the DEVICE_UNPLUG_ERROR event in that code path as well to cover the case of older kernels.

---
 hw/ppc/spapr.c     |  9 ++++++++-
 hw/ppc/spapr_drc.c | 18 ++++++++++++------
 2 files changed, 20 insertions(+), 7 deletions(-)
hw/ppc/spapr.c +4 -1
index 1611d7ab05..5459f9a7e9 100644
@@ -29,6 +29,7 @@
 #include "qemu/datadir.h"
 #include "qapi/error.h"
 #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
 #include "qapi/visitor.h"
 #include "sysemu/sysemu.h"
 #include "sysemu/hostmem.h"
@@ -3686,13 +3687,19 @@ void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev)
 
     /*
      * Tell QAPI that something happened and the memory
-     * hotunplug wasn't successful.
+     * hotunplug wasn't successful. Keep sending
+     * MEM_UNPLUG_ERROR even while sending DEVICE_UNPLUG_ERROR
+     * until the deprecation MEM_UNPLUG_ERROR is due.
      */
     if (dev->id) {

hw/ppc/spapr.c void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev) · line 3704

Quoted 10 lines
hw/ppc/spapr.c +4 -0
         qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
                                      "for device %s", dev->id);
         qapi_event_send_mem_unplug_error(dev->id, qapi_error);
     }
+
+    qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                        dev->canonical_path,
+                                        qapi_error != NULL, qapi_error);
 }
 

When dev->id is null, we send something like

    {"event": "DEVICE_UNPLUG_ERROR",
     "data": {"path": "/machine/..."},
     "timestamp": ...}

Unless I'm missing something, this is all the information the management application really needs.

When dev->id is non-null, we add to "data":

"device": "dev123",
"msg": "Memory hotunplug rejected by the guest for device dev123",

I'm fine with emitting the device ID when we have it.

What's the intended use of "msg"?

Could DEVICE_UNPLUG_ERROR ever be emitted for this device with a different "msg"?

If "msg" is useful when dev->id is non-null, then it's likely useful when dev->id is null. Why not

"msg": "Memory hotunplug rejected by the guest",

always?

If we do that here, we'll likely do it everywhere, and then member @msg isn't actually optional.

Quoted 15 lines
hw/ppc/spapr.c +0 -0
 /* Callback to be called during DRC release. */
hw/ppc/spapr_drc.c +2 -0
index a4d9496f76..8f0479631f 100644
@@ -17,6 +17,8 @@
 #include "hw/ppc/spapr_drc.h"
 #include "qom/object.h"
 #include "migration/vmstate.h"
+#include "qapi/error.h"
+#include "qapi/qapi-events-qdev.h"
 #include "qapi/visitor.h"
 #include "qemu/error-report.h"
 #include "hw/ppc/spapr.h" /* for RTAS return codes */

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 165

Quoted 4 lines
hw/ppc/spapr_drc.c +1 -0
@@ -160,6 +162,11 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
          * means that the kernel is refusing the removal.
          */
         if (drc->unplug_requested && drc->dev) {
+            const char qapi_error_fmt[] = \

Drop the superfluous \

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 166

Quoted 1 line
hw/ppc/spapr_drc.c +1 -0
+"Device hotunplug rejected by the guest for device %s";

Unusual indentation.

Quoted 7 lines
hw/ppc/spapr_drc.c +3 -0
+
+            g_autofree char *qapi_error = NULL;
+
             if (spapr_drc_type(drc) == SPAPR_DR_CONNECTOR_TYPE_LMB) {
                 spapr = SPAPR_MACHINE(qdev_get_machine());
 

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 180

Quoted 7 lines
hw/ppc/spapr_drc.c +2 -2
@@ -169,14 +176,13 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
             drc->unplug_requested = false;
 
             if (drc->dev->id) {
-                error_report("Device hotunplug rejected by the guest "
-                             "for device %s", drc->dev->id);
+                qapi_error = g_strdup_printf(qapi_error_fmt, drc->dev->id);
+                error_report(qapi_error_fmt, drc->dev->id);

Simpler:

                   qapi_error = ...
                   error_report("%s", qapi_error);

Matter of taste. Maintainer decides.

