* [PATCH V15 01/14] qapi/misc: Fix missed query-iothreads items
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 7:24 ` Michael Tokarev
2026-09-07 6:37 ` [PATCH V15 02/14] iothread: introduce holder tracking Zhang Chen
` (12 subsequent siblings)
13 siblings, 1 reply; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen, qemu-stable
The example is incomplete: it misses members @poll-max-ns, @poll-grow,
@poll-shrink, @aio-max-batch. Messed up in commit 5fc00480ab1
(monitor: add poll-* properties into query-iothreads result) and
commit 1793ad0247c (iothread: add aio-max-batch parameter).
cc: qemu-stable@nongnu.org
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
qapi/misc.json | 12 ++++++++++--
1 file changed, 10 insertions(+), 2 deletions(-)
diff --git a/qapi/misc.json b/qapi/misc.json
index a21ba61b96..baa50cc844 100644
--- a/qapi/misc.json
+++ b/qapi/misc.json
@@ -123,11 +123,19 @@
# <- { "return": [
# {
# "id":"iothread0",
-# "thread-id":3134
+# "thread-id":3134,
+# "poll-max-ns":32768,
+# "poll-grow":0,
+# "poll-shrink":0,
+# "aio-max-batch":0
# },
# {
# "id":"iothread1",
-# "thread-id":3135
+# "thread-id":3135,
+# "poll-max-ns":32768,
+# "poll-grow":0,
+# "poll-shrink":0,
+# "aio-max-batch":0
# }
# ]
# }
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* Re: [PATCH V15 01/14] qapi/misc: Fix missed query-iothreads items
2026-09-07 6:37 ` [PATCH V15 01/14] qapi/misc: Fix missed query-iothreads items Zhang Chen
@ 2026-09-07 7:24 ` Michael Tokarev
2026-09-10 5:28 ` Zhang Chen
0 siblings, 1 reply; 26+ messages in thread
From: Michael Tokarev @ 2026-09-07 7:24 UTC (permalink / raw)
To: Zhang Chen, qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: qemu-stable
On 9/7/26 09:37, Zhang Chen wrote:
> The example is incomplete: it misses members @poll-max-ns, @poll-grow,
> @poll-shrink, @aio-max-batch. Messed up in commit 5fc00480ab1
> (monitor: add poll-* properties into query-iothreads result) and
> commit 1793ad0247c (iothread: add aio-max-batch parameter).
>
> cc: qemu-stable@nongnu.org
>
> Reviewed-by: Markus Armbruster <armbru@redhat.com>
> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
> Signed-off-by: Zhang Chen <zhangckid@gmail.com>
This is not technically a part of "iothread: Support tracking and querying
IOThread holder" series, it's a bugfix by its own. Can we apply this one
without the rest? I can do it through the trivial-patches tree but it
looks like it should go to the qapi tree.
Thanks,
/mjt
^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [PATCH V15 01/14] qapi/misc: Fix missed query-iothreads items
2026-09-07 7:24 ` Michael Tokarev
@ 2026-09-10 5:28 ` Zhang Chen
2026-09-10 6:01 ` Markus Armbruster
0 siblings, 1 reply; 26+ messages in thread
From: Zhang Chen @ 2026-09-10 5:28 UTC (permalink / raw)
To: Michael Tokarev
Cc: qemu-devel, Eric Blake, Markus Armbruster, Michael S . Tsirkin,
Stefan Hajnoczi, Paolo Bonzini, Daniel P . Berrangé,
Jason Wang, qemu-stable
On Mon, Sep 7, 2026 at 3:24 PM Michael Tokarev <mjt@tls.msk.ru> wrote:
>
> On 9/7/26 09:37, Zhang Chen wrote:
> > The example is incomplete: it misses members @poll-max-ns, @poll-grow,
> > @poll-shrink, @aio-max-batch. Messed up in commit 5fc00480ab1
> > (monitor: add poll-* properties into query-iothreads result) and
> > commit 1793ad0247c (iothread: add aio-max-batch parameter).
> >
> > cc: qemu-stable@nongnu.org
> >
> > Reviewed-by: Markus Armbruster <armbru@redhat.com>
> > Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
> > Signed-off-by: Zhang Chen <zhangckid@gmail.com>
>
>
> This is not technically a part of "iothread: Support tracking and querying
> IOThread holder" series, it's a bugfix by its own. Can we apply this one
> without the rest? I can do it through the trivial-patches tree but it
> looks like it should go to the qapi tree.
It's OK for me. It looks like the qapi maintainer isn't responding.
Please apply to the trivial-patches tree and I will drop this patch in
next version.
Thanks
Chen
>
> Thanks,
>
> /mjt
^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [PATCH V15 01/14] qapi/misc: Fix missed query-iothreads items
2026-09-10 5:28 ` Zhang Chen
@ 2026-09-10 6:01 ` Markus Armbruster
0 siblings, 0 replies; 26+ messages in thread
From: Markus Armbruster @ 2026-09-10 6:01 UTC (permalink / raw)
To: Zhang Chen
Cc: Michael Tokarev, qemu-devel, Eric Blake, Michael S . Tsirkin,
Stefan Hajnoczi, Paolo Bonzini, Daniel P . Berrangé,
Jason Wang, qemu-stable
Zhang Chen <zhangckid@gmail.com> writes:
> On Mon, Sep 7, 2026 at 3:24 PM Michael Tokarev <mjt@tls.msk.ru> wrote:
>>
>> On 9/7/26 09:37, Zhang Chen wrote:
>> > The example is incomplete: it misses members @poll-max-ns, @poll-grow,
>> > @poll-shrink, @aio-max-batch. Messed up in commit 5fc00480ab1
>> > (monitor: add poll-* properties into query-iothreads result) and
>> > commit 1793ad0247c (iothread: add aio-max-batch parameter).
>> >
>> > cc: qemu-stable@nongnu.org
>> >
>> > Reviewed-by: Markus Armbruster <armbru@redhat.com>
>> > Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
>> > Signed-off-by: Zhang Chen <zhangckid@gmail.com>
>>
>>
>> This is not technically a part of "iothread: Support tracking and querying
>> IOThread holder" series, it's a bugfix by its own. Can we apply this one
>> without the rest? I can do it through the trivial-patches tree but it
>> looks like it should go to the qapi tree.
>
> It's OK for me. It looks like the qapi maintainer isn't responding.
> Please apply to the trivial-patches tree and I will drop this patch in
> next version.
I've been responding :) There's nothing else for QAPI at the moment, so
please take it through qemu-trivial. Thanks!
^ permalink raw reply [flat|nested] 26+ messages in thread
* [PATCH V15 02/14] iothread: introduce holder tracking
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
2026-09-07 6:37 ` [PATCH V15 01/14] qapi/misc: Fix missed query-iothreads items Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 03/14] iothread: add holder-aware AioContext accessors Zhang Chen
` (11 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
IOThreads do not record the QOM objects and block exports that use
them, making runtime introspection and hotplug debugging difficult.
Add:
- holders: a list describing what is holding an iothread
- iothread_ref(): add a holder to the list
- iothread_unref(): remove a holder from the list
A holder can identify either a QOM object or a block export. A later
patch will expose this information through query-iothreads.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
include/system/iothread.h | 5 +++
iothread.c | 63 ++++++++++++++++++++++++++
qapi/misc.json | 94 ++++++++++++++++++++++++++++++++++-----
3 files changed, 150 insertions(+), 12 deletions(-)
diff --git a/include/system/iothread.h b/include/system/iothread.h
index a1ef7696cb..b8aeb32b0e 100644
--- a/include/system/iothread.h
+++ b/include/system/iothread.h
@@ -18,6 +18,7 @@
#include "qemu/thread.h"
#include "qom/object.h"
#include "system/event-loop-base.h"
+#include "qapi/qapi-types-misc.h"
#define TYPE_IOTHREAD "iothread"
@@ -50,6 +51,7 @@ struct IOThread {
bool stopping; /* has iothread_stop() been called? */
bool running; /* should iothread_run() continue? */
int thread_id;
+ IOThreadHolderList *holders;
/* AioContext poll parameters */
int64_t poll_max_ns;
@@ -82,4 +84,7 @@ void iothread_destroy(IOThread *iothread);
*/
bool qemu_in_iothread(void);
+void iothread_ref(IOThread *iothread, const IOThreadHolder *holder);
+void iothread_unref(IOThread *iothread, const IOThreadHolder *holder);
+
#endif /* IOTHREAD_H */
diff --git a/iothread.c b/iothread.c
index 3558535b40..66dfde626e 100644
--- a/iothread.c
+++ b/iothread.c
@@ -21,10 +21,71 @@
#include "system/iothread.h"
#include "qapi/error.h"
#include "qapi/qapi-commands-misc.h"
+#include "qapi/clone-visitor.h"
+#include "qapi/qapi-visit-misc.h"
#include "qemu/error-report.h"
#include "qemu/rcu.h"
#include "qemu/main-loop.h"
+/* Add a deep copy of @holder to @iothread's list of holders. */
+void iothread_ref(IOThread *iothread, const IOThreadHolder *holder)
+{
+ assert(holder);
+
+ QAPI_LIST_PREPEND(iothread->holders, QAPI_CLONE(IOThreadHolder, holder));
+}
+
+static int iothread_holder_compare(const IOThreadHolder *holder_a,
+ const IOThreadHolder *holder_b)
+{
+ const char *name_a, *name_b;
+
+ if (holder_a->type != holder_b->type) {
+ return holder_a->type - holder_b->type;
+ }
+
+ switch (holder_a->type) {
+ case IO_THREAD_HOLDER_KIND_QOM_OBJECT:
+ name_a = holder_a->u.qom_object.qom_path;
+ name_b = holder_b->u.qom_object.qom_path;
+ break;
+ case IO_THREAD_HOLDER_KIND_BLOCK_EXPORT:
+ name_a = holder_a->u.block_export.export_id;
+ name_b = holder_b->u.block_export.export_id;
+ break;
+ default:
+ g_assert_not_reached();
+ }
+
+ return strcmp(name_a, name_b);
+}
+
+/*
+ * Remove @holder from @iothread's list of holders.
+ *
+ * It is a programming error if @holder is not present.
+ */
+void iothread_unref(IOThread *iothread, const IOThreadHolder *holder)
+{
+ IOThreadHolderList **prev = &iothread->holders;
+ IOThreadHolderList *curr;
+
+ assert(holder);
+
+ while (*prev) {
+ curr = *prev;
+ if (iothread_holder_compare(curr->value, holder) == 0) {
+ *prev = curr->next;
+ curr->next = NULL;
+ qapi_free_IOThreadHolderList(curr);
+ return;
+ }
+ prev = &curr->next;
+ }
+
+ g_assert_not_reached();
+}
+
static void *iothread_run(void *opaque)
{
IOThread *iothread = opaque;
@@ -129,6 +190,7 @@ static void iothread_instance_finalize(Object *obj)
iothread->main_loop = NULL;
}
qemu_sem_destroy(&iothread->init_done_sem);
+ qapi_free_IOThreadHolderList(iothread->holders);
}
static void iothread_init_gcontext(IOThread *iothread, const char *thread_name)
@@ -373,6 +435,7 @@ static int query_one_iothread(Object *object, void *opaque)
info = g_new0(IOThreadInfo, 1);
info->id = iothread_get_id(iothread);
info->thread_id = iothread->thread_id;
+ info->holders = QAPI_CLONE(IOThreadHolderList, iothread->holders);
info->poll_max_ns = iothread->poll_max_ns;
info->poll_grow = iothread->poll_grow;
info->poll_shrink = iothread->poll_shrink;
diff --git a/qapi/misc.json b/qapi/misc.json
index baa50cc844..bb49969b4b 100644
--- a/qapi/misc.json
+++ b/qapi/misc.json
@@ -67,6 +67,55 @@
##
{ 'command': 'query-name', 'returns': 'NameInfo', 'allow-preconfig': true }
+##
+# @IOThreadHolderBlockExport:
+#
+# @export-id: The unique block export identifier.
+#
+# Since: 11.2
+#
+##
+{ 'struct': 'IOThreadHolderBlockExport',
+ 'data': { 'export-id': 'str' } }
+
+##
+# @IOThreadHolderQomObject:
+#
+# @qom-path: Path to the object in the QOM tree.
+#
+# Since: 11.2
+#
+##
+{ 'struct': 'IOThreadHolderQomObject',
+ 'data': { 'qom-path': 'str' } }
+
+##
+# @IOThreadHolderKind:
+#
+# @block-export: A block export.
+# @qom-object: A QOM object.
+#
+# Since: 11.2
+##
+{ 'enum': 'IOThreadHolderKind',
+ 'data': [ 'block-export', 'qom-object' ] }
+
+##
+# @IOThreadHolder:
+#
+# The block export or QOM object holding the iothread.
+#
+# @type: the kind of iothread holder.
