llvm-svn: 266760

Some codes that use TLS fail intermittently because one thread tries to write
TLS values after the TLS key has been destroyed by another thread. This happens
when one thread executes library shutdown (and destroys TLS keys), while another
thread starts to execute the TLS key destructor routine. Before this change, the
kmp_init_runtime flag was checked before calling pthread_* TLS functions, but
this flag is set to FALSE later than the destruction of the TLS keys, which
leads to failure. The fix is to check kmp_init_gtid instead, as this flag is
unset *before* the destruction of TLS keys.

Differential Revision: http://reviews.llvm.org/D19022

llvm-svn: 266674

llvm-svn: 264776

From the standard: A doacross loop nest is a loop nest that has cross-iteration
dependence. An iteration is dependent on one or more lexicographically earlier
iterations. The ordered clause parameter on a loop directive identifies the
loop(s) associated with the doacross loop nest.

The init/fini routines allocate/free doacross buffer(s) for each loop for each
thread.  The wait routine waits for a flag designated by the dependence vector.
The post routine sets the flag designated by current iteration vector.  We use
a similar technique of shared buffer indices that covers up to 7 nowait loops
executed simultaneously by different threads (number 7 has no real meaning,
just heuristic value).  Also, the size of structures are kept intact via
reducing dummy arrays.

This needs to be put into the OpenMP runtime library in order for the compiler
team to develop the compiler side of the implementation.

Differential Revision: http://reviews.llvm.org/D17399

llvm-svn: 262532

This change introduces the new OpenMP 4.5 affinity api surrounding
OpenMP Places. There are six new entry points:

Typically called in serial region:
 * omp_get_num_places - returns the number of places available to the execution
       environment in the place list.
 * omp_get_place_num_procs - returns the number of processors available to the
       execution environment in the specified place.
 * omp_get_place_proc_ids - returns the numerical identifiers of the processors
       available to the execution environment in the specified place.

Typically called inside parallel region:
 * omp_get_place_num - returns the place number of the place to which the
       encountering thread is bound.
 * omp_get_partition_num_places - returns the number of places in the place
       partition of the innermost implicit task.
 * omp_get_partition_place_nums - returns the list of place numbers
       corresponding to the places in the place-var ICV of the innermost
       implicit task.

Differential Revision: http://reviews.llvm.org/D17417

llvm-svn: 261915

The problem is that the master's thread state was not saved before entering a
parallel region so it does not remember tasks when it returns.

llvm-svn: 260306

When the code behind the barrier is executed, the master thread may have
already resumed execution. That's why we cannot safely assume that *pteam
is not yet freed.

This has been introduced by r258866.

llvm-svn: 259037

Removing extraneous { } bracket sections.  Unindenting blocks of
code as a result.  Also removing empty #ifdef KMP_STUB

llvm-svn: 258986

Removing references to non-existent functions, fixing typos.

llvm-svn: 258985

llvm-svn: 258984

For implcit barriers in simple parallel for loops, the order of the OMPT events
was wrong.  The barrier_{begin,end} events came after the implcit_task_end
event for the implcit barrier at the end of the parallel region. This is wrong
because the implicit task executes the barrier before ending. This patch fixes
the order of the event: It will be triggerd now just before
__kmp_pop_current_task_from_thread() is called.

Patch by Tim Cramer

Differential Revision: http://reviews.llvm.org/D16347

llvm-svn: 258866

Change (__kmp_mic_type != non_mic) to (__kmp_mic_type == mic2)

llvm-svn: 257380

llvm-svn: 255901

Fix for crash in the teams construct in case user sets OMP_THREAD_LIMIT to a
number less than the number of processors. Now the number of threads will be
silently reduced if the user didn't specify teams parameters or with a
warning if the user specified teams parameters conflicting with
OMP_THREAD_LIMIT.

