Hi Thomas,
Thanks for the review. See the updated patch in attachment and my
comments below.
On 07/07/2026 08:58, Thomas Schwinge wrote:
Hi Paul-Antoine!
On 2026-07-01T16:57:43+0200, Paul-Antoine Arras <[email protected]> wrote:
While working on a follow-up patch, I realised that I did not include
target-side versions of GOMP_distribute_static_worksharing_start et al.
The attached fixup patch does so and adds testcases.
ACK.
On 30/06/2026 14:56, Paul-Antoine Arras wrote:
--- gcc/common.opt
+++ gcc/common.opt
@@ -2503,6 +2503,13 @@ fomit-frame-pointer
Common Var(flag_omit_frame_pointer) Optimization
When possible do not generate stack frames.
+fopenmp-ompt
+Common Var(flag_openmp_ompt) EnabledBy(fopenmp-ompt-detailed)
+
+
+fopenmp-ompt-detailed
+Common Var(flag_openmp_ompt_detailed)
+
fopenmp-target-simd-clone
Common Alias(fopenmp-target-simd-clone=,any,none)
Some short "help text" for the new options, please.
Also, need documenting in 'gcc/doc/invoke.texi', in the "Option Summary"
as well as "OpenMP and OpenACC Options". Probably don't need anything
additional in 'gcc/fortran/gfortran.texi'.
Added documentation as needed.
We have '-fopenmp', '-fopenmp-simd' defined in 'gcc/c-family/c.opt',
'gcc/fortran/lang.opt', as, from the user perspective, they only apply to
those respective front ends. I did wonder whether we shouldn't handle
'-fopenmp-ompt', '-fopenmp-ompt-detailed' similarly? But on the other
head, those are 'Common' infrastructure, in the sense that OpenMP/OMPT
functionality also applies if OpenMP is used "as an implementation
detail" in GCC, such as when 'Common' '-ftree-parallelize-loops'
synthesizes OpenMP constructs (and therefore implicitly enables OpenMP
support) to parallelize Algol 68 code, for example. In that case, it's
then certainly valid to expect to see OpenMP/OMPT events reported, and
'-fopenmp-ompt', '-fopenmp-ompt-detailed' than likewise apply.
Therefore, it appears correct to have those 'Common' options. (We shall
add a corresponding Algol 68 test case (or other non-C, C++, Fortran, of
course), once that's feasible to do.)
Indeed, my reasoning was that, contrary to '-fopenmp', '-fopenmp-ompt'
has no effect on either front end and does not affect user-written code,
only code generation.
--- gcc/omp-builtins.def
+++ gcc/omp-builtins.def
@@ -445,6 +445,11 @@ DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SECTIONS_END_NOWAIT,
BT_FN_VOID, ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_START, "GOMP_single_start",
BT_FN_BOOL, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_START_WITH_END,
+ "GOMP_single_start_with_end", BT_FN_BOOL,
+ ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_END, "GOMP_single_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_COPY_START, "GOMP_single_copy_start",
BT_FN_PTR, ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_COPY_END, "GOMP_single_copy_end",
@@ -503,9 +508,29 @@ DEF_GOMP_BUILTIN (BUILT_IN_GOMP_ERROR, "GOMP_error",
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_HAS_MASKED_THREAD_NUM,
"GOMP_has_masked_thread_num", BT_FN_BOOL_INT,
ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_MASKED_END, "GOMP_masked_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING,
"GOMP_loop_static_worksharing", BT_FN_COMPLEX_INT,
ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_START,
+ "GOMP_loop_static_worksharing_start", BT_FN_COMPLEX_INT,
+ ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_END,
+ "GOMP_loop_static_worksharing_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_DISPATCH,
+ "GOMP_loop_static_worksharing_dispatch", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING,
"GOMP_distribute_static_worksharing", BT_FN_COMPLEX_INT,
ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_START,
+ "GOMP_distribute_static_worksharing_start", BT_FN_COMPLEX_INT,
+ ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_END,
+ "GOMP_distribute_static_worksharing_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_DISPATCH,
+ "GOMP_distribute_static_worksharing_dispatch", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
All these need prototypes in 'libgomp/libgomp_g.h', too. (This is
especially important for my libgomp/C++ work.) In 'libgomp/libgomp_g.h',
please sort them in the right place: in sections corresponding to their
libgomp implementation file, in order of definition therein.
Added prototypes and ensured proper sorting.
'GOMP_loop_static_worksharing_dispatch',
'GOMP_distribute_static_worksharing_dispatch' are really just a signal,
but don't transmit any further data into libgomp/OMPT?
Yes, that is the requirement.
--- gcc/omp-expand.cc
+++ gcc/omp-expand.cc
@@ -5201,10 +5201,14 @@ expand_omp_for_static_nochunk (struct omp_region
*region,
@@ -5969,10 +6019,14 @@ expand_omp_for_static_chunk (struct omp_region *region,
I've not verified the positions of the individual 'GOMP_[...]' calls that
you're synthesizing.
Let me know if anything needs more in-depth review.
--- gcc/omp-low.cc
+++ gcc/omp-low.cc
@@ -8786,7 +8786,9 @@ lower_omp_single_simple (gomp_single *single_stmt,
gimple_seq *pre_p)
@@ -8923,6 +8925,15 @@ lower_omp_single (gimple_stmt_iterator *gsi_p,
omp_context *ctx)
@@ -9099,6 +9110,14 @@ lower_omp_master (gimple_stmt_iterator *gsi_p,
omp_context *ctx)
Likewise.
--- gcc/opts.cc
+++ gcc/opts.cc
@@ -1538,6 +1538,12 @@ finish_options (struct gcc_options *opts, struct
gcc_options *opts_set,
" %<-fstrict-flex-arrays%> is not present");
}
+ if ((opts->x_flag_openmp_ompt || opts->x_flag_openmp_ompt_detailed)
+ && !opts->x_flag_openmp)
+ error_at (
+ loc,
+ "%<-fopenmp-ompt%> and %<-fopenmp-ompt-detailed%> require %<-fopenmp%>");
+
diagnose_options (opts, opts_set, loc);
}
Looks like the appropriate place for these diagnostics.
