Bug#43230: SELECT ... FOR UPDATE can hang with FLUSH TABLES WITH READ LOCK indefinitely

The problem is that a SELECT .. FOR UPDATE statement might open
a table and later wait for a impeding global read lock without
noticing whether it is holding a table that is being waited upon
the the flush phase of the process that took the global read
lock.

The same problem also affected the following statements:

LOCK TABLES .. WRITE
UPDATE .. SET (update and multi-table update)
TRUNCATE TABLE ..
LOAD DATA ..

The solution is to make the above statements wait for a impending
global read lock before opening the tables. If there is no
impending global read lock, the statement raises a temporary
protection against global read locks and progresses smoothly
towards completion.

Important notice: the patch does not try to address all possible
cases, only those which are common and can be fixed unintrusively
enough for 5.0.

mysql-test/r/lock_multi.result:
  Add test case result for Bug#43230
mysql-test/t/lock_multi.test:
  Add test case for Bug#43230
sql/sql_lex.cc:
  Initialize flag.
sql/sql_lex.h:
  Add a flag to the lexer.
sql/sql_parse.cc:
  Wait for the global read lock is a write lock is going to be
  taken. The wait is done before opening tables.
sql/sql_yacc.yy:
  Protect against the GRL if its a SELECT .. FOR UPDATE or LOCK TABLES
  .. WRITE statement.

1 files changed, 16 insertions, 0 deletions

diff --git a/sql/sql_lex.h b/sql/sql_lex.h
index 9f020f4adc5..3e762581e04 100644
--- a/sql/sql_lex.h
+++ b/sql/sql_lex.h
@@ -1176,6 +1176,22 @@ typedef struct st_lex : public Query_tables_list
 
   bool escape_used;
 
+  /*
+    Special case for SELECT .. FOR UPDATE and LOCK TABLES .. WRITE.
+
+    Protect from a impending GRL as otherwise the thread might deadlock
+    if it starts waiting for the GRL in mysql_lock_tables.
+
+    The protection is needed because there is a race between setting
+    the global read lock and waiting for all open tables to be closed.
+    The problem is a circular wait where a thread holding "old" open
+    tables will wait for the global read lock to be released while the
+    thread holding the global read lock will wait for all "old" open
+    tables to be closed -- the flush part of flush tables with read
+    lock.
+  */
+  bool protect_against_global_read_lock;
+
   st_lex();
 
   virtual ~st_lex()