Merge 10.5 into 10.6

mysql-test/suite/encryption/r/innodb-redo-nokeys.result

-call mtr.add_suppression("mariadbd.*: File .*");
-call mtr.add_suppression("Plugin 'file_key_management' .*");
-call mtr.add_suppression("InnoDB: We do not continue the crash recovery");
+call mtr.add_suppression("mariadbd.*: File ");
+call mtr.add_suppression("Plugin 'file_key_management' ");
+call mtr.add_suppression("InnoDB: Recovery cannot access file");
 call mtr.add_suppression("InnoDB: Plugin initialization aborted");
 call mtr.add_suppression("Plugin 'InnoDB' init function returned error\\.");
 call mtr.add_suppression("Plugin 'InnoDB' registration as a STORAGE ENGINE failed.");

mysql-test/suite/encryption/t/innodb-redo-nokeys.test

@@ -3,9 +3,9 @@
 # embedded does not support restart
 -- source include/not_embedded.inc
 
-call mtr.add_suppression("mariadbd.*: File .*");
-call mtr.add_suppression("Plugin 'file_key_management' .*");
-call mtr.add_suppression("InnoDB: We do not continue the crash recovery");
+call mtr.add_suppression("mariadbd.*: File ");
+call mtr.add_suppression("Plugin 'file_key_management' ");
+call mtr.add_suppression("InnoDB: Recovery cannot access file");
 call mtr.add_suppression("InnoDB: Plugin initialization aborted");
 call mtr.add_suppression("Plugin 'InnoDB' init function returned error\\.");
 call mtr.add_suppression("Plugin 'InnoDB' registration as a STORAGE ENGINE failed.");

mysql-test/suite/innodb/r/ibuf_not_empty.result

@@ -12,7 +12,7 @@ check table t1;
 Table	Op	Msg_type	Msg_text
 test.t1	check	Warning	InnoDB: Index 'b' contains 990 entries, should be 1024.
 test.t1	check	error	Corrupt
-# restart
+# restart: --innodb-force_recovery=0
 SET GLOBAL innodb_fast_shutdown=0;
-# restart
+# restart: --innodb-force_recovery=0
 DROP TABLE t1;

mysql-test/suite/innodb/t/ibuf_not_empty.test

@@ -11,6 +11,8 @@ call mtr.add_suppression("InnoDB: Failed to find tablespace for table `test`\\.`
 call mtr.add_suppression("InnoDB: Allocated tablespace ID \\d+ for test.t1, old maximum was");
 call mtr.add_suppression("InnoDB: Failed to find tablespace for table `mysql`\\.`transaction_registry` in the cache\\. Attempting to load the tablespace with space id");
 call mtr.add_suppression("InnoDB: Allocated tablespace ID \\d+ for mysql.transaction_registry, old maximum was");
+call mtr.add_suppression("InnoDB: Trying to read 4096 bytes");
+call mtr.add_suppression("InnoDB: File './test/t1.ibd' is corrupted");
 --enable_query_log
 
 CREATE TABLE t1(
@@ -106,7 +108,7 @@ truncate(FILE, $ps * $pages);
 close(FILE) || die "Unable to close $file";
 EOF
 
---let $restart_parameters=
+--let $restart_parameters=--innodb-force_recovery=0
 --source include/start_mysqld.inc
 SET GLOBAL innodb_fast_shutdown=0;
 --source include/restart_mysqld.inc

storage/innobase/buf/buf0buf.cc

@@ -2698,7 +2698,7 @@ buf_page_get_low(
 re_evict:
 	if (mode != BUF_GET_IF_IN_POOL
 	    && mode != BUF_GET_IF_IN_POOL_OR_WATCH) {
-	} else if (!ibuf_debug) {
+	} else if (!ibuf_debug || recv_recovery_is_on()) {
 	} else if (fil_space_t* space = fil_space_t::get(page_id.space())) {
 		/* Try to evict the block from the buffer pool, to use the
 		insert buffer (change buffer) as much as possible. */

storage/innobase/ibuf/ibuf0ibuf.cc

@@ -2261,16 +2261,74 @@ static MY_ATTRIBUTE((warn_unused_result, nonnull))
 bool ibuf_delete_rec(const page_id_t page_id, btr_pcur_t* pcur,
 		     const dtuple_t* search_tuple, mtr_t* mtr);
 
