[PATCH] D56430: Incorrect implicit data-sharing for nested tasks
smateo added a comment. I don't have commit access yet. Would you mind to commit it for me? Thanks!, Sergi Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D56430/new/ https://reviews.llvm.org/D56430 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D56430: Incorrect implicit data-sharing for nested tasks
smateo updated this revision to Diff 180844.
smateo added a comment.
Done! Thanks,
Sergi
Repository:
rC Clang
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D56430/new/
https://reviews.llvm.org/D56430
Files:
lib/Sema/SemaOpenMP.cpp
test/OpenMP/task_messages.cpp
Index: test/OpenMP/task_messages.cpp
===
--- test/OpenMP/task_messages.cpp
+++ test/OpenMP/task_messages.cpp
@@ -8,7 +8,7 @@
#pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp task'}}
class S {
- S(const S &s) { a = s.a + 12; } // expected-note 14 {{implicitly declared private here}}
+ S(const S &s) { a = s.a + 12; } // expected-note 16 {{implicitly declared private here}}
int a;
public:
@@ -51,6 +51,15 @@
++a; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task
+#pragma omp task
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task default(shared)
+#pragma omp task
++a;
#pragma omp task
#pragma omp parallel
@@ -205,6 +214,15 @@
++sa; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task
+#pragma omp task
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task default(shared)
+#pragma omp task
++sa;
#pragma omp task
#pragma omp parallel
Index: lib/Sema/SemaOpenMP.cpp
===
--- lib/Sema/SemaOpenMP.cpp
+++ lib/Sema/SemaOpenMP.cpp
@@ -676,9 +676,13 @@
}
};
-bool isParallelOrTaskRegion(OpenMPDirectiveKind DKind) {
- return isOpenMPParallelDirective(DKind) || isOpenMPTaskingDirective(DKind) ||
- isOpenMPTeamsDirective(DKind) || DKind == OMPD_unknown;
+
+bool isImplicitTaskingRegion(OpenMPDirectiveKind DKind) {
+ return isOpenMPParallelDirective(DKind) || isOpenMPTeamsDirective(DKind);
+}
+
+bool isImplicitOrExplicitTaskingRegion(OpenMPDirectiveKind DKind) {
+ return isImplicitTaskingRegion(DKind) || isOpenMPTaskingDirective(DKind) || DKind == OMPD_unknown;
}
} // namespace
@@ -819,7 +823,7 @@
DVar.CKind = OMPC_firstprivate;
return DVar;
}
- } while (I != E && !isParallelOrTaskRegion(I->Directive));
+ } while (I != E && !isImplicitTaskingRegion(I->Directive));
DVar.CKind =
(DVarTemp.CKind == OMPC_unknown) ? OMPC_firstprivate : OMPC_shared;
return DVar;
@@ -1066,7 +1070,7 @@
if (!isStackEmpty()) {
iterator I = Iter, E = Stack.back().first.rend();
Scope *TopScope = nullptr;
-while (I != E && !isParallelOrTaskRegion(I->Directive) &&
+while (I != E && !isImplicitOrExplicitTaskingRegion(I->Directive) &&
!isOpenMPTargetExecutionDirective(I->Directive))
++I;
if (I == E)
@@ -1292,7 +1296,7 @@
if (FromParent && I != EndI)
std::advance(I, 1);
for (; I != EndI; std::advance(I, 1)) {
-if (!DPred(I->Directive) && !isParallelOrTaskRegion(I->Directive))
+if (!DPred(I->Directive) && !isImplicitOrExplicitTaskingRegion(I->Directive))
continue;
iterator NewI = I;
DSAVarData DVar = getDSA(NewI, D);
@@ -1636,7 +1640,7 @@
auto &&Info = DSAStack->isLoopControlVariable(D);
if (Info.first ||
(VD && VD->hasLocalStorage() &&
- isParallelOrTaskRegion(DSAStack->getCurrentDirective())) ||
+ isImplicitOrExplicitTaskingRegion(DSAStack->getCurrentDirective())) ||
(VD && DSAStack->isForceVarCapturing()))
return VD ? VD : Info.second;
DSAStackTy::DSAVarData DVarPrivate =
@@ -2244,7 +2248,7 @@
// attribute, must have its data-sharing attribute explicitly determined
// by being listed in a data-sharing attribute clause.
