[M68k] always use movem for register spills / spill 8 bit registers i…

…nto 16 bit slots
llvm · Jan 9, 2025 · 3cdcef9 · 3cdcef9
1 parent fdf72c9
commit 3cdcef9
Show file tree

Hide file tree

Showing 6 changed files with 524 additions and 52 deletions.
diff --git a/llvm/lib/Target/M68k/M68kExpandPseudo.cpp b/llvm/lib/Target/M68k/M68kExpandPseudo.cpp
@@ -193,31 +193,23 @@ bool M68kExpandPseudo::ExpandMI(MachineBasicBlock &MBB,
   case M68k::MOV8dc:
     return TII->ExpandCCR(MIB, /*IsToCCR=*/false);
 
-  case M68k::MOVM8jm_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32jm), /*IsRM=*/false);
   case M68k::MOVM16jm_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32jm), /*IsRM=*/false);
+    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM16jm), /*IsRM=*/false);
   case M68k::MOVM32jm_P:
     return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32jm), /*IsRM=*/false);
 
-  case M68k::MOVM8pm_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32pm), /*IsRM=*/false);
   case M68k::MOVM16pm_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32pm), /*IsRM=*/false);
+    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM16pm), /*IsRM=*/false);
   case M68k::MOVM32pm_P:
     return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32pm), /*IsRM=*/false);
 
-  case M68k::MOVM8mj_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32mj), /*IsRM=*/true);
   case M68k::MOVM16mj_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32mj), /*IsRM=*/true);
+    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM16mj), /*IsRM=*/true);
   case M68k::MOVM32mj_P:
     return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32mj), /*IsRM=*/true);
 
-  case M68k::MOVM8mp_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32mp), /*IsRM=*/true);
   case M68k::MOVM16mp_P:
-    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32mp), /*IsRM=*/true);
+    return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM16mp), /*IsRM=*/true);
   case M68k::MOVM32mp_P:
     return TII->ExpandMOVEM(MIB, TII->get(M68k::MOVM32mp), /*IsRM=*/true);
 

diff --git a/llvm/lib/Target/M68k/M68kInstrData.td b/llvm/lib/Target/M68k/M68kInstrData.td
@@ -337,20 +337,16 @@ class MxMOVEM_RM_Pseudo<MxType TYPE, MxOperand MEMOp>
     : MxPseudo<(outs TYPE.ROp:$dst), (ins MEMOp:$src)>;
 
 // Mem <- Reg
-def MOVM8jm_P  : MxMOVEM_MR_Pseudo<MxType8d,  MxType8.JOp>;
 def MOVM16jm_P : MxMOVEM_MR_Pseudo<MxType16r, MxType16.JOp>;
 def MOVM32jm_P : MxMOVEM_MR_Pseudo<MxType32r, MxType32.JOp>;
 
-def MOVM8pm_P  : MxMOVEM_MR_Pseudo<MxType8d,  MxType8.POp>;
 def MOVM16pm_P : MxMOVEM_MR_Pseudo<MxType16r, MxType16.POp>;
 def MOVM32pm_P : MxMOVEM_MR_Pseudo<MxType32r, MxType32.POp>;
 
 // Reg <- Mem
-def MOVM8mj_P  : MxMOVEM_RM_Pseudo<MxType8d,  MxType8.JOp>;
 def MOVM16mj_P : MxMOVEM_RM_Pseudo<MxType16r, MxType16.JOp>;
 def MOVM32mj_P : MxMOVEM_RM_Pseudo<MxType32r, MxType32.JOp>;
 
-def MOVM8mp_P  : MxMOVEM_RM_Pseudo<MxType8d,  MxType8.POp>;
 def MOVM16mp_P : MxMOVEM_RM_Pseudo<MxType16r, MxType16.POp>;
 def MOVM32mp_P : MxMOVEM_RM_Pseudo<MxType32r, MxType32.POp>;
 

diff --git a/llvm/lib/Target/M68k/M68kInstrInfo.cpp b/llvm/lib/Target/M68k/M68kInstrInfo.cpp
@@ -542,7 +542,6 @@ bool M68kInstrInfo::ExpandCCR(MachineInstrBuilder &MIB, bool IsToCCR) const {
 bool M68kInstrInfo::ExpandMOVEM(MachineInstrBuilder &MIB,
                                 const MCInstrDesc &Desc, bool IsRM) const {
   int Reg = 0, Offset = 0, Base = 0;
-  auto XR32 = RI.getRegClass(M68k::XR32RegClassID);
   auto DL = MIB->getDebugLoc();
   auto MI = MIB.getInstr();
   auto &MBB = *MIB->getParent();
@@ -557,13 +556,6 @@ bool M68kInstrInfo::ExpandMOVEM(MachineInstrBuilder &MIB,
     Reg = MIB->getOperand(2).getReg();
   }
 
-  // If the register is not in XR32 then it is smaller than 32 bit, we
-  // implicitly promote it to 32
-  if (!XR32->contains(Reg)) {
-    Reg = RI.getMatchingMegaReg(Reg, XR32);
-    assert(Reg && "Has not meaningful MEGA register");
-  }
-
   unsigned Mask = 1 << RI.getSpillRegisterOrder(Reg);
   if (IsRM) {
     BuildMI(MBB, MI, DL, Desc)
@@ -734,22 +726,25 @@ namespace {
 unsigned getLoadStoreRegOpcode(unsigned Reg, const TargetRegisterClass *RC,
                                const TargetRegisterInfo *TRI,
                                const M68kSubtarget &STI, bool load) {
-  switch (TRI->getRegSizeInBits(*RC)) {
+  switch (TRI->getSpillSize(*RC)) {
   default:
+    LLVM_DEBUG(
+        dbgs() << "Cannot determine appropriate opcode for load/store to/from "
+               << TRI->getName(Reg) << " of class " << TRI->getRegClassName(RC)
+               << " with spill size " << TRI->getSpillSize(*RC) << '\n');
     llvm_unreachable("Unknown spill size");
-  case 8:
+  case 2:
+    if (M68k::XR16RegClass.hasSubClassEq(RC))
+      return load ? M68k::MOVM16mp_P : M68k::MOVM16pm_P;
     if (M68k::DR8RegClass.hasSubClassEq(RC))
-      return load ? M68k::MOV8dp : M68k::MOV8pd;
+      return load ? M68k::MOVM16mp_P : M68k::MOVM16pm_P;
     if (M68k::CCRCRegClass.hasSubClassEq(RC))
-      return load ? M68k::MOV16cp : M68k::MOV16pc;
-
-    llvm_unreachable("Unknown 1-byte regclass");
-  case 16:
-    assert(M68k::XR16RegClass.hasSubClassEq(RC) && "Unknown 2-byte regclass");
-    return load ? M68k::MOVM16mp_P : M68k::MOVM16pm_P;
-  case 32:
-    assert(M68k::XR32RegClass.hasSubClassEq(RC) && "Unknown 4-byte regclass");
-    return load ? M68k::MOVM32mp_P : M68k::MOVM32pm_P;
+      return load ? M68k::MOVM16mp_P : M68k::MOVM16pm_P;
+    llvm_unreachable("Unknown 2-byte regclass");
+  case 4:
+    if (M68k::XR32RegClass.hasSubClassEq(RC))
+      return load ? M68k::MOVM32mp_P : M68k::MOVM32pm_P;
+    llvm_unreachable("Unknown 4-byte regclass");
   }
 }
 

diff --git a/llvm/lib/Target/M68k/M68kRegisterInfo.td b/llvm/lib/Target/M68k/M68kRegisterInfo.td
@@ -99,52 +99,77 @@ class MxRegClass<list<ValueType> regTypes, int alignment, dag regList>
     : RegisterClass<"M68k", regTypes, alignment, regList>;
 
 // Data Registers
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<8,16,16>]> in
 def DR8  : MxRegClass<[i8],  16, (sequence "BD%u", 0, 7)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<16,16,16>]> in
 def DR16 : MxRegClass<[i16], 16, (sequence "WD%u", 0, 7)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def DR32 : MxRegClass<[i32], 32, (sequence "D%u",  0, 7)>;
 
 // Address Registers
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<16,16,16>]> in
 def AR16 : MxRegClass<[i16], 16, (add (sequence "WA%u", 0, 6), WSP)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def AR32 : MxRegClass<[i32], 32, (add (sequence "A%u", 0, 6), SP)>;
 
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def AR32_NOSP : MxRegClass<[i32], 32, (sequence "A%u", 0, 6)>;
 
 // Index Register Classes
 // FIXME try alternative ordering like `D0, D1, A0, A1, ...`
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<16,16,16>]> in
 def XR16 : MxRegClass<[i16], 16, (add DR16, AR16)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def XR32 : MxRegClass<[i32], 32, (add DR32, AR32)>;
 
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def SPC  : MxRegClass<[i32], 32, (add SP)>;
 
 // Floating Point Data Registers
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def FPDR32 : MxRegClass<[f32], 32, (sequence "FP%u", 0, 7)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<64,64,32>]> in
 def FPDR64 : MxRegClass<[f64], 32, (add FPDR32)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<80,128,32>]> in
 def FPDR80 : MxRegClass<[f80], 32, (add FPDR32)>;
 
 let CopyCost = -1 in {
+  let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<8,16,16>]> in
   def CCRC : MxRegClass<[i8],  16, (add CCR)>;
+  let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<16,16,16>]> in
   def SRC  : MxRegClass<[i16], 16, (add SR)>;
 
   // Float Point System Control Registers
-  def FPIC   : MxRegClass<[i32], 32, (add FPIAR)>;
-  def FPCSC  : MxRegClass<[i32], 32, (add FPC, FPS)>;
-  def FPSYSC : MxRegClass<[i32], 32, (add FPCSC, FPIC)>;
+  let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in {
+    def FPIC   : MxRegClass<[i32], 32, (add FPIAR)>;
+    def FPCSC  : MxRegClass<[i32], 32, (add FPC, FPS)>;
+    def FPSYSC : MxRegClass<[i32], 32, (add FPCSC, FPIC)>;
+  }
 }
 
 let isAllocatable = 0 in {
+  let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
   def PCC  : MxRegClass<[i32], 32, (add PC)>;
 }
 
 // Register used with tail call
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<16,16,16>]> in
 def DR16_TC : MxRegClass<[i16], 16, (add D0, D1)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def DR32_TC : MxRegClass<[i32], 32, (add D0, D1)>;
 
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<16,16,16>]> in
 def AR16_TC : MxRegClass<[i16], 16, (add A0, A1)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def AR32_TC : MxRegClass<[i32], 32, (add A0, A1)>;
 
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<16,16,16>]> in
 def XR16_TC : MxRegClass<[i16], 16, (add DR16_TC, AR16_TC)>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in
 def XR32_TC : MxRegClass<[i32], 32, (add DR32_TC, AR32_TC)>;
 
 // These classes provide spill/restore order if used with MOVEM instruction
-def SPILL   : MxRegClass<[i32], 32, (add XR32)>;
-def SPILL_R : MxRegClass<[i32], 32, (add SP, (sequence "A%u", 6, 0), (sequence "D%u", 7, 0))>;
+let RegInfos = RegInfoByHwMode<[DefaultMode], [RegInfo<32,32,32>]> in {
+  def SPILL   : MxRegClass<[i32], 32, (add XR32)>;
+  def SPILL_R : MxRegClass<[i32], 32, (add SP, (sequence "A%u", 6, 0), (sequence "D%u", 7, 0))>;
+}
diff --git a/llvm/test/CodeGen/M68k/PR57660.ll b/llvm/test/CodeGen/M68k/PR57660.ll
@@ -8,10 +8,10 @@ define dso_local void @foo1() {
 ; CHECK-NEXT:    suba.l #2, %sp
 ; CHECK-NEXT:    .cfi_def_cfa_offset -6
 ; CHECK-NEXT:    moveq #0, %d0
-; CHECK-NEXT:    move.b %d0, (0,%sp) ; 1-byte Folded Spill
+; CHECK-NEXT:    movem.w %d0, (0,%sp)
 ; CHECK-NEXT:  .LBB0_1: ; %do.body
 ; CHECK-NEXT:    ; =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    move.b (0,%sp), %d0 ; 1-byte Folded Reload
+; CHECK-NEXT:    movem.w (0,%sp), %d0
 ; CHECK-NEXT:    cmpi.b #0, %d0
 ; CHECK-NEXT:    bne .LBB0_1
 ; CHECK-NEXT:  ; %bb.2: ; %do.end
@@ -39,24 +39,24 @@ define i32 @foo2(ptr noundef %0) {
 ; CHECK-NEXT:    .cfi_def_cfa_offset -8
 ; CHECK-NEXT:    move.l (8,%sp), %a0
 ; CHECK-NEXT:    move.b (%a0), %d0
-; CHECK-NEXT:    move.b %d0, (0,%sp) ; 1-byte Folded Spill
+; CHECK-NEXT:    movem.w %d0, (0,%sp)
 ; CHECK-NEXT:    and.b #1, %d0
-; CHECK-NEXT:    move.b %d0, (2,%sp) ; 1-byte Folded Spill
+; CHECK-NEXT:    movem.w %d0, (2,%sp)
 ; CHECK-NEXT:    sub.b #1, %d0
 ; CHECK-NEXT:    bgt .LBB1_2
 ; CHECK-NEXT:  ; %bb.1: ; %if
-; CHECK-NEXT:    move.b (2,%sp), %d0 ; 1-byte Folded Reload
-; CHECK-NEXT:    move.b (0,%sp), %d1 ; 1-byte Folded Reload
+; CHECK-NEXT:    movem.w (2,%sp), %d0
+; CHECK-NEXT:    movem.w (0,%sp), %d1
 ; CHECK-NEXT:    add.b %d1, %d0
 ; CHECK-NEXT:    bra .LBB1_3
 ; CHECK-NEXT:  .LBB1_2: ; %else
-; CHECK-NEXT:    move.b (2,%sp), %d1 ; 1-byte Folded Reload
-; CHECK-NEXT:    move.b (0,%sp), %d0 ; 1-byte Folded Reload
+; CHECK-NEXT:    movem.w (2,%sp), %d1
+; CHECK-NEXT:    movem.w (0,%sp), %d0
 ; CHECK-NEXT:    sub.b %d1, %d0
-; CHECK-NEXT:    move.b %d0, (0,%sp) ; 1-byte Folded Spill
+; CHECK-NEXT:    movem.w %d0, (0,%sp)
 ; CHECK-NEXT:  .LBB1_3: ; %cont
-; CHECK-NEXT:    move.b %d0, (2,%sp) ; 1-byte Folded Spill
-; CHECK-NEXT:    move.b (2,%sp), %d0 ; 1-byte Folded Reload
+; CHECK-NEXT:    movem.w %d0, (2,%sp)
+; CHECK-NEXT:    movem.w (2,%sp), %d0
 ; CHECK-NEXT:    ext.w %d0
 ; CHECK-NEXT:    ext.l %d0
 ; CHECK-NEXT:    adda.l #4, %sp