-%
-% (c) The AQUA Project, Glasgow University, 1993-1995
-%
+-- -----------------------------------------------------------------------------
+--
+-- (c) The University of Glasgow 1993-2004
+--
+-- This is the top-level module in the native code generator.
+--
+-- -----------------------------------------------------------------------------
\begin{code}
-#include "HsVersions.h"
-#include "../../includes/platform.h"
-#include "../../includes/GhcConstants.h"
-
-module AsmCodeGen (
-#ifdef __GLASGOW_HASKELL__
- writeRealAsm,
-#endif
- dumpRealAsm,
-
- -- And, I guess we need these...
- AbstractC, GlobalSwitch, SwitchResult,
- SplitUniqSupply, SUniqSM(..)
- ) where
-
-import AbsCSyn ( AbstractC )
-import AbsCStixGen ( genCodeAbstractC )
-import AbsPrel ( PrimKind, PrimOp(..)
- IF_ATTACK_PRAGMAS(COMMA tagOf_PrimOp)
- IF_ATTACK_PRAGMAS(COMMA pprPrimOp)
- )
-import CmdLineOpts ( GlobalSwitch(..), stringSwitchSet, switchIsOn, SwitchResult(..) )
-import MachDesc
-import Maybes ( Maybe(..) )
-import Outputable
-#if alpha_dec_osf1_TARGET
-import AlphaDesc ( mkAlpha )
-#else
-#if sparc_TARGET_ARCH
-import SparcDesc ( mkSparc )
-#endif
-#endif
-import Stix
-import SplitUniq
-import Unique
-import Unpretty
-import Util
-#if defined(__HBC__)
-import
- Word
-#endif
-\end{code}
-
-This is a generic assembly language generator for the Glasgow Haskell
-Compiler. It has been a long time in germinating, basically due to
-time constraints and the large spectrum of design possibilities.
-Presently it generates code for:
-\begin{itemize}
-\item Sparc
-\end{itemize}
-In the pipeline (sic) are plans and/or code for 680x0, 386/486.
-
-The code generator presumes the presence of a working C port. This is
-because any code that cannot be compiled (e.g. @casm@s) is re-directed
-via this route. It also help incremental development. Because this
-code generator is specially written for the Abstract C produced by the
-Glasgow Haskell Compiler, several optimisation opportunities are open
-to us that are not open to @gcc@. In particular, we know that the A
-and B stacks and the Heap are all mutually exclusive wrt. aliasing,
-and that expressions have no side effects (all state transformations
-are top level objects).
-
-There are two main components to the code generator.
-\begin{itemize}
-\item Abstract C is considered in statements,
- with a Twig-like system handling each statement in turn.
-\item A scheduler turns the tree of assembly language orderings
- into a sequence suitable for input to an assembler.
-\end{itemize}
-The @codeGenerate@ function returns the final assembly language output
-(as a String). We can return a string, because there is only one way
-of printing the output suitable for assembler consumption. It also
-allows limited abstraction of different machines from the Main module.
-
-The first part is the actual assembly language generation. First we
-split up the Abstract C into individual functions, then consider
-chunks in isolation, giving back an @OrdList@ of assembly language
-instructions. The generic algorithm is heavily inspired by Twig
-(ref), but also draws concepts from (ref). The basic idea is to
-(dynamically) walk the Abstract C syntax tree, annotating it with
-possible code matches. For example, on the Sparc, a possible match
-(with its translation) could be
-@
- :=
- / \
- i r2 => ST r2,[r1]
- |
- r1
-@
-where @r1,r2@ are registers, and @i@ is an indirection. The Twig
-bit twiddling algorithm for tree matching has been abandoned. It is
-replaced with a more direct scheme. This is because, after careful
-consideration it is felt that the overhead of handling many bit
-patterns would be heavier that simply looking at the syntax of the
-tree at the node being considered, and dynamically choosing and
-pruning rules.
-
-The ultimate result of the first part is a Set of ordering lists of
-ordering lists of assembly language instructions (yes, really!), where
-each element in the set is basic chunk. Now several (generic)
-simplifications and transformations can be performed. This includes
-ones that turn the the ordering of orderings into just a single
-ordering list. (The equivalent of applying @concat@ to a list of
-lists.) A lot of the re-ordering and optimisation is actually done
-(generically) here! The final part, the scheduler, can now be used on
-this structure. The code sequence is optimised (obviously) to avoid
-stalling the pipeline. This part {\em has} to be heavily machine
-dependent.
-
-[The above seems to describe mostly dreamware. -- JSM]
-
-The flag that needs to be added is -fasm-<platform> where platform is one of
-the choices below.
