Fix uninitialized access to InitialRunningXacts during decoding after ERROR.

The transactions and subtransactions array that was allocated under
snapshot builder memory context and recorded during decoding was not
cleared in case of errors. This can result in an assertion failure if we
attempt to retry logical decoding within the same session. To address this
issue, we register a callback function under the snapshot builder memory
context to clear the recorded transactions and subtransactions array along
with the context.

This problem doesn't exist in PG16 and HEAD as instead of using
InitialRunningXacts, we added the list of transaction IDs and
sub-transaction IDs, that have modified catalogs and are running during
snapshot serialization, to the serialized snapshot (see commit 7f13ac8123).

Author: Hou Zhijie
Reviewed-by: Amit Kapila
Backpatch-through: 11
Discussion: http://postgr.es/m/18055-ab3beed9f4b7b7d6@postgresql.org

diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c
index 02d73e5c9396f9d57b6f809c24b545bfeacc1f2d..e32ed39e28241341cb96b2ff8bbe8bc7e9913804 100644 (file)
--- a/src/backend/replication/logical/snapbuild.c
+++ b/src/backend/replication/logical/snapbuild.c
@@ -336,6 +336,17 @@ SnapBuildStartNextPhaseAt(SnapBuild *builder, TransactionId at)
        builder->was_running.was_xmax = at;
 }
 
+/*
+ * Memory context reset callback for clearing the array of running transactions
+ * and subtransactions.
+ */
+static void
+SnapBuildResetRunningXactsCallback(void *arg)
+{
+       NInitialRunningXacts = 0;
+       InitialRunningXacts = NULL;
+}
+
 /*
  * Allocate a new snapshot builder.
  *
@@ -351,6 +362,7 @@ AllocateSnapshotBuilder(ReorderBuffer *reorder,
        MemoryContext context;
        MemoryContext oldcontext;
        SnapBuild  *builder;
+       MemoryContextCallback *mcallback;
 
        /* allocate memory in own context, to have better accountability */
        context = AllocSetContextCreate(CurrentMemoryContext,
@@ -375,6 +387,10 @@ AllocateSnapshotBuilder(ReorderBuffer *reorder,
        builder->start_decoding_at = start_lsn;
        builder->building_full_snapshot = need_full_snapshot;
 
+       mcallback = palloc0(sizeof(MemoryContextCallback));
+       mcallback->func = SnapBuildResetRunningXactsCallback;
+       MemoryContextRegisterResetCallback(CurrentMemoryContext, mcallback);
+
        MemoryContextSwitchTo(oldcontext);
 
        /* The initial running transactions array must be empty. */
@@ -400,10 +416,6 @@ FreeSnapshotBuilder(SnapBuild *builder)
 
        /* other resources are deallocated via memory context reset */
        MemoryContextDelete(context);
-
-       /* InitialRunningXacts is freed along with the context */
-       NInitialRunningXacts = 0;
-       InitialRunningXacts = NULL;
 }
 
 /*