From 3e40f9a7f3bbe82d96c8acccbb017deebfa00647 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marko=20M=C3=A4kel=C3=A4?= <marko.makela@mariadb.com>
Date: Thu, 8 Jun 2023 09:17:52 +0300
Subject: [PATCH] MDEV-31355 innodb_undo_log_truncate=ON fails to wait for
 purge of enough transaction history

purge_sys_t::sees(): Wrapper for view.sees().

trx_purge_truncate_history(): Invoke purge_sys.sees() instead of
comparing to head.trx_no, to determine if undo pages can be safely freed.

The test innodb.cursor-restore-locking was adjusted by Vladislav Lesin,
as was the the debug instrumentation in row_purge_del_mark().

Reviewed by: Vladislav Lesin
---
 .../gcol/r/innodb_virtual_debug_purge.result  |  2 +-
 .../gcol/t/innodb_virtual_debug_purge.test    |  2 ++
 .../innodb/r/cursor-restore-locking.result    | 13 ++++---
 mysql-test/suite/innodb/r/dml_purge.result    |  2 +-
 .../r/instant_alter_debug,dynamic.rdiff       |  6 ----
 .../suite/innodb/r/instant_alter_debug.result |  3 +-
 .../suite/innodb/r/instant_alter_purge.result |  3 +-
 .../innodb/t/cursor-restore-locking.test      | 35 ++++++++++---------
 mysql-test/suite/innodb/t/dml_purge.test      |  2 +-
 .../suite/innodb/t/instant_alter_debug.test   |  3 ++
 .../suite/innodb/t/instant_alter_purge.test   |  4 +--
 storage/innobase/include/trx0purge.h          | 12 +++++++
 storage/innobase/row/row0purge.cc             | 14 +++++++-
 storage/innobase/trx/trx0purge.cc             |  4 +--
 14 files changed, 68 insertions(+), 37 deletions(-)
 delete mode 100644 mysql-test/suite/innodb/r/instant_alter_debug,dynamic.rdiff

diff --git a/mysql-test/suite/gcol/r/innodb_virtual_debug_purge.result b/mysql-test/suite/gcol/r/innodb_virtual_debug_purge.result
index 4a204532630..2820aa7cdbb 100644
--- a/mysql-test/suite/gcol/r/innodb_virtual_debug_purge.result
+++ b/mysql-test/suite/gcol/r/innodb_virtual_debug_purge.result
@@ -76,7 +76,7 @@ SET DEBUG_SYNC= 'now WAIT_FOR uncommitted';
 # enable purge
 COMMIT;
 # wait for purge to process the deleted records.
-InnoDB		0 transactions not purged
+InnoDB		1 transactions not purged
 SET DEBUG_SYNC= 'now SIGNAL purged';
 connection default;
 /* connection default */ ALTER TABLE t1 ADD COLUMN c INT GENERATED ALWAYS AS(a+b), ADD INDEX idx (c), ALGORITHM=INPLACE, LOCK=SHARED;
diff --git a/mysql-test/suite/gcol/t/innodb_virtual_debug_purge.test b/mysql-test/suite/gcol/t/innodb_virtual_debug_purge.test
index ca60ed84a98..c8f0cc4c414 100644
--- a/mysql-test/suite/gcol/t/innodb_virtual_debug_purge.test
+++ b/mysql-test/suite/gcol/t/innodb_virtual_debug_purge.test
@@ -111,7 +111,9 @@ SET DEBUG_SYNC= 'now WAIT_FOR uncommitted';
 COMMIT;
 
 --echo # wait for purge to process the deleted records.
+let $wait_all_purged = 1;
 --source ../../innodb/include/wait_all_purged.inc
+let $wait_all_purged = 0;
 
 SET DEBUG_SYNC= 'now SIGNAL purged';
 
