[PATCH] D8688: Update MergedLoadStoreMotion to use MemorySSA

Chad Rosier via llvm-commits llvm-commits at lists.llvm.org
Fri Jun 17 08:54:07 PDT 2016


mcrosier added inline comments.

================
Comment at: lib/Transforms/Scalar/MergedLoadStoreMotion.cpp:97
@@ -93,1 +96,3 @@
 #define DEBUG_TYPE "mldst-motion"
+static cl::opt<bool>
+    UseMemorySSA("use-memoryssa-mldst", cl::Hidden,
----------------
I agree we should have a flag per pass to simplify debugging.

================
Comment at: lib/Transforms/Scalar/MergedLoadStoreMotion.cpp:761
@@ +760,3 @@
+    LoadInfo.insert({Load0, CachingWalker->getClobberingMemoryAccess(Load0)});
+    ++NLoads;
+    if (NLoads * Size1 >= MagicCompileTimeControl)
----------------
Sorry if I wasn't clear.  I was suggesting we remove the pre-increment above and do it in the if statement.

  ++NLoads;
  if (NLoads * Size1 ...)

vs

  if (++NLoads * Size1 ...)



http://reviews.llvm.org/D8688





More information about the llvm-commits mailing list