Hi, Sergey! On Apr 24, Sergey Vojtovich wrote:

revision-id: 70c5ab5658d (mariadb-10.4.4-29-g70c5ab5658d) parent(s): c5b1799435b author: Sergey Vojtovich committer: Sergey Vojtovich timestamp: 2019-04-20 14:45:16 +0400 message:

Move XID_STATE::rm_error to XID_cache_element

It's somewhat confusing. What if it's an internal 2PC and some engine's rollback returns an error? The error won't be stored in xid_state.rm_error anymore. Basically, you're changing the semantics of rm_error from "an error that happened during rollback" to "an error that happened during rollback or an external XA transaction". Mention this change explicitly in the commit comment, please.

Also removed redundant xid reset from THD::init_for_queries(). Must've been done already either by THD::transaction constructor or by THD::cleanup().

Part of MDEV-7974 - backport fix for mysql bug#12161 (XA and binlog)

diff --git a/sql/sql_class.h b/sql/sql_class.h index c68628be65d..e7584a0d825 100644 --- a/sql/sql_class.h +++ b/sql/sql_class.h @@ -2606,13 +2606,8 @@ class THD: public THD_count, /* this must be first */ DBUG_ENTER("thd::cleanup");

rename to "THD::st_transaction::cleanup" please

changed_tables= 0; savepoints= 0; - /* - If rm_error is raised, it means that this piece of a distributed - transaction has failed and must be rolled back. But the user must - rollback it explicitly, so don't start a new distributed XA until - then. - */ - if (!xid_state.rm_error) + /* xid_cache_delete() resets xid of explicitly started XA transaction */ + if (!xid_state.is_explicit_XA()) xid_state.xid.null(); free_root(&mem_root,MYF(MY_KEEP_PREALLOC)); DBUG_VOID_RETURN;

Regards, Sergei Chief Architect MariaDB and security@mariadb.org