diff --git a/mysql-test/suite/innodb/r/cursor-restore-locking.result b/mysql-test/suite/innodb/r/cursor-restore-locking.result
index fc56f0935fa..a792babe5cd 100644
--- a/mysql-test/suite/innodb/r/cursor-restore-locking.result
+++ b/mysql-test/suite/innodb/r/cursor-restore-locking.result
@@ -1,31 +1,34 @@
 SET @save_freq=@@GLOBAL.innodb_purge_rseg_truncate_frequency;
 SET GLOBAL innodb_purge_rseg_truncate_frequency=1;
-CREATE TABLE t (a int PRIMARY KEY, b int NOT NULL UNIQUE) engine = InnoDB;
+CREATE TABLE t (a int PRIMARY KEY, b int NOT NULL UNIQUE) engine = InnoDB, STATS_PERSISTENT=0;
 InnoDB		0 transactions not purged
 connect prevent_purge,localhost,root,,;
 start transaction with consistent snapshot;
 connect con_del_1,localhost,root,,;
 INSERT INTO t VALUES (20,20);
 SET DEBUG_SYNC = 'innodb_row_search_for_mysql_exit SIGNAL first_del_row_search_mvcc_finished WAIT_FOR first_del_cont';
-DELETE FROM t WHERE b = 20;
+DELETE FROM t WHERE b = 20 # trx_1;
 connect con_ins_1,localhost,root,,;
 SET DEBUG_SYNC = 'now WAIT_FOR first_del_row_search_mvcc_finished';
 SET DEBUG_SYNC = 'lock_wait_suspend_thread_enter SIGNAL first_ins_locked';
 SET DEBUG_SYNC = 'ib_after_row_insert SIGNAL first_ins_row_inserted WAIT_FOR first_ins_cont';
-INSERT INTO t VALUES(10, 20);
+INSERT INTO t VALUES(10, 20) # trx_2;
 connect con_del_2,localhost,root,,;
 SET TRANSACTION ISOLATION LEVEL READ COMMITTED;
 SET DEBUG_SYNC = 'now WAIT_FOR first_ins_locked';
 SET DEBUG_SYNC = 'lock_wait_suspend_thread_enter SIGNAL second_del_locked';
-DELETE FROM t WHERE b = 20;
+DELETE FROM t WHERE b = 20 # trx_3;
 connection default;
 SET DEBUG_SYNC = 'now WAIT_FOR second_del_locked';
+SET @saved_dbug = @@GLOBAL.debug_dbug;
+SET @@GLOBAL.debug_dbug="d,enable_row_purge_del_mark_exit_sync_point";
 SET DEBUG_SYNC = 'now SIGNAL first_del_cont';
 SET DEBUG_SYNC = 'now WAIT_FOR first_ins_row_inserted';
 connection con_del_1;
 connection default;
 disconnect prevent_purge;
-InnoDB		0 transactions not purged
+SET DEBUG_SYNC = 'now WAIT_FOR row_purge_del_mark_finished';
+SET @@GLOBAL.debug_dbug = @saved_dbug;
 SET DEBUG_SYNC = 'now SIGNAL first_ins_cont';
 connection con_del_2;
 connection con_ins_1;
diff --git a/mysql-test/suite/innodb/r/dml_purge.result b/mysql-test/suite/innodb/r/dml_purge.result
index 38273d571c0..75a5f0fec6c 100644
--- a/mysql-test/suite/innodb/r/dml_purge.result
+++ b/mysql-test/suite/innodb/r/dml_purge.result
@@ -19,7 +19,7 @@ BEGIN;
 UPDATE t1 SET b=4 WHERE a=3;
 disconnect prevent_purge;
 connection default;
-InnoDB		0 transactions not purged
+SET GLOBAL innodb_max_purge_lag_wait=1;
 connection con1;
 ROLLBACK;
 disconnect con1;
diff --git a/mysql-test/suite/innodb/r/instant_alter_debug,dynamic.rdiff b/mysql-test/suite/innodb/r/instant_alter_debug,dynamic.rdiff
deleted file mode 100644
index 379514edad9..00000000000
--- a/mysql-test/suite/innodb/r/instant_alter_debug,dynamic.rdiff
+++ /dev/null
@@ -1,6 +0,0 @@
-@@ -470,4 +470,4 @@
- FROM information_schema.global_status
- WHERE variable_name = 'innodb_instant_alter_column';
- instants
--33
-+32
diff --git a/mysql-test/suite/innodb/r/instant_alter_debug.result b/mysql-test/suite/innodb/r/instant_alter_debug.result
index 82230573c44..39846a0e329 100644
--- a/mysql-test/suite/innodb/r/instant_alter_debug.result
+++ b/mysql-test/suite/innodb/r/instant_alter_debug.result
@@ -391,11 +391,12 @@ connection stop_purge;
 COMMIT;
 disconnect stop_purge;
 connection default;
-InnoDB		0 transactions not purged
+InnoDB		1 transactions not purged
 SET DEBUG_SYNC='now SIGNAL s2';
 connection dml;
 disconnect dml;
 connection default;
+InnoDB		0 transactions not purged
 SET DEBUG_SYNC=RESET;
 DROP TABLE t1;
 # End of 10.3 tests
diff --git a/mysql-test/suite/innodb/r/instant_alter_purge.result b/mysql-test/suite/innodb/r/instant_alter_purge.result
index 1179ff62ecc..261356bad12 100644
--- a/mysql-test/suite/innodb/r/instant_alter_purge.result
+++ b/mysql-test/suite/innodb/r/instant_alter_purge.result
@@ -1,5 +1,6 @@
 SET @saved_frequency = @@GLOBAL.innodb_purge_rseg_truncate_frequency;
 SET GLOBAL innodb_purge_rseg_truncate_frequency=1;
+InnoDB		0 transactions not purged
 #
 # MDEV-17793 Crash in purge after instant DROP and emptying the table
 #
@@ -16,7 +17,7 @@ COMMIT;
 START TRANSACTION WITH CONSISTENT SNAPSHOT;
 connection default;
 ALTER TABLE t1 ADD COLUMN extra TINYINT UNSIGNED NOT NULL DEFAULT 42;
-InnoDB		1 transactions not purged
+SET GLOBAL innodb_max_purge_lag_wait=1;
 ALTER TABLE t1 DROP extra;
 disconnect prevent_purge;
 InnoDB		0 transactions not purged
diff --git a/mysql-test/suite/innodb/t/cursor-restore-locking.test b/mysql-test/suite/innodb/t/cursor-restore-locking.test
index 0f083f9295b..815542c32db 100644
--- a/mysql-test/suite/innodb/t/cursor-restore-locking.test
+++ b/mysql-test/suite/innodb/t/cursor-restore-locking.test
@@ -5,7 +5,7 @@ source include/have_debug_sync.inc;
 
