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/distribute_lastprivate_code...

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

406 lines
26 KiB
C++
Raw Normal View History

[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
// RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-32
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-32
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY0 %s
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY0 %s
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY0 %s
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY1 %s
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY1 %s
// RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY1 %s
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
[OPENMP]Fix PR39372: Does not complain about loop bound variable not being shared. According to the standard, the variables with unspecified data-sharing attributes in presence of `default(none)` clause must be reported to users. Compiler did not generate error reports for the variables used in other OpenMP regions. Patch fixes this. llvm-svn: 345533
2018-10-30 04:17:42 +08:00
// RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-target | FileCheck --check-prefix SIMD-ONLY1 %s
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// SIMD-ONLY1-NOT: {{__kmpc|__tgt}}
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
template <class T>
struct S {
T f;
S(T a) : f(a) {}
S() : f() {}
operator T() { return T(); }
~S() {}
};
// CHECK: [[S_FLOAT_TY:%.+]] = type { float }
// CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
template <typename T>
T tmain() {
S<T> test;
T t_var = T();
T vec[] = {1, 2};
S<T> s_arr[] = {1, 2};
S<T> &var = test;
#pragma omp target
#pragma omp teams
#pragma omp distribute lastprivate(t_var, vec, s_arr, s_arr, var, var)
for (int i = 0; i < 2; ++i) {
vec[i] = t_var;
s_arr[i] = var;
}
return T();
}
int main() {
static int svar;
volatile double g;
volatile double &g1 = g;
#ifdef LAMBDA
// LAMBDA-LABEL: @main
// LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
[&]() {
static float sfvar;
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
// LAMBDA: call i{{[0-9]+}} @__tgt_target_teams(
// LAMBDA: call void [[OFFLOADING_FUN:@.+]](
// LAMBDA: define{{.+}} void [[OFFLOADING_FUN]](
// LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_teams({{.+}}, i32 4, {{.+}}* [[OMP_OUTLINED:@.+]] to {{.+}})
#pragma omp target
#pragma omp teams
#pragma omp distribute lastprivate(g, g1, svar, sfvar)
for (int i = 0; i < 2; ++i) {
// LAMBDA: define{{.*}} internal{{.*}} void [[OMP_OUTLINED]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, double*{{.+}} [[G_IN:%.+]], double*{{.+}} [[G1_IN:%.+]], i{{[0-9]+}}*{{.+}} [[SVAR_IN:%.+]], float*{{.+}} [[SFVAR_IN:%.+]])
// LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double*,
// LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca double*,
// LAMBDA: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}*,
// LAMBDA: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float*,
// LAMBDA: [[TMP_G1:%.+]] = alloca double*,
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// LAMBDA: [[TMP1_G1:%.+]] = alloca double*,
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// loop variables
// 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] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// LAMBDA: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}},
// LAMBDA: [[G_PRIVATE:%.+]] = alloca double,
// LAMBDA: [[G1_PRIVATE:%.+]] = alloca double,
// LAMBDA: [[TMP_G1_PRIVATE:%.+]] = alloca double*,
// LAMBDA: [[SVAR_PRIVATE:%.+]] = alloca i{{[0-9]+}},
// LAMBDA: [[SFVAR_PRIVATE:%.+]] = alloca float,
// LAMBDA: store double* [[G_IN]], double** [[G_PRIVATE_ADDR]],