-
-\begin{code}
-
-#ifdef __GLASGOW_HASKELL__
-# if __GLASGOW_HASKELL__ < 23
-# define _FILE _Addr
-# endif
-writeRealAsm :: (GlobalSwitch -> SwitchResult) -> _FILE -> AbstractC -> SplitUniqSupply -> PrimIO ()
-
-writeRealAsm flags file absC uniq_supply
- = uppAppendFile file 80 (runNCG (code flags absC) uniq_supply)
+module AsmCodeGen ( nativeCodeGen ) where
+#include "HsVersions.h"
+#include "NCG.h"
+
+import MachInstrs
+import MachRegs
+import MachCodeGen
+import PprMach
+import RegisterAlloc
+import RegAllocInfo ( jumpDests )
+import NCGMonad
+import PositionIndependentCode
+
+import Cmm
+import CmmOpt ( cmmMiniInline, cmmMachOpFold )
+import PprCmm ( pprStmt, pprCmms )
+import MachOp
+import CLabel ( CLabel, mkSplitMarkerLabel, mkAsmTempLabel )
+#if powerpc_TARGET_ARCH
+import CLabel ( mkRtsCodeLabel )
#endif
-dumpRealAsm :: (GlobalSwitch -> SwitchResult) -> AbstractC -> SplitUniqSupply -> String
-
-dumpRealAsm flags absC uniq_supply = uppShow 80 (runNCG (code flags absC) uniq_supply)
-
-runNCG m uniq_supply = m uniq_supply
+import UniqFM
+import Unique ( Unique, getUnique )
+import UniqSupply
+import FastTypes
+import List ( groupBy, sortBy )
+import CLabel ( pprCLabel )
+import ErrUtils ( dumpIfSet_dyn )
+import DynFlags ( DynFlags, DynFlag(..), dopt )
+import StaticFlags ( opt_Static, opt_PIC )
+
+import Digraph
+import qualified Pretty
+import Outputable
+import FastString
-code flags absC =
- genCodeAbstractC target absC `thenSUs` \ treelists ->
- let
- stix = map (map (genericOpt target)) treelists
- in
- codeGen target sty stix
- where
- sty = PprForAsm (switchIsOn flags) (underscore target) (fmtAsmLbl target)
+-- DEBUGGING ONLY
+--import OrdList
- target = case stringSwitchSet flags AsmTarget of
-#if ! OMIT_NATIVE_CODEGEN
-#if sparc_sun_sunos4_TARGET
- Just _ {-???"sparc-sun-sunos4"-} -> mkSparc True flags
+#ifdef NCG_DEBUG
+import List ( intersperse )
#endif
-#if sparc_sun_solaris2_TARGET
- Just _ {-???"sparc-sun-solaris2"-} -> mkSparc False flags
-#endif
-#if alpha_TARGET_ARCH
- Just _ {-???"alpha-dec-osf1"-} -> mkAlpha flags
-#endif
-#endif
- _ -> error
- ("ERROR:Trying to generate assembly language for an unsupported architecture\n"++
- "(or one for which this build is not configured).")
-
-\end{code}
-
-%************************************************************************
-%* *
-\subsection[NCOpt]{The Generic Optimiser}
-%* *
-%************************************************************************
-
-This is called between translating Abstract C to its Tree
-and actually using the Native Code Generator to generate
-the annotations. It's a chance to do some strength reductions.
-
-** Remember these all have to be machine independent ***
-
-Note that constant-folding should have already happened, but we might have
-introduced some new opportunities for constant-folding wrt address manipulations.
-
-\begin{code}
-
-genericOpt
- :: Target
- -> StixTree
- -> StixTree
-
-\end{code}
-
-For most nodes, just optimize the children.
-
-\begin{code}
-
-genericOpt target (StInd pk addr) =
- StInd pk (genericOpt target addr)
-
-genericOpt target (StAssign pk dst src) =
- StAssign pk (genericOpt target dst) (genericOpt target src)
-
-genericOpt target (StJump addr) =
- StJump (genericOpt target addr)
-
-genericOpt target (StCondJump addr test) =
- StCondJump addr (genericOpt target test)
-
-genericOpt target (StCall fn pk args) =
- StCall fn pk (map (genericOpt target) args)
-
-\end{code}
-
-Fold indices together when the types match.
-
-\begin{code}
-
-genericOpt target (StIndex pk (StIndex pk' base off) off')
- | pk == pk' =
- StIndex pk (genericOpt target base)
- (genericOpt target (StPrim IntAddOp [off, off']))
-
-genericOpt target (StIndex pk base off) =
- StIndex pk (genericOpt target base)
- (genericOpt target off)
-
-\end{code}
-For primOps, we first optimize the children, and then we try our hand
-at some constant-folding.