 SET @save_freq=@@GLOBAL.innodb_purge_rseg_truncate_frequency;
 SET GLOBAL innodb_purge_rseg_truncate_frequency=1;
-CREATE TABLE t (a int PRIMARY KEY, b int NOT NULL UNIQUE) engine = InnoDB;
+CREATE TABLE t (a int PRIMARY KEY, b int NOT NULL UNIQUE) engine = InnoDB, STATS_PERSISTENT=0;
 
 --source include/wait_all_purged.inc
 --connect(prevent_purge,localhost,root,,)
@@ -14,20 +14,20 @@ start transaction with consistent snapshot;
 --connect(con_del_1,localhost,root,,)
 INSERT INTO t VALUES (20,20);
 SET DEBUG_SYNC = 'innodb_row_search_for_mysql_exit SIGNAL first_del_row_search_mvcc_finished WAIT_FOR first_del_cont';
---send DELETE FROM t WHERE b = 20
+--send DELETE FROM t WHERE b = 20 # trx_1
 
 --connect(con_ins_1,localhost,root,,)
 SET DEBUG_SYNC = 'now WAIT_FOR first_del_row_search_mvcc_finished';
 # It's supposed the following INSERT will be suspended just after
 # lock_wait_suspend_thread_enter syncpoint, and will be awaken
-# after the previous DELETE commits. ib_after_row_insert will be executed
-# after the INSERT is woken up. The previous DELETE will wait for
+# after trx_1 DELETE commits. ib_after_row_insert will be executed
+# after the trx_2 INSERT is woken up. The trx_1 DELETE will wait for
 # first_del_cont signal before commit, and this signal will be sent later.
 # So it's safe to use two signals in a row here, it's guaranted the first
 # signal will be received before the second signal is sent.
 SET DEBUG_SYNC = 'lock_wait_suspend_thread_enter SIGNAL first_ins_locked';
 SET DEBUG_SYNC = 'ib_after_row_insert SIGNAL first_ins_row_inserted WAIT_FOR first_ins_cont';
---send INSERT INTO t VALUES(10, 20)
+--send INSERT INTO t VALUES(10, 20) # trx_2
 