--- libgomp/loop.c
+++ libgomp/loop.c
@@ -1186,3 +1186,53 @@ GOMP_loop_ordered_guided_next (long *istart, long *iend)
return gomp_loop_ordered_guided_next (istart, iend);
}
#endif
+
+/* For a worksharing-loop construct with static schedule, return the thread ID
+ and number of threads packed into a single complex value. */
+
+_Complex int
+GOMP_loop_static_worksharing (void)
+{
+ struct gomp_team *team = gomp_thread ()->ts.team;
+ unsigned tid = gomp_thread ()->ts.team_id;
+ unsigned nthreads = team ? team->nthreads : 1;
+ return nthreads + tid * 1I;
+}
+
+/* OMPT variant enabled by -fopenmp-ompt. */
+
+_Complex int
+GOMP_loop_static_worksharing_start (void)
+{
+ struct gomp_team *team = gomp_thread ()->ts.team;
+ unsigned tid = gomp_thread ()->ts.team_id;
+ unsigned nthreads = team ? team->nthreads : 1;
+ return nthreads + tid * 1I;
+}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt-detailed. */
+
+void
+GOMP_loop_static_worksharing_dispatch (void)
+{}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_loop_static_worksharing_end (void)
+{}
+
+/* Return true if the current thread number equals TID.
+ Used to implement the masked construct's filter clause. */
+
+bool
+GOMP_has_masked_thread_num (int tid)
+{
+ return tid == gomp_thread ()->ts.team_id;
+}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_masked_end (void)
+{}
Please update 'libgomp/libgomp_g.h' also for the functions you're moving
into 'libgomp/loop.c' from:
--- libgomp/parallel.c
+++ libgomp/parallel.c
@@ -271,27 +271,6 @@ GOMP_cancel (int which, bool do_cancel)
return true;
}
-/* For a worksharing-loop construct with static schedule, return the thread ID
- and number of threads packed into a single complex value. */
-
-_Complex int
-GOMP_loop_static_worksharing (void)
-{
- struct gomp_team *team = gomp_thread ()->ts.team;
- unsigned tid = gomp_thread ()->ts.team_id;
- unsigned nthreads = team ? team->nthreads : 1;
- return nthreads + tid * 1I;
-}
-
-/* Return true if the current thread number equals TID.
- Used to implement the masked construct's filter clause. */
-
-bool
-GOMP_has_masked_thread_num (int tid)
-{
- return tid == gomp_thread ()->ts.team_id;
-}
... here.
Done.
(I'm not disagreeing -- but was there any specific reason for moving
them?)
The initial reason to put GOMP_loop_static_worksharing in parallel.c was
that its implementation was based on omp_get_num_threads and
omp_get_thread_num which lived there. However, it is not used in
parallel constructs, just in loops.
It was probably a mistake though to move GOMP_has_masked_thread_num
along. It is now reverted.
-
Keep the page break, I guess?
/* The public OpenMP API for thread and team related inquiries. */
int
Restored form feed.
Best,
--
PA
commit 52792ed16b33dead397ecfb2f06805d91f87c6cd
Author: Paul-Antoine Arras <[email protected]>
Date: Tue Jun 30 11:06:44 2026 +0200
openmp: Add OMPT entry points for worksharing-loop, single, and masked constructs
Add two new compiler flags, -fopenmp-ompt and -fopenmp-ompt-detailed, that
cause the compiler to emit OMPT-aware runtime entry points around OpenMP
worksharing constructs. With -fopenmp-ompt, the _start and _end variants
are called instead of the plain worksharing functions; with
-fopenmp-ompt-detailed, a _dispatch stub is additionally called after the
chunk bounds are computed. New stubs are added to libgomp as placeholders
for future OMPT callback invocations.
gcc/ChangeLog:
* common.opt (fopenmp-ompt): New option.
(fopenmp-ompt-detailed): New option.
* doc/invoke.texi (Option Summary): Add -fopenmp-ompt and
-fopenmp-ompt-detailed.
(OpenMP and OpenACC Options): Document -fopenmp-ompt and
-fopenmp-ompt-detailed.
* omp-builtins.def (BUILT_IN_GOMP_SINGLE_START_WITH_END): New builtin.
(BUILT_IN_GOMP_SINGLE_END): New builtin.
(BUILT_IN_GOMP_MASKED_END): New builtin.
(BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_START): New builtin.
(BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_END): New builtin.
(BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_DISPATCH): New builtin.
(BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_START): New builtin.
(BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_END): New builtin.
(BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_DISPATCH): New builtin.
* omp-expand.cc (expand_omp_for_static_nochunk): With -fopenmp-ompt,
emit _start and _end variants instead of the plain worksharing call;
with -fopenmp-ompt-detailed, also emit _dispatch.
(expand_omp_for_static_chunk): Likewise.
* omp-low.cc (lower_omp_single_simple): With -fopenmp-ompt, use
GOMP_single_start_with_end instead of GOMP_single_start.
(lower_omp_single): With -fopenmp-ompt, emit GOMP_single_end.
(lower_omp_master): With -fopenmp-ompt, emit GOMP_masked_end.
* opts.cc (finish_options): Diagnose -fopenmp-ompt and
-fopenmp-ompt-detailed used without -fopenmp.
libgomp/ChangeLog:
* config/gcn/teams.c (GOMP_distribute_static_worksharing_start,
GOMP_distribute_static_worksharing_dispatch,
GOMP_distribute_static_worksharing_end): New functions.
* config/nvptx/teams.c (GOMP_distribute_static_worksharing_start,
GOMP_distribute_static_worksharing_dispatch,
GOMP_distribute_static_worksharing_end): Likewise.
* libgomp.map (GOMP_6.0.2): Export new entry points.
* libgomp_g.h: Add prototypes for GOMP_loop_static_worksharing,
GOMP_loop_static_worksharing_start, GOMP_loop_static_worksharing_dispatch,
GOMP_loop_static_worksharing_end, GOMP_has_masked_thread_num,
GOMP_masked_end, GOMP_single_start_with_end, GOMP_single_end,
GOMP_distribute_static_worksharing_start,
GOMP_distribute_static_worksharing_dispatch, and
GOMP_distribute_static_worksharing_end.
* loop.c (GOMP_loop_static_worksharing): Move here from parallel.c.
(GOMP_loop_static_worksharing_start): New function.
(GOMP_loop_static_worksharing_dispatch): New stub.
(GOMP_loop_static_worksharing_end): New stub.
(GOMP_has_masked_thread_num): Move here from parallel.c.
(GOMP_masked_end): New stub.
* parallel.c (GOMP_loop_static_worksharing): Move to loop.c.
(GOMP_has_masked_thread_num): Move to loop.c.
* single.c (GOMP_single_start_with_end): New function.
(GOMP_single_end): New stub.
(GOMP_single_copy_start): Fix trailing whitespace.
* teams.c (GOMP_distribute_static_worksharing_start): New function.
(GOMP_distribute_static_worksharing_dispatch): New stub.
(GOMP_distribute_static_worksharing_end): New stub.
* testsuite/libgomp.c-c++-common/for-static-1.c: New test.