if (DVar.CKind == OMPC_unknown && Stack->getDefaultDSA() == DSA_none &&
- isParallelOrTaskRegion(DKind) &&
+ isImplicitOrExplicitTaskingRegion(DKind) &&
VarsWithInheritedDSA.count(VD) == 0) {
VarsWithInheritedDSA[VD] = E;
return;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D56430: Incorrect implicit data-sharing for nested tasks
smateo updated this revision to Diff 180792.
smateo added a comment.
Renaming the `isParallelRegion` function to `isImplicitTaskingRegion`.
Repository:
rC Clang
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D56430/new/
https://reviews.llvm.org/D56430
Files:
lib/Sema/SemaOpenMP.cpp
test/OpenMP/task_messages.cpp
Index: test/OpenMP/task_messages.cpp
===
--- test/OpenMP/task_messages.cpp
+++ test/OpenMP/task_messages.cpp
@@ -8,7 +8,7 @@
#pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp
task'}}
class S {
- S(const S &s) { a = s.a + 12; } // expected-note 14 {{implicitly declared
private here}}
+ S(const S &s) { a = s.a + 12; } // expected-note 16 {{implicitly declared
private here}}
int a;
public:
@@ -51,6 +51,15 @@
++a; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task
+#pragma omp task
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task default(shared)
+#pragma omp task
++a;
#pragma omp task
#pragma omp parallel
@@ -205,6 +214,15 @@
++sa; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task
+#pragma omp task
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task default(shared)
+#pragma omp task
++sa;
#pragma omp task
#pragma omp parallel
Index: lib/Sema/SemaOpenMP.cpp
===
--- lib/Sema/SemaOpenMP.cpp
+++ lib/Sema/SemaOpenMP.cpp
@@ -676,9 +676,13 @@
}
};
+
+bool isImplicitTaskingRegion(OpenMPDirectiveKind DKind) {
+ return isOpenMPParallelDirective(DKind) || isOpenMPTeamsDirective(DKind);
+}
+
bool isParallelOrTaskRegion(OpenMPDirectiveKind DKind) {
- return isOpenMPParallelDirective(DKind) || isOpenMPTaskingDirective(DKind) ||
- isOpenMPTeamsDirective(DKind) || DKind == OMPD_unknown;
+ return isImplicitTaskingRegion(DKind) || isOpenMPTaskingDirective(DKind) ||
DKind == OMPD_unknown;
}
} // namespace
@@ -819,7 +823,7 @@
DVar.CKind = OMPC_firstprivate;
return DVar;
}
- } while (I != E && !isParallelOrTaskRegion(I->Directive));
+ } while (I != E && !isImplicitTaskingRegion(I->Directive));
DVar.CKind =
(DVarTemp.CKind == OMPC_unknown) ? OMPC_firstprivate : OMPC_shared;
return DVar;
Index: test/OpenMP/task_messages.cpp
===
--- test/OpenMP/task_messages.cpp
+++ test/OpenMP/task_messages.cpp
@@ -8,7 +8,7 @@
#pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp task'}}
class S {
- S(const S &s) { a = s.a + 12; } // expected-note 14 {{implicitly declared private here}}
+ S(const S &s) { a = s.a + 12; } // expected-note 16 {{implicitly declared private here}}
int a;
public:
@@ -51,6 +51,15 @@
++a; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task
+#pragma omp task
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task default(shared)
