Approved by Evan:

$ svn merge -c 113297 https://llvm.org/svn/llvm-project/llvm/trunk
--- Merging r113297 into '.':
U    lib/Target/ARM/Thumb2SizeReduction.cpp
$ svn merge -c 113322 https://llvm.org/svn/llvm-project/llvm/trunk
--- Merging r113322 into '.':
U    lib/Target/ARM/ARMInstrNEON.td
U    lib/Target/ARM/ARMInstrVFP.td
U    lib/Target/ARM/ARMInstrFormats.td
$ svn merge -c 113365 https://llvm.org/svn/llvm-project/llvm/trunk
--- Merging r113365 into '.':
U    lib/Target/ARM/ARMBaseRegisterInfo.cpp
$ svn merge -c 113366 https://llvm.org/svn/llvm-project/llvm/trunk
--- Merging r113366 into '.':
G    lib/Target/ARM/ARMBaseRegisterInfo.cpp
$ svn merge -c 113394 https://llvm.org/svn/llvm-project/llvm/trunk
--- Merging r113394 into '.':
G    lib/Target/ARM/ARMBaseRegisterInfo.cpp



git-svn-id: https://llvm.org/svn/llvm-project/llvm/branches/release_28@113540 91177308-0d34-0410-b5e6-96231b3b80d8

lib/Target/ARM/ARMBaseRegisterInfo.cpp

@@ -667,8 +667,14 @@ bool ARMBaseRegisterInfo::hasBasePointer(const MachineFunction &MF) const {
 }
 
 bool ARMBaseRegisterInfo::canRealignStack(const MachineFunction &MF) const {
+  const MachineFrameInfo *MFI = MF.getFrameInfo();
   const ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
-  return (RealignStack && !AFI->isThumb1OnlyFunction());
+  // We can't realign the stack if:
+  // 1. Dynamic stack realignment is explicitly disabled,
+  // 2. This is a Thumb1 function (it's not useful, so we don't bother), or
+  // 3. There are VLAs in the function and the base pointer is disabled.
+  return (RealignStack && !AFI->isThumb1OnlyFunction() &&
+          (!MFI->hasVarSizedObjects() || EnableBasePointer));
 }
 
 bool ARMBaseRegisterInfo::
@@ -1057,8 +1063,11 @@ ARMBaseRegisterInfo::ResolveFrameIndexReference(const MachineFunction &MF,
     if (isFixed) {
       FrameReg = getFrameRegister(MF);
       Offset = FPOffset;
-    } else if (MFI->hasVarSizedObjects())
+    } else if (MFI->hasVarSizedObjects()) {
+      assert(hasBasePointer(MF) &&
+             "VLAs and dynamic stack alignment, but missing base pointer!");
       FrameReg = BasePtr;
+    }
     return Offset;
   }
 
@@ -1068,7 +1077,7 @@ ARMBaseRegisterInfo::ResolveFrameIndexReference(const MachineFunction &MF,
     // there are VLAs (and thus the SP isn't reliable as a base).
     if (isFixed || (MFI->hasVarSizedObjects() && !hasBasePointer(MF))) {
       FrameReg = getFrameRegister(MF);
-      Offset = FPOffset;
+      return FPOffset;
     } else if (MFI->hasVarSizedObjects()) {
       assert(hasBasePointer(MF) && "missing base pointer!");
       // Use the base register since we have it.
@@ -1078,12 +1087,12 @@ ARMBaseRegisterInfo::ResolveFrameIndexReference(const MachineFunction &MF,
       // out of range references.
       if (FPOffset >= -255 && FPOffset < 0) {
         FrameReg = getFrameRegister(MF);
-        Offset = FPOffset;
+        return FPOffset;
       }
     } else if (Offset > (FPOffset < 0 ? -FPOffset : FPOffset)) {
       // Otherwise, use SP or FP, whichever is closer to the stack slot.
       FrameReg = getFrameRegister(MF);
-      Offset = FPOffset;
+      return FPOffset;
     }
   }
   // Use the base pointer if we have one.
@@ -1887,7 +1896,8 @@ emitPrologue(MachineFunction &MF) const {
   AFI->setGPRCalleeSavedArea2Size(GPRCS2Size);
   AFI->setDPRCalleeSavedAreaSize(DPRCSSize);
 