+/** Delete the change buffer records for the given page id
+@param page_id page identifier */
+static void ibuf_delete_recs(const page_id_t page_id)
+{
+  if (!ibuf.index || srv_read_only_mode)
+    return;
+  dfield_t dfield[IBUF_REC_FIELD_METADATA];
+  dtuple_t tuple {0,IBUF_REC_FIELD_METADATA,IBUF_REC_FIELD_METADATA,
+                  dfield,0,nullptr
+#ifdef UNIV_DEBUG
+                  ,DATA_TUPLE_MAGIC_N
+#endif
+  };
+  byte space_id[4], page_no[4];
+
+  mach_write_to_4(space_id, page_id.space());
+  mach_write_to_4(page_no, page_id.page_no());
+
+  dfield_set_data(&dfield[0], space_id, 4);
+  dfield_set_data(&dfield[1], field_ref_zero, 1);
+  dfield_set_data(&dfield[2], page_no, 4);
+  dtuple_set_types_binary(&tuple, IBUF_REC_FIELD_METADATA);
+
+  mtr_t mtr;
+loop:
+  btr_pcur_t pcur;
+  pcur.btr_cur.page_cur.index= ibuf.index;
+  ibuf_mtr_start(&mtr);
+  if (btr_pcur_open(&tuple, PAGE_CUR_GE, BTR_MODIFY_LEAF, &pcur, 0, &mtr))
+    goto func_exit;
+  if (!btr_pcur_is_on_user_rec(&pcur))
+  {
+    ut_ad(btr_pcur_is_after_last_on_page(&pcur));
+    goto func_exit;
+  }
+
+  for (;;)
+  {
+    ut_ad(btr_pcur_is_on_user_rec(&pcur));
+    const rec_t* ibuf_rec = btr_pcur_get_rec(&pcur);
+    if (ibuf_rec_get_space(&mtr, ibuf_rec) != page_id.space()
+        || ibuf_rec_get_page_no(&mtr, ibuf_rec) != page_id.page_no())
+      break;
+    /* Delete the record from ibuf */
+    if (ibuf_delete_rec(page_id, &pcur, &tuple, &mtr))
+    {
+      /* Deletion was pessimistic and mtr was committed:
+      we start from the beginning again */
+      ut_ad(mtr.has_committed());
+      goto loop;
+    }
+
+    if (btr_pcur_is_after_last_on_page(&pcur))
+    {
+      ibuf_mtr_commit(&mtr);
+      btr_pcur_close(&pcur);
+      goto loop;
+    }
+  }
+func_exit:
+  ibuf_mtr_commit(&mtr);
+  btr_pcur_close(&pcur);
+}
+
 /** Merge the change buffer to some pages. */
 static void ibuf_read_merge_pages(const uint32_t* space_ids,
 				  const uint32_t* page_nos, ulint n_stored)
 {
-#ifndef DBUG_OFF
-	mem_heap_t* heap = mem_heap_create(512);
-	ulint dops[IBUF_OP_COUNT];
-	memset(dops, 0, sizeof(dops));
-#endif
-
 	for (ulint i = 0; i < n_stored; i++) {
 		const ulint space_id = space_ids[i];
 		fil_space_t* s = fil_space_t::get(space_id);
@@ -2293,28 +2351,36 @@ static void ibuf_read_merge_pages(const uint32_t* space_ids,
 		if (UNIV_LIKELY(page_nos[i] < size)) {
 			mtr.start();
 			dberr_t err;
-			buf_block_t *b =
+			buf_block_t *block =
 			buf_page_get_gen(page_id_t(space_id, page_nos[i]),
 					 zip_size, RW_X_LATCH, nullptr,
 					 BUF_GET_POSSIBLY_FREED,
 					 &mtr, &err, true);
+			bool remove = !block
+				|| fil_page_get_type(block->page.frame)
+				!= FIL_PAGE_INDEX
+				|| !page_is_leaf(block->page.frame);
 			mtr.commit();
-			if (b) {
-			} else if (err == DB_TABLESPACE_DELETED) {
+			if (err == DB_TABLESPACE_DELETED) {
 				goto tablespace_deleted;
-			} else {
+			}
+			if (!remove) {
 				continue;
 			}
 		}
-#ifndef DBUG_OFF
-		DBUG_EXECUTE_IF("ibuf_merge_corruption", goto work_around;);
-		continue;
+
+		if (srv_shutdown_state == SRV_SHUTDOWN_NONE
+		    || srv_fast_shutdown) {
+			continue;
+		}
 