+#pragma omp task
++a;
#pragma omp task
#pragma omp parallel
@@ -205,6 +214,15 @@
++sa; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task
+#pragma omp task
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task default(shared)
+#pragma omp task
++sa;
#pragma omp task
#pragma omp parallel
Index: lib/Sema/SemaOpenMP.cpp
===
--- lib/Sema/SemaOpenMP.cpp
+++ lib/Sema/SemaOpenMP.cpp
@@ -676,9 +676,13 @@
}
};
+
+bool isImplicitTaskingRegion(OpenMPDirectiveKind DKind) {
+ return isOpenMPParallelDirective(DKind) || isOpenMPTeamsDirective(DKind);
+}
+
bool isParallelOrTaskRegion(OpenMPDirectiveKind DKind) {
- return isOpenMPParallelDirective(DKind) || isOpenMPTaskingDirective(DKind) ||
- isOpenMPTeamsDirective(DKind) || DKind == OMPD_unknown;
+ return isImplicitTaskingRegion(DKind) || isOpenMPTaskingDirective(DKind) || DKind == OMPD_unknown;
}
} // namespace
@@ -819,7 +823,7 @@
DVar.CKind = OMPC_firstprivate;
return DVar;
}
- } while (I != E && !isParallelOrTaskRegion(I->Directive));
+ } while (I != E && !isImplicitTaskingRegion(I->Directive));
DVar.CKind =
(DVa
[PATCH] D56430: Incorrect implicit data-sharing for nested tasks
smateo updated this revision to Diff 180631.
smateo added a comment.
Adding more context
Repository:
rC Clang
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D56430/new/
https://reviews.llvm.org/D56430
Files:
lib/Sema/SemaOpenMP.cpp
test/OpenMP/task_messages.cpp
Index: test/OpenMP/task_messages.cpp
===
--- test/OpenMP/task_messages.cpp
+++ test/OpenMP/task_messages.cpp
@@ -8,7 +8,7 @@
#pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp
task'}}
class S {
- S(const S &s) { a = s.a + 12; } // expected-note 14 {{implicitly declared
private here}}
+ S(const S &s) { a = s.a + 12; } // expected-note 16 {{implicitly declared
private here}}
int a;
public:
@@ -51,6 +51,15 @@
++a; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task
+#pragma omp task
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task default(shared)
+#pragma omp task
++a;
#pragma omp task
#pragma omp parallel
@@ -205,6 +214,15 @@
++sa; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task
+#pragma omp task
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task default(shared)
+#pragma omp task
++sa;
#pragma omp task
#pragma omp parallel
Index: lib/Sema/SemaOpenMP.cpp
===
--- lib/Sema/SemaOpenMP.cpp
+++ lib/Sema/SemaOpenMP.cpp
@@ -676,9 +676,13 @@
}
};
+
+bool isParallelRegion(OpenMPDirectiveKind DKind) {
+ return isOpenMPParallelDirective(DKind) || isOpenMPTeamsDirective(DKind);
+}
+
bool isParallelOrTaskRegion(OpenMPDirectiveKind DKind) {
- return isOpenMPParallelDirective(DKind) || isOpenMPTaskingDirective(DKind) ||
- isOpenMPTeamsDirective(DKind) || DKind == OMPD_unknown;
+ return isParallelRegion(DKind) || isOpenMPTaskingDirective(DKind) || DKind
== OMPD_unknown;
}
} // namespace
@@ -819,7 +823,7 @@
DVar.CKind = OMPC_firstprivate;
return DVar;
}
- } while (I != E && !isParallelOrTaskRegion(I->Directive));
+ } while (I != E && !isParallelRegion(I->Directive));
DVar.CKind =
(DVarTemp.CKind == OMPC_unknown) ? OMPC_firstprivate : OMPC_shared;
return DVar;
Index: test/OpenMP/task_messages.cpp
===
--- test/OpenMP/task_messages.cpp
+++ test/OpenMP/task_messages.cpp
@@ -8,7 +8,7 @@
#pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp task'}}
class S {
- S(const S &s) { a = s.a + 12; } // expected-note 14 {{implicitly declared private here}}
+ S(const S &s) { a = s.a + 12; } // expected-note 16 {{implicitly declared private here}}
int a;
public:
@@ -51,6 +51,15 @@
++a; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task
+#pragma omp task
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task default(shared)
+#pragma omp task
++a;
#pragma omp task
#pragma omp parallel
@@ -205,6 +214,15 @@
++sa; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task
+#pragma omp task
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task default(shared)
+#pragma omp task
++sa;
#pragma omp task
#pragma omp parallel
Index: lib/Sema/SemaOpenMP.cpp
===
--- lib/Sema/SemaOpenMP.cpp
+++ lib/Sema/SemaOpenMP.cpp
@@ -676,9 +676,13 @@
}
};
+
+bool isParallelRegion(OpenMPDirectiveKind DKind) {
+ return isOpenMPParallelDirective(DKind) || isOpenMPTeamsDirective(DKind);
+}
+
bool isParallelOrTaskRegion(OpenMPDirectiveKind DKind) {
- return isOpenMPParallelDirective(DKind) || isOpenMPTaskingDirective(DKind) ||
- isOpenMPTeamsDirective(DKind) || DKind == OMPD_unknown;
+ return isParallelRegion(DKind) || isOpenMPTaskingDirective(DKind) || DKind == OMPD_unknown;
}
} // namespace
@@ -819,7 +823,7 @@
DVar.CKind = OMPC_firstprivate;
return DVar;
}
- } while (I != E && !isParallelOrTaskRegion(I->Directive));
+ } while (I != E && !isParallelRegion(I->Directive));
DVar.CKind =
(DVarTemp.CKind == OMPC_unknown) ? OMPC_firstprivate : OMPC_shared;
return DVar;
_
[PATCH] D56430: Incorrect implicit data-sharing for nested tasks
smateo created this revision.
smateo added a reviewer: ABataev.
smateo added a project: OpenMP.
Herald added a subscriber: cfe-commits.
There is a minor issue in how the implicit data-sharings for nested tasks are
computed.
For the following example:
int x;
#pragma omp task shared(x)
#pragma omp task
x++;
We compute an implicit data-sharing of shared for `x` in the second task
although I think that it should be firstprivate. Below you can find the part of
the OpenMP spec that covers this example:
- // In a task generating construct, if no default clause is present, a
variable for which the data-sharing attribute is not determined by the rules
above and that in the enclosing context is determined to be shared by all
implicit tasks bound to the current team is shared.//
- //In a task generating construct, if no default clause is present, a variable
for which the data-sharing attribute is not determined by the rules above is
firstprivate.//
Since each implicit-task has its own copy of `x`, we shouldn't apply the first
rule.
Repository:
rC Clang
https://reviews.llvm.org/D56430
Files:
lib/Sema/SemaOpenMP.cpp
test/OpenMP/task_messages.cpp
Index: test/OpenMP/task_messages.cpp
===
--- test/OpenMP/task_messages.cpp
+++ test/OpenMP/task_messages.cpp
@@ -8,7 +8,7 @@
#pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp
task'}}
class S {
- S(const S &s) { a = s.a + 12; } // expected-note 14 {{implicitly declared
private here}}
+ S(const S &s) { a = s.a + 12; } // expected-note 16 {{implicitly declared
private here}}
int a;
public:
@@ -51,6 +51,15 @@
++a; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task
+#pragma omp task
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task default(shared)
+#pragma omp task
++a;
#pragma omp task
#pragma omp parallel
@@ -205,6 +214,15 @@
++sa; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task
+#pragma omp task
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task default(shared)
+#pragma omp task
++sa;
#pragma omp task
#pragma omp parallel
Index: lib/Sema/SemaOpenMP.cpp
===
--- lib/Sema/SemaOpenMP.cpp
+++ lib/Sema/SemaOpenMP.cpp
@@ -676,9 +676,13 @@
}
};
+
+bool isParallelRegion(OpenMPDirectiveKind DKind) {
+ return isOpenMPParallelDirective(DKind) || isOpenMPTeamsDirective(DKind);
+}
+
bool isParallelOrTaskRegion(OpenMPDirectiveKind DKind) {
- return isOpenMPParallelDirective(DKind) || isOpenMPTaskingDirective(DKind) ||
- isOpenMPTeamsDirective(DKind) || DKind == OMPD_unknown;
+ return isParallelRegion(DKind) || isOpenMPTaskingDirective(DKind) || DKind
== OMPD_unknown;
}
} // namespace
@@ -819,7 +823,7 @@
DVar.CKind = OMPC_firstprivate;
return DVar;
}
- } while (I != E && !isParallelOrTaskRegion(I->Directive));
+ } while (I != E && !isParallelRegion(I->Directive));
DVar.CKind =
(DVarTemp.CKind == OMPC_unknown) ? OMPC_firstprivate : OMPC_shared;
return DVar;
Index: test/OpenMP/task_messages.cpp
===
--- test/OpenMP/task_messages.cpp
+++ test/OpenMP/task_messages.cpp
@@ -8,7 +8,7 @@
#pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp task'}}
class S {
- S(const S &s) { a = s.a + 12; } // expected-note 14 {{implicitly declared private here}}
+ S(const S &s) { a = s.a + 12; } // expected-note 16 {{implicitly declared private here}}
int a;
public:
@@ -51,6 +51,15 @@
++a; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task
+#pragma omp task
+ ++a;
+#pragma omp parallel shared(a)
+#pragma omp task default(shared)
+#pragma omp task
++a;
#pragma omp task
#pragma omp parallel
@@ -205,6 +214,15 @@
++sa; // expected-error {{calling a private constructor of class 'S'}}
#pragma omp task default(shared)
#pragma omp task
+ // expected-error@+1 {{calling a private constructor of class 'S'}}
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task
+#pragma omp task
+ ++sa;
+#pragma omp parallel shared(sa)
+#pragma omp task default(shared)
+#pragma omp task
++sa;
#pragma omp task
#pragma omp parallel
Index: lib/Sema/SemaOpenMP.cpp
=
[PATCH] D53636: Do not always request an implicit taskgroup region inside the kmpc_taskloop function
smateo added a comment. Hi Alexey, Thanks for the prompt review! I don't have commit access yet, do you mind to commit it for me? Thanks! Repository: rC Clang https://reviews.llvm.org/D53636 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D53636: Do not always request an implicit taskgroup region inside the kmpc_taskloop function
smateo created this revision.
smateo added a reviewer: ABataev.
Herald added a subscriber: cfe-commits.
For the following code:
int i;
#pragma omp taskloop
for (i = 0; i < 100; ++i)
{}
#pragma omp taskloop nogroup
for (i = 0; i < 100; ++i)
{}
Clang emits the following LLVM IR:
...
call void @__kmpc_taskgroup(%struct.ident_t* @0, i32 %0)
%2 = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @0, i32 %0, i32 1, i64
80, i64 8, i32 (i32, i8*)* bitcast (i32 (i32,
%struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
...
call void @__kmpc_taskloop(%struct.ident_t* @0, i32 %0, i8* %2, i32 1, i64*
%8, i64* %9, i64 %13, i32 0, i32 0, i64 0, i8* null)
call void @__kmpc_end_taskgroup(%struct.ident_t* @0, i32 %0)
...
%15 = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @0, i32 %0, i32 1,
i64 80, i64 8, i32 (i32, i8*)* bitcast (i32 (i32,
%struct.kmp_task_t_with_privates.1*)* @.omp_task_entry..2 to i32 (i32, i8*)*))
...
call void @__kmpc_taskloop(%struct.ident_t* @0, i32 %0, i8* %15, i32 1, i64*
%21, i64* %22, i64 %26, i32 0, i32 0, i64 0, i8* null)
The first set of instructions corresponds to the first taskloop construct. It
is important to note that the implicit taskgroup region associated with the
taskloop construct has been materialized in our IR: the `__kmpc_taskloop`
occurs inside a taskgroup region. Note also that this taskgroup region does not
exist in our second taskloop because we are using the `nogroup` clause.
The issue here is the 4th argument of the kmpc_taskloop call, starting from the
end, is always a zero. Checking the LLVM OpenMP RT implementation, we see that
this argument corresponds to the nogroup parameter:
void __kmpc_taskloop(ident_t *loc, int gtid, kmp_task_t *task, int if_val,
kmp_uint64 *lb, kmp_uint64 *ub, kmp_int64 st, int
nogroup,
int sched, kmp_uint64 grainsize, void *task_dup);
So basically we always tell to the RT to do another taskgroup region. For the
first taskloop, this means that we create two taskgroup regions. For the second
example, it means that despite the fact we had a nogroup clause we are going to
have a taskgroup region, so we unnecessary wait until all descendant tasks have
been executed.
Repository:
rC Clang
https://reviews.llvm.org/D53636
Files:
lib/CodeGen/CGOpenMPRuntime.cpp
test/OpenMP/taskloop_codegen.cpp
test/OpenMP/taskloop_firstprivate_codegen.cpp
test/OpenMP/taskloop_lastprivate_codegen.cpp
test/OpenMP/taskloop_private_codegen.cpp
test/OpenMP/taskloop_reduction_codegen.cpp
test/OpenMP/taskloop_simd_codegen.cpp
test/OpenMP/taskloop_simd_firstprivate_codegen.cpp
test/OpenMP/taskloop_simd_lastprivate_codegen.cpp
test/OpenMP/taskloop_simd_private_codegen.cpp
test/OpenMP/taskloop_simd_reduction_codegen.cpp
Index: test/OpenMP/taskloop_simd_reduction_codegen.cpp
===
--- test/OpenMP/taskloop_simd_reduction_codegen.cpp
+++ test/OpenMP/taskloop_simd_reduction_codegen.cpp
@@ -148,7 +148,7 @@
// CHECK:[[SUB12:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK:store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]],
// CHECK:[[TMP65:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* %{{.+}}, i32 [[TMP0]], i32 1, i64 888, i64 72, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @{{.+}} to i32 (i32, i8*)*))
-// CHECK:call void @__kmpc_taskloop(%struct.ident_t* %{{.+}}, i32 [[TMP0]], i8* [[TMP65]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 0, i32 0, i64 0, i8* null)
+// CHECK:call void @__kmpc_taskloop(%struct.ident_t* %{{.+}}, i32 [[TMP0]], i8* [[TMP65]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 1, i32 0, i64 0, i8* null)
// CHECK:call void @__kmpc_end_taskgroup(%struct.ident_t*
// CHECK:ret i32
Index: test/OpenMP/taskloop_simd_private_codegen.cpp
===
--- test/OpenMP/taskloop_simd_private_codegen.cpp
+++ test/OpenMP/taskloop_simd_private_codegen.cpp
@@ -65,7 +65,7 @@
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
// LAMBDA: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 96, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
// LAMBDA: [[PRIVATES:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
-// LAMBDA: call void @__kmpc_taskloop(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 0, i32 0, i64 0, i8* null)
+// LAMBDA: call void @__kmpc_taskloop(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]], i32 1, i64* %{{.+}}, i64* %{{.+}}, i64 %{{.+}}, i32 1, i32 0, i64 0, i8* null)
// LAMBDA: ret
#pragma omp taskloop simd private(g, sivar)
for (int i = 0; i < 10; ++i) {
@@ -101,7 +101,7 @@
// BLOCKS: define{{.*}} internal{{.*}} vo