* testsuite/libgomp.c-c++-common/for-static-2.c: New test.
* testsuite/libgomp.c-c++-common/for-static-3.c: New test.
* testsuite/libgomp.c-c++-common/for-static.h: New test.
gcc/testsuite/ChangeLog:
* c-c++-common/gomp/masked-1.c: Add scan-tree-dump-not check that
GOMP_masked_end is not emitted without -fopenmp-ompt.
* c-c++-common/gomp/for-8.c: Remove; superseded by for-static-*.c.
* c-c++-common/gomp/for-static-1.c: New test; checks that the plain
worksharing builtins are emitted without -fopenmp-ompt.
* c-c++-common/gomp/for-static-2.c: New test; checks that _start and
_end variants are emitted with -fopenmp-ompt.
* c-c++-common/gomp/for-static-3.c: New test; checks that _dispatch is
also emitted with -fopenmp-ompt-detailed.
* c-c++-common/gomp/for-static.h: New shared test header.
* c-c++-common/gomp/masked-3.c: New test for GOMP_masked_end emission
with -fopenmp-ompt.
* c-c++-common/gomp/openmp-ompt-1.c: New test; checks that
-fopenmp-ompt without -fopenmp is diagnosed.
* c-c++-common/gomp/openmp-ompt-2.c: New test; checks that
-fopenmp-ompt-detailed without -fopenmp is diagnosed.
* c-c++-common/gomp/single2.c: New test for GOMP_single_start_with_end
and GOMP_single_end emission with -fopenmp-ompt.
diff --git gcc/common.opt gcc/common.opt
index 218dddf5dfe..4d1e1d0f928 100644
--- gcc/common.opt
+++ gcc/common.opt
@@ -2503,6 +2503,14 @@ fomit-frame-pointer
Common Var(flag_omit_frame_pointer) Optimization
When possible do not generate stack frames.
+fopenmp-ompt
+Common Var(flag_openmp_ompt) EnabledBy(fopenmp-ompt-detailed)
+Emit additional entry points for OMPT before and after OpenMP worksharing constructs.
+
+fopenmp-ompt-detailed
+Common Var(flag_openmp_ompt_detailed)
+Like -fopenmp-ompt, and also emit OMPT dispatch entry points.
+
fopenmp-target-simd-clone
Common Alias(fopenmp-target-simd-clone=,any,none)
diff --git gcc/doc/invoke.texi gcc/doc/invoke.texi
index 4fbd25e49dd..5cbc22e1e76 100644
--- gcc/doc/invoke.texi
+++ gcc/doc/invoke.texi
@@ -323,7 +323,8 @@ Objective-C and Objective-C++ Dialects}.
@xref{OpenMP and OpenACC Options,,Options Controlling OpenMP and OpenACC}.
@gccoptlist{-foffload=@var{arg} -foffload-options=@var{arg}
-fopenacc -fopenacc-dim=@var{geom}
--fopenmp -fopenmp-simd -fopenmp-target-simd-clone@r{[}=@var{device-type}@r{]}}
+-fopenmp -fopenmp-simd -fopenmp-ompt -fopenmp-ompt-detailed
+-fopenmp-target-simd-clone@r{[}=@var{device-type}@r{]}}
@item Diagnostic Message Formatting Options
@xref{Diagnostic Message Formatting Options,,Options to Control Diagnostic Messages Formatting}.
@@ -5600,6 +5601,20 @@ Fortran. In fixed source form Fortran, the sentinels can also start with
@option{-fopenmp} is additionally specified, the @code{loop} region binds
to the current task region, independent of the specified @code{bind} clause.
+@opindex fopenmp-ompt
+@opindex fno-openmp-ompt
+@cindex OpenMP OMPT
+@item -fopenmp-ompt
+Emit OMPT entry points around OpenMP worksharing constructs, so that an
+OMPT tool can observe their start and end. Requires @option{-fopenmp}.
+
+@opindex fopenmp-ompt-detailed
+@opindex fno-openmp-ompt-detailed
+@item -fopenmp-ompt-detailed
+Like @option{-fopenmp-ompt}, and additionally emit an OMPT dispatch
+entry point that gets called at the beginning of each workshare chunk.
+Requires @option{-fopenmp}.
+
@opindex fopenmp-target-simd-clone
@opindex fno-openmp-target-simd-clone
@cindex OpenMP target SIMD clone
diff --git gcc/omp-builtins.def gcc/omp-builtins.def
index d7440a42761..465aeebefdb 100644
--- gcc/omp-builtins.def
+++ gcc/omp-builtins.def
@@ -445,6 +445,11 @@ DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SECTIONS_END_NOWAIT,
BT_FN_VOID, ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_START, "GOMP_single_start",
BT_FN_BOOL, ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_START_WITH_END,
+ "GOMP_single_start_with_end", BT_FN_BOOL,
+ ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_END, "GOMP_single_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_COPY_START, "GOMP_single_copy_start",
BT_FN_PTR, ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_SINGLE_COPY_END, "GOMP_single_copy_end",
@@ -503,9 +508,29 @@ DEF_GOMP_BUILTIN (BUILT_IN_GOMP_ERROR, "GOMP_error",
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_HAS_MASKED_THREAD_NUM,
"GOMP_has_masked_thread_num", BT_FN_BOOL_INT,
ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_MASKED_END, "GOMP_masked_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING,
"GOMP_loop_static_worksharing", BT_FN_COMPLEX_INT,
ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_START,
+ "GOMP_loop_static_worksharing_start", BT_FN_COMPLEX_INT,
+ ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_END,
+ "GOMP_loop_static_worksharing_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_DISPATCH,
+ "GOMP_loop_static_worksharing_dispatch", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING,
"GOMP_distribute_static_worksharing", BT_FN_COMPLEX_INT,
ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_START,
+ "GOMP_distribute_static_worksharing_start", BT_FN_COMPLEX_INT,
+ ATTR_CONST_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_END,
+ "GOMP_distribute_static_worksharing_end", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
+DEF_GOMP_BUILTIN (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_DISPATCH,
+ "GOMP_distribute_static_worksharing_dispatch", BT_FN_VOID,
+ ATTR_NOTHROW_LEAF_LIST)
diff --git gcc/omp-expand.cc gcc/omp-expand.cc
index 3d799d28032..ad380b27d06 100644
--- gcc/omp-expand.cc
+++ gcc/omp-expand.cc
@@ -5201,10 +5201,14 @@ expand_omp_for_static_nochunk (struct omp_region *region,
switch (gimple_omp_for_kind (fd->for_stmt))
{
case GF_OMP_FOR_KIND_FOR:
- decl = builtin_decl_explicit (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING);
+ decl = builtin_decl_explicit (
+ flag_openmp_ompt ? BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_START
+ : BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING);
break;
case GF_OMP_FOR_KIND_DISTRIBUTE:
- decl = builtin_decl_explicit (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING);
+ decl = builtin_decl_explicit (
+ flag_openmp_ompt ? BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_START
+ : BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING);
break;
default:
gcc_unreachable ();
@@ -5576,8 +5580,53 @@ expand_omp_for_static_nochunk (struct omp_region *region,
cont_bb, body_bb);
}
- /* Replace the GIMPLE_OMP_RETURN with a barrier, or nothing. */
+ if (flag_openmp_ompt_detailed)
+ {
+ /* Insert call to GOMP_*_static_worksharing_dispatch at the end of
+ seq_start_bb. */
+ gsi = gsi_last_nondebug_bb (seq_start_bb);
+ tree decl;
+ switch (gimple_omp_for_kind (fd->for_stmt))
+ {
+ case GF_OMP_FOR_KIND_FOR:
+ decl = builtin_decl_explicit (
+ BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_DISPATCH);
+ break;
+ case GF_OMP_FOR_KIND_DISTRIBUTE:
+ decl = builtin_decl_explicit (
+ BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_DISPATCH);
+ break;
+ default:
+ gcc_unreachable ();
+ }
+ gcall *g = gimple_build_call (decl, 0);
+ gsi_insert_before (&gsi, g, GSI_SAME_STMT);
+ }
+
gsi = gsi_last_nondebug_bb (exit_bb);
+ if (flag_openmp_ompt)
+ {
+ /* Insert call to GOMP_*_static_worksharing_end at the end of exit_bb.
