[PowerPC] Change CTR clobber estimation for 128-bit floating types

Reviewed By: shchenz

Differential Revision: https://reviews.llvm.org/D117459

2 files changed, 58 insertions, 4 deletions

diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
index ed28731b8ef2..707c1396e572 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -653,11 +653,17 @@ bool PPCTTIImpl::mightUseCTR(BasicBlock *BB, TargetLibraryInfo *LibInfo,
       }
 
       return true;
-    } else if (isa<BinaryOperator>(J) &&
-               (J->getType()->getScalarType()->isFP128Ty() ||
+    } else if ((J->getType()->getScalarType()->isFP128Ty() ||
                 J->getType()->getScalarType()->isPPC_FP128Ty())) {
       // Most operations on f128 or ppc_f128 values become calls.
       return true;
+    } else if (isa<FCmpInst>(J) &&
+               J->getOperand(0)->getType()->getScalarType()->isFP128Ty()) {
+      return true;
+    } else if ((isa<FPTruncInst>(J) || isa<FPExtInst>(J)) &&
+               (cast<CastInst>(J)->getSrcTy()->getScalarType()->isFP128Ty() ||
+                cast<CastInst>(J)->getDestTy()->getScalarType()->isFP128Ty())) {
+      return true;
     } else if (isa<UIToFPInst>(J) || isa<SIToFPInst>(J) ||
                isa<FPToUIInst>(J) || isa<FPToSIInst>(J)) {
       CastInst *CI = cast<CastInst>(J);
diff --git a/llvm/test/CodeGen/PowerPC/ctrloop-fp128.ll b/llvm/test/CodeGen/PowerPC/ctrloop-fp128.ll
index 57095413cdb2..fde8e20212c0 100644
--- a/llvm/test/CodeGen/PowerPC/ctrloop-fp128.ll
+++ b/llvm/test/CodeGen/PowerPC/ctrloop-fp128.ll
@@ -1,5 +1,7 @@
-; RUN: llc -verify-machineinstrs -stop-after=hardware-loops -mcpu=pwr9 \
-; RUN:   -mtriple=powerpc64le-unknown-unknown < %s | FileCheck %s
+; RUN: llc < %s -verify-machineinstrs -stop-after=hardware-loops -mcpu=pwr9 \
+; RUN:   -mtriple=powerpc64le-unknown-unknown | FileCheck %s
+; RUN: llc < %s -verify-machineinstrs -stop-after=hardware-loops -mcpu=pwr8 \
+; RUN:   -mtriple=powerpc64le-unknown-unknown | FileCheck %s
 
 @a = internal global fp128 0xL00000000000000000000000000000000, align 16
 @x = internal global [4 x fp128] zeroinitializer, align 16
@@ -29,4 +31,50 @@ for.end:                                          ; preds = %for.body
 ; CHECK-NOT:     call i1 @llvm.loop.decrement.i64(i64 1)
 }
 
+define void @fpext_ctrloop_fp128(double* %a) {
+entry:
+  br label %for.body
+
+for.body:
+  %i.06 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+  %arrayidx = getelementptr inbounds double, double* %a, i64 %i.06
+  %0 = load double, double* %arrayidx, align 8
+  %ext = fpext double %0 to fp128
+  %arrayidx1 = getelementptr inbounds [4 x fp128], [4 x fp128]* @y, i64 0, i64 %i.06
+  store fp128 %ext, fp128* %arrayidx1, align 16
+  %inc = add nuw nsw i64 %i.06, 1
+  %exitcond = icmp eq i64 %inc, 4
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:
+  ret void
+
+; CHECK-LABEL: fpext_ctrloop_fp128
+; CHECK-NOT: call void @llvm.set.loop.iterations.i64(i64 4)
+; CHECK-NOT: call i1 @llvm.loop.decrement.i64(i64 1)
+}
+
+define void @fptrunc_ctrloop_fp128(double* %a) {
+entry:
+  br label %for.body
+
+for.body:
+  %i.06 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
+  %arrayidx = getelementptr inbounds [4 x fp128], [4 x fp128]* @x, i64 0, i64 %i.06
+  %0 = load fp128, fp128* %arrayidx, align 16
+  %trunc = fptrunc fp128 %0 to double
+  %arrayidx1 = getelementptr inbounds double, double* %a, i64 %i.06
+  store double %trunc, double* %arrayidx1, align 16
+  %inc = add nuw nsw i64 %i.06, 1
+  %exitcond = icmp eq i64 %inc, 4
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:
+  ret void
+
+; CHECK-LABEL: fptrunc_ctrloop_fp128
+; CHECK-NOT: call void @llvm.set.loop.iterations.i64(i64 4)
+; CHECK-NOT: call i1 @llvm.loop.decrement.i64(i64 1)
+}
+
 declare void @obfuscate(i8*, ...) local_unnamed_addr #2