Differential Revision: http://reviews.llvm.org/D14732

llvm-svn: 254322

llvm-svn: 253264

llvm-svn: 252953

in __kmp_free_team(), the team's number of processors can be == 1.

llvm-svn: 252086

llvm-svn: 252084

This is a refactoring of the task_team code that more elegantly handles the two
task_team case. Two task_teams per team are kept in use for the lifetime of the
team. Thus no reference counting is needed.

Differential Revision: http://reviews.llvm.org/D13993

llvm-svn: 252082

The th.th_task_state for the master thread at the start of a nested parallel
should not be zeroed in __kmp_allocate_team() because it is later put in the
stack of states in __kmp_fork_call() for further re-use after exiting the
nested region. It is zeroed after being put in the stack.

Differential Revision: http://reviews.llvm.org/D13702

llvm-svn: 250847

Without this fix, cancellation requests in one parallel region cause
cancellation of the second region even though the second one was
not intended to be cancelled.

llvm-svn: 250727

llvm-svn: 249725

These changes improve/update the trace messages and debug asserts related to
the previous wait/release checkin.

llvm-svn: 249717

These changes improve the wait/release mechanism for threads spinning in 
barriers that are handling tasks while spinnin by providing feedback to the 
barriers about any task stealing that occurs.

Differential Revision: http://reviews.llvm.org/D13353

llvm-svn: 249711

llvm-svn: 248204

Prior to this change, OMPT had a status flag ompt_status, which could take
several values. This was due to an earlier OMPT design that had several levels
of enablement (ready, disabled, tracking state, tracking callbacks). The
current OMPT design has OMPT support either on or off.
This revision replaces ompt_status with a boolean flag ompt_enabled, which 
simplifies the runtime logic for OMPT.

Patch by John Mellor-Crummey

Differential Revision: http://reviews.llvm.org/D12999

llvm-svn: 248189

The OMPT specification has changed. This revision brings the LLVM OpenMP
implementation up to date.

Technical overview of changes:
Previously, a public weak symbol ompt_initialize was called after the OpenMP
runtime is initialized. The new interface calls a global weak symbol ompt_tool
prior to initialization. If a tool is present, ompt_tool returns a pointer to
a function that matches the signature for ompt_initialize. After OpenMP is 
initialized the function pointer is called to initialize a tool.
Knowing that OMPT will be enabled before initialization allows OMPT support to
be initialized as part of initialization instead of back patching
initialization of OMPT support after the fact.
Post OpenMP initialization support has been generalized moves from
ompt-specific.c into ompt-general.c, since the OMPT initialization logic is no
longer implementation specific.

Patch by John Mellor-Crummey

Differential Revision: http://reviews.llvm.org/D12998

llvm-svn: 248187

This change adds guards to the code in places where they are missing to enable
the OpenMP 3.0 build.

Patch by Diego Caballero and Johnny Peyton

Mailing List: http://lists.llvm.org/pipermail/openmp-dev/2015-September/000935.html

llvm-svn: 248178

This only triggered when built in debug mode with OMPT enabled:
__kmp_wait_template expected the state of the current thread to be either
ompt_state_idle or ompt_state_wait_barrier{,_implicit,_explicit}.

Patch by Jonas Hahnfeld

Differential Revision: http://reviews.llvm.org/D12754

llvm-svn: 247339

Some of this is improvement to code suggested by Hal Finkel. Four changes here:
1.Cleanup of hierarchy code to handle all hierarchy cases whether affinity is available or not
2.Separated this and other classes and common functions out to a header file
3.Added a destructor-like fini function for the hierarchy (and call in __kmp_cleanup)
4.Remove some redundant code that is hopefully no longer needed

Differential Revision: http://reviews.llvm.org/D12449

llvm-svn: 247326

The fix is to make b_arrived flag 64 bit in both structures - kmp_balign_team_t
and kmp_balign_t. Otherwise when flag in kmp_balign_team_t wrapped over
UINT_MAX the library hangs.