-  // If we need dynamic stack realignment, do it here.
+  // If we need dynamic stack realignment, do it here. Be paranoid and make
+  // sure if we also have VLAs, we have a base pointer for frame access.
   if (needsStackRealignment(MF)) {
     unsigned MaxAlign = MFI->getMaxAlignment();
     assert (!AFI->isThumb1OnlyFunction());

lib/Target/ARM/ARMInstrFormats.td

@@ -1332,9 +1332,9 @@ class ASI5<bits<4> opcod1, bits<2> opcod2, dag oops, dag iops,
 }
 
 // Load / store multiple
-class AXDI5<dag oops, dag iops, IndexMode im, InstrItinClass itin,
+class AXDI4<dag oops, dag iops, IndexMode im, InstrItinClass itin,
             string asm, string cstr, list<dag> pattern>
-  : VFPXI<oops, iops, AddrMode5, Size4Bytes, im,
+  : VFPXI<oops, iops, AddrMode4, Size4Bytes, im,
           VFPLdStMulFrm, itin, asm, cstr, pattern> {
   // TODO: Mark the instructions with the appropriate subtarget info.
   let Inst{27-25} = 0b110;
@@ -1344,9 +1344,9 @@ class AXDI5<dag oops, dag iops, IndexMode im, InstrItinClass itin,
   let D = VFPNeonDomain;
 }
 
-class AXSI5<dag oops, dag iops, IndexMode im, InstrItinClass itin,
+class AXSI4<dag oops, dag iops, IndexMode im, InstrItinClass itin,
             string asm, string cstr, list<dag> pattern>
-  : VFPXI<oops, iops, AddrMode5, Size4Bytes, im,
+  : VFPXI<oops, iops, AddrMode4, Size4Bytes, im,
           VFPLdStMulFrm, itin, asm, cstr, pattern> {
   // TODO: Mark the instructions with the appropriate subtarget info.
   let Inst{27-25} = 0b110;

lib/Target/ARM/ARMInstrNEON.td

@@ -133,7 +133,7 @@ def nModImm : Operand<i32> {
 // This is equivalent to VLDMD except that it has a Q register operand
 // instead of a pair of D registers.
 def VLDMQ
-  : AXDI5<(outs QPR:$dst), (ins addrmode4:$addr, pred:$p),
+  : AXDI4<(outs QPR:$dst), (ins addrmode4:$addr, pred:$p),
           IndexModeNone, IIC_fpLoadm,
           "vldm${addr:submode}${p}\t$addr, ${dst:dregpair}", "",
           [(set QPR:$dst, (v2f64 (load addrmode4:$addr)))]>;
@@ -151,7 +151,7 @@ def VLD1q
 // This is equivalent to VSTMD except that it has a Q register operand
 // instead of a pair of D registers.
 def VSTMQ
-  : AXDI5<(outs), (ins QPR:$src, addrmode4:$addr, pred:$p),
+  : AXDI4<(outs), (ins QPR:$src, addrmode4:$addr, pred:$p),
           IndexModeNone, IIC_fpStorem,
           "vstm${addr:submode}${p}\t$addr, ${src:dregpair}", "",
           [(store (v2f64 QPR:$src), addrmode4:$addr)]>;

lib/Target/ARM/ARMInstrVFP.td

@@ -77,19 +77,19 @@ def VSTRS  : ASI5<0b1101, 0b00, (outs), (ins SPR:$src, addrmode5:$addr),
 //
 
 let mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1 in {
-def VLDMD : AXDI5<(outs), (ins addrmode4:$addr, pred:$p, reglist:$dsts,
+def VLDMD : AXDI4<(outs), (ins addrmode4:$addr, pred:$p, reglist:$dsts,
                            variable_ops), IndexModeNone, IIC_fpLoadm,
                   "vldm${addr:submode}${p}\t$addr, $dsts", "", []> {
   let Inst{20} = 1;
 }
 
-def VLDMS : AXSI5<(outs), (ins addrmode4:$addr, pred:$p, reglist:$dsts,
+def VLDMS : AXSI4<(outs), (ins addrmode4:$addr, pred:$p, reglist:$dsts,
                            variable_ops), IndexModeNone, IIC_fpLoadm,
                   "vldm${addr:submode}${p}\t$addr, $dsts", "", []> {
   let Inst{20} = 1;
 }
 
-def VLDMD_UPD : AXDI5<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
+def VLDMD_UPD : AXDI4<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
                                        reglist:$dsts, variable_ops),
                       IndexModeUpd, IIC_fpLoadm,
                       "vldm${addr:submode}${p}\t$addr!, $dsts",
@@ -97,7 +97,7 @@ def VLDMD_UPD : AXDI5<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
   let Inst{20} = 1;
 }
 
-def VLDMS_UPD : AXSI5<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
+def VLDMS_UPD : AXSI4<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
                                        reglist:$dsts, variable_ops),
                       IndexModeUpd, IIC_fpLoadm, 
                       "vldm${addr:submode}${p}\t$addr!, $dsts",
@@ -107,19 +107,19 @@ def VLDMS_UPD : AXSI5<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
 } // mayLoad, neverHasSideEffects, hasExtraDefRegAllocReq
 
 let mayStore = 1, neverHasSideEffects = 1, hasExtraSrcRegAllocReq = 1 in {
-def VSTMD : AXDI5<(outs), (ins addrmode4:$addr, pred:$p, reglist:$srcs,
+def VSTMD : AXDI4<(outs), (ins addrmode4:$addr, pred:$p, reglist:$srcs,
                            variable_ops), IndexModeNone, IIC_fpStorem,
                   "vstm${addr:submode}${p}\t$addr, $srcs", "", []> {
   let Inst{20} = 0;
 }
 
-def VSTMS : AXSI5<(outs), (ins addrmode4:$addr, pred:$p, reglist:$srcs,
+def VSTMS : AXSI4<(outs), (ins addrmode4:$addr, pred:$p, reglist:$srcs,
                            variable_ops), IndexModeNone, IIC_fpStorem,
                   "vstm${addr:submode}${p}\t$addr, $srcs", "", []> {
   let Inst{20} = 0;
 }
 
-def VSTMD_UPD : AXDI5<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
+def VSTMD_UPD : AXDI4<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
                                        reglist:$srcs, variable_ops),
                       IndexModeUpd, IIC_fpStorem,
                       "vstm${addr:submode}${p}\t$addr!, $srcs",
@@ -127,7 +127,7 @@ def VSTMD_UPD : AXDI5<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
   let Inst{20} = 0;
 }
 
-def VSTMS_UPD : AXSI5<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
+def VSTMS_UPD : AXSI4<(outs GPR:$wb), (ins addrmode4:$addr, pred:$p,
                                        reglist:$srcs, variable_ops),
                       IndexModeUpd, IIC_fpStorem,
                       "vstm${addr:submode}${p}\t$addr!, $srcs",

lib/Target/ARM/Thumb2SizeReduction.cpp

@@ -315,6 +315,18 @@ Thumb2SizeReduce::ReduceLoadStore(MachineBasicBlock &MBB, MachineInstr *MI,
     ARM_AM::AMSubMode Mode = ARM_AM::getAM4SubMode(MI->getOperand(1).getImm());
     if (!isARMLowRegister(BaseReg) || Mode != ARM_AM::ia)
       return false;
+    // For the non-writeback version (this one), the base register must be
+    // one of the registers being loaded.
+    bool isOK = false;
+    for (unsigned i = 4; i < MI->getNumOperands(); ++i) {
+      if (MI->getOperand(i).getReg() == BaseReg) {
+        isOK = true;
+        break;
+      }
+    }
+    if (!isOK)
+      return false;
+
     OpNum = 0;
     isLdStMul = true;
     break;