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 185

Quoted 9 lines
hw/ppc/spapr_drc.c +3 -4
             }
 
-            /*
-             * TODO: send a QAPI DEVICE_UNPLUG_ERROR event when
-             * it is implemented.
-             */
+            qapi_event_send_device_unplug_error(!!drc->dev->id, drc->dev->id,
+                                                drc->dev->canonical_path,
+                                                qapi_error != NULL, qapi_error);

My questions on "msg" apply.

Quoted 3 lines
hw/ppc/spapr_drc.c +0 -0
         }
 
         return RTAS_OUT_SUCCESS; /* Nothing to do */
#15 lore raw

Re: [PATCH for-6.2 v6 6/7] spapr: use DEVICE_UNPLUG_ERROR to report unplug errors

Quoted 67 lines · Markus Armbruster
Quoted 40 lines · Daniel Henrique Barboza

Linux Kernel 5.12 is now unisolating CPU DRCs in the device_removal error path, signalling that the hotunplug process wasn't successful. This allow us to send a DEVICE_UNPLUG_ERROR in drc_unisolate_logical() to signal this error to the management layer.

We also have another error path in spapr_memory_unplug_rollback() for configured LMB DRCs. Kernels older than 5.13 will not unisolate the LMBs in the hotunplug error path, but it will reconfigure them. Let's send the DEVICE_UNPLUG_ERROR event in that code path as well to cover the case of older kernels.

---
  hw/ppc/spapr.c     |  9 ++++++++-
  hw/ppc/spapr_drc.c | 18 ++++++++++++------
  2 files changed, 20 insertions(+), 7 deletions(-)
hw/ppc/spapr.c +4 -1
index 1611d7ab05..5459f9a7e9 100644
@@ -29,6 +29,7 @@
  #include "qemu/datadir.h"
  #include "qapi/error.h"
  #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
  #include "qapi/visitor.h"
  #include "sysemu/sysemu.h"
  #include "sysemu/hostmem.h"
@@ -3686,13 +3687,19 @@ void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev)
 
      /*
       * Tell QAPI that something happened and the memory
-     * hotunplug wasn't successful.
+     * hotunplug wasn't successful. Keep sending
+     * MEM_UNPLUG_ERROR even while sending DEVICE_UNPLUG_ERROR
+     * until the deprecation MEM_UNPLUG_ERROR is due.
       */
      if (dev->id) {

hw/ppc/spapr.c void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev) · line 3704

Quoted 10 lines
hw/ppc/spapr.c +4 -0
          qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
                                       "for device %s", dev->id);
          qapi_event_send_mem_unplug_error(dev->id, qapi_error);
      }
+
+    qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                        dev->canonical_path,
+                                        qapi_error != NULL, qapi_error);
  }
 

When dev->id is null, we send something like

     {"event": "DEVICE_UNPLUG_ERROR",
      "data": {"path": "/machine/..."},
      "timestamp": ...}

Unless I'm missing something, this is all the information the management application really needs.

When dev->id is non-null, we add to "data":

"device": "dev123",
"msg": "Memory hotunplug rejected by the guest for device dev123",

Quoted 6 lines

I'm fine with emitting the device ID when we have it.

What's the intended use of "msg"?

Could DEVICE_UNPLUG_ERROR ever be emitted for this device with a different "msg"?

It won't have a different 'msg' for the current use of the event in both ppc64 and x86. It'll always be the same '<dev> hotunplug rejected by the guest' message.

The idea is that a future caller might want to insert a more informative message, such as "hotunplug failed: memory is being used by kernel space" or any other more specific condition. But then I guess we can argue that, if that time comes, one can just add this new optional 'msg' member in this event, and for now we can live without it.

Would you oppose to renaming this new event to "DEVICE_UNPLUG_GUEST_ERROR" and then remove the 'msg' member? I guess this rename would make it clearer for management that we're reporting a guest side error, making any further clarifications via 'msg' unneeded.

Thanks,

Daniel

Quoted 75 lines

If "msg" is useful when dev->id is non-null, then it's likely useful when dev->id is null. Why not

"msg": "Memory hotunplug rejected by the guest",

always?