// LAMBDA: store double* [[G1_IN]], double** [[G1_PRIVATE_ADDR]],
// LAMBDA: store i{{[0-9]+}}* [[SVAR_IN]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]],
// LAMBDA: store float* [[SFVAR_IN]], float** [[SFVAR_PRIVATE_ADDR]],
// init private variables
// LAMBDA: [[G_IN_REF:%.+]] = load double*, double** [[G_PRIVATE_ADDR]],
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** [[G1_PRIVATE_ADDR]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// LAMBDA: [[SVAR_IN_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR]],
// LAMBDA: [[SFVAR_IN_REF:%.+]] = load float*, float** [[SFVAR_PRIVATE_ADDR]],
// LAMBDA: store double* [[G1_IN_REF]], double** [[TMP_G1]],
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// LAMBDA: [[G1_IN_REF:%.+]] = load double*, double** [[TMP_G1]],
// LAMBDA: store double* [[G1_IN_REF]], double** [[TMP1_G1]],
// LAMBDA: [[TMP_G1_VAL:%.+]] = load double*, double** [[TMP1_G1]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// LAMBDA: store double* [[G1_PRIVATE]], double** [[TMP_G1_PRIVATE]],
g = 1;
g1 = 1;
svar = 3;
sfvar = 4.0;
// LAMBDA: call {{.*}}void @__kmpc_for_static_init_4(
// LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE]],
// LAMBDA: [[TMP_G1_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]],
// LAMBDA: store{{.+}} double 1.0{{.+}}, double* [[TMP_G1_REF]],
// LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SVAR_PRIVATE]],
// LAMBDA: store float 4.0{{.+}}, float* [[SFVAR_PRIVATE]],
// LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// LAMBDA: store double* [[G_PRIVATE]], double** [[G_PRIVATE_ADDR_REF]],
// LAMBDA: [[TMP_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// LAMBDA: [[G1_PRIVATE_ADDR_FROM_TMP:%.+]] = load double*, double** [[TMP_G1_PRIVATE]],
// LAMBDA: store double* [[G1_PRIVATE_ADDR_FROM_TMP]], double** [[TMP_PRIVATE_ADDR_REF]],
// LAMBDA: [[SVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// LAMBDA: store i{{[0-9]+}}* [[SVAR_PRIVATE]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR_REF]]
// LAMBDA: [[SFVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
// LAMBDA: store float* [[SFVAR_PRIVATE]], float** [[SFVAR_PRIVATE_ADDR_REF]]
// LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
// LAMBDA: call {{.*}}void @__kmpc_for_static_fini(
// LAMBDA: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]],
// LAMBDA: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0
// LAMBDA: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]]
// LAMBDA: [[OMP_LASTPRIV_BLOCK]]:
// LAMBDA: [[G_PRIV_VAL:%.+]] = load double, double* [[G_PRIVATE]],
// LAMBDA: store{{.*}} double [[G_PRIV_VAL]], double* [[G_IN_REF]],
// LAMBDA: [[TMP_G1_PRIV_REF:%.+]] = load double*, double** [[TMP_G1_PRIVATE]],
// LAMBDA: [[TMP_G1_PRIV_VAL:%.+]] = load double, double* [[TMP_G1_PRIV_REF]],
// LAMBDA: store{{.*}} double [[TMP_G1_PRIV_VAL]], double* [[TMP_G1_VAL]],
// LAMBDA: [[SVAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SVAR_PRIVATE]],
// LAMBDA: store i{{[0-9]+}} [[SVAR_PRIV_VAL]], i{{[0-9]+}}* [[SVAR_IN_REF]],
// LAMBDA: [[SFVAR_PRIV_VAL:%.+]] = load float, float* [[SFVAR_PRIVATE]],
// LAMBDA: store float [[SFVAR_PRIV_VAL]], float* [[SFVAR_IN_REF]],
// LAMBDA: br label %[[OMP_LASTPRIV_DONE]]
// LAMBDA: [[OMP_LASTPRIV_DONE]]:
// LAMBDA: ret
[&]() {
// LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
// LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
g = 2;
g1 = 2;
svar = 4;
sfvar = 8.0;
// 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 double*, double** [[G_PTR_REF]]
// LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
// LAMBDA: [[TMP_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// LAMBDA: [[G1_REF:%.+]] = load double*, double** [[TMP_PTR_REF]]
// LAMBDA: store double 2.0{{.+}}, double* [[G1_REF]],