+ */
+ tree decl;
+ switch (gimple_omp_for_kind (fd->for_stmt))
+ {
+ case GF_OMP_FOR_KIND_FOR:
+ decl
+ = builtin_decl_explicit (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_END);
+ break;
+ case GF_OMP_FOR_KIND_DISTRIBUTE:
+ decl = builtin_decl_explicit (
+ BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_END);
+ break;
+ default:
+ gcc_unreachable ();
+ }
+ gcall *g = gimple_build_call (decl, 0);
+ gsi_insert_after (&gsi, g, GSI_SAME_STMT);
+ }
+
+ /* Replace the GIMPLE_OMP_RETURN with a barrier, or nothing. */
if (!gimple_omp_return_nowait_p (gsi_stmt (gsi)))
{
t = gimple_omp_return_lhs (gsi_stmt (gsi));
@@ -5646,6 +5695,7 @@ expand_omp_for_static_nochunk (struct omp_region *region,
exit3_bb = split_block (exit2_bb, g)->dest;
gsi = gsi_after_labels (exit3_bb);
}
+
gsi_remove (&gsi, true);
/* Connect all the blocks. */
@@ -5969,10 +6019,14 @@ expand_omp_for_static_chunk (struct omp_region *region,
switch (gimple_omp_for_kind (fd->for_stmt))
{
case GF_OMP_FOR_KIND_FOR:
- decl = builtin_decl_explicit (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING);
+ decl = builtin_decl_explicit (
+ flag_openmp_ompt ? BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_START
+ : BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING);
break;
case GF_OMP_FOR_KIND_DISTRIBUTE:
- decl = builtin_decl_explicit (BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING);
+ decl = builtin_decl_explicit (
+ flag_openmp_ompt ? BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_START
+ : BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING);
break;
default:
gcc_unreachable ();
@@ -6303,8 +6357,30 @@ expand_omp_for_static_chunk (struct omp_region *region,
gsi_insert_after (&gsi, assign_stmt, GSI_CONTINUE_LINKING);
}
- /* Replace the GIMPLE_OMP_RETURN with a barrier, or nothing. */
gsi = gsi_last_nondebug_bb (exit_bb);
+ if (flag_openmp_ompt)
+ {
+ /* Insert call to GOMP_*_static_worksharing_end at the end of exit_bb.
+ */
+ tree decl;
+ switch (gimple_omp_for_kind (fd->for_stmt))
+ {
+ case GF_OMP_FOR_KIND_FOR:
+ decl
+ = builtin_decl_explicit (BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_END);
+ break;
+ case GF_OMP_FOR_KIND_DISTRIBUTE:
+ decl = builtin_decl_explicit (
+ BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_END);
+ break;
+ default:
+ gcc_unreachable ();
+ }
+ gcall *g = gimple_build_call (decl, 0);
+ gsi_insert_after (&gsi, g, GSI_SAME_STMT);
+ }
+
+ /* Replace the GIMPLE_OMP_RETURN with a barrier, or nothing. */
if (!gimple_omp_return_nowait_p (gsi_stmt (gsi)))
{
t = gimple_omp_return_lhs (gsi_stmt (gsi));
@@ -6340,6 +6416,29 @@ expand_omp_for_static_chunk (struct omp_region *region,
}
gsi_remove (&gsi, true);
+ if (flag_openmp_ompt_detailed)
+ {
+ /* Insert call to GOMP_*_static_worksharing_dispatch at the end of
+ seq_start_bb. */
+ gsi = gsi_last_nondebug_bb (seq_start_bb);
+ tree decl;
+ switch (gimple_omp_for_kind (fd->for_stmt))
+ {
+ case GF_OMP_FOR_KIND_FOR:
+ decl = builtin_decl_explicit (
+ BUILT_IN_GOMP_LOOP_STATIC_WORKSHARING_DISPATCH);
+ break;
+ case GF_OMP_FOR_KIND_DISTRIBUTE:
+ decl = builtin_decl_explicit (
+ BUILT_IN_GOMP_DISTRIBUTE_STATIC_WORKSHARING_DISPATCH);
+ break;
+ default:
+ gcc_unreachable ();
+ }
+ gcall *g = gimple_build_call (decl, 0);
+ gsi_insert_before (&gsi, g, GSI_SAME_STMT);
+ }
+
/* Connect the new blocks. */
find_edge (iter_part_bb, seq_start_bb)->flags = EDGE_TRUE_VALUE;
find_edge (iter_part_bb, fin_bb)->flags = EDGE_FALSE_VALUE;
diff --git gcc/omp-low.cc gcc/omp-low.cc
index dbdbc4079f4..50da8581a8d 100644
--- gcc/omp-low.cc
+++ gcc/omp-low.cc
@@ -8786,7 +8786,9 @@ lower_omp_single_simple (gomp_single *single_stmt, gimple_seq *pre_p)
gimple *call, *cond;
tree lhs, decl;
- decl = builtin_decl_explicit (BUILT_IN_GOMP_SINGLE_START);
+ decl = builtin_decl_explicit (flag_openmp_ompt
+ ? BUILT_IN_GOMP_SINGLE_START_WITH_END
+ : BUILT_IN_GOMP_SINGLE_START);
lhs = create_tmp_var (TREE_TYPE (TREE_TYPE (decl)));
call = gimple_build_call (decl, 0);
gimple_call_set_lhs (call, lhs);
@@ -8923,6 +8925,15 @@ lower_omp_single (gimple_stmt_iterator *gsi_p, omp_context *ctx)
gimple *g = gimple_build_omp_return (nowait);
gimple_seq_add_stmt (&bind_body_tail, g);
maybe_add_implicit_barrier_cancel (ctx, g, &bind_body_tail);
+
+ if (flag_openmp_ompt && !ctx->record_type)
+ {
+ /* Insert call to GOMP_single_end. */
+ tree decl = builtin_decl_explicit (BUILT_IN_GOMP_SINGLE_END);
+ gimple *stmt = gimple_build_call (decl, 0);
+ gimple_seq_add_stmt (&bind_body_tail, stmt);
+ }
+
if (ctx->record_type)
{
gimple_stmt_iterator gsi = gsi_start (bind_body_tail);
@@ -9099,6 +9110,14 @@ lower_omp_master (gimple_stmt_iterator *gsi_p, omp_context *ctx)
gimple_bind_add_seq (bind, gimple_omp_body (stmt));
gimple_omp_set_body (stmt, NULL);
+ if (flag_openmp_ompt)
+ {
+ /* Insert call to GOMP_masked_end at the end of the body. */
+ tree decl = builtin_decl_explicit (BUILT_IN_GOMP_MASKED_END);
+ gcall *g = gimple_build_call (decl, 0);
+ gimple_bind_add_stmt (bind, g);
+ }
+
gimple_bind_add_stmt (bind, gimple_build_label (lab));