+import DATA_INT
+import DATA_WORD
+import DATA_BITS
+import GLAEXTS
-\begin{code}
-
-genericOpt target (StPrim op args) =
- primOpt op (map (genericOpt target) args)
+{-
+The native-code generator has machine-independent and
+machine-dependent modules.
-\end{code}
+This module ("AsmCodeGen") is the top-level machine-independent
+module. Before entering machine-dependent land, we do some
+machine-independent optimisations (defined below) on the
+'CmmStmts's.
-Replace register leaves with appropriate StixTrees for the given target.
-(Oh, so this is why we've been hauling the target around!)
+We convert to the machine-specific 'Instr' datatype with
+'cmmCodeGen', assuming an infinite supply of registers. We then use
+a machine-independent register allocator ('regAlloc') to rejoin
+reality. Obviously, 'regAlloc' has machine-specific helper
+functions (see about "RegAllocInfo" below).
-\begin{code}
+Finally, we order the basic blocks of the function so as to minimise
+the number of jumps between blocks, by utilising fallthrough wherever
+possible.
-genericOpt target leaf@(StReg (StixMagicId id)) =
- case stgReg target id of
- Always tree -> genericOpt target tree
- Save _ -> leaf
-
-genericOpt target other = other
-
-\end{code}
-
-Now, try to constant-fold the primOps. The arguments have
-already been optimized and folded.
-
-\begin{code}
-
-primOpt
- :: PrimOp -- The operation from an StPrim
- -> [StixTree] -- The optimized arguments
- -> StixTree
-
-primOpt op arg@[StInt x] =
- case op of
- IntNegOp -> StInt (-x)
- IntAbsOp -> StInt (abs x)
- _ -> StPrim op arg
-
-primOpt op args@[StInt x, StInt y] =
- case op of
- CharGtOp -> StInt (if x > y then 1 else 0)
- CharGeOp -> StInt (if x >= y then 1 else 0)
- CharEqOp -> StInt (if x == y then 1 else 0)
- CharNeOp -> StInt (if x /= y then 1 else 0)
- CharLtOp -> StInt (if x < y then 1 else 0)
- CharLeOp -> StInt (if x <= y then 1 else 0)
- IntAddOp -> StInt (x + y)
- IntSubOp -> StInt (x - y)
- IntMulOp -> StInt (x * y)
- IntQuotOp -> StInt (x `quot` y)
- IntDivOp -> StInt (x `div` y)
- IntRemOp -> StInt (x `rem` y)
- IntGtOp -> StInt (if x > y then 1 else 0)
- IntGeOp -> StInt (if x >= y then 1 else 0)
- IntEqOp -> StInt (if x == y then 1 else 0)
- IntNeOp -> StInt (if x /= y then 1 else 0)
- IntLtOp -> StInt (if x < y then 1 else 0)
- IntLeOp -> StInt (if x <= y then 1 else 0)
- _ -> StPrim op args
-
-\end{code}
+The machine-dependent bits break down as follows:
-When possible, shift the constants to the right-hand side, so that we
-can match for strength reductions. Note that the code generator will
-also assume that constants have been shifted to the right when possible.
+ * ["MachRegs"] Everything about the target platform's machine
+ registers (and immediate operands, and addresses, which tend to
+ intermingle/interact with registers).
-\begin{code}
+ * ["MachInstrs"] Includes the 'Instr' datatype (possibly should
+ have a module of its own), plus a miscellany of other things
+ (e.g., 'targetDoubleSize', 'smStablePtrTable', ...)
-primOpt op [x@(StInt _), y] | commutableOp op = primOpt op [y, x]
---OLD:
---primOpt op [x@(StDouble _), y] | commutableOp op = primOpt op [y, x]
+ * ["MachCodeGen"] is where 'Cmm' stuff turns into
+ machine instructions.
-\end{code}
+ * ["PprMach"] 'pprInstr' turns an 'Instr' into text (well, really
+ a 'Doc').
-We can often do something with constants of 0 and 1 ...
+ * ["RegAllocInfo"] In the register allocator, we manipulate
+ 'MRegsState's, which are 'BitSet's, one bit per machine register.
+ When we want to say something about a specific machine register
+ (e.g., ``it gets clobbered by this instruction''), we set/unset
+ its bit. Obviously, we do this 'BitSet' thing for efficiency
+ reasons.
-\begin{code}
+ The 'RegAllocInfo' module collects together the machine-specific
+ info needed to do register allocation.
