[llvm] r330667 - Reland r301880(!): "[InstSimplify] Handle selects of GEPs with 0 offset"

[George Burgess IV via llvm-commits]

Author: gbiv
Date: Mon Apr 23 17:25:01 2018
New Revision: 330667

URL: http://llvm.org/viewvc/llvm-project?rev=330667&view=rev
Log:
Reland r301880(!): "[InstSimplify] Handle selects of GEPs with 0 offset"

I was reminded today that this patch got reverted in r301885. I can no
longer reproduce the failure that caused the revert locally (...almost
one year later), and the patch applied pretty cleanly, so I guess we'll
see if the bots still get angry about it.

The original breakage was InstSimplify complaining (in "assertion
failed" form) about getting passed some crazy IR when running `ninja
check-sanitizer`. I'm unable to find traces of what, exactly, said crazy
IR was. I suppose we'll find out pretty soon if that's still the case.
:)

Original commit:

  Author: gbiv
  Date: Mon May  1 18:12:08 2017
  New Revision: 301880

  URL: http://llvm.org/viewvc/llvm-project?rev=301880&view=rev
  Log:
  [InstSimplify] Handle selects of GEPs with 0 offset

  In particular (since it wouldn't fit nicely in the summary):
  (select (icmp eq V 0) P (getelementptr P V)) -> (getelementptr P V)

  Differential Revision: https://reviews.llvm.org/D31435


Modified:
    llvm/trunk/lib/Analysis/InstructionSimplify.cpp
    llvm/trunk/test/Transforms/InstSimplify/select.ll

Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=330667&r1=330666&r2=330667&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
+++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Mon Apr 23 17:25:01 2018
@@ -62,6 +62,8 @@ static Value *SimplifyOrInst(Value *, Va
 static Value *SimplifyXorInst(Value *, Value *, const SimplifyQuery &, unsigned);
 static Value *SimplifyCastInst(unsigned, Value *, Type *,
                                const SimplifyQuery &, unsigned);
+static Value *SimplifyGEPInst(Type *, ArrayRef<Value *>, const SimplifyQuery &,
+                              unsigned);
 
 /// For a boolean type or a vector of boolean type, return false or a vector
 /// with every element false.
@@ -3528,6 +3530,17 @@ static const Value *SimplifyWithOpReplac
     }
   }
 
+  // Same for GEPs.
+  if (auto *GEP = dyn_cast<GetElementPtrInst>(I)) {
+    if (MaxRecurse) {
+      SmallVector<Value *, 8> NewOps(GEP->getNumOperands());
+      transform(GEP->operands(), NewOps.begin(),
+                [&](Value *V) { return V == Op ? RepOp : V; });
+      return SimplifyGEPInst(GEP->getSourceElementType(), NewOps, Q,
+                             MaxRecurse - 1);
+    }
+  }
+
   // TODO: We could hand off more cases to instsimplify here.
 
   // If all operands are constant after substituting Op for RepOp then we can

Modified: llvm/trunk/test/Transforms/InstSimplify/select.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstSimplify/select.ll?rev=330667&r1=330666&r2=330667&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/InstSimplify/select.ll (original)
+++ llvm/trunk/test/Transforms/InstSimplify/select.ll Mon Apr 23 17:25:01 2018
@@ -545,3 +545,22 @@ define i8 @do_not_assume_sel_cond(i1 %co
   ret i8 %sel
 }
 
+define i32* @select_icmp_eq_0_gep_operand(i32* %base, i64 %n) {
+; CHECK-LABEL: @select_icmp_eq_0_gep_operand(
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr
+; CHECK-NEXT: ret i32* [[GEP]]
+  %cond = icmp eq i64 %n, 0
+  %gep = getelementptr i32, i32* %base, i64 %n
+  %r = select i1 %cond, i32* %base, i32* %gep
+  ret i32* %r
+}
+
+define i32* @select_icmp_ne_0_gep_operand(i32* %base, i64 %n) {
+; CHECK-LABEL: @select_icmp_ne_0_gep_operand(
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr
+; CHECK-NEXT: ret i32* [[GEP]]
+  %cond = icmp ne i64 %n, 0
+  %gep = getelementptr i32, i32* %base, i64 %n
+  %r = select i1 %cond, i32* %gep, i32* %base
+  ret i32* %r
+}