If we do that here, we'll likely do it everywhere, and then member @msg isn't actually optional.

Quoted 15 lines
hw/ppc/spapr.c +0 -0
  /* Callback to be called during DRC release. */
hw/ppc/spapr_drc.c +2 -0
index a4d9496f76..8f0479631f 100644
@@ -17,6 +17,8 @@
  #include "hw/ppc/spapr_drc.h"
  #include "qom/object.h"
  #include "migration/vmstate.h"
+#include "qapi/error.h"
+#include "qapi/qapi-events-qdev.h"
  #include "qapi/visitor.h"
  #include "qemu/error-report.h"
  #include "hw/ppc/spapr.h" /* for RTAS return codes */

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 165

Quoted 4 lines
hw/ppc/spapr_drc.c +1 -0
@@ -160,6 +162,11 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
           * means that the kernel is refusing the removal.
           */
          if (drc->unplug_requested && drc->dev) {
+            const char qapi_error_fmt[] = \

Drop the superfluous \

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 166

Quoted 1 line
hw/ppc/spapr_drc.c +1 -0
+"Device hotunplug rejected by the guest for device %s";

Unusual indentation.

Quoted 7 lines
hw/ppc/spapr_drc.c +3 -0
+
+            g_autofree char *qapi_error = NULL;
+
              if (spapr_drc_type(drc) == SPAPR_DR_CONNECTOR_TYPE_LMB) {
                  spapr = SPAPR_MACHINE(qdev_get_machine());
 

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 180

Quoted 7 lines
hw/ppc/spapr_drc.c +2 -2
@@ -169,14 +176,13 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
              drc->unplug_requested = false;
 
              if (drc->dev->id) {
-                error_report("Device hotunplug rejected by the guest "
-                             "for device %s", drc->dev->id);
+                qapi_error = g_strdup_printf(qapi_error_fmt, drc->dev->id);
+                error_report(qapi_error_fmt, drc->dev->id);

Simpler:

                    qapi_error = ...
                    error_report("%s", qapi_error);

Matter of taste. Maintainer decides.

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 185

Quoted 9 lines
hw/ppc/spapr_drc.c +3 -4
              }
 
-            /*
-             * TODO: send a QAPI DEVICE_UNPLUG_ERROR event when
-             * it is implemented.
-             */
+            qapi_event_send_device_unplug_error(!!drc->dev->id, drc->dev->id,
+                                                drc->dev->canonical_path,
+                                                qapi_error != NULL, qapi_error);

My questions on "msg" apply.

Quoted 3 lines
hw/ppc/spapr_drc.c +0 -0
          }
 
          return RTAS_OUT_SUCCESS; /* Nothing to do */
#16 lore raw

Re: [PATCH for-6.2 v6 6/7] spapr: use DEVICE_UNPLUG_ERROR to report unplug errors

Quoted 80 lines · Daniel Henrique Barboza
Quoted 65 lines · Markus Armbruster
Quoted 38 lines · Daniel Henrique Barboza

Linux Kernel 5.12 is now unisolating CPU DRCs in the device_removal error path, signalling that the hotunplug process wasn't successful. This allow us to send a DEVICE_UNPLUG_ERROR in drc_unisolate_logical() to signal this error to the management layer.

We also have another error path in spapr_memory_unplug_rollback() for configured LMB DRCs. Kernels older than 5.13 will not unisolate the LMBs in the hotunplug error path, but it will reconfigure them. Let's send the DEVICE_UNPLUG_ERROR event in that code path as well to cover the case of older kernels.