// LAMBDA: [[SVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// LAMBDA: [[SVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PTR_REF]]
// LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SVAR_REF]]
// LAMBDA: [[SFVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
// LAMBDA: [[SFVAR_REF:%.+]] = load float*, float** [[SFVAR_PTR_REF]]
// LAMBDA: store float 8.0{{.+}}, float* [[SFVAR_REF]]
}();
}
}();
return 0;
#else
S<float> test;
int t_var = 0;
int vec[] = {1, 2};
S<float> s_arr[] = {1, 2};
S<float> &var = test;
#pragma omp target
#pragma omp teams
#pragma omp distribute lastprivate(t_var, vec, s_arr, s_arr, var, var, svar)
for (int i = 0; i < 2; ++i) {
vec[i] = t_var;
s_arr[i] = var;
}
int i;
return tmain<int>();
#endif
}
// CHECK: define{{.*}} i{{[0-9]+}} @main()
// CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
Revert "[CodeGenCXX] Treat 'this' as noalias in constructors" This reverts commit https://reviews.llvm.org/rL344150 which causes MachineOutliner related failures on the ppc64le multistage buildbot. llvm-svn: 344526
2018-10-15 23:43:00 +08:00
// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: call i{{[0-9]+}} @__tgt_target_teams(
// CHECK: call void [[OFFLOAD_FUN:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_FLOAT_TY]]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, i{{[0-9]+}} {{.+}})
// CHECK: ret
// CHECK: define{{.+}} [[OFFLOAD_FUN]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]*{{.+}} {{.+}}, [2 x [[S_FLOAT_TY]]]*{{.+}} {{.+}}, [[S_FLOAT_TY]]*{{.+}} {{.+}}, i{{[0-9]+}} {{.+}})
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(
// CHECK: ret
//
// CHECK: define internal void [[OMP_OUTLINED:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i{{[0-9]+}}*{{.+}} [[T_VAR_IN:%.+]], [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN:%.+]], [2 x [[S_FLOAT_TY]]]*{{.+}} [[S_ARR_IN:%.+]], [[S_FLOAT_TY]]*{{.+}} [[VAR_IN:%.+]], i{{[0-9]+}}*{{.*}} [[S_VAR_IN:%.+]])
// CHECK: {{.+}} = alloca i{{[0-9]+}}*,
// CHECK: {{.+}} = alloca i{{[0-9]+}}*,
// CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}*,
// CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*,
// CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*,
// CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*,
// CHECK: [[SVAR_ADDR:%.+]] = alloca i{{[0-9]+}}*,
// CHECK: [[TMP:%.*]] = alloca [[S_FLOAT_TY]]*,
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// CHECK: [[TMP1:%.*]] = alloca [[S_FLOAT_TY]]*,
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// skip loop variables
// 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] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: [[OMP_IS_LAST:%.+]] = alloca i{{[0-9]+}},
// 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]],
// CHECK: [[TMP_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*,
// CHECK: [[S_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
// copy from parameters to local address variables
// CHECK: store i{{[0-9]+}}* [[T_VAR_IN]], i{{[0-9]+}}** [[T_VAR_ADDR]],
// CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN]], [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
// CHECK: store [2 x [[S_FLOAT_TY]]]* [[S_ARR_IN]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]],
// CHECK: store [[S_FLOAT_TY]]* [[VAR_IN]], [[S_FLOAT_TY]]** [[VAR_ADDR]],
// CHECK: store i{{[0-9]+}}* [[S_VAR_IN]], i{{[0-9]+}}** [[SVAR_ADDR]],
// load content of local address variables
// CHECK: [[T_VAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR]],
// CHECK: [[VEC_ADDR_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]],
// CHECK: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]],
// CHECK: [[VAR_ADDR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_ADDR]],
