Deadlock in DbugParse, on Linux.
In 10.1, DBUG recursive mutex was improperly implemented.
CODE_STATE::locked counter was never updated.
Copy the code around LockMutex/UnlockMutex from 10.2
FreeState(): replace pointer to freed memory with NULL. This actually fixes a crash
which is use-after-free as reported by ASAN
DbugParse(): unconditionally lock mutex because we're touching shared init_settings.keywords
DbugParse(): removed mutex lock/unlock which should protect file writes only.
And no file writes happen in this function.
DbugFlush(): move mutex_unlock out of this method because fflush() doesn't
need any locking.
Slow stuff like mutex lock/unlock and accessing errno (TLS)
is moved to a more narrow scope.
MYSQL_PLUGIN_IMPORT did not work correctly for the RocksDB helper library
rocksdb_aux_lib, because that library was not compiled with
-DMYSQL_DYNAMIC_PLUGIN.
Fix DBUG such that it does not depend on exported data, only on functions
(which do not need MYSQL_PLUGIN_IMPORT decoration)
Use a "getter" function _db_my_assert() instead of DLL-exported variable.
Also, reduce object code duplication by moving more of the DBUG_ASSERT
logic inside the _db_my_assert() function, and add unlikely() and
ATTRIBUTE_COLD hints to ensure that the 'assertion failed' code will
be separated from the main control flow logic. Thus, the compiler can
move the unlikely() code to the end of the compiled function, reachable
via a forward conditional branch, which the processor's branch predictor
could assume 'not taken'.
- Fix win64 pointer truncation warnings
(usually coming from misusing 0x%lx and long cast in DBUG)
- Also fix printf-format warnings
Make the above mentioned warnings fatal.
- fix pthread_join on Windows to set return value.
If compiling a non DBUG binary with
-DDBUG_ASSERT_AS_PRINTF asserts will be
changed to printf + stack trace (of stack
trace are enabled).
- Changed #ifndef DBUG_OFF to
#ifdef DBUG_ASSERT_EXISTS
for those DBUG_OFF that was just used to enable
assert
- Assert checking that could greatly impact
performance where changed to DBUG_ASSERT_SLOW which
is not affected by DBUG_ASSERT_AS_PRINTF
- Added one extra option to my_print_stacktrace() to
get more silent in case of stack trace printing as
part of assert.
Define my_thread_id as an unsigned type, to avoid mismatch with
ulonglong. Change some parameters to this type.
Use size_t in a few more places.
Declare many flag constants as unsigned to avoid sign mismatch
when shifting bits or applying the unary ~ operator.
When applying the unary ~ operator to enum constants, explictly
cast the result to an unsigned type, because enum constants can
be treated as signed.
In InnoDB, change the source code line number parameters from
ulint to unsigned type. Also, make some InnoDB functions return
a narrower type (unsigned or uint32_t instead of ulint;
bool instead of ibool).
MariaDB Server is unnecessarily evaluating the arguments of
DBUG_PRINT() macros when the label is not defined.
The macro DBUG_LOG() for C++ operator<< output which was added for
InnoDB diagnostics in MySQL 5.7 is missing from MariaDB. Unlike the
MySQL 5.7 implementation, MariaDB will avoid allocating and
initializing the output string when the label is not defined.
Introduce DBUG_OUT("crypt") and DBUG_OUT("checksum") for some InnoDB
diagnostics, replacing some use of ib::info().
when func1 calls func2 from DBUG_RETURN, dbug shows the trace as
| > func1
| < func1
| > func2
| < func2
because DBUG_LEAVE happens before func2(). Change that to invoke
DBUG_LEAVE when the local variable goes out of scope. This uses
gcc specific __attribute__((cleanup)).
- Reset static variables that are used to signal "init done" for DBUG, in dbug_end()
- Set string server variables to NULL after memory for the value is freed - avoids double free()
- fix DBUG_ASSERTs that happened during reinitialization.
client/mysqldump.c:
Slave needs to be initialized with 0
dbug/dbug.c:
Removed not existing function
plugin/semisync/semisync_master.cc:
Fixed compiler warning
sql/opt_range.cc:
thd needs to be set early as it's used in some error conditions.
sql/sql_table.cc:
Changed to use uchar* to make array indexing portable
storage/innobase/handler/ha_innodb.cc:
Removed not used variable
storage/maria/ma_delete.c:
Fixed compiler warning
storage/maria/ma_write.c:
Fixed compiler warning
sql/sql_insert.cc:
CREATE ... IF NOT EXISTS may do nothing, but
it is still not a failure. don't forget to my_ok it.