---
  hw/ppc/spapr.c     |  9 ++++++++-
  hw/ppc/spapr_drc.c | 18 ++++++++++++------
  2 files changed, 20 insertions(+), 7 deletions(-)
hw/ppc/spapr.c +4 -1
index 1611d7ab05..5459f9a7e9 100644
@@ -29,6 +29,7 @@
  #include "qemu/datadir.h"
  #include "qapi/error.h"
  #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
  #include "qapi/visitor.h"
  #include "sysemu/sysemu.h"
  #include "sysemu/hostmem.h"
@@ -3686,13 +3687,19 @@ void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev)
      /*
       * Tell QAPI that something happened and the memory
-     * hotunplug wasn't successful.
+     * hotunplug wasn't successful. Keep sending
+     * MEM_UNPLUG_ERROR even while sending DEVICE_UNPLUG_ERROR
+     * until the deprecation MEM_UNPLUG_ERROR is due.
       */

hw/ppc/spapr.c void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev) · line 3702

Quoted 10 lines
hw/ppc/spapr.c +4 -0
      if (dev->id) {
          qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
                                       "for device %s", dev->id);
          qapi_event_send_mem_unplug_error(dev->id, qapi_error);
      }
+
+    qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                        dev->canonical_path,
+                                        qapi_error != NULL, qapi_error);
  }

When dev->id is null, we send something like

     {"event": "DEVICE_UNPLUG_ERROR",
      "data": {"path": "/machine/..."},
      "timestamp": ...}

Unless I'm missing something, this is all the information the management application really needs.

When dev->id is non-null, we add to "data":

"device": "dev123",
"msg": "Memory hotunplug rejected by the guest for device dev123",

Quoted 6 lines

I'm fine with emitting the device ID when we have it.

What's the intended use of "msg"?

Could DEVICE_UNPLUG_ERROR ever be emitted for this device with a different "msg"?

It won't have a different 'msg' for the current use of the event in both ppc64 and x86. It'll always be the same '<dev> hotunplug rejected by the guest' message.

Quoted 5 lines

The idea is that a future caller might want to insert a more informative message, such as "hotunplug failed: memory is being used by kernel space" or any other more specific condition. But then I guess we can argue that, if that time comes, one can just add this new optional 'msg' member in this event, and for now we can live without it.

Right. We could also consider making the current message more specific about why we chose to cancel the unplug: e.g. "guest unisolated DRC after unplug request" for PAPR, and something appropriate to the ACPI specifics for x86. Not sure if that's useful enough to justify it.

Quoted 86 lines

Would you oppose to renaming this new event to "DEVICE_UNPLUG_GUEST_ERROR" and then remove the 'msg' member? I guess this rename would make it clearer for management that we're reporting a guest side error, making any further clarifications via 'msg' unneeded.

Thanks,

Daniel

Quoted 71 lines

If "msg" is useful when dev->id is non-null, then it's likely useful when dev->id is null. Why not

"msg": "Memory hotunplug rejected by the guest",

always?

If we do that here, we'll likely do it everywhere, and then member @msg isn't actually optional.

Quoted 15 lines
hw/ppc/spapr.c +0 -0
  /* Callback to be called during DRC release. */
hw/ppc/spapr_drc.c +2 -0
index a4d9496f76..8f0479631f 100644
@@ -17,6 +17,8 @@
  #include "hw/ppc/spapr_drc.h"
  #include "qom/object.h"
  #include "migration/vmstate.h"
+#include "qapi/error.h"
+#include "qapi/qapi-events-qdev.h"
  #include "qapi/visitor.h"
  #include "qemu/error-report.h"
  #include "hw/ppc/spapr.h" /* for RTAS return codes */

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 165

Quoted 4 lines
hw/ppc/spapr_drc.c +1 -0
@@ -160,6 +162,11 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
           * means that the kernel is refusing the removal.
           */
          if (drc->unplug_requested && drc->dev) {
+            const char qapi_error_fmt[] = \

Drop the superfluous \

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 166

Quoted 1 line
hw/ppc/spapr_drc.c +1 -0
+"Device hotunplug rejected by the guest for device %s";

Unusual indentation.

