[PATCH] D106154: [AbstractAttributor] Fold __kmpc_parallel_level if possible

[Johannes Doerfert via Phabricator via llvm-commits]

jdoerfert accepted this revision.
jdoerfert added a comment.
This revision is now accepted and ready to land.

LG, some minor comments below.



================
Comment at: llvm/lib/Transforms/IPO/OpenMPOpt.cpp:3733-3734
+        CallerKernelInfoAA.ParallelLevels.size() > 1)
+      SimplifiedValue = nullptr;
+    else {
+      const uint8_t Level = CallerKernelInfoAA.ParallelLevels[0];
----------------



================
Comment at: llvm/lib/Transforms/IPO/OpenMPOpt.cpp:3691
+        CallerKernelInfoAA.ParallelLevels.size() > 1)
+      SimplifiedValue = nullptr;
+    else {
----------------



================
Comment at: llvm/lib/Transforms/IPO/OpenMPOpt.cpp:3296
+          for (uint8_t L : CAA.ParallelLevels)
+            ParallelLevels.insert(L + 1);
+
----------------
tianshilei1992 wrote:
> The logic seems problematic here. Need to revise later.
mark `__kmpc_parallel_51` as noinline to make this sound.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D106154/new/

https://reviews.llvm.org/D106154