/* * This file is part of the Jikes RVM project (http://jikesrvm.org). * * This file is licensed to You under the Eclipse Public License (EPL); * You may not use this file except in compliance with the License. You * may obtain a copy of the License at * * http://www.opensource.org/licenses/eclipse-1.0.php * * See the COPYRIGHT.txt file distributed with this work for information * regarding copyright ownership. */ package org.jikesrvm.compilers.opt.regalloc.ppc; import static org.jikesrvm.VM.NOT_REACHED; import static org.jikesrvm.compilers.opt.ir.Operators.IR_PROLOGUE_opcode; import static org.jikesrvm.compilers.opt.ir.Operators.YIELDPOINT_OSR; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.CALL_SAVE_VOLATILE; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_ADDI; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_BCTRL_SYS; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_BLR_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_BL_SYS; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_CMPI; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_FMR; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_FMR_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LAddr; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LAddr_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LDI; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LDIS; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LFD; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LFD_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LFS; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LFS_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LInt; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LInt_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LMW; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LWZ; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_LWZ_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_MFSPR; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_MOVE; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_MOVE_opcode; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_MTSPR; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_ORI; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_STAddr; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_STAddrU; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_STFD; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_STFS; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_STMW; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_STW; import static org.jikesrvm.compilers.opt.ir.ppc.ArchOperators.PPC_TAddr; import static org.jikesrvm.compilers.opt.regalloc.ppc.PhysicalRegisterConstants.CONDITION_REG; import static org.jikesrvm.compilers.opt.regalloc.ppc.PhysicalRegisterConstants.DOUBLE_REG; import static org.jikesrvm.compilers.opt.regalloc.ppc.PhysicalRegisterConstants.INT_REG; import static org.jikesrvm.compilers.opt.regalloc.ppc.PhysicalRegisterConstants.SPECIAL_REG; import static org.jikesrvm.ppc.RegisterConstants.FIRST_SCRATCH_GPR; import static org.jikesrvm.ppc.RegisterConstants.LAST_SCRATCH_GPR; import static org.jikesrvm.ppc.StackframeLayoutConstants.STACKFRAME_ALIGNMENT; import static org.jikesrvm.ppc.StackframeLayoutConstants.STACKFRAME_METHOD_ID_OFFSET; import static org.jikesrvm.ppc.StackframeLayoutConstants.STACKFRAME_RETURN_ADDRESS_OFFSET; import static org.jikesrvm.ppc.StackframeLayoutConstants.STACK_SIZE_GUARD; import static org.jikesrvm.runtime.JavaSizeConstants.BYTES_IN_DOUBLE; import static org.jikesrvm.runtime.JavaSizeConstants.BYTES_IN_FLOAT; import static org.jikesrvm.runtime.JavaSizeConstants.BYTES_IN_INT; import static org.jikesrvm.runtime.UnboxedSizeConstants.BYTES_IN_ADDRESS; import static org.jikesrvm.util.Bits.PPCMaskLower16; import static org.jikesrvm.util.Bits.fits; import java.util.Enumeration; import java.util.Iterator; import org.jikesrvm.VM; import