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// CHECK: [[SVAR_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_ADDR]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: store [[S_FLOAT_TY]]* [[VAR_ADDR_REF]], [[S_FLOAT_TY]]** [[TMP]],
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// CHECK: [[VAR_ADDR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP]],
// CHECK: store [[S_FLOAT_TY]]* [[VAR_ADDR_REF]], [[S_FLOAT_TY]]** [[TMP1]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST]],
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// CHECK: [[TMP_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP1]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// the distribute loop
// CHECK: call void @__kmpc_for_static_init_4(
// assignment: vec[i] = t_var;
// CHECK: [[T_VAR_PRIV_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV]],
// CHECK: [[VEC_PTR:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}}
// CHECK: store i{{[0-9]+}} [[T_VAR_PRIV_VAL]], i{{[0-9]+}}* [[VEC_PTR]],
// assignment: s_arr[i] = var;
// CHECK-DAG: [[S_ARR_PTR:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]],
// CHECK-DAG: [[TMP_VAL:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]],
// CHECK-DAG: [[S_ARR_PTR_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_PTR]] to i8*
// CHECK-DAG: [[TMP_VAL_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: call void @__kmpc_for_static_fini(
// lastprivates
// CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST]],
// CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0
// CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]]
// CHECK: [[OMP_LASTPRIV_BLOCK]]:
// 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_ADDR_REF]],
// CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF]] to i8*
// CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]] to [[S_FLOAT_TY]]*
// CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2
// CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]]
// CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]]
// CHECK: [[S_ARR_COPY_BLOCK]]:
// CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}}
// CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_FLOAT_TY]]*{{.+}}
// CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_DST_EL]] to i8*
// CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[S_ARR_SRC_EL]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}})
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1
// CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}}
// CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_FLOAT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]]
// CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]]
// CHECK: [[S_ARR_COPY_DONE]]:
// CHECK: [[TMP_VAL1:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[TMP_PRIV]],
// CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_REF]] to i8*
// CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[TMP_VAL1]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}})
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: [[SVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[S_VAR_PRIV]],
// CHECK: store i{{[0-9]+}} [[SVAR_VAL]], i{{[0-9]+}}* [[SVAR_ADDR_REF]],
// CHECK: ret void
// template tmain
// CHECK: define{{.*}} i{{[0-9]+}} [[TMAIN_INT:@.+]]()
// CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
Revert "[CodeGenCXX] Treat 'this' as noalias in constructors" This reverts commit https://reviews.llvm.org/rL344150 which causes MachineOutliner related failures on the ppc64le multistage buildbot. llvm-svn: 344526
2018-10-15 23:43:00 +08:00
// CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: call i{{[0-9]+}} @__tgt_target_teams(
