{-# LANGUAGE CPP, GADTs #-} ----------------------------------------------------------------------------- -- -- Generating machine code (instruction selection) -- -- (c) The University of Glasgow 1996-2004 -- ----------------------------------------------------------------------------- -- This is a big module, but, if you pay attention to -- (a) the sectioning, and (b) the type signatures, -- the structure should not be too overwhelming. module PPC.CodeGen ( cmmTopCodeGen, generateJumpTableForInstr, InstrBlock ) where #include "HsVersions.h" #include "nativeGen/NCG.h" #include "MachDeps.h" -- NCG stuff: import GhcPrelude import CodeGen.Platform import PPC.Instr import PPC.Cond import PPC.Regs import CPrim import NCGMonad import Instruction import PIC import Format import RegClass import Reg import TargetReg import Platform -- Our intermediate code: import BlockId import PprCmm ( pprExpr ) import Cmm import CmmUtils import CmmSwitch import CLabel import Hoopl.Block import Hoopl.Graph -- The rest: import OrdList import Outputable import DynFlags import Control.Monad ( mapAndUnzipM, when ) import Data.Bits import Data.Word import BasicTypes import FastString import Util -- ----------------------------------------------------------------------------- -- Top-level of the instruction selector -- | 'InstrBlock's are the insn sequences generated by the insn selectors. -- They are really trees of insns to facilitate fast appending, where a -- left-to-right traversal (pre-order?) yields the insns in the correct -- order. cmmTopCodeGen :: RawCmmDecl -> NatM [NatCmmDecl CmmStatics Instr] cmmTopCodeGen (CmmProc info lab live graph) = do let blocks = toBlockListEntryFirst graph (nat_blocks,statics) <- mapAndUnzipM basicBlockCodeGen blocks dflags <- getDynFlags let proc = CmmProc info lab live (ListGraph $ concat nat_blocks) tops = proc : concat statics os = platformOS $ targetPlatform dflags arch = platformArch $ targetPlatform dflags case arch of ArchPPC | os == OSAIX -> return tops | otherwise -> do picBaseMb <- getPicBaseMaybeNat case picBaseMb of Just picBase -> initializePicBase_ppc arch os picBase tops Nothing -> return tops ArchPPC_64 ELF_V1 -> fixup_entry tops -- generating function descriptor is handled in -- pretty printer ArchPPC_64 ELF_V2 -> fixup_entry tops -- generating function prologue is handled in -- pretty printer _ -> panic "PPC.cmmTopCodeGen: unknown arch" where fixup_entry (CmmProc info lab live (ListGraph (entry:blocks)) : statics) = do let BasicBlock bID insns = entry bID' <- if lab == (blockLbl bID) then newBlockId else return bID let b' = BasicBlock bID' insns return (CmmProc info lab live (ListGraph (b':blocks)) : statics) fixup_entry _ = panic "cmmTopCodegen: Broken CmmProc" cmmTopCodeGen (CmmData sec dat) = do return [CmmData sec dat] -- no translation, we just use CmmStatic basicBlockCodeGen :: Block CmmNode C C -> NatM ( [NatBasicBlock Instr] , [NatCmmDecl CmmStatics Instr]) basicBlockCodeGen block = do let (_, nodes, tail) = blockSplit block id = entryLabel block stmts = blockToList nodes mid_instrs <- stmtsToInstrs stmts tail_instrs <- stmtToInstrs tail let instrs = mid_instrs `appOL` tail_instrs -- code generation may introduce new basic block boundaries, which -- are indicated by the NEWBLOCK instruction. We must split up the -- instruction stream into basic blocks again. Also, we extract -- LDATAs here too. let (top,other_blocks,statics) = foldrOL mkBlocks ([],[],[]) instrs mkBlocks (NEWBLOCK id) (instrs,blocks,statics) = ([], BasicBlock id instrs : blocks, statics) mkBlocks (LDATA sec dat) (instrs,blocks,statics) = (instrs, blocks, CmmData sec dat:statics) mkBlocks instr (instrs,blocks,statics) = (instr:instrs, blocks, statics) return (BasicBlock id top : other_blocks, statics) stmtsToInstrs :: [CmmNode e x] -> NatM InstrBlock stmtsToInstrs stmts = do instrss <- mapM stmtToInstrs stmts return (concatOL instrss) stmtToInstrs :: CmmNode e x -> NatM InstrBlock stmtToInstrs stmt = do dflags <- getDynFlags case stmt of CmmComment s -> return (unitOL (COMMENT s)) CmmTick {} -> return nilOL CmmUnwind {} -> return nilOL CmmAssign reg src | isFloatType ty -> assignReg_FltCode format reg src | target32Bit (targetPlatform dflags) && isWord64 ty -> assignReg_I64Code reg src | otherwise -> assignReg_IntCode format reg src where ty = cmmRegType dflags reg format = cmmTypeFormat ty CmmStore addr src | isFloatType ty -> assignMem_FltCode format addr src | target32Bit (targetPlatform dflags) && isWord64 ty -> assignMem_I64Code addr src | otherwise -> assignMem_IntCode format addr src where ty = cmmExprType dflags src format = cmmTypeFormat ty CmmUnsafeForeignCall target result_regs args -> genCCall target result_regs args CmmBranch id -> genBranch id CmmCondBranch arg true false prediction -> do b1 <- genCondJump true arg prediction b2 <- genBranch false return (b1 `appOL` b2) CmmSwitch arg ids -> do dflags <- getDynFlags genSwitch dflags arg ids CmmCall { cml_target = arg } -> genJump arg _ -> panic "stmtToInstrs: statement should have been cps'd away" -------------------------------------------------------------------------------- -- | 'InstrBlock's are the insn sequences generated by the insn selectors. -- They are really trees of insns to facilitate fast appending, where a -- left-to-right traversal yields the insns in the correct order. -- type InstrBlock = OrdList Instr -- | Register's passed up the tree. If the stix code forces the register -- to live in a pre-decided machine register, it comes out as @Fixed@; -- otherwise, it comes out as @Any@, and the parent can decide which -- register to put it in. -- data Register = Fixed Format Reg InstrBlock | Any Format (Reg -> InstrBlock) swizzleRegisterRep :: Register -> Format -> Register swizzleRegisterRep (Fixed _ reg code) format = Fixed format reg code swizzleRegisterRep (Any _ codefn) format = Any format codefn -- | Grab the Reg for a CmmReg getRegisterReg :: Platform -> CmmReg -> Reg getRegisterReg _ (CmmLocal (LocalReg u pk)) = RegVirtual $ mkVirtualReg u (cmmTypeFormat pk) getRegisterReg platform (CmmGlobal mid) = case globalRegMaybe platform mid of Just reg -> RegReal reg Nothing -> pprPanic "getRegisterReg-memory" (ppr $ CmmGlobal mid) -- By this stage, the only MagicIds remaining should be the -- ones which map to a real machine register on this -- platform. Hence ... -- | Convert a BlockId to some CmmStatic data jumpTableEntry :: DynFlags -> Maybe BlockId -> CmmStatic jumpTableEntry dflags Nothing = CmmStaticLit (CmmInt 0 (wordWidth dflags)) jumpTableEntry _ (Just blockid) = CmmStaticLit (CmmLabel blockLabel) where blockLabel = blockLbl blockid -- ----------------------------------------------------------------------------- -- General things for putting together code sequences -- Expand CmmRegOff. ToDo: should we do it this way around, or convert -- CmmExprs into CmmRegOff? mangleIndexTree :: DynFlags -> CmmExpr -> CmmExpr mangleIndexTree dflags (CmmRegOff reg off) = CmmMachOp (MO_Add width) [CmmReg reg, CmmLit (CmmInt (fromIntegral off) width)] where width = typeWidth (cmmRegType dflags reg) mangleIndexTree _ _ = panic "PPC.CodeGen.mangleIndexTree: no match" -- ----------------------------------------------------------------------------- -- Code gen for 64-bit arithmetic on 32-bit platforms {- Simple support for generating 64-bit code (ie, 64 bit values and 64 bit assignments) on 32-bit platforms. Unlike the main code generator we merely shoot for generating working code as simply as possible, and pay little attention to code quality. Specifically, there is no attempt to deal cleverly with the fixed-vs-floating register distinction; all values are generated into (pairs of) floating registers, even if this would mean some redundant reg-reg moves as a result. Only one of the VRegUniques is returned, since it will be of the VRegUniqueLo form, and the upper-half VReg can be determined by applying getHiVRegFromLo to it. -} data ChildCode64 -- a.k.a "Register64" = ChildCode64 InstrBlock -- code Reg -- the lower 32-bit temporary which contains the -- result; use getHiVRegFromLo to find the other -- VRegUnique. Rules of this simplified insn -- selection game are therefore that the returned -- Reg may be modified -- | Compute an expression into a register, but -- we don't mind which one it is. getSomeReg :: CmmExpr -> NatM (Reg, InstrBlock) getSomeReg expr = do r <- getRegister expr case r of Any rep code -> do tmp <- getNewRegNat rep return (tmp, code tmp) Fixed _ reg code -> return (reg, code) getI64Amodes :: CmmExpr -> NatM (AddrMode, AddrMode, InstrBlock) getI64Amodes addrTree = do Amode hi_addr addr_code <- getAmode D addrTree case addrOffset hi_addr 4 of Just lo_addr -> return (hi_addr, lo_addr, addr_code) Nothing -> do (hi_ptr, code) <- getSomeReg addrTree return (AddrRegImm hi_ptr (ImmInt 0), AddrRegImm hi_ptr (ImmInt 4), code) assignMem_I64Code :: CmmExpr -> CmmExpr -> NatM InstrBlock assignMem_I64Code addrTree valueTree = do (hi_addr, lo_addr, addr_code) <- getI64Amodes addrTree ChildCode64 vcode rlo <- iselExpr64 valueTree let rhi = getHiVRegFromLo rlo -- Big-endian store mov_hi = ST II32 rhi hi_addr mov_lo = ST II32 rlo lo_addr return (vcode `appOL` addr_code `snocOL` mov_lo `snocOL` mov_hi) assignReg_I64Code :: CmmReg -> CmmExpr -> NatM InstrBlock assignReg_I64Code (CmmLocal (LocalReg u_dst _)) valueTree = do ChildCode64 vcode r_src_lo <- iselExpr64 valueTree let r_dst_lo = RegVirtual $ mkVirtualReg u_dst II32 r_dst_hi = getHiVRegFromLo r_dst_lo r_src_hi = getHiVRegFromLo r_src_lo mov_lo = MR r_dst_lo r_src_lo mov_hi = MR r_dst_hi r_src_hi return ( vcode `snocOL` mov_lo `snocOL` mov_hi ) assignReg_I64Code _ _ = panic "assignReg_I64Code(powerpc): invalid lvalue" iselExpr64 :: CmmExpr -> NatM ChildCode64 iselExpr64 (CmmLoad addrTree ty) | isWord64 ty = do (hi_addr, lo_addr, addr_code) <- getI64Amodes addrTree (rlo, rhi) <- getNewRegPairNat II32 let mov_hi = LD II32 rhi hi_addr mov_lo = LD II32 rlo lo_addr return $ ChildCode64 (addr_code `snocOL` mov_lo `snocOL` mov_hi) rlo iselExpr64 (CmmReg (CmmLocal (LocalReg vu ty))) | isWord64 ty = return (ChildCode64 nilOL (RegVirtual $ mkVirtualReg vu II32)) iselExpr64 (CmmLit (CmmInt i _)) = do (rlo,rhi) <- getNewRegPairNat II32 let half0 = fromIntegral (fromIntegral i :: Word16) half1 = fromIntegral (fromIntegral (i `shiftR` 16) :: Word16) half2 = fromIntegral (fromIntegral (i `shiftR` 32) :: Word16) half3 = fromIntegral (fromIntegral (i `shiftR` 48) :: Word16) code = toOL [ LIS rlo (ImmInt half1), OR rlo rlo (RIImm $ ImmInt half0), LIS rhi (ImmInt half3), OR rhi rhi (RIImm $ ImmInt half2) ] return (ChildCode64 code rlo) iselExpr64 (CmmMachOp (MO_Add _) [e1,e2]) = do ChildCode64 code1 r1lo <- iselExpr64 e1 ChildCode64 code2 r2lo <- iselExpr64 e2 (rlo,rhi) <- getNewRegPairNat II32 let r1hi = getHiVRegFromLo r1lo r2hi = getHiVRegFromLo r2lo code = code1 `appOL` code2 `appOL` toOL [ ADDC rlo r1lo r2lo, ADDE rhi r1hi r2hi ] return (ChildCode64 code rlo) iselExpr64 (CmmMachOp (MO_Sub _) [e1,e2]) = do ChildCode64 code1 r1lo <- iselExpr64 e1 ChildCode64 code2 r2lo <- iselExpr64 e2 (rlo,rhi) <- getNewRegPairNat II32 let r1hi = getHiVRegFromLo r1lo r2hi = getHiVRegFromLo r2lo code = code1 `appOL` code2 `appOL` toOL [ SUBFC rlo r2lo (RIReg r1lo), SUBFE rhi r2hi r1hi ] return (ChildCode64 code rlo) iselExpr64 (CmmMachOp (MO_UU_Conv W32 W64) [expr]) = do (expr_reg,expr_code) <- getSomeReg expr (rlo, rhi) <- getNewRegPairNat II32 let mov_hi = LI rhi (ImmInt 0) mov_lo = MR rlo expr_reg return $ ChildCode64 (expr_code `snocOL` mov_lo `snocOL` mov_hi) rlo iselExpr64 (CmmMachOp (MO_SS_Conv W32 W64) [expr]) = do (expr_reg,expr_code) <- getSomeReg expr (rlo, rhi) <- getNewRegPairNat II32 let mov_hi = SRA II32 rhi expr_reg (RIImm (ImmInt 31)) mov_lo = MR rlo expr_reg return $ ChildCode64 (expr_code `snocOL` mov_lo `snocOL` mov_hi) rlo iselExpr64 expr = pprPanic "iselExpr64(powerpc)" (pprExpr expr) getRegister :: CmmExpr -> NatM Register getRegister e = do dflags <- getDynFlags getRegister' dflags e getRegister' :: DynFlags -> CmmExpr -> NatM Register getRegister' dflags (CmmReg (CmmGlobal PicBaseReg)) | OSAIX <- platformOS (targetPlatform dflags) = do let code dst = toOL [ LD II32 dst tocAddr ] tocAddr = AddrRegImm toc (ImmLit (text "ghc_toc_table[TC]")) return (Any II32 code) | target32Bit (targetPlatform dflags) = do reg <- getPicBaseNat $ archWordFormat (target32Bit (targetPlatform dflags)) return (Fixed (archWordFormat (target32Bit (targetPlatform dflags))) reg nilOL) | otherwise = return (Fixed II64 toc nilOL) getRegister' dflags (CmmReg reg) = return (Fixed (cmmTypeFormat (cmmRegType dflags reg)) (getRegisterReg (targetPlatform dflags) reg) nilOL) getRegister' dflags tree@(CmmRegOff _ _) = getRegister' dflags (mangleIndexTree dflags tree) -- for 32-bit architectuers, support some 64 -> 32 bit conversions: -- TO_W_(x), TO_W_(x >> 32) getRegister' dflags (CmmMachOp (MO_UU_Conv W64 W32) [CmmMachOp (MO_U_Shr W64) [x,CmmLit (CmmInt 32 _)]]) | target32Bit (targetPlatform dflags) = do ChildCode64 code rlo <- iselExpr64 x return $ Fixed II32 (getHiVRegFromLo rlo) code getRegister' dflags (CmmMachOp (MO_SS_Conv W64 W32) [CmmMachOp (MO_U_Shr W64) [x,CmmLit (CmmInt 32 _)]]) | target32Bit (targetPlatform dflags) = do ChildCode64 code rlo <- iselExpr64 x return $ Fixed II32 (getHiVRegFromLo rlo) code getRegister' dflags (CmmMachOp (MO_UU_Conv W64 W32) [x]) | target32Bit (targetPlatform dflags) = do ChildCode64 code rlo <- iselExpr64 x return $ Fixed II32 rlo code getRegister' dflags (CmmMachOp (MO_SS_Conv W64 W32) [x]) | target32Bit (targetPlatform dflags) = do ChildCode64 code rlo <- iselExpr64 x return $ Fixed II32 rlo code getRegister' dflags (CmmLoad mem pk) | not (isWord64 pk) = do let platform = targetPlatform dflags Amode addr addr_code <- getAmode D mem let code dst = ASSERT((targetClassOfReg platform dst == RcDouble) == isFloatType pk) addr_code `snocOL` LD format dst addr return (Any format code) | not (target32Bit (targetPlatform dflags)) = do Amode addr addr_code <- getAmode DS mem let code dst = addr_code `snocOL` LD II64 dst addr return (Any II64 code) where format = cmmTypeFormat pk -- catch simple cases of zero- or sign-extended load getRegister' _ (CmmMachOp (MO_UU_Conv W8 W32) [CmmLoad mem _]) = do Amode addr addr_code <- getAmode D mem return (Any II32 (\dst -> addr_code `snocOL` LD II8 dst addr)) getRegister' _ (CmmMachOp (MO_UU_Conv W8 W64) [CmmLoad mem _]) = do Amode addr addr_code <- getAmode D mem return (Any II64 (\dst -> addr_code `snocOL` LD II8 dst addr)) -- Note: there is no Load Byte Arithmetic instruction, so no signed case here getRegister' _ (CmmMachOp (MO_UU_Conv W16 W32) [CmmLoad mem _]) = do Amode addr addr_code <- getAmode D mem return (Any II32 (\dst -> addr_code `snocOL` LD II16 dst addr)) getRegister' _ (CmmMachOp (MO_SS_Conv W16 W32) [CmmLoad mem _]) = do Amode addr addr_code <- getAmode D mem return (Any II32 (\dst -> addr_code `snocOL` LA II16 dst addr)) getRegister' _ (CmmMachOp (MO_UU_Conv W16 W64) [CmmLoad mem _]) = do Amode addr addr_code <- getAmode D mem return (Any II64 (\dst -> addr_code `snocOL` LD II16 dst addr)) getRegister' _ (CmmMachOp (MO_SS_Conv W16 W64) [CmmLoad mem _]) = do Amode addr addr_code <- getAmode D mem return (Any II64 (\dst -> addr_code `snocOL` LA II16 dst addr)) getRegister' _ (CmmMachOp (MO_UU_Conv W32 W64) [CmmLoad mem _]) = do Amode addr addr_code <- getAmode D mem return (Any II64 (\dst -> addr_code `snocOL` LD II32 dst addr)) getRegister' _ (CmmMachOp (MO_SS_Conv W32 W64) [CmmLoad mem _]) = do -- lwa is DS-form. See Note [Power instruction format] Amode addr addr_code <- getAmode DS mem return (Any II64 (\dst -> addr_code `snocOL` LA II32 dst addr)) getRegister' dflags (CmmMachOp mop [x]) -- unary MachOps = case mop of MO_Not rep -> triv_ucode_int rep NOT MO_F_Neg w -> triv_ucode_float w FNEG MO_S_Neg w -> triv_ucode_int w NEG MO_FF_Conv W64 W32 -> trivialUCode FF32 FRSP x MO_FF_Conv W32 W64 -> conversionNop FF64 x MO_FS_Conv from to -> coerceFP2Int from to x MO_SF_Conv from to -> coerceInt2FP from to x MO_SS_Conv from to | from >= to -> conversionNop (intFormat to) x | otherwise -> triv_ucode_int to (EXTS (intFormat from)) MO_UU_Conv from to | from >= to -> conversionNop (intFormat to) x | otherwise -> clearLeft from to _ -> panic "PPC.CodeGen.getRegister: no match" where triv_ucode_int width instr = trivialUCode (intFormat width) instr x triv_ucode_float width instr = trivialUCode (floatFormat width) instr x conversionNop new_format expr = do e_code <- getRegister' dflags expr return (swizzleRegisterRep e_code new_format) clearLeft from to = do (src1, code1) <- getSomeReg x let arch_fmt = intFormat (wordWidth dflags) arch_bits = widthInBits (wordWidth dflags) size = widthInBits from code dst = code1 `snocOL` CLRLI arch_fmt dst src1 (arch_bits - size) return (Any (intFormat to) code) getRegister' _ (CmmMachOp mop [x, y]) -- dyadic PrimOps = case mop of MO_F_Eq _ -> condFltReg EQQ x y MO_F_Ne _ -> condFltReg NE x y MO_F_Gt _ -> condFltReg GTT x y MO_F_Ge _ -> condFltReg GE x y MO_F_Lt _ -> condFltReg LTT x y MO_F_Le _ -> condFltReg LE x y MO_Eq rep -> condIntReg EQQ rep x y MO_Ne rep -> condIntReg NE rep x y MO_S_Gt rep -> condIntReg GTT rep x y MO_S_Ge rep -> condIntReg GE rep x y MO_S_Lt rep -> condIntReg LTT rep x y MO_S_Le rep -> condIntReg LE rep x y MO_U_Gt rep -> condIntReg GU rep x y MO_U_Ge rep -> condIntReg GEU rep x y MO_U_Lt rep -> condIntReg LU rep x y MO_U_Le rep -> condIntReg LEU rep x y MO_F_Add w -> triv_float w FADD MO_F_Sub w -> triv_float w FSUB MO_F_Mul w -> triv_float w FMUL MO_F_Quot w -> triv_float w FDIV -- optimize addition with 32-bit immediate -- (needed for PIC) MO_Add W32 -> case y of CmmLit (CmmInt imm immrep) | Just _ <- makeImmediate W32 True imm -> trivialCode W32 True ADD x (CmmLit $ CmmInt imm immrep) CmmLit lit -> do (src, srcCode) <- getSomeReg x let imm = litToImm lit code dst = srcCode `appOL` toOL [ ADDIS dst src (HA imm), ADD dst dst (RIImm (LO imm)) ] return (Any II32 code) _ -> trivialCode W32 True ADD x y MO_Add rep -> trivialCode rep True ADD x y MO_Sub rep -> case y of CmmLit (CmmInt imm immrep) | Just _ <- makeImmediate rep True (-imm) -> trivialCode rep True ADD x (CmmLit $ CmmInt (-imm) immrep) _ -> case x of CmmLit (CmmInt imm _) | Just _ <- makeImmediate rep True imm -- subfi ('substract from' with immediate) doesn't exist -> trivialCode rep True SUBFC y x _ -> trivialCodeNoImm' (intFormat rep) SUBF y x MO_Mul rep -> shiftMulCode rep True MULL x y MO_S_MulMayOflo rep -> do (src1, code1) <- getSomeReg x (src2, code2) <- getSomeReg y let format = intFormat rep