 		/* The following code works around a hang when the
-		change buffer is corrupted, likely due to the race
-		condition in crash recovery that was fixed in
-		MDEV-24449. But, it also introduces corruption by
-		itself in the following scenario:
+		change buffer is corrupted, likely due to the
+		failure of ibuf_merge_or_delete_for_page() to
+		invoke ibuf_delete_recs() if (!bitmap_bits).
+
+		It also introduced corruption by itself in the
+		following scenario:
 
 		(1) We merged buffered changes in buf_page_get_gen()
 		(2) We committed the mini-transaction
@@ -2323,67 +2389,16 @@ static void ibuf_read_merge_pages(const uint32_t* space_ids,
 		(5) Other threads buffer changes for that page.
 		(6) We will wrongly discard those newly buffered changes below.
 
-		This code will be available in debug builds, so that
-		users may try to fix a shutdown hang that occurs due
-		to a corrupted change buffer. */
+		To prevent this scenario, we will only invoke this code
+		on shutdown. A call to ibuf_max_size_update(0) will cause
+		ibuf_insert_low() to refuse to insert anything into the
+		change buffer. */
 
-work_around:
 		/* Prevent an infinite loop, by removing entries from
-		the change buffer also in the case the bitmap bits were
+		the change buffer in the case the bitmap bits were
 		wrongly clear even though buffered changes exist. */
-		const dtuple_t* tuple = ibuf_search_tuple_build(
-			space_id, page_nos[i], heap);
-loop:
-		btr_pcur_t pcur;
-		pcur.btr_cur.page_cur.index = ibuf.index;
-
-		ibuf_mtr_start(&mtr);
-		if (btr_pcur_open(tuple, PAGE_CUR_GE,
-				  BTR_MODIFY_LEAF, &pcur, 0, &mtr)
-		    != DB_SUCCESS) {
-			goto done;
-		}
-		if (!btr_pcur_is_on_user_rec(&pcur)) {
-			ut_ad(btr_pcur_is_after_last_on_page(&pcur));
-			goto done;
-		}
-
-		for (;;) {
-			ut_ad(btr_pcur_is_on_user_rec(&pcur));
-
-			const rec_t* ibuf_rec = btr_pcur_get_rec(&pcur);
-			if (ibuf_rec_get_space(&mtr, ibuf_rec) != space_id
-			    || ibuf_rec_get_page_no(&mtr, ibuf_rec)
-			    != page_nos[i]) {
-				break;
-			}
-
-			dops[ibuf_rec_get_op_type(&mtr, ibuf_rec)]++;
-			/* Delete the record from ibuf */
-			if (ibuf_delete_rec(page_id_t(space_id, page_nos[i]),
-					    &pcur, tuple, &mtr)) {
-				/* Deletion was pessimistic and mtr
-				was committed: we start from the
-				beginning again */
-				ut_ad(mtr.has_committed());
-				goto loop;
-			}
-
-			if (btr_pcur_is_after_last_on_page(&pcur)) {
-				ibuf_mtr_commit(&mtr);
-				goto loop;
-			}
-		}
-done:
-		ibuf_mtr_commit(&mtr);
-		mem_heap_empty(heap);
-#endif
+		ibuf_delete_recs(page_id_t(space_ids[i], page_nos[i]));
 	}
-
-#ifndef DBUG_OFF
-	ibuf_add_ops(ibuf.n_discarded_ops, dops);
-	mem_heap_free(heap);
-#endif
 }
 