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 179

Quoted 12 lines
hw/ppc/spapr_drc.c +5 -2
+
+            g_autofree char *qapi_error = NULL;
+
              if (spapr_drc_type(drc) == SPAPR_DR_CONNECTOR_TYPE_LMB) {
                  spapr = SPAPR_MACHINE(qdev_get_machine());
@@ -169,14 +176,13 @@ static uint32_t drc_unisolate_logical(SpaprDrc *drc)
              drc->unplug_requested = false;
              if (drc->dev->id) {
-                error_report("Device hotunplug rejected by the guest "
-                             "for device %s", drc->dev->id);
+                qapi_error = g_strdup_printf(qapi_error_fmt, drc->dev->id);
+                error_report(qapi_error_fmt, drc->dev->id);

Simpler:

                    qapi_error = ...
                    error_report("%s", qapi_error);

Matter of taste. Maintainer decides.

hw/ppc/spapr_drc.c static uint32_t drc_unisolate_logical(SpaprDrc *drc) · line 185

Quoted 8 lines
hw/ppc/spapr_drc.c +3 -4
              }
-            /*
-             * TODO: send a QAPI DEVICE_UNPLUG_ERROR event when
-             * it is implemented.
-             */
+            qapi_event_send_device_unplug_error(!!drc->dev->id, drc->dev->id,
+                                                drc->dev->canonical_path,
+                                                qapi_error != NULL, qapi_error);

My questions on "msg" apply.

Quoted 2 lines
hw/ppc/spapr_drc.c +0 -0
          }
          return RTAS_OUT_SUCCESS; /* Nothing to do */
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#17 lore raw

Re: [PATCH for-6.2 v6 6/7] spapr: use DEVICE_UNPLUG_ERROR to report unplug errors

In reply to Daniel Henrique Barboza
Quoted 84 lines · Daniel Henrique Barboza
Quoted 65 lines · Markus Armbruster
Quoted 39 lines · Daniel Henrique Barboza

Linux Kernel 5.12 is now unisolating CPU DRCs in the device_removal error path, signalling that the hotunplug process wasn't successful. This allow us to send a DEVICE_UNPLUG_ERROR in drc_unisolate_logical() to signal this error to the management layer.

We also have another error path in spapr_memory_unplug_rollback() for configured LMB DRCs. Kernels older than 5.13 will not unisolate the LMBs in the hotunplug error path, but it will reconfigure them. Let's send the DEVICE_UNPLUG_ERROR event in that code path as well to cover the case of older kernels.

---
  hw/ppc/spapr.c     |  9 ++++++++-
  hw/ppc/spapr_drc.c | 18 ++++++++++++------
  2 files changed, 20 insertions(+), 7 deletions(-)
hw/ppc/spapr.c +4 -1
index 1611d7ab05..5459f9a7e9 100644
@@ -29,6 +29,7 @@
  #include "qemu/datadir.h"
  #include "qapi/error.h"
  #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
  #include "qapi/visitor.h"
  #include "sysemu/sysemu.h"
  #include "sysemu/hostmem.h"
@@ -3686,13 +3687,19 @@ void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev)
        /*
       * Tell QAPI that something happened and the memory
-     * hotunplug wasn't successful.
+     * hotunplug wasn't successful. Keep sending
+     * MEM_UNPLUG_ERROR even while sending DEVICE_UNPLUG_ERROR
+     * until the deprecation MEM_UNPLUG_ERROR is due.
       */
      if (dev->id) {

hw/ppc/spapr.c void spapr_memory_unplug_rollback(SpaprMachineState *spapr, DeviceState *dev) · line 3703

Quoted 10 lines
hw/ppc/spapr.c +4 -0
          qapi_error = g_strdup_printf("Memory hotunplug rejected by the guest "
                                       "for device %s", dev->id);
          qapi_event_send_mem_unplug_error(dev->id, qapi_error);
      }
+
+    qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                        dev->canonical_path,
+                                        qapi_error != NULL, qapi_error);
  }
 

When dev->id is null, we send something like

     {"event": "DEVICE_UNPLUG_ERROR",
      "data": {"path": "/machine/..."},
      "timestamp": ...}

Unless I'm missing something, this is all the information the management application really needs.

When dev->id is non-null, we add to "data":

"device": "dev123",
"msg": "Memory hotunplug rejected by the guest for device dev123",

Quoted 6 lines

I'm fine with emitting the device ID when we have it.

What's the intended use of "msg"?

Could DEVICE_UNPLUG_ERROR ever be emitted for this device with a different "msg"?

