Valgrind warnings in maria-5.3-mwl34

[Philip Stoev]

A simple RQG test produced the misc valgrind warnings from the attached file. Most seem to be related to handling malformed datetimes, however note that there are 2 warnings about innodb and myisam using an undefined value for some internal operations.

Let me know if individual test cases are required -- I can provide the RQG grammar and/or MTR.

[Philip Stoev (pstoev-askmonty)]

valgrind warnings

[Michael Widenius (monty)]

I looked at the valgrind result, but unfortunately without the test cases that caused the failures there is not much I can do.
So please provide the RQG and MTR grammar so that I can test this myself.

I think that I manged to fix a couple of the valgrind errors as part of other bug fixes, but looking at the log there are some left that would be nice to get rid off.

[Philip Stoev (pstoev-askmonty)]

Unsimplified test case. When run without valgrind, causes the following crash as well:

# 2011-05-11T13:15:59 Error: Memory allocated at string.c:39 was overrun, discovered at 'string.c:84'
# 2011-05-11T13:15:59 Error: Memory allocated at string.c:39 was overrun, discovered at 'string.c:84'
# 2011-05-11T13:15:59 Error: Memory allocated at string.c:39 was overrun, discovered at 'string.c:84'

# 2011-05-11T13:15:59 #0 0x00857424 in __kernel_vsyscall ()
# 2011-05-11T13:15:59 #0 0x00857424 in __kernel_vsyscall ()
# 2011-05-11T13:15:59 #1 0x008263c3 in pthread_kill () from /lib/libpthread.so.0
# 2011-05-11T13:15:59 #2 0x0878175c in my_write_core (sig=11) at stacktrace.c:426
# 2011-05-11T13:15:59 #3 0x0828ebf4 in handle_segfault (sig=11) at mysqld.cc:2776
# 2011-05-11T13:15:59 #4 <signal handler called>
# 2011-05-11T13:15:59 #5 0x006cfc91 in vfprintf () from /lib/libc.so.6
# 2011-05-11T13:15:59 #6 0x006d1572 in buffered_vfprintf () from /lib/libc.so.6
# 2011-05-11T13:15:59 #7 0x006cc5a3 in vfprintf () from /lib/libc.so.6
# 2011-05-11T13:15:59 #8 0x006d6b9f in fprintf () from /lib/libc.so.6
# 2011-05-11T13:15:59 #9 0x087658b7 in _checkchunk (irem=0xae585be8, filename=0x89746ff "string.c", lineno=84) at safemalloc.c:492
# 2011-05-11T13:15:59 #10 0x08765aa2 in _sanity (filename=0x89746ff "string.c", lineno=84) at safemalloc.c:535
# 2011-05-11T13:15:59 #11 0x087650b9 in _myfree (ptr=0xae5859b0, filename=0x89746ff "string.c", lineno=84, myflags=0) at safemalloc.c:280
# 2011-05-11T13:15:59 #12 0x08764fc8 in _myrealloc (ptr=0xae5859b0, size=353, filename=0x89746ff "string.c", lineno=84, MyFlags=16) at safemalloc.c:258
# 2011-05-11T13:15:59 #13 0x0877326f in dynstr_realloc (str=0xae89ec98, additional_size=10) at string.c:84
# 2011-05-11T13:15:59 #14 0x0875b9ee in dynamic_column_var_uint_store (str=0xae89ec98, val=8) at ma_dyncol.c:132
# 2011-05-11T13:15:59 #15 0x0875bfb5 in dynamic_column_string_store (str=0xae89ec98, string=0xae5079d4, charset=0x8ab5360) at ma_dyncol.c:409
# 2011-05-11T13:15:59 #16 0x0875ca3a in data_store (str=0xae89ec98, value=0xae5079d0) at ma_dyncol.c:816
# 2011-05-11T13:15:59 #17 0x0875ebc8 in dynamic_column_update_many (str=0xae89ed84, add_column_count=1, column_numbers=0xae507a10, values=0xae5079d0) at ma_dyncol.c:2010
# 2011-05-11T13:15:59 #18 0x082211b4 in Item_func_dyncol_add::val_str (this=0xae5075c8, str=0xae89ef30) at item_strfunc.cc:3745
# 2011-05-11T13:15:59 #19 0x082213d2 in Item_dyncol_get::get_dyn_value (this=0xae5076d8, val=0xae89eec4, tmp=0xae89ef30) at item_strfunc.cc:3808
# 2011-05-11T13:15:59 #20 0x082217fc in Item_dyncol_get::val_int (this=0xae5076d8) at item_strfunc.cc:3915
# 2011-05-11T13:15:59 #21 0x081eed11 in Item_func_signed::val_int (this=0xae507758) at item_func.cc:996
# 2011-05-11T13:15:59 #22 0x081daeb6 in Item::update_null_value (this=0xae507758) at item.h:909
# 2011-05-11T13:15:59 #23 0x081ff268 in Item_func::is_null (this=0xae507758) at item_func.h:159
# 2011-05-11T13:15:59 #24 0x0820f35c in Item_func_isnull::val_int (this=0xae5077d8) at item_cmpfunc.cc:4748
# 2011-05-11T13:15:59 #25 0x0832f44a in evaluate_join_record (join=0xae547d98, join_tab=0xae507d60, error=0) at sql_select.cc:14061
# 2...

[Michael Widenius (monty)]

The problem was using a wrong (too short) strong to column_add().
I have now added a check to detect this + a test case.