org.jikesrvm.compilers.opt.OptimizingCompilerException; import org.jikesrvm.compilers.opt.ir.GenericPhysicalRegisterSet; import org.jikesrvm.compilers.opt.ir.IR; import org.jikesrvm.compilers.opt.ir.Instruction; import org.jikesrvm.compilers.opt.ir.Register; import org.jikesrvm.compilers.opt.ir.operand.IntConstantOperand; import org.jikesrvm.compilers.opt.ir.operand.Operand; import org.jikesrvm.compilers.opt.ir.operand.RegisterOperand; import org.jikesrvm.compilers.opt.ir.operand.TrapCodeOperand; import org.jikesrvm.compilers.opt.ir.operand.ppc.PowerPCTrapOperand; import org.jikesrvm.compilers.opt.ir.ppc.MIR_Binary; import org.jikesrvm.compilers.opt.ir.ppc.MIR_Load; import org.jikesrvm.compilers.opt.ir.ppc.MIR_Move; import org.jikesrvm.compilers.opt.ir.ppc.MIR_Store; import org.jikesrvm.compilers.opt.ir.ppc.MIR_StoreUpdate; import org.jikesrvm.compilers.opt.ir.ppc.MIR_Trap; import org.jikesrvm.compilers.opt.ir.ppc.MIR_Unary; import org.jikesrvm.compilers.opt.ir.ppc.PhysicalRegisterSet; import org.jikesrvm.compilers.opt.regalloc.GenericStackManager; import org.jikesrvm.runtime.Entrypoints; import org.vmmagic.unboxed.Offset; /** * Class to manage the allocation of the "compiler-specific" portion of * the stackframe. This class holds only the architecture-specific * functions. * <p> */ public final class StackManager extends GenericStackManager { /** * stack location to save the XER register */ private int saveXERLocation; /** * stack location to save the CTR register */ private int saveCTRLocation; /** * Return the size of the fixed portion of the stack. * @return size in bytes of the fixed portion of the stackframe */ @Override public int getFrameFixedSize() { return frameSize; } /** * Allocate a new spill location and grow the * frame size to reflect the new layout. * * @param type the type to spill * @return the spill location */ @Override public int allocateNewSpillLocation(int type) { int spillSize = getSpillSize(type); // Naturally align the spill pointer spillPointer = align(spillPointer, spillSize); // increment by the spill size spillPointer += spillSize; if (spillPointer > frameSize) { frameSize = spillPointer; } return spillPointer - spillSize; } /** * Gets the spill size for a register with a particular type * @param type one of INT_REG, DOUBLE_REG, CONDITION_REG, SPECIAL_REG * @return the spill size in bytes */ @Override public int getSpillSize(int type) { if (VM.VerifyAssertions) { VM._assert((type == INT_REG) || (type == DOUBLE_REG) || (type == CONDITION_REG) || (type == SPECIAL_REG)); } if (type == DOUBLE_REG) { return BYTES_IN_DOUBLE; } else { return BYTES_IN_ADDRESS; } } /** * Clean up some junk that's left in the IR after register allocation, * and add epilogue code. */ @Override public void cleanUpAndInsertEpilogue() { Instruction inst = ir.firstInstructionInCodeOrder().nextInstructionInCodeOrder(); for (; inst != null; inst = inst.nextInstructionInCodeOrder()) { switch (inst.getOpcode()) { case PPC_MOVE_opcode: case PPC_FMR_opcode: // remove frivolous moves if (MIR_Move.getResult(inst).getRegister().number == MIR_Move.getValue(inst).getRegister().number) { inst = inst.remove(); } break; case PPC_BLR_opcode: if (frameIsRequired()) { insertEpilogue(inst); } break; case PPC_LFS_opcode: case PPC_LFD_opcode: case PPC_LInt_opcode: case PPC_LWZ_opcode: case PPC_LAddr_opcode: // the following to handle spilled parameters // SJF: this is ugly. clean it up someday. if (MIR_Load.getAddress(inst).getRegister() == ir.regpool.getPhysicalRegisterSet().getFP()) { Operand one = MIR_Load.getOffset(inst); if (one instanceof IntConstantOperand) { int offset = ((IntConstantOperand) one).value; if (offset <= -256) { if (frameIsRequired()) { MIR_Load.setOffset(inst, IC(frameSize - offset - 256)); } else { MIR_Load.setOffset(inst, IC(-offset - 256)); } } } } default: break; } } } /** * Insert a spill of a physical register before instruction s. * * @param s the instruction before which the spill should occur * @param r the register (should be physical) to spill * @param type one of INT_VALUE, FLOAT_VALUE, DOUBLE_VALUE, or * CONDITION_VALUE * @param location the spill location */ @Override public void insertSpillBefore(Instruction s, Register r, Register type, int location) { Register FP = ir.regpool.getPhysicalRegisterSet().getFP(); if (type.isFloat()) { s.insertBefore(MIR_Store.create(PPC_STFS, F(r), A(FP), IC(location + BYTES_IN_ADDRESS - BYTES_IN_FLOAT))); } else if (type.isDouble()) { s.insertBefore(MIR_Store.create(PPC_STFD, D(r), A(FP), IC(location))); } else if (type.isNatural()) { // integer or half of long s.insertBefore(MIR_Store.create(PPC_STAddr, A(r), A(FP), IC(location))); } else { throw new OptimizingCompilerException("insertSpillBefore", "unsupported type " + type); } } /** * Create an MIR instruction to move rhs into lhs */ Instruction makeMoveInstruction(Register lhs, Register rhs) { if (rhs.isFloatingPoint() && lhs.isFloatingPoint()) { return MIR_Move.create(PPC_FMR, D(lhs), D(rhs)); //} else if (rhs.isInteger() && lhs.isInteger()) { // integer } else if (rhs.isAddress() && lhs.isAddress()) { // integer return MIR_Move.create(PPC_MOVE, A(lhs), A(rhs)); } else { throw new OptimizingCompilerException("RegAlloc", "unknown register:", lhs.toString()); } } /** * Insert a load of a physical register from a spill location before * instruction s. * * @param s the instruction before which the spill should occur * @param r the register (should be physical) to spill * @param type one of INT_VALUE, FLOAT_VALUE, DOUBLE_VALUE, or * CONDITION_VALUE * @param location the spill location */ @Override public void insertUnspillBefore(Instruction s, Register r, Register type, int location) { PhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet().asPPC(); Register FP = phys.getFP(); if (type.isCondition()) { Register temp = phys.getTemp(); s.insertBefore(MIR_Load.create(PPC_LWZ, I(temp), A(FP), IC(location + BYTES_IN_ADDRESS - BYTES_IN_INT))); } else if (type.isDouble()) { s.insertBefore(MIR_Load.create(PPC_LFD, D(r), A(FP), IC(location))); } else if (type.isFloat()) { s.insertBefore(MIR_Load.create(PPC_LFS, F(r), A(FP), IC(location + BYTES_IN_ADDRESS - BYTES_IN_FLOAT))); } else if (type.isNatural()) { // integer or half of long s.insertBefore(MIR_Load.create(PPC_LAddr, A(r), A(FP), IC(location))); } else { throw new OptimizingCompilerException("insertUnspillBefore", "unknown type:" + type); } } /** * Insert the epilogue before a particular return instruction. * * @param ret the return instruction. */ private void insertEpilogue(Instruction ret) { // 1. Restore any saved registers if (ir.compiledMethod.isSaveVolatile()) { restoreVolatileRegisters(ret); } restoreNonVolatiles(ret); // 2. Restore return address PhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet().asPPC(); Register temp = phys.getTemp(); Register FP = phys.getFP(); ret.insertBefore(MIR_Load.create(PPC_LAddr, A(temp), A(FP), IC(STACKFRAME_RETURN_ADDRESS_OFFSET.toInt() + frameSize))); // 3. Load return address into LR ret.insertBefore(MIR_Move.create(PPC_MTSPR, A(phys.getLR()), A(phys.getTemp()))); // 4. Restore old FP ret.insertBefore(MIR_Binary.create(PPC_ADDI, A(FP), A(FP), IC(frameSize))); } /** * Insert code in the prologue to save the * volatile registers. * * @param inst */ private void saveVolatiles(Instruction inst) { PhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet().asPPC(); // 1. save the volatile GPRs Register FP = phys.getFP(); int i = 0; for (Enumeration<Register> e = phys.enumerateVolatileGPRs(); e.hasMoreElements(); i++) { Register r = e.nextElement(); int location = saveVolatileGPRLocation[i]; inst.insertBefore(MIR_Store.create(PPC_STAddr, A(r), A(FP), IC(location))); } // 2. save the volatile FPRs i = 0; for (Enumeration<Register> e = phys.enumerateVolatileFPRs(); e.hasMoreElements(); i++) { Register r = e.nextElement(); int location = saveVolatileFPRLocation[i]; inst.insertBefore(MIR_Store.create(PPC_STFD, D(r), A(FP), IC(location))); } // 3. Save some special registers Register temp = phys.getTemp(); inst.insertBefore(MIR_Move.create(PPC_MFSPR, I(temp), I(phys.getXER()))); inst.insertBefore(MIR_Store.create(PPC_STW, I(temp), A(FP), IC(saveXERLocation))); inst.insertBefore(MIR_Move.create(PPC_MFSPR, A(temp), A(phys.getCTR()))); inst.insertBefore(MIR_Store.create(PPC_STAddr, A(temp), A(FP), IC(saveCTRLocation))); } /** * Insert code into the prologue to save any used non-volatile * registers. * * @param inst the first instruction after the prologue. */ private void saveNonVolatiles(Instruction inst) { GenericPhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet(); int nNonvolatileGPRS = ir.compiledMethod.getNumberOfNonvolatileGPRs(); if (ir.compiledMethod.isSaveVolatile()) { // pretend we use all non-volatiles nNonvolatileGPRS = PhysicalRegisterSet.getNumberOfNonvolatileGPRs(); } // 1. save the nonvolatile GPRs int n = nNonvolatileGPRS - 1; Register FP = phys.getFP(); if (VM.BuildFor32Addr && n > MULTIPLE_CUTOFF) { // use a stm Register nv = null; for (Enumeration<Register> e = phys.enumerateNonvolatileGPRsBackwards(); e.hasMoreElements() && n >= 0; n--) { nv = e.nextElement(); } n++; RegisterOperand range = I(nv); // YUCK!!! Why is this crap in register operand?? int offset = getNonvolatileGPROffset(n); inst.insertBefore(MIR_Store.create(PPC_STMW, range, A(FP), IC(offset))); } else { // use a sequence of load instructions for (Enumeration<Register> e = phys.enumerateNonvolatileGPRsBackwards(); e.hasMoreElements() && n >= 0; n--) { Register nv = e.nextElement(); int offset = getNonvolatileGPROffset(n); inst.insertBefore(MIR_Store.create(PPC_STAddr, A(nv), A(FP), IC(offset))); } } // 1. save the nonvolatile FPRs if (ir.compiledMethod.isSaveVolatile()) { // DANGER: as an optimization, we assume that a SaveVolatile method // will never use nonvolatile FPRs. // this invariant is not checked!!!!! // TODO: We really need some way to verify that this is true. } else { int nNonvolatileFPRS = ir.compiledMethod.getNumberOfNonvolatileFPRs(); n = nNonvolatileFPRS - 1; // use a sequence of load instructions for (Enumeration<Register> e = phys.enumerateNonvolatileFPRsBackwards(); e.hasMoreElements() && n >= 0; n--) { Register nv = e.nextElement(); int offset = getNonvolatileFPROffset(n); inst.insertBefore(MIR_Store.create(PPC_STFD, D(nv), A(FP), IC(offset))); } } } /** * Insert code before a return instruction to restore the nonvolatile * registers. * * @param inst the return instruction */ private void restoreNonVolatiles(Instruction inst) { GenericPhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet(); int nNonvolatileGPRS = ir.compiledMethod.getNumberOfNonvolatileGPRs(); // 1. restore the nonvolatile GPRs int n = nNonvolatileGPRS - 1; Register FP = phys.getFP(); if (VM.BuildFor32Addr && n > MULTIPLE_CUTOFF) { // use an lm Register nv = null; for (Enumeration<Register> e = phys.enumerateNonvolatileGPRsBackwards(); e.hasMoreElements() && n >= 0; n--) { nv = e.nextElement(); } n++; RegisterOperand range = I(nv); // YUCK!!! Why is this crap in register operand?? int offset = getNonvolatileGPROffset(n); inst.insertBefore(MIR_Load.create(PPC_LMW, range, A(FP), IC(offset))); } else { for (Enumeration<Register> e = phys.enumerateNonvolatileGPRsBackwards(); e.hasMoreElements() && n >= 0; n--) { Register nv = e.nextElement(); int offset = getNonvolatileGPROffset(n); inst.insertBefore(MIR_Load.create(PPC_LAddr, A(nv), A(FP), IC(offset))); } } // Note that save-volatiles are forbidden from using nonvolatile FPRs. if (!ir.compiledMethod.isSaveVolatile()) { // 1. restore the nonvolatile FPRs int nNonvolatileFPRS = ir.compiledMethod.getNumberOfNonvolatileFPRs(); n = nNonvolatileFPRS - 1; // use a sequence of load instructions for (Enumeration<Register> e = phys.enumerateNonvolatileFPRsBackwards(); e.hasMoreElements() && n >= 0; n--) { Register nv = e.nextElement(); int offset = getNonvolatileFPROffset(n); inst.insertBefore(MIR_Load.create(PPC_LFD, D(nv), A(FP), IC(offset))); } } } /** * Insert code before a return instruction to restore the * volatile registers. * * @param inst the return instruction */ private void restoreVolatileRegisters(Instruction inst) { PhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet().asPPC(); // 1. restore the volatile GPRs Register FP = phys.getFP(); int i = 0; for (Enumeration<Register> e = phys.enumerateVolatileGPRs(); e.hasMoreElements(); i++) { Register r = e.nextElement(); int location = saveVolatileGPRLocation[i]; inst.insertBefore(MIR_Load.create(PPC_LAddr, A(r), A(FP), IC(location))); } // 2. restore the volatile FPRs i = 0; for (Enumeration<Register> e = phys.enumerateVolatileFPRs(); e.hasMoreElements(); i++) { Register r = e.nextElement(); int location = saveVolatileFPRLocation[i]; inst.insertBefore(MIR_Load.create(PPC_LFD, D(r), A(FP), IC(location))); } // 3. Restore some special registers Register temp = phys.getTemp(); inst.insertBefore(MIR_Load.create(PPC_LInt, I(temp), A(FP), IC(saveXERLocation))); inst.insertBefore(MIR_Move.create(PPC_MTSPR, I(phys.getXER()), I(temp))); inst.insertBefore(MIR_Load.create(PPC_LAddr, A(temp), A(FP), IC(saveCTRLocation))); inst.insertBefore(MIR_Move.create(PPC_MTSPR, A(phys.getCTR()), A(temp))); } /* * Insert the prologue. * The available scratch registers are normally: R0, S0, S1 * However, if this is the prologue for a 'save volatile' frame, * then R0 is the only available scratch register. * The "normal" prologue must perform the following tasks: * stack overflow check * set TSR for the yieldpoint if there is a prologue yieldpoint instruction * save lr * store cmid * buy stack frame * store any used non volatiles * We schedule the prologue for this combination of operations, * since it is currently the common case. * When this changes, this code should be modifed accordingly. * The desired sequence is: * 1 mflr 00 # return addr * 2 l S1 takeYieldpointOffset(PR) # setting TSR for yield point * 3 stu FP -frameSize(FP) # buy frame, save caller's fp * 4 l S0 stackLimitOffset(S0) # stack overflow check * 5 <save used non volatiles> * 6 cmpi TSR S1 0x0 # setting TSR for yield point (S1 is now free) * 7 lil S1 CMID # cmid * 8 st 00 STACKFRAME_RETURN_ADDRESS_OFFSET(FP) # return addr (00 is now free) * 9 st S1 STACKFRAME_METHOD_ID_OFFSET(FP) # cmid * 10 tgt S0, FP # stack overflow check (already bought frame) */ /** * Schedule prologue for 'normal' case (see above) */ @Override public void insertNormalPrologue() { PhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet().asPPC(); Register FP = phys.getFP(); Register TR = phys.getTR(); Register TSR = phys.getTSR(); Register R0 = phys.getTemp(); Register S0 = phys.getGPR(FIRST_SCRATCH_GPR); Register S1 = phys.getGPR(LAST_SCRATCH_GPR); boolean interruptible = ir.method.isInterruptible(); boolean stackOverflow = interruptible; boolean yp = hasPrologueYieldpoint(); int frameFixedSize = getFrameFixedSize(); ir.compiledMethod.setFrameFixedSize(frameFixedSize); if (frameFixedSize >= STACK_SIZE_GUARD || ir.compiledMethod.isSaveVolatile()) { insertExceptionalPrologue(); return; } Instruction ptr = ir.firstInstructionInCodeOrder().nextInstructionInCodeOrder(); if (VM.VerifyAssertions) VM._assert(ptr.getOpcode() == IR_PROLOGUE_opcode); ptr.insertBefore(MIR_Move.create(PPC_MFSPR, A(R0), A(phys.getLR()))); // 1 if (yp) { Offset offset = Entrypoints.takeYieldpointField.getOffset(); if (VM.VerifyAssertions) VM._assert(fits(offset, 16)); ptr.insertBefore(MIR_Load.create(PPC_LInt, I(S1), A(TR), IC(PPCMaskLower16(offset)))); // 2 } ptr.insertBefore(MIR_StoreUpdate.create(PPC_STAddrU, A(FP), A(FP), IC(-frameSize))); // 3 if (stackOverflow) { Offset offset = Entrypoints.stackLimitField.getOffset(); if (VM.VerifyAssertions) VM._assert(fits(offset, 16)); ptr.insertBefore(MIR_Load.create(PPC_LAddr, A(S0), A(phys.getTR()), IC(PPCMaskLower16(offset)))); // 4 } // Now add any instructions to save the volatiles and nonvolatiles (5) saveNonVolatiles(ptr); if (yp) { ptr.insertBefore(MIR_Binary.create(PPC_CMPI, I(TSR), I(S1), IC(0))); // 6 } int cmid = ir.compiledMethod.getId(); if (cmid <= 0x7fff) { ptr.insertBefore(MIR_Unary.create(PPC_LDI, I(S1), IC(cmid))); // 7 } else { ptr.insertBefore(MIR_Unary.create(PPC_LDIS, I(S1), IC(cmid >>> 16))); // 7 (a) ptr.insertBefore(MIR_Binary.create(PPC_ORI, I(S1), I(S1), IC(cmid & 0xffff))); // 7 (b) } ptr.insertBefore(MIR_Store.create(PPC_STAddr, A(R0), A(FP), IC(frameSize + STACKFRAME_RETURN_ADDRESS_OFFSET.toInt()))); // 8 ptr.insertBefore(MIR_Store.create(PPC_STW, I(S1), A(FP), IC(STACKFRAME_METHOD_ID_OFFSET.toInt()))); // 9 if (stackOverflow) { // Mutate the Prologue instruction into the trap MIR_Trap.mutate(ptr, PPC_TAddr, PowerPCTrapOperand.GREATER(), A(S0), A(FP), TrapCodeOperand.StackOverflow()); // 10 } else { // no stack overflow test, so we remove the IR_Prologue instruction ptr.remove(); } } /** * prologue for the exceptional case. * (1) R0 is the only available scratch register. * (2) stack overflow check has to come first. */ void insertExceptionalPrologue() { if (VM.VerifyAssertions) { VM._assert((frameSize & (STACKFRAME_ALIGNMENT - 1)) == 0, "Stack frame alignment error"); } if (frameSize >= 0x7ff0) { throw new OptimizingCompilerException("Stackframe size exceeded!"); } PhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet().asPPC(); Register FP = phys.getFP(); Register TR = phys.getTR(); Register TSR = phys.getTSR(); Register R0 = phys.getTemp(); Register S1 = phys.getGPR(LAST_SCRATCH_GPR); boolean interruptible = ir.method.isInterruptible(); boolean stackOverflow = interruptible; boolean yp = hasPrologueYieldpoint(); Instruction ptr = ir.firstInstructionInCodeOrder().nextInstructionInCodeOrder(); if (VM.VerifyAssertions) VM._assert(ptr.getOpcode() == IR_PROLOGUE_opcode); // Stack overflow check if (stackOverflow) { // R0 is fairly useless (can't be operand 1 of an addi or the base ptr // of a load) so, free up S1 for use by briefly saving its contents in the // return address slot of my caller's frame ptr.insertBefore(MIR_Store.create(PPC_STAddr, A(S1), A(FP), IC(STACKFRAME_RETURN_ADDRESS_OFFSET.toInt()))); Offset offset = Entrypoints.stackLimitField.getOffset(); if (VM.VerifyAssertions) VM._assert(fits(offset, 16)); ptr.insertBefore(MIR_Load.create(PPC_LAddr, A(S1), A(phys.getTR()), IC(PPCMaskLower16(offset)))); ptr.insertBefore(MIR_Binary.create(PPC_ADDI, A(R0), A(S1), IC(frameSize))); ptr.insertBefore(MIR_Load.create(PPC_LAddr, A(S1), A(FP), IC(STACKFRAME_RETURN_ADDRESS_OFFSET.toInt()))); // Mutate the Prologue holder instruction into the trap MIR_Trap.mutate(ptr, PPC_TAddr, PowerPCTrapOperand.LESS(), A(FP), A(R0), TrapCodeOperand.StackOverflow()); // advance ptr because we want the remaining instructions to come after // the trap ptr = ptr.nextInstructionInCodeOrder(); } else { // no stack overflow test, so we must remove the IR_Prologue instruction Instruction next = ptr.nextInstructionInCodeOrder(); ptr.remove(); ptr = next; } // Buy stack frame, save LR, caller's FP ptr.insertBefore(MIR_Move.create(PPC_MFSPR, A(R0), A(phys.getLR()))); ptr.insertBefore(MIR_StoreUpdate.create(PPC_STAddrU, A(FP), A(FP), IC(-frameSize))); ptr.insertBefore(MIR_Store.create(PPC_STAddr, A(R0), A(FP), IC(frameSize + STACKFRAME_RETURN_ADDRESS_OFFSET.toInt()))); // Store cmid int cmid = ir.compiledMethod.getId(); if (cmid <= 0x7fff) { ptr.insertBefore(MIR_Unary.create(PPC_LDI, I(R0), IC(cmid))); } else { ptr.insertBefore(MIR_Unary.create(PPC_LDIS, I(R0), IC(cmid >>> 16))); ptr.insertBefore(MIR_Binary.create(PPC_ORI, I(R0), I(R0), IC(cmid & 0xffff))); } ptr.insertBefore(MIR_Store.create(PPC_STW, I(R0), A(FP), IC(STACKFRAME_METHOD_ID_OFFSET.toInt()))); // Now save the volatile/nonvolatile registers if (ir.compiledMethod.isSaveVolatile()) { saveVolatiles(ptr); } saveNonVolatiles(ptr); // Threadswitch if (yp) { Offset offset = Entrypoints.takeYieldpointField.getOffset(); if (VM.VerifyAssertions) VM._assert(fits(offset, 16)); ptr.insertBefore(MIR_Load.create(PPC_LInt, I(R0), A(TR), IC(PPCMaskLower16(offset)))); ptr.insertBefore(MIR_Binary.create(PPC_CMPI, I(TSR), I(R0), IC(0))); } } /** * Compute the number of stack words needed to hold nonvolatile * registers. * * Side effects: * <ul> * <li> updates the OptCompiler structure * <li> updates the <code>frameSize</code> field of this object * <li> updates the <code>frameRequired</code> field of this object * </ul> */ @Override public void computeNonVolatileArea() { GenericPhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet(); if (ir.compiledMethod.isSaveVolatile()) { // Record that we use every nonvolatile GPR int numGprNv = PhysicalRegisterSet.getNumberOfNonvolatileGPRs(); ir.compiledMethod.setNumberOfNonvolatileGPRs((short) numGprNv); // set the frame size frameSize += numGprNv * BYTES_IN_ADDRESS; int numFprNv = PhysicalRegisterSet.getNumberOfNonvolatileFPRs(); ir.compiledMethod.setNumberOfNonvolatileFPRs((short) numFprNv); frameSize += numFprNv * BYTES_IN_DOUBLE; // Record that we need a stack frame. setFrameRequired(); // Map each volatile GPR to a spill location. int i = 0; for (Enumeration<Register> e = phys.enumerateVolatileGPRs(); e.hasMoreElements(); i++) { e.nextElement(); // Note that as a side effect, the following call bumps up the // frame size. saveVolatileGPRLocation[i] = allocateNewSpillLocation(INT_REG); } // Map each non-volatile GPR register to a spill location. i = 0; for (Enumeration<Register> e = phys.enumerateNonvolatileGPRs(); e.hasMoreElements(); i++) { e.nextElement(); // Note that as a side effect, the following call bumps up the // frame size. nonVolatileGPRLocation[i] = allocateNewSpillLocation(INT_REG); } // Map some special registers to spill locations. saveXERLocation = allocateNewSpillLocation(INT_REG); saveCTRLocation = allocateNewSpillLocation(INT_REG); i = 0; for (Enumeration<Register> e = phys.enumerateVolatileFPRs(); e.hasMoreElements(); i++) { e.nextElement(); // Note that as a side effect, the following call bumps up the // frame size. saveVolatileFPRLocation[i] = allocateNewSpillLocation(DOUBLE_REG); } // Set the offset to find non-volatiles. int gprOffset = getNonvolatileGPROffset(0); ir.compiledMethod.setUnsignedNonVolatileOffset(gprOffset); } else { // Count the number of nonvolatiles used. int numGprNv = 0; int i = 0; for (Enumeration<Register> e = phys.enumerateNonvolatileGPRs(); e.hasMoreElements();) { Register r = e.nextElement(); if (r.isTouched()) { // Note that as a side effect, the following call bumps up the // frame size. nonVolatileGPRLocation[i++] = allocateNewSpillLocation(INT_REG); numGprNv++; } } i = 0; int numFprNv = 0; for (Enumeration<Register> e = phys.enumerateNonvolatileFPRs(); e.hasMoreElements();) { Register r = e.nextElement(); if (r.isTouched()) { // Note that as a side effect, the following call bumps up the // frame size. nonVolatileFPRLocation[i++] = allocateNewSpillLocation(DOUBLE_REG); numFprNv++; } } // Update the OptCompiledMethod object. ir.compiledMethod.setNumberOfNonvolatileGPRs((short) numGprNv); ir.compiledMethod.setNumberOfNonvolatileFPRs((short) numFprNv); if (numGprNv > 0 || numFprNv > 0) { int gprOffset = getNonvolatileGPROffset(0); ir.compiledMethod.setUnsignedNonVolatileOffset(gprOffset); // record that we need a stack frame setFrameRequired(); } else { ir.compiledMethod.setUnsignedNonVolatileOffset(0); } } frameSize = align(frameSize, STACKFRAME_ALIGNMENT); } /** * Walk over the currently available scratch registers. * * <p>For any scratch register r which is def'ed by instruction s, * spill r before s and remove r from the pool of available scratch * registers. * * <p>For any scratch register r which is used by instruction s, * restore r before s and remove r from the pool of available scratch * registers. * * <p>For any scratch register r which has current contents symb, and * symb is spilled to location M, and s defs M: the old value of symb is * dead. Mark this. * * <p>Invalidate any scratch register assignments that are illegal in s. */ @Override public void restoreScratchRegistersBefore(Instruction s) { for (Iterator<ScratchRegister> i = scratchInUse.iterator(); i.hasNext();) { ScratchRegister scratch = i.next(); if (scratch.getCurrentContents() == null) continue; if (VERBOSE_DEBUG) { System.out.println("RESTORE: consider " + scratch); } boolean removed = false; boolean unloaded = false; if (definedIn(scratch.scratch, s) || (s.isCall() && s.operator() != CALL_SAVE_VOLATILE && scratch.scratch.isVolatile())) { // s defines the scratch register, so save its contents before they // are killed. if (VERBOSE_DEBUG) { System.out.println("RESTORE : unload because defined " + scratch); } unloadScratchRegisterBefore(s, scratch); // update mapping information if (VERBOSE_DEBUG) { System.out.println("RSRB: End scratch interval " + scratch.scratch + " " + s); } scratchMap.endScratchInterval(scratch.scratch, s); Register scratchContents = scratch.getCurrentContents(); if (scratchContents != null) { if (VERBOSE_DEBUG) { System.out.println("RSRB: End symbolic interval " + scratch.getCurrentContents() + " " + s); } scratchMap.endSymbolicInterval(scratch.getCurrentContents(), s); } i.remove(); removed = true; unloaded = true; } if (usedIn(scratch.scratch, s) || !isLegal(scratch.getCurrentContents(), scratch.scratch, s)) { // first spill the currents contents of the scratch register to // memory if (!unloaded) { if (VERBOSE_DEBUG) { System.out.println("RESTORE : unload because used " + scratch); } unloadScratchRegisterBefore(s, scratch); // update mapping information if (VERBOSE_DEBUG) { System.out.println("RSRB2: End scratch interval " + scratch.scratch + " " + s); } scratchMap.endScratchInterval(scratch.scratch, s); Register scratchContents = scratch.getCurrentContents(); if (scratchContents != null) { if (VERBOSE_DEBUG) { System.out.println("RSRB2: End symbolic interval " + scratch.getCurrentContents() + " " + s); } scratchMap.endSymbolicInterval(scratch.getCurrentContents(), s); } } // s or some future instruction uses the scratch register, // so restore the correct contents. if (VERBOSE_DEBUG) { System.out.println("RESTORE : reload because used " + scratch); } reloadScratchRegisterBefore(s, scratch); if (!removed) { i.remove(); removed = true; } } } } protected static final int MULTIPLE_CUTOFF = 4; /** * Initializes the "tmp" regs for this object * @param ir the governing ir */ @Override public void initForArch(IR ir) { PhysicalRegisterSet phys = ir.regpool.getPhysicalRegisterSet().asPPC(); phys.getJTOC().reserveRegister(); phys.getFirstConditionRegister().reserveRegister(); } /** * Is a particular instruction a system call? */ @Override public boolean isSysCall(Instruction s) { return s.operator() == PPC_BCTRL_SYS || s.operator() == PPC_BL_SYS; } /** * Given symbolic register r in instruction s, do we need to ensure that * r is in a scratch register is s (as opposed to a memory operand) */ @Override public boolean needScratch(Register r, Instruction s) { if (s.operator() == YIELDPOINT_OSR) return false; // PowerPC does not support memory operands. return true; } /** * In instruction s, replace all appearances of a symbolic register * operand with uses of the appropriate spill location, as cached by the * register allocator. * * @param s the instruction to mutate. * @param symb the symbolic register operand to replace */ @Override public void replaceOperandWithSpillLocation(Instruction s, RegisterOperand symb) { // PowerPC does not support memory operands. if (VM.VerifyAssertions) VM._assert(NOT_REACHED); } }