StringStore NULL check includes a check for _CpNuLl_ in the
StringStore. This is a case should never happen but we keep it just in
case.
Unfortunately this check was skipping 4*8 bytes instead of just 4 bytes.
This is definitely bad behaviour but it could cause an out-of-bounds read
based crash.
In the FLOAT/DOUBLE to string conversions a class global string was used
to store the result. Unfortunately it is possible for an instance of
this class to be used by multiple threads of PrimProc simultaneously.
This would cause a race and data corruption or more likely a crash.
This fix passes a string object from the caller to use instead.
Anything that links against joblist will spin up a threadpool upon
startup. This includes the tools setConfig/getConfig. It is possible on
a low core count machine or low CPU speed that the signal to the prune
thread to shutdown is sent before the thread has completed startup when
these quick-running tools are used.
This fix adds a mutex so that spin up and shutdown can't happen at the
same time as well as a stop watch in case we are shutting down when
either the thread is running or we haven't fully started.
When a thread has been idle for 10 minutes and we have too many threads
in the threadpool the thread will be pruned. This is done by the
thread's main function just returning. Unfortunately this does not free
up the memory, the thread either needs to be joined or detatched.
We cannot use detached threads since there are mutexes and conditional
variables between the main thread and the threadpool threads. If the
main thread finishes before the threadpool threads (as would happen in
cpimport) then crashes occur. The parent needs to wait on the child
threads which is the whole point in joining.
So this fix spawns a new thread which every minute will check the list
of threads to be joined due to timeout and join them.
We have had to use an adapted version of boost::thread_group so that we
can join a single thread based off its thread ID.
In addition with have modified PriorityThreadPool to use detached
threads since this does not need to signal the child threads at the end.
PriorityThreadPool didn't have very good error handling. If something
failed it would just ignore whatever was being processed. This could
lead to a query continuing without retreiving all of the required data.
This patch adds error handling, sending a message back to the client
and a log message. It also destroys and recreates the pool thread.
If negative count number is more than the number of characters in the
string then it should always return the string.
For example if a table contains SUBSTRING_INDEX('zzz', 'z', -5) should
return 'zzz'. Before this patch it would return NULL.
StringStore originally worked by returning a 32bit pointer to a memory
location and storing the length with that pointer. This allowed 4GB to
be stored in 64KB blocks. With 1.1 we used the high bit to signify a
TEXT/BLOB string of > 64KB reducing the max capacity to 2GB but without
any bounds checking.
So, if you went over the 2GB mark the getter would think you are trying
to get a long string instead of a short one and come up empty. It would
then return NULL.
This patch uses 64bit memory points still retaining the high bit to
signify long strings. It also now stores the length with the string
rather than with the pointer to allow the full 64bits for pointers.
It also adds a bounds check for small strings.
MariaDB Server 10.2.14 changed the order that CASE items are processed.
This broke the engine's CASE handling. This patch uses the new order
instead since this is what will be used in 10.2 and 10.3 going forward.
In ha_calpont_execplan, Allow OR to be parsed; in searched_case parsing, reverse the order of processing arguments so that ptWorkStack.pop() is executed in the same order as the arguments being processed.
In func_case, modify to pass left and right to getBoolVal, if they exist.
SQL-92 basically specifies for a NOPAD collation that only space should
be ignored for matches. Tabs and other whitespace are handled
differently. We don't fully support collations yet so we assume the
defaults.
For equality string matches other engines ignore trailing whitespace
(this does not apply to LIKE matches). So we should do the same. This
patch trims whitespace for MIN/MAX extent elimination checks, fixed
width columns and dictionary columns during equality matches against
constants (SELECT * FROM t1 WHERE b = 'ABC').