 --connect(con_del_2,localhost,root,,)
 # After MDEV-30225 is fixed, the following DELETE creates next-key lock for
@@ -36,24 +36,26 @@ SET DEBUG_SYNC = 'ib_after_row_insert SIGNAL first_ins_row_inserted WAIT_FOR fir
 SET TRANSACTION ISOLATION LEVEL READ COMMITTED;
 SET DEBUG_SYNC = 'now WAIT_FOR first_ins_locked';
 SET DEBUG_SYNC = 'lock_wait_suspend_thread_enter SIGNAL second_del_locked';
-###############################################################################
-# This DELETE is locked by the previous DELETE, after that DELETE is
-# committed, it will still be locked by the next INSERT on delete-marked
-# heap_no 2 record. After that INSERT inserted the record with heap_no 3,
-# and after heap_no 2 record is purged, this DELETE will be unlocked and
+##############################################################################
+# trx_3 DELETE is locked by trx_1 DELETE, after trx_1 DELETE is
+# committed, it will still be locked by trx_2 INSERT on delete-marked
+# heap_no 2 record. After trx_2 INSERT inserted the record with heap_no 3,
+# and after heap_no 2 record is purged, trx_3 DELETE will be unlocked and
 # must restore persistent cursor position at heap_no 3 record, as it has the
 # same secondary key value as former heap_no 2 record. Then it must be blocked
-# by the previous INSERT, and after the INSERT is committed, it must
-# delete the record, inserted by the previous INSERT, and the last INSERT(see
+# by trx_2 INSERT, and after trx_2 INSERT is committed, it must
+# delete the record, inserted by trx_2 INSERT, and trx_4 INSERT(see
 # below) must be finished without error. But instead this DELETE restores
 # persistent cursor position to supremum, as a result, it does not delete the
-# record, inserted by the previous INSERT, and the last INSERT is finished with
+# record, inserted by trx_2 INSERT, and trx_4 INSERT is finished with
 # duplicate key check error.
 ###############################################################################
---send DELETE FROM t WHERE b = 20
+--send DELETE FROM t WHERE b = 20 # trx_3
 
 --connection default
 SET DEBUG_SYNC = 'now WAIT_FOR second_del_locked';
+SET @saved_dbug = @@GLOBAL.debug_dbug;
+SET @@GLOBAL.debug_dbug="d,enable_row_purge_del_mark_exit_sync_point";
 SET DEBUG_SYNC = 'now SIGNAL first_del_cont';
 SET DEBUG_SYNC = 'now WAIT_FOR first_ins_row_inserted';
 --connection con_del_1
@@ -61,7 +63,8 @@ SET DEBUG_SYNC = 'now WAIT_FOR first_ins_row_inserted';
 
 --connection default
 --disconnect prevent_purge
---source include/wait_all_purged.inc
+SET DEBUG_SYNC = 'now WAIT_FOR row_purge_del_mark_finished';
+SET @@GLOBAL.debug_dbug = @saved_dbug;
 SET DEBUG_SYNC = 'now SIGNAL first_ins_cont';
 
 --connection con_del_2
@@ -74,7 +77,7 @@ SET DEBUG_SYNC = 'now SIGNAL first_ins_cont';
 ###############################################################################
 # Duplicate key error is expected if the bug is not fixed.
 ###############################################################################
-INSERT INTO t VALUES(30, 20);
+INSERT INTO t VALUES(30, 20); # trx_4
 
 --disconnect con_ins_1
 --disconnect con_del_1
diff --git a/mysql-test/suite/innodb/t/dml_purge.test b/mysql-test/suite/innodb/t/dml_purge.test
index 7034939aa4e..c13ff22572b 100644
--- a/mysql-test/suite/innodb/t/dml_purge.test
+++ b/mysql-test/suite/innodb/t/dml_purge.test
@@ -32,7 +32,7 @@ UPDATE t1 SET b=4 WHERE a=3;
 
 --connection default
 # Initiate a full purge, which should reset the DB_TRX_ID except for a=3.
---source include/wait_all_purged.inc
+SET GLOBAL innodb_max_purge_lag_wait=1;
 # Initiate a ROLLBACK of the update, which should reset the DB_TRX_ID for a=3.
 --connection con1
 ROLLBACK;
diff --git a/mysql-test/suite/innodb/t/instant_alter_debug.test b/mysql-test/suite/innodb/t/instant_alter_debug.test
index f102185c27f..c49ab758f24 100644
--- a/mysql-test/suite/innodb/t/instant_alter_debug.test
+++ b/mysql-test/suite/innodb/t/instant_alter_debug.test
@@ -451,7 +451,9 @@ COMMIT;
 disconnect stop_purge;
 
 connection default;
+let $wait_all_purged = 1;
 --source include/wait_all_purged.inc
+let $wait_all_purged = 0;
 SET DEBUG_SYNC='now SIGNAL s2';
 
 connection dml;
@@ -459,6 +461,7 @@ reap;
 disconnect dml;
 
 connection default;
+--source include/wait_all_purged.inc
 SET DEBUG_SYNC=RESET;
 DROP TABLE t1;
 
diff --git a/mysql-test/suite/innodb/t/instant_alter_purge.test b/mysql-test/suite/innodb/t/instant_alter_purge.test
index 9ccf3347d7b..88a56141a1f 100644
--- a/mysql-test/suite/innodb/t/instant_alter_purge.test
+++ b/mysql-test/suite/innodb/t/instant_alter_purge.test
@@ -6,6 +6,7 @@ if ($have_debug) {
 