// CHECK: call void [[OFFLOAD_FUN_1:@.+]](i{{[0-9]+}} {{.+}}, [2 x i{{[0-9]+}}]* {{.+}}, [2 x [[S_INT_TY]]]* {{.+}}, [[S_INT_TY]]* {{.+}})
// CHECK: ret
// CHECK: define internal void [[OFFLOAD_FUN_1]](
// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_teams(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4,
// CHECK: ret
// CHECK: define internal void [[OMP_OUTLINED_1:@.+]](i{{[0-9]+}}* noalias [[GTID_ADDR1:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i{{[0-9]+}}*{{.+}} [[T_VAR_IN1:%.+]], [2 x i{{[0-9]+}}]*{{.+}} [[VEC_IN1:%.+]], [2 x [[S_INT_TY]]]*{{.+}} [[S_ARR_IN1:%.+]], [[S_INT_TY]]*{{.+}} [[VAR_IN1:%.+]])
// skip alloca of global_tid and bound_tid
// CHECK: {{.+}} = alloca i{{[0-9]+}}*,
// CHECK: {{.+}} = alloca i{{[0-9]+}}*,
// CHECK: [[T_VAR_ADDR1:%.+]] = alloca i{{[0-9]+}}*,
// CHECK: [[VEC_ADDR1:%.+]] = alloca [2 x i{{[0-9]+}}]*,
// CHECK: [[S_ARR_ADDR1:%.+]] = alloca [2 x [[S_INT_TY]]]*,
// CHECK: [[VAR_ADDR1:%.+]] = alloca [[S_INT_TY]]*,
// CHECK: [[TMP:%.+]] = alloca [[S_INT_TY]]*,
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// CHECK: [[TMP1:%.+]] = alloca [[S_INT_TY]]*,
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// skip loop variables
// 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] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: [[OMP_IS_LAST1:%.+]] = alloca i{{[0-9]+}},
// CHECK: [[T_VAR_PRIV1:%.+]] = alloca i{{[0-9]+}},
// CHECK: [[VEC_PRIV1:%.+]] = alloca [2 x i{{[0-9]+}}],
// CHECK: [[S_ARR_PRIV1:%.+]] = alloca [2 x [[S_INT_TY]]],
// CHECK: [[VAR_PRIV1:%.+]] = alloca [[S_INT_TY]],
// CHECK: [[TMP_PRIV1:%.+]] = alloca [[S_INT_TY]]*,
// skip init of bound and global tid
// CHECK: store i{{[0-9]+}}* {{.*}},
// CHECK: store i{{[0-9]+}}* {{.*}},
// copy from parameters to local address variables
// CHECK: store i{{[0-9]+}}* [[T_VAR_IN1]], i{{[0-9]+}}** [[T_VAR_ADDR1]],
// CHECK: store [2 x i{{[0-9]+}}]* [[VEC_IN1]], [2 x i{{[0-9]+}}]** [[VEC_ADDR1]],
// CHECK: store [2 x [[S_INT_TY]]]* [[S_ARR_IN1]], [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]],
// CHECK: store [[S_INT_TY]]* [[VAR_IN1]], [[S_INT_TY]]** [[VAR_ADDR1]],
// load content of local address variables
// CHECK: [[T_VAR_ADDR_REF1:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_ADDR1]],
// CHECK: [[VEC_ADDR_REF1:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR1]],
// CHECK: [[S_ARR_ADDR_REF1:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR1]],
// CHECK: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_ADDR1]],
// CHECK-DAG: store [[S_INT_TY]]* [[VAR_ADDR1_REF]], [[S_INT_TY]]** [[TMP]],
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// CHECK-DAG: [[VAR_ADDR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP]],
// CHECK-DAG: store [[S_INT_TY]]* [[VAR_ADDR1_REF]], [[S_INT_TY]]** [[TMP1]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK-DAG: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[OMP_IS_LAST1]],
[OPENMP]Simplify codegen for the outlined regions. Simplified codegen for the outlined regions, excluding duplication code for handling variables with the reference types. llvm-svn: 361529
2019-05-24 02:19:54 +08:00
// CHECK-DAG: [[TMP_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP1]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: call void @__kmpc_for_static_init_4(
// assignment: vec[i] = t_var;
// CHECK: [[IV_VAL1:%.+]] =
// CHECK: [[T_VAR_PRIV_VAL1:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]],
// CHECK: [[VEC_PTR1:%.+]] = getelementptr inbounds [2 x i{{[0-9]+}}], [2 x i{{[0-9]+}}]* [[VEC_PRIV1]], i{{[0-9]+}} 0, i{{[0-9]+}} {{.+}}
// CHECK: store i{{[0-9]+}} [[T_VAR_PRIV_VAL1]], i{{[0-9]+}}* [[VEC_PTR1]],
// assignment: s_arr[i] = var;
// CHECK-DAG: [[S_ARR_PTR1:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]],
// CHECK-DAG: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]],