 /** Contract the change buffer by reading pages to the buffer pool.
@@ -2453,8 +2468,23 @@ ibuf_merge_space(
 {
 	mtr_t		mtr;
 	btr_pcur_t	pcur;
-	mem_heap_t*	heap = mem_heap_create(512);
-	dtuple_t*	tuple = ibuf_search_tuple_build(space, 0, heap);
+
+	dfield_t dfield[IBUF_REC_FIELD_METADATA];
+	dtuple_t tuple {0, IBUF_REC_FIELD_METADATA,
+			IBUF_REC_FIELD_METADATA,dfield,0,nullptr
+#ifdef UNIV_DEBUG
+			, DATA_TUPLE_MAGIC_N
+#endif
+	};
+	byte space_id[4];
+
+	mach_write_to_4(space_id, space);
+
+	dfield_set_data(&dfield[0], space_id, 4);
+	dfield_set_data(&dfield[1], field_ref_zero, 1);
+	dfield_set_data(&dfield[2], field_ref_zero, 4);
+
+	dtuple_set_types_binary(&tuple, IBUF_REC_FIELD_METADATA);
 	ulint		n_pages = 0;
 
 	ut_ad(space < SRV_SPACE_ID_UPPER_BOUND);
@@ -2465,13 +2495,11 @@ ibuf_merge_space(
 	/* Position the cursor on the first matching record. */
 
 	pcur.btr_cur.page_cur.index = ibuf.index;
-	dberr_t err = btr_pcur_open(tuple, PAGE_CUR_GE,
+	dberr_t err = btr_pcur_open(&tuple, PAGE_CUR_GE,
 				    BTR_SEARCH_LEAF, &pcur, 0, &mtr);
 	ut_ad(err != DB_SUCCESS || page_validate(btr_pcur_get_page(&pcur),
 						 ibuf.index));
 
-	mem_heap_free(heap);
-
 	ulint		sum_sizes = 0;
 	uint32_t	pages[IBUF_MAX_N_PAGES_MERGED];
 	uint32_t	spaces[IBUF_MAX_N_PAGES_MERGED];
@@ -4164,6 +4192,11 @@ dberr_t ibuf_merge_or_delete_for_page(buf_block_t *block,
 			ibuf_reset_bitmap(block, page_id, zip_size, &mtr);
 			ibuf_mtr_commit(&mtr);
 			bitmap_bits = 0;
+			if (!block
+			    || btr_page_get_index_id(block->page.frame)
+			    != DICT_IBUF_ID_MIN + IBUF_SPACE_ID) {
+				ibuf_delete_recs(page_id);
+			}
 		}
 
 		if (!bitmap_bits) {
@@ -4389,22 +4422,31 @@ in DISCARD TABLESPACE, IMPORT TABLESPACE, or read-ahead.
 @param[in]	space		missing or to-be-discarded tablespace */
 void ibuf_delete_for_discarded_space(ulint space)
 {
-	mem_heap_t*	heap;
 	btr_pcur_t	pcur;
-	dtuple_t*	search_tuple;
 	const rec_t*	ibuf_rec;
 	mtr_t		mtr;
 
 	/* Counts for discarded operations. */
 	ulint		dops[IBUF_OP_COUNT];
 