-primOpt op args@[x, y@(StInt 0)] =
- case op of
- IntAddOp -> x
- IntSubOp -> x
- IntMulOp -> y
- AndOp -> y
- OrOp -> x
- SllOp -> x
- SraOp -> x
- SrlOp -> x
- ISllOp -> x
- ISraOp -> x
- ISrlOp -> x
- _ -> StPrim op args
-
-primOpt op args@[x, y@(StInt 1)] =
- case op of
- IntMulOp -> x
- IntDivOp -> x
- IntQuotOp -> x
- IntRemOp -> StInt 0
- _ -> StPrim op args
-
--- The following code tweaks a bug in early versions of GHC (pre-0.21)
-
-{- OLD: (death to constant folding in ncg)
-primOpt op args@[x, y@(StDouble 0.0)] =
- case op of
- FloatAddOp -> x
- FloatSubOp -> x
- FloatMulOp -> y
- DoubleAddOp -> x
- DoubleSubOp -> x
- DoubleMulOp -> y
- _ -> StPrim op args
-
-primOpt op args@[x, y@(StDouble 1.0)] =
- case op of
- FloatMulOp -> x
- FloatDivOp -> x
- DoubleMulOp -> x
- DoubleDivOp -> x
- _ -> StPrim op args
-
-primOpt op args@[x, y@(StDouble 2.0)] =
- case op of
- FloatMulOp -> StPrim FloatAddOp [x, x]
- DoubleMulOp -> StPrim DoubleAddOp [x, x]
- _ -> StPrim op args
+ * ["RegisterAlloc"] The (machine-independent) register allocator.
-}
-\end{code}
-
-Now look for multiplication/division by powers of 2 (integers).
-
-\begin{code}
-
-primOpt op args@[x, y@(StInt n)] =
- case op of
- IntMulOp -> case exact_log2 n of
- Nothing -> StPrim op args
- Just p -> StPrim SllOp [x, StInt p]
- IntQuotOp -> case exact_log2 n of
- Nothing -> StPrim op args
- Just p -> StPrim SraOp [x, StInt p]
- _ -> StPrim op args
-
-\end{code}
+-- -----------------------------------------------------------------------------
+-- Top-level of the native codegen
+
+-- NB. We *lazilly* compile each block of code for space reasons.
+
+nativeCodeGen :: DynFlags -> [Cmm] -> UniqSupply -> IO Pretty.Doc
+nativeCodeGen dflags cmms us
+ = let (res, _) = initUs us $
+ cgCmm (concat (map add_split cmms))
+
+ cgCmm :: [CmmTop] -> UniqSM (Cmm, Pretty.Doc, [CLabel])
+ cgCmm tops =
+ lazyMapUs (cmmNativeGen dflags) tops `thenUs` \ results ->
+ case unzip3 results of { (cmms,docs,imps) ->
+ returnUs (Cmm cmms, my_vcat docs, concat imps)
+ }
+ in
+ case res of { (ppr_cmms, insn_sdoc, imports) -> do
+ dumpIfSet_dyn dflags Opt_D_dump_opt_cmm "Optimised Cmm" (pprCmms [ppr_cmms])
+ return (insn_sdoc Pretty.$$ dyld_stubs imports
+#if HAVE_SUBSECTIONS_VIA_SYMBOLS
+ -- On recent versions of Darwin, the linker supports
+ -- dead-stripping of code and data on a per-symbol basis.
+ -- There's a hack to make this work in PprMach.pprNatCmmTop.
+ Pretty.$$ Pretty.text ".subsections_via_symbols"
+#endif
+ )
+ }
-Anything else is just too hard.
+ where
-\begin{code}
+ add_split (Cmm tops)
+ | dopt Opt_SplitObjs dflags = split_marker : tops
+ | otherwise = tops
+
+ split_marker = CmmProc [] mkSplitMarkerLabel [] []
+
+ -- Generate "symbol stubs" for all external symbols that might
+ -- come from a dynamic library.
+{- dyld_stubs imps = Pretty.vcat $ map pprDyldSymbolStub $
+ map head $ group $ sort imps-}
+
+ -- (Hack) sometimes two Labels pretty-print the same, but have
+ -- different uniques; so we compare their text versions...
+ dyld_stubs imps
+ | needImportedSymbols
+ = Pretty.vcat $
+ (pprGotDeclaration :) $
+ map (pprImportedSymbol . fst . head) $
+ groupBy (\(_,a) (_,b) -> a == b) $
+ sortBy (\(_,a) (_,b) -> compare a b) $
+ map doPpr $
+ imps
+ | otherwise
+ = Pretty.empty
+
+ where doPpr lbl = (lbl, Pretty.render $ pprCLabel lbl astyle)
+ astyle = mkCodeStyle AsmStyle
+
+#ifndef NCG_DEBUG
+ my_vcat sds = Pretty.vcat sds
+#else
+ my_vcat sds = Pretty.vcat (
+ intersperse (
+ Pretty.char ' '
+ Pretty.$$ Pretty.ptext SLIT("# ___ncg_debug_marker")
+ Pretty.$$ Pretty.char ' '
+ )
+ sds
+ )
+#endif
-primOpt op args = StPrim op args
-\end{code}
+-- Complete native code generation phase for a single top-level chunk
+-- of Cmm.