It won't have a different 'msg' for the current use of the event in both ppc64 and x86. It'll always be the same '<dev> hotunplug rejected by the guest' message.

The idea is that a future caller might want to insert a more informative message, such as "hotunplug failed: memory is being used by kernel space" or any other more specific condition. But then I guess we can argue that, if that time comes, one can just add this new optional 'msg' member in this event, and for now we can live without it.

Quoted 4 lines

Would you oppose to renaming this new event to "DEVICE_UNPLUG_GUEST_ERROR" and then remove the 'msg' member? I guess this rename would make it clearer for management that we're reporting a guest side error, making any further clarifications via 'msg' unneeded.

No objection.

#18 lore raw

PATCH v6 7/7 memory_hotplug.c: send DEVICE_UNPLUG_ERROR in acpi_memory_hotplug_write()

In reply to Daniel Henrique Barboza

MEM_UNPLUG_ERROR is deprecated since the introduction of DEVICE_UNPLUG_ERROR. Keep emitting both while the deprecation of MEM_UNPLUG_ERROR is pending.

---
 hw/acpi/memory_hotplug.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
hw/acpi/memory_hotplug.c +10 -0
index e37acb0367..a0772fe083 100644
@@ -8,6 +8,7 @@
 #include "qapi/error.h"
 #include "qapi/qapi-events-acpi.h"
 #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
 
 #define MEMORY_SLOTS_NUMBER          "MDNR"
 #define MEMORY_HOTPLUG_IO_REGION     "HPMR"
@@ -181,10 +182,19 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
 
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
 
+                /*
+                 * Send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
+                 * while the deprecation of MEM_UNPLUG_ERROR is
+                 * pending.
+                 */
                 if (dev->id) {
                     qapi_event_send_mem_unplug_error(dev->id, error_pretty);
                 }
 
+                qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                                    dev->canonical_path,
+                                                    true, error_pretty);
+
                 error_free(local_err);
                 break;
             }
Signature · 1 line
2.31.1
Acked-by on v6 7/7
#19 lore raw

Re: [PATCH for-6.2 v6 7/7] memory_hotplug.c: send DEVICE_UNPLUG_ERROR in acpi_memory_hotplug_write()

Quoted 8 lines · Daniel Henrique Barboza

MEM_UNPLUG_ERROR is deprecated since the introduction of DEVICE_UNPLUG_ERROR. Keep emitting both while the deprecation of MEM_UNPLUG_ERROR is pending.

Pls merge with rest of series.

Quoted 38 lines
---
 hw/acpi/memory_hotplug.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
hw/acpi/memory_hotplug.c +10 -0
index e37acb0367..a0772fe083 100644
@@ -8,6 +8,7 @@
 #include "qapi/error.h"
 #include "qapi/qapi-events-acpi.h"
 #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
 
 #define MEMORY_SLOTS_NUMBER          "MDNR"
 #define MEMORY_HOTPLUG_IO_REGION     "HPMR"
@@ -181,10 +182,19 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
 
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
 
+                /*
+                 * Send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
+                 * while the deprecation of MEM_UNPLUG_ERROR is
+                 * pending.
+                 */
                 if (dev->id) {
                     qapi_event_send_mem_unplug_error(dev->id, error_pretty);
                 }
 
+                qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                                    dev->canonical_path,
+                                                    true, error_pretty);
+
                 error_free(local_err);
                 break;
             }
Signature · 1 line
2.31.1
#20 lore raw

Re: [PATCH for-6.2 v6 7/7] memory_hotplug.c: send DEVICE_UNPLUG_ERROR in acpi_memory_hotplug_write()

Quoted 11 lines · Michael S. Tsirkin
Quoted 8 lines · Daniel Henrique Barboza

MEM_UNPLUG_ERROR is deprecated since the introduction of DEVICE_UNPLUG_ERROR. Keep emitting both while the deprecation of MEM_UNPLUG_ERROR is pending.

Michael, can I assume I have your ack for 1/7 as well?

Quoted 39 lines

Pls merge with rest of series.

