2 % (c) The GRASP/AQUA Project, Glasgow University, 1992-1998
4 % $Id: AbsCSyn.lhs,v 1.24 1999/06/24 13:04:13 simonmar Exp $
6 \section[AbstractC]{Abstract C: the last stop before machine code}
8 This ``Abstract C'' data type describes the raw Spineless Tagless
9 machine model at a C-ish level; it is ``abstract'' in that it only
10 includes C-like structures that we happen to need. The conversion of
11 programs from @StgSyntax@ (basically a functional language) to
12 @AbstractC@ (basically imperative C) is the heart of code generation.
13 From @AbstractC@, one may convert to real C (for portability) or to
14 raw assembler/machine code.
24 mkAbstractCs, mkAbsCStmts, mkAlgAltsCSwitch,
33 MagicId(..), node, infoptr,
38 #include "HsVersions.h"
40 import {-# SOURCE #-} ClosureInfo ( ClosureInfo )
42 #if ! OMIT_NATIVE_CODEGEN
43 import {-# SOURCE #-} MachMisc
47 import Constants ( mAX_Vanilla_REG, mAX_Float_REG,
48 mAX_Double_REG, spRelToInt )
49 import CostCentre ( CostCentre, CostCentreStack )
50 import Const ( mkMachInt, Literal(..) )
51 import PrimRep ( PrimRep(..) )
52 import PrimOp ( PrimOp )
53 import Unique ( Unique )
54 import StgSyn ( SRT(..) )
55 import TyCon ( TyCon )
56 import BitSet -- for liveness masks
60 @AbstractC@ is a list of Abstract~C statements, but the data structure
61 is tree-ish, for easier and more efficient putting-together.
67 | AbsCStmts AbstractC AbstractC
69 -- and the individual stmts...
72 A note on @CAssign@: In general, the type associated with an assignment
73 is the type of the lhs. However, when the lhs is a pointer to mixed
74 types (e.g. SpB relative), the type of the assignment is the type of
75 the rhs for float types, or the generic StgWord for all other types.
76 (In particular, a CharRep on the rhs is promoted to IntRep when
77 stored in a mixed type location.)
85 CAddrMode -- Put this in the program counter
86 -- eg `CJump (CReg (VanillaReg PtrRep 1))' puts Ret1 in PC
87 -- Enter can be done by:
88 -- CJump (CVal NodeRel zeroOff)
91 CAddrMode -- Fall through into this routine
92 -- (for the benefit of the native code generators)
93 -- Equivalent to CJump in C land
95 | CReturn -- Perform a return
96 CAddrMode -- Address of a RET_<blah> info table
97 ReturnInfo -- Whether it's a direct or vectored return
100 [(Literal, AbstractC)] -- alternatives
101 AbstractC -- default; if there is no real Abstract C in here
102 -- (e.g., all comments; see function "nonemptyAbsC"),
103 -- then that means the default _cannot_ occur.
104 -- If there is only one alternative & no default code,
105 -- then there is no need to check the tag.
107 -- CSwitch m [(tag,code)] AbsCNop == code
109 | CCodeBlock CLabel AbstractC
110 -- A labelled block of code; this "statement" is not
111 -- executed; rather, the labelled code will be hoisted
112 -- out to the top level (out of line) & it can be
115 | CInitHdr -- to initialise the header of a closure (both fixed/var parts)
117 CAddrMode -- address of the info ptr
118 CAddrMode -- cost centre to place in closure
119 -- CReg CurCostCentre or CC_HDR(R1.p{-Node-})
122 [CAddrMode] -- Results
124 [CAddrMode] -- Arguments
125 [MagicId] -- Potentially volatile/live registers
126 -- (to save/restore around the call/op)
128 -- INVARIANT: When a PrimOp which can cause GC is used, the
129 -- only live data is tidily on the STG stacks or in the STG
130 -- registers (the code generator ensures this).
132 -- Why this? Because if the arguments were arbitrary
133 -- addressing modes, they might be things like (Hp+6) which
134 -- will get utterly spongled by GC.
136 | CSimultaneous -- Perform simultaneously all the statements
137 AbstractC -- in the nested AbstractC. They are only
138 -- allowed to be CAssigns, COpStmts and AbsCNops, so the
139 -- "simultaneous" part just concerns making
140 -- sure that permutations work.
