[llvm] ffcaeca - [CodeGen] Fix partial phi input removal in TailDuplicator. (#158265)

[via llvm-commits]

Author: Afanasyev Ivan
Date: 2025-09-13T10:45:54+09:00
New Revision: ffcaeca90a3c0965acace6645f775ab1d876fa6e

URL: https://github.com/llvm/llvm-project/commit/ffcaeca90a3c0965acace6645f775ab1d876fa6e
DIFF: https://github.com/llvm/llvm-project/commit/ffcaeca90a3c0965acace6645f775ab1d876fa6e.diff

LOG: [CodeGen] Fix partial phi input removal in TailDuplicator. (#158265)

Tail duplicator removes the first PHI income from the predecessor basic
block, while it should remove all operands for this block.

PHI instructions happen to have duplicated values for the same
predecessor block:
* `UnreachableMachineBlockElim` assumes that PHI instruction might have
duplicates:
https://github.com/llvm/llvm-project/blob/7289f2cd0c371b2539faa628ec0eea58fa61892c/llvm/lib/CodeGen/UnreachableBlockElim.cpp#L160
* `AArch64` directly states that PHI instruction might have duplicates:
https://github.com/llvm/llvm-project/blob/7289f2cd0c371b2539faa628ec0eea58fa61892c/llvm/lib/Target/AArch64/AArch64ConditionalCompares.cpp#L244
* And `Hexagon`:
https://github.com/llvm/llvm-project/blob/7289f2cd0c371b2539faa628ec0eea58fa61892c/llvm/lib/Target/Hexagon/HexagonConstPropagation.cpp#L844

We have caught the bug on custom out-of-tree backend. `TailDuplicator`
should remove all operands corresponding to the removing block.

Please note, that bug likely does not affect in-tree backends, because:
* It happens only in scenario of **partial** tail duplication (i.e. tail
block is duplicated in some predecessors, but not in all of them)
* It happens in **Pre-RA** tail duplication only (Post-RA does not
contain PHIs, obviously)
* The only backend (I know) uses Pre-RA tail duplication is X86. It uses
tail duplication via `early-tailduplication` pass which declines partial
tail duplication via `canCompletelyDuplicateBB` check, because it uses
`TailDuplicator::tailDuplicateBlocks` public API.

So, bug happens only in the case of pre-ra partial tail duplication if
backend uses `TailDuplicator::tailDuplicate` public API directly.

That's why I can not add reproducer test for in-tree backends.

Added: 
    

Modified: 
    llvm/lib/CodeGen/TailDuplicator.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/CodeGen/TailDuplicator.cpp b/llvm/lib/CodeGen/TailDuplicator.cpp
index 5d720fbbf1c61..9b1420a94142d 100644
--- a/llvm/lib/CodeGen/TailDuplicator.cpp
+++ b/llvm/lib/CodeGen/TailDuplicator.cpp
@@ -375,9 +375,14 @@ void TailDuplicator::processPHI(
   if (!Remove)
     return;
 
-  // Remove PredBB from the PHI node.
-  MI->removeOperand(SrcOpIdx + 1);
-  MI->removeOperand(SrcOpIdx);
+  // MI might have multiple entries for PredBB. Need to remove them all.
+  for (unsigned N = MI->getNumOperands(); N > 2; N -= 2) {
+    if (MI->getOperand(N - 1).getMBB() == PredBB) {
+      MI->removeOperand(N - 1);
+      MI->removeOperand(N - 2);
+    }
+  }
+
   if (MI->getNumOperands() == 1 && !TailBB->hasAddressTaken())
     MI->eraseFromParent();
   else if (MI->getNumOperands() == 1)