-	heap = mem_heap_create(512);
+	dfield_t	dfield[IBUF_REC_FIELD_METADATA];
+	dtuple_t	search_tuple {0,IBUF_REC_FIELD_METADATA,
+				      IBUF_REC_FIELD_METADATA,dfield,0
+				      ,nullptr
+#ifdef UNIV_DEBUG
+				      ,DATA_TUPLE_MAGIC_N
+#endif /* UNIV_DEBUG */
+	};
+	byte space_id[4];
+	mach_write_to_4(space_id, space);
 
+	dfield_set_data(&dfield[0], space_id, 4);
+	dfield_set_data(&dfield[1], field_ref_zero, 1);
+	dfield_set_data(&dfield[2], field_ref_zero, 4);
+	dtuple_set_types_binary(&search_tuple, IBUF_REC_FIELD_METADATA);
 	/* Use page number 0 to build the search tuple so that we get the
 	cursor positioned at the first entry for this space id */
 
-	search_tuple = ibuf_search_tuple_build(space, 0, heap);
-
 	memset(dops, 0, sizeof(dops));
 	pcur.btr_cur.page_cur.index = ibuf.index;
 
@@ -4414,7 +4456,7 @@ void ibuf_delete_for_discarded_space(ulint space)
 
 	/* Position pcur in the insert buffer at the first entry for the
 	space */
-	if (btr_pcur_open_on_user_rec(search_tuple, PAGE_CUR_GE,
+	if (btr_pcur_open_on_user_rec(&search_tuple, PAGE_CUR_GE,
 				      BTR_MODIFY_LEAF, &pcur, &mtr)
 	    != DB_SUCCESS) {
 		goto leave_loop;
@@ -4442,7 +4484,7 @@ void ibuf_delete_for_discarded_space(ulint space)
 
 		/* Delete the record from ibuf */
 		if (ibuf_delete_rec(page_id_t(space, page_no),
-				    &pcur, search_tuple, &mtr)) {
+				    &pcur, &search_tuple, &mtr)) {
 			/* Deletion was pessimistic and mtr was committed:
 			we start from the beginning again */
 
@@ -4463,8 +4505,6 @@ void ibuf_delete_for_discarded_space(ulint space)
 	ut_free(pcur.old_rec_buf);
 
 	ibuf_add_ops(ibuf.n_discarded_ops, dops);
-
-	mem_heap_free(heap);
 }
 
 /******************************************************************//**
@@ -4639,15 +4679,20 @@ dberr_t ibuf_check_bitmap_on_import(const trx_t* trx, fil_space_t* space)
 void ibuf_set_bitmap_for_bulk_load(buf_block_t *block, mtr_t *mtr, bool reset)
 {
   ut_a(page_is_leaf(block->page.frame));
+  const page_id_t id{block->page.id()};
+  const auto zip_size= block->zip_size();
 
-  if (buf_block_t *bitmap_page=
-      ibuf_bitmap_get_map_page(block->page.id(), block->zip_size(), mtr))
+  if (buf_block_t *bitmap_page= ibuf_bitmap_get_map_page(id, zip_size, mtr))
   {
+    if (ibuf_bitmap_page_get_bits(bitmap_page->page.frame, id, zip_size,
+                                  IBUF_BITMAP_BUFFERED, mtr))
+      ibuf_delete_recs(id);
+
     ulint free_val= reset ? 0 : ibuf_index_page_calc_free(block);
     /* FIXME: update the bitmap byte only once! */
     ibuf_bitmap_page_set_bits<IBUF_BITMAP_FREE>
-      (bitmap_page, block->page.id(), block->physical_size(), free_val, mtr);
+      (bitmap_page, id, block->physical_size(), free_val, mtr);
     ibuf_bitmap_page_set_bits<IBUF_BITMAP_BUFFERED>
-      (bitmap_page, block->page.id(), block->physical_size(), false, mtr);
+      (bitmap_page, id, block->physical_size(), false, mtr);
   }
 }

storage/innobase/log/log0log.cc

@@ -1204,8 +1204,9 @@ ATTRIBUTE_COLD void logs_empty_and_mark_files_at_shutdown()
 
 	if (srv_fast_shutdown == 2 || !srv_was_started) {
 		if (!srv_read_only_mode && srv_was_started) {
-			ib::info() << "Executing innodb_fast_shutdown=2."
-				" Next startup will execute crash recovery!";
+			sql_print_information(
+				"InnoDB: Executing innodb_fast_shutdown=2."
+				" Next startup will execute crash recovery!");
 