141 -- For example { a := b, b := a }
142 -- needs to go via (at least one) temporary
144 | CCheck -- heap or stack checks, or both.
145 CCheckMacro -- These might include some code to fill in tags
146 [CAddrMode] -- on the stack, so we can't use CMacroStmt below.
149 | CRetDirect -- Direct return
150 !Unique -- for making labels
151 AbstractC -- return code
152 (CLabel,SRT) -- SRT info
153 Liveness -- stack liveness at the return point
155 -- see the notes about these next few; they follow below...
156 | CMacroStmt CStmtMacro [CAddrMode]
157 | CCallProfCtrMacro FAST_STRING [CAddrMode]
158 | CCallProfCCMacro FAST_STRING [CAddrMode]
160 {- The presence of this constructor is a makeshift solution;
161 it being used to work around a gcc-related problem of
162 handling typedefs within statement blocks (or, rather,
163 the inability to do so.)
165 The AbstractC flattener takes care of lifting out these
166 typedefs if needs be (i.e., when generating .hc code and
167 compiling 'foreign import dynamic's)
169 | CCallTypedef PrimOp{-CCallOp-} [CAddrMode] [CAddrMode]
171 -- *** the next three [or so...] are DATA (those above are CODE) ***
174 CLabel -- The (full, not base) label to use for labelling the closure.
176 CAddrMode -- cost centre identifier to place in closure
177 [CAddrMode] -- free vars; ptrs, then non-ptrs.
179 | CSRT CLabel [CLabel] -- SRT declarations: basically an array of
180 -- pointers to static closures.
182 | CBitmap CLabel LivenessMask -- A larger-than-32-bits bitmap.
184 | CClosureInfoAndCode
185 ClosureInfo -- Explains placement and layout of closure
186 AbstractC -- Slow entry point code
188 -- Fast entry point code, if any
189 String -- Closure description; NB we can't get this
190 -- from ClosureInfo, because the latter refers
191 -- to the *right* hand side of a defn, whereas
192 -- the "description" refers to *left* hand side
194 | CRetVector -- A labelled block of static data
197 (CLabel,SRT) -- SRT info
198 Liveness -- stack liveness at the return point
200 | CClosureTbl -- table of constructors for enumerated types
201 TyCon -- which TyCon this table is for
203 | CCostCentreDecl -- A cost centre *declaration*
204 Bool -- True <=> local => full declaration
205 -- False <=> extern; just say so
208 | CCostCentreStackDecl -- A cost centre stack *declaration*
209 CostCentreStack -- this is the declaration for a
210 -- pre-defined singleton CCS (see
213 | CSplitMarker -- Split into separate object modules here
216 About @CMacroStmt@, etc.: notionally, they all just call some
217 arbitrary C~macro or routine, passing the @CAddrModes@ as arguments.
218 However, we distinguish between various flavours of these things,
219 mostly just to keep things somewhat less wild and wooly.
223 Some {\em essential} bits of the STG execution model are done with C
224 macros. An example is @STK_CHK@, which checks for stack-space
225 overflow. This enumeration type lists all such macros:
228 = ARGS_CHK -- arg satisfaction check
229 | ARGS_CHK_LOAD_NODE -- arg check for top-level functions
230 | UPD_CAF -- update CAF closure with indirection
231 | UPD_BH_UPDATABLE -- eager backholing
232 | UPD_BH_SINGLE_ENTRY -- more eager blackholing
233 | PUSH_UPD_FRAME -- push update frame
234 | PUSH_SEQ_FRAME -- push seq frame
235 | UPDATE_SU_FROM_UPD_FRAME -- pull Su out of the update frame
236 | SET_TAG -- set TagReg if it exists
237 | GRAN_FETCH -- for GrAnSim only -- HWL
238 | GRAN_RESCHEDULE -- for GrAnSim only -- HWL
239 | GRAN_FETCH_AND_RESCHEDULE -- for GrAnSim only -- HWL
240 | THREAD_CONTEXT_SWITCH -- for GrAnSim only -- HWL
241 | GRAN_YIELD -- for GrAnSim only -- HWL
244 Heap/Stack checks. There are far too many of these.
249 = HP_CHK_NP -- heap/stack checks when
250 | STK_CHK_NP -- node points to the closure
252 | HP_CHK_SEQ_NP -- for 'seq' style case alternatives
254 | HP_CHK -- heap/stack checks when
255 | STK_CHK -- node doesn't point
257 -- case alternative heap checks:
259 | HP_CHK_NOREGS -- no registers live
260 | HP_CHK_UNPT_R1 -- R1 is boxed/unlifted
261 | HP_CHK_UNBX_R1 -- R1 is unboxed
262 | HP_CHK_F1 -- FloatReg1 (only) is live
263 | HP_CHK_D1 -- DblReg1 (only) is live
264 | HP_CHK_L1 -- LngReg1 (only) is live
265 | HP_CHK_UT_ALT -- unboxed tuple return.
267 | HP_CHK_GEN -- generic heap check
270 \item[@CCallProfCtrMacro@:]
271 The @String@ names a macro that, if \tr{#define}d, will bump one/some
272 of the STG-event profiling counters.
274 \item[@CCallProfCCMacro@:]
275 The @String@ names a macro that, if \tr{#define}d, will perform some
276 cost-centre-profiling-related action.
279 %************************************************************************
281 \subsection[CAddrMode]{C addressing modes}
283 %************************************************************************
287 = CVal RegRelative PrimRep
288 -- On RHS of assign: Contents of Magic[n]
289 -- On LHS of assign: location Magic[n]
290 -- (ie at addr Magic+n)
293 -- On RHS of assign: Address of Magic[n]; ie Magic+n
294 -- n=0 gets the Magic location itself
295 -- (NB: n=0 case superceded by CReg)
296 -- On LHS of assign: only sensible if n=0,
297 -- which gives the magic location itself
298 -- (NB: superceded by CReg)
300 | CReg MagicId -- To replace (CAddr MagicId 0)
302 | CTemp !Unique !PrimRep -- Temporary locations
303 -- ``Temporaries'' correspond to local variables in C, and registers in
306 | CLbl CLabel -- Labels in the runtime system, etc.
307 PrimRep -- the kind is so we can generate accurate C decls
309 | CCharLike CAddrMode -- The address of a static char-like closure for
310 -- the specified character. It is guaranteed to be in
313 | CIntLike CAddrMode -- The address of a static int-like closure for the
314 -- specified small integer. It is guaranteed to be in
315 -- the range mIN_INTLIKE..mAX_INTLIKE
319 | CLitLit FAST_STRING -- completely literal literal: just spit this String
323 | CJoinPoint -- This is used as the amode of a let-no-escape-bound
325 VirtualSpOffset -- Sp value after any volatile free vars
326 -- of the rhs have been saved on stack.
327 -- Just before the code for the thing is jumped to,
328 -- Sp will be set to this value,
329 -- and then any stack-passed args pushed,
330 -- then the code for this thing will be entered
332 !PrimRep -- the kind of the result
333 CExprMacro -- the macro to generate a value
334 [CAddrMode] -- and its arguments
337 Various C macros for values which are dependent on the back-end layout.
343 | ARG_TAG -- stack argument tagging
344 | GET_TAG -- get current constructor tag
349 Convenience functions:
352 mkIntCLit :: Int -> CAddrMode
353 mkIntCLit i = CLit (mkMachInt (toInteger i))
355 mkCString :: FAST_STRING -> CAddrMode
356 mkCString s = CLit (MachStr s)
358 mkCCostCentre :: CostCentre -> CAddrMode
359 mkCCostCentre cc = CLbl (mkCC_Label cc) DataPtrRep
361 mkCCostCentreStack :: CostCentreStack -> CAddrMode
362 mkCCostCentreStack ccs = CLbl (mkCCS_Label ccs) DataPtrRep
365 %************************************************************************
367 \subsection[RegRelative]{@RegRelatives@: ???}
369 %************************************************************************
373 = HpRel FAST_INT -- }
374 | SpRel FAST_INT -- }- offsets in StgWords
375 | NodeRel FAST_INT -- }
376 | CIndex CAddrMode CAddrMode PrimRep -- pointer arithmetic :-)
377 -- CIndex a b k === (k*)a[b]
380 = DirectReturn -- Jump directly, if possible
381 | StaticVectoredReturn Int -- Fixed tag, starting at zero
382 | DynamicVectoredReturn CAddrMode -- Dynamic tag given by amode, starting at zero
384 hpRel :: VirtualHeapOffset -- virtual offset of Hp
385 -> VirtualHeapOffset -- virtual offset of The Thing
386 -> RegRelative -- integer offset
387 hpRel IBOX(hp) IBOX(off) = HpRel (hp _SUB_ off)
389 spRel :: VirtualSpOffset -- virtual offset of Sp
390 -> VirtualSpOffset -- virtual offset of The Thing
391 -> RegRelative -- integer offset
392 spRel sp off = SpRel (case spRelToInt sp off of { IBOX(i) -> i })
394 nodeRel :: VirtualHeapOffset
396 nodeRel IBOX(off) = NodeRel off
400 %************************************************************************
402 \subsection[Liveness]{Liveness Masks}
404 %************************************************************************
406 We represent liveness bitmaps as a BitSet (whose internal
407 representation really is a bitmap). These are pinned onto case return
408 vectors to indicate the state of the stack for the garbage collector.