+#
+# Since: 11.2
+##
+{ 'union': 'IOThreadHolder',
+ 'base': { 'type': 'IOThreadHolderKind' },
+ 'discriminator': 'type',
+ 'data': {
+ 'block-export': 'IOThreadHolderBlockExport',
+ 'qom-object': 'IOThreadHolderQomObject' } }
+
##
# @IOThreadInfo:
#
@@ -76,6 +125,10 @@
#
# @thread-id: ID of the underlying host thread
#
+# @holders: the QOM objects or block exports currently holding this
+# iothread. When a holder is detached or destroyed, it is removed
+# from this list. (Since 11.2)
+#
# @poll-max-ns: maximum polling time in ns, 0 means polling is
# disabled (since 2.9)
#
@@ -98,6 +151,7 @@
{ 'struct': 'IOThreadInfo',
'data': {'id': 'str',
'thread-id': 'int',
+ 'holders': ['IOThreadHolder'],
'poll-max-ns': 'int',
'poll-grow': 'int',
'poll-shrink': 'int',
@@ -122,20 +176,36 @@
# -> { "execute": "query-iothreads" }
# <- { "return": [
# {
-# "id":"iothread0",
-# "thread-id":3134,
-# "poll-max-ns":32768,
-# "poll-grow":0,
-# "poll-shrink":0,
-# "aio-max-batch":0
+# "id": "iothread0",
+# "thread-id": 3134,
+# "holders": [
+# {
+# "qom-path": "/machine/peripheral/blk1/virtio-backend",
+# "type": "qom-object"
+# },
+# {
+# "qom-path": "/machine/peripheral/blk2/virtio-backend",
+# "type": "qom-object"
+# }
+# ],
+# "poll-max-ns": 32768,
+# "poll-grow": 0,
+# "poll-shrink": 0,
+# "aio-max-batch": 0
# },
# {
-# "id":"iothread1",
-# "thread-id":3135,
-# "poll-max-ns":32768,
-# "poll-grow":0,
-# "poll-shrink":0,
-# "aio-max-batch":0
+# "id": "iothread1",
+# "thread-id": 3135,
+# "holders": [
+# {
+# "export-id": "export0",
+# "type": "block-export"
+# }
+# ],
+# "poll-max-ns": 32768,
+# "poll-grow": 0,
+# "poll-shrink": 0,
+# "aio-max-batch": 0
# }
# ]
# }
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 03/14] iothread: add holder-aware AioContext accessors
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
2026-09-07 6:37 ` [PATCH V15 01/14] qapi/misc: Fix missed query-iothreads items Zhang Chen
2026-09-07 6:37 ` [PATCH V15 02/14] iothread: introduce holder tracking Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 04/14] block/export: track IOThread references Zhang Chen
` (10 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Introduce iothread_ref_and_get_aio_context() with a holder argument
and its counterpart iothread_unref_and_put_aio_context().
Previously, users of an IOThread AioContext did not explicitly record
their identity, making it difficult to debug which devices or
subsystems were pinning an IOThread. These new helpers provide the
infrastructure for callers to register their identity. Subsequent
patches convert existing users to the new helpers.
Registering a holder takes an IOThread object reference so that the
IOThread and its AioContext stay alive until the matching put
operation. Document the ownership and BQL requirements.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
include/system/iothread.h | 22 ++++++++++++++++++++++
iothread.c | 24 +++++++++++++++++++++++-
2 files changed, 45 insertions(+), 1 deletion(-)
diff --git a/include/system/iothread.h b/include/system/iothread.h
index b8aeb32b0e..f6c5f95dd9 100644
--- a/include/system/iothread.h
+++ b/include/system/iothread.h
@@ -67,6 +67,28 @@ DECLARE_INSTANCE_CHECKER(IOThread, IOTHREAD,
char *iothread_get_id(IOThread *iothread);
IOThread *iothread_by_id(const char *id);
AioContext *iothread_get_aio_context(IOThread *iothread);
+
+/*
+ * Register @holder and return @iothread's AioContext. The holder is copied,
+ * and a reference is taken on @iothread so that both the IOThread and its
+ * AioContext remain alive.
+ *
+ * The caller must eventually call iothread_unref_and_put_aio_context() with an
+ * equivalent holder. This function is not thread-safe and must be called
+ * under the Big QEMU Lock (BQL).
+ */
+AioContext *iothread_ref_and_get_aio_context(IOThread *iothread,
+ const IOThreadHolder *holder);
+
+/*
+ * Unregister @holder and release the corresponding reference on @iothread.
+ * Calling this function without a matching
+ * iothread_ref_and_get_aio_context() call is a programming error.
+ *
+ * This function is not thread-safe and must be called under the BQL.
+ */
+void iothread_unref_and_put_aio_context(IOThread *iothread,
+ const IOThreadHolder *holder);
GMainContext *iothread_get_g_main_context(IOThread *iothread);
/*
diff --git a/iothread.c b/iothread.c
index 66dfde626e..58dace351c 100644
--- a/iothread.c
+++ b/iothread.c
@@ -33,6 +33,11 @@ void iothread_ref(IOThread *iothread, const IOThreadHolder *holder)
assert(holder);
QAPI_LIST_PREPEND(iothread->holders, QAPI_CLONE(IOThreadHolder, holder));
+ /*
+ * This guarantees that the IOThread and its AioContext remain alive
+ * as long as there is a holder.
+ */
+ object_ref(OBJECT(iothread));
}
static int iothread_holder_compare(const IOThreadHolder *holder_a,
@@ -78,6 +83,7 @@ void iothread_unref(IOThread *iothread, const IOThreadHolder *holder)
*prev = curr->next;
curr->next = NULL;
qapi_free_IOThreadHolderList(curr);
+ object_unref(OBJECT(iothread));
return;
}
prev = &curr->next;
@@ -199,7 +205,7 @@ static void iothread_init_gcontext(IOThread *iothread, const char *thread_name)
g_autofree char *name = g_strdup_printf("%s aio-context", thread_name);
iothread->worker_context = g_main_context_new();
- source = aio_get_g_source(iothread_get_aio_context(iothread));
+ source = aio_get_g_source(iothread->ctx);
g_source_set_name(source, name);
g_source_attach(source, iothread->worker_context);
g_source_unref(source);
@@ -421,6 +427,22 @@ AioContext *iothread_get_aio_context(IOThread *iothread)
return iothread->ctx;
}
+AioContext *iothread_ref_and_get_aio_context(IOThread *iothread,
+ const IOThreadHolder *holder)
+{
+ /* Add IOThreadHolder to the list */
+ iothread_ref(iothread, holder);
+
+ return iothread->ctx;
+}
+
+void iothread_unref_and_put_aio_context(IOThread *iothread,
+ const IOThreadHolder *holder)
+{
+ /* Delete IOThreadHolder from the list */
+ iothread_unref(iothread, holder);
+}
+
static int query_one_iothread(Object *object, void *opaque)
{
IOThreadInfoList ***tail = opaque;
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 04/14] block/export: track IOThread references
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (2 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 03/14] iothread: add holder-aware AioContext accessors Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 05/14] monitor: track IOThread users with QOM paths Zhang Chen
` (9 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Track the IOThreads used by a block export and identify the holder
with the unique BlockExportOptions id.
Acquire holder-aware references during export creation and release
them on error or export deletion. Support both single- and
multi-iothread exports.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
block/export/export.c | 62 ++++++++++++++++++++++++++++++++++++------
include/block/export.h | 5 ++++
2 files changed, 58 insertions(+), 9 deletions(-)
diff --git a/block/export/export.c b/block/export/export.c
index b733f269f3..0e019545ef 100644
--- a/block/export/export.c
+++ b/block/export/export.c
@@ -15,7 +15,6 @@
#include "block/block.h"
#include "system/block-backend.h"
-#include "system/iothread.h"
#include "block/export.h"
#include "block/fuse.h"
#include "block/nbd.h"
@@ -72,6 +71,32 @@ static const BlockExportDriver *blk_exp_find_driver(BlockExportType type)
return NULL;
}
+static void init_iothreads(const char *holder_id, IOThread **iothreads,
+ size_t num_iothreads, AioContext **aio_ctxs)
+{
+ const IOThreadHolder holder = {
+ .type = IO_THREAD_HOLDER_KIND_BLOCK_EXPORT,
+ .u.block_export.export_id = (char *)holder_id,
+ };
+
+ for (size_t i = 0; i < num_iothreads; i++) {
+ aio_ctxs[i] = iothread_ref_and_get_aio_context(iothreads[i], &holder);
+ }
+}
+
+static void cleanup_iothreads(const char *holder_id, IOThread **iothreads,
+ size_t num_iothreads)
+{
+ const IOThreadHolder holder = {
+ .type = IO_THREAD_HOLDER_KIND_BLOCK_EXPORT,
+ .u.block_export.export_id = (char *)holder_id,
+ };
+
+ for (size_t i = 0; i < num_iothreads; i++) {
+ iothread_unref_and_put_aio_context(iothreads[i], &holder);
+ }
+}
+
BlockExport *blk_exp_add(BlockExportOptions *export, Error **errp)
{
bool fixed_iothread = export->has_fixed_iothread && export->fixed_iothread;
@@ -85,6 +110,8 @@ BlockExport *blk_exp_add(BlockExportOptions *export, Error **errp)
AioContext *ctx;
AioContext **multithread_ctxs = NULL;
size_t multithread_count = 0;
+ g_autofree IOThread **local_iothreads = NULL;
+ size_t iothread_count = 0;
uint64_t perm;
int ret;
@@ -139,7 +166,10 @@ BlockExport *blk_exp_add(BlockExportOptions *export, Error **errp)
goto fail;
}
- new_ctx = iothread_get_aio_context(iothread);
+ local_iothreads = g_new0(IOThread *, 1);
+ local_iothreads[0] = iothread;
+ init_iothreads(export->id, local_iothreads, 1, &new_ctx);
+ iothread_count = 1;
/* Ignore errors with fixed-iothread=false */
set_context_errp = fixed_iothread ? errp : NULL;
@@ -163,8 +193,10 @@ BlockExport *blk_exp_add(BlockExportOptions *export, Error **errp)
return NULL;
}
+ local_iothreads = g_new0(IOThread *, multithread_count);
multithread_ctxs = g_new(AioContext *, multithread_count);
i = 0;
+
for (strList *e = iothread_list; e; e = e->next) {
IOThread *iothread = iothread_by_id(e->value);
@@ -172,9 +204,12 @@ BlockExport *blk_exp_add(BlockExportOptions *export, Error **errp)
error_setg(errp, "iothread \"%s\" not found", e->value);
goto fail;
}
- multithread_ctxs[i++] = iothread_get_aio_context(iothread);
+ local_iothreads[i++] = iothread;
}
assert(i == multithread_count);
+ init_iothreads(export->id, local_iothreads, multithread_count,
+ multithread_ctxs);
+ iothread_count = multithread_count;
}
bdrv_graph_rdlock_main_loop();
@@ -225,12 +260,14 @@ BlockExport *blk_exp_add(BlockExportOptions *export, Error **errp)
assert(drv->instance_size >= sizeof(BlockExport));
exp = g_malloc0(drv->instance_size);
*exp = (BlockExport) {
- .drv = drv,
- .refcount = 1,
- .user_owned = true,
- .id = g_strdup(export->id),
- .ctx = ctx,
- .blk = blk,
+ .drv = drv,
+ .refcount = 1,
+ .user_owned = true,
+ .id = g_strdup(export->id),
+ .ctx = ctx,
+ .blk = blk,
+ .iothreads = g_steal_pointer(&local_iothreads),
+ .iothread_count = iothread_count,
};
ret = drv->create(exp, export, multithread_ctxs, multithread_count, errp);
@@ -250,8 +287,12 @@ fail:
blk_unref(blk);
}
if (exp) {
+ cleanup_iothreads(exp->id, exp->iothreads, exp->iothread_count);
+ g_free(exp->iothreads);
g_free(exp->id);
g_free(exp);
+ } else {
+ cleanup_iothreads(export->id, local_iothreads, iothread_count);
}
g_free(multithread_ctxs);
return NULL;
@@ -273,6 +314,9 @@ static void blk_exp_delete_bh(void *opaque)
exp->drv->delete(exp);
blk_set_dev_ops(exp->blk, NULL, NULL);
blk_unref(exp->blk);
+
+ cleanup_iothreads(exp->id, exp->iothreads, exp->iothread_count);
+ g_free(exp->iothreads);
qapi_event_send_block_export_deleted(exp->id);
g_free(exp->id);
g_free(exp);
diff --git a/include/block/export.h b/include/block/export.h
index ca45da928c..ce8e4eb604 100644
--- a/include/block/export.h
+++ b/include/block/export.h
@@ -16,6 +16,7 @@
#include "qapi/qapi-types-block-export.h"
#include "qemu/queue.h"
+#include "system/iothread.h"
typedef struct BlockExport BlockExport;
@@ -89,6 +90,10 @@ struct BlockExport {
/* List entry for block_exports */
QLIST_ENTRY(BlockExport) next;
+
+ /* The IOThreads utilized by this specific block export */
+ IOThread **iothreads;
+ size_t iothread_count;
};
BlockExport *blk_exp_add(BlockExportOptions *export, Error **errp);
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 05/14] monitor: track IOThread users with QOM paths
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (3 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 04/14] block/export: track IOThread references Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 12:44 ` Markus Armbruster
2026-09-07 6:37 ` [PATCH V15 06/14] virtio-vq-mapping: track iothread-vq-mapping references using device path Zhang Chen
` (8 subsequent siblings)
13 siblings, 1 reply; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Monitors that require an IOThread share the internal mon_iothread.
Account for each monitor explicitly in IOThread holder tracking, using
the monitor's canonical QOM path to distinguish holders.
Acquire the shared monitor IOThread AioContext once and store the
IOThread and canonical QOM path so the holder can be released after the
monitor has been removed from the QOM tree. Reuse the stored context for
later operations.
Release the holder from monitor_finalize(), after resources associated
with the AioContext have been destroyed. This pairs the reference taken
in complete() with instance finalization and keeps the IOThread alive
throughout monitor teardown.
The internal mon_iothread remains hidden from query-iothreads, as before.
