min() and max() functions are implemented in MySQL as macros.
This means that max(a,b) is expanded to: ((a) > (b) ? (a) : (b))
Note how 'a' is quoted two times.
Now imagine 'a' is a recursive function call that's several 10s of levels deep.
And the recursive function does max() with a function arg as well to dive into
recursion.
This means that simple function call can take most of the clock time.
Identified and fixed several such calls to max()/min() : including the IF()
sql function implementation.
Calling List<Cached_item>::delete_elements for the same list twice
caused a crash of the server in the function JOIN::cleaunup.
Ensured that delete_elements() in JOIN::cleanup would be called only once.
Tables in the table definition cache are keeping a cache buffer for blob
fields which can consume a lot of memory.
This patch introduces a maximum size threshold for these buffers.
Range scan in descending order for c <= <col> <= c type of
ranges was ignoring the DESC flag.
However some engines like InnoDB have the primary key parts
as a suffix for every secondary key.
When such primary key suffix is used for ordering ignoring
the DESC is not valid.
But we generally would like to do this because it's faster.
Fixed by performing only reverse scan if the primary key is used.
Removed some dead code in the process.
Range scan in descending order for c <= <col> <= c type of
ranges was ignoring the DESC flag.
However some engines like InnoDB have the primary key parts
as a suffix for every secondary key.
When such primary key suffix is used for ordering ignoring
the DESC is not valid.
But we generally would like to do this because it's faster.
Fixed by performing only reverse scan if the primary key is used.
Removed some dead code in the process.
- In QUICK_INDEX_MERGE_SELECT::read_keys_and_merge: when we got table->sort from Unique,
tell init_read_record() not to use rr_from_cache() because a) rowids are already sorted
and b) it might be that the the data is used by filesort(), which will need record rowids
(which rr_from_cache() cannot provide).
- Fully de-initialize the table->sort read in QUICK_INDEX_MERGE_SELECT::get_next(). This fixes BUG#35477.
(bk trigger: file as fix for BUG#35478).
build)
The crash was caused by freeing the internal parser stack during the parser
execution.
This occured only for complex stored procedures, after reallocating the parser
stack using my_yyoverflow(), with the following C call stack:
- MYSQLparse()
- any rule calling sp_head::restore_lex()
- lex_end()
- x_free(lex->yacc_yyss), xfree(lex->yacc_yyvs)
The root cause is the implementation of stored procedures, which breaks the
assumption from 4.1 that there is only one LEX structure per parser call.
The solution is to separate the LEX structure into:
- attributes that represent a statement (the current LEX structure),
- attributes that relate to the syntax parser itself (Yacc_state),
so that parsing multiple statements in stored programs can create multiple
LEX structures while not changing the unique Yacc_state.
Now, Yacc_state and the existing Lex_input_stream are aggregated into
Parser_state, a structure that represent the complete state of the (Lexical +
Syntax) parser.
and value-list
The server returns unexpected results if a right side of the
NOT IN clause consists of NULL value and some constants of
the same type, for example:
SELECT * FROM t WHERE NOT t.id IN (NULL, 1, 2)
may return 3, 4, 5 etc if a table contains these values.
The Item_func_in::val_int method has been modified:
unnecessary resets of an Item_func_case::has_null field
value has been moved outside of an argument comparison
loop. (Also unnecessary re-initialization of the null_value
field has been moved).
offset for time part in UUIDs was 1/1000 of what it
should be. In other words, offset was off.
Also handle the case where we count into the future
when several UUIDs are generated in one "tick", and
then the next call is late enough for us to unwind
some but not all of those borrowed ticks.
Lastly, handle the case where we keep borrowing and
borrowing until the tick-counter overflows by also
changing into a new "numberspace" by creating a new
random suffix.
offset for time part in UUIDs was 1/1000 of what it
should be. In other words, offset was off.
Also handle the case where we count into the future
when several UUIDs are generated in one "tick", and
then the next call is late enough for us to unwind
some but not all of those borrowed ticks.
Lastly, handle the case where we keep borrowing and
borrowing until the tick-counter overflows by also
changing into a new "numberspace" by creating a new
random suffix.
Problem: reading/writing data from/to an address without proper
alignment leads to SIGBUS on some platforms.
Fix: use the correct data type when dereferencing variable values.