+
+cmmNativeGen :: DynFlags -> CmmTop -> UniqSM (CmmTop, Pretty.Doc, [CLabel])
+cmmNativeGen dflags cmm
+ = {-# SCC "fixAssigns" #-}
+ fixAssignsTop cmm `thenUs` \ fixed_cmm ->
+ {-# SCC "genericOpt" #-}
+ cmmToCmm fixed_cmm `bind` \ (cmm, imports) ->
+ (if dopt Opt_D_dump_opt_cmm dflags -- space leak avoidance
+ then cmm
+ else CmmData Text []) `bind` \ ppr_cmm ->
+ {-# SCC "genMachCode" #-}
+ genMachCode cmm `thenUs` \ (pre_regalloc, lastMinuteImports) ->
+ {-# SCC "regAlloc" #-}
+ mapUs regAlloc pre_regalloc `thenUs` \ with_regs ->
+ {-# SCC "sequenceBlocks" #-}
+ map sequenceTop with_regs `bind` \ sequenced ->
+ {-# SCC "x86fp_kludge" #-}
+ map x86fp_kludge sequenced `bind` \ final_mach_code ->
+ {-# SCC "vcat" #-}
+ Pretty.vcat (map pprNatCmmTop final_mach_code) `bind` \ final_sdoc ->
+
+ returnUs (ppr_cmm, final_sdoc Pretty.$$ Pretty.text "", lastMinuteImports ++ imports)
+ where
+ x86fp_kludge :: NatCmmTop -> NatCmmTop
+ x86fp_kludge top@(CmmData _ _) = top
+#if i386_TARGET_ARCH
+ x86fp_kludge top@(CmmProc info lbl params code) =
+ CmmProc info lbl params (map bb_i386_insert_ffrees code)
+ where
+ bb_i386_insert_ffrees (BasicBlock id instrs) =
+ BasicBlock id (i386_insert_ffrees instrs)
+#else
+ x86fp_kludge top = top
+#endif
-The commutable ops are those for which we will try to move constants to the
-right hand side for strength reduction.
+-- -----------------------------------------------------------------------------
+-- Sequencing the basic blocks
+
+-- Cmm BasicBlocks are self-contained entities: they always end in a
+-- jump, either non-local or to another basic block in the same proc.
+-- In this phase, we attempt to place the basic blocks in a sequence
+-- such that as many of the local jumps as possible turn into
+-- fallthroughs.
+
+sequenceTop :: NatCmmTop -> NatCmmTop
+sequenceTop top@(CmmData _ _) = top
+sequenceTop (CmmProc info lbl params blocks) =
+ CmmProc info lbl params (sequenceBlocks blocks)
+
+-- The algorithm is very simple (and stupid): we make a graph out of
+-- the blocks where there is an edge from one block to another iff the
+-- first block ends by jumping to the second. Then we topologically
+-- sort this graph. Then traverse the list: for each block, we first
+-- output the block, then if it has an out edge, we move the
+-- destination of the out edge to the front of the list, and continue.
+
+sequenceBlocks :: [NatBasicBlock] -> [NatBasicBlock]
+sequenceBlocks [] = []
+sequenceBlocks (entry:blocks) =
+ seqBlocks (mkNode entry : reverse (flattenSCCs (sccBlocks blocks)))
+ -- the first block is the entry point ==> it must remain at the start.
+
+sccBlocks :: [NatBasicBlock] -> [SCC (NatBasicBlock,Unique,[Unique])]
+sccBlocks blocks = stronglyConnCompR (map mkNode blocks)
+
+getOutEdges :: [Instr] -> [Unique]
+getOutEdges instrs = case jumpDests (last instrs) [] of
+ [one] -> [getUnique one]
+ _many -> []
+ -- we're only interested in the last instruction of
+ -- the block, and only if it has a single destination.
+
+mkNode block@(BasicBlock id instrs) = (block, getUnique id, getOutEdges instrs)
+
+seqBlocks [] = []
+seqBlocks ((block,_,[]) : rest)
+ = block : seqBlocks rest
+seqBlocks ((block@(BasicBlock id instrs),_,[next]) : rest)
+ | can_fallthrough = BasicBlock id (init instrs) : seqBlocks rest'
+ | otherwise = block : seqBlocks rest'
+ where
+ (can_fallthrough, rest') = reorder next [] rest
+ -- TODO: we should do a better job for cycles; try to maximise the
+ -- fallthroughs within a loop.
+seqBlocks _ = panic "AsmCodegen:seqBlocks"
+
+reorder id accum [] = (False, reverse accum)
+reorder id accum (b@(block,id',out) : rest)
+ | id == id' = (True, (block,id,out) : reverse accum ++ rest)
+ | otherwise = reorder id (b:accum) rest
+
+-- -----------------------------------------------------------------------------
+-- Instruction selection
+
+-- Native code instruction selection for a chunk of stix code. For
+-- this part of the computation, we switch from the UniqSM monad to
+-- the NatM monad. The latter carries not only a Unique, but also an
+-- Int denoting the current C stack pointer offset in the generated
+-- code; this is needed for creating correct spill offsets on
+-- architectures which don't offer, or for which it would be
+-- prohibitively expensive to employ, a frame pointer register. Viz,
+-- x86.
+
+-- The offset is measured in bytes, and indicates the difference
+-- between the current (simulated) C stack-ptr and the value it was at
+-- the beginning of the block. For stacks which grow down, this value
+-- should be either zero or negative.
+
+-- Switching between the two monads whilst carrying along the same
+-- Unique supply breaks abstraction. Is that bad?
+
+genMachCode :: CmmTop -> UniqSM ([NatCmmTop], [CLabel])
+
+genMachCode cmm_top initial_us
+ = let initial_st = mkNatM_State initial_us 0
+ (new_tops, final_st) = initNat initial_st (cmmTopCodeGen cmm_top)
+ final_us = natm_us final_st
+ final_delta = natm_delta final_st
+ final_imports = natm_imports final_st
+ in
+ if final_delta == 0
+ then ((new_tops, final_imports), final_us)
+ else pprPanic "genMachCode: nonzero final delta"
+ (int final_delta)
+
+-- -----------------------------------------------------------------------------
+-- Fixup assignments to global registers so that they assign to
+-- locations within the RegTable, if appropriate.
+
+-- Note that we currently don't fixup reads here: they're done by
+-- the generic optimiser below, to avoid having two separate passes
+-- over the Cmm.
+
+fixAssignsTop :: CmmTop -> UniqSM CmmTop
+fixAssignsTop top@(CmmData _ _) = returnUs top
+fixAssignsTop (CmmProc info lbl params blocks) =
+ mapUs fixAssignsBlock blocks `thenUs` \ blocks' ->
+ returnUs (CmmProc info lbl params blocks')
+
+fixAssignsBlock :: CmmBasicBlock -> UniqSM CmmBasicBlock
+fixAssignsBlock (BasicBlock id stmts) =
+ fixAssigns stmts `thenUs` \ stmts' ->
+ returnUs (BasicBlock id stmts')
+
+fixAssigns :: [CmmStmt] -> UniqSM [CmmStmt]
+fixAssigns stmts =
+ mapUs fixAssign stmts `thenUs` \ stmtss ->
+ returnUs (concat stmtss)
+
+fixAssign :: CmmStmt -> UniqSM [CmmStmt]
+fixAssign (CmmAssign (CmmGlobal BaseReg) src)
+ = panic "cmmStmtConFold: assignment to BaseReg";
+
+fixAssign (CmmAssign (CmmGlobal reg) src)
+ | Left realreg <- reg_or_addr
+ = returnUs [CmmAssign (CmmGlobal reg) src]
+ | Right baseRegAddr <- reg_or_addr
+ = returnUs [CmmStore baseRegAddr src]
+ -- Replace register leaves with appropriate StixTrees for
+ -- the given target. GlobalRegs which map to a reg on this
+ -- arch are left unchanged. Assigning to BaseReg is always
+ -- illegal, so we check for that.
+ where
+ reg_or_addr = get_GlobalReg_reg_or_addr reg
+
+fixAssign (CmmCall target results args vols)
+ = mapAndUnzipUs fixResult results `thenUs` \ (results',stores) ->
+ returnUs (caller_save ++
+ CmmCall target results' args vols :
+ caller_restore ++
+ concat stores)
+ where
+ -- we also save/restore any caller-saves STG registers here
+ (caller_save, caller_restore) = callerSaveVolatileRegs vols
+
+ fixResult g@(CmmGlobal reg,hint) =
+ case get_GlobalReg_reg_or_addr reg of
+ Left realreg -> returnUs (g, [])
+ Right baseRegAddr ->
+ getUniqueUs `thenUs` \ uq ->
+ let local = CmmLocal (LocalReg uq (globalRegRep reg)) in
+ returnUs ((local,hint),
+ [CmmStore baseRegAddr (CmmReg local)])
+ fixResult other =
+ returnUs (other,[])
+
+fixAssign other_stmt = returnUs [other_stmt]
+
+-- -----------------------------------------------------------------------------
+-- Generic Cmm optimiser
+
+{-
+Here we do:
+
+ (a) Constant folding
+ (b) Simple inlining: a temporary which is assigned to and then
+ used, once, can be shorted.
+ (c) Replacement of references to GlobalRegs which do not have
+ machine registers by the appropriate memory load (eg.
+ Hp ==> *(BaseReg + 34) ).
+ (d) Position independent code and dynamic linking
+ (i) introduce the appropriate indirections
+ and position independent refs
+ (ii) compile a list of imported symbols
+
+Ideas for other things we could do (ToDo):
+
+ - shortcut jumps-to-jumps
+ - eliminate dead code blocks
+ - simple CSE: if an expr is assigned to a temp, then replace later occs of
+ that expr with the temp, until the expr is no longer valid (can push through
+ temp assignments, and certain assigns to mem...)
+-}
-\begin{code}
+cmmToCmm :: CmmTop -> (CmmTop, [CLabel])
+cmmToCmm top@(CmmData _ _) = (top, [])
+cmmToCmm (CmmProc info lbl params blocks) = runCmmOpt $ do
+ blocks' <- mapM cmmBlockConFold (cmmMiniInline blocks)
+ return $ CmmProc info lbl params blocks'
+
+newtype CmmOptM a = CmmOptM ([CLabel] -> (# a, [CLabel] #))
+
+instance Monad CmmOptM where
+ return x = CmmOptM $ \imports -> (# x,imports #)
+ (CmmOptM f) >>= g =
+ CmmOptM $ \imports ->
+ case f imports of
+ (# x, imports' #) ->
+ case g x of
+ CmmOptM g' -> g' imports'
+
+addImportCmmOpt :: CLabel -> CmmOptM ()
+addImportCmmOpt lbl = CmmOptM $ \imports -> (# (), lbl:imports #)
+
+runCmmOpt :: CmmOptM a -> (a, [CLabel])
+runCmmOpt (CmmOptM f) = case f [] of
+ (# result, imports #) -> (result, imports)
+
+cmmBlockConFold :: CmmBasicBlock -> CmmOptM CmmBasicBlock
+cmmBlockConFold (BasicBlock id stmts) = do
+ stmts' <- mapM cmmStmtConFold stmts
+ return $ BasicBlock id stmts'
+
+cmmStmtConFold stmt
+ = case stmt of
+ CmmAssign reg src
+ -> do src' <- cmmExprConFold False src
+ return $ case src' of
+ CmmReg reg' | reg == reg' -> CmmNop
+ new_src -> CmmAssign reg new_src
+
+ CmmStore addr src
+ -> do addr' <- cmmExprConFold False addr
+ src' <- cmmExprConFold False src
+ return $ CmmStore addr' src'
+
+ CmmJump addr regs
+ -> do addr' <- cmmExprConFold True addr
+ return $ CmmJump addr' regs
+
+ CmmCall target regs args vols
+ -> do target' <- case target of
+ CmmForeignCall e conv -> do
+ e' <- cmmExprConFold True e
+ return $ CmmForeignCall e' conv
+ other -> return other
+ args' <- mapM (\(arg, hint) -> do
+ arg' <- cmmExprConFold False arg
+ return (arg', hint)) args
+ return $ CmmCall target' regs args' vols
+
+ CmmCondBranch test dest
+ -> do test' <- cmmExprConFold False test
+ return $ case test' of
+ CmmLit (CmmInt 0 _) ->
+ CmmComment (mkFastString ("deleted: " ++
+ showSDoc (pprStmt stmt)))
+
+ CmmLit (CmmInt n _) -> CmmBranch dest
+ other -> CmmCondBranch test' dest
+
+ CmmSwitch expr ids
+ -> do expr' <- cmmExprConFold False expr
+ return $ CmmSwitch expr' ids
+
+ other
+ -> return other
+
+
+cmmExprConFold isJumpTarget expr
+ = case expr of
+ CmmLoad addr rep
+ -> do addr' <- cmmExprConFold False addr
+ return $ CmmLoad addr' rep
+
+ CmmMachOp mop args
+ -- For MachOps, we first optimize the children, and then we try
+ -- our hand at some constant-folding.
