From 0882fd6c4f90f1cbaa4bb6f6ceec289428cca734 Mon Sep 17 00:00:00 2001
From: Hal Finkel <hfinkel@anl.gov>
Date: Fri, 29 Mar 2013 19:41:55 +0000
Subject: Implement FRINT lowering on PPC using frin

Like nearbyint, rint can be implemented on PPC using the frin instruction. The
complication comes from the fact that rint needs to set the FE_INEXACT flag
when the result does not equal the input value (and frin does not do that). As
a result, we use a custom inserter which, after the rounding, compares the
rounded value with the original, and if they differ, explicitly sets the XX bit
in the FPSCR register (which corresponds to FE_INEXACT).

Once LLVM has better modeling of the floating-point environment we should be
able to (often) eliminate this extra complexity.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@178362 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/PowerPC/PPCISelLowering.cpp | 49 ++++++++++++++++++++++++++++++++++
 lib/Target/PowerPC/PPCInstrInfo.td     |  9 +++++++
 2 files changed, 58 insertions(+)

(limited to 'lib/Target')

diff --git a/lib/Target/PowerPC/PPCISelLowering.cpp b/lib/Target/PowerPC/PPCISelLowering.cpp
index aadcb45359..0574f047c2 100644
--- a/lib/Target/PowerPC/PPCISelLowering.cpp
+++ b/lib/Target/PowerPC/PPCISelLowering.cpp
@@ -172,6 +172,10 @@ PPCTargetLowering::PPCTargetLowering(PPCTargetMachine &TM)
     if (TM.Options.UnsafeFPMath) {
       setOperationAction(ISD::FNEARBYINT, MVT::f64, Legal);
       setOperationAction(ISD::FNEARBYINT, MVT::f32, Legal);
+
+      // These need to set FE_INEXACT, and use a custom inserter.
+      setOperationAction(ISD::FRINT, MVT::f64, Legal);
+      setOperationAction(ISD::FRINT, MVT::f32, Legal);
     }
   }
 
@@ -6524,6 +6528,51 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
     // Restore FPSCR value.
     BuildMI(*BB, MI, dl, TII->get(PPC::MTFSF)).addImm(1).addReg(MFFSReg);
+  } else if (MI->getOpcode() == PPC::FRINDrint ||
+             MI->getOpcode() == PPC::FRINSrint) {
+    bool isf32 = MI->getOpcode() == PPC::FRINSrint;
+    unsigned Dest = MI->getOperand(0).getReg();
+    unsigned Src = MI->getOperand(1).getReg();
+    DebugLoc dl   = MI->getDebugLoc();
+
+    MachineRegisterInfo &RegInfo = F->getRegInfo();
+    unsigned CRReg = RegInfo.createVirtualRegister(&PPC::CRRCRegClass);
+
+    // Perform the rounding.
+    BuildMI(*BB, MI, dl, TII->get(isf32 ? PPC::FRINS : PPC::FRIND), Dest)
+      .addReg(Src);
+
+    // Compare the results.
+    BuildMI(*BB, MI, dl, TII->get(isf32 ? PPC::FCMPUS : PPC::FCMPUD), CRReg)
+      .addReg(Dest).addReg(Src);
+
+    // If the results were not equal, then set the FPSCR XX bit.
+    MachineBasicBlock *midMBB = F->CreateMachineBasicBlock(LLVM_BB);
+    MachineBasicBlock *exitMBB = F->CreateMachineBasicBlock(LLVM_BB);
+    F->insert(It, midMBB);
+    F->insert(It, exitMBB);
+    exitMBB->splice(exitMBB->begin(), BB,
+                    llvm::next(MachineBasicBlock::iterator(MI)),
+                    BB->end());
+    exitMBB->transferSuccessorsAndUpdatePHIs(BB);
+
+    BuildMI(*BB, MI, dl, TII->get(PPC::BCC))
+      .addImm(PPC::PRED_EQ).addReg(CRReg).addMBB(exitMBB);
+
+    BB->addSuccessor(midMBB);
+    BB->addSuccessor(exitMBB);
+
+    BB = midMBB;
+
+    // Set the FPSCR XX bit (FE_INEXACT). Note that we cannot just set
+    // the FI bit here because that will not automatically set XX also,
+    // and XX is what libm interprets as the FE_INEXACT flag.
+    BuildMI(BB, dl, TII->get(PPC::MTFSB1)).addImm(/* 38 - 32 = */ 6);
+    BuildMI(BB, dl, TII->get(PPC::B)).addMBB(exitMBB);
+
+    BB->addSuccessor(exitMBB);
+
+    BB = exitMBB;
   } else {
     llvm_unreachable("Unexpected instr type to insert");
   }
diff --git a/lib/Target/PowerPC/PPCInstrInfo.td b/lib/Target/PowerPC/PPCInstrInfo.td
index af3b57f2ed..f897a47800 100644
--- a/lib/Target/PowerPC/PPCInstrInfo.td
+++ b/lib/Target/PowerPC/PPCInstrInfo.td
@@ -1141,6 +1141,15 @@ let Uses = [RM] in {
                         "frin $frD, $frB", FPGeneral,
                         [(set f32:$frD, (fnearbyint f32:$frB))]>;
 
+  // These pseudos expand to rint but also set FE_INEXACT when the result does
+  // not equal the argument.
+  let usesCustomInserter = 1, Defs = [RM] in { // FIXME: Model FPSCR!
+    def FRINDrint : Pseudo<(outs F8RC:$frD), (ins F8RC:$frB),
+                            "#FRINDrint", [(set f64:$frD, (frint f64:$frB))]>;
+    def FRINSrint : Pseudo<(outs F4RC:$frD), (ins F4RC:$frB),
+                            "#FRINSrint", [(set f32:$frD, (frint f32:$frB))]>;
+  }
+
   def FRIPD  : XForm_26<63, 456, (outs F8RC:$frD), (ins F8RC:$frB),
                         "frip $frD, $frB", FPGeneral,
                         [(set f64:$frD, (fceil f64:$frB))]>;
-- 
cgit v1.2.3-18-g5258