Management applications therefore cannot use it to determine which
monitors are using the IOThread. Nevertheless, track the monitor
holders so that they follow the same explicit holder and
IOThread reference lifecycle as other AioContext users.
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
monitor/monitor-internal.h | 5 +++++
monitor/monitor.c | 26 +++++++++++++++++++++-----
monitor/qmp.c | 5 +++--
3 files changed, 29 insertions(+), 7 deletions(-)
diff --git a/monitor/monitor-internal.h b/monitor/monitor-internal.h
index 822a66d06d..1d5aeae5e6 100644
--- a/monitor/monitor-internal.h
+++ b/monitor/monitor-internal.h
@@ -77,6 +77,11 @@ struct Monitor {
GString *outbuf;
guint out_watch;
int mux_out;
+
+ /* iothread context and holder identity */
+ IOThread *iothread;
+ char *iothread_qom_path;
+ AioContext *ctx;
};
struct MonitorQMPClass {
diff --git a/monitor/monitor.c b/monitor/monitor.c
index 2654b81890..cc7f6785b5 100644
--- a/monitor/monitor.c
+++ b/monitor/monitor.c
@@ -93,6 +93,16 @@ static void monitor_finalize(Object *obj)
qemu_chr_fe_deinit(&mon->chr, false);
g_string_free(mon->outbuf, true);
qemu_mutex_destroy(&mon->mon_lock);
+
+ if (mon->iothread) {
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = mon->iothread_qom_path,
+ };
+
+ iothread_unref_and_put_aio_context(mon->iothread, &io_holder);
+ }
+ g_free(mon->iothread_qom_path);
}
static char *monitor_get_chardev_id(Object *obj, Error **errp)
@@ -521,7 +531,7 @@ void monitor_suspend(Monitor *mon)
* Kick I/O thread to make sure this takes effect. It'll be
* evaluated again in prepare() of the watch object.
*/
- aio_notify(iothread_get_aio_context(mon_iothread));
+ aio_notify(mon->ctx);
}
trace_monitor_suspend(mon, 1);
@@ -661,7 +671,6 @@ char *monitor_compat_id(void)
static void monitor_complete(UserCreatable *uc, Error **errp)
{
Monitor *mon = MONITOR(uc);
- AioContext *ctx;
if (mon->chardev_id) {
Chardev *chr = qemu_chr_find(mon->chardev_id);
@@ -680,11 +689,18 @@ static void monitor_complete(UserCreatable *uc, Error **errp)
mon_iothread = iothread_create("mon_iothread", &error_abort);
}
- ctx = iothread_get_aio_context(mon_iothread);
+ mon->iothread = mon_iothread;
+ mon->iothread_qom_path = object_get_canonical_path(OBJECT(mon));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = mon->iothread_qom_path,
+ };
+
+ mon->ctx = iothread_ref_and_get_aio_context(mon->iothread, &io_holder);
} else {
- ctx = qemu_get_aio_context();
+ mon->ctx = qemu_get_aio_context();
}
- mon->accept_input_bh = aio_bh_new(ctx, monitor_accept_input, mon);
+ mon->accept_input_bh = aio_bh_new(mon->ctx, monitor_accept_input, mon);
}
int monitor_new(MonitorOptions *opts, bool allow_hmp, Error **errp)
diff --git a/monitor/qmp.c b/monitor/qmp.c
index aec0315775..de56345456 100644
--- a/monitor/qmp.c
+++ b/monitor/qmp.c
@@ -734,7 +734,8 @@ static void monitor_qmp_complete(UserCreatable *uc, Error **errp)
* thread. Schedule a bottom half.
*/
mon->setup_pending = true;
- aio_bh_schedule_oneshot(iothread_get_aio_context(mon_iothread),
+
+ aio_bh_schedule_oneshot(MONITOR(mon)->ctx,
monitor_qmp_setup_handlers_bh, mon);
/* The bottom half will add @mon to @mon_list */
} else {
@@ -789,7 +790,7 @@ static bool monitor_qmp_prepare_delete(UserCreatable *uc, Error **errp)
/* Synchronize with in-flight iothread callbacks. */
if (monitor_requires_iothread(mon)) {
- aio_wait_bh_oneshot(iothread_get_aio_context(mon_iothread),
+ aio_wait_bh_oneshot(MONITOR(mon)->ctx,
monitor_qmp_iothread_quiesce, NULL);
}
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* Re: [PATCH V15 05/14] monitor: track IOThread users with QOM paths
2026-09-07 6:37 ` [PATCH V15 05/14] monitor: track IOThread users with QOM paths Zhang Chen
@ 2026-09-07 12:44 ` Markus Armbruster
2026-09-09 7:57 ` Zhang Chen
0 siblings, 1 reply; 26+ messages in thread
From: Markus Armbruster @ 2026-09-07 12:44 UTC (permalink / raw)
To: Zhang Chen
Cc: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Zhang Chen <zhangckid@gmail.com> writes:
> Monitors that require an IOThread share the internal mon_iothread.
> Account for each monitor explicitly in IOThread holder tracking, using
> the monitor's canonical QOM path to distinguish holders.
>
> Acquire the shared monitor IOThread AioContext once and store the
> IOThread and canonical QOM path so the holder can be released after the
> monitor has been removed from the QOM tree. Reuse the stored context for
> later operations.
>
> Release the holder from monitor_finalize(), after resources associated
> with the AioContext have been destroyed. This pairs the reference taken
> in complete() with instance finalization and keeps the IOThread alive
> throughout monitor teardown.
>
> The internal mon_iothread remains hidden from query-iothreads, as before.
> Management applications therefore cannot use it to determine which
> monitors are using the IOThread. Nevertheless, track the monitor
> holders so that they follow the same explicit holder and
> IOThread reference lifecycle as other AioContext users.
>
> Signed-off-by: Zhang Chen <zhangckid@gmail.com>
> ---
> monitor/monitor-internal.h | 5 +++++
> monitor/monitor.c | 26 +++++++++++++++++++++-----
> monitor/qmp.c | 5 +++--
> 3 files changed, 29 insertions(+), 7 deletions(-)
>
> diff --git a/monitor/monitor-internal.h b/monitor/monitor-internal.h
> index 822a66d06d..1d5aeae5e6 100644
> --- a/monitor/monitor-internal.h
> +++ b/monitor/monitor-internal.h
> @@ -77,6 +77,11 @@ struct Monitor {
> GString *outbuf;
> guint out_watch;
> int mux_out;
> +
> + /* iothread context and holder identity */
> + IOThread *iothread;
> + char *iothread_qom_path;
> + AioContext *ctx;
> };
>
> struct MonitorQMPClass {
> diff --git a/monitor/monitor.c b/monitor/monitor.c
> index 2654b81890..cc7f6785b5 100644
> --- a/monitor/monitor.c
> +++ b/monitor/monitor.c
> @@ -93,6 +93,16 @@ static void monitor_finalize(Object *obj)
> qemu_chr_fe_deinit(&mon->chr, false);
> g_string_free(mon->outbuf, true);
> qemu_mutex_destroy(&mon->mon_lock);
> +
> + if (mon->iothread) {
As we'll see in monitor_complete() below, mon->iothread is mon_iothread
if monitor_requires_iothread(mon), else null. Which means
mon->mon_iothread is non-null if and only if
monitor_requires_iothread(mon).
Here, you check whether mon->mon_iothread is non-null to find out
whether @mon uses the monitor I/O thread, and you read the monitor I/O
thread from mon->mon_iothread.
Elsewhere in this file, we check monitor_requires_iothread(mon) and you
read mon_iothread, e.g. in monitor_cancel_out_watch()
if (monitor_requires_iothread(mon)) {
ctx = iothread_get_g_main_context(mon_iothread);
}
I don't like this inconsistency. Two ways to avoid it:
1. Always use monitor_requires_iothread(mon) to check and mon_iothread
to read. Drop struct Monitor member iothread.
2. Always use mon_iothread, except where we initialize it in
monitor_complete().
> + const IOThreadHolder io_holder = {
> + .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
> + .u.qom_object.qom_path = mon->iothread_qom_path,
> + };
> +
> + iothread_unref_and_put_aio_context(mon->iothread, &io_holder);
> + }
> + g_free(mon->iothread_qom_path);
> }
>
> static char *monitor_get_chardev_id(Object *obj, Error **errp)
> @@ -521,7 +531,7 @@ void monitor_suspend(Monitor *mon)
> * Kick I/O thread to make sure this takes effect. It'll be
> * evaluated again in prepare() of the watch object.
> */
> - aio_notify(iothread_get_aio_context(mon_iothread));
> + aio_notify(mon->ctx);
> }
>
> trace_monitor_suspend(mon, 1);
> @@ -661,7 +671,6 @@ char *monitor_compat_id(void)
> static void monitor_complete(UserCreatable *uc, Error **errp)
> {
> Monitor *mon = MONITOR(uc);
> - AioContext *ctx;
>
> if (mon->chardev_id) {
> Chardev *chr = qemu_chr_find(mon->chardev_id);
> @@ -680,11 +689,18 @@ static void monitor_complete(UserCreatable *uc, Error **errp)
if (monitor_requires_iothread(mon)) {
if (!mon_iothread) {
> mon_iothread = iothread_create("mon_iothread", &error_abort);
Create the single, shared monitor I/O thread on demand, and store it in
mon_iothread.
> }
>
> - ctx = iothread_get_aio_context(mon_iothread);
> + mon->iothread = mon_iothread;
Set mon->iothread to the monitor I/O thread.
> + mon->iothread_qom_path = object_get_canonical_path(OBJECT(mon));
> + const IOThreadHolder io_holder = {
> + .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
> + .u.qom_object.qom_path = mon->iothread_qom_path,
> + };
> +
> + mon->ctx = iothread_ref_and_get_aio_context(mon->iothread, &io_holder);
> } else {
> - ctx = qemu_get_aio_context();
> + mon->ctx = qemu_get_aio_context();
Leave mon->iothread null.
> }
mon->iothread is mon_iothread if monitor_requires_iothread(mon), else
null.
> - mon->accept_input_bh = aio_bh_new(ctx, monitor_accept_input, mon);
> + mon->accept_input_bh = aio_bh_new(mon->ctx, monitor_accept_input, mon);
> }
>
> int monitor_new(MonitorOptions *opts, bool allow_hmp, Error **errp)
> diff --git a/monitor/qmp.c b/monitor/qmp.c
> index aec0315775..de56345456 100644
> --- a/monitor/qmp.c
> +++ b/monitor/qmp.c
> @@ -734,7 +734,8 @@ static void monitor_qmp_complete(UserCreatable *uc, Error **errp)
> * thread. Schedule a bottom half.
> */
> mon->setup_pending = true;
> - aio_bh_schedule_oneshot(iothread_get_aio_context(mon_iothread),
> +
> + aio_bh_schedule_oneshot(MONITOR(mon)->ctx,
> monitor_qmp_setup_handlers_bh, mon);
> /* The bottom half will add @mon to @mon_list */
> } else {
> @@ -789,7 +790,7 @@ static bool monitor_qmp_prepare_delete(UserCreatable *uc, Error **errp)
>
> /* Synchronize with in-flight iothread callbacks. */
> if (monitor_requires_iothread(mon)) {
> - aio_wait_bh_oneshot(iothread_get_aio_context(mon_iothread),
> + aio_wait_bh_oneshot(MONITOR(mon)->ctx,
> monitor_qmp_iothread_quiesce, NULL);
> }
^ permalink raw reply [flat|nested] 26+ messages in thread* Re: [PATCH V15 05/14] monitor: track IOThread users with QOM paths
2026-09-07 12:44 ` Markus Armbruster
@ 2026-09-09 7:57 ` Zhang Chen
2026-09-09 10:40 ` Markus Armbruster
0 siblings, 1 reply; 26+ messages in thread
From: Zhang Chen @ 2026-09-09 7:57 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel, Eric Blake, Michael S . Tsirkin, Stefan Hajnoczi,
Paolo Bonzini, Daniel P . Berrangé, Jason Wang
On Mon, Sep 7, 2026 at 8:44 PM Markus Armbruster <armbru@redhat.com> wrote:
>
> Zhang Chen <zhangckid@gmail.com> writes:
>
> > Monitors that require an IOThread share the internal mon_iothread.
> > Account for each monitor explicitly in IOThread holder tracking, using
> > the monitor's canonical QOM path to distinguish holders.
> >
> > Acquire the shared monitor IOThread AioContext once and store the
> > IOThread and canonical QOM path so the holder can be released after the
> > monitor has been removed from the QOM tree. Reuse the stored context for
> > later operations.
> >
> > Release the holder from monitor_finalize(), after resources associated
> > with the AioContext have been destroyed. This pairs the reference taken
> > in complete() with instance finalization and keeps the IOThread alive
> > throughout monitor teardown.
> >
> > The internal mon_iothread remains hidden from query-iothreads, as before.
> > Management applications therefore cannot use it to determine which
> > monitors are using the IOThread. Nevertheless, track the monitor
> > holders so that they follow the same explicit holder and
> > IOThread reference lifecycle as other AioContext users.