gimple_bind_add_stmt (bind, gimple_build_omp_return (true));
diff --git gcc/opts.cc gcc/opts.cc
index 342517528e8..3dae57af0b4 100644
--- gcc/opts.cc
+++ gcc/opts.cc
@@ -1538,6 +1538,12 @@ finish_options (struct gcc_options *opts, struct gcc_options *opts_set,
" %<-fstrict-flex-arrays%> is not present");
}
+ if ((opts->x_flag_openmp_ompt || opts->x_flag_openmp_ompt_detailed)
+ && !opts->x_flag_openmp)
+ error_at (
+ loc,
+ "%<-fopenmp-ompt%> and %<-fopenmp-ompt-detailed%> require %<-fopenmp%>");
+
diagnose_options (opts, opts_set, loc);
}
diff --git gcc/testsuite/c-c++-common/gomp/for-8.c gcc/testsuite/c-c++-common/gomp/for-8.c
deleted file mode 100644
index a894aaaaed1..00000000000
--- gcc/testsuite/c-c++-common/gomp/for-8.c
+++ /dev/null
@@ -1,67 +0,0 @@
-/* { dg-do compile } */
-/* { dg-additional-options "-fdump-tree-ompexp" } */
-
-/* Check that the static loop and distribute expanders fetch the thread/team
- id and count through a single GOMP_loop_static_worksharing /
- GOMP_distribute_static_worksharing call rather than separate
- omp_get_thread_num/omp_get_num_threads or omp_get_team_num/omp_get_num_teams
- calls. */
-
-void bar (int);
-
-/* Static schedule without a chunk size goes through
- expand_omp_for_static_nochunk. */
-
-void
-f1 (int n)
-{
- int i;
- #pragma omp for schedule(static)
- for (i = 0; i < n; ++i)
- bar (i);
-}
-
-/* Static schedule with a chunk size goes through
- expand_omp_for_static_chunk. */
-
-void
-f2 (int n)
-{
- int i;
- #pragma omp for schedule(static, 4)
- for (i = 0; i < n; ++i)
- bar (i);
-}
-
-/* Distribute without a chunk size goes through
- expand_omp_for_static_nochunk. */
-
-void
-f3 (int n)
-{
- int i;
-#pragma omp teams
-#pragma omp distribute
- for (i = 0; i < n; ++i)
- bar (i);
-}
-
-/* Distribute with a chunk size goes through
- expand_omp_for_static_chunk. */
-
-void
-f4 (int n)
-{
- int i;
-#pragma omp teams
-#pragma omp distribute dist_schedule(static, 4)
- for (i = 0; i < n; ++i)
- bar (i);
-}
-
-/* { dg-final { scan-tree-dump "GOMP_loop_static_worksharing" "ompexp" } } */
-/* { dg-final { scan-tree-dump "GOMP_distribute_static_worksharing" "ompexp" } } */
-/* { dg-final { scan-tree-dump-not "omp_get_num_threads" "ompexp" } } */
-/* { dg-final { scan-tree-dump-not "omp_get_thread_num" "ompexp" } } */
-/* { dg-final { scan-tree-dump-not "omp_get_num_teams" "ompexp" } } */
-/* { dg-final { scan-tree-dump-not "omp_get_team_num" "ompexp" } } */
diff --git gcc/testsuite/c-c++-common/gomp/for-static-1.c gcc/testsuite/c-c++-common/gomp/for-static-1.c
new file mode 100644
index 00000000000..50d4e176f3d
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/for-static-1.c
@@ -0,0 +1,17 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-fdump-tree-ompexp" } */
+
+/* Check that the static loop and distribute expanders fetch the thread/team
+ id and count through a single GOMP_loop_static_worksharing /
+ GOMP_distribute_static_worksharing call rather than separate
+ omp_get_thread_num/omp_get_num_threads or omp_get_team_num/omp_get_num_teams
+ calls. */
+
+#include "for-static.h"
+
+/* { dg-final { scan-tree-dump-times "GOMP_loop_static_worksharing \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_distribute_static_worksharing \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "omp_get_num_threads" "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "omp_get_thread_num" "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "omp_get_num_teams" "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "omp_get_team_num" "ompexp" } } */
diff --git gcc/testsuite/c-c++-common/gomp/for-static-2.c gcc/testsuite/c-c++-common/gomp/for-static-2.c
new file mode 100644
index 00000000000..c3d25b97bed
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/for-static-2.c
@@ -0,0 +1,15 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-fopenmp-ompt -fdump-tree-ompexp" } */
+
+/* Check that, with -fopenmp-ompt, the _start and _end variants are called. */
+
+#include "for-static.h"
+
+/* { dg-final { scan-tree-dump-times "GOMP_loop_static_worksharing_start \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_loop_static_worksharing_end \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_distribute_static_worksharing_start \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_distribute_static_worksharing_end \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "GOMP_loop_static_worksharing \\(" "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "GOMP_loop_static_worksharing_dispatch" "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "GOMP_distribute_static_worksharing \\(" "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "GOMP_distribute_static_worksharing_dispatch" "ompexp" } } */
diff --git gcc/testsuite/c-c++-common/gomp/for-static-3.c gcc/testsuite/c-c++-common/gomp/for-static-3.c
new file mode 100644
index 00000000000..6fd1da6c407
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/for-static-3.c
@@ -0,0 +1,14 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-fopenmp-ompt-detailed -fdump-tree-ompexp" } */