 SET @saved_frequency = @@GLOBAL.innodb_purge_rseg_truncate_frequency;
 SET GLOBAL innodb_purge_rseg_truncate_frequency=1;
+--source include/wait_all_purged.inc
 
 --echo #
 --echo # MDEV-17793 Crash in purge after instant DROP and emptying the table
@@ -27,8 +28,7 @@ START TRANSACTION WITH CONSISTENT SNAPSHOT;
 connection default;
 
 ALTER TABLE t1 ADD COLUMN extra TINYINT UNSIGNED NOT NULL DEFAULT 42;
-let $wait_all_purged= 1;
---source include/wait_all_purged.inc
+SET GLOBAL innodb_max_purge_lag_wait=1;
 ALTER TABLE t1 DROP extra;
 disconnect prevent_purge;
 let $wait_all_purged= 0;
diff --git a/storage/innobase/include/trx0purge.h b/storage/innobase/include/trx0purge.h
index 14cf6a2958b..b8d349af2b0 100644
--- a/storage/innobase/include/trx0purge.h
+++ b/storage/innobase/include/trx0purge.h
@@ -253,6 +253,18 @@ public:
 #endif
     return view.low_limit_no();
   }
+  /** A wrapper around ReadView::sees(). */
+  trx_id_t sees(trx_id_t id) const
+  {
+    /* This function may only be called by purge_coordinator_callback().
+
+    The purge coordinator task may call this without holding any latch,
+    because it is the only thread that may modify purge_sys.view.
+
+    Any other threads that access purge_sys.view must hold purge_sys.latch,
+    typically via purge_sys_t::view_guard. */
+    return view.sees(id);
+  }
   /** A wrapper around trx_sys_t::clone_oldest_view(). */
   void clone_oldest_view()
   {
diff --git a/storage/innobase/row/row0purge.cc b/storage/innobase/row/row0purge.cc
index 74bbc61df52..6942f5b7af8 100644
--- a/storage/innobase/row/row0purge.cc
+++ b/storage/innobase/row/row0purge.cc
@@ -46,6 +46,7 @@ Created 3/14/1997 Heikki Tuuri
 #include "handler.h"
 #include "ha_innodb.h"
 #include "fil0fil.h"
+#include "debug_sync.h"
 
 /*************************************************************************
 IMPORTANT NOTE: Any operation that generates redo MUST check that there
@@ -646,7 +647,18 @@ row_purge_del_mark(
 
 	mem_heap_free(heap);
 
-	return(row_purge_remove_clust_if_poss(node));
+	bool result = row_purge_remove_clust_if_poss(node);
+
+#ifdef ENABLED_DEBUG_SYNC
+	DBUG_EXECUTE_IF(
+		"enable_row_purge_del_mark_exit_sync_point",
+		debug_sync_set_action(
+			current_thd,
+			STRING_WITH_LEN(
+				"now SIGNAL row_purge_del_mark_finished")););
+#endif
+
+	return result;
 }
 
 /** Reset DB_TRX_ID, DB_ROLL_PTR of a clustered index record
diff --git a/storage/innobase/trx/trx0purge.cc b/storage/innobase/trx/trx0purge.cc
index 2f21a4de1e6..7140b2c8816 100644
--- a/storage/innobase/trx/trx0purge.cc
+++ b/storage/innobase/trx/trx0purge.cc
@@ -547,7 +547,7 @@ static void trx_purge_truncate_history()
       mutex_enter(&rseg->mutex);
       trx_purge_truncate_rseg_history(*rseg, head,
                                       !rseg->trx_ref_count &&
-                                      rseg->needs_purge <= head.trx_no);
+                                      purge_sys.sees(rseg->needs_purge));
       mutex_exit(&rseg->mutex);
     }
   }
@@ -604,7 +604,7 @@ static void trx_purge_truncate_history()
       transactions to finish and to be purged. */
       rseg->skip_allocation = true;
 
-      if (rseg->trx_ref_count || rseg->needs_purge > head.trx_no)
+      if (rseg->trx_ref_count || !purge_sys.sees(rseg->needs_purge))
       {
       not_free:
         mutex_exit(&rseg->mutex);