+ -> do args' <- mapM (cmmExprConFold False) args
+ return $ cmmMachOpFold mop args'
+
+ CmmLit (CmmLabel lbl)
+ -> cmmMakeDynamicReference addImportCmmOpt isJumpTarget lbl
+ CmmLit (CmmLabelOff lbl off)
+ -> do dynRef <- cmmMakeDynamicReference addImportCmmOpt isJumpTarget lbl
+ return $ cmmMachOpFold (MO_Add wordRep) [
+ dynRef,
+ (CmmLit $ CmmInt (fromIntegral off) wordRep)
+ ]
+
+#if powerpc_TARGET_ARCH
+ -- On powerpc (non-PIC), it's easier to jump directly to a label than
+ -- to use the register table, so we replace these registers
+ -- with the corresponding labels:
+ CmmReg (CmmGlobal GCEnter1)
+ | not opt_PIC
+ -> cmmExprConFold isJumpTarget $
+ CmmLit (CmmLabel (mkRtsCodeLabel SLIT( "__stg_gc_enter_1")))
+ CmmReg (CmmGlobal GCFun)
+ | not opt_PIC
+ -> cmmExprConFold isJumpTarget $
+ CmmLit (CmmLabel (mkRtsCodeLabel SLIT( "__stg_gc_fun")))
+#endif
-commutableOp :: PrimOp -> Bool
-commutableOp CharEqOp = True
-commutableOp CharNeOp = True
-commutableOp IntAddOp = True
-commutableOp IntMulOp = True
-commutableOp AndOp = True
-commutableOp OrOp = True
-commutableOp IntEqOp = True
-commutableOp IntNeOp = True
-commutableOp IntegerAddOp = True
-commutableOp IntegerMulOp = True
-commutableOp FloatAddOp = True
-commutableOp FloatMulOp = True
-commutableOp FloatEqOp = True
-commutableOp FloatNeOp = True
-commutableOp DoubleAddOp = True
-commutableOp DoubleMulOp = True
-commutableOp DoubleEqOp = True
-commutableOp DoubleNeOp = True
-commutableOp _ = False
+ CmmReg (CmmGlobal mid)
+ -- Replace register leaves with appropriate StixTrees for
+ -- the given target. MagicIds which map to a reg on this
+ -- arch are left unchanged. For the rest, BaseReg is taken
+ -- to mean the address of the reg table in MainCapability,
+ -- and for all others we generate an indirection to its
+ -- location in the register table.
+ -> case get_GlobalReg_reg_or_addr mid of
+ Left realreg -> return expr
+ Right baseRegAddr
+ -> case mid of
+ BaseReg -> cmmExprConFold False baseRegAddr
+ other -> cmmExprConFold False (CmmLoad baseRegAddr
+ (globalRegRep mid))
+ -- eliminate zero offsets
+ CmmRegOff reg 0
+ -> cmmExprConFold False (CmmReg reg)
+
+ CmmRegOff (CmmGlobal mid) offset
+ -- RegOf leaves are just a shorthand form. If the reg maps
+ -- to a real reg, we keep the shorthand, otherwise, we just
+ -- expand it and defer to the above code.
+ -> case get_GlobalReg_reg_or_addr mid of
+ Left realreg -> return expr
+ Right baseRegAddr
+ -> cmmExprConFold False (CmmMachOp (MO_Add wordRep) [
+ CmmReg (CmmGlobal mid),
+ CmmLit (CmmInt (fromIntegral offset)
+ wordRep)])
+ other
+ -> return other
+
+-- -----------------------------------------------------------------------------
+-- Utils
+
+bind f x = x $! f
\end{code}
-This algorithm for determining the $\log_2$ of exact powers of 2 comes from gcc. It
-requires bit manipulation primitives, so we have a ghc version and an hbc version.
-Other Haskell compilers are on their own.
-
-\begin{code}
-
-#ifdef __GLASGOW_HASKELL__
-
-w2i x = word2Int# x
-i2w x = int2Word# x
-i2w_s x = (x::Int#)
-
-exact_log2 :: Integer -> Maybe Integer
-exact_log2 x
- | x <= 0 || x >= 2147483648 = Nothing
- | otherwise = case fromInteger x of
- I# x# -> if (w2i ((i2w x#) `and#` (i2w (0# -# x#))) /=# x#) then Nothing
- else Just (toInteger (I# (pow2 x#)))
-
- where pow2 x# | x# ==# 1# = 0#
- | otherwise = 1# +# pow2 (w2i (i2w x# `shiftr` i2w_s 1#))
-
-# if __GLASGOW_HASKELL__ >= 23
- shiftr x y = shiftRA# x y
-# else
- shiftr x y = shiftR# x y
-# endif
-
-#else {-probably HBC-}
-
-exact_log2 :: Integer -> Maybe Integer
-exact_log2 x
- | x <= 0 || x >= 2147483648 = Nothing
- | otherwise =
- if x' `bitAnd` (-x') /= x' then Nothing
- else Just (toInteger (pow2 x'))
-
- where x' = ((fromInteger x) :: Word)
- pow2 x | x == bit0 = 0 :: Int
- | otherwise = 1 + pow2 (x `bitRsh` 1)
-
-#endif {-probably HBC-}
-
-\end{code}
projects / ghc-hetmet.git / blobdiff