+
+/* Check that, with -fopenmp-ompt-detailed, _dispatch builtins are called along
+ with the _start and _end variants. */
+
+#include "for-static.h"
+
+/* { dg-final { scan-tree-dump-times "GOMP_loop_static_worksharing_start \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_loop_static_worksharing_dispatch \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_loop_static_worksharing_end \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_distribute_static_worksharing_start \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_distribute_static_worksharing_dispatch \\(" 2 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_distribute_static_worksharing_end \\(" 2 "ompexp" } } */
diff --git gcc/testsuite/c-c++-common/gomp/for-static.h gcc/testsuite/c-c++-common/gomp/for-static.h
new file mode 100644
index 00000000000..6504c75a13e
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/for-static.h
@@ -0,0 +1,51 @@
+void bar (int);
+
+/* Static schedule without a chunk size goes through
+ expand_omp_for_static_nochunk. */
+
+void
+f1 (int n)
+{
+ int i;
+ #pragma omp for schedule(static)
+ for (i = 0; i < n; ++i)
+ bar (i);
+}
+
+/* Static schedule with a chunk size goes through
+ expand_omp_for_static_chunk. */
+
+void
+f2 (int n)
+{
+ int i;
+ #pragma omp for schedule(static, 4)
+ for (i = 0; i < n; ++i)
+ bar (i);
+}
+
+/* Distribute without a chunk size goes through
+ expand_omp_for_static_nochunk. */
+
+void
+f3 (int n)
+{
+ int i;
+#pragma omp teams
+#pragma omp distribute
+ for (i = 0; i < n; ++i)
+ bar (i);
+}
+
+/* Distribute with a chunk size goes through
+ expand_omp_for_static_chunk. */
+
+void
+f4 (int n)
+{
+ int i;
+#pragma omp teams
+#pragma omp distribute dist_schedule(static, 4)
+ for (i = 0; i < n; ++i)
+ bar (i);
+}
diff --git gcc/testsuite/c-c++-common/gomp/masked-1.c gcc/testsuite/c-c++-common/gomp/masked-1.c
index 8de87b551e3..28d2735d96e 100644
--- gcc/testsuite/c-c++-common/gomp/masked-1.c
+++ gcc/testsuite/c-c++-common/gomp/masked-1.c
@@ -26,4 +26,5 @@ foo (int x, int *a)
}
/* { dg-final { scan-tree-dump "GOMP_has_masked_thread_num" "omplower" } } */
+/* { dg-final { scan-tree-dump-not "GOMP_masked_end" "omplower" } } */
/* { dg-final { scan-tree-dump-not "omp_get_thread_num" "omplower" } } */
diff --git gcc/testsuite/c-c++-common/gomp/masked-3.c gcc/testsuite/c-c++-common/gomp/masked-3.c
new file mode 100644
index 00000000000..adea59fdb69
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/masked-3.c
@@ -0,0 +1,15 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-fopenmp-ompt -fdump-tree-omplower" } */
+
+void bar (void);
+
+void
+foo (void)
+{
+ #pragma omp masked
+ bar ();
+}
+
+/* { dg-final { scan-tree-dump-times "GOMP_has_masked_thread_num" 1 "omplower" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_masked_end" 1 "omplower" } } */
+/* { dg-final { scan-tree-dump-not "omp_get_thread_num" "omplower" } } */
diff --git gcc/testsuite/c-c++-common/gomp/openmp-ompt-1.c gcc/testsuite/c-c++-common/gomp/openmp-ompt-1.c
new file mode 100644
index 00000000000..0a45c2bd52c
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/openmp-ompt-1.c
@@ -0,0 +1,4 @@
+/* { dg-do compile } */
+/* { dg-options "-fopenmp-ompt" } */
+
+/* { dg-error ".-fopenmp-ompt. and .-fopenmp-ompt-detailed. require .-fopenmp." "" { target *-*-* } 0 } */
diff --git gcc/testsuite/c-c++-common/gomp/openmp-ompt-2.c gcc/testsuite/c-c++-common/gomp/openmp-ompt-2.c
new file mode 100644
index 00000000000..f93f6920327
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/openmp-ompt-2.c
@@ -0,0 +1,4 @@
+/* { dg-do compile } */
+/* { dg-options "-fopenmp-ompt-detailed" } */
+
+/* { dg-error ".-fopenmp-ompt. and .-fopenmp-ompt-detailed. require .-fopenmp." "" { target *-*-* } 0 } */
diff --git gcc/testsuite/c-c++-common/gomp/single2.c gcc/testsuite/c-c++-common/gomp/single2.c
new file mode 100644
index 00000000000..215f0368ce4
--- /dev/null
+++ gcc/testsuite/c-c++-common/gomp/single2.c
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-fopenmp-ompt -fdump-tree-ompexp" } */
+
+void
+foo (void)
+{
+ #pragma omp single
+ ;
+}
+
+/* { dg-final { scan-tree-dump-times "GOMP_single_start_with_end" 1 "ompexp" } } */
+/* { dg-final { scan-tree-dump-times "GOMP_single_end" 1 "ompexp" } } */
+/* { dg-final { scan-tree-dump-not "GOMP_single_start \\(" "ompexp" } } */
diff --git libgomp/config/gcn/teams.c libgomp/config/gcn/teams.c
index 5aa8b15502a..99f807ac876 100644
--- libgomp/config/gcn/teams.c
+++ libgomp/config/gcn/teams.c
@@ -50,6 +50,29 @@ GOMP_distribute_static_worksharing (void)
return nteams + tid * 1I;
}
+/* OMPT variant enabled by -fopenmp-ompt. */
+
+_Complex int
+GOMP_distribute_static_worksharing_start (void)
+{
+ int __lds *gomp_team_num = (int __lds *) GOMP_TEAM_NUM;
+ unsigned tid = *gomp_team_num;
+ unsigned nteams = gomp_num_teams_var + 1;
+ return nteams + tid * 1I;
+}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt-detailed. */
+
+void
+GOMP_distribute_static_worksharing_dispatch (void)
