[OpenMP] Avoid calling `isSPMDMode` during RT initialization
Until we hit the first barrier we should not call `mapping::isSPMDMode`
with all threads. Instead, we now have (and use during initialization) a
`mapping::isMainThreadInGenericMode` overload that takes the known
SPMD-mode state and one that queries it.
Reviewed By: tianshilei1992
Differential Revision: https://reviews.llvm.org/D111381
GitOrigin-RevId: 85ad56633593d011ab00f4eb431ffd879677bffc
diff --git a/libomptarget/DeviceRTL/include/Mapping.h b/libomptarget/DeviceRTL/include/Mapping.h
index b34ecf4..a3193f3 100644
--- a/libomptarget/DeviceRTL/include/Mapping.h
+++ b/libomptarget/DeviceRTL/include/Mapping.h
@@ -35,6 +35,7 @@
/// Return true if the executing thread is the main thread in generic mode.
bool isMainThreadInGenericMode();
+bool isMainThreadInGenericMode(bool IsSPMD);
/// Return true if the executing thread has the lowest Id of the active threads
/// in the warp.
diff --git a/libomptarget/DeviceRTL/src/Kernel.cpp b/libomptarget/DeviceRTL/src/Kernel.cpp
index ead6085..f834754 100644
--- a/libomptarget/DeviceRTL/src/Kernel.cpp
+++ b/libomptarget/DeviceRTL/src/Kernel.cpp
@@ -81,7 +81,7 @@
return -1;
}
- if (mapping::isMainThreadInGenericMode())
+ if (mapping::isMainThreadInGenericMode(IsSPMD))
return -1;
if (UseGenericStateMachine)
diff --git a/libomptarget/DeviceRTL/src/Mapping.cpp b/libomptarget/DeviceRTL/src/Mapping.cpp
index 96a612d..66089ea 100644
--- a/libomptarget/DeviceRTL/src/Mapping.cpp
+++ b/libomptarget/DeviceRTL/src/Mapping.cpp
@@ -165,8 +165,8 @@
} // namespace impl
} // namespace _OMP
-bool mapping::isMainThreadInGenericMode() {
- if (mapping::isSPMDMode() || icv::Level)
+bool mapping::isMainThreadInGenericMode(bool IsSPMD) {
+ if (IsSPMD || icv::Level)
return false;
// Check if this is the last warp in the block.
@@ -175,6 +175,10 @@
return mapping::getThreadIdInBlock() == MainTId;
}
+bool mapping::isMainThreadInGenericMode() {
+ return mapping::isMainThreadInGenericMode(mapping::isSPMDMode());
+}
+
bool mapping::isLeaderInWarp() {
__kmpc_impl_lanemask_t Active = mapping::activemask();
__kmpc_impl_lanemask_t LaneMaskLT = mapping::lanemaskLT();
diff --git a/libomptarget/DeviceRTL/src/Reduction.cpp b/libomptarget/DeviceRTL/src/Reduction.cpp
index cd56581..a06ac23 100644
--- a/libomptarget/DeviceRTL/src/Reduction.cpp
+++ b/libomptarget/DeviceRTL/src/Reduction.cpp
@@ -72,7 +72,7 @@
InterWarpCopyFnTy cpyFct,
bool isSPMDExecutionMode, bool) {
uint32_t BlockThreadId = mapping::getThreadIdInBlock();
- if (mapping::isMainThreadInGenericMode())
+ if (mapping::isMainThreadInGenericMode(/* IsSPMD */ false))
BlockThreadId = 0;
uint32_t NumThreads = omp_get_num_threads();
if (NumThreads == 1)