> >
> > Signed-off-by: Zhang Chen <zhangckid@gmail.com>
> > ---
> > monitor/monitor-internal.h | 5 +++++
> > monitor/monitor.c | 26 +++++++++++++++++++++-----
> > monitor/qmp.c | 5 +++--
> > 3 files changed, 29 insertions(+), 7 deletions(-)
> >
> > diff --git a/monitor/monitor-internal.h b/monitor/monitor-internal.h
> > index 822a66d06d..1d5aeae5e6 100644
> > --- a/monitor/monitor-internal.h
> > +++ b/monitor/monitor-internal.h
> > @@ -77,6 +77,11 @@ struct Monitor {
> > GString *outbuf;
> > guint out_watch;
> > int mux_out;
> > +
> > + /* iothread context and holder identity */
> > + IOThread *iothread;
> > + char *iothread_qom_path;
> > + AioContext *ctx;
> > };
> >
> > struct MonitorQMPClass {
> > diff --git a/monitor/monitor.c b/monitor/monitor.c
> > index 2654b81890..cc7f6785b5 100644
> > --- a/monitor/monitor.c
> > +++ b/monitor/monitor.c
> > @@ -93,6 +93,16 @@ static void monitor_finalize(Object *obj)
> > qemu_chr_fe_deinit(&mon->chr, false);
> > g_string_free(mon->outbuf, true);
> > qemu_mutex_destroy(&mon->mon_lock);
> > +
> > + if (mon->iothread) {
>
> As we'll see in monitor_complete() below, mon->iothread is mon_iothread
> if monitor_requires_iothread(mon), else null. Which means
> mon->mon_iothread is non-null if and only if
> monitor_requires_iothread(mon).
>
> Here, you check whether mon->mon_iothread is non-null to find out
> whether @mon uses the monitor I/O thread, and you read the monitor I/O
> thread from mon->mon_iothread.
>
> Elsewhere in this file, we check monitor_requires_iothread(mon) and you
> read mon_iothread, e.g. in monitor_cancel_out_watch()
>
> if (monitor_requires_iothread(mon)) {
> ctx = iothread_get_g_main_context(mon_iothread);
> }
>
> I don't like this inconsistency. Two ways to avoid it:
>
> 1. Always use monitor_requires_iothread(mon) to check and mon_iothread
> to read. Drop struct Monitor member iothread.
>
> 2. Always use mon_iothread, except where we initialize it in
> monitor_complete().
Just to clarify: by option 2, do you mean using mon->iothread
consistently for per-monitor checks and accesses, except in
monitor_complete(), where monitor_requires_iothread() determines
whether to initialize it?
Thanks
Chen
>
> > + const IOThreadHolder io_holder = {
> > + .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
> > + .u.qom_object.qom_path = mon->iothread_qom_path,
> > + };
> > +
> > + iothread_unref_and_put_aio_context(mon->iothread, &io_holder);
> > + }
> > + g_free(mon->iothread_qom_path);
> > }
> >
> > static char *monitor_get_chardev_id(Object *obj, Error **errp)
> > @@ -521,7 +531,7 @@ void monitor_suspend(Monitor *mon)
> > * Kick I/O thread to make sure this takes effect. It'll be
> > * evaluated again in prepare() of the watch object.
> > */
> > - aio_notify(iothread_get_aio_context(mon_iothread));
> > + aio_notify(mon->ctx);
> > }
> >
> > trace_monitor_suspend(mon, 1);
> > @@ -661,7 +671,6 @@ char *monitor_compat_id(void)
> > static void monitor_complete(UserCreatable *uc, Error **errp)
> > {
> > Monitor *mon = MONITOR(uc);
> > - AioContext *ctx;
> >
> > if (mon->chardev_id) {
> > Chardev *chr = qemu_chr_find(mon->chardev_id);
> > @@ -680,11 +689,18 @@ static void monitor_complete(UserCreatable *uc, Error **errp)
> if (monitor_requires_iothread(mon)) {
> if (!mon_iothread) {
> > mon_iothread = iothread_create("mon_iothread", &error_abort);
>
> Create the single, shared monitor I/O thread on demand, and store it in
> mon_iothread.
>
> > }
> >
> > - ctx = iothread_get_aio_context(mon_iothread);
> > + mon->iothread = mon_iothread;
>
> Set mon->iothread to the monitor I/O thread.
>
> > + mon->iothread_qom_path = object_get_canonical_path(OBJECT(mon));
> > + const IOThreadHolder io_holder = {
> > + .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
> > + .u.qom_object.qom_path = mon->iothread_qom_path,
> > + };
> > +
> > + mon->ctx = iothread_ref_and_get_aio_context(mon->iothread, &io_holder);
> > } else {
> > - ctx = qemu_get_aio_context();
> > + mon->ctx = qemu_get_aio_context();
>
> Leave mon->iothread null.
>
> > }
>
> mon->iothread is mon_iothread if monitor_requires_iothread(mon), else
> null.
>
> > - mon->accept_input_bh = aio_bh_new(ctx, monitor_accept_input, mon);
> > + mon->accept_input_bh = aio_bh_new(mon->ctx, monitor_accept_input, mon);
> > }
> >
> > int monitor_new(MonitorOptions *opts, bool allow_hmp, Error **errp)
> > diff --git a/monitor/qmp.c b/monitor/qmp.c
> > index aec0315775..de56345456 100644
> > --- a/monitor/qmp.c
> > +++ b/monitor/qmp.c
> > @@ -734,7 +734,8 @@ static void monitor_qmp_complete(UserCreatable *uc, Error **errp)
> > * thread. Schedule a bottom half.
> > */
> > mon->setup_pending = true;
> > - aio_bh_schedule_oneshot(iothread_get_aio_context(mon_iothread),
> > +
> > + aio_bh_schedule_oneshot(MONITOR(mon)->ctx,
> > monitor_qmp_setup_handlers_bh, mon);
> > /* The bottom half will add @mon to @mon_list */
> > } else {
> > @@ -789,7 +790,7 @@ static bool monitor_qmp_prepare_delete(UserCreatable *uc, Error **errp)
> >
> > /* Synchronize with in-flight iothread callbacks. */
> > if (monitor_requires_iothread(mon)) {
> > - aio_wait_bh_oneshot(iothread_get_aio_context(mon_iothread),
> > + aio_wait_bh_oneshot(MONITOR(mon)->ctx,
> > monitor_qmp_iothread_quiesce, NULL);
> > }
>
^ permalink raw reply [flat|nested] 26+ messages in thread* Re: [PATCH V15 05/14] monitor: track IOThread users with QOM paths
2026-09-09 7:57 ` Zhang Chen
@ 2026-09-09 10:40 ` Markus Armbruster
2026-09-10 2:24 ` Zhang Chen
0 siblings, 1 reply; 26+ messages in thread
From: Markus Armbruster @ 2026-09-09 10:40 UTC (permalink / raw)
To: Zhang Chen
Cc: qemu-devel, Eric Blake, Michael S . Tsirkin, Stefan Hajnoczi,
Paolo Bonzini, Daniel P . Berrangé, Jason Wang
Zhang Chen <zhangckid@gmail.com> writes:
> On Mon, Sep 7, 2026 at 8:44 PM Markus Armbruster <armbru@redhat.com> wrote:
>>
>> Zhang Chen <zhangckid@gmail.com> writes:
>>
>> > Monitors that require an IOThread share the internal mon_iothread.
>> > Account for each monitor explicitly in IOThread holder tracking, using
>> > the monitor's canonical QOM path to distinguish holders.
>> >
>> > Acquire the shared monitor IOThread AioContext once and store the
>> > IOThread and canonical QOM path so the holder can be released after the
>> > monitor has been removed from the QOM tree. Reuse the stored context for
>> > later operations.
>> >
>> > Release the holder from monitor_finalize(), after resources associated
>> > with the AioContext have been destroyed. This pairs the reference taken
>> > in complete() with instance finalization and keeps the IOThread alive
>> > throughout monitor teardown.
>> >
>> > The internal mon_iothread remains hidden from query-iothreads, as before.
>> > Management applications therefore cannot use it to determine which
>> > monitors are using the IOThread. Nevertheless, track the monitor
>> > holders so that they follow the same explicit holder and
>> > IOThread reference lifecycle as other AioContext users.
>> >
>> > Signed-off-by: Zhang Chen <zhangckid@gmail.com>
>> > ---
>> > monitor/monitor-internal.h | 5 +++++
>> > monitor/monitor.c | 26 +++++++++++++++++++++-----
>> > monitor/qmp.c | 5 +++--
>> > 3 files changed, 29 insertions(+), 7 deletions(-)
>> >
>> > diff --git a/monitor/monitor-internal.h b/monitor/monitor-internal.h
>> > index 822a66d06d..1d5aeae5e6 100644
>> > --- a/monitor/monitor-internal.h
>> > +++ b/monitor/monitor-internal.h
>> > @@ -77,6 +77,11 @@ struct Monitor {
>> > GString *outbuf;
>> > guint out_watch;
>> > int mux_out;
>> > +
>> > + /* iothread context and holder identity */
>> > + IOThread *iothread;
>> > + char *iothread_qom_path;
>> > + AioContext *ctx;
>> > };
>> >
>> > struct MonitorQMPClass {
>> > diff --git a/monitor/monitor.c b/monitor/monitor.c
>> > index 2654b81890..cc7f6785b5 100644
>> > --- a/monitor/monitor.c
>> > +++ b/monitor/monitor.c
>> > @@ -93,6 +93,16 @@ static void monitor_finalize(Object *obj)
>> > qemu_chr_fe_deinit(&mon->chr, false);
>> > g_string_free(mon->outbuf, true);
>> > qemu_mutex_destroy(&mon->mon_lock);
>> > +
>> > + if (mon->iothread) {
>>
>> As we'll see in monitor_complete() below, mon->iothread is mon_iothread
>> if monitor_requires_iothread(mon), else null. Which means
>> mon->mon_iothread is non-null if and only if
>> monitor_requires_iothread(mon).
>>
>> Here, you check whether mon->mon_iothread is non-null to find out
>> whether @mon uses the monitor I/O thread, and you read the monitor I/O
>> thread from mon->mon_iothread.
>>
>> Elsewhere in this file, we check monitor_requires_iothread(mon) and you
>> read mon_iothread, e.g. in monitor_cancel_out_watch()
>>
>> if (monitor_requires_iothread(mon)) {
>> ctx = iothread_get_g_main_context(mon_iothread);
>> }
>>
>> I don't like this inconsistency. Two ways to avoid it:
>>
>> 1. Always use monitor_requires_iothread(mon) to check and mon_iothread
>> to read. Drop struct Monitor member iothread.
>>
>> 2. Always use mon_iothread, except where we initialize it in
>> monitor_complete().
>
> Just to clarify: by option 2, do you mean using mon->iothread
> consistently for per-monitor checks and accesses, except in
> monitor_complete(), where monitor_requires_iothread() determines
> whether to initialize it?
Yes. I meant to write 2. Always use mon->iothread, but mistyped.
^ permalink raw reply [flat|nested] 26+ messages in thread* Re: [PATCH V15 05/14] monitor: track IOThread users with QOM paths
2026-09-09 10:40 ` Markus Armbruster
@ 2026-09-10 2:24 ` Zhang Chen
2026-09-10 4:42 ` Markus Armbruster
0 siblings, 1 reply; 26+ messages in thread
From: Zhang Chen @ 2026-09-10 2:24 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel, Eric Blake, Michael S . Tsirkin, Stefan Hajnoczi,
Paolo Bonzini, Daniel P . Berrangé, Jason Wang
On Wed, Sep 9, 2026 at 6:40 PM Markus Armbruster <armbru@redhat.com> wrote:
>
> Zhang Chen <zhangckid@gmail.com> writes:
>
> > On Mon, Sep 7, 2026 at 8:44 PM Markus Armbruster <armbru@redhat.com> wrote:
> >>
> >> Zhang Chen <zhangckid@gmail.com> writes:
> >>
> >> > Monitors that require an IOThread share the internal mon_iothread.
> >> > Account for each monitor explicitly in IOThread holder tracking, using
> >> > the monitor's canonical QOM path to distinguish holders.
> >> >
> >> > Acquire the shared monitor IOThread AioContext once and store the
> >> > IOThread and canonical QOM path so the holder can be released after the
> >> > monitor has been removed from the QOM tree. Reuse the stored context for
> >> > later operations.
> >> >
> >> > Release the holder from monitor_finalize(), after resources associated
> >> > with the AioContext have been destroyed. This pairs the reference taken
> >> > in complete() with instance finalization and keeps the IOThread alive
> >> > throughout monitor teardown.
> >> >
> >> > The internal mon_iothread remains hidden from query-iothreads, as before.
> >> > Management applications therefore cannot use it to determine which
> >> > monitors are using the IOThread. Nevertheless, track the monitor
> >> > holders so that they follow the same explicit holder and
> >> > IOThread reference lifecycle as other AioContext users.
> >> >
> >> > Signed-off-by: Zhang Chen <zhangckid@gmail.com>
> >> > ---
> >> > monitor/monitor-internal.h | 5 +++++
> >> > monitor/monitor.c | 26 +++++++++++++++++++++-----
> >> > monitor/qmp.c | 5 +++--
> >> > 3 files changed, 29 insertions(+), 7 deletions(-)
> >> >
> >> > diff --git a/monitor/monitor-internal.h b/monitor/monitor-internal.h
> >> > index 822a66d06d..1d5aeae5e6 100644
> >> > --- a/monitor/monitor-internal.h
> >> > +++ b/monitor/monitor-internal.h
> >> > @@ -77,6 +77,11 @@ struct Monitor {
> >> > GString *outbuf;
> >> > guint out_watch;
> >> > int mux_out;
> >> > +
> >> > + /* iothread context and holder identity */
> >> > + IOThread *iothread;
> >> > + char *iothread_qom_path;
> >> > + AioContext *ctx;
> >> > };
> >> >
> >> > struct MonitorQMPClass {
> >> > diff --git a/monitor/monitor.c b/monitor/monitor.c
> >> > index 2654b81890..cc7f6785b5 100644
> >> > --- a/monitor/monitor.c
> >> > +++ b/monitor/monitor.c
> >> > @@ -93,6 +93,16 @@ static void monitor_finalize(Object *obj)
> >> > qemu_chr_fe_deinit(&mon->chr, false);
> >> > g_string_free(mon->outbuf, true);
> >> > qemu_mutex_destroy(&mon->mon_lock);
> >> > +
> >> > + if (mon->iothread) {
> >>
> >> As we'll see in monitor_complete() below, mon->iothread is mon_iothread
> >> if monitor_requires_iothread(mon), else null. Which means
> >> mon->mon_iothread is non-null if and only if
> >> monitor_requires_iothread(mon).