+{}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_distribute_static_worksharing_end (void)
+{}
+
int
omp_get_num_teams (void)
{
diff --git libgomp/config/nvptx/teams.c libgomp/config/nvptx/teams.c
index 9763f6f4108..3e6ea287276 100644
--- libgomp/config/nvptx/teams.c
+++ libgomp/config/nvptx/teams.c
@@ -52,6 +52,27 @@ GOMP_distribute_static_worksharing (void)
return nteams + tid * 1I;
}
+/* OMPT variant enabled by -fopenmp-ompt. */
+
+_Complex int
+GOMP_distribute_static_worksharing_start (void)
+{
+ unsigned tid = __gomp_team_num;
+ unsigned nteams = gomp_num_teams_var + 1;
+ return nteams + tid * 1I;
+}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt-detailed. */
+
+void
+GOMP_distribute_static_worksharing_dispatch (void)
+{}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_distribute_static_worksharing_end (void)
+{}
int
omp_get_num_teams (void)
{
diff --git libgomp/libgomp.map libgomp/libgomp.map
index e2985347bf7..3f465cc1491 100644
--- libgomp/libgomp.map
+++ libgomp/libgomp.map
@@ -485,12 +485,21 @@ GOMP_6.0.1 {
GOMP_6.0.2 {
global:
GOMP_has_masked_thread_num;
+ GOMP_masked_end;
GOMP_loop_static_worksharing;
+ GOMP_loop_static_worksharing_start;
+ GOMP_loop_static_worksharing_dispatch;
+ GOMP_loop_static_worksharing_end;
GOMP_distribute_static_worksharing;
+ GOMP_distribute_static_worksharing_start;
+ GOMP_distribute_static_worksharing_dispatch;
+ GOMP_distribute_static_worksharing_end;
GOMP_barrier_ext;
GOMP_barrier_cancel_ext;
GOMP_reduction_start;
GOMP_reduction_end;
+ GOMP_single_start_with_end;
+ GOMP_single_end;
} GOMP_6.0.1;
OACC_2.0 {
diff --git libgomp/libgomp_g.h libgomp/libgomp_g.h
index a3097d5b093..4a3565684cc 100644
--- libgomp/libgomp_g.h
+++ libgomp/libgomp_g.h
@@ -144,6 +144,11 @@ extern void GOMP_loop_end (void);
extern void GOMP_loop_end_nowait (void);
extern bool GOMP_loop_end_cancel (void);
+extern __complex__ int GOMP_loop_static_worksharing (void);
+extern __complex__ int GOMP_loop_static_worksharing_start (void);
+extern void GOMP_loop_static_worksharing_dispatch (void);
+extern void GOMP_loop_static_worksharing_end (void);
+
/* loop_ull.c */
extern bool GOMP_loop_ull_static_start (bool, unsigned long long,
@@ -298,7 +303,7 @@ extern unsigned GOMP_parallel_reductions (void (*) (void *), void *, unsigned,
extern bool GOMP_cancel (int, bool);
extern bool GOMP_cancellation_point (int);
extern bool GOMP_has_masked_thread_num (int);
-extern __complex__ int GOMP_loop_static_worksharing (void);
+extern void GOMP_masked_end (void);
/* task.c */
@@ -339,6 +344,8 @@ extern bool GOMP_sections_end_cancel (void);
/* single.c */
extern bool GOMP_single_start (void);
+extern bool GOMP_single_start_with_end (void);
+extern void GOMP_single_end (void);
extern void *GOMP_single_copy_start (void);
extern void GOMP_single_copy_end (void *);
@@ -380,6 +387,9 @@ extern void *GOMP_target_map_indirect_ptr (void *);
extern void GOMP_teams_reg (void (*) (void *), void *, unsigned, unsigned,
unsigned);
extern __complex__ int GOMP_distribute_static_worksharing (void);
+extern __complex__ int GOMP_distribute_static_worksharing_start (void);
+extern void GOMP_distribute_static_worksharing_dispatch (void);
+extern void GOMP_distribute_static_worksharing_end (void);
/* allocator.c */
diff --git libgomp/loop.c libgomp/loop.c
index e3fbeb1c4de..f286ce6b6db 100644
--- libgomp/loop.c
+++ libgomp/loop.c
@@ -1186,3 +1186,38 @@ GOMP_loop_ordered_guided_next (long *istart, long *iend)
return gomp_loop_ordered_guided_next (istart, iend);
}
#endif
+
+/* For a worksharing-loop construct with static schedule, return the thread ID
+ and number of threads packed into a single complex value. */
+
+_Complex int
+GOMP_loop_static_worksharing (void)
+{
+ struct gomp_team *team = gomp_thread ()->ts.team;
+ unsigned tid = gomp_thread ()->ts.team_id;
+ unsigned nthreads = team ? team->nthreads : 1;
+ return nthreads + tid * 1I;
+}
+
+/* OMPT variant enabled by -fopenmp-ompt. */
+
+_Complex int
+GOMP_loop_static_worksharing_start (void)
+{
+ struct gomp_team *team = gomp_thread ()->ts.team;
+ unsigned tid = gomp_thread ()->ts.team_id;
+ unsigned nthreads = team ? team->nthreads : 1;
+ return nthreads + tid * 1I;
+}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt-detailed. */
+
+void
+GOMP_loop_static_worksharing_dispatch (void)
+{}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_loop_static_worksharing_end (void)
+{}
diff --git libgomp/parallel.c libgomp/parallel.c
index 92472ecb4f0..bb522b39ecb 100644
--- libgomp/parallel.c
+++ libgomp/parallel.c
@@ -271,18 +271,6 @@ GOMP_cancel (int which, bool do_cancel)
return true;
}
-/* For a worksharing-loop construct with static schedule, return the thread ID
- and number of threads packed into a single complex value. */
-
-_Complex int
-GOMP_loop_static_worksharing (void)
-{
- struct gomp_team *team = gomp_thread ()->ts.team;
- unsigned tid = gomp_thread ()->ts.team_id;
- unsigned nthreads = team ? team->nthreads : 1;
- return nthreads + tid * 1I;
-}
-
/* Return true if the current thread number equals TID.
