diff mbox series

[v7,2/2] memory: Do not create circular reference with subregion

Message ID 20250109-san-v7-2-93c432a73024@daynix.com (mailing list archive)
State New
Headers show
Series Fix check-qtest-ppc64 sanitizer errors | expand

Commit Message

Akihiko Odaki Jan. 9, 2025, 5:50 a.m. UTC
memory_region_update_container_subregions() used to call
memory_region_ref(), which creates a reference to the owner of the
subregion, on behalf of the owner of the container. This results in a
circular reference if the subregion and container have the same owner.

memory_region_ref() creates a reference to the owner instead of the
memory region to match the lifetime of the owner and memory region. We
do not need such a hack if the subregion and container have the same
owner because the owner will be alive as long as the container is.
Therefore, create a reference to the subregion itself instead ot its
owner in such a case; the reference to the subregion is still necessary
to ensure that the subregion gets finalized after the container.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
 system/memory.c | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

Comments

Peter Xu Jan. 9, 2025, 3:55 p.m. UTC | #1
On Thu, Jan 09, 2025 at 02:50:21PM +0900, Akihiko Odaki wrote:
> memory_region_update_container_subregions() used to call
> memory_region_ref(), which creates a reference to the owner of the
> subregion, on behalf of the owner of the container. This results in a
> circular reference if the subregion and container have the same owner.
> 
> memory_region_ref() creates a reference to the owner instead of the
> memory region to match the lifetime of the owner and memory region. We
> do not need such a hack if the subregion and container have the same
> owner because the owner will be alive as long as the container is.
> Therefore, create a reference to the subregion itself instead ot its
> owner in such a case; the reference to the subregion is still necessary
> to ensure that the subregion gets finalized after the container.
> 
> Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>

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

Patch

diff --git a/system/memory.c b/system/memory.c
index 78e17e0efa83..972617b0192d 100644
--- a/system/memory.c
+++ b/system/memory.c
@@ -1855,6 +1855,24 @@  void memory_region_unref(MemoryRegion *mr)
     }
 }
 
+static void memory_region_ref_subregion(MemoryRegion *subregion)
+{
+    if (subregion->container->owner == subregion->owner) {
+        object_ref(OBJECT(subregion));
+    } else {
+        memory_region_ref(subregion);
+    }
+}
+
+static void memory_region_unref_subregion(MemoryRegion *subregion)
+{
+    if (subregion->container->owner == subregion->owner) {
+        object_unref(OBJECT(subregion));
+     } else {
+        memory_region_unref(subregion);
+     }
+}
+
 uint64_t memory_region_size(MemoryRegion *mr)
 {
     if (int128_eq(mr->size, int128_2_64())) {
@@ -2644,7 +2662,8 @@  static void memory_region_update_container_subregions(MemoryRegion *subregion)
 
     memory_region_transaction_begin();
 
-    memory_region_ref(subregion);
+    memory_region_ref_subregion(subregion);
+
     QTAILQ_FOREACH(other, &mr->subregions, subregions_link) {
         if (subregion->priority >= other->priority) {
             QTAILQ_INSERT_BEFORE(other, subregion, subregions_link);
@@ -2702,7 +2721,8 @@  void memory_region_del_subregion(MemoryRegion *mr,
         assert(alias->mapped_via_alias >= 0);
     }
     QTAILQ_REMOVE(&mr->subregions, subregion, subregions_link);
-    memory_region_unref(subregion);
+    memory_region_unref_subregion(subregion);
+
     memory_region_update_pending |= mr->enabled && subregion->enabled;
     memory_region_transaction_commit();
 }