Merge pull request #66 from dylanmckay/avr-pick-upstream-llvm-fixes

[AVR] Cherry-pick 17 upstream AVR backend fixes into the Rust LLVM fork

Author: nikic

llvm/lib/Target/AVR/AVRCallingConv.td

@@ -6,21 +6,13 @@
 //
 //===----------------------------------------------------------------------===//
 // This describes the calling conventions for AVR architecture.
+// Normal functions use a special calling convention, solved in code.
 //===----------------------------------------------------------------------===//
 
 //===----------------------------------------------------------------------===//
 // AVR Return Value Calling Convention
 //===----------------------------------------------------------------------===//
 
-def RetCC_AVR : CallingConv
-<[
-  // i8 is returned in R24.
-  CCIfType<[i8], CCAssignToReg<[R24]>>,
-
-  // i16 are returned in R25:R24, R23:R22, R21:R20 and R19:R18.
-  CCIfType<[i16], CCAssignToReg<[R25R24, R23R22, R21R20, R19R18]>>
-]>;
-
 // Special return value calling convention for runtime functions.
 def RetCC_AVR_BUILTIN : CallingConv
 <[
@@ -41,14 +33,6 @@ def ArgCC_AVR_Vararg : CallingConv
   CCAssignToStack<2, 1>
 ]>;
 
-// Special argument calling convention for
-// division runtime functions.
-def ArgCC_AVR_BUILTIN_DIV : CallingConv
-<[
-  CCIfType<[i8], CCAssignToReg<[R24,R22]>>,
-  CCIfType<[i16], CCAssignToReg<[R25R24, R23R22]>>
-]>;
-
 //===----------------------------------------------------------------------===//
 // Callee-saved register lists.
 //===----------------------------------------------------------------------===//

llvm/lib/Target/AVR/AVRDevices.td

@@ -121,6 +121,11 @@ def FeatureTinyEncoding   : SubtargetFeature<"tinyencoding",
                                   "The device has Tiny core specific "
                                   "instruction encodings">;
 
+// The device has CPU registers mapped in data address space
+def FeatureMMR : SubtargetFeature<"memmappedregs", "m_hasMemMappedGPR",
+                                  "true", "The device has CPU registers "
+                                  "mapped in data address space">;
+
 class ELFArch<string name>  : SubtargetFeature<"", "ELFArch",
                                     !strconcat("ELF::",name), "">;
 
@@ -152,7 +157,7 @@ def ELFArchXMEGA7  : ELFArch<"EF_AVR_ARCH_XMEGA7">;
 // device should have.
 def FamilyAVR0           : Family<"avr0", []>;
 
-def FamilyAVR1           : Family<"avr1", [FamilyAVR0, FeatureLPM]>;
+def FamilyAVR1           : Family<"avr1", [FamilyAVR0, FeatureLPM, FeatureMMR]>;
 
 def FamilyAVR2           : Family<"avr2",
                                  [FamilyAVR1, FeatureIJMPCALL, FeatureADDSUBIW,
@@ -190,11 +195,14 @@ def FamilyAVR6           : Family<"avr6",
 
 def FamilyTiny           : Family<"avrtiny",
                                  [FamilyAVR0, FeatureBREAK, FeatureSRAM,
-                                  FeatureTinyEncoding]>;
+                                  FeatureTinyEncoding, FeatureMMR]>;
 
 def FamilyXMEGA          : Family<"xmega",
-                                 [FamilyAVR51, FeatureEIJMPCALL, FeatureSPMX,
-                                  FeatureDES]>;
+                                 [FamilyAVR0, FeatureLPM, FeatureIJMPCALL, FeatureADDSUBIW,
+                                  FeatureSRAM, FeatureJMPCALL, FeatureMultiplication,
+                                  FeatureMOVW, FeatureLPMX, FeatureSPM,
+                                  FeatureBREAK, FeatureEIJMPCALL, FeatureSPMX,
+                                  FeatureDES, FeatureELPM, FeatureELPMX]>;
 
 def FamilyXMEGAU         : Family<"xmegau",
                                   [FamilyXMEGA, FeatureRMW]>;
@@ -208,7 +216,7 @@ def FeatureSetSpecial    : FeatureSet<"special",
                                        FeatureLPM, FeatureLPMX, FeatureELPM,
                                        FeatureELPMX, FeatureSPM, FeatureSPMX,
                                        FeatureDES, FeatureRMW,
-                                       FeatureMultiplication, FeatureBREAK]>;
+                                       FeatureMultiplication, FeatureBREAK, FeatureMMR]>;
 
 //===---------------------------------------------------------------------===//
 // AVR microcontrollers supported.

llvm/lib/Target/AVR/AVRExpandPseudoInsts.cpp

@@ -597,8 +597,8 @@ bool AVRExpandPseudo::expand<AVR::LDWRdPtr>(Block &MBB, BlockIt MBBI) {
 
   // Load low byte.
   auto MIBLO = buildMI(MBB, MBBI, OpLo)
-    .addReg(CurDstLoReg, RegState::Define)
-    .addReg(SrcReg, RegState::Define);
+                   .addReg(CurDstLoReg, RegState::Define)
+                   .addReg(SrcReg);
 
   // Push low byte onto stack if necessary.
   if (TmpReg)

llvm/lib/Target/AVR/AVRFrameLowering.cpp

@@ -57,26 +57,19 @@ void AVRFrameLowering::emitPrologue(MachineFunction &MF,
   DebugLoc DL = (MBBI != MBB.end()) ? MBBI->getDebugLoc() : DebugLoc();
   const AVRSubtarget &STI = MF.getSubtarget<AVRSubtarget>();
   const AVRInstrInfo &TII = *STI.getInstrInfo();
+  const AVRMachineFunctionInfo *AFI = MF.getInfo<AVRMachineFunctionInfo>();
   bool HasFP = hasFP(MF);
 
   // Interrupt handlers re-enable interrupts in function entry.
-  if (CallConv == CallingConv::AVR_INTR) {
+  if (AFI->isInterruptHandler()) {
     BuildMI(MBB, MBBI, DL, TII.get(AVR::BSETs))
         .addImm(0x07)
         .setMIFlag(MachineInstr::FrameSetup);
   }
 
-  // Save the frame pointer if we have one.
-  if (HasFP) {
-    BuildMI(MBB, MBBI, DL, TII.get(AVR::PUSHWRr))
-        .addReg(AVR::R29R28, RegState::Kill)
-        .setMIFlag(MachineInstr::FrameSetup);
-  }
-
   // Emit special prologue code to save R1, R0 and SREG in interrupt/signal
   // handlers before saving any other registers.
-  if (CallConv == CallingConv::AVR_INTR ||
-      CallConv == CallingConv::AVR_SIGNAL) {
+  if (AFI->isInterruptOrSignalHandler()) {
     BuildMI(MBB, MBBI, DL, TII.get(AVR::PUSHWRr))
         .addReg(AVR::R1R0, RegState::Kill)
         .setMIFlag(MachineInstr::FrameSetup);
@@ -100,7 +93,6 @@ void AVRFrameLowering::emitPrologue(MachineFunction &MF,
   }
 
   const MachineFrameInfo &MFI = MF.getFrameInfo();
-  const AVRMachineFunctionInfo *AFI = MF.getInfo<AVRMachineFunctionInfo>();
   unsigned FrameSize = MFI.getStackSize() - AFI->getCalleeSavedFrameSize();
 
   // Skip the callee-saved push instructions.
@@ -143,13 +135,11 @@ void AVRFrameLowering::emitPrologue(MachineFunction &MF,
 
 void AVRFrameLowering::emitEpilogue(MachineFunction &MF,
                                     MachineBasicBlock &MBB) const {
-  CallingConv::ID CallConv = MF.getFunction().getCallingConv();
-  bool isHandler = (CallConv == CallingConv::AVR_INTR ||
-                    CallConv == CallingConv::AVR_SIGNAL);
+  const AVRMachineFunctionInfo *AFI = MF.getInfo<AVRMachineFunctionInfo>();
 
   // Early exit if the frame pointer is not needed in this function except for
   // signal/interrupt handlers where special code generation is required.
-  if (!hasFP(MF) && !isHandler) {
+  if (!hasFP(MF) && !AFI->isInterruptOrSignalHandler()) {
     return;
   }
 
@@ -159,24 +149,20 @@ void AVRFrameLowering::emitEpilogue(MachineFunction &MF,
 
   DebugLoc DL = MBBI->getDebugLoc();
   const MachineFrameInfo &MFI = MF.getFrameInfo();
-  const AVRMachineFunctionInfo *AFI = MF.getInfo<AVRMachineFunctionInfo>();
   unsigned FrameSize = MFI.getStackSize() - AFI->getCalleeSavedFrameSize();
   const AVRSubtarget &STI = MF.getSubtarget<AVRSubtarget>();
   const AVRInstrInfo &TII = *STI.getInstrInfo();
 
   // Emit special epilogue code to restore R1, R0 and SREG in interrupt/signal
   // handlers at the very end of the function, just before reti.
-  if (isHandler) {
+  if (AFI->isInterruptOrSignalHandler()) {
     BuildMI(MBB, MBBI, DL, TII.get(AVR::POPRd), AVR::R0);
     BuildMI(MBB, MBBI, DL, TII.get(AVR::OUTARr))
         .addImm(0x3f)
         .addReg(AVR::R0, RegState::Kill);
     BuildMI(MBB, MBBI, DL, TII.get(AVR::POPWRd), AVR::R1R0);
   }
 
-  if (hasFP(MF))
-    BuildMI(MBB, MBBI, DL, TII.get(AVR::POPWRd), AVR::R29R28);
-
   // Early exit if there is no need to restore the frame pointer.
   if (!FrameSize) {
     return;
@@ -299,15 +285,10 @@ bool AVRFrameLowering::restoreCalleeSavedRegisters(
 }
 
 /// Replace pseudo store instructions that pass arguments through the stack with
-/// real instructions. If insertPushes is true then all instructions are
-/// replaced with push instructions, otherwise regular std instructions are
-/// inserted.
+/// real instructions.
 static void fixStackStores(MachineBasicBlock &MBB,
                            MachineBasicBlock::iterator MI,
-                           const TargetInstrInfo &TII, bool insertPushes) {
-  const AVRSubtarget &STI = MBB.getParent()->getSubtarget<AVRSubtarget>();
-  const TargetRegisterInfo &TRI = *STI.getRegisterInfo();
-
+                           const TargetInstrInfo &TII, Register FP) {
   // Iterate through the BB until we hit a call instruction or we reach the end.
   for (auto I = MI, E = MBB.end(); I != E && !I->isCall();) {
     MachineBasicBlock::iterator NextMI = std::next(I);
@@ -322,37 +303,14 @@ static void fixStackStores(MachineBasicBlock &MBB,
 
     assert(MI.getOperand(0).getReg() == AVR::SP &&
            "Invalid register, should be SP!");
-    if (insertPushes) {
-      // Replace this instruction with a push.
-      Register SrcReg = MI.getOperand(2).getReg();
-      bool SrcIsKill = MI.getOperand(2).isKill();
-
-      // We can't use PUSHWRr here because when expanded the order of the new
-      // instructions are reversed from what we need. Perform the expansion now.
-      if (Opcode == AVR::STDWSPQRr) {
-        BuildMI(MBB, I, MI.getDebugLoc(), TII.get(AVR::PUSHRr))
-            .addReg(TRI.getSubReg(SrcReg, AVR::sub_hi),
-                    getKillRegState(SrcIsKill));
-        BuildMI(MBB, I, MI.getDebugLoc(), TII.get(AVR::PUSHRr))
-            .addReg(TRI.getSubReg(SrcReg, AVR::sub_lo),
-                    getKillRegState(SrcIsKill));
-      } else {
-        BuildMI(MBB, I, MI.getDebugLoc(), TII.get(AVR::PUSHRr))
-            .addReg(SrcReg, getKillRegState(SrcIsKill));
-      }
-
-      MI.eraseFromParent();
-      I = NextMI;
-      continue;
-    }
 
     // Replace this instruction with a regular store. Use Y as the base
     // pointer since it is guaranteed to contain a copy of SP.
     unsigned STOpc =
         (Opcode == AVR::STDWSPQRr) ? AVR::STDWPtrQRr : AVR::STDPtrQRr;
 
     MI.setDesc(TII.get(STOpc));
-    MI.getOperand(0).setReg(AVR::R29R28);
+    MI.getOperand(0).setReg(FP);
 
     I = NextMI;
   }
@@ -368,26 +326,45 @@ MachineBasicBlock::iterator AVRFrameLowering::eliminateCallFramePseudoInstr(
   // function entry. Delete the call frame pseudo and replace all pseudo stores
   // with real store instructions.
   if (hasReservedCallFrame(MF)) {
-    fixStackStores(MBB, MI, TII, false);
+    fixStackStores(MBB, MI, TII, AVR::R29R28);
     return MBB.erase(MI);
   }
 
   DebugLoc DL = MI->getDebugLoc();
   unsigned int Opcode = MI->getOpcode();
   int Amount = TII.getFrameSize(*MI);
 
-  // Adjcallstackup does not need to allocate stack space for the call, instead
-  // we insert push instructions that will allocate the necessary stack.
-  // For adjcallstackdown we convert it into an 'adiw reg, <amt>' handling
-  // the read and write of SP in I/O space.
+  // ADJCALLSTACKUP and ADJCALLSTACKDOWN are converted to adiw/subi
+  // instructions to read and write the stack pointer in I/O space.
   if (Amount != 0) {
     assert(getStackAlignment() == 1 && "Unsupported stack alignment");
 
     if (Opcode == TII.getCallFrameSetupOpcode()) {
-      fixStackStores(MBB, MI, TII, true);
+      // Update the stack pointer.
+      // In many cases this can be done far more efficiently by pushing the
+      // relevant values directly to the stack. However, doing that correctly
+      // (in the right order, possibly skipping some empty space for undef
+      // values, etc) is tricky and thus left to be optimized in the future.
+      BuildMI(MBB, MI, DL, TII.get(AVR::SPREAD), AVR::R31R30).addReg(AVR::SP);
+
+      MachineInstr *New = BuildMI(MBB, MI, DL, TII.get(AVR::SUBIWRdK), AVR::R31R30)
+                              .addReg(AVR::R31R30, RegState::Kill)
+                              .addImm(Amount);
+      New->getOperand(3).setIsDead();
+
+      BuildMI(MBB, MI, DL, TII.get(AVR::SPWRITE), AVR::SP)
+          .addReg(AVR::R31R30, RegState::Kill);
+
+      // Make sure the remaining stack stores are converted to real store
+      // instructions.
+      fixStackStores(MBB, MI, TII, AVR::R31R30);
     } else {
       assert(Opcode == TII.getCallFrameDestroyOpcode());
 
+      // Note that small stack changes could be implemented more efficiently
+      // with a few pop instructions instead of the 8-9 instructions now
+      // required.
+
       // Select the best opcode to adjust SP based on the offset size.
       unsigned addOpcode;
       if (isUInt<6>(Amount)) {
@@ -419,8 +396,10 @@ void AVRFrameLowering::determineCalleeSaves(MachineFunction &MF,
   TargetFrameLowering::determineCalleeSaves(MF, SavedRegs, RS);
 
   // If we have a frame pointer, the Y register needs to be saved as well.
-  // We don't do that here however - the prologue and epilogue generation
-  // code will handle it specially.
+  if (hasFP(MF)) {
+    SavedRegs.set(AVR::R29);
+    SavedRegs.set(AVR::R28);
+  }
 }
 /// The frame analyzer pass.
 ///