Quoted 36 lines
---
 hw/acpi/memory_hotplug.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
hw/acpi/memory_hotplug.c +10 -0
index e37acb0367..a0772fe083 100644
@@ -8,6 +8,7 @@
 #include "qapi/error.h"
 #include "qapi/qapi-events-acpi.h"
 #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
 
 #define MEMORY_SLOTS_NUMBER          "MDNR"
 #define MEMORY_HOTPLUG_IO_REGION     "HPMR"
@@ -181,10 +182,19 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
 
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
 
+                /*
+                 * Send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
+                 * while the deprecation of MEM_UNPLUG_ERROR is
+                 * pending.
+                 */
                 if (dev->id) {
                     qapi_event_send_mem_unplug_error(dev->id, error_pretty);
                 }
 
+                qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                                    dev->canonical_path,
+                                                    true, error_pretty);
+
                 error_free(local_err);
                 break;
             }
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
Reviewed-by on v6 7/7
#21 lore raw

Re: [PATCH for-6.2 v6 7/7] memory_hotplug.c: send DEVICE_UNPLUG_ERROR in acpi_memory_hotplug_write()

In reply to Daniel Henrique Barboza
Quoted 34 lines · Daniel Henrique Barboza

MEM_UNPLUG_ERROR is deprecated since the introduction of DEVICE_UNPLUG_ERROR. Keep emitting both while the deprecation of MEM_UNPLUG_ERROR is pending.

---
 hw/acpi/memory_hotplug.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
hw/acpi/memory_hotplug.c +6 -0
index e37acb0367..a0772fe083 100644
@@ -8,6 +8,7 @@
 #include "qapi/error.h"
 #include "qapi/qapi-events-acpi.h"
 #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
 
 #define MEMORY_SLOTS_NUMBER          "MDNR"
 #define MEMORY_HOTPLUG_IO_REGION     "HPMR"
@@ -181,10 +182,19 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
 
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
 
+                /*
+                 * Send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
+                 * while the deprecation of MEM_UNPLUG_ERROR is
+                 * pending.
+                 */
                 if (dev->id) {

hw/acpi/memory_hotplug.c static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data, · line 200

Quoted 10 lines
hw/acpi/memory_hotplug.c +4 -0
                     qapi_event_send_mem_unplug_error(dev->id, error_pretty);
                 }
 
+                qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                                    dev->canonical_path,
+                                                    true, error_pretty);
+
                 error_free(local_err);
                 break;
             }

Unlike the previous patch, "msg" is present even when !dev->id. Makes me doubt the previous patch's conditional passing of "msg" some more.

#22 lore raw

Re: [PATCH for-6.2 v6 7/7] memory_hotplug.c: send DEVICE_UNPLUG_ERROR in acpi_memory_hotplug_write()

Quoted 39 lines · Markus Armbruster
Quoted 37 lines · Daniel Henrique Barboza

MEM_UNPLUG_ERROR is deprecated since the introduction of DEVICE_UNPLUG_ERROR. Keep emitting both while the deprecation of MEM_UNPLUG_ERROR is pending.

---
 hw/acpi/memory_hotplug.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
hw/acpi/memory_hotplug.c +6 -0
index e37acb0367..a0772fe083 100644
@@ -8,6 +8,7 @@
 #include "qapi/error.h"
 #include "qapi/qapi-events-acpi.h"
 #include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-qdev.h"
 
 #define MEMORY_SLOTS_NUMBER          "MDNR"
 #define MEMORY_HOTPLUG_IO_REGION     "HPMR"
@@ -181,10 +182,19 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data,
 
                 trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
 
+                /*
+                 * Send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
+                 * while the deprecation of MEM_UNPLUG_ERROR is
+                 * pending.
+                 */
                 if (dev->id) {
                     qapi_event_send_mem_unplug_error(dev->id, error_pretty);
                 }
 
Quoted 10 lines

hw/acpi/memory_hotplug.c static void acpi_memory_hotplug_write(void *opaque, hwaddr addr, uint64_t data, · line 200

Quoted 7 lines
hw/acpi/memory_hotplug.c +4 -0
+                qapi_event_send_device_unplug_error(!!dev->id, dev->id,
+                                                    dev->canonical_path,
+                                                    true, error_pretty);
+
                 error_free(local_err);
                 break;
             }

Unlike the previous patch, "msg" is present even when !dev->id. Makes me doubt the previous patch's conditional passing of "msg" some more.