// CHECK-DAG: [[S_ARR_PTR_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_PTR1]] to i8*
// CHECK-DAG: [[TMP_VAL_BCAST1:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK-DAG: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_PTR_BCAST1]], i8* align {{[0-9]+}} [[TMP_VAL_BCAST1]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: call void @__kmpc_for_static_fini(
// lastprivates
// CHECK: [[OMP_IS_LAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[OMP_IS_LAST1]],
// CHECK: [[IS_LAST_IT:%.+]] = icmp ne i{{[0-9]+}} [[OMP_IS_LAST_VAL]], 0
// CHECK: br i1 [[IS_LAST_IT]], label %[[OMP_LASTPRIV_BLOCK:.+]], label %[[OMP_LASTPRIV_DONE:.+]]
// CHECK: [[OMP_LASTPRIV_BLOCK]]:
// CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_PRIV1]],
// CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_ADDR_REF1]],
// CHECK: [[BCAST_VEC_ADDR_REF:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_REF1]] to i8*
// CHECK: [[BCAST_VEC_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV1]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BCAST_VEC_ADDR_REF]], i8* align {{[0-9]+}} [[BCAST_VEC_PRIV]],
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_ADDR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[S_ARR_PRIV_BCAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV1]] to [[S_INT_TY]]*
// CHECK: [[S_ARR_BEGIN_GEP:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2
// CHECK: [[S_ARR_IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_BEGIN_GEP]]
// CHECK: br i1 [[S_ARR_IS_EMPTY]], label %[[S_ARR_COPY_DONE:.+]], label %[[S_ARR_COPY_BLOCK:.+]]
// CHECK: [[S_ARR_COPY_BLOCK]]:
// CHECK: [[S_ARR_SRC_EL:%.+]] = phi [[S_INT_TY]]*{{.+}}
// CHECK: [[S_ARR_DST_EL:%.+]] = phi [[S_INT_TY]]*{{.+}}
// CHECK: [[S_ARR_DST_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_DST_EL]] to i8*
// CHECK: [[S_ARR_SRC_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[S_ARR_SRC_EL]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[S_ARR_DST_BCAST]], i8* align {{[0-9]+}} [[S_ARR_SRC_BCAST]]{{.+}})
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: [[S_ARR_DST_NEXT:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_DST_EL]], i{{[0-9]+}} 1
// CHECK: [[S_ARR_SRC_NEXT:%.+]] = getelementptr{{.+}}
// CHECK: [[CPY_IS_FINISHED:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_DST_NEXT]], [[S_ARR_BEGIN_GEP]]
// CHECK: br i1 [[CPY_IS_FINISHED]], label %[[S_ARR_COPY_DONE]], label %[[S_ARR_COPY_BLOCK]]
// CHECK: [[S_ARR_COPY_DONE]]:
// CHECK: [[TMP_VAL1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TMP_PRIV1]],
// CHECK: [[VAR_ADDR_REF_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[TMP_REF]] to i8*
// CHECK: [[TMP_VAL1_BCAST:%.+]] = bitcast [[S_INT_TY]]* [[TMP_VAL1]] to i8*
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VAR_ADDR_REF_BCAST]], i8* align {{[0-9]+}} [[TMP_VAL1_BCAST]],{{.+}})
[OPENMP] Private, firstprivate, and lastprivate clauses for distribute, host code generation https://reviews.llvm.org/D17840 This patch enables private, firstprivate, and lastprivate clauses for the OpenMP distribute directive. Regression tests differ from the similar case of the same clauses on the for directive, by removing a reference to two global variables g and g1. This is necessary because: 1. a distribute pragma is only allowed inside a target region; 2. referring a global variable (e.g. g and g1) in a target region requires the program to enclose the variable in a "declare target" region; 3. declare target pragmas, which are used to define a declare target region, are currently unavailable in clang (patch being prepared). For this reason, I moved the global declarations into local variables. llvm-svn: 290898
2017-01-04 02:24:42 +08:00
// CHECK: ret void
#endif