diff --git a/libomptarget/DeviceRTL/include/Synchronization.h b/libomptarget/DeviceRTL/include/Synchronization.h
index 4b8898f..e33f37a 100644
--- a/libomptarget/DeviceRTL/include/Synchronization.h
+++ b/libomptarget/DeviceRTL/include/Synchronization.h
@@ -29,13 +29,15 @@
 
 /// Synchronizing threads is allowed even if they all hit different instances of
 /// `synchronize::threads()`. However, `synchronize::threadsAligned()` is more
-/// restrictive in that it requires all threads to hit the same instance.
+/// restrictive in that it requires all threads to hit the same instance. The
+/// noinline is removed by the openmp-opt pass and helps to preserve the
+/// information till then.
 ///{
 #pragma omp begin assumes ext_aligned_barrier
 
 /// Synchronize all threads in a block, they are are reaching the same
 /// instruction (hence all threads in the block are "aligned").
-void threadsAligned();
+__attribute__((noinline)) void threadsAligned();
 
 #pragma omp end assumes
 ///}
diff --git a/libomptarget/DeviceRTL/src/Mapping.cpp b/libomptarget/DeviceRTL/src/Mapping.cpp
index b161c55..172bbbf 100644
--- a/libomptarget/DeviceRTL/src/Mapping.cpp
+++ b/libomptarget/DeviceRTL/src/Mapping.cpp
@@ -289,17 +289,17 @@
 ///}
 
 extern "C" {
-uint32_t __kmpc_get_hardware_thread_id_in_block() {
+__attribute__((noinline)) uint32_t __kmpc_get_hardware_thread_id_in_block() {
   FunctionTracingRAII();
   return mapping::getThreadIdInBlock();
 }
 
-uint32_t __kmpc_get_hardware_num_threads_in_block() {
+__attribute__((noinline)) uint32_t __kmpc_get_hardware_num_threads_in_block() {
   FunctionTracingRAII();
   return impl::getNumHardwareThreadsInBlock();
 }
 
-uint32_t __kmpc_get_warp_size() {
+__attribute__((noinline)) uint32_t __kmpc_get_warp_size() {
   FunctionTracingRAII();
   return impl::getWarpSize();
 }
diff --git a/libomptarget/DeviceRTL/src/Parallelism.cpp b/libomptarget/DeviceRTL/src/Parallelism.cpp
index 5b133b0..27d1ff2 100644
--- a/libomptarget/DeviceRTL/src/Parallelism.cpp
+++ b/libomptarget/DeviceRTL/src/Parallelism.cpp
@@ -243,7 +243,8 @@
     __kmpc_end_sharing_variables();
 }
 
-bool __kmpc_kernel_parallel(ParallelRegionFnTy *WorkFn) {
+__attribute__((noinline)) bool
+__kmpc_kernel_parallel(ParallelRegionFnTy *WorkFn) {
   FunctionTracingRAII();
   // Work function and arguments for L1 parallel region.
   *WorkFn = state::ParallelRegionFn;
@@ -258,7 +259,7 @@
   return ThreadIsActive;
 }
 
-void __kmpc_kernel_end_parallel() {
+__attribute__((noinline)) void __kmpc_kernel_end_parallel() {
   FunctionTracingRAII();
   // In case we have modified an ICV for this thread before a ThreadState was
   // created. We drop it now to not contaminate the next parallel region.
diff --git a/libomptarget/DeviceRTL/src/State.cpp b/libomptarget/DeviceRTL/src/State.cpp
index 92847f7..7a73330 100644
--- a/libomptarget/DeviceRTL/src/State.cpp
+++ b/libomptarget/DeviceRTL/src/State.cpp
@@ -393,12 +393,12 @@
 }
 
 extern "C" {
-void *__kmpc_alloc_shared(uint64_t Bytes) {
+__attribute__((noinline)) void *__kmpc_alloc_shared(uint64_t Bytes) {
   FunctionTracingRAII();
   return memory::allocShared(Bytes, "Frontend alloc shared");
 }
 
-void __kmpc_free_shared(void *Ptr, uint64_t Bytes) {
+__attribute__((noinline)) void __kmpc_free_shared(void *Ptr, uint64_t Bytes) {
   FunctionTracingRAII();
   memory::freeShared(Ptr, Bytes, "Frontend free shared");
 }
diff --git a/libomptarget/DeviceRTL/src/Synchronization.cpp b/libomptarget/DeviceRTL/src/Synchronization.cpp
index 350da0b..4327871 100644
--- a/libomptarget/DeviceRTL/src/Synchronization.cpp
+++ b/libomptarget/DeviceRTL/src/Synchronization.cpp
@@ -358,12 +358,14 @@
   impl::namedBarrier();
 }
 
-void __kmpc_barrier_simple_spmd(IdentTy *Loc, int32_t TId) {
+__attribute__((noinline)) void __kmpc_barrier_simple_spmd(IdentTy *Loc,
+                                                          int32_t TId) {
   FunctionTracingRAII();
   synchronize::threadsAligned();
 }
 
-void __kmpc_barrier_simple_generic(IdentTy *Loc, int32_t TId) {
+__attribute__((noinline)) void __kmpc_barrier_simple_generic(IdentTy *Loc,
+                                                             int32_t TId) {
   FunctionTracingRAII();
   synchronize::threads();
 }