> >>
> >> Here, you check whether mon->mon_iothread is non-null to find out
> >> whether @mon uses the monitor I/O thread, and you read the monitor I/O
> >> thread from mon->mon_iothread.
> >>
> >> Elsewhere in this file, we check monitor_requires_iothread(mon) and you
> >> read mon_iothread, e.g. in monitor_cancel_out_watch()
> >>
> >> if (monitor_requires_iothread(mon)) {
> >> ctx = iothread_get_g_main_context(mon_iothread);
> >> }
> >>
> >> I don't like this inconsistency. Two ways to avoid it:
> >>
> >> 1. Always use monitor_requires_iothread(mon) to check and mon_iothread
> >> to read. Drop struct Monitor member iothread.
> >>
> >> 2. Always use mon_iothread, except where we initialize it in
> >> monitor_complete().
> >
> > Just to clarify: by option 2, do you mean using mon->iothread
> > consistently for per-monitor checks and accesses, except in
> > monitor_complete(), where monitor_requires_iothread() determines
> > whether to initialize it?
>
> Yes. I meant to write 2. Always use mon->iothread, but mistyped.
>
Thanks for clarifying. I'll use mon->iothread consistently for
per-monitor checks and accesses after initialization, keeping
monitor_requires_iothread() for the decision in monitor_complete()
and the global mon_iothread for shared thread creation and cleanup.
By the way, any comments for other patches in this series?
Thanks
Chen
^ permalink raw reply [flat|nested] 26+ messages in thread* Re: [PATCH V15 05/14] monitor: track IOThread users with QOM paths
2026-09-10 2:24 ` Zhang Chen
@ 2026-09-10 4:42 ` Markus Armbruster
0 siblings, 0 replies; 26+ messages in thread
From: Markus Armbruster @ 2026-09-10 4:42 UTC (permalink / raw)
To: Zhang Chen
Cc: qemu-devel, Eric Blake, Michael S . Tsirkin, Stefan Hajnoczi,
Paolo Bonzini, Daniel P . Berrangé, Jason Wang
Zhang Chen <zhangckid@gmail.com> writes:
> Thanks for clarifying. I'll use mon->iothread consistently for
> per-monitor checks and accesses after initialization, keeping
> monitor_requires_iothread() for the decision in monitor_complete()
> and the global mon_iothread for shared thread creation and cleanup.
>
> By the way, any comments for other patches in this series?
No, I'm done reviewing v15 :)
^ permalink raw reply [flat|nested] 26+ messages in thread
* [PATCH V15 06/14] virtio-vq-mapping: track iothread-vq-mapping references using device path
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (4 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 05/14] monitor: track IOThread users with QOM paths Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 07/14] virtio: track IOThread references for thread pinning Zhang Chen
` (7 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Replace raw object_ref/unref calls with iothread_ref_and_get/
unref_and_put_aio_context in iothread-vq-mapping. This allows
tracking IOThread users via the device's canonical QOM path,
improving lifecycle traceability for virtio-blk and
virtio-scsi devices.
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
hw/block/virtio-blk.c | 8 +++++++-
hw/scsi/virtio-scsi-dataplane.c | 9 +++++++--
hw/virtio/iothread-vq-mapping.c | 20 +++++++++++++++-----
include/hw/virtio/iothread-vq-mapping.h | 6 +++++-
4 files changed, 34 insertions(+), 9 deletions(-)
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index 6b92066aff..4b2cf3d52f 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -1493,9 +1493,12 @@ static bool virtio_blk_vq_aio_context_init(VirtIOBlock *s, Error **errp)
s->vq_aio_context = g_new(AioContext *, conf->num_queues);
if (conf->iothread_vq_mapping_list) {
+ g_autofree char *path = object_get_canonical_path(OBJECT(vdev));
+
if (!iothread_vq_mapping_apply(conf->iothread_vq_mapping_list,
s->vq_aio_context,
conf->num_queues,
+ path,
errp)) {
g_free(s->vq_aio_context);
s->vq_aio_context = NULL;
@@ -1527,7 +1530,10 @@ static void virtio_blk_vq_aio_context_cleanup(VirtIOBlock *s)
assert(!s->ioeventfd_started);
if (conf->iothread_vq_mapping_list) {
- iothread_vq_mapping_cleanup(conf->iothread_vq_mapping_list);
+ g_autofree char *path = object_get_canonical_path(
+ OBJECT(VIRTIO_DEVICE(s)));
+
+ iothread_vq_mapping_cleanup(conf->iothread_vq_mapping_list, path);
}
if (conf->iothread) {
diff --git a/hw/scsi/virtio-scsi-dataplane.c b/hw/scsi/virtio-scsi-dataplane.c
index 95f13fb7c2..26ecefd547 100644
--- a/hw/scsi/virtio-scsi-dataplane.c
+++ b/hw/scsi/virtio-scsi-dataplane.c
@@ -65,9 +65,11 @@ void virtio_scsi_dataplane_setup(VirtIOSCSI *s, Error **errp)
s->vq_aio_context[1] = qemu_get_aio_context();
if (vs->conf.iothread_vq_mapping_list) {
+ g_autofree char *path = object_get_canonical_path(OBJECT(vdev));
+
if (!iothread_vq_mapping_apply(vs->conf.iothread_vq_mapping_list,
&s->vq_aio_context[VIRTIO_SCSI_VQ_NUM_FIXED],
- vs->conf.num_queues, errp)) {
+ vs->conf.num_queues, path, errp)) {
g_free(s->vq_aio_context);
s->vq_aio_context = NULL;
return;
@@ -94,7 +96,10 @@ void virtio_scsi_dataplane_cleanup(VirtIOSCSI *s)
VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
if (vs->conf.iothread_vq_mapping_list) {
- iothread_vq_mapping_cleanup(vs->conf.iothread_vq_mapping_list);
+ g_autofree char *path = object_get_canonical_path(
+ OBJECT(VIRTIO_DEVICE(s)));
+
+ iothread_vq_mapping_cleanup(vs->conf.iothread_vq_mapping_list, path);
}
if (vs->conf.iothread) {
diff --git a/hw/virtio/iothread-vq-mapping.c b/hw/virtio/iothread-vq-mapping.c
index 55ce62986c..60e1d55005 100644
--- a/hw/virtio/iothread-vq-mapping.c
+++ b/hw/virtio/iothread-vq-mapping.c
@@ -77,6 +77,7 @@ bool iothread_vq_mapping_apply(
IOThreadVirtQueueMappingList *list,
AioContext **vq_aio_context,
uint16_t num_queues,
+ const char *holder,
Error **errp)
{
IOThreadVirtQueueMappingList *node;
@@ -93,10 +94,13 @@ bool iothread_vq_mapping_apply(
for (node = list; node; node = node->next) {
IOThread *iothread = iothread_by_id(node->value->iothread);
- AioContext *ctx = iothread_get_aio_context(iothread);
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = (char *)holder,
+ };
- /* Released in virtio_blk_vq_aio_context_cleanup() */
- object_ref(OBJECT(iothread));
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread,
+ &io_holder);
if (node->value->vqs) {
uint16List *vq;
@@ -120,13 +124,19 @@ bool iothread_vq_mapping_apply(
return true;
}
-void iothread_vq_mapping_cleanup(IOThreadVirtQueueMappingList *list)
+void iothread_vq_mapping_cleanup(IOThreadVirtQueueMappingList *list,
+ const char *holder)
{
IOThreadVirtQueueMappingList *node;
for (node = list; node; node = node->next) {
IOThread *iothread = iothread_by_id(node->value->iothread);
- object_unref(OBJECT(iothread));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = (char *)holder,
+ };
+
+ iothread_unref_and_put_aio_context(iothread, &io_holder);
}
}
diff --git a/include/hw/virtio/iothread-vq-mapping.h b/include/hw/virtio/iothread-vq-mapping.h
index 57335c3703..0d39caddf3 100644
--- a/include/hw/virtio/iothread-vq-mapping.h
+++ b/include/hw/virtio/iothread-vq-mapping.h
@@ -17,6 +17,7 @@
* @list: The mapping of virtqueues to IOThreads.
* @vq_aio_context: The array of AioContext pointers to fill in.
* @num_queues: The length of @vq_aio_context.
+ * @holder: The QOM paths for attached device.
* @errp: If an error occurs, a pointer to the area to store the error.
*
* Fill in the AioContext for each virtqueue in the @vq_aio_context array given
@@ -31,15 +32,18 @@ bool iothread_vq_mapping_apply(
IOThreadVirtQueueMappingList *list,
AioContext **vq_aio_context,
uint16_t num_queues,
+ const char *holder,
Error **errp);
/**
* iothread_vq_mapping_cleanup:
* @list: The mapping of virtqueues to IOThreads.
+ * @holder: The QOM paths for attached device.
*
* Release IOThread object references that were acquired by
* iothread_vq_mapping_apply().
*/
-void iothread_vq_mapping_cleanup(IOThreadVirtQueueMappingList *list);
+void iothread_vq_mapping_cleanup(IOThreadVirtQueueMappingList *list,
+ const char *holder);
#endif /* HW_VIRTIO_IOTHREAD_VQ_MAPPING_H */
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 07/14] virtio: track IOThread references for thread pinning
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (5 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 06/14] virtio-vq-mapping: track iothread-vq-mapping references using device path Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 08/14] net/colo: track IOThread references using path-based holder Zhang Chen
` (6 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Use the virtio device canonical QOM path as the holder when virtio-blk
and virtio-scsi acquire an IOThread AioContext.
Replace the raw object references with the holder-aware
acquire/release helpers so the users are visible through IOThread
introspection.
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
hw/block/virtio-blk.c | 23 +++++++++++++----------
hw/scsi/virtio-scsi-dataplane.c | 23 +++++++++++++----------
2 files changed, 26 insertions(+), 20 deletions(-)
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index 4b2cf3d52f..db7f513da5 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -1469,6 +1469,7 @@ static bool virtio_blk_vq_aio_context_init(VirtIOBlock *s, Error **errp)
VirtIOBlkConf *conf = &s->conf;
BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));
VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
+ g_autofree char *path = object_get_canonical_path(OBJECT(vdev));
if (conf->iothread && conf->iothread_vq_mapping_list) {
error_setg(errp,
@@ -1493,8 +1494,6 @@ static bool virtio_blk_vq_aio_context_init(VirtIOBlock *s, Error **errp)
s->vq_aio_context = g_new(AioContext *, conf->num_queues);
if (conf->iothread_vq_mapping_list) {
- g_autofree char *path = object_get_canonical_path(OBJECT(vdev));
-
if (!iothread_vq_mapping_apply(conf->iothread_vq_mapping_list,
s->vq_aio_context,
conf->num_queues,
@@ -1505,13 +1504,15 @@ static bool virtio_blk_vq_aio_context_init(VirtIOBlock *s, Error **errp)
return false;
}
} else if (conf->iothread) {
- AioContext *ctx = iothread_get_aio_context(conf->iothread);
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+ AioContext *ctx = iothread_ref_and_get_aio_context(conf->iothread,
+ &io_holder);
for (unsigned i = 0; i < conf->num_queues; i++) {
s->vq_aio_context[i] = ctx;
}
-
- /* Released in virtio_blk_vq_aio_context_cleanup() */
- object_ref(OBJECT(conf->iothread));
} else {
AioContext *ctx = qemu_get_aio_context();
for (unsigned i = 0; i < conf->num_queues; i++) {
@@ -1526,18 +1527,20 @@ static bool virtio_blk_vq_aio_context_init(VirtIOBlock *s, Error **errp)
static void virtio_blk_vq_aio_context_cleanup(VirtIOBlock *s)
{
VirtIOBlkConf *conf = &s->conf;
+ g_autofree char *path = object_get_canonical_path(OBJECT(VIRTIO_DEVICE(s)));
assert(!s->ioeventfd_started);
if (conf->iothread_vq_mapping_list) {
- g_autofree char *path = object_get_canonical_path(
- OBJECT(VIRTIO_DEVICE(s)));
-
iothread_vq_mapping_cleanup(conf->iothread_vq_mapping_list, path);
}
if (conf->iothread) {
- object_unref(OBJECT(conf->iothread));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+ iothread_unref_and_put_aio_context(conf->iothread, &io_holder);
}
g_free(s->vq_aio_context);
diff --git a/hw/scsi/virtio-scsi-dataplane.c b/hw/scsi/virtio-scsi-dataplane.c
index 26ecefd547..55bbf045ce 100644
--- a/hw/scsi/virtio-scsi-dataplane.c
+++ b/hw/scsi/virtio-scsi-dataplane.c
@@ -28,6 +28,7 @@ void virtio_scsi_dataplane_setup(VirtIOSCSI *s, Error **errp)
VirtIODevice *vdev = VIRTIO_DEVICE(s);
BusState *qbus = qdev_get_parent_bus(DEVICE(vdev));
VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
+ g_autofree char *path = object_get_canonical_path(OBJECT(vdev));
if (vs->conf.iothread && vs->conf.iothread_vq_mapping_list) {
error_setg(errp,
@@ -65,8 +66,6 @@ void virtio_scsi_dataplane_setup(VirtIOSCSI *s, Error **errp)
s->vq_aio_context[1] = qemu_get_aio_context();
if (vs->conf.iothread_vq_mapping_list) {
- g_autofree char *path = object_get_canonical_path(OBJECT(vdev));
-
if (!iothread_vq_mapping_apply(vs->conf.iothread_vq_mapping_list,
&s->vq_aio_context[VIRTIO_SCSI_VQ_NUM_FIXED],
vs->conf.num_queues, path, errp)) {
@@ -75,13 +74,15 @@ void virtio_scsi_dataplane_setup(VirtIOSCSI *s, Error **errp)
return;
}
} else if (vs->conf.iothread) {
- AioContext *ctx = iothread_get_aio_context(vs->conf.iothread);
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+ AioContext *ctx = iothread_ref_and_get_aio_context(vs->conf.iothread,
+ &io_holder);
for (uint16_t i = 0; i < vs->conf.num_queues; i++) {
s->vq_aio_context[VIRTIO_SCSI_VQ_NUM_FIXED + i] = ctx;
}
-
- /* Released in virtio_scsi_dataplane_cleanup() */
- object_ref(OBJECT(vs->conf.iothread));
} else {
AioContext *ctx = qemu_get_aio_context();
for (unsigned i = 0; i < vs->conf.num_queues; i++) {
@@ -94,16 +95,18 @@ void virtio_scsi_dataplane_setup(VirtIOSCSI *s, Error **errp)
void virtio_scsi_dataplane_cleanup(VirtIOSCSI *s)
{
VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
+ g_autofree char *path = object_get_canonical_path(OBJECT(VIRTIO_DEVICE(s)));
if (vs->conf.iothread_vq_mapping_list) {
- g_autofree char *path = object_get_canonical_path(
- OBJECT(VIRTIO_DEVICE(s)));
-
iothread_vq_mapping_cleanup(vs->conf.iothread_vq_mapping_list, path);
}
if (vs->conf.iothread) {
- object_unref(OBJECT(vs->conf.iothread));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+ iothread_unref_and_put_aio_context(vs->conf.iothread, &io_holder);
}
g_free(s->vq_aio_context);
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 08/14] net/colo: track IOThread references using path-based holder
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (6 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 07/14] virtio: track IOThread references for thread pinning Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 09/14] virtio-balloon: Update tracking iothread users with holder Zhang Chen
` (5 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Convert colo-compare to use the iothread_ref_and_get_aio_context() and
iothread_unref_and_put_aio_context() APIs. This ensures that
IOThread references are tracked using the COLO object's canonical
QOM path as the holder ID.
