[llvm] 94ffc75 - [MemCpyOpt] keep src/dest alloca ordering (#176012)

[via llvm-commits]

Author: Jameson Nash
Date: 2026-01-14T15:18:37-05:00
New Revision: 94ffc754d2e292e81e6f7baaef394d86974aea7c

URL: https://github.com/llvm/llvm-project/commit/94ffc754d2e292e81e6f7baaef394d86974aea7c
DIFF: https://github.com/llvm/llvm-project/commit/94ffc754d2e292e81e6f7baaef394d86974aea7c.diff

LOG: [MemCpyOpt] keep src/dest alloca ordering (#176012)

Rather than test dominator of every use, just check which of src or dest
is first, and use that insert location. This minimizes unnecessary
dominator queries while also helping to preserve the order of allocas
(for better code readability / diff).

Extracted from PR optimization improvement series at
https://github.com/llvm/llvm-project/pull/150792

Added: 
    

Modified: 
    llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
index f7e761550f211..8683576890081 100644
--- a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
+++ b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
@@ -1530,6 +1530,13 @@ bool MemCpyOptPass::performStackMoveOptzn(Instruction *Load, Instruction *Store,
   if (!SrcAlloca->isStaticAlloca() || !DestAlloca->isStaticAlloca())
     return false;
 
+  // Check if it will be legal to combine allocas without breaking dominator.
+  bool MoveSrc = !DT->dominates(SrcAlloca, DestAlloca);
+  if (MoveSrc) {
+    if (!DT->dominates(DestAlloca, SrcAlloca))
+      return false;
+  }
+
   // Check that src and dest are never captured, unescaped allocas. Also
   // find the nearest common dominator and postdominator for all users in
   // order to shrink wrap the lifetimes, and instructions with noalias metadata
@@ -1537,7 +1544,6 @@ bool MemCpyOptPass::performStackMoveOptzn(Instruction *Load, Instruction *Store,
 
   SmallVector<Instruction *, 4> LifetimeMarkers;
   SmallPtrSet<Instruction *, 4> AAMetadataInstrs;
-  bool SrcNotDom = false;
 
   auto CaptureTrackingWithModRef =
       [&](Instruction *AI, function_ref<bool(Instruction *)> ModRefCallback,
@@ -1551,10 +1557,6 @@ bool MemCpyOptPass::performStackMoveOptzn(Instruction *Load, Instruction *Store,
       Instruction *I = Worklist.pop_back_val();
       for (const Use &U : I->uses()) {
         auto *UI = cast<Instruction>(U.getUser());
-        // If any use that isn't dominated by SrcAlloca exists, we move src
-        // alloca to the entry before the transformation.
-        if (!DT->dominates(SrcAlloca, UI))
-          SrcNotDom = true;
 
         if (Visited.size() >= MaxUsesToExplore) {
           LLVM_DEBUG(
@@ -1672,11 +1674,10 @@ bool MemCpyOptPass::performStackMoveOptzn(Instruction *Load, Instruction *Store,
   if (DestAddressCaptured && SrcAddressCaptured)
     return false;
 
-  // We can do the transformation. First, move the SrcAlloca to the start of the
-  // BB.
-  if (SrcNotDom)
-    SrcAlloca->moveBefore(*SrcAlloca->getParent(),
-                          SrcAlloca->getParent()->getFirstInsertionPt());
+  // We can now do the transformation. First move the Src if it was after Dest.
+  if (MoveSrc)
+    SrcAlloca->moveBefore(DestAlloca->getIterator());
+
   // Align the allocas appropriately.
   SrcAlloca->setAlignment(
       std::max(SrcAlloca->getAlign(), DestAlloca->getAlign()));