[llvm] r354058 - Canonicalize all integer "idempotent" atomicrmw ops

[Philip Reames via llvm-commits]

Author: reames
Date: Thu Feb 14 12:41:17 2019
New Revision: 354058

URL: http://llvm.org/viewvc/llvm-project?rev=354058&view=rev
Log:
Canonicalize all integer "idempotent" atomicrmw ops

For "idempotent" atomicrmw instructions which we can't simply turn into load, canonicalize the operation and constant. This reduces the matching needed elsewhere in the optimizer, but doesn't directly impact codegen.

For any architecture where OR/Zero is not a good default choice, you can extend the AtomicExpand lowerIdempotentRMWIntoFencedLoad mechanism. I reviewed X86 to make sure this works well, haven't audited other backends.

Differential Revision: https://reviews.llvm.org/D58244


Modified:
    llvm/trunk/lib/Transforms/InstCombine/InstCombineAtomicRMW.cpp
    llvm/trunk/test/Transforms/InstCombine/atomicrmw.ll

Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineAtomicRMW.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineAtomicRMW.cpp?rev=354058&r1=354057&r2=354058&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/InstCombine/InstCombineAtomicRMW.cpp (original)
+++ llvm/trunk/lib/Transforms/InstCombine/InstCombineAtomicRMW.cpp Thu Feb 14 12:41:17 2019
@@ -54,14 +54,22 @@ Instruction *InstCombiner::visitAtomicRM
   if (!isIdempotentRMW(RMWI))
     return nullptr;
 
-  // TODO: Canonicalize the operation for an idempotent operation we can't
-  // convert into a simple load.
-
-  // Volatile RMWs perform a load and a store, we cannot replace
-  // this by just a load.
+  // Volatile RMWs perform a load and a store, we cannot replace this by just a
+  // load. We chose not to canonicalize out of general paranoia about user
+  // expectations around volatile. 
   if (RMWI.isVolatile())
     return nullptr;
 
+  // We chose to canonicalize all idempotent operations to an single
+  // operation code and constant.  This makes it easier for the rest of the
+  // optimizer to match easily.  The choice of or w/zero is arbitrary.
+  if (RMWI.getType()->isIntegerTy() &&
+      RMWI.getOperation() != AtomicRMWInst::Or) {
+    RMWI.setOperation(AtomicRMWInst::Or);
+    RMWI.setOperand(1, ConstantInt::get(RMWI.getType(), 0));
+    return &RMWI;
+  }
+
   // Check if the required ordering is compatible with an atomic load.
   AtomicOrdering Ordering = RMWI.getOrdering();
   assert(Ordering != AtomicOrdering::NotAtomic &&

Modified: llvm/trunk/test/Transforms/InstCombine/atomicrmw.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstCombine/atomicrmw.ll?rev=354058&r1=354057&r2=354058&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/InstCombine/atomicrmw.ll (original)
+++ llvm/trunk/test/Transforms/InstCombine/atomicrmw.ll Thu Feb 14 12:41:17 2019
@@ -87,14 +87,13 @@ define i16 @atomic_syncscope(i16* %addr)
   ret i16 %res
 }
 
-; Don't transform seq_cst ordering.
 ; By eliminating the store part of the atomicrmw, we would get rid of the
-; release semantic, which is incorrect.
-; CHECK-LABEL: atomic_or_zero_seq_cst
+; release semantic, which is incorrect.  We can canonicalize the operation.
+; CHECK-LABEL: atomic_seq_cst
 ; CHECK-NEXT: %res = atomicrmw or i16* %addr, i16 0 seq_cst
 ; CHECK-NEXT: ret i16 %res
-define i16 @atomic_or_zero_seq_cst(i16* %addr) {
-  %res = atomicrmw or i16* %addr, i16 0 seq_cst
+define i16 @atomic_seq_cst(i16* %addr) {
+  %res = atomicrmw add i16* %addr, i16 0 seq_cst
   ret i16 %res
 }
 
@@ -117,21 +116,21 @@ define i16 @atomic_xor_zero(i16* %addr)
 }
 
 ; Check that the transformation does not apply when the ordering is
-; incompatible with a load (release).
-; CHECK-LABEL: atomic_or_zero_release
+; incompatible with a load (release).  Do canonicalize.
+; CHECK-LABEL: atomic_release
 ; CHECK-NEXT: %res = atomicrmw or i16* %addr, i16 0 release
 ; CHECK-NEXT: ret i16 %res
-define i16 @atomic_or_zero_release(i16* %addr) {
-  %res = atomicrmw or i16* %addr, i16 0 release
+define i16 @atomic_release(i16* %addr) {
+  %res = atomicrmw sub i16* %addr, i16 0 release
   ret i16 %res
 }
 
 ; Check that the transformation does not apply when the ordering is
-; incompatible with a load (acquire, release).
-; CHECK-LABEL: atomic_or_zero_acq_rel
+; incompatible with a load (acquire, release).  Do canonicalize.
+; CHECK-LABEL: atomic_acq_rel
 ; CHECK-NEXT: %res = atomicrmw or i16* %addr, i16 0 acq_rel
 ; CHECK-NEXT: ret i16 %res
-define i16 @atomic_or_zero_acq_rel(i16* %addr) {
-  %res = atomicrmw or i16* %addr, i16 0 acq_rel
+define i16 @atomic_acq_rel(i16* %addr) {
+  %res = atomicrmw xor i16* %addr, i16 0 acq_rel
   ret i16 %res
 }