This refactoring improves IOThread lifecycle traceability and aligns
the code with modern QEMU iothread reference management patterns.
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
net/colo-compare.c | 29 +++++++++++++++++++----------
1 file changed, 19 insertions(+), 10 deletions(-)
diff --git a/net/colo-compare.c b/net/colo-compare.c
index 5986fb1e88..361ee32729 100644
--- a/net/colo-compare.c
+++ b/net/colo-compare.c
@@ -130,6 +130,7 @@ struct CompareState {
GHashTable *connection_track_table;
IOThread *iothread;
+ AioContext *iothread_ctx;
GMainContext *worker_context;
QEMUTimer *packet_check_timer;
@@ -922,9 +923,7 @@ void colo_notify_compares_event(void *opaque, int event, Error **errp)
static void colo_compare_timer_init(CompareState *s)
{
- AioContext *ctx = iothread_get_aio_context(s->iothread);
-
- s->packet_check_timer = aio_timer_new(ctx, QEMU_CLOCK_HOST,
+ s->packet_check_timer = aio_timer_new(s->iothread_ctx, QEMU_CLOCK_HOST,
SCALE_MS, check_old_packet_regular,
s);
timer_mod(s->packet_check_timer, qemu_clock_get_ms(QEMU_CLOCK_HOST) +
@@ -964,8 +963,15 @@ static void colo_compare_handle_event(void *opaque)
static void colo_compare_iothread(CompareState *s)
{
- AioContext *ctx = iothread_get_aio_context(s->iothread);
- object_ref(OBJECT(s->iothread));
+ g_autofree char *path = object_get_canonical_path(OBJECT(s));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+
+ AioContext *ctx = iothread_ref_and_get_aio_context(s->iothread, &io_holder);
+
+ s->iothread_ctx = ctx;
s->worker_context = iothread_get_g_main_context(s->iothread);
qemu_chr_fe_set_handlers(&s->chr_pri_in, compare_chr_can_read,
@@ -1404,6 +1410,7 @@ static void colo_compare_finalize(Object *obj)
{
CompareState *s = COLO_COMPARE(obj);
CompareState *tmp;
+ g_autofree char *path = object_get_canonical_path(OBJECT(s));
qemu_mutex_lock(&colo_compare_mutex);
QTAILQ_FOREACH(tmp, &net_compares, next) {
@@ -1430,18 +1437,20 @@ static void colo_compare_finalize(Object *obj)
g_clear_pointer(&s->event_bh, qemu_bh_delete);
if (s->iothread) {
- AioContext *ctx = iothread_get_aio_context(s->iothread);
-
- AIO_WAIT_WHILE(ctx, !s->out_sendco.done);
+ AIO_WAIT_WHILE(s->iothread_ctx, !s->out_sendco.done);
if (s->notify_dev) {
- AIO_WAIT_WHILE(ctx, !s->notify_sendco.done);
+ AIO_WAIT_WHILE(s->iothread_ctx, !s->notify_sendco.done);
}
/* Release all unhandled packets after compare thread exited */
g_queue_foreach(&s->conn_list, colo_flush_packets, s);
AIO_WAIT_WHILE(NULL, !s->out_sendco.done);
- object_unref(OBJECT(s->iothread));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+ iothread_unref_and_put_aio_context(s->iothread, &io_holder);
}
g_queue_clear(&s->conn_list);
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 09/14] virtio-balloon: Update tracking iothread users with holder
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (7 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 08/14] net/colo: track IOThread references using path-based holder Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 10/14] vfio-user/proxy: Update tracking iothread users with holder name Zhang Chen
` (4 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Replace raw object_ref/unref calls with
iothread_ref_and_get/unref_and_put_aio_context.
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
hw/virtio/virtio-balloon.c | 22 +++++++++++++++++-----
1 file changed, 17 insertions(+), 5 deletions(-)
diff --git a/hw/virtio/virtio-balloon.c b/hw/virtio/virtio-balloon.c
index 4c5f486ba2..7c9c8d2a41 100644
--- a/hw/virtio/virtio-balloon.c
+++ b/hw/virtio/virtio-balloon.c
@@ -892,14 +892,20 @@ static void virtio_balloon_device_realize(DeviceState *dev, Error **errp)
s->svq = virtio_add_queue(vdev, 128, virtio_balloon_receive_stats);
if (virtio_has_feature(s->host_features, VIRTIO_BALLOON_F_FREE_PAGE_HINT)) {
+ g_autofree char *path = object_get_canonical_path(OBJECT(s));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+
s->free_page_vq = virtio_add_queue(vdev, VIRTQUEUE_MAX_SIZE,
virtio_balloon_handle_free_page_vq);
precopy_add_notifier(&s->free_page_hint_notify);
- object_ref(OBJECT(s->iothread));
- s->free_page_bh = aio_bh_new_guarded(iothread_get_aio_context(s->iothread),
- virtio_ballloon_get_free_page_hints, s,
- &dev->mem_reentrancy_guard);
+ s->free_page_bh = aio_bh_new_guarded(
+ iothread_ref_and_get_aio_context(s->iothread, &io_holder),
+ virtio_ballloon_get_free_page_hints, s,
+ &dev->mem_reentrancy_guard);
}
if (virtio_has_feature(s->host_features, VIRTIO_BALLOON_F_REPORTING)) {
@@ -919,9 +925,15 @@ static void virtio_balloon_device_unrealize(DeviceState *dev)
qemu_unregister_resettable(OBJECT(dev));
if (s->free_page_bh) {
+ g_autofree char *path = object_get_canonical_path(OBJECT(s));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+
qemu_bh_delete(s->free_page_bh);
- object_unref(OBJECT(s->iothread));
virtio_balloon_free_page_stop(s);
+ iothread_unref_and_put_aio_context(s->iothread, &io_holder);
precopy_remove_notifier(&s->free_page_hint_notify);
}
balloon_stats_destroy_timer(s);
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 10/14] vfio-user/proxy: Update tracking iothread users with holder name
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (8 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 09/14] virtio-balloon: Update tracking iothread users with holder Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 11/14] xen-block: " Zhang Chen
` (3 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Add object_ref/unref calls with
iothread_ref_and_get/unref_and_put_aio_context.
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
hw/vfio-user/proxy.c | 16 +++++++++++++++-
1 file changed, 15 insertions(+), 1 deletion(-)
diff --git a/hw/vfio-user/proxy.c b/hw/vfio-user/proxy.c
index 197aee07bf..e642fb0ad3 100644
--- a/hw/vfio-user/proxy.c
+++ b/hw/vfio-user/proxy.c
@@ -898,6 +898,7 @@ VFIOUserProxy *vfio_user_connect_dev(SocketAddress *addr, Error **errp)
QIOChannelSocket *sioc;
QIOChannel *ioc;
char *sockname;
+ g_autofree char *path = NULL;
if (addr->type != SOCKET_ADDRESS_TYPE_UNIX) {
error_setg(errp, "vfio_user_connect - bad address family");
@@ -917,6 +918,11 @@ VFIOUserProxy *vfio_user_connect_dev(SocketAddress *addr, Error **errp)
proxy = g_malloc0(sizeof(VFIOUserProxy));
proxy->sockname = g_strdup_printf("unix:%s", sockname);
proxy->ioc = ioc;
+ path = object_get_canonical_path(OBJECT(proxy->ioc));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
/* init defaults */
proxy->max_xfer_size = VFIO_USER_DEF_MAX_XFER;
@@ -936,7 +942,8 @@ VFIOUserProxy *vfio_user_connect_dev(SocketAddress *addr, Error **errp)
vfio_user_iothread = iothread_create("vfio-user", errp);
}
- proxy->ctx = iothread_get_aio_context(vfio_user_iothread);
+ proxy->ctx = iothread_ref_and_get_aio_context(vfio_user_iothread,
+ &io_holder);
proxy->req_bh = qemu_bh_new(vfio_user_request, proxy);
QTAILQ_INIT(&proxy->outgoing);
@@ -967,6 +974,11 @@ void vfio_user_set_handler(VFIODevice *vbasedev,
void vfio_user_disconnect(VFIOUserProxy *proxy)
{
VFIOUserMsg *r1, *r2;
+ g_autofree char *path = object_get_canonical_path(OBJECT(proxy->ioc));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
qemu_mutex_lock(&proxy->lock);
@@ -1021,6 +1033,8 @@ void vfio_user_disconnect(VFIOUserProxy *proxy)
qemu_cond_destroy(&proxy->close_cv);
qemu_mutex_destroy(&proxy->lock);
+ iothread_unref_and_put_aio_context(vfio_user_iothread, &io_holder);
+
QLIST_REMOVE(proxy, next);
if (QLIST_EMPTY(&vfio_user_sockets)) {
iothread_destroy(vfio_user_iothread);
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 11/14] xen-block: Update tracking iothread users with holder name
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (9 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 10/14] vfio-user/proxy: Update tracking iothread users with holder name Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 12/14] monitor/hmp: display IOThread holders Zhang Chen
` (2 subsequent siblings)
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Replace raw object_ref/unref calls with
iothread_ref_and_get/unref_and_put_aio_context.
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
hw/block/dataplane/xen-block.c | 19 ++++++++++++++++---
1 file changed, 16 insertions(+), 3 deletions(-)
diff --git a/hw/block/dataplane/xen-block.c b/hw/block/dataplane/xen-block.c
index 48c2e315f3..9dfe06e1be 100644
--- a/hw/block/dataplane/xen-block.c
+++ b/hw/block/dataplane/xen-block.c
@@ -621,9 +621,15 @@ XenBlockDataPlane *xen_block_dataplane_create(XenDevice *xendev,
QLIST_INIT(&dataplane->freelist);
if (iothread) {
+ g_autofree char *path = object_get_canonical_path(OBJECT(xendev));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+
dataplane->iothread = iothread;
- object_ref(OBJECT(dataplane->iothread));
- dataplane->ctx = iothread_get_aio_context(dataplane->iothread);
+ dataplane->ctx = iothread_ref_and_get_aio_context(dataplane->iothread,
+ &io_holder);
} else {
dataplane->ctx = qemu_get_aio_context();
}
@@ -652,7 +658,14 @@ void xen_block_dataplane_destroy(XenBlockDataPlane *dataplane)
qemu_bh_delete(dataplane->bh);
if (dataplane->iothread) {
- object_unref(OBJECT(dataplane->iothread));
+ g_autofree char *path = object_get_canonical_path(
+ OBJECT(dataplane->xendev));
+ const IOThreadHolder io_holder = {
+ .type = IO_THREAD_HOLDER_KIND_QOM_OBJECT,
+ .u.qom_object.qom_path = path,
+ };
+
+ iothread_unref_and_put_aio_context(dataplane->iothread, &io_holder);
}
g_free(dataplane);
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 12/14] monitor/hmp: display IOThread holders
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (10 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 11/14] xen-block: " Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 6:37 ` [PATCH V15 13/14] iothread: make legacy AioContext getter explicitly unsafe Zhang Chen
2026-09-07 6:37 ` [PATCH V15 14/14] tests/unit/iothread: update AioContext ref/put helpers Zhang Chen
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Display holder information in info iothreads to aid manual debugging.
