The multifd threads still may access the colo cache, so release it
only after they terminate.

Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Lukas Straub <lukasstraub2@web.de>
---
 migration/colo.c      | 3 ---
 migration/migration.c | 3 +++
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/migration/colo.c b/migration/colo.c
index 8dfd39b035c48590fcebeb20459f01fb37fb67d1..d3534d1a32ad82f02101ac092ebf818a0caee6f2 100644
--- a/migration/colo.c
+++ b/migration/colo.c
@@ -947,7 +947,4 @@ void coroutine_fn colo_incoming_co(void)
     /* Wait checkpoint incoming thread exit before free resource */
     qemu_thread_join(&th);
     bql_lock();
-
-    /* We hold the global BQL, so it is safe here */
-    colo_release_ram_cache();
 }
diff --git a/migration/migration.c b/migration/migration.c
index 5515be1bf305b40ba0b590136df18a53451872c5..9e3f73f27766196ea8673bf9a58c97d5b8b1672f 100644
--- a/migration/migration.c
+++ b/migration/migration.c
@@ -454,6 +454,9 @@ void migration_incoming_state_destroy(void)
      * BQL and retake unconditionally.
      */
     assert(bql_locked());
+    if (migrate_colo()) {
+        colo_release_ram_cache();
+    }
     qemu_loadvm_state_cleanup(mis);
 
     if (mis->to_src_file) {

-- 
2.39.5

On Tue, Feb 03, 2026 at 11:15:16AM +0100, Lukas Straub wrote:
> The multifd threads still may access the colo cache, so release it
> only after they terminate.
> 
> Reviewed-by: Fabiano Rosas <farosas@suse.de>
> Signed-off-by: Lukas Straub <lukasstraub2@web.de>

Reviewed-by: Peter Xu <peterx@redhat.com>

-- 
Peter Xu