From: Fiona Ebner <f.ebner@proxmox.com>
To: pve-devel@lists.proxmox.com
Subject: [PATCH qemu 3/3] async snapshot: keep track of savevm-end status
Date: Wed, 18 Nov 2026 13:21:49 +0100 [thread overview]
Message-ID: <20261118122359.123647-4-f.ebner@proxmox.com> (raw)
In-Reply-To: <20261118122359.123647-1-f.ebner@proxmox.com>
If a snapshot ends with an error, qemu-server had no way to detect
when it's safe to consider 'savevm-end' done. The result of
'query-savevm' only changes when the status is set to SAVE_STATE_DONE,
not reporting the 'bytes' value anymore, and qemu-server looked at
that. Keep track of the 'savevm-end' state and report it to allow
distinguishing between 'savevm-end' being done or still waiting for
the target file close in case of an error.
Checking for an already active 'savevm-end' invocation at the
beginning of qmp_savevm_end() fixes a crash: when a second
'savevm-end' QMP call came in, while a previous one was still sleeping
in target_close_wait, the following assertion would trigger:
> kvm: ../util/qemu-coroutine-sleep.c:77: qemu_co_sleep: \
> Assertion `prev == NULL' failed.
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
migration/savevm-async.c | 13 +++++++++++++
qapi/migration.json | 6 +++++-
2 files changed, 18 insertions(+), 1 deletion(-)
diff --git a/migration/savevm-async.c b/migration/savevm-async.c
index 3ab0bd4539..e9037173af 100644
--- a/migration/savevm-async.c
+++ b/migration/savevm-async.c
@@ -60,6 +60,7 @@ static struct SnapshotState {
Coroutine *co;
QemuCoSleep target_close_wait;
IOThread *iothread;
+ bool savevm_end_active;
} snap_state;
static bool savevm_aborted(void)
@@ -77,6 +78,8 @@ SaveVMInfo *qmp_query_savevm(Error **errp)
SaveVMInfo *info = g_malloc0(sizeof(*info));
struct SnapshotState *s = &snap_state;
+ info->savevm_end_active = s->savevm_end_active;
+
if (s->state != SAVE_STATE_DONE) {
info->has_bytes = true;
info->bytes = s->bs_pos;
@@ -498,6 +501,7 @@ static void coroutine_fn wait_for_close_co(void *opaque)
"qmp_savevm_end");
/* we cannot assume the snapshot finished in this case, so leave the
* state alone - caller has to figure something out */
+ snap_state.savevm_end_active = false;
return;
}
} else {
@@ -511,6 +515,8 @@ static void coroutine_fn wait_for_close_co(void *opaque)
snap_state.state = SAVE_STATE_DONE;
}
+ snap_state.savevm_end_active = false;
+
DPRINTF("savevm-end: cleanup done\n");
}
@@ -521,6 +527,13 @@ void qmp_savevm_end(Error **errp)
return;
}
+ if (snap_state.savevm_end_active) {
+ error_setg(errp, "savevm-end handler already running");
+ return;
+ }
+
+ snap_state.savevm_end_active = true;
+
Coroutine *wait_for_close = qemu_coroutine_create(wait_for_close_co, NULL);
if (snap_state.state == SAVE_STATE_ACTIVE) {
diff --git a/qapi/migration.json b/qapi/migration.json
index 0794a3a7a3..d7504a0d4d 100644
--- a/qapi/migration.json
+++ b/qapi/migration.json
@@ -354,11 +354,15 @@
#
# @bytes: total amount of data transfered
#
+# @savevm-end-active: Whether a @savevm-end handler is currently
+# active.
+#
# Since: 1.3
##
{ 'struct': 'SaveVMInfo',
'data': {'*status': 'str', '*error': 'str',
- '*total-time': 'int', '*bytes': 'int'} }
+ '*total-time': 'int', '*bytes': 'int',
+ 'savevm-end-active': 'bool'} }
##
# @query-savevm:
--
2.47.3
next prev parent reply other threads:[~2026-11-18 12:24 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-18 12:21 [PATCH-SERIES qemu-server/qemu 0/3] fix #8331: snapshot: fix regression with endless loop after error Fiona Ebner
2026-11-18 12:21 ` [PATCH qemu-server 1/3] " Fiona Ebner
2026-11-18 12:21 ` [PATCH qemu-server 2/3] snapshot: directly check for savevm-end status Fiona Ebner
2026-11-18 12:21 ` Fiona Ebner [this message]
2026-11-18 21:02 ` applied: [PATCH-SERIES qemu-server/qemu 0/3] fix #8331: snapshot: fix regression with endless loop after error Thomas Lamprecht
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20261118122359.123647-4-f.ebner@proxmox.com \
--to=f.ebner@proxmox.com \
--cc=pve-devel@lists.proxmox.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox