Bug#12612184 Race condition after btr_cur_pessimistic_update()

btr_cur_compress_if_useful(), btr_compress(): Add the parameter ibool
adjust. If adjust=TRUE, adjust the cursor position after compressing
the page.

btr_lift_page_up(): Return a pointer to the father page.

BTR_KEEP_POS_FLAG: A new flag for btr_cur_pessimistic_update().

btr_cur_pessimistic_update(): If *big_rec != NULL and flags &
BTR_KEEP_POS_FLAG, keep the cursor positioned on the updated record.
Also, do not release the index tree x-lock if *big_rec != NULL.

btr_cur_mtr_commit_and_start(): Commits and restarts a
mini-transaction so that it will retain an x-lock on index->lock and
the page of the cursor. This is invoked when
btr_cur_pessimistic_update() returns *big_rec != NULL.

In all callers of btr_cur_pessimistic_update() that do not pass
BTR_KEEP_POS_FLAG, assert that *big_rec == NULL.

btr_cur_compress(): Unused function [in the built-in MySQL 5.1], remove.

page_rec_get_nth(): Return the nth record on the page (an inverse
function of page_rec_get_n_recs_before()). Refactored from
page_get_middle_rec().

page_get_middle_rec(): Invoke page_rec_get_nth().

page_cur_insert_rec_zip_reorg(): Make use of the page directory
shortcuts in page_rec_get_nth() instead of scanning the whole list of
records.

row_ins_clust_index_entry_by_modify(): Pass BTR_KEEP_POS_FLAG to
btr_cur_pessimistic_update().

row_ins_index_entry_low(): If row_ins_clust_index_entry_by_modify()
returns a big_rec, invoke btr_cur_mtr_commit_and_start() in order to
commit and start the mini-transaction without releasing the x-locks on
index->lock and the cursor page, and write the big_rec. Releasing the
page latch in mtr_commit() caused a race condition.

row_upd_clust_rec(): Pass BTR_KEEP_POS_FLAG to
btr_cur_pessimistic_update(). If it returns a big_rec, invoke
btr_cur_mtr_commit_and_start() in order to commit and start the
mini-transaction without releasing the x-locks on index->lock and the
cursor page, and write the big_rec. Releasing the page latch in
mtr_commit() caused a race condition.

sync_thread_add_level(): Add the parameter ibool relock. When TRUE,
bypass the latching order rules.

rw_lock_add_debug_info(): For nested X-lock requests, pass relock=TRUE
to sync_thread_add_level().

rb:678 approved by Jimmy Yang

1 files changed, 25 insertions, 9 deletions

diff --git a/storage/innobase/row/row0upd.c b/storage/innobase/row/row0upd.c
index 0790cfe02e2..694b00ea265 100644
--- a/storage/innobase/row/row0upd.c
+++ b/storage/innobase/row/row0upd.c
@@ -1580,32 +1580,48 @@ row_upd_clust_rec(
 	ut_ad(!rec_get_deleted_flag(btr_pcur_get_rec(pcur),
 				    dict_table_is_comp(index->table)));
 
-	err = btr_cur_pessimistic_update(BTR_NO_LOCKING_FLAG, btr_cur,
-					 &big_rec, node->update,
-					 node->cmpl_info, thr, mtr);
-	mtr_commit(mtr);
+	err = btr_cur_pessimistic_update(
+		BTR_NO_LOCKING_FLAG | BTR_KEEP_POS_FLAG, btr_cur,
+		&big_rec, node->update, node->cmpl_info, thr, mtr);
 
-	if (err == DB_SUCCESS && big_rec) {
+	if (big_rec) {
 		mem_heap_t*	heap		= NULL;
 		ulint		offsets_[REC_OFFS_NORMAL_SIZE];
 		rec_t*		rec;
 		*offsets_ = (sizeof offsets_) / sizeof *offsets_;
 
-		mtr_start(mtr);
+		ut_a(err == DB_SUCCESS);
+		/* Write out the externally stored columns while still
+		x-latching index->lock and block->lock. We have to
+		mtr_commit(mtr) first, so that the redo log will be
+		written in the correct order. Otherwise, we would run
+		into trouble on crash recovery if mtr freed B-tree
+		pages on which some of the big_rec fields will be
+		written. */
+		btr_cur_mtr_commit_and_start(btr_cur, mtr);
 
-		ut_a(btr_pcur_restore_position(BTR_MODIFY_TREE, pcur, mtr));
 		rec = btr_cur_get_rec(btr_cur);
 		err = btr_store_big_rec_extern_fields(
 			index, rec,
 			rec_get_offsets(rec, index, offsets_,
 					ULINT_UNDEFINED, &heap),
-			 big_rec, mtr);
+			big_rec, mtr);
 		if (UNIV_LIKELY_NULL(heap)) {
 			mem_heap_free(heap);
 		}
-		mtr_commit(mtr);
+		/* If writing big_rec fails (for example, because of
+		DB_OUT_OF_FILE_SPACE), the record will be corrupted.
+		Even if we did not update any externally stored
+		columns, our update could cause the record to grow so
+		that a non-updated column was selected for external
+		storage. This non-update would not have been written
+		to the undo log, and thus the record cannot be rolled
+		back. */
+		ut_a(err == DB_SUCCESS);
 	}
 
+	mtr_commit(mtr);
+
 	if (big_rec) {
 		dtuple_big_rec_free(big_rec);
 	}