Used to implement the masked construct's filter clause. */
@@ -291,6 +279,12 @@ GOMP_has_masked_thread_num (int tid)
{
return tid == gomp_thread ()->ts.team_id;
}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_masked_end (void)
+{}
/* The public OpenMP API for thread and team related inquiries. */
diff --git libgomp/single.c libgomp/single.c
index 65126000ea9..00a84e1c9c1 100644
--- libgomp/single.c
+++ libgomp/single.c
@@ -55,6 +55,37 @@ GOMP_single_start (void)
#endif
}
+/* OMPT variant enabled by -fopenmp-ompt. */
+
+bool
+GOMP_single_start_with_end (void)
+{
+#ifdef HAVE_SYNC_BUILTINS
+ struct gomp_thread *thr = gomp_thread ();
+ struct gomp_team *team = thr->ts.team;
+ unsigned long single_count;
+
+ if (__builtin_expect (team == NULL, 0))
+ return true;
+
+ single_count = thr->ts.single_count++;
+ return __sync_bool_compare_and_swap (&team->single_count, single_count,
+ single_count + 1L);
+#else
+ bool ret = gomp_work_share_start (0);
+ if (ret)
+ gomp_work_share_init_done ();
+ gomp_work_share_end_nowait ();
+ return ret;
+#endif
+}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_single_end (void)
+{}
+
/* This routine is called when first encountering a SINGLE construct that
does have a COPYPRIVATE clause. Returns NULL if this is the thread
that should execute the clause; otherwise the return value is pointer
@@ -69,7 +100,7 @@ GOMP_single_copy_start (void)
void *ret;
first = gomp_work_share_start (0);
-
+
if (first)
{
gomp_work_share_init_done ();
diff --git libgomp/teams.c libgomp/teams.c
index 7038623b2df..f501123d647 100644
--- libgomp/teams.c
+++ libgomp/teams.c
@@ -70,6 +70,29 @@ GOMP_distribute_static_worksharing (void)
return nteams + tid * 1I;
}
+/* OMPT variant enabled by -fopenmp-ompt. */
+
+_Complex int
+GOMP_distribute_static_worksharing_start (void)
+{
+ struct gomp_thread *thr = gomp_thread ();
+ unsigned tid = thr->team_num;
+ unsigned nteams = thr->num_teams + 1;
+ return nteams + tid * 1I;
+}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt-detailed. */
+
+void
+GOMP_distribute_static_worksharing_dispatch (void)
+{}
+
+/* Stub for OMPT callback enabled by -fopenmp-ompt. */
+
+void
+GOMP_distribute_static_worksharing_end (void)
+{}
+
int
omp_get_num_teams (void)
{
diff --git libgomp/testsuite/libgomp.c-c++-common/for-static-1.c libgomp/testsuite/libgomp.c-c++-common/for-static-1.c
new file mode 100644
index 00000000000..b304c8c10ec
--- /dev/null
+++ libgomp/testsuite/libgomp.c-c++-common/for-static-1.c
@@ -0,0 +1,13 @@
+/* { dg-do link { target offload_target_any } } */
+/* { dg-additional-options "-O0 -foffload=-fdump-tree-optimized" } */
+
+/* Check that the static distribute expanders fetch the team id and count
+ through a single GOMP_distribute_static_worksharing call rather than separate
+ omp_get_team_num and omp_get_num_teams calls, and that this function exists
+ on the target side. */
+
+#include "for-static.h"
+
+/* { dg-final { scan-offload-tree-dump-times "GOMP_distribute_static_worksharing \\(" 2 "optimized" } } */
+/* { dg-final { scan-offload-tree-dump-not "omp_get_num_teams" "optimized" } } */
+/* { dg-final { scan-offload-tree-dump-not "omp_get_team_num" "optimized" } } */
diff --git libgomp/testsuite/libgomp.c-c++-common/for-static-2.c libgomp/testsuite/libgomp.c-c++-common/for-static-2.c
new file mode 100644
index 00000000000..83e577a3298
--- /dev/null
+++ libgomp/testsuite/libgomp.c-c++-common/for-static-2.c
@@ -0,0 +1,11 @@
+/* { dg-do link { target offload_target_any } } */
+/* { dg-additional-options "-O0 -foffload=-fdump-tree-optimized -fopenmp-ompt" } */
+
+/* Check that, with -fopenmp-ompt, the _start and _end variants are called. */
+
+#include "for-static.h"
+
+/* { dg-final { scan-offload-tree-dump-times "GOMP_distribute_static_worksharing_start \\(" 2 "optimized" } } */
+/* { dg-final { scan-offload-tree-dump-times "GOMP_distribute_static_worksharing_end \\(" 2 "optimized" } } */
+/* { dg-final { scan-offload-tree-dump-not "GOMP_distribute_static_worksharing \\(" "optimized" } } */
+/* { dg-final { scan-offload-tree-dump-not "GOMP_distribute_static_worksharing_dispatch" "optimized" } } */
diff --git libgomp/testsuite/libgomp.c-c++-common/for-static-3.c libgomp/testsuite/libgomp.c-c++-common/for-static-3.c
new file mode 100644
index 00000000000..e7b904eedc3
--- /dev/null
+++ libgomp/testsuite/libgomp.c-c++-common/for-static-3.c
@@ -0,0 +1,11 @@
+/* { dg-do link { target offload_target_any } } */
+/* { dg-additional-options "-O0 -foffload=-fdump-tree-optimized -fopenmp-ompt-detailed" } */
+
+/* Check that, with -fopenmp-ompt-detailed, _dispatch builtins are called along
+ with the _start and _end variants. */
+
+#include "for-static.h"
+
+/* { dg-final { scan-offload-tree-dump-times "GOMP_distribute_static_worksharing_start \\(" 2 "optimized" } } */
+/* { dg-final { scan-offload-tree-dump-times "GOMP_distribute_static_worksharing_dispatch \\(" 2 "optimized" } } */
+/* { dg-final { scan-offload-tree-dump-times "GOMP_distribute_static_worksharing_end \\(" 2 "optimized" } } */
diff --git libgomp/testsuite/libgomp.c-c++-common/for-static.h libgomp/testsuite/libgomp.c-c++-common/for-static.h
new file mode 100644
index 00000000000..a6c0e055b83
--- /dev/null
+++ libgomp/testsuite/libgomp.c-c++-common/for-static.h
@@ -0,0 +1,37 @@
+void bar (int a)
+{}
+
+/* Distribute without a chunk size goes through
+ expand_omp_for_static_nochunk. */
+
+void
+f3 (int n)
+{
+ int i;
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+ for (i = 0; i < n; ++i)
+ bar (i);
+}
+
+/* Distribute with a chunk size goes through
+ expand_omp_for_static_chunk. */
+
+void
+f4 (int n)
+{
+ int i;
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute dist_schedule(static, 4)
+ for (i = 0; i < n; ++i)
+ bar (i);
+}
+
+int main (void)
+{
+ f3(0);
+ f4(1);
+ return 0;
+}