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
monitor/hmp-cmds.c | 20 ++++++++++++++++++++
1 file changed, 20 insertions(+)
diff --git a/monitor/hmp-cmds.c b/monitor/hmp-cmds.c
index 91701ddf33..d7dd24d951 100644
--- a/monitor/hmp-cmds.c
+++ b/monitor/hmp-cmds.c
@@ -246,11 +246,31 @@ void hmp_info_iothreads(MonitorHMP *hmp, const QDict *qdict)
IOThreadInfoList *info_list = qmp_query_iothreads(NULL);
IOThreadInfoList *info;
IOThreadInfo *value;
+ IOThreadHolderList *h;
for (info = info_list; info; info = info->next) {
value = info->value;
monitor_hmp_printf(hmp, "%s:\n", value->id);
monitor_hmp_printf(hmp, " thread_id=%" PRId64 "\n", value->thread_id);
+ monitor_hmp_printf(hmp, " holders=");
+ for (h = value->holders; h; h = h->next) {
+ IOThreadHolder *holder = h->value;
+
+ switch (holder->type) {
+ case IO_THREAD_HOLDER_KIND_BLOCK_EXPORT:
+ monitor_hmp_printf(hmp, "[block-export: %s]",
+ holder->u.block_export.export_id);
+ break;
+ case IO_THREAD_HOLDER_KIND_QOM_OBJECT:
+ monitor_hmp_printf(hmp, "[qom-path: %s]",
+ holder->u.qom_object.qom_path);
+ break;
+ default:
+ g_assert_not_reached();
+ }
+ }
+ monitor_hmp_printf(hmp, "\n");
+
monitor_hmp_printf(hmp, " poll-max-ns=%" PRId64 "\n", value->poll_max_ns);
monitor_hmp_printf(hmp, " poll-grow=%" PRId64 "\n", value->poll_grow);
monitor_hmp_printf(hmp, " poll-shrink=%" PRId64 "\n", value->poll_shrink);
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* [PATCH V15 13/14] iothread: make legacy AioContext getter explicitly unsafe
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (11 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 12/14] monitor/hmp: display IOThread holders Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
2026-09-07 12:55 ` Markus Armbruster
2026-09-07 6:37 ` [PATCH V15 14/14] tests/unit/iothread: update AioContext ref/put helpers Zhang Chen
13 siblings, 1 reply; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Now that holder-aware callers have been converted, rename
iothread_get_aio_context() to iothread_unsafe_get_aio_context() and
update qmp_x_blockdev_set_iothread(). This avoids adding a duplicate
accessor only to remove the old one later.
The block graph retains only the AioContext, while detachment happens
outside qmp_x_blockdev_set_iothread(), so this call site cannot provide
a matching holder lifecycle. The new name makes its existing lifetime
assumption explicit.
New code should use iothread_ref_and_get_aio_context().
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
blockdev.c | 7 ++++++-
include/system/iothread.h | 11 ++++++++++-
iothread.c | 2 +-
3 files changed, 17 insertions(+), 3 deletions(-)
diff --git a/blockdev.c b/blockdev.c
index 6e86c6262f..97d161079a 100644
--- a/blockdev.c
+++ b/blockdev.c
@@ -3683,7 +3683,12 @@ void qmp_x_blockdev_set_iothread(const char *node_name, StrOrNull *iothread,
goto out;
}
- new_context = iothread_get_aio_context(obj);
+ /*
+ * The block graph retains only the AioContext, not the IOThread.
+ * Context detachment happens outside this command, so there is no
+ * point here where a matching IOThread holder can be released.
+ */
+ new_context = iothread_unsafe_get_aio_context(obj);
} else {
new_context = qemu_get_aio_context();
}
diff --git a/include/system/iothread.h b/include/system/iothread.h
index f6c5f95dd9..9850cdb3dc 100644
--- a/include/system/iothread.h
+++ b/include/system/iothread.h
@@ -66,7 +66,16 @@ DECLARE_INSTANCE_CHECKER(IOThread, IOTHREAD,
char *iothread_get_id(IOThread *iothread);
IOThread *iothread_by_id(const char *id);
-AioContext *iothread_get_aio_context(IOThread *iothread);
+
+/*
+ * Return @iothread's AioContext without registering a holder or taking a
+ * reference on @iothread. The caller must ensure that the IOThread remains
+ * alive for as long as the returned AioContext is used.
+ *
+ * This API exists for legacy callers without a clear ref/unref lifecycle. Do
+ * not use it in new code; use iothread_ref_and_get_aio_context() instead.
+ */
+AioContext *iothread_unsafe_get_aio_context(IOThread *iothread);
/*
* Register @holder and return @iothread's AioContext. The holder is copied,
diff --git a/iothread.c b/iothread.c
index 58dace351c..ae9290c733 100644
--- a/iothread.c
+++ b/iothread.c
@@ -422,7 +422,7 @@ char *iothread_get_id(IOThread *iothread)
return g_strdup(object_get_canonical_path_component(OBJECT(iothread)));
}
-AioContext *iothread_get_aio_context(IOThread *iothread)
+AioContext *iothread_unsafe_get_aio_context(IOThread *iothread)
{
return iothread->ctx;
}
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread* Re: [PATCH V15 13/14] iothread: make legacy AioContext getter explicitly unsafe
2026-09-07 6:37 ` [PATCH V15 13/14] iothread: make legacy AioContext getter explicitly unsafe Zhang Chen
@ 2026-09-07 12:55 ` Markus Armbruster
2026-09-09 6:24 ` Zhang Chen
0 siblings, 1 reply; 26+ messages in thread
From: Markus Armbruster @ 2026-09-07 12:55 UTC (permalink / raw)
To: Zhang Chen
Cc: qemu-devel, Eric Blake, 'Michael S . Tsirkin',
Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Zhang Chen <zhangckid@gmail.com> writes:
> Now that holder-aware callers have been converted, rename
> iothread_get_aio_context() to iothread_unsafe_get_aio_context() and
> update qmp_x_blockdev_set_iothread(). This avoids adding a duplicate
> accessor only to remove the old one later.
The last sentence appears to justify the change since v14. If that's
the intent, drop it, please. If not, what is the intent?
> The block graph retains only the AioContext, while detachment happens
> outside qmp_x_blockdev_set_iothread(), so this call site cannot provide
> a matching holder lifecycle. The new name makes its existing lifetime
> assumption explicit.
Suggest to start like "The only remaining user is
qmp_x_blockdev_set_iothread()" and then explain why this caller cannot
use iothread_ref_and_get_aio_context().
> New code should use iothread_ref_and_get_aio_context().
>
> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
> Signed-off-by: Zhang Chen <zhangckid@gmail.com>
> ---
> blockdev.c | 7 ++++++-
> include/system/iothread.h | 11 ++++++++++-
> iothread.c | 2 +-
> 3 files changed, 17 insertions(+), 3 deletions(-)
>
> diff --git a/blockdev.c b/blockdev.c
> index 6e86c6262f..97d161079a 100644
> --- a/blockdev.c
> +++ b/blockdev.c
> @@ -3683,7 +3683,12 @@ void qmp_x_blockdev_set_iothread(const char *node_name, StrOrNull *iothread,
> goto out;
> }
>
> - new_context = iothread_get_aio_context(obj);
> + /*
> + * The block graph retains only the AioContext, not the IOThread.
> + * Context detachment happens outside this command, so there is no
> + * point here where a matching IOThread holder can be released.
> + */
This is fine for readers who already know what this is about: why we
can't use iothread_ref_and_get_aio_context() /
iothread_unsafe_get_aio_context() here. Others will lack context.
Suggest to start like
/*
* We cannot use iothread_ref_and_get_aio_context() /
* iothread_unref_and_put_aio_context() here.
then explain why.
> + new_context = iothread_unsafe_get_aio_context(obj);
> } else {
> new_context = qemu_get_aio_context();
> }
[...]
^ permalink raw reply [flat|nested] 26+ messages in thread* Re: [PATCH V15 13/14] iothread: make legacy AioContext getter explicitly unsafe
2026-09-07 12:55 ` Markus Armbruster
@ 2026-09-09 6:24 ` Zhang Chen
2026-09-09 10:41 ` Markus Armbruster
0 siblings, 1 reply; 26+ messages in thread
From: Zhang Chen @ 2026-09-09 6:24 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel, Eric Blake, Michael S . Tsirkin, Stefan Hajnoczi,
Paolo Bonzini, Daniel P . Berrangé, Jason Wang
On Mon, Sep 7, 2026 at 8:55 PM Markus Armbruster <armbru@redhat.com> wrote:
>
> Zhang Chen <zhangckid@gmail.com> writes:
>
> > Now that holder-aware callers have been converted, rename
> > iothread_get_aio_context() to iothread_unsafe_get_aio_context() and
> > update qmp_x_blockdev_set_iothread(). This avoids adding a duplicate
> > accessor only to remove the old one later.
>
> The last sentence appears to justify the change since v14. If that's
> the intent, drop it, please. If not, what is the intent?
Thanks. Yes, that sentence was about the changes since v14; I'll
drop it from the commit message.
>
> > The block graph retains only the AioContext, while detachment happens
> > outside qmp_x_blockdev_set_iothread(), so this call site cannot provide
> > a matching holder lifecycle. The new name makes its existing lifetime
> > assumption explicit.
>
> Suggest to start like "The only remaining user is
> qmp_x_blockdev_set_iothread()" and then explain why this caller cannot
> use iothread_ref_and_get_aio_context().
>
> > New code should use iothread_ref_and_get_aio_context().
> >
> > Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
> > Signed-off-by: Zhang Chen <zhangckid@gmail.com>
> > ---
> > blockdev.c | 7 ++++++-
> > include/system/iothread.h | 11 ++++++++++-
> > iothread.c | 2 +-
> > 3 files changed, 17 insertions(+), 3 deletions(-)
> >
> > diff --git a/blockdev.c b/blockdev.c
> > index 6e86c6262f..97d161079a 100644
> > --- a/blockdev.c
> > +++ b/blockdev.c
> > @@ -3683,7 +3683,12 @@ void qmp_x_blockdev_set_iothread(const char *node_name, StrOrNull *iothread,
> > goto out;
> > }
> >
> > - new_context = iothread_get_aio_context(obj);
> > + /*
> > + * The block graph retains only the AioContext, not the IOThread.
> > + * Context detachment happens outside this command, so there is no
> > + * point here where a matching IOThread holder can be released.
> > + */
>
> This is fine for readers who already know what this is about: why we
> can't use iothread_ref_and_get_aio_context() /
> iothread_unsafe_get_aio_context() here. Others will lack context.
>
> Suggest to start like
>
> /*
> * We cannot use iothread_ref_and_get_aio_context() /
> * iothread_unref_and_put_aio_context() here.
>
> then explain why.
OK, will change it to:
/*
* We cannot use iothread_ref_and_get_aio_context() /
* iothread_unref_and_put_aio_context() here. The block graph stores
* the AioContext but not the IOThread, and continues using the
* context after this command returns. A matching put would need
* to be tied to the block graph's context lifecycle.
*/
Thanks
Chen
>
> > + new_context = iothread_unsafe_get_aio_context(obj);
> > } else {
> > new_context = qemu_get_aio_context();
> > }
>
> [...]
>
^ permalink raw reply [flat|nested] 26+ messages in thread* Re: [PATCH V15 13/14] iothread: make legacy AioContext getter explicitly unsafe
2026-09-09 6:24 ` Zhang Chen
@ 2026-09-09 10:41 ` Markus Armbruster
0 siblings, 0 replies; 26+ messages in thread
From: Markus Armbruster @ 2026-09-09 10:41 UTC (permalink / raw)
To: Zhang Chen
Cc: qemu-devel, Eric Blake, Michael S . Tsirkin, Stefan Hajnoczi,
Paolo Bonzini, Daniel P . Berrangé, Jason Wang
Zhang Chen <zhangckid@gmail.com> writes:
> On Mon, Sep 7, 2026 at 8:55 PM Markus Armbruster <armbru@redhat.com> wrote:
>>
>> Zhang Chen <zhangckid@gmail.com> writes:
>>
>> > Now that holder-aware callers have been converted, rename
>> > iothread_get_aio_context() to iothread_unsafe_get_aio_context() and
>> > update qmp_x_blockdev_set_iothread(). This avoids adding a duplicate
>> > accessor only to remove the old one later.
>>
>> The last sentence appears to justify the change since v14. If that's
>> the intent, drop it, please. If not, what is the intent?
>
> Thanks. Yes, that sentence was about the changes since v14; I'll
> drop it from the commit message.
>
>>
>> > The block graph retains only the AioContext, while detachment happens
>> > outside qmp_x_blockdev_set_iothread(), so this call site cannot provide
>> > a matching holder lifecycle. The new name makes its existing lifetime
>> > assumption explicit.