 			/* In this fastest shutdown we do not flush the
 			buffer pool:

storage/innobase/log/log0recv.cc

@@ -1343,35 +1343,22 @@ static void fil_name_process(const char *name, ulint len, uint32_t space_id,
 		case FIL_LOAD_INVALID:
 			ut_ad(space == NULL);
 			if (srv_force_recovery == 0) {
-				ib::warn() << "We do not continue the crash"
-					" recovery, because the table may"
-					" become corrupt if we cannot apply"
-					" the log records in the InnoDB log to"
-					" it. To fix the problem and start"
-					" mysqld:";
-				ib::info() << "1) If there is a permission"
-					" problem in the file and mysqld"
-					" cannot open the file, you should"
-					" modify the permissions.";
-				ib::info() << "2) If the tablespace is not"
-					" needed, or you can restore an older"
-					" version from a backup, then you can"
-					" remove the .ibd file, and use"
-					" --innodb_force_recovery=1 to force"
-					" startup without this file.";
-				ib::info() << "3) If the file system or the"
-					" disk is broken, and you cannot"
-					" remove the .ibd file, you can set"
-					" --innodb_force_recovery.";
+				sql_print_error("InnoDB: Recovery cannot access"
+						" file %s (tablespace "
+						UINT32PF ")", name, space_id);
+				sql_print_information("InnoDB: You may set "
+						      "innodb_force_recovery=1"
+						      " to ignore this and"
+						      " possibly get a"
+						      " corrupted database.");
 				recv_sys.set_corrupt_fs();
 				break;
 			}
 
-			ib::info() << "innodb_force_recovery was set to "
-				<< srv_force_recovery << ". Continuing crash"
-				" recovery even though we cannot access the"
-				" files for tablespace " << space_id << ".";
-			break;
+			sql_print_warning("InnoDB: Ignoring changes to"
+					  " file %s (tablespace " UINT32PF ")"
+					  " due to innodb_force_recovery",
+					  name, space_id);
 		}
 	}
 }

storage/innobase/row/row0mysql.cc

@@ -704,7 +704,7 @@ row_mysql_handle_errors(
 			" table. You have to dump + drop + reimport the"
 			" table or, in a case of widespread corruption,"
 			" dump all InnoDB tables and recreate the whole"
-			" tablespace. If the mysqld server crashes after"
+			" tablespace. If the mariadbd server crashes after"
 			" the startup or when you dump the tables. "
 			<< FORCE_RECOVERY_MSG;
 		goto rollback_to_savept;

storage/innobase/srv/srv0srv.cc

@@ -1556,6 +1556,10 @@ void srv_shutdown(bool ibuf_merge)
 
 		if (ibuf_merge) {
 			srv_main_thread_op_info = "doing insert buffer merge";
+			/* Disallow the use of change buffer to
+			avoid a race condition with
+			ibuf_read_merge_pages() */
+			ibuf_max_size_update(0);
 			log_free_check();
 			n_read = ibuf_contract();
 			srv_shutdown_print(now, n_read);

storage/innobase/ut/ut0dbg.cc

 /*****************************************************************************
 
 Copyright (c) 1994, 2016, Oracle and/or its affiliates. All Rights Reserved.
-Copyright (c) 2017, 2018, MariaDB Corporation.
+Copyright (c) 2017, 2022, MariaDB Corporation.
 
 This program is free software; you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free Software