******
CREATE ... IF NOT EXISTS may do nothing, but
it is still not a failure. don't forget to my_ok it.
sql/sql_table.cc:
small cleanup
******
small cleanup
All client functions that can block on I/O have alternate _start() and
_cont() versions that do not block but return control back to the
application, which can then issue I/O wait in its own fashion and later
call back into the library to continue the operation.
Works behind the scenes by spawning a co-routine/fiber to run the
blocking operation and suspend it while waiting for I/O. This
co-routine/fiber use is invisible to applications.
For i368/x86_64 on GCC, uses very fast assembler co-routine support. On
Windows uses native Win32 Fibers. Falls back to POSIX ucontext on other
platforms. Assembler routines for more platforms are relatively easy to
add by extending mysys/my_context.c, eg. similar to the Lua lcoco
library.
For testing, mysqltest and mysql_client_test are extended with the
option --non-blocking-api. This causes the programs to use the
non-blocking API for database access. mysql-test-run.pl has a similar
option --non-blocking-api that uses this, as well as additional
testcases.
An example program tests/async_queries.c is included that uses the new
non-blocking API with libevent to show how, in a single-threaded
program, to issue many queries in parallel against a database.
client/async_example.c:
Fix const warning
******
Fix bug with wrong timeout value for poll().
include/Makefile.am:
Fix missing include for `make dist`
include/mysql.h:
Add prototypes for all non-blocking API calls.
include/mysql.h.pp:
Add prototypes for all non-blocking API calls.
mysys/my_context.c:
Fix type warning for makecontext() function pointer argument.
sql-common/mysql_async.c:
Fix crashes in the non-blocking API for functions that can take MYSQL argument
that is NULL.
tests/Makefile.am:
Add header file to `make dist`
tests/mysql_client_test.c:
Replace blocking calls with wrappers around the non-blocking calls, used in
mysql_client_test to test the new non-blocking API.
tests/nonblock-wrappers.h:
Replace blocking calls with wrappers around the non-blocking calls, used in
mysql_client_test to test the new non-blocking API.
... but differently
client/mysqltest.cc:
my_safe_print_str() don't append \n anymore
dbug/dbug.c:
restore safemalloc as a part of dbug suite
dbug/user.r:
restore 'S' flag documentation
include/my_dbug.h:
restore safemalloc as a part of dbug suite
include/my_sys.h:
move valgrind defines to a dedicated header
mysys/my_malloc.c:
use new safemalloc
mysys/stacktrace.c:
don't append \n. let the calller do it, if needed
sql/mysqld.cc:
my_safe_print_str() don't append \n anymore
GCC 4.6 has new -Wunused-but-set-variable flag, which is enabled
by -Wall, that causes GCC to emit a warning whenever a local variable
is assigned to, but otherwise unused (aside from its declaration).
Since the maintainer mode uses -Wall and -Werror, source code which
triggers these warnings will be rejected. That is, these warnings
become hard errors.
The solution is to fix the code which triggers these specific warnings.
In most of the cases, this is a welcome cleanup as code which triggers
this warning is probably dead anyway.
dbug/dbug.c:
Unused but set.
libmysqld/lib_sql.cc:
Length is not necessary as the converted error message is always
null-terminated.
sql/item_func.cc:
Make get_var_with_binlog private to this compilation unit.
If a error was raised, do not attempt to evaluate the user
variable as the statement execution will be interrupted
anyway.
sql/mysqld.cc:
Use a void expression to silence the warning. Avoids the use of
macros that would make the code more unreadable than it already is.
sql/protocol.cc:
Length is not necessary as the converted error message is always
null-terminated. Remove unnecessary casts and assignment.
sql/sql_class.h:
Function is only used in a single compilation unit.
sql/sql_load.cc:
Only use the variable outside of EMBEDDED_LIBRARY.
storage/innobase/btr/btr0cur.c:
Do not retrieve field, only the record length is being used.
storage/perfschema/pfs.cc:
Use a void expression to silence the warning.
tests/mysql_client_test.c:
Unused but set.
unittest/mysys/lf-t.c:
Unused but set.