>>
>> Suggest to start like "The only remaining user is
>> qmp_x_blockdev_set_iothread()" and then explain why this caller cannot
>> use iothread_ref_and_get_aio_context().
>>
>> > New code should use iothread_ref_and_get_aio_context().
>> >
>> > Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
>> > Signed-off-by: Zhang Chen <zhangckid@gmail.com>
>> > ---
>> > blockdev.c | 7 ++++++-
>> > include/system/iothread.h | 11 ++++++++++-
>> > iothread.c | 2 +-
>> > 3 files changed, 17 insertions(+), 3 deletions(-)
>> >
>> > diff --git a/blockdev.c b/blockdev.c
>> > index 6e86c6262f..97d161079a 100644
>> > --- a/blockdev.c
>> > +++ b/blockdev.c
>> > @@ -3683,7 +3683,12 @@ void qmp_x_blockdev_set_iothread(const char *node_name, StrOrNull *iothread,
>> > goto out;
>> > }
>> >
>> > - new_context = iothread_get_aio_context(obj);
>> > + /*
>> > + * The block graph retains only the AioContext, not the IOThread.
>> > + * Context detachment happens outside this command, so there is no
>> > + * point here where a matching IOThread holder can be released.
>> > + */
>>
>> This is fine for readers who already know what this is about: why we
>> can't use iothread_ref_and_get_aio_context() /
>> iothread_unsafe_get_aio_context() here. Others will lack context.
>>
>> Suggest to start like
>>
>> /*
>> * We cannot use iothread_ref_and_get_aio_context() /
>> * iothread_unref_and_put_aio_context() here.
>>
>> then explain why.
>
> OK, will change it to:
> /*
> * We cannot use iothread_ref_and_get_aio_context() /
> * iothread_unref_and_put_aio_context() here. The block graph stores
> * the AioContext but not the IOThread, and continues using the
> * context after this command returns. A matching put would need
> * to be tied to the block graph's context lifecycle.
> */
Works for me, thanks!
^ permalink raw reply [flat|nested] 26+ messages in thread
* [PATCH V15 14/14] tests/unit/iothread: update AioContext ref/put helpers
2026-09-07 6:37 [PATCH V15 00/14] iothread: Support tracking and querying IOThread holder Zhang Chen
` (12 preceding siblings ...)
2026-09-07 6:37 ` [PATCH V15 13/14] iothread: make legacy AioContext getter explicitly unsafe Zhang Chen
@ 2026-09-07 6:37 ` Zhang Chen
13 siblings, 0 replies; 26+ messages in thread
From: Zhang Chen @ 2026-09-07 6:37 UTC (permalink / raw)
To: qemu-devel, Eric Blake, Markus Armbruster,
'Michael S . Tsirkin', Stefan Hajnoczi, Paolo Bonzini,
'Daniel P . Berrangé', Jason Wang
Cc: Zhang Chen
Update the unit-test IOThread stubs and callers to use
iothread_ref_and_get_aio_context() together with the matching
iothread_unref_and_put_aio_context().
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
---
tests/unit/iothread.c | 18 ++++++++++++------
tests/unit/iothread.h | 7 ++++++-
tests/unit/test-aio-multithread.c | 5 ++++-
tests/unit/test-bdrv-drain.c | 18 +++++++++++++-----
tests/unit/test-block-iothread.c | 21 ++++++++++++++-------
5 files changed, 49 insertions(+), 20 deletions(-)
diff --git a/tests/unit/iothread.c b/tests/unit/iothread.c
index a363bf8f70..790cd2dce7 100644
--- a/tests/unit/iothread.c
+++ b/tests/unit/iothread.c
@@ -30,12 +30,23 @@ struct IOThread {
bool stopping;
};
+AioContext *iothread_ref_and_get_aio_context(IOThread *iothread,
+ IOThreadHolder *holder)
+{
+ return iothread->ctx;
+}
+
+void iothread_unref_and_put_aio_context(IOThread *iothread,
+ IOThreadHolder *holder)
+{
+}
+
static void iothread_init_gcontext(IOThread *iothread)
{
GSource *source;
iothread->worker_context = g_main_context_new();
- source = aio_get_g_source(iothread_get_aio_context(iothread));
+ source = aio_get_g_source(iothread->ctx);
g_source_attach(source, iothread->worker_context);
g_source_unref(source);
iothread->main_loop = g_main_loop_new(iothread->worker_context, TRUE);
@@ -113,8 +124,3 @@ IOThread *iothread_new(void)
qemu_mutex_unlock(&iothread->init_done_lock);
return iothread;
}
-
-AioContext *iothread_get_aio_context(IOThread *iothread)
-{
- return iothread->ctx;
-}
diff --git a/tests/unit/iothread.h b/tests/unit/iothread.h
index eb4d0c77f8..9bdac074c0 100644
--- a/tests/unit/iothread.h
+++ b/tests/unit/iothread.h
@@ -17,9 +17,14 @@
#include "qemu/thread.h"
typedef struct IOThread IOThread;
+typedef struct IOThreadHolder IOThreadHolder;
+
+AioContext *iothread_ref_and_get_aio_context(IOThread *iothread,
+ IOThreadHolder *holder);
+void iothread_unref_and_put_aio_context(IOThread *iothread,
+ IOThreadHolder *holder);
IOThread *iothread_new(void);
void iothread_join(IOThread *iothread);
-AioContext *iothread_get_aio_context(IOThread *iothread);
#endif
diff --git a/tests/unit/test-aio-multithread.c b/tests/unit/test-aio-multithread.c
index 9179cdc6a3..66ba5a466b 100644
--- a/tests/unit/test-aio-multithread.c
+++ b/tests/unit/test-aio-multithread.c
@@ -69,7 +69,7 @@ static void create_aio_contexts(void)
for (i = 0; i < NUM_CONTEXTS; i++) {
threads[i] = iothread_new();
- ctx[i] = iothread_get_aio_context(threads[i]);
+ ctx[i] = iothread_ref_and_get_aio_context(threads[i], NULL);
}
qemu_event_init(&done_event, false);
@@ -87,6 +87,9 @@ static void join_aio_contexts(void)
for (i = 0; i < NUM_CONTEXTS; i++) {
aio_context_ref(ctx[i]);
}
+ for (i = 0; i < NUM_CONTEXTS; i++) {
+ iothread_unref_and_put_aio_context(threads[i], NULL);
+ }
for (i = 0; i < NUM_CONTEXTS; i++) {
iothread_join(threads[i]);
}
diff --git a/tests/unit/test-bdrv-drain.c b/tests/unit/test-bdrv-drain.c
index 43b0ba8648..05c31ffca8 100644
--- a/tests/unit/test-bdrv-drain.c
+++ b/tests/unit/test-bdrv-drain.c
@@ -537,8 +537,8 @@ static void test_iothread_common(enum drain_type drain_type, int drain_thread)
IOThread *a = iothread_new();
IOThread *b = iothread_new();
- AioContext *ctx_a = iothread_get_aio_context(a);
- AioContext *ctx_b = iothread_get_aio_context(b);
+ AioContext *ctx_a = iothread_ref_and_get_aio_context(a, NULL);
+ AioContext *ctx_b = iothread_ref_and_get_aio_context(b, NULL);
QEMUIOVector qiov = QEMU_IOVEC_INIT_BUF(qiov, NULL, 0);
@@ -612,6 +612,8 @@ static void test_iothread_common(enum drain_type drain_type, int drain_thread)
bdrv_unref(bs);
blk_unref(blk);
+ iothread_unref_and_put_aio_context(a, NULL);
+ iothread_unref_and_put_aio_context(b, NULL);
out:
iothread_join(a);
@@ -762,7 +764,7 @@ static void test_blockjob_common_drain_node(enum drain_type drain_type,
AioContext *ctx;
iothread = iothread_new();
- ctx = iothread_get_aio_context(iothread);
+ ctx = iothread_ref_and_get_aio_context(iothread, NULL);
blk_set_aio_context(blk_src, ctx, &error_abort);
}
@@ -892,6 +894,10 @@ static void test_blockjob_common_drain_node(enum drain_type drain_type,
bdrv_unref(src_overlay);
bdrv_unref(target);
+ if (use_iothread) {
+ iothread_unref_and_put_aio_context(iothread, NULL);
+ }
+
if (iothread) {
iothread_join(iothread);
}
@@ -1398,8 +1404,8 @@ static void test_set_aio_context(void)
BlockDriverState *bs;
IOThread *a = iothread_new();
IOThread *b = iothread_new();
- AioContext *ctx_a = iothread_get_aio_context(a);
- AioContext *ctx_b = iothread_get_aio_context(b);
+ AioContext *ctx_a = iothread_ref_and_get_aio_context(a, NULL);
+ AioContext *ctx_b = iothread_ref_and_get_aio_context(b, NULL);
bs = bdrv_new_open_driver(&bdrv_test, "test-node", BDRV_O_RDWR,
&error_abort);
@@ -1410,6 +1416,8 @@ static void test_set_aio_context(void)
bdrv_try_change_aio_context(bs, qemu_get_aio_context(), NULL, &error_abort);
bdrv_unref(bs);
+ iothread_unref_and_put_aio_context(a, NULL);
+ iothread_unref_and_put_aio_context(b, NULL);
iothread_join(a);
iothread_join(b);
}
diff --git a/tests/unit/test-block-iothread.c b/tests/unit/test-block-iothread.c
index 5273ff235a..fc9735d056 100644
--- a/tests/unit/test-block-iothread.c
+++ b/tests/unit/test-block-iothread.c
@@ -466,7 +466,7 @@ static void test_sync_op(const void *opaque)
{
const SyncOpTest *t = opaque;
IOThread *iothread = iothread_new();
- AioContext *ctx = iothread_get_aio_context(iothread);
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread, NULL);
BlockBackend *blk;
BlockDriverState *bs;
BdrvChild *c;
@@ -493,6 +493,7 @@ static void test_sync_op(const void *opaque)
bdrv_unref(bs);
blk_unref(blk);
+ iothread_unref_and_put_aio_context(iothread, NULL);
}
typedef struct TestBlockJob {
@@ -549,7 +550,7 @@ BlockJobDriver test_job_driver = {
static void test_attach_blockjob(void)
{
IOThread *iothread = iothread_new();
- AioContext *ctx = iothread_get_aio_context(iothread);
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread, NULL);
BlockBackend *blk;
BlockDriverState *bs;
TestBlockJob *tjob;
@@ -595,6 +596,7 @@ static void test_attach_blockjob(void)
bdrv_unref(bs);
blk_unref(blk);
+ iothread_unref_and_put_aio_context(iothread, NULL);
}
/*
@@ -612,7 +614,7 @@ static void test_attach_blockjob(void)
static void test_propagate_basic(void)
{
IOThread *iothread = iothread_new();
- AioContext *ctx = iothread_get_aio_context(iothread);
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread, NULL);
AioContext *main_ctx;
BlockBackend *blk;
BlockDriverState *bs_a, *bs_b, *bs_verify;
@@ -658,6 +660,7 @@ static void test_propagate_basic(void)
bdrv_unref(bs_b);
bdrv_unref(bs_a);
blk_unref(blk);
+ iothread_unref_and_put_aio_context(iothread, NULL);
}
/*
@@ -676,7 +679,7 @@ static void test_propagate_basic(void)
static void test_propagate_diamond(void)
{
IOThread *iothread = iothread_new();
- AioContext *ctx = iothread_get_aio_context(iothread);
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread, NULL);
AioContext *main_ctx;
BlockBackend *blk;
BlockDriverState *bs_a, *bs_b, *bs_c, *bs_verify;
@@ -736,12 +739,13 @@ static void test_propagate_diamond(void)
bdrv_unref(bs_c);
bdrv_unref(bs_b);
bdrv_unref(bs_a);
+ iothread_unref_and_put_aio_context(iothread, NULL);
}
static void test_propagate_mirror(void)
{
IOThread *iothread = iothread_new();
- AioContext *ctx = iothread_get_aio_context(iothread);
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread, NULL);
AioContext *main_ctx = qemu_get_aio_context();
BlockDriverState *src, *target, *filter;
BlockBackend *blk;
@@ -807,12 +811,13 @@ static void test_propagate_mirror(void)
blk_unref(blk);
bdrv_unref(src);
bdrv_unref(target);
+ iothread_unref_and_put_aio_context(iothread, NULL);
}
static void test_attach_second_node(void)
{
IOThread *iothread = iothread_new();
- AioContext *ctx = iothread_get_aio_context(iothread);
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread, NULL);
AioContext *main_ctx = qemu_get_aio_context();
BlockBackend *blk;
BlockDriverState *bs, *filter;
@@ -840,12 +845,13 @@ static void test_attach_second_node(void)
bdrv_unref(filter);
bdrv_unref(bs);
blk_unref(blk);
+ iothread_unref_and_put_aio_context(iothread, NULL);
}
static void test_attach_preserve_blk_ctx(void)
{
IOThread *iothread = iothread_new();
- AioContext *ctx = iothread_get_aio_context(iothread);
+ AioContext *ctx = iothread_ref_and_get_aio_context(iothread, NULL);
BlockBackend *blk;
BlockDriverState *bs;
@@ -871,6 +877,7 @@ static void test_attach_preserve_blk_ctx(void)
blk_set_aio_context(blk, qemu_get_aio_context(), &error_abort);
bdrv_unref(bs);
blk_unref(blk);
+ iothread_unref_and_put_aio_context(iothread, NULL);
}
int main(int argc, char **argv)
--
2.43.0
^ permalink raw reply related [flat|nested] 26+ messages in thread