Differential Revision: http://reviews.llvm.org/D12563

llvm-svn: 247320

The th.th_team_nproc is assigned in __kmp_allocate_thread() just 3 lines above,
so there is no need to assign the same value again.

llvm-svn: 246703

Conditionally include the fork_context parameter to __kmp_join_call()
only if OMPT_SUPPORT=1

Differential Revision: http://reviews.llvm.org/D12495

llvm-svn: 246460

llvm-svn: 245286

llvm-svn: 245209

llvm-svn: 245206

This removes some statistics counters and timers which were not used,
adds new counters and timers for some language features that were not
monitored previously and separates the counters and timers into those
which are of interest for investigating user code and those which are
only of interest to the developer of the runtime itself.
The runtime developer statistics are now ony collected if the
additional #define KMP_DEVELOPER_STATS is set.

Additional user statistics which are now collected include:
* Count of nested parallelism (omp parallel inside a parallel region)
* Count of omp distribute occurrences
* Count of omp teams occurrences
* Counts of task related statistics (taskyield, task execution, task
  cancellation, task steal)
* Values passed to omp_set_numtheads
* Time spent in omp single and omp master

None of this affects code compiled without stats gathering enabled,
which is the normal library build mode.

This also fixes the CMake build by linking to the standard c++ library
when building the stats library as it is a requirement.  The normal library
does not have this requirement and its link phase is left alone.

Differential Revision: http://reviews.llvm.org/D11759

llvm-svn: 244677

Compiling simple testcase with g++ and linking it to the LLVM OpenMP runtime
compiled in debug mode trips an assertion that produces a fatal error. When
the assertion is skipped, the program runs successfully to completion and 
produces the same answer as the sequential code. Intel will restore the
assertion with a patch that fixes the issues that cause it to trip.

Patch by John Mellor-Crummey

Differential Revision: http://reviews.llvm.org/D11269

llvm-svn: 243032

This patch makes it possible for a performance tool that uses call stack
unwinding to map implementation-level call stacks from master and worker
threads into a unified global view. There are several components to this patch.

include/*/ompt.h.var
  Add a new enumeration type that indicates whether the code for a master task
    for a parallel region is invoked by the user program or the runtime system
  Change the signature for OMPT parallel begin/end callbacks to indicate whether
    the master task will be invoked by the program or the runtime system. This
    enables a performance tool using call stack unwinding to handle these two
    cases differently. For this case, a profiler that uses call stack unwinding
    needs to know that the call path prefix for the master task may differ from
    those available within the begin/end callbacks if the program invokes the
    master.

kmp.h
  Change the signature for __kmp_join_call to take an additional parameter
  indicating the fork_context type. This is needed to supply the OMPT parallel
  end callback with information about whether the compiler or the runtime
  invoked the master task for a parallel region.

kmp_csupport.c
  Ensure that the OMPT task frame field reenter_runtime_frame is properly set
    and cleared before and after calls to fork and join threads for a parallel
    region.
  Adjust the code for the new signature for __kmp_join_call.
  Adjust the OMPT parallel begin callback invocations to carry the extra
    parameter indicating whether the program or the runtime invokes the master
    task for a parallel region.

kmp_gsupport.c
  Apply all of the analogous changes described for kmp_csupport.c for the GOMP
    interface
  Add OMPT support for the GOMP combined parallel region + loop API to
    maintain the OMPT task frame field reenter_runtime_frame.

kmp_runtime.c:
  Use the new information passed by __kmp_join_call to adjust the OMPT
    parallel end callback invocations to carry the extra parameter indicating
    whether the program or the runtime invokes the master task for a parallel
    region.

ompt_internal.h:
  Use the flavor of the parallel region API (GNU or Intel) to determine who
    invokes the master task.

Differential Revision: http://reviews.llvm.org/D11259

llvm-svn: 242817

These changes enable external debuggers to conveniently interface with 
the LLVM OpenMP Library.  Structures are added which describe the important
internal structures of the OpenMP Library e.g., teams, threads, etc.
This feature is turned on by default (CMake variable LIBOMP_USE_DEBUGGER)
and can be turned off with -DLIBOMP_USE_DEBUGGER=off.

Differential Revision: http://reviews.llvm.org/D10038

llvm-svn: 241832