Daniel, if you can address Markus' comments and send another spin, I'll replace the draft I have in ppc-for-6.2 with the new version.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#23 lore raw
In reply to Daniel Henrique Barboza
Quoted 4 lines · Daniel Henrique Barboza

Hi,

This version was rebased with QEMU master @ 7457b407edd6e8555e4b4 and has a few simple changes based on Greg's review.

Tentatively applied to ppc-for-6.2.

Quoted 79 lines

changes from v5:
- patch 1:
* fixed function name in commit msg
-patch 6:
* kept error message in a single line
- all patches:
* added Greg's R-b
- v5 link: https://lists.gnu.org/archive/html/qemu-devel/2021-07/msg03239.html

changes from v4:
- patch 1 (new):
* avoid emitting MEM_UNPLUG_ERROR when dev->id == NULL in
memory_hotplug.c
- patch 2 (new):
* avoid emitting MEM_UNPLUG_ERROR when dev->id == NULL in
spapr.c
- patch 3 (new):
* do not error_report() when drc->dev->id == NULL
- patch 4 (new):
* update DEVICE_DELETED API docs
- patch 5 (former 1):
* added 2 spaces after each sentence
* appended "Some errors cannot be detected." in DEVICE_UNPLUG_ERROR
docs
* changed release from 6.1 to 6.2 in docs
* changed DEVICE_UNPLUG_ERROR API to behave more like device_deleted
- patches 6 and 7:
* changed to use the new DEVICE_UNPLUG_ERROR API
- v4 link: https://lists.gnu.org/archive/html/qemu-devel/2021-07/msg01355.html

changes from v3:
- patch 1:
* fixed format
- all patches:
* rebased with master
* added David's R-b
- v3 link: https://lists.gnu.org/archive/html/qemu-devel/2021-06/msg05842.html

changes from v2:
- patch 1:
* moved DEVICE_UNPLUG_ERROR declaration to qapi/qdev.json
* updated 'device_del' description
* added 'deprecated' notice on MEM_UNPLUG_ERROR
* added MEM_UNPLUG_ERROR 'deprecated' info in docs/system/deprecated.rst
- patch 2:
* send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
- patch 3 (new):
* send DEVICE_UNPLUG_ERROR in acpi/memory_hotplug.c
- v2 link: https://lists.gnu.org/archive/html/qemu-devel/2021-06/msg01304.html

changes from v1:
- former patches 1 and 2: dropped
- patch 1 (former 3): changed the version to '6.1'
- patch 2 (former 4): add a DEVICE_UNPLUG_ERROR event in the device unplug error path of CPUs and DIMMs
- v1 link: https://lists.gnu.org/archive/html/qemu-devel/2021-03/msg04682.html

Daniel Henrique Barboza (7):
hw/acpi/memory_hotplug.c: avoid sending MEM_UNPLUG_ERROR if dev->id is
NULL
spapr.c: avoid sending MEM_UNPLUG_ERROR if dev->id is NULL spapr_drc.c: do not error_report() when drc->dev->id == NULL qapi/qdev.json: fix DEVICE_DELETED parameters doc qapi/qdev.json: add DEVICE_UNPLUG_ERROR QAPI event
spapr: use DEVICE_UNPLUG_ERROR to report unplug errors memory_hotplug.c: send DEVICE_UNPLUG_ERROR in
acpi_memory_hotplug_write()

 docs/about/deprecated.rst | 10 ++++++++++
 hw/acpi/memory_hotplug.c  | 19 +++++++++++++++++--
 hw/ppc/spapr.c            | 17 +++++++++++++----
 hw/ppc/spapr_drc.c        | 21 +++++++++++++++------
 qapi/machine.json         |  6 +++++-
 qapi/qdev.json            | 34 +++++++++++++++++++++++++++++++---
 stubs/qdev.c              |  7 +++++++
 7 files changed, 98 insertions(+), 16 deletions(-)
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson