maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/OpenMP/for_lastprivate_codegen.cpp

773 lines
36 KiB
C++
Raw Normal View History

[OPENMP] -fopenmp enables OpenMP support (fix for http://llvm.org/PR23492) -fopenmp turns on OpenMP support and links libiomp5 as OpenMP library. Also there is -fopenmp={libiomp5|libgomp} option that allows to override effect of -fopenmp and link libgomp library (if -fopenmp=libgomp is specified). Differential Revision: http://reviews.llvm.org/D9736 llvm-svn: 237769
2015-05-20 12:24:19 +08:00
// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s
// RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
struct SS {
int a;
int b : 4;
int &c;
SS(int &d) : a(0), b(0), c(d) {
#pragma omp parallel
#pragma omp for lastprivate(a, b, c)
for (int i = 0; i < 2; ++i)
#ifdef LAMBDA
[&]() {
++this->a, --b, (this)->c /= 1;
#pragma omp parallel
#pragma omp for lastprivate(a, b, c)
for (int i = 0; i < 2; ++i)
++(this)->a, --b, this->c /= 1;
}();
#elif defined(BLOCKS)
^{
++a;
--this->b;
(this)->c /= 1;
#pragma omp parallel
#pragma omp for lastprivate(a, b, c)
for (int i = 0; i < 2; ++i)
++(this)->a, --b, this->c /= 1;
}();
#else
++this->a, --b, c /= 1;
[OPENMP 4.5] Allow data members as loop counters in loop-based directives. OpenMP 4.5 allows privatization of non-static data members in non-static member functions. Patch allows to use and implicit privatization of data members used as counters in loop-based directives. llvm-svn: 265121
2016-04-01 17:23:34 +08:00
#endif
#pragma omp for
for (a = 0; a < 2; ++a)
#ifdef LAMBDA
[&]() {
[OPENMP] Fix for PR30632: Name mangling issue. Reworked fix after comments from Richard Smith. We must skip all CapturedDecl-based contexts when trying to get correct mangling number context. llvm-svn: 286953
2016-11-15 21:15:20 +08:00
--this->a, ++b, (this)->c *= 2;
[OPENMP 4.5] Allow data members as loop counters in loop-based directives. OpenMP 4.5 allows privatization of non-static data members in non-static member functions. Patch allows to use and implicit privatization of data members used as counters in loop-based directives. llvm-svn: 265121
2016-04-01 17:23:34 +08:00
#pragma omp parallel
#pragma omp for lastprivate(b)
for (b = 0; b < 2; ++b)
++(this)->a, --b, this->c /= 1;
}();
#elif defined(BLOCKS)
^{
++a;
--this->b;
(this)->c /= 1;
#pragma omp parallel
#pragma omp for
for (c = 0; c < 2; ++c)
++(this)->a, --b, this->c /= 1;
}();
#else
++this->a, --b, c /= 1;
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
#endif
}
};
template <typename T>
struct SST {
T a;
SST() : a(T()) {
#pragma omp parallel
#pragma omp for lastprivate(a)
for (int i = 0; i < 2; ++i)
#ifdef LAMBDA
[&]() {
[&]() {
++this->a;
#pragma omp parallel
#pragma omp for lastprivate(a)
for (int i = 0; i < 2; ++i)
++(this)->a;
}();
}();
#elif defined(BLOCKS)
^{
^{
++a;
#pragma omp parallel
#pragma omp for lastprivate(a)
for (int i = 0; i < 2; ++i)
++(this)->a;
}();
}();
#else
++(this)->a;
[OPENMP 4.5] Allow data members as loop counters in loop-based directives. OpenMP 4.5 allows privatization of non-static data members in non-static member functions. Patch allows to use and implicit privatization of data members used as counters in loop-based directives. llvm-svn: 265121
2016-04-01 17:23:34 +08:00
#endif
#pragma omp for
for (a = 0; a < 2; ++a)
#ifdef LAMBDA
[&]() {
++this->a;
#pragma omp parallel
#pragma omp for
for (a = 0; a < 2; ++(this)->a)
++(this)->a;
}();
#elif defined(BLOCKS)
^{
++a;
#pragma omp parallel
#pragma omp for
for (this->a = 0; a < 2; ++a)
++(this)->a;
}();
#else
++(this)->a;
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
#endif
}
};
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
template <class T>
struct S {
T f;
S(T a) : f(a) {}
S() : f() {}
S<T> &operator=(const S<T> &);
operator T() { return T(); }
~S() {}
};
[OPENMP] Propagate alignment from original variables to the private copies. Currently private copies of captured variables have default alignment. Patch makes private variables to have same alignment as original variables. llvm-svn: 247260
2015-09-10 17:48:30 +08:00
volatile int g __attribute__((aligned(128)))= 1212;
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
volatile int &g1 = g;
[OPENMP] Code cleanup for capturing of variables in OpenMP regions. llvm-svn: 236821
2015-05-08 18:41:21 +08:00
float f;
[OPENMP] Fixed codegen for lastprivate LCV in worksharing constructs. If loop control variable in a worksharing construct is marked as lastprivate, we should copy last calculated value of private counter back to original variable. llvm-svn: 237879
2015-05-21 15:59:51 +08:00
char cnt;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// CHECK: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8
// LAMBDA: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8
// BLOCKS: [[SS_TY:%.+]] = type { i{{[0-9]+}}, i8
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: [[S_FLOAT_TY:%.+]] = type { float }
// CHECK: [[S_INT_TY:%.+]] = type { i32 }
// CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
// CHECK-DAG: [[X:@.+]] = global double 0.0
[OPENMP] Code cleanup for capturing of variables in OpenMP regions. llvm-svn: 236821
2015-05-08 18:41:21 +08:00
// CHECK-DAG: [[F:@.+]] = global float 0.0
[OPENMP] Fixed codegen for lastprivate LCV in worksharing constructs. If loop control variable in a worksharing construct is marked as lastprivate, we should copy last calculated value of private counter back to original variable. llvm-svn: 237879
2015-05-21 15:59:51 +08:00
// CHECK-DAG: [[CNT:@.+]] = global i8 0
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
template <typename T>
T tmain() {
S<T> test;
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
SST<T> sst;
[OPENMP] Propagate alignment from original variables to the private copies. Currently private copies of captured variables have default alignment. Patch makes private variables to have same alignment as original variables. llvm-svn: 247260
2015-09-10 17:48:30 +08:00
T t_var __attribute__((aligned(128))) = T();
T vec[] __attribute__((aligned(128))) = {1, 2};
S<T> s_arr[] __attribute__((aligned(128))) = {1, 2};
S<T> &var __attribute__((aligned(128))) = test;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
#pragma omp parallel
#pragma omp for lastprivate(t_var, vec, s_arr, var)
for (int i = 0; i < 2; ++i) {
vec[i] = t_var;
s_arr[i] = var;
}
return T();
}
namespace A {
double x;
}
namespace B {
using A::x;
}
int main() {
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
static int sivar;
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
SS ss(sivar);
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
#ifdef LAMBDA
// LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0,
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// LAMBDA-LABEL: @main
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// LAMBDA: alloca [[SS_TY]],
// LAMBDA: alloca [[CAP_TY:%.+]],
// LAMBDA: call void [[OUTER_LAMBDA:@.+]]([[CAP_TY]]*
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
[&]() {
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* %{{.+}})
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
#pragma omp parallel
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
#pragma omp for lastprivate(g, g1, sivar)
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
for (int i = 0; i < 2; ++i) {
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// LAMBDA: define {{.+}} @{{.+}}([[SS_TY]]*
// LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
// LAMBDA: store i{{[0-9]+}} 0, i{{[0-9]+}}* %
// LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1
// LAMBDA: store i8
// LAMBDA: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2
// LAMBDA: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*)* [[SS_MICROTASK:@.+]] to void
[OPENMP 4.5] Allow data members as loop counters in loop-based directives. OpenMP 4.5 allows privatization of non-static data members in non-static member functions. Patch allows to use and implicit privatization of data members used as counters in loop-based directives. llvm-svn: 265121
2016-04-01 17:23:34 +08:00
// LAMBDA: call void @__kmpc_for_static_init_4(
// LAMBDA-NOT: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
[OPENMP] Fix for PR30632: Name mangling issue. Reworked fix after comments from Richard Smith. We must skip all CapturedDecl-based contexts when trying to get correct mangling number context. llvm-svn: 286953
2016-11-15 21:15:20 +08:00
// LAMBDA: call{{.*}} void [[SS_LAMBDA1:@[^ ]+]]
[OPENMP 4.5] Allow data members as loop counters in loop-based directives. OpenMP 4.5 allows privatization of non-static data members in non-static member functions. Patch allows to use and implicit privatization of data members used as counters in loop-based directives. llvm-svn: 265121
2016-04-01 17:23:34 +08:00
// LAMBDA: call void @__kmpc_for_static_fini(%
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// LAMBDA: ret
// LAMBDA: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}})
// LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0
[OPENMP 4.5] Codegen for data members in 'reduction' clause. OpenMP 4.5 allows to privatize non-static data members of current class in non-static member functions. Patch supports codegen for non-static data members in 'reduction' clauses. llvm-svn: 262460
2016-03-02 12:57:40 +08:00
// LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// LAMBDA: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2
// LAMBDA: call void @__kmpc_for_static_init_4(
// LAMBDA-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]*
[OPENMP] Fix for PR30632: Name mangling issue. Reworked fix after comments from Richard Smith. We must skip all CapturedDecl-based contexts when trying to get correct mangling number context. llvm-svn: 286953
2016-11-15 21:15:20 +08:00
// LAMBDA: call{{.*}} void [[SS_LAMBDA:@[^ ]+]]
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// LAMBDA: call void @__kmpc_for_static_fini(
// LAMBDA: br i1
// LAMBDA: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1
// LAMBDA: store i8 %{{.+}}, i8* [[B_REF]],
// LAMBDA: br label
// LAMBDA: ret void
// LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}})
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
[OPENMP] Fix PR35486: crash when collapsing loops with dependent iteration spaces. Though it is incorrect from point of view of OpenMP standard to have dependent iteration space in OpenMP loops, compiler should not crash. Patch fixes this problem. llvm-svn: 319700
2017-12-05 05:30:42 +08:00
// LAMBDA: alloca i{{[0-9]+}},
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// LAMBDA: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
// LAMBDA: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
// LAMBDA: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
// LAMBDA: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
// LAMBDA: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
// LAMBDA: call void @__kmpc_for_static_init_4(
// LAMBDA: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]],
// LAMBDA-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
// LAMBDA-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
// LAMBDA-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
// LAMBDA-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
// LAMBDA-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
// LAMBDA-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
// LAMBDA-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
// LAMBDA-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
// LAMBDA-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
// LAMBDA-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]],
// LAMBDA: call void @__kmpc_for_static_fini(
// LAMBDA: br i1
// LAMBDA: br label
// LAMBDA: ret void
[OPENMP] Fix for PR30632: Name mangling issue. Reworked fix after comments from Richard Smith. We must skip all CapturedDecl-based contexts when trying to get correct mangling number context. llvm-svn: 286953
2016-11-15 21:15:20 +08:00
// LAMBDA: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}})
// LAMBDA: ret void
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
// LAMBDA: alloca i{{[0-9]+}},
[OPENMP] Propagate alignment from original variables to the private copies. Currently private copies of captured variables have default alignment. Patch makes private variables to have same alignment as original variables. llvm-svn: 247260
2015-09-10 17:48:30 +08:00
// LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
[OPENMP] Capture references to global variables. In C++11 variable to global variables are considered as constant expressions and these variables are not captured in the outlined regions. Patch allows capturing of such variables in the OpenMP regions. llvm-svn: 315074
2017-10-07 00:17:25 +08:00
// LAMBDA: [[G1_PRIVATE_REF:%.+]] = alloca i{{[0-9]+}}*,
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
// LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}},
[OPENMP] Fix use of unsigned counters in loops with zero trip count. Patch fixes bugs in codegen for loops with unsigned counters and zero trip count. Previously preconditions for all loops were built using logic (Upper - Lower) > 0. But if the loop is a loop with zero trip count, then Upper - Lower is < 0 only for signed integer, for unsigned we're running into an underflow situation. In this patch we're using original Lower<Upper condition to check that loop body can be executed at least once. Also this allows to skip code generation for loops, if it is known that preconditions for the loop are always false. Differential Revision: http://reviews.llvm.org/D9103 llvm-svn: 235500
2015-04-22 19:59:37 +08:00
// LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
// LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
[OPENMP] Fix use of unsigned counters in loops with zero trip count. Patch fixes bugs in codegen for loops with unsigned counters and zero trip count. Previously preconditions for all loops were built using logic (Upper - Lower) > 0. But if the loop is a loop with zero trip count, then Upper - Lower is < 0 only for signed integer, for unsigned we're running into an underflow situation. In this patch we're using original Lower<Upper condition to check that loop body can be executed at least once. Also this allows to skip code generation for loops, if it is known that preconditions for the loop are always false. Differential Revision: http://reviews.llvm.org/D9103 llvm-svn: 235500
2015-04-22 19:59:37 +08:00
// LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
[OPENMP] Drop type qualifiers from private variables. If the variable is marked as private in OpenMP construct, the reference to this variable should not keep type qualifiers for the original variable. Private copy is not volatile or constant, so we can use unqualified type for private copy. llvm-svn: 242133
2015-07-14 18:32:29 +08:00
// LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
[OPENMP] Capture references to global variables. In C++11 variable to global variables are considered as constant expressions and these variables are not captured in the outlined regions. Patch allows capturing of such variables in the OpenMP regions. llvm-svn: 315074
2017-10-07 00:17:25 +08:00
// LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PRIVATE_REF]],
// LAMBDA: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G1_PRIVATE_ADDR]],
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
[OPENMP] Capture references to global variables. In C++11 variable to global variables are considered as constant expressions and these variables are not captured in the outlined regions. Patch allows capturing of such variables in the OpenMP regions. llvm-svn: 315074
2017-10-07 00:17:25 +08:00
// LAMBDA: [[G1_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PRIVATE_REF]],
// LAMBDA: store i{{[0-9]+}}* [[G1_PRIVATE_ADDR]], i{{[0-9]+}}** [[G1_PRIVATE_ADDR_REF]]
// LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
[OPENMP] Fix use of unsigned counters in loops with zero trip count. Patch fixes bugs in codegen for loops with unsigned counters and zero trip count. Previously preconditions for all loops were built using logic (Upper - Lower) > 0. But if the loop is a loop with zero trip count, then Upper - Lower is < 0 only for signed integer, for unsigned we're running into an underflow situation. In this patch we're using original Lower<Upper condition to check that loop body can be executed at least once. Also this allows to skip code generation for loops, if it is known that preconditions for the loop are always false. Differential Revision: http://reviews.llvm.org/D9103 llvm-svn: 235500
2015-04-22 19:59:37 +08:00
// LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
g = 1;
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
g1 = 1;
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
sivar = 2;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// Check for final copying of private values back to original vars.
// LAMBDA: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
// LAMBDA: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
// LAMBDA: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]]
// LAMBDA: [[LAST_THEN]]
// Actual copying.
// original g=private_g;
// LAMBDA: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
// LAMBDA: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]],
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// original sivar=private_sivar;
// LAMBDA: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
// LAMBDA: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* %{{.+}},
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// LAMBDA: br label %[[LAST_DONE]]
// LAMBDA: [[LAST_DONE]]
[OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found. Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then ``` if (__kmpc_cancel_barrier()) exit from outer construct; ``` code is generated. Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive. llvm-svn: 247681
2015-09-15 20:52:43 +08:00
// LAMBDA: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
[&]() {
// LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
// LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
g = 2;
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
g1 = 2;
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
sivar = 4;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
// LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
[OPENMP] Drop type qualifiers from private variables. If the variable is marked as private in OpenMP construct, the reference to this variable should not keep type qualifiers for the original variable. Private copy is not volatile or constant, so we can use unqualified type for private copy. llvm-svn: 242133
2015-07-14 18:32:29 +08:00
// LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
[OPENMP] Capture references to global variables. In C++11 variable to global variables are considered as constant expressions and these variables are not captured in the outlined regions. Patch allows capturing of such variables in the OpenMP regions. llvm-svn: 315074
2017-10-07 00:17:25 +08:00
// LAMBDA: [[G1_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// LAMBDA: [[G1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PTR_REF]]
// LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G1_REF]]
// LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
// LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]]
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
}();
}
}();
return 0;
#elif defined(BLOCKS)
// BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212,
// BLOCKS-LABEL: @main
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// BLOCKS: call
[opaque pointer types] Explicit non-pointer type for call expressions (migration for recent LLVM change to textual IR for calls) llvm-svn: 235147
2015-04-17 07:25:00 +08:00
// BLOCKS: call void {{%.+}}(i8
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
^{
// BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
#pragma omp parallel
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
#pragma omp for lastprivate(g, g1, sivar)
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
for (int i = 0; i < 2; ++i) {
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
[OPENMP] Propagate alignment from original variables to the private copies. Currently private copies of captured variables have default alignment. Patch makes private variables to have same alignment as original variables. llvm-svn: 247260
2015-09-10 17:48:30 +08:00
// BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// BLOCKS: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 4
// BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
// BLOCKS: store i{{[0-9]+}}* [[SIVAR]], i{{[0-9]+}}** [[SIVAR_ADDR:%.+]],
// BLOCKS: {{.+}} = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_ADDR]]
[OPENMP] Fix use of unsigned counters in loops with zero trip count. Patch fixes bugs in codegen for loops with unsigned counters and zero trip count. Previously preconditions for all loops were built using logic (Upper - Lower) > 0. But if the loop is a loop with zero trip count, then Upper - Lower is < 0 only for signed integer, for unsigned we're running into an underflow situation. In this patch we're using original Lower<Upper condition to check that loop body can be executed at least once. Also this allows to skip code generation for loops, if it is known that preconditions for the loop are always false. Differential Revision: http://reviews.llvm.org/D9103 llvm-svn: 235500
2015-04-22 19:59:37 +08:00
// BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
// BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
// BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
[OPENMP] Drop type qualifiers from private variables. If the variable is marked as private in OpenMP construct, the reference to this variable should not keep type qualifiers for the original variable. Private copy is not volatile or constant, so we can use unqualified type for private copy. llvm-svn: 242133
2015-07-14 18:32:29 +08:00
// BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// BLOCKS-NOT: [[G]]{{[[^:word:]]}}
// BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
// BLOCKS-NOT: [[G]]{{[[^:word:]]}}
[opaque pointer types] Explicit non-pointer type for call expressions (migration for recent LLVM change to textual IR for calls) llvm-svn: 235147
2015-04-17 07:25:00 +08:00
// BLOCKS: call void {{%.+}}(i8
[OPENMP] Fix use of unsigned counters in loops with zero trip count. Patch fixes bugs in codegen for loops with unsigned counters and zero trip count. Previously preconditions for all loops were built using logic (Upper - Lower) > 0. But if the loop is a loop with zero trip count, then Upper - Lower is < 0 only for signed integer, for unsigned we're running into an underflow situation. In this patch we're using original Lower<Upper condition to check that loop body can be executed at least once. Also this allows to skip code generation for loops, if it is known that preconditions for the loop are always false. Differential Revision: http://reviews.llvm.org/D9103 llvm-svn: 235500
2015-04-22 19:59:37 +08:00
// BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
g = 1;
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
g1 = 1;
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
sivar = 2;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// Check for final copying of private values back to original vars.
// BLOCKS: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
// BLOCKS: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
// BLOCKS: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]]
// BLOCKS: [[LAST_THEN]]
// Actual copying.
// original g=private_g;
// BLOCKS: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
// BLOCKS: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]],
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// BLOCKS: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
// BLOCKS: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* %{{.+}},
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// BLOCKS: br label %[[LAST_DONE]]
// BLOCKS: [[LAST_DONE]]
[OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found. Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then ``` if (__kmpc_cancel_barrier()) exit from outer construct; ``` code is generated. Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive. llvm-svn: 247681
2015-09-15 20:52:43 +08:00
// BLOCKS: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
g = 1;
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
g1 = 1;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
^{
// BLOCKS: define {{.+}} void {{@.+}}(i8*
g = 2;
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
g1 = 1;
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
sivar = 4;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// BLOCKS-NOT: [[G]]{{[[^:word:]]}}
[OPENMP] Drop type qualifiers from private variables. If the variable is marked as private in OpenMP construct, the reference to this variable should not keep type qualifiers for the original variable. Private copy is not volatile or constant, so we can use unqualified type for private copy. llvm-svn: 242133
2015-07-14 18:32:29 +08:00
// BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// BLOCKS-NOT: [[G]]{{[[^:word:]]}}
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
// BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
// BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// BLOCKS: ret
}();
}
}();
return 0;
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// BLOCKS: define {{.+}} @{{.+}}([[SS_TY]]*
// BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
// BLOCKS: store i{{[0-9]+}} 0, i{{[0-9]+}}* %
// BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1
// BLOCKS: store i8
// BLOCKS: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2
// BLOCKS: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*)* [[SS_MICROTASK:@.+]] to void
[OPENMP 4.5] Allow data members as loop counters in loop-based directives. OpenMP 4.5 allows privatization of non-static data members in non-static member functions. Patch allows to use and implicit privatization of data members used as counters in loop-based directives. llvm-svn: 265121
2016-04-01 17:23:34 +08:00
// BLOCKS: call void @__kmpc_for_static_init_4(
// BLOCKS-NOT: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
// BLOCKS: call void
// BLOCKS: call void @__kmpc_for_static_fini(%
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// BLOCKS: ret
// BLOCKS: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}})
// BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 0
[OPENMP 4.5] Codegen for data members in 'reduction' clause. OpenMP 4.5 allows to privatize non-static data members of current class in non-static member functions. Patch supports codegen for non-static data members in 'reduction' clauses. llvm-svn: 262460
2016-03-02 12:57:40 +08:00
// BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// BLOCKS: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 2
// BLOCKS: call void @__kmpc_for_static_init_4(
// BLOCKS-NOT: getelementptr {{.*}}[[SS_TY]], [[SS_TY]]*
// BLOCKS: call{{.*}} void
// BLOCKS: call void @__kmpc_for_static_fini(
// BLOCKS: br i1
// BLOCKS: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1
// BLOCKS: store i8 %{{.+}}, i8* [[B_REF]],
// BLOCKS: br label
// BLOCKS: ret void
// BLOCKS: define internal void @{{.+}}(i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}}, i32* {{.+}}, i32* {{.+}}, i32* {{.+}})
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
// BLOCKS: alloca i{{[0-9]+}},
[OPENMP] Fix PR35486: crash when collapsing loops with dependent iteration spaces. Though it is incorrect from point of view of OpenMP standard to have dependent iteration space in OpenMP loops, compiler should not crash. Patch fixes this problem. llvm-svn: 319700
2017-12-05 05:30:42 +08:00
// BLOCKS: alloca i{{[0-9]+}},
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// BLOCKS: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
// BLOCKS: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
// BLOCKS: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
// BLOCKS: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
// BLOCKS: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
// BLOCKS: call void @__kmpc_for_static_init_4(
// BLOCKS: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]],
// BLOCKS-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
// BLOCKS-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
// BLOCKS-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
// BLOCKS-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
// BLOCKS-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
// BLOCKS-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
// BLOCKS-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
// BLOCKS-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
// BLOCKS-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
// BLOCKS-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]],
// BLOCKS: call void @__kmpc_for_static_fini(
// BLOCKS: br i1
// BLOCKS: br label
// BLOCKS: ret void
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
#else
S<float> test;
int t_var = 0;
int vec[] = {1, 2};
S<float> s_arr[] = {1, 2};
S<float> var(3);
#pragma omp parallel
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
#pragma omp for lastprivate(t_var, vec, s_arr, var, sivar)
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
for (int i = 0; i < 2; ++i) {
vec[i] = t_var;
s_arr[i] = var;
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
sivar += i;
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
}
#pragma omp parallel
[OPENMP] Code cleanup for capturing of variables in OpenMP regions. llvm-svn: 236821
2015-05-08 18:41:21 +08:00
#pragma omp for lastprivate(A::x, B::x) firstprivate(f) lastprivate(f)
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
for (int i = 0; i < 2; ++i) {
A::x++;
}
[OPENMP] Fixed codegen for firstprivate variables, also marked as lastprivate. In some rare cases shared copies of lastprivate/firstprivate variables were not updated after the loop directive. llvm-svn: 237243
2015-05-13 18:23:02 +08:00
#pragma omp parallel
#pragma omp for firstprivate(f) lastprivate(f)
for (int i = 0; i < 2; ++i) {
A::x++;
}
[OPENMP] Fixed codegen for lastprivate LCV in worksharing constructs. If loop control variable in a worksharing construct is marked as lastprivate, we should copy last calculated value of private counter back to original variable. llvm-svn: 237879
2015-05-21 15:59:51 +08:00
#pragma omp parallel
#pragma omp for lastprivate(cnt)
for (cnt = 0; cnt < 2; ++cnt) {
A::x++;
}
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
return tmain<int>();
#endif
}
// CHECK: define i{{[0-9]+}} @main()
// CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i32*)* [[MAIN_MICROTASK:@.+]] to void
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK1:@.+]] to void
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK2:@.+]] to void
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK3:@.+]] to void
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: = call {{.+}} [[TMAIN_INT:@.+]]()
// CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
// CHECK: ret
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: define internal void [[MAIN_MICROTASK]](i32* noalias [[GTID_ADDR:%.+]], i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}})
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
[OPENMP] Fix PR35486: crash when collapsing loops with dependent iteration spaces. Though it is incorrect from point of view of OpenMP standard to have dependent iteration space in OpenMP loops, compiler should not crash. Patch fixes this problem. llvm-svn: 319700
2017-12-05 05:30:42 +08:00
// CHECK: alloca i{{[0-9]+}},
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
// CHECK: [[VEC_REF:%.+]] = load [2 x i32]*, [2 x i32]** %
// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** %
// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// Check for default initialization.
// CHECK-NOT: [[T_VAR_PRIV]]
// CHECK-NOT: [[VEC_PRIV]]
// CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 %{{.+}}, i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
// <Skip loop body>
// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}})
// Check for final copying of private values back to original vars.
// CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
// CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
// CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]]
// CHECK: [[LAST_THEN]]
// Actual copying.
// original t_var=private_t_var;
// CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]],
// CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_REF]],
// original vec[]=private_vec[];
// CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
// CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
Revert "Change memcpy/memset/memmove to have dest and source alignments." This reverts commit r253512. This likely broke the bots in: http://lab.llvm.org:8011/builders/clang-ppc64-elf-linux2/builds/20202 http://bb.pgr.jp/builders/clang-3stage-i686-linux/builds/3787 llvm-svn: 253542
2015-11-19 13:55:59 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// original s_arr[]=private_s_arr[];
// CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]] to [[S_FLOAT_TY]]*
// CHECK: [[S_ARR_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2
// CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_END]]
// CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]]
// CHECK: [[S_ARR_BODY]]
// CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}})
// CHECK: br i1 {{.+}}, label %[[S_ARR_BODY_DONE]], label %[[S_ARR_BODY]]
// CHECK: [[S_ARR_BODY_DONE]]
// original var=private_var;
// CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* [[VAR_REF]], [[S_FLOAT_TY]]* {{.*}} [[VAR_PRIV]])
Allow static local variables specified on data-sharing attribute clauses. http://reviews.llvm.org/D11619 llvm-svn: 247715
2015-09-16 02:56:58 +08:00
// CHECK: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIV]],
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: br label %[[LAST_DONE]]
// CHECK: [[LAST_DONE]]
// CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
// CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
[OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found. Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then ``` if (__kmpc_cancel_barrier()) exit from outer construct; ``` code is generated. Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive. llvm-svn: 247681
2015-09-15 20:52:43 +08:00
// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: ret void
//
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
[OPENMP] Code cleanup for capturing of variables in OpenMP regions. llvm-svn: 236821
2015-05-08 18:41:21 +08:00
// CHECK: [[F_PRIV:%.+]] = alloca float,
// CHECK-NOT: alloca float
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: [[X_PRIV:%.+]] = alloca double,
[OPENMP] Code cleanup for capturing of variables in OpenMP regions. llvm-svn: 236821
2015-05-08 18:41:21 +08:00
// CHECK-NOT: alloca float
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK-NOT: alloca double
// Check for default initialization.
// CHECK-NOT: [[X_PRIV]]
[OPENMP] Code cleanup for capturing of variables in OpenMP regions. llvm-svn: 236821
2015-05-08 18:41:21 +08:00
// CHECK: [[F_VAL:%.+]] = load float, float* [[F]],
// CHECK: store float [[F_VAL]], float* [[F_PRIV]],
// CHECK-NOT: [[X_PRIV]]
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
[OPENMP] Fix use of unsigned counters in loops with zero trip count. Patch fixes bugs in codegen for loops with unsigned counters and zero trip count. Previously preconditions for all loops were built using logic (Upper - Lower) > 0. But if the loop is a loop with zero trip count, then Upper - Lower is < 0 only for signed integer, for unsigned we're running into an underflow situation. In this patch we're using original Lower<Upper condition to check that loop body can be executed at least once. Also this allows to skip code generation for loops, if it is known that preconditions for the loop are always false. Differential Revision: http://reviews.llvm.org/D9103 llvm-svn: 235500
2015-04-22 19:59:37 +08:00
// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// <Skip loop body>
[OPENMP] Fix use of unsigned counters in loops with zero trip count. Patch fixes bugs in codegen for loops with unsigned counters and zero trip count. Previously preconditions for all loops were built using logic (Upper - Lower) > 0. But if the loop is a loop with zero trip count, then Upper - Lower is < 0 only for signed integer, for unsigned we're running into an underflow situation. In this patch we're using original Lower<Upper condition to check that loop body can be executed at least once. Also this allows to skip code generation for loops, if it is known that preconditions for the loop are always false. Differential Revision: http://reviews.llvm.org/D9103 llvm-svn: 235500
2015-04-22 19:59:37 +08:00
// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// Check for final copying of private values back to original vars.
// CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
// CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
// CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]]
// CHECK: [[LAST_THEN]]
// Actual copying.
// original x=private_x;
// CHECK: [[X_VAL:%.+]] = load double, double* [[X_PRIV]],
// CHECK: store double [[X_VAL]], double* [[X]],
[OPENMP] Code cleanup for capturing of variables in OpenMP regions. llvm-svn: 236821
2015-05-08 18:41:21 +08:00
// original f=private_f;
// CHECK: [[F_VAL:%.+]] = load float, float* [[F_PRIV]],
// CHECK: store float [[F_VAL]], float* [[F]],
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK-NEXT: br label %[[LAST_DONE]]
// CHECK: [[LAST_DONE]]
[OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found. Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then ``` if (__kmpc_cancel_barrier()) exit from outer construct; ``` code is generated. Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive. llvm-svn: 247681
2015-09-15 20:52:43 +08:00
// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: ret void
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: define internal void [[MAIN_MICROTASK2]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
[OPENMP] Fixed codegen for firstprivate variables, also marked as lastprivate. In some rare cases shared copies of lastprivate/firstprivate variables were not updated after the loop directive. llvm-svn: 237243
2015-05-13 18:23:02 +08:00
// CHECK: [[F_PRIV:%.+]] = alloca float,
// CHECK-NOT: alloca float
// Check for default initialization.
// CHECK: [[F_VAL:%.+]] = load float, float* [[F]],
// CHECK: store float [[F_VAL]], float* [[F_PRIV]],
// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
// <Skip loop body>
// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
// Check for final copying of private values back to original vars.
// CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
// CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
// CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]]
// CHECK: [[LAST_THEN]]
// Actual copying.
// original f=private_f;
// CHECK: [[F_VAL:%.+]] = load float, float* [[F_PRIV]],
// CHECK: store float [[F_VAL]], float* [[F]],
// CHECK-NEXT: br label %[[LAST_DONE]]
// CHECK: [[LAST_DONE]]
[OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found. Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then ``` if (__kmpc_cancel_barrier()) exit from outer construct; ``` code is generated. Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive. llvm-svn: 247681
2015-09-15 20:52:43 +08:00
// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
[OPENMP] Fixed codegen for firstprivate variables, also marked as lastprivate. In some rare cases shared copies of lastprivate/firstprivate variables were not updated after the loop directive. llvm-svn: 237243
2015-05-13 18:23:02 +08:00
// CHECK: ret void
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: define internal void [[MAIN_MICROTASK3]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
[OPENMP] Fixed codegen for lastprivate LCV in worksharing constructs. If loop control variable in a worksharing construct is marked as lastprivate, we should copy last calculated value of private counter back to original variable. llvm-svn: 237879
2015-05-21 15:59:51 +08:00
// CHECK: [[CNT_PRIV:%.+]] = alloca i8,
// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* [[OMP_LB:%[^,]+]], i32* [[OMP_UB:%[^,]+]], i32* [[OMP_ST:%[^,]+]], i32 1, i32 1)
// UB = min(UB, GlobalUB)
// CHECK-NEXT: [[UB:%.+]] = load i32, i32* [[OMP_UB]]
// CHECK-NEXT: [[UBCMP:%.+]] = icmp sgt i32 [[UB]], 1
// CHECK-NEXT: br i1 [[UBCMP]], label [[UB_TRUE:%[^,]+]], label [[UB_FALSE:%[^,]+]]
// CHECK: [[UBRESULT:%.+]] = phi i32 [ 1, [[UB_TRUE]] ], [ [[UBVAL:%[^,]+]], [[UB_FALSE]] ]
// CHECK-NEXT: store i32 [[UBRESULT]], i32* [[OMP_UB]]
// CHECK-NEXT: [[LB:%.+]] = load i32, i32* [[OMP_LB]]
// CHECK-NEXT: store i32 [[LB]], i32* [[OMP_IV:[^,]+]]
// <Skip loop body>
// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
// Check for final copying of private values back to original vars.
// CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
// CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
// CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]]
// CHECK: [[LAST_THEN]]
// Calculate private cnt value.
[OPENMP] Fix codegen for lastprivate loop counters. Patch fixes bug with codegen for lastprivate loop counters. Also it may improve performance for lastprivates calculations in some cases. llvm-svn: 261209
2016-02-18 21:48:15 +08:00
// CHECK: store i8 2, i8* [[CNT_PRIV]]
// original cnt=private_cnt;
[OPENMP] Fixed codegen for lastprivate LCV in worksharing constructs. If loop control variable in a worksharing construct is marked as lastprivate, we should copy last calculated value of private counter back to original variable. llvm-svn: 237879
2015-05-21 15:59:51 +08:00
// CHECK: [[CNT_VAL:%.+]] = load i8, i8* [[CNT_PRIV]],
// CHECK: store i8 [[CNT_VAL]], i8* [[CNT]],
// CHECK-NEXT: br label %[[LAST_DONE]]
// CHECK: [[LAST_DONE]]
[OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found. Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then ``` if (__kmpc_cancel_barrier()) exit from outer construct; ``` code is generated. Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive. llvm-svn: 247681
2015-09-15 20:52:43 +08:00
// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
[OPENMP] Fixed codegen for lastprivate LCV in worksharing constructs. If loop control variable in a worksharing construct is marked as lastprivate, we should copy last calculated value of private counter back to original variable. llvm-svn: 237879
2015-05-21 15:59:51 +08:00
// CHECK: ret void
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
// CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
// CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
// CHECK: ret
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// CHECK: define {{.+}} @{{.+}}([[SS_TY]]*
// CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
// CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* %
// CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1
// CHECK: store i8
// CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[SS_TY]]*)* [[SS_MICROTASK:@.+]] to void
[OPENMP 4.5] Allow data members as loop counters in loop-based directives. OpenMP 4.5 allows privatization of non-static data members in non-static member functions. Patch allows to use and implicit privatization of data members used as counters in loop-based directives. llvm-svn: 265121
2016-04-01 17:23:34 +08:00
// CHECK: call void @__kmpc_for_static_init_4(
// CHECK-NOT: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 0
// CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 1
// CHECK: getelementptr inbounds [[SS_TY]], [[SS_TY]]* %{{.+}}, i32 0, i32 2
// CHECK: call void @__kmpc_for_static_fini(%
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// CHECK: ret
// CHECK: define internal void [[SS_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [[SS_TY]]* %{{.+}})
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
[OPENMP] Fix PR35486: crash when collapsing loops with dependent iteration spaces. Though it is incorrect from point of view of OpenMP standard to have dependent iteration space in OpenMP loops, compiler should not crash. Patch fixes this problem. llvm-svn: 319700
2017-12-05 05:30:42 +08:00
// CHECK: alloca i{{[0-9]+}},
[OPENMP 4.5] Codegen for member decls in 'lastprivate' clause. OpenMP 4.5 allows to privatize non-static member decls in non-static member functions. Patch captures such decls by reference in general (for bitfields, by value) and then operates with this capture. For bitfields, at the end of codegen for lastprivates original bitfield is updated with the value of captured copy. llvm-svn: 261824
2016-02-25 13:25:57 +08:00
// CHECK: [[A_PRIV:%.+]] = alloca i{{[0-9]+}},
// CHECK: [[B_PRIV:%.+]] = alloca i{{[0-9]+}},
// CHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}},
// CHECK: store i{{[0-9]+}}* [[A_PRIV]], i{{[0-9]+}}** [[REFA:%.+]],
// CHECK: store i{{[0-9]+}}* [[C_PRIV]], i{{[0-9]+}}** [[REFC:%.+]],
// CHECK: call void @__kmpc_for_static_init_4(
// CHECK: [[A_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFA]],
// CHECK-NEXT: [[A_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A_PRIV]],
// CHECK-NEXT: [[INC:%.+]] = add nsw i{{[0-9]+}} [[A_VAL]], 1
// CHECK-NEXT: store i{{[0-9]+}} [[INC]], i{{[0-9]+}}* [[A_PRIV]],
// CHECK-NEXT: [[B_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[B_PRIV]],
// CHECK-NEXT: [[DEC:%.+]] = add nsw i{{[0-9]+}} [[B_VAL]], -1
// CHECK-NEXT: store i{{[0-9]+}} [[DEC]], i{{[0-9]+}}* [[B_PRIV]],
// CHECK-NEXT: [[C_PRIV:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[REFC]],
// CHECK-NEXT: [[C_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[C_PRIV]],
// CHECK-NEXT: [[DIV:%.+]] = sdiv i{{[0-9]+}} [[C_VAL]], 1
// CHECK-NEXT: store i{{[0-9]+}} [[DIV]], i{{[0-9]+}}* [[C_PRIV]],
// CHECK: call void @__kmpc_for_static_fini(
// CHECK: br i1
// CHECK: [[B_REF:%.+]] = getelementptr {{.*}}[[SS_TY]], [[SS_TY]]* %{{.*}}, i32 0, i32 1
// CHECK: store i8 %{{.+}}, i8* [[B_REF]],
// CHECK: br label
// CHECK: ret void
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
// CHECK: alloca i{{[0-9]+}},
[OPENMP] Propagate alignment from original variables to the private copies. Currently private copies of captured variables have default alignment. Patch makes private variables to have same alignment as original variables. llvm-svn: 247260
2015-09-10 17:48:30 +08:00
// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128
// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128
// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
// CHECK: [[VAR_PRIV_REF:%.+]] = alloca [[S_INT_TY]]*,
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// Check for default initialization.
// CHECK-NOT: [[T_VAR_PRIV]]
// CHECK-NOT: [[VEC_PRIV]]
// CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
// CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
[OPENMP] Outlined function for parallel and other regions with list of captured variables. Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least. Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record. llvm-svn: 247251
2015-09-10 16:12:02 +08:00
// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]])
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
// CHECK: store [[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]** [[VAR_PRIV_REF]]
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 %{{.+}}, i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
// <Skip loop body>
// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}})
// Check for final copying of private values back to original vars.
// CHECK: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
// CHECK: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
// CHECK: br i1 [[IS_LAST_ITER:%.+]], label %[[LAST_THEN:.+]], label %[[LAST_DONE:.+]]
// CHECK: [[LAST_THEN]]
// Actual copying.
// original t_var=private_t_var;
// CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]],
// CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_REF]],
// original vec[]=private_vec[];
// CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
// CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
Revert "Change memcpy/memset/memmove to have dest and source alignments." This reverts commit r253512. This likely broke the bots in: http://lab.llvm.org:8011/builders/clang-ppc64-elf-linux2/builds/20202 http://bb.pgr.jp/builders/clang-3stage-i686-linux/builds/3787 llvm-svn: 253542
2015-11-19 13:55:59 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// original s_arr[]=private_s_arr[];
// CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]] to [[S_INT_TY]]*
// CHECK: [[S_ARR_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2
// CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_END]]
// CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]]
// CHECK: [[S_ARR_BODY]]
// CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}})
// CHECK: br i1 {{.+}}, label %[[S_ARR_BODY_DONE]], label %[[S_ARR_BODY]]
// CHECK: [[S_ARR_BODY_DONE]]
// original var=private_var;
[OPENMP] Fix for http://llvm.org/PR24674: assertion failed and and abort trap Fix processing of shared variables with reference types in OpenMP constructs. Previously, if the variable was not marked in one of the private clauses, the reference to this variable was emitted incorrectly and caused an assertion later. llvm-svn: 246846
2015-09-04 19:26:21 +08:00
// CHECK: [[VAR_PRIV1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PRIV_REF]],
// CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[VAR_REF]], [[S_INT_TY]]* {{.*}} [[VAR_PRIV1]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: br label %[[LAST_DONE]]
// CHECK: [[LAST_DONE]]
// CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
// CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]*
// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
[OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found. Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then ``` if (__kmpc_cancel_barrier()) exit from outer construct; ``` code is generated. Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive. llvm-svn: 247681
2015-09-15 20:52:43 +08:00
// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
[OPENMP] Codegen for 'lastprivate' clause in 'for' directive. #pragma omp for lastprivate(<var>) for (i = a; i < b; ++b) <BODY>; This construct is translated into something like: <last_iter> = alloca i32 <lastprivate_var> = alloca <type> <last_iter> = 0 ; No initializer for simple variables or a default constructor is called for objects. ; For arrays perform element by element initialization by the call of the default constructor. ... OMP_FOR_START(...,<last_iter>, ..); sets <last_iter> to 1 if this is the last iteration. <BODY> ... OMP_FOR_END if (<last_iter> != 0) { <var> = <lastprivate_var> ; Update original variable with the lastprivate value. } call __kmpc_cancel_barrier() ; an implicit barrier to avoid possible data race. Differential Revision: http://reviews.llvm.org/D8658 llvm-svn: 235074
2015-04-16 12:54:05 +08:00
// CHECK: ret void
#endif