code dst = code1 `appOL` code2 `appOL` toOL [ MULLO format dst src1 src2 , MFOV format dst ] return (Any format code) MO_S_Quot rep -> divCode rep True x y MO_U_Quot rep -> divCode rep False x y MO_S_Rem rep -> remainderCode rep True x y MO_U_Rem rep -> remainderCode rep False x y MO_And rep -> case y of (CmmLit (CmmInt imm _)) | imm == -8 || imm == -4 -> do (src, srcCode) <- getSomeReg x let clear_mask = if imm == -4 then 2 else 3 fmt = intFormat rep code dst = srcCode `appOL` unitOL (CLRRI fmt dst src clear_mask) return (Any fmt code) _ -> trivialCode rep False AND x y MO_Or rep -> trivialCode rep False OR x y MO_Xor rep -> trivialCode rep False XOR x y MO_Shl rep -> shiftMulCode rep False SL x y MO_S_Shr rep -> srCode rep True SRA x y MO_U_Shr rep -> srCode rep False SR x y _ -> panic "PPC.CodeGen.getRegister: no match" where triv_float :: Width -> (Format -> Reg -> Reg -> Reg -> Instr) -> NatM Register triv_float width instr = trivialCodeNoImm (floatFormat width) instr x y getRegister' _ (CmmLit (CmmInt i rep)) | Just imm <- makeImmediate rep True i = let code dst = unitOL (LI dst imm) in return (Any (intFormat rep) code) getRegister' _ (CmmLit (CmmFloat f frep)) = do lbl <- getNewLabelNat dflags <- getDynFlags dynRef <- cmmMakeDynamicReference dflags DataReference lbl Amode addr addr_code <- getAmode D dynRef let format = floatFormat frep code dst = LDATA (Section ReadOnlyData lbl) (Statics lbl [CmmStaticLit (CmmFloat f frep)]) `consOL` (addr_code `snocOL` LD format dst addr) return (Any format code) getRegister' dflags (CmmLit lit) | target32Bit (targetPlatform dflags) = let rep = cmmLitType dflags lit imm = litToImm lit code dst = toOL [ LIS dst (HA imm), ADD dst dst (RIImm (LO imm)) ] in return (Any (cmmTypeFormat rep) code) | otherwise = do lbl <- getNewLabelNat dflags <- getDynFlags dynRef <- cmmMakeDynamicReference dflags DataReference lbl Amode addr addr_code <- getAmode D dynRef let rep = cmmLitType dflags lit format = cmmTypeFormat rep code dst = LDATA (Section ReadOnlyData lbl) (Statics lbl [CmmStaticLit lit]) `consOL` (addr_code `snocOL` LD format dst addr) return (Any format code) getRegister' _ other = pprPanic "getRegister(ppc)" (pprExpr other) -- extend?Rep: wrap integer expression of type `from` -- in a conversion to `to` extendSExpr :: Width -> Width -> CmmExpr -> CmmExpr extendSExpr from to x = CmmMachOp (MO_SS_Conv from to) [x] extendUExpr :: Width -> Width -> CmmExpr -> CmmExpr extendUExpr from to x = CmmMachOp (MO_UU_Conv from to) [x] -- ----------------------------------------------------------------------------- -- The 'Amode' type: Memory addressing modes passed up the tree. data Amode = Amode AddrMode InstrBlock {- Now, given a tree (the argument to a CmmLoad) that references memory, produce a suitable addressing mode. A Rule of the Game (tm) for Amodes: use of the addr bit must immediately follow use of the code part, since the code part puts values in registers which the addr then refers to. So you can't put anything in between, lest it overwrite some of those registers. If you need to do some other computation between the code part and use of the addr bit, first store the effective address from the amode in a temporary, then do the other computation, and then use the temporary: code LEA amode, tmp ... other computation ... ... (tmp) ... -} {- Note [Power instruction format] In some instructions the 16 bit offset must be a multiple of 4, i.e. the two least significant bits must be zero. The "Power ISA" specification calls these instruction formats "DS-FORM" and the instructions with arbitrary 16 bit offsets are "D-FORM". The Power ISA specification document can be obtained from www.power.org. -} data InstrForm = D | DS getAmode :: InstrForm -> CmmExpr -> NatM Amode getAmode inf tree@(CmmRegOff _ _) = do dflags <- getDynFlags getAmode inf (mangleIndexTree dflags tree) getAmode _ (CmmMachOp (MO_Sub W32) [x, CmmLit (CmmInt i _)]) | Just off <- makeImmediate W32 True (-i) = do (reg, code) <- getSomeReg x return (Amode (AddrRegImm reg off) code) getAmode _ (CmmMachOp (MO_Add W32) [x, CmmLit (CmmInt i _)]) | Just off <- makeImmediate W32 True i = do (reg, code) <- getSomeReg x return (Amode (AddrRegImm reg off) code) getAmode D (CmmMachOp (MO_Sub W64) [x, CmmLit (CmmInt i _)]) | Just off <- makeImmediate W64 True (-i) = do (reg, code) <- getSomeReg x return (Amode (AddrRegImm reg off) code) getAmode D (CmmMachOp (MO_Add W64) [x, CmmLit (CmmInt i _)]) | Just off <- makeImmediate W64 True i = do (reg, code) <- getSomeReg x return (Amode (AddrRegImm reg off) code) getAmode DS (CmmMachOp (MO_Sub W64) [x, CmmLit (CmmInt i _)]) | Just off <- makeImmediate W64 True (-i) = do (reg, code) <- getSomeReg x (reg', off', code') <- if i `mod` 4 == 0 then do return (reg, off, code) else do tmp <- getNewRegNat II64 return (tmp, ImmInt 0, code `snocOL` ADD tmp reg (RIImm off)) return (Amode (AddrRegImm reg' off') code') getAmode DS (CmmMachOp (MO_Add W64) [x, CmmLit (CmmInt i _)]) | Just off <- makeImmediate W64 True i = do (reg, code) <- getSomeReg x (reg', off', code') <- if i `mod` 4 == 0 then do return (reg, off, code) else do tmp <- getNewRegNat II64 return (tmp, ImmInt 0, code `snocOL` ADD tmp reg (RIImm off)) return (Amode (AddrRegImm reg' off') code') -- optimize addition with 32-bit immediate -- (needed for PIC) getAmode _ (CmmMachOp (MO_Add W32) [x, CmmLit lit]) = do dflags <- getDynFlags (src, srcCode) <- getSomeReg x let imm = litToImm lit case () of _ | OSAIX <- platformOS (targetPlatform dflags) , isCmmLabelType lit -> -- HA16/LO16 relocations on labels not supported on AIX return (Amode (AddrRegImm src imm) srcCode) | otherwise -> do tmp <- getNewRegNat II32 let code = srcCode `snocOL` ADDIS tmp src (HA imm) return (Amode (AddrRegImm tmp (LO imm)) code) where isCmmLabelType (CmmLabel {}) = True isCmmLabelType (CmmLabelOff {}) = True isCmmLabelType (CmmLabelDiffOff {}) = True isCmmLabelType _ = False getAmode _ (CmmLit lit) = do dflags <- getDynFlags case platformArch $ targetPlatform dflags of ArchPPC -> do tmp <- getNewRegNat II32 let imm = litToImm lit code = unitOL (LIS tmp (HA imm)) return (Amode (AddrRegImm tmp (LO imm)) code) _ -> do -- TODO: Load from TOC, -- see getRegister' _ (CmmLit lit) tmp <- getNewRegNat II64 let imm = litToImm lit code = toOL [ LIS tmp (HIGHESTA imm), OR tmp tmp (RIImm (HIGHERA imm)), SL II64 tmp tmp (RIImm (ImmInt 32)), ORIS tmp tmp (HA imm) ] return (Amode (AddrRegImm tmp (LO imm)) code) getAmode _ (CmmMachOp (MO_Add W32) [x, y]) = do (regX, codeX) <- getSomeReg x (regY, codeY) <- getSomeReg y return (Amode (AddrRegReg regX regY) (codeX `appOL` codeY)) getAmode _ (CmmMachOp (MO_Add W64) [x, y]) = do (regX, codeX) <- getSomeReg x (regY, codeY) <- getSomeReg y return (Amode (AddrRegReg regX regY) (codeX `appOL` codeY)) getAmode _ other = do (reg, code) <- getSomeReg other let off = ImmInt 0 return (Amode (AddrRegImm reg off) code) -- The 'CondCode' type: Condition codes passed up the tree. data CondCode = CondCode Bool Cond InstrBlock -- Set up a condition code for a conditional branch. getCondCode :: CmmExpr -> NatM CondCode -- almost the same as everywhere else - but we need to -- extend small integers to 32 bit or 64 bit first getCondCode (CmmMachOp mop [x, y]) = do case mop of MO_F_Eq W32 -> condFltCode EQQ x y MO_F_Ne W32 -> condFltCode NE x y MO_F_Gt W32 -> condFltCode GTT x y MO_F_Ge W32 -> condFltCode GE x y MO_F_Lt W32 -> condFltCode LTT x y MO_F_Le W32 -> condFltCode LE x y MO_F_Eq W64 -> condFltCode EQQ x y MO_F_Ne W64 -> condFltCode NE x y MO_F_Gt W64 -> condFltCode GTT x y MO_F_Ge W64 -> condFltCode GE x y MO_F_Lt W64 -> condFltCode LTT x y MO_F_Le W64 -> condFltCode LE x y MO_Eq rep -> condIntCode EQQ rep x y MO_Ne rep -> condIntCode NE rep x y MO_S_Gt rep -> condIntCode GTT rep x y MO_S_Ge rep -> condIntCode GE rep x y MO_S_Lt rep -> condIntCode LTT rep x y MO_S_Le rep -> condIntCode LE rep x y MO_U_Gt rep -> condIntCode GU rep x y MO_U_Ge rep -> condIntCode GEU rep x y MO_U_Lt rep -> condIntCode LU rep x y MO_U_Le rep -> condIntCode LEU rep x y _ -> pprPanic "getCondCode(powerpc)" (pprMachOp mop) getCondCode _ = panic "getCondCode(2)(powerpc)" -- @cond(Int|Flt)Code@: Turn a boolean expression into a condition, to be -- passed back up the tree. condIntCode :: Cond -> Width -> CmmExpr -> CmmExpr -> NatM CondCode condIntCode cond width x y = do dflags <- getDynFlags condIntCode' (target32Bit (targetPlatform dflags)) cond width x y condIntCode' :: Bool -> Cond -> Width -> CmmExpr -> CmmExpr -> NatM CondCode -- simple code for 64-bit on 32-bit platforms condIntCode' True cond W64 x y | condUnsigned cond = do ChildCode64 code_x x_lo <- iselExpr64 x ChildCode64 code_y y_lo <- iselExpr64 y let x_hi = getHiVRegFromLo x_lo y_hi = getHiVRegFromLo y_lo end_lbl <- getBlockIdNat let code = code_x `appOL` code_y `appOL` toOL [ CMPL II32 x_hi (RIReg y_hi) , BCC NE end_lbl Nothing , CMPL II32 x_lo (RIReg y_lo) , BCC ALWAYS end_lbl Nothing , NEWBLOCK end_lbl ] return (CondCode False cond code) | otherwise = do ChildCode64 code_x x_lo <- iselExpr64 x ChildCode64 code_y y_lo <- iselExpr64 y let x_hi = getHiVRegFromLo x_lo y_hi = getHiVRegFromLo y_lo end_lbl <- getBlockIdNat cmp_lo <- getBlockIdNat let code = code_x `appOL` code_y `appOL` toOL [ CMP II32 x_hi (RIReg y_hi) , BCC NE end_lbl Nothing , CMP II32 x_hi (RIImm (ImmInt 0)) , BCC LE cmp_lo Nothing , CMPL II32 x_lo (RIReg y_lo) , BCC ALWAYS end_lbl Nothing , CMPL II32 y_lo (RIReg x_lo) , BCC ALWAYS end_lbl Nothing , NEWBLOCK end_lbl ] return (CondCode False cond code) -- optimize pointer tag checks. Operation andi. sets condition register -- so cmpi ..., 0 is redundant. condIntCode' _ cond _ (CmmMachOp (MO_And _) [x, CmmLit (CmmInt imm rep)]) (CmmLit (CmmInt 0 _)) | not $ condUnsigned cond, Just src2 <- makeImmediate rep False imm = do (src1, code) <- getSomeReg x let code' = code `snocOL` AND r0 src1 (RIImm src2) return (CondCode False cond code') condIntCode' _ cond width x (CmmLit (CmmInt y rep)) | Just src2 <- makeImmediate rep (not $ condUnsigned cond) y = do let op_len = max W32 width let extend = extendSExpr width op_len (src1, code) <- getSomeReg (extend x) let format = intFormat op_len code' = code `snocOL` (if condUnsigned cond then CMPL else CMP) format src1 (RIImm src2) return (CondCode False cond code') condIntCode' _ cond width x y = do let op_len = max W32 width let extend = if condUnsigned cond then extendUExpr width op_len else extendSExpr width op_len (src1, code1) <- getSomeReg (extend x) (src2, code2) <- getSomeReg (extend y) let format = intFormat op_len code' = code1 `appOL` code2 `snocOL` (if condUnsigned cond then CMPL else CMP) format src1 (RIReg src2) return (CondCode False cond code') condFltCode :: Cond -> CmmExpr -> CmmExpr -> NatM CondCode condFltCode cond x y = do (src1, code1) <- getSomeReg x (src2, code2) <- getSomeReg y let code' = code1 `appOL` code2 `snocOL` FCMP src1 src2 code'' = case cond of -- twiddle CR to handle unordered case GE -> code' `snocOL` CRNOR ltbit eqbit gtbit LE -> code' `snocOL` CRNOR gtbit eqbit ltbit _ -> code' where ltbit = 0 ; eqbit = 2 ; gtbit = 1 return (CondCode True cond code'') -- ----------------------------------------------------------------------------- -- Generating assignments -- Assignments are really at the heart of the whole code generation -- business. Almost all top-level nodes of any real importance are -- assignments, which correspond to loads, stores, or register -- transfers. If we're really lucky, some of the register transfers -- will go away, because we can use the destination register to -- complete the code generation for the right hand side. This only -- fails when the right hand side is forced into a fixed register -- (e.g. the result of a call). assignMem_IntCode :: Format -> CmmExpr -> CmmExpr -> NatM InstrBlock assignReg_IntCode :: Format -> CmmReg -> CmmExpr -> NatM InstrBlock assignMem_FltCode :: Format -> CmmExpr -> CmmExpr -> NatM InstrBlock assignReg_FltCode :: Format -> CmmReg -> CmmExpr -> NatM InstrBlock assignMem_IntCode pk addr src = do (srcReg, code) <- getSomeReg src Amode dstAddr addr_code <- case pk of II64 -> getAmode DS addr _ -> getAmode D addr return $ code `appOL` addr_code `snocOL` ST pk srcReg dstAddr -- dst is a reg, but src could be anything assignReg_IntCode _ reg src = do dflags <- getDynFlags let dst = getRegisterReg (targetPlatform dflags) reg r <- getRegister src return $ case r of Any _ code -> code dst Fixed _ freg fcode -> fcode `snocOL` MR dst freg -- Easy, isn't it? assignMem_FltCode = assignMem_IntCode assignReg_FltCode = assignReg_IntCode genJump :: CmmExpr{-the branch target-} -> NatM InstrBlock genJump (CmmLit (CmmLabel lbl)) = return (unitOL $ JMP lbl) genJump tree = do dflags <- getDynFlags genJump' tree (platformToGCP (targetPlatform dflags)) genJump' :: CmmExpr -> GenCCallPlatform -> NatM InstrBlock genJump' tree (GCPLinux64ELF 1) = do (target,code) <- getSomeReg tree return (code `snocOL` LD II64 r11 (AddrRegImm target (ImmInt 0)) `snocOL` LD II64 toc (AddrRegImm target (ImmInt 8)) `snocOL` MTCTR r11 `snocOL` LD II64 r11 (AddrRegImm target (ImmInt 16)) `snocOL` BCTR [] Nothing) genJump' tree (GCPLinux64ELF 2) = do (target,code) <- getSomeReg tree return (code `snocOL` MR r12 target `snocOL` MTCTR r12 `snocOL` BCTR [] Nothing) genJump' tree _ = do (target,code) <- getSomeReg tree return (code `snocOL` MTCTR target `snocOL` BCTR [] Nothing) -- ----------------------------------------------------------------------------- -- Unconditional branches genBranch :: BlockId -> NatM InstrBlock genBranch = return . toOL . mkJumpInstr -- ----------------------------------------------------------------------------- -- Conditional jumps {- Conditional jumps are always to local labels, so we can use branch instructions. We peek at the arguments to decide what kind of comparison to do. -} genCondJump :: BlockId -- the branch target -> CmmExpr -- the condition on which to branch -> Maybe Bool -> NatM InstrBlock genCondJump id bool prediction = do CondCode _ cond code <- getCondCode bool return (code `snocOL` BCC cond id prediction) -- ----------------------------------------------------------------------------- -- Generating C calls -- Now the biggest nightmare---calls. Most of the nastiness is buried in -- @get_arg@, which moves the arguments to the correct registers/stack -- locations. Apart from that, the code is easy. genCCall :: ForeignTarget -- function to call -> [CmmFormal] -- where to put the result -> [CmmActual] -- arguments (of mixed type) -> NatM InstrBlock genCCall (PrimTarget MO_WriteBarrier) _ _ = return $ unitOL LWSYNC genCCall (PrimTarget MO_Touch) _ _ = return $ nilOL genCCall (PrimTarget (MO_Prefetch_Data _)) _ _ = return $ nilOL genCCall (PrimTarget (MO_AtomicRMW width amop)) [dst] [addr, n] = do dflags <- getDynFlags let platform = targetPlatform dflags fmt = intFormat width reg_dst = getRegisterReg platform (CmmLocal dst) (instr, n_code) <- case amop of AMO_Add -> getSomeRegOrImm ADD True reg_dst AMO_Sub -> case n of CmmLit (CmmInt i _) | Just imm <- makeImmediate width True (-i) -> return (ADD reg_dst reg_dst (RIImm imm), nilOL) _ -> do (n_reg, n_code) <- getSomeReg n return (SUBF reg_dst n_reg reg_dst, n_code) AMO_And -> getSomeRegOrImm AND False reg_dst AMO_Nand -> do (n_reg, n_code) <- getSomeReg n return (NAND reg_dst reg_dst n_reg, n_code) AMO_Or -> getSomeRegOrImm OR False reg_dst AMO_Xor -> getSomeRegOrImm XOR False reg_dst Amode addr_reg addr_code <- getAmodeIndex addr lbl_retry <- getBlockIdNat return $ n_code `appOL` addr_code `appOL` toOL [ HWSYNC , BCC ALWAYS lbl_retry Nothing , NEWBLOCK lbl_retry , LDR fmt reg_dst addr_reg , instr , STC fmt reg_dst addr_reg , BCC NE lbl_retry (Just False) , ISYNC ] where getAmodeIndex (CmmMachOp (MO_Add _) [x, y]) = do (regX, codeX) <- getSomeReg x (regY, codeY) <- getSomeReg y return (Amode (AddrRegReg regX regY) (codeX `appOL` codeY)) getAmodeIndex other = do (reg, code) <- getSomeReg other return (Amode (AddrRegReg r0 reg) code) -- NB: r0 is 0 here! getSomeRegOrImm op sign dst = case n of CmmLit (CmmInt i _) | Just imm <- makeImmediate width sign i -> return (op dst dst (RIImm imm), nilOL) _ -> do (n_reg, n_code) <- getSomeReg n return (op dst dst (RIReg n_reg), n_code) genCCall (PrimTarget (MO_AtomicRead width)) [dst] [addr] = do dflags <- getDynFlags let platform = targetPlatform dflags fmt = intFormat width reg_dst = getRegisterReg platform (CmmLocal dst) form = if widthInBits width == 64 then DS else D Amode addr_reg addr_code <- getAmode form addr lbl_end <- getBlockIdNat return $ addr_code `appOL` toOL [ HWSYNC , LD fmt reg_dst addr_reg , CMP fmt reg_dst (RIReg reg_dst) , BCC NE lbl_end (Just False) , BCC ALWAYS lbl_end Nothing -- See Note [Seemingly useless cmp and bne] , NEWBLOCK lbl_end , ISYNC ] -- Note [Seemingly useless cmp and bne] -- In Power ISA, Book II, Section 4.4.1, Instruction Synchronize Instruction -- the second paragraph says that isync may complete before storage accesses -- "associated" with a preceding instruction have been performed. The cmp -- operation and the following bne introduce a data and control dependency -- on the load instruction (See also Power ISA, Book II, Appendix B.2.3, Safe -- Fetch). -- This is also what gcc does. genCCall (PrimTarget (MO_AtomicWrite width)) [] [addr, val] = do code <- assignMem_IntCode (intFormat width) addr val return $ unitOL(HWSYNC) `appOL` code genCCall (PrimTarget (MO_Clz width)) [dst] [src] = do dflags <- getDynFlags let platform = targetPlatform dflags reg_dst = getRegisterReg platform (CmmLocal dst) if target32Bit platform && width == W64 then do ChildCode64 code vr_lo <- iselExpr64 src lbl1 <- getBlockIdNat lbl2 <- getBlockIdNat lbl3 <- getBlockIdNat let vr_hi = getHiVRegFromLo vr_lo cntlz = toOL [ CMPL II32 vr_hi (RIImm (ImmInt 0)) , BCC NE lbl2 Nothing , BCC ALWAYS lbl1 Nothing , NEWBLOCK lbl1 , CNTLZ II32 reg_dst vr_lo , ADD reg_dst reg_dst (RIImm (ImmInt 32)) , BCC ALWAYS lbl3 Nothing , NEWBLOCK lbl2 , CNTLZ II32 reg_dst vr_hi , BCC ALWAYS lbl3 Nothing , NEWBLOCK lbl3 ] return $ code `appOL` cntlz else do let format = if width == W64 then II64 else II32 (s_reg, s_code) <- getSomeReg src (pre, reg , post) <- case width of W64 -> return (nilOL, s_reg, nilOL) W32 -> return (nilOL, s_reg, nilOL) W16 -> do reg_tmp <- getNewRegNat format return ( unitOL $ AND reg_tmp s_reg (RIImm (ImmInt 65535)) , reg_tmp , unitOL $ ADD reg_dst reg_dst (RIImm (ImmInt (-16))) ) W8 -> do reg_tmp <- getNewRegNat format return ( unitOL $ AND reg_tmp s_reg (RIImm (ImmInt 255)) , reg_tmp , unitOL $ ADD reg_dst reg_dst (RIImm (ImmInt (-24))) ) _ -> panic "genCall: Clz wrong format" let cntlz = unitOL (CNTLZ format reg_dst reg) return $ s_code `appOL` pre `appOL` cntlz `appOL` post genCCall (PrimTarget (MO_Ctz width)) [dst] [src] = do dflags <- getDynFlags let platform = targetPlatform dflags reg_dst = getRegisterReg platform (CmmLocal dst) if target32Bit platform && width == W64 then do let format = II32 ChildCode64 code vr_lo <- iselExpr64 src lbl1 <- getBlockIdNat lbl2 <- getBlockIdNat lbl3 <- getBlockIdNat x' <- getNewRegNat format x'' <- getNewRegNat format r' <- getNewRegNat format cnttzlo <- cnttz format reg_dst vr_lo let vr_hi = getHiVRegFromLo vr_lo cnttz64 = toOL [ CMPL format vr_lo (RIImm (ImmInt 0)) , BCC NE lbl2 Nothing , BCC ALWAYS lbl1 Nothing , NEWBLOCK lbl1 , ADD x' vr_hi (RIImm (ImmInt (-1))) , ANDC x'' x' vr_hi , CNTLZ format r' x'' -- 32 + (32 - clz(x'')) , SUBFC reg_dst r' (RIImm (ImmInt 64)) , BCC ALWAYS lbl3 Nothing , NEWBLOCK lbl2 ] `appOL` cnttzlo `appOL` toOL [ BCC ALWAYS lbl3 Nothing , NEWBLOCK lbl3 ] return $ code `appOL` cnttz64 else do let format = if width == W64 then II64 else II32 (s_reg, s_code) <- getSomeReg src (reg_ctz, pre_code) <- case width of W64 -> return (s_reg, nilOL) W32 -> return (s_reg, nilOL) W16 -> do reg_tmp <- getNewRegNat format return (reg_tmp, unitOL $ ORIS reg_tmp s_reg (ImmInt 1)) W8 -> do reg_tmp <- getNewRegNat format return (reg_tmp, unitOL $ OR reg_tmp s_reg (RIImm (ImmInt 256))) _ -> panic "genCall: Ctz wrong format" ctz_code <- cnttz format reg_dst reg_ctz return $ s_code `appOL` pre_code `appOL` ctz_code where -- cnttz(x) = sizeof(x) - cntlz(~x & (x - 1)) -- see Henry S. Warren, Hacker's Delight, p 107 cnttz format dst src = do let format_bits = 8 * formatInBytes format x' <- getNewRegNat format x'' <- getNewRegNat format r' <- getNewRegNat format return $ toOL [ ADD x' src (RIImm (ImmInt (-1))) , ANDC x'' x' src , CNTLZ format r' x'' , SUBFC dst r' (RIImm (ImmInt (format_bits))) ] genCCall target dest_regs argsAndHints = do dflags <- getDynFlags let platform = targetPlatform dflags case target of PrimTarget (MO_S_QuotRem width) -> divOp1 platform True width dest_regs argsAndHints PrimTarget (MO_U_QuotRem width) -> divOp1 platform False width dest_regs argsAndHints PrimTarget (MO_U_QuotRem2 width) -> divOp2 platform width dest_regs argsAndHints PrimTarget (MO_U_Mul2 width) -> multOp2 platform width dest_regs argsAndHints PrimTarget (MO_Add2 _) -> add2Op platform dest_regs argsAndHints PrimTarget (MO_AddWordC _) -> addcOp platform dest_regs argsAndHints PrimTarget (MO_SubWordC _) -> subcOp platform dest_regs argsAndHints PrimTarget (MO_AddIntC width) -> addSubCOp ADDO platform width dest_regs argsAndHints PrimTarget (MO_SubIntC width) -> addSubCOp SUBFO platform width dest_regs argsAndHints PrimTarget MO_F64_Fabs -> fabs platform dest_regs argsAndHints PrimTarget MO_F32_Fabs -> fabs platform dest_regs argsAndHints _ -> genCCall' dflags (platformToGCP platform) target dest_regs argsAndHints where divOp1 platform signed width [res_q, res_r] [arg_x, arg_y] = do let reg_q = getRegisterReg platform (CmmLocal res_q) reg_r = getRegisterReg platform (CmmLocal res_r) fmt = intFormat width (x_reg, x_code) <- getSomeReg arg_x (y_reg, y_code) <- getSomeReg arg_y return $ y_code `appOL` x_code `appOL` toOL [ DIV fmt signed reg_q x_reg y_reg , MULL fmt reg_r reg_q (RIReg y_reg) , SUBF reg_r reg_r x_reg ] divOp1 _ _ _ _ _ = panic "genCCall: Wrong number of arguments for divOp1" divOp2 platform width [res_q, res_r] [arg_x_high, arg_x_low, arg_y] = do let reg_q = getRegisterReg platform (CmmLocal res_q) reg_r = getRegisterReg platform (CmmLocal res_r) fmt = intFormat width half = 4 * (formatInBytes fmt) (xh_reg, xh_code) <- getSomeReg arg_x_high (xl_reg, xl_code) <- getSomeReg arg_x_low (y_reg, y_code) <- getSomeReg arg_y s <- getNewRegNat fmt b <- getNewRegNat fmt v <- getNewRegNat fmt vn1 <- getNewRegNat fmt vn0 <- getNewRegNat fmt un32 <- getNewRegNat fmt tmp <- getNewRegNat fmt un10 <- getNewRegNat fmt un1 <- getNewRegNat fmt un0 <- getNewRegNat fmt q1 <- getNewRegNat fmt rhat <- getNewRegNat fmt tmp1 <- getNewRegNat fmt q0 <- getNewRegNat fmt un21 <- getNewRegNat fmt again1 <- getBlockIdNat no1 <- getBlockIdNat then1 <- getBlockIdNat endif1 <- getBlockIdNat again2 <- getBlockIdNat no2 <- getBlockIdNat then2 <- getBlockIdNat endif2 <- getBlockIdNat return $ y_code `appOL` xl_code `appOL` xh_code `appOL` -- see Hacker's Delight p 196 Figure 9-3 toOL [ -- b = 2 ^ (bits_in_word / 2) LI b (ImmInt 1) , SL fmt b b (RIImm (ImmInt half)) -- s = clz(y) , CNTLZ fmt s y_reg -- v = y << s , SL fmt v y_reg (RIReg s) -- vn1 = upper half of v , SR fmt vn1 v (RIImm (ImmInt half)) -- vn0 = lower half of v , CLRLI fmt vn0 v half -- un32 = (u1 << s) -- | (u0 >> (bits_in_word - s)) , SL fmt un32 xh_reg (RIReg s) , SUBFC tmp s (RIImm (ImmInt (8 * formatInBytes fmt))) , SR fmt tmp xl_reg (RIReg tmp) , OR un32 un32 (RIReg tmp) -- un10 = u0 << s , SL fmt un10 xl_reg (RIReg s) -- un1 = upper half of un10 , SR fmt un1 un10 (RIImm (ImmInt half)) -- un0 = lower half of un10 , CLRLI fmt un0 un10 half -- q1 = un32/vn1 , DIV fmt False q1 un32 vn1 -- rhat = un32 - q1*vn1 , MULL fmt tmp q1 (RIReg vn1) , SUBF rhat tmp un32 , BCC ALWAYS again1 Nothing , NEWBLOCK again1 -- if (q1 >= b || q1*vn0 > b*rhat + un1) , CMPL fmt q1 (RIReg b) , BCC GEU then1 Nothing , BCC ALWAYS no1 Nothing , NEWBLOCK no1 , MULL fmt tmp q1 (RIReg vn0) , SL fmt tmp1 rhat (RIImm (ImmInt half)) , ADD tmp1 tmp1 (RIReg un1) , CMPL fmt tmp (RIReg tmp1) , BCC LEU endif1 Nothing , BCC ALWAYS then1 Nothing , NEWBLOCK then1 -- q1 = q1 - 1 , ADD q1 q1 (RIImm (ImmInt (-1))) -- rhat = rhat + vn1 , ADD rhat rhat (RIReg vn1) -- if (rhat < b) goto again1 , CMPL fmt rhat (RIReg b) , BCC LTT again1 Nothing , BCC ALWAYS endif1 Nothing , NEWBLOCK endif1 -- un21 = un32*b + un1 - q1*v , SL fmt un21 un32 (RIImm (ImmInt half)) , ADD un21 un21 (RIReg un1) , MULL fmt tmp q1 (RIReg v) , SUBF un21 tmp un21 -- compute second quotient digit -- q0 = un21/vn1 , DIV fmt False q0 un21 vn1 -- rhat = un21- q0*vn1 , MULL fmt tmp q0 (RIReg vn1) , SUBF rhat tmp un21 , BCC ALWAYS again2 Nothing , NEWBLOCK again2 -- if (q0>b || q0*vn0 > b*rhat + un0) , CMPL fmt q0 (RIReg b) , BCC GEU then2 Nothing , BCC ALWAYS no2 Nothing , NEWBLOCK no2 , MULL fmt tmp q0 (RIReg vn0) , SL fmt tmp1 rhat (RIImm (ImmInt half)) , ADD tmp1 tmp1 (RIReg un0) , CMPL fmt tmp (RIReg tmp1) , BCC LEU endif2 Nothing , BCC ALWAYS then2 Nothing , NEWBLOCK then2 -- q0 = q0 - 1 , ADD q0 q0 (RIImm (ImmInt (-1))) -- rhat = rhat + vn1 , ADD rhat rhat (RIReg vn1) -- if (rhat> s , SL fmt reg_r un21 (RIImm (ImmInt half)) , ADD reg_r reg_r (RIReg un0) , MULL fmt tmp q0 (RIReg v) , SUBF reg_r tmp reg_r , SR fmt reg_r reg_r (RIReg s) -- compute quotient -- q = q1*b + q0 , SL fmt reg_q q1 (RIImm (ImmInt half)) , ADD reg_q reg_q (RIReg q0) ] divOp2 _ _ _ _ = panic "genCCall: Wrong number of arguments for divOp2" multOp2 platform width [res_h, res_l] [arg_x, arg_y] = do let reg_h = getRegisterReg platform (CmmLocal res_h) reg_l = getRegisterReg platform (CmmLocal res_l) fmt = intFormat width (x_reg, x_code) <- getSomeReg arg_x (y_reg, y_code) <- getSomeReg arg_y return $ y_code `appOL` x_code `appOL` toOL [ MULL fmt reg_l x_reg (RIReg y_reg) , MULHU fmt reg_h x_reg y_reg ] multOp2 _ _ _ _ = panic "genCall: Wrong number of arguments for multOp2" add2Op platform [res_h, res_l] [arg_x, arg_y] = do let reg_h = getRegisterReg platform (CmmLocal res_h) reg_l = getRegisterReg platform (CmmLocal res_l) (x_reg, x_code) <- getSomeReg arg_x (y_reg, y_code) <- getSomeReg arg_y return $ y_code `appOL` x_code `appOL` toOL [ LI reg_h (ImmInt 0) , ADDC reg_l x_reg y_reg , ADDZE reg_h reg_h ] add2Op _ _ _ = panic "genCCall: Wrong number of arguments/results for add2" addcOp platform [res_r, res_c] [arg_x, arg_y] = add2Op platform [res_c {-hi-}, res_r {-lo-}] [arg_x, arg_y] addcOp _ _ _ = panic "genCCall: Wrong number of arguments/results for addc" -- PowerPC subfc sets the carry for rT = ~(rA) + rB + 1, -- which is 0 for borrow and 1 otherwise. We need 1 and 0 -- so xor with 1. subcOp platform [res_r, res_c] [arg_x, arg_y] = do let reg_r = getRegisterReg platform (CmmLocal res_r) reg_c = getRegisterReg platform (CmmLocal res_c) (x_reg, x_code) <- getSomeReg arg_x (y_reg, y_code) <- getSomeReg arg_y return $ y_code `appOL` x_code `appOL` toOL [ LI reg_c (ImmInt 0) , SUBFC reg_r y_reg (RIReg x_reg) , ADDZE reg_c reg_c , XOR reg_c reg_c (RIImm (ImmInt 1)) ] subcOp _ _ _ = panic "genCCall: Wrong number of arguments/results for subc" addSubCOp instr platform width [res_r, res_c] [arg_x, arg_y] = do let reg_r = getRegisterReg platform (CmmLocal res_r) reg_c = getRegisterReg platform (CmmLocal res_c) (x_reg, x_code) <- getSomeReg arg_x (y_reg, y_code) <- getSomeReg arg_y return $ y_code `appOL` x_code `appOL` toOL [ instr reg_r y_reg x_reg, -- SUBFO argument order reversed! MFOV (intFormat width) reg_c ] addSubCOp _ _ _ _ _ = panic "genCall: Wrong number of arguments/results for addC" fabs platform [res] [arg] = do let res_r = getRegisterReg platform (CmmLocal res) (arg_reg, arg_code) <- getSomeReg arg return $ arg_code `snocOL` FABS res_r arg_reg fabs _ _ _ = panic "genCall: Wrong number of arguments/results for fabs" -- TODO: replace 'Int' by an enum such as 'PPC_64ABI' data GenCCallPlatform = GCPLinux | GCPDarwin | GCPLinux64ELF !Int | GCPAIX platformToGCP :: Platform -> GenCCallPlatform platformToGCP platform = case platformOS platform of OSLinux -> case platformArch platform of ArchPPC -> GCPLinux ArchPPC_64 ELF_V1 -> GCPLinux64ELF 1 ArchPPC_64 ELF_V2 -> GCPLinux64ELF 2 _ -> panic "PPC.CodeGen.platformToGCP: Unknown Linux" OSAIX -> GCPAIX OSDarwin -> GCPDarwin _ -> panic "PPC.CodeGen.platformToGCP: not defined for this OS" genCCall' :: DynFlags -> GenCCallPlatform -> ForeignTarget -- function to call -> [CmmFormal] -- where to put the result -> [CmmActual] -- arguments (of mixed type) -> NatM InstrBlock {- The PowerPC calling convention for Darwin/Mac OS X is described in Apple's document "Inside Mac OS X - Mach-O Runtime Architecture". PowerPC Linux uses the System V Release 4 Calling Convention for PowerPC. It is described in the "System V Application Binary Interface PowerPC Processor Supplement". Both conventions are similar: Parameters may be passed in general-purpose registers starting at r3, in floating point registers starting at f1, or on the stack. But there are substantial differences: * The number of registers used for parameter passing and the exact set of nonvolatile registers differs (see MachRegs.hs). * On Darwin, stack space is always reserved for parameters, even if they are passed in registers. The called routine may choose to save parameters from registers to the corresponding space on the stack. * On Darwin, a corresponding amount of GPRs is skipped when a floating point parameter is passed in an FPR. * SysV insists on either passing I64 arguments on the stack, or in two GPRs, starting with an odd-numbered GPR. It may skip a GPR to achieve this. Darwin just treats an I64 like two separate II32s (high word first). * I64 and FF64 arguments are 8-byte aligned on the stack for SysV, but only 4-byte aligned like everything else on Darwin. * The SysV spec claims that FF32 is represented as FF64 on the stack. GCC on PowerPC Linux does not agree, so neither do we. PowerPC 64 Linux uses the System V Release 4 Calling Convention for 64-bit PowerPC. It is specified in "64-bit PowerPC ELF Application Binary Interface Supplement 1.9" (PPC64 ELF v1.9). PowerPC 64 Linux in little endian mode uses the "Power Architecture 64-Bit ELF V2 ABI Specification -- OpenPOWER ABI for Linux Supplement" (PPC64 ELF v2). AIX follows the "PowerOpen ABI: Application Binary Interface Big-Endian 32-Bit Hardware Implementation" According to all conventions, the parameter area should be part of the caller's stack frame, allocated in the caller's prologue code (large enough to hold the parameter lists for all called routines). The NCG already uses the stack for register spilling, leaving 64 bytes free at the top. If we need a larger parameter area than that, we just allocate a new stack frame just before ccalling. -} genCCall' dflags gcp target dest_regs args = ASSERT(not $ any (`elem` [II16]) $ map cmmTypeFormat argReps) -- we rely on argument promotion in the codeGen do (finalStack,passArgumentsCode,usedRegs) <- passArguments (zip args argReps) allArgRegs (allFPArgRegs platform) initialStackOffset (toOL []) [] (labelOrExpr, reduceToFF32) <- case target of ForeignTarget (CmmLit (CmmLabel lbl)) _ -> do uses_pic_base_implicitly return (Left lbl, False) ForeignTarget expr _ -> do uses_pic_base_implicitly return (Right expr, False) PrimTarget mop -> outOfLineMachOp mop let codeBefore = move_sp_down finalStack `appOL` passArgumentsCode codeAfter = move_sp_up finalStack `appOL` moveResult reduceToFF32 case labelOrExpr of Left lbl -> do -- the linker does all the work for us return ( codeBefore `snocOL` BL lbl usedRegs `appOL` maybeNOP -- some ABI require a NOP after BL `appOL` codeAfter) Right dyn -> do -- implement call through function pointer (dynReg, dynCode) <- getSomeReg dyn case gcp of GCPLinux64ELF 1 -> return ( dynCode `appOL` codeBefore `snocOL` ST spFormat toc (AddrRegImm sp (ImmInt 40)) `snocOL` LD II64 r11 (AddrRegImm dynReg (ImmInt 0)) `snocOL` LD II64 toc (AddrRegImm dynReg (ImmInt 8)) `snocOL` MTCTR r11 `snocOL` LD II64 r11 (AddrRegImm dynReg (ImmInt 16)) `snocOL` BCTRL usedRegs `snocOL` LD spFormat toc (AddrRegImm sp (ImmInt 40)) `appOL` codeAfter) GCPLinux64ELF 2 -> return ( dynCode `appOL` codeBefore `snocOL` ST spFormat toc (AddrRegImm sp (ImmInt 24)) `snocOL` MR r12 dynReg `snocOL` MTCTR r12 `snocOL` BCTRL usedRegs `snocOL` LD spFormat toc (AddrRegImm sp (ImmInt 24)) `appOL` codeAfter) GCPAIX -> return ( dynCode -- AIX/XCOFF follows the PowerOPEN ABI -- which is quite similiar to LinuxPPC64/ELFv1 `appOL` codeBefore `snocOL` ST spFormat toc (AddrRegImm sp (ImmInt 20)) `snocOL` LD II32 r11 (AddrRegImm dynReg (ImmInt 0)) `snocOL` LD II32 toc (AddrRegImm dynReg (ImmInt 4)) `snocOL` MTCTR r11 `snocOL` LD II32 r11 (AddrRegImm dynReg (ImmInt 8)) `snocOL` BCTRL usedRegs `snocOL` LD spFormat toc (AddrRegImm sp (ImmInt 20)) `appOL` codeAfter) _ -> return ( dynCode `snocOL` MTCTR dynReg `appOL` codeBefore `snocOL` BCTRL usedRegs `appOL` codeAfter) where platform = targetPlatform dflags uses_pic_base_implicitly = do -- See Note [implicit register in PPC PIC code] -- on why we claim to use PIC register here when (positionIndependent dflags && target32Bit platform) $ do _ <- getPicBaseNat $ archWordFormat True return () initialStackOffset = case gcp of GCPAIX -> 24 GCPDarwin -> 24 GCPLinux -> 8 GCPLinux64ELF 1 -> 48 GCPLinux64ELF 2 -> 32 _ -> panic "genCall': unknown calling convention" -- size of linkage area + size of arguments, in bytes stackDelta finalStack = case gcp of GCPAIX -> roundTo 16 $ (24 +) $ max 32 $ sum $ map (widthInBytes . typeWidth) argReps GCPDarwin -> roundTo 16 $ (24 +) $ max 32 $ sum $ map (widthInBytes . typeWidth) argReps GCPLinux -> roundTo 16 finalStack GCPLinux64ELF 1 -> roundTo 16 $ (48 +) $ max 64 $ sum $ map (roundTo 8 . widthInBytes . typeWidth) argReps GCPLinux64ELF 2 -> roundTo 16 $ (32 +) $ max 64 $ sum $ map (roundTo 8 . widthInBytes . typeWidth) argReps _ -> panic "genCall': unknown calling conv." argReps = map (cmmExprType dflags) args roundTo a x | x `mod` a == 0 = x | otherwise = x + a - (x `mod` a) spFormat = if target32Bit platform then II32 else II64 -- TODO: Do not create a new stack frame if delta is too large. move_sp_down finalStack | delta > stackFrameHeaderSize dflags = toOL [STU spFormat sp (AddrRegImm sp (ImmInt (-delta))), DELTA (-delta)] | otherwise = nilOL where delta = stackDelta finalStack move_sp_up finalStack | delta > stackFrameHeaderSize dflags = toOL [ADD sp sp (RIImm (ImmInt delta)), DELTA 0] | otherwise = nilOL where delta = stackDelta finalStack -- A NOP instruction is required after a call (bl instruction) -- on AIX and 64-Bit Linux. -- If the call is to a function with a different TOC (r2) the -- link editor replaces the NOP instruction with a load of the TOC -- from the stack to restore the TOC. maybeNOP = case gcp of -- See Section 3.9.4 of OpenPower ABI GCPAIX -> unitOL NOP -- See Section 3.5.11 of PPC64 ELF v1.9 GCPLinux64ELF 1 -> unitOL NOP -- See Section 2.3.6 of PPC64 ELF v2 GCPLinux64ELF 2 -> unitOL NOP _ -> nilOL passArguments [] _ _ stackOffset accumCode accumUsed = return (stackOffset, accumCode, accumUsed) passArguments ((arg,arg_ty):args) gprs fprs stackOffset accumCode accumUsed | isWord64 arg_ty && target32Bit (targetPlatform dflags) = do ChildCode64 code vr_lo <- iselExpr64 arg let vr_hi = getHiVRegFromLo vr_lo case gcp of GCPAIX -> -- same as for Darwin do let storeWord vr (gpr:_) _ = MR gpr vr storeWord vr [] offset = ST II32 vr (AddrRegImm sp (ImmInt offset)) passArguments args (drop 2 gprs) fprs (stackOffset+8) (accumCode `appOL` code `snocOL` storeWord vr_hi gprs stackOffset `snocOL` storeWord vr_lo (drop 1 gprs) (stackOffset+4)) ((take 2 gprs) ++ accumUsed) GCPDarwin -> do let storeWord vr (gpr:_) _ = MR gpr vr storeWord vr [] offset = ST II32 vr (AddrRegImm sp (ImmInt offset)) passArguments args (drop 2 gprs) fprs (stackOffset+8) (accumCode `appOL` code `snocOL` storeWord vr_hi gprs stackOffset `snocOL` storeWord vr_lo (drop 1 gprs) (stackOffset+4)) ((take 2 gprs) ++ accumUsed) GCPLinux -> do let stackOffset' = roundTo 8 stackOffset stackCode = accumCode `appOL` code `snocOL` ST II32 vr_hi (AddrRegImm sp (ImmInt stackOffset')) `snocOL` ST II32 vr_lo (AddrRegImm sp (ImmInt (stackOffset'+4))) regCode hireg loreg = accumCode `appOL` code `snocOL` MR hireg vr_hi `snocOL` MR loreg vr_lo case gprs of hireg : loreg : regs | even (length gprs) -> passArguments args regs fprs stackOffset (regCode hireg loreg) (hireg : loreg : accumUsed) _skipped : hireg : loreg : regs -> passArguments args regs fprs stackOffset (regCode hireg loreg) (hireg : loreg : accumUsed) _ -> -- only one or no regs left passArguments args [] fprs (stackOffset'+8) stackCode accumUsed GCPLinux64ELF _ -> panic "passArguments: 32 bit code" passArguments ((arg,rep):args) gprs fprs stackOffset accumCode accumUsed | reg : _ <- regs = do register <- getRegister arg let code = case register of Fixed _ freg fcode -> fcode `snocOL` MR reg freg Any _ acode -> acode reg stackOffsetRes = case gcp of -- The Darwin ABI requires that we reserve -- stack slots for register parameters GCPDarwin -> stackOffset + stackBytes -- ... so does the PowerOpen ABI. GCPAIX -> stackOffset + stackBytes -- ... the SysV ABI 32-bit doesn't. GCPLinux -> stackOffset -- ... but SysV ABI 64-bit does. GCPLinux64ELF _ -> stackOffset + stackBytes passArguments args (drop nGprs gprs) (drop nFprs fprs) stackOffsetRes (accumCode `appOL` code) (reg : accumUsed) | otherwise = do (vr, code) <- getSomeReg arg passArguments args (drop nGprs gprs) (drop nFprs fprs) (stackOffset' + stackBytes) (accumCode `appOL` code `snocOL` ST (cmmTypeFormat rep) vr stackSlot) accumUsed where stackOffset' = case gcp of GCPDarwin -> -- stackOffset is at least 4-byte aligned -- The Darwin ABI is happy with that. stackOffset GCPAIX -> -- The 32bit PowerOPEN ABI is happy with -- 32bit-alignment as well... stackOffset GCPLinux -- ... the SysV ABI requires 8-byte -- alignment for doubles. | isFloatType rep && typeWidth rep == W64 -> roundTo 8 stackOffset | otherwise -> stackOffset GCPLinux64ELF _ -> -- Everything on the stack is mapped to -- 8-byte aligned doublewords stackOffset stackOffset'' | isFloatType rep && typeWidth rep == W32 = case gcp of -- The ELF v1 ABI Section 3.2.3 requires: -- "Single precision floating point values -- are mapped to the second word in a single -- doubleword" GCPLinux64ELF 1 -> stackOffset' + 4 _ -> stackOffset' | otherwise = stackOffset' stackSlot = AddrRegImm sp (ImmInt stackOffset'') (nGprs, nFprs, stackBytes, regs) = case gcp of GCPAIX -> case cmmTypeFormat rep of II8 -> (1, 0, 4, gprs) II16 -> (1, 0, 4, gprs) II32 -> (1, 0, 4, gprs) -- The PowerOpen ABI requires that we skip a -- corresponding number of GPRs when we use -- the FPRs. -- -- E.g. for a `double` two GPRs are skipped, -- whereas for a `float` one GPR is skipped -- when parameters are assigned to -- registers. -- -- The PowerOpen ABI specification can be found at -- ftp://www.sourceware.org/pub/binutils/ppc-docs/ppc-poweropen/ FF32 -> (1, 1, 4, fprs) FF64 -> (2, 1, 8, fprs) II64 -> panic "genCCall' passArguments II64" FF80 -> panic "genCCall' passArguments FF80" GCPDarwin -> case cmmTypeFormat rep of II8 -> (1, 0, 4, gprs) II16 -> (1, 0, 4, gprs) II32 -> (1, 0, 4, gprs) -- The Darwin ABI requires that we skip a -- corresponding number of GPRs when we use -- the FPRs. FF32 -> (1, 1, 4, fprs) FF64 -> (2, 1, 8, fprs) II64 -> panic "genCCall' passArguments II64" FF80 -> panic "genCCall' passArguments FF80" GCPLinux -> case cmmTypeFormat rep of II8 -> (1, 0, 4, gprs) II16 -> (1, 0, 4, gprs) II32 -> (1, 0, 4, gprs) -- ... the SysV ABI doesn't. FF32 -> (0, 1, 4, fprs) FF64 -> (0, 1, 8, fprs) II64 -> panic "genCCall' passArguments II64" FF80 -> panic "genCCall' passArguments FF80" GCPLinux64ELF _ -> case cmmTypeFormat rep of II8 -> (1, 0, 8, gprs) II16 -> (1, 0, 8, gprs) II32 -> (1, 0, 8, gprs) II64 -> (1, 0, 8, gprs) -- The ELFv1 ABI requires that we skip a -- corresponding number of GPRs when we use -- the FPRs. FF32 -> (1, 1, 8, fprs) FF64 -> (1, 1, 8, fprs) FF80 -> panic "genCCall' passArguments FF80" moveResult reduceToFF32 = case dest_regs of [] -> nilOL [dest] | reduceToFF32 && isFloat32 rep -> unitOL (FRSP r_dest f1) | isFloat32 rep || isFloat64 rep -> unitOL (MR r_dest f1) | isWord64 rep && target32Bit (targetPlatform dflags) -> toOL [MR (getHiVRegFromLo r_dest) r3, MR r_dest r4] | otherwise -> unitOL (MR r_dest r3) where rep = cmmRegType dflags (CmmLocal dest) r_dest = getRegisterReg platform (CmmLocal dest) _ -> panic "genCCall' moveResult: Bad dest_regs" outOfLineMachOp mop = do dflags <- getDynFlags mopExpr <- cmmMakeDynamicReference dflags CallReference $ mkForeignLabel functionName Nothing ForeignLabelInThisPackage IsFunction let mopLabelOrExpr = case mopExpr of CmmLit (CmmLabel lbl) -> Left lbl _ -> Right mopExpr return (mopLabelOrExpr, reduce) where (functionName, reduce) = case mop of MO_F32_Exp -> (fsLit "exp", True) MO_F32_Log -> (fsLit "log", True) MO_F32_Sqrt -> (fsLit "sqrt", True) MO_F32_Fabs -> unsupported MO_F32_Sin -> (fsLit "sin", True) MO_F32_Cos -> (fsLit "cos", True) MO_F32_Tan -> (fsLit "tan", True) MO_F32_Asin -> (fsLit "asin", True) MO_F32_Acos -> (fsLit "acos", True) MO_F32_Atan -> (fsLit "atan", True) MO_F32_Sinh -> (fsLit "sinh", True) MO_F32_Cosh -> (fsLit "cosh", True) MO_F32_Tanh -> (fsLit "tanh", True) MO_F32_Pwr -> (fsLit "pow", True) MO_F64_Exp -> (fsLit "exp", False) MO_F64_Log -> (fsLit "log", False) MO_F64_Sqrt -> (fsLit "sqrt", False) MO_F64_Fabs -> unsupported MO_F64_Sin -> (fsLit "sin", False) MO_F64_Cos -> (fsLit "cos", False) MO_F64_Tan -> (fsLit "tan", False) MO_F64_Asin -> (fsLit "asin", False) MO_F64_Acos -> (fsLit "acos", False) MO_F64_Atan -> (fsLit "atan", False) MO_F64_Sinh -> (fsLit "sinh", False) MO_F64_Cosh -> (fsLit "cosh", False) MO_F64_Tanh -> (fsLit "tanh", False) MO_F64_Pwr -> (fsLit "pow", False) MO_UF_Conv w -> (fsLit $ word2FloatLabel w, False) MO_Memcpy _ -> (fsLit "memcpy", False) MO_Memset _ -> (fsLit "memset", False) MO_Memmove _ -> (fsLit "memmove", False) MO_Memcmp _ -> (fsLit "memcmp", False) MO_BSwap w -> (fsLit $ bSwapLabel w, False) MO_PopCnt w -> (fsLit $ popCntLabel w, False) MO_Pdep w -> (fsLit $ pdepLabel w, False) MO_Pext w -> (fsLit $ pextLabel w, False) MO_Clz _ -> unsupported MO_Ctz _ -> unsupported MO_AtomicRMW {} -> unsupported MO_Cmpxchg w -> (fsLit $ cmpxchgLabel w, False) MO_AtomicRead _ -> unsupported MO_AtomicWrite _ -> unsupported MO_S_QuotRem {} -> unsupported MO_U_QuotRem {} -> unsupported MO_U_QuotRem2 {} -> unsupported MO_Add2 {} -> unsupported MO_AddWordC {} -> unsupported MO_SubWordC {} -> unsupported MO_AddIntC {} -> unsupported MO_SubIntC {} -> unsupported MO_U_Mul2 {} -> unsupported MO_WriteBarrier -> unsupported MO_Touch -> unsupported MO_Prefetch_Data _ -> unsupported unsupported = panic ("outOfLineCmmOp: " ++ show mop ++ " not supported") -- ----------------------------------------------------------------------------- -- Generating a table-branch genSwitch :: DynFlags -> CmmExpr -> SwitchTargets -> NatM InstrBlock genSwitch dflags expr targets | OSAIX <- platformOS (targetPlatform dflags) = do (reg,e_code) <- getSomeReg (cmmOffset dflags expr offset) let fmt = archWordFormat $ target32Bit $ targetPlatform dflags sha = if target32Bit $ targetPlatform dflags then 2 else 3 tmp <- getNewRegNat fmt lbl <- getNewLabelNat dynRef <- cmmMakeDynamicReference dflags DataReference lbl (tableReg,t_code) <- getSomeReg $ dynRef let code = e_code `appOL` t_code `appOL` toOL [ SL fmt tmp reg (RIImm (ImmInt sha)), LD fmt tmp (AddrRegReg tableReg tmp), MTCTR tmp, BCTR ids (Just lbl) ] return code | (positionIndependent dflags) || (not $ target32Bit $ targetPlatform dflags) = do (reg,e_code) <- getSomeReg (cmmOffset dflags expr offset) let fmt = archWordFormat $ target32Bit $ targetPlatform dflags sha = if target32Bit $ targetPlatform dflags then 2 else 3 tmp <- getNewRegNat fmt lbl <- getNewLabelNat dynRef <- cmmMakeDynamicReference dflags DataReference lbl (tableReg,t_code) <- getSomeReg $ dynRef let code = e_code `appOL` t_code `appOL` toOL [ SL fmt tmp reg (RIImm (ImmInt sha)), LD fmt tmp (AddrRegReg tableReg tmp), ADD tmp tmp (RIReg tableReg), MTCTR tmp, BCTR ids (Just lbl) ] return code | otherwise = do (reg,e_code) <- getSomeReg (cmmOffset dflags expr offset) let fmt = archWordFormat $ target32Bit $ targetPlatform dflags sha = if target32Bit $ targetPlatform dflags then 2 else 3 tmp <- getNewRegNat fmt lbl <- getNewLabelNat let code = e_code `appOL` toOL [ SL fmt tmp reg (RIImm (ImmInt sha)), ADDIS tmp tmp (HA (ImmCLbl lbl)), LD fmt tmp (AddrRegImm tmp (LO (ImmCLbl lbl))), MTCTR tmp, BCTR ids (Just lbl) ] return code where (offset, ids) = switchTargetsToTable targets generateJumpTableForInstr :: DynFlags -> Instr -> Maybe (NatCmmDecl CmmStatics Instr) generateJumpTableForInstr dflags (BCTR ids (Just lbl)) = let jumpTable | (positionIndependent dflags) || (not $ target32Bit $ targetPlatform dflags) = map jumpTableEntryRel ids | otherwise = map (jumpTableEntry dflags) ids where jumpTableEntryRel Nothing = CmmStaticLit (CmmInt 0 (wordWidth dflags)) jumpTableEntryRel (Just blockid) = CmmStaticLit (CmmLabelDiffOff blockLabel lbl 0 (wordWidth dflags)) where blockLabel = blockLbl blockid in Just (CmmData (Section ReadOnlyData lbl) (Statics lbl jumpTable)) generateJumpTableForInstr _ _ = Nothing -- ----------------------------------------------------------------------------- -- 'condIntReg' and 'condFltReg': condition codes into registers -- Turn those condition codes into integers now (when they appear on -- the right hand side of an assignment). condReg :: NatM CondCode -> NatM Register condReg getCond = do CondCode _ cond cond_code <- getCond dflags <- getDynFlags let code dst = cond_code `appOL` negate_code `appOL` toOL [ MFCR dst, RLWINM dst dst (bit + 1) 31 31 ] negate_code | do_negate = unitOL (CRNOR bit bit bit) | otherwise = nilOL (bit, do_negate) = case cond of LTT -> (0, False) LE -> (1, True) EQQ -> (2, False) GE -> (0, True) GTT -> (1, False) NE -> (2, True) LU -> (0, False) LEU -> (1, True) GEU -> (0, True) GU -> (1, False) _ -> panic "PPC.CodeGen.codeReg: no match" format = archWordFormat $ target32Bit $ targetPlatform dflags return (Any format code) condIntReg :: Cond -> Width -> CmmExpr -> CmmExpr -> NatM Register condIntReg cond width x y = condReg (condIntCode cond width x y) condFltReg :: Cond -> CmmExpr -> CmmExpr -> NatM Register condFltReg cond x y = condReg (condFltCode cond x y) -- ----------------------------------------------------------------------------- -- 'trivial*Code': deal with trivial instructions -- Trivial (dyadic: 'trivialCode', floating-point: 'trivialFCode', -- unary: 'trivialUCode', unary fl-pt:'trivialUFCode') instructions. -- Only look for constants on the right hand side, because that's -- where the generic optimizer will have put them. -- Similarly, for unary instructions, we don't have to worry about -- matching an StInt as the argument, because genericOpt will already -- have handled the constant-folding. {- Wolfgang's PowerPC version of The Rules: A slightly modified version of The Rules to take advantage of the fact that PowerPC instructions work on all registers and don't implicitly clobber any fixed registers. * The only expression for which getRegister returns Fixed is (CmmReg reg). * If getRegister returns Any, then the code it generates may modify only: (a) fresh temporaries (b) the destination register It may *not* modify global registers, unless the global register happens to be the destination register. It may not clobber any other registers. In fact, only ccalls clobber any fixed registers. Also, it may not modify the counter register (used by genCCall). Corollary: If a getRegister for a subexpression returns Fixed, you need not move it to a fresh temporary before evaluating the next subexpression. The Fixed register won't be modified. Therefore, we don't need a counterpart for the x86's getStableReg on PPC. * SDM's First Rule is valid for PowerPC, too: subexpressions can depend on the value of the destination register. -} trivialCode :: Width -> Bool -> (Reg -> Reg -> RI -> Instr) -> CmmExpr -> CmmExpr -> NatM Register trivialCode rep signed instr x (CmmLit (CmmInt y _)) | Just imm <- makeImmediate rep signed y = do (src1, code1) <- getSomeReg x let code dst = code1 `snocOL` instr dst src1 (RIImm imm) return (Any (intFormat rep) code) trivialCode rep _ instr x y = do (src1, code1) <- getSomeReg x (src2, code2) <- getSomeReg y let code dst = code1 `appOL` code2 `snocOL` instr dst src1 (RIReg src2) return (Any (intFormat rep) code) shiftMulCode :: Width -> Bool -> (Format-> Reg -> Reg -> RI -> Instr) -> CmmExpr -> CmmExpr -> NatM Register shiftMulCode width sign instr x (CmmLit (CmmInt y _)) | Just imm <- makeImmediate width sign y = do (src1, code1) <- getSomeReg x let format = intFormat width let ins_fmt = intFormat (max W32 width) let code dst = code1 `snocOL` instr ins_fmt dst src1 (RIImm imm) return (Any format code) shiftMulCode width _ instr x y = do (src1, code1) <- getSomeReg x (src2, code2) <- getSomeReg y let format = intFormat width let ins_fmt = intFormat (max W32 width) let code dst = code1 `appOL` code2 `snocOL` instr ins_fmt dst src1 (RIReg src2) return (Any format code) trivialCodeNoImm' :: Format -> (Reg -> Reg -> Reg -> Instr) -> CmmExpr -> CmmExpr -> NatM Register trivialCodeNoImm' format instr x y = do (src1, code1) <- getSomeReg x (src2, code2) <- getSomeReg y let code dst = code1 `appOL` code2 `snocOL` instr dst src1 src2 return (Any format code) trivialCodeNoImm :: Format -> (Format -> Reg -> Reg -> Reg -> Instr) -> CmmExpr -> CmmExpr -> NatM Register trivialCodeNoImm format instr x y = trivialCodeNoImm' format (instr format) x y srCode :: Width -> Bool -> (Format-> Reg -> Reg -> RI -> Instr) -> CmmExpr -> CmmExpr -> NatM Register srCode width sgn instr x (CmmLit (CmmInt y _)) | Just imm <- makeImmediate width sgn y = do let op_len = max W32 width extend = if sgn then extendSExpr else extendUExpr (src1, code1) <- getSomeReg (extend width op_len x) let code dst = code1 `snocOL` instr (intFormat op_len) dst src1 (RIImm imm) return (Any (intFormat width) code) srCode width sgn instr x y = do let op_len = max W32 width extend = if sgn then extendSExpr else extendUExpr (src1, code1) <- getSomeReg (extend width op_len x) (src2, code2) <- getSomeReg (extendUExpr width op_len y) -- Note: Shift amount `y` is unsigned let code dst = code1 `appOL` code2 `snocOL` instr (intFormat op_len) dst src1 (RIReg src2) return (Any (intFormat width) code) divCode :: Width -> Bool -> CmmExpr -> CmmExpr -> NatM Register divCode width sgn x y = do let op_len = max W32 width extend = if sgn then extendSExpr else extendUExpr (src1, code1) <- getSomeReg (extend width op_len x) (src2, code2) <- getSomeReg (extend width op_len y) let code dst = code1 `appOL` code2 `snocOL` DIV (intFormat op_len) sgn dst src1 src2 return (Any (intFormat width) code) trivialUCode :: Format -> (Reg -> Reg -> Instr) -> CmmExpr -> NatM Register trivialUCode rep instr x = do (src, code) <- getSomeReg x let code' dst = code `snocOL` instr dst src return (Any rep code') -- There is no "remainder" instruction on the PPC, so we have to do -- it the hard way. -- The "sgn" parameter is the signedness for the division instruction remainderCode :: Width -> Bool -> CmmExpr -> CmmExpr -> NatM Register remainderCode rep sgn x y = do let op_len = max W32 rep ins_fmt = intFormat op_len extend = if sgn then extendSExpr else extendUExpr (src1, code1) <- getSomeReg (extend rep op_len x) (src2, code2) <- getSomeReg (extend rep op_len y) let code dst = code1 `appOL` code2 `appOL` toOL [ DIV ins_fmt sgn dst src1 src2, MULL ins_fmt dst dst (RIReg src2), SUBF dst dst src1 ] return (Any (intFormat rep) code) coerceInt2FP :: Width -> Width -> CmmExpr -> NatM Register coerceInt2FP fromRep toRep x = do dflags <- getDynFlags let arch = platformArch $ targetPlatform dflags coerceInt2FP' arch fromRep toRep x coerceInt2FP' :: Arch -> Width -> Width -> CmmExpr -> NatM Register coerceInt2FP' ArchPPC fromRep toRep x = do (src, code) <- getSomeReg x lbl <- getNewLabelNat itmp <- getNewRegNat II32 ftmp <- getNewRegNat FF64 dflags <- getDynFlags dynRef <- cmmMakeDynamicReference dflags DataReference lbl Amode addr addr_code <- getAmode D dynRef let code' dst = code `appOL` maybe_exts `appOL` toOL [ LDATA (Section ReadOnlyData lbl) $ Statics lbl [CmmStaticLit (CmmInt 0x43300000 W32), CmmStaticLit (CmmInt 0x80000000 W32)], XORIS itmp src (ImmInt 0x8000), ST II32 itmp (spRel dflags 3), LIS itmp (ImmInt 0x4330), ST II32 itmp (spRel dflags 2), LD FF64 ftmp (spRel dflags 2) ] `appOL` addr_code `appOL` toOL [ LD FF64 dst addr, FSUB FF64 dst ftmp dst ] `appOL` maybe_frsp dst maybe_exts = case fromRep of W8 -> unitOL $ EXTS II8 src src W16 -> unitOL $ EXTS II16 src src W32 -> nilOL _ -> panic "PPC.CodeGen.coerceInt2FP: no match" maybe_frsp dst = case toRep of W32 -> unitOL $ FRSP dst dst W64 -> nilOL _ -> panic "PPC.CodeGen.coerceInt2FP: no match" return (Any (floatFormat toRep) code') -- On an ELF v1 Linux we use the compiler doubleword in the stack frame -- this is the TOC pointer doubleword on ELF v2 Linux. The latter is only -- set right before a call and restored right after return from the call. -- So it is fine. coerceInt2FP' (ArchPPC_64 _) fromRep toRep x = do (src, code) <- getSomeReg x dflags <- getDynFlags let code' dst = code `appOL` maybe_exts `appOL` toOL [ ST II64 src (spRel dflags 3), LD FF64 dst (spRel dflags 3), FCFID dst dst ] `appOL` maybe_frsp dst maybe_exts = case fromRep of W8 -> unitOL $ EXTS II8 src src W16 -> unitOL $ EXTS II16 src src W32 -> unitOL $ EXTS II32 src src W64 -> nilOL _ -> panic "PPC.CodeGen.coerceInt2FP: no match" maybe_frsp dst = case toRep of W32 -> unitOL $ FRSP dst dst W64 -> nilOL _ -> panic "PPC.CodeGen.coerceInt2FP: no match" return (Any (floatFormat toRep) code') coerceInt2FP' _ _ _ _ = panic "PPC.CodeGen.coerceInt2FP: unknown arch" coerceFP2Int :: Width -> Width -> CmmExpr -> NatM Register coerceFP2Int fromRep toRep x = do dflags <- getDynFlags let arch = platformArch $ targetPlatform dflags coerceFP2Int' arch fromRep toRep x coerceFP2Int' :: Arch -> Width -> Width -> CmmExpr -> NatM Register coerceFP2Int' ArchPPC _ toRep x = do dflags <- getDynFlags -- the reps don't really matter: F*->FF64 and II32->I* are no-ops (src, code) <- getSomeReg x tmp <- getNewRegNat FF64 let code' dst = code `appOL` toOL [ -- convert to int in FP reg FCTIWZ tmp src, -- store value (64bit) from FP to stack ST FF64 tmp (spRel dflags 2), -- read low word of value (high word is undefined) LD II32 dst (spRel dflags 3)] return (Any (intFormat toRep) code') coerceFP2Int' (ArchPPC_64 _) _ toRep x = do dflags <- getDynFlags -- the reps don't really matter: F*->FF64 and II64->I* are no-ops (src, code) <- getSomeReg x tmp <- getNewRegNat FF64 let code' dst = code `appOL` toOL [ -- convert to int in FP reg FCTIDZ tmp src, -- store value (64bit) from FP to compiler word on stack ST FF64 tmp (spRel dflags 3), LD II64 dst (spRel dflags 3)] return (Any (intFormat toRep) code') coerceFP2Int' _ _ _ _ = panic "PPC.CodeGen.coerceFP2Int: unknown arch" -- Note [.LCTOC1 in PPC PIC code] -- The .LCTOC1 label is defined to point 32768 bytes into the GOT table -- to make the most of the PPC's 16-bit displacements. -- As 16-bit signed offset is used (usually via addi/lwz instructions) -- first element will have '-32768' offset against .LCTOC1. -- Note [implicit register in PPC PIC code] -- PPC generates calls by labels in assembly -- in form of: -- bl puts+32768@plt -- in this form it's not seen directly (by GHC NCG) -- that r30 (PicBaseReg) is used, -- but r30 is a required part of PLT code setup: -- puts+32768@plt: -- lwz r11,-30484(r30) ; offset in .LCTOC1 -- mtctr r11 -- bctr