411 type LivenessMask = [BitSet]
413 data Liveness = LvSmall BitSet
417 %************************************************************************
419 \subsection[HeapOffset]{@Heap Offsets@}
421 %************************************************************************
423 This used to be a grotesquely complicated datatype in an attempt to
424 hide the details of header sizes from the compiler itself. Now these
425 constants are imported from the RTS, and we deal in real Ints.
428 type HeapOffset = Int -- ToDo: remove
430 type VirtualHeapOffset = HeapOffset
431 type VirtualSpOffset = Int
433 type HpRelOffset = HeapOffset
434 type SpRelOffset = Int
437 %************************************************************************
439 \subsection[MagicId]{@MagicIds@: registers and such}
441 %************************************************************************
445 = BaseReg -- mentioned only in nativeGen
447 -- Argument and return registers
448 | VanillaReg -- pointers, unboxed ints and chars
450 FAST_INT -- its number (1 .. mAX_Vanilla_REG)
452 | FloatReg -- single-precision floating-point registers
453 FAST_INT -- its number (1 .. mAX_Float_REG)
455 | DoubleReg -- double-precision floating-point registers
456 FAST_INT -- its number (1 .. mAX_Double_REG)
459 | Sp -- Stack ptr; points to last occupied stack location.
460 | Su -- Stack update frame pointer
461 | SpLim -- Stack limit
462 | Hp -- Heap ptr; points to last occupied heap location.
463 | HpLim -- Heap limit register
464 | CurCostCentre -- current cost centre register.
465 | VoidReg -- see "VoidPrim" type; just a placeholder;
466 -- no actual register
467 | LongReg -- long int registers (64-bit, really)
468 PrimRep -- Int64Rep or Word64Rep
469 FAST_INT -- its number (1 .. mAX_Long_REG)
472 node = VanillaReg PtrRep ILIT(1) -- A convenient alias for Node
473 tagreg = VanillaReg WordRep ILIT(2) -- A convenient alias for TagReg
477 We need magical @Eq@ because @VanillaReg@s come in multiple flavors.
480 instance Eq MagicId where
481 reg1 == reg2 = tag reg1 _EQ_ tag reg2
483 tag BaseReg = (ILIT(0) :: FAST_INT)
489 tag CurCostCentre = ILIT(6)
490 tag VoidReg = ILIT(7)
492 tag (VanillaReg _ i) = ILIT(8) _ADD_ i
494 tag (FloatReg i) = ILIT(8) _ADD_ maxv _ADD_ i
495 tag (DoubleReg i) = ILIT(8) _ADD_ maxv _ADD_ maxf _ADD_ i
496 tag (LongReg _ i) = ILIT(8) _ADD_ maxv _ADD_ maxf _ADD_ maxd _ADD_ i
498 maxv = case mAX_Vanilla_REG of { IBOX(x) -> x }
499 maxf = case mAX_Float_REG of { IBOX(x) -> x }
500 maxd = case mAX_Double_REG of { IBOX(x) -> x }
503 Returns True for any register that {\em potentially} dies across
504 C calls (or anything near equivalent). We just say @True@ and
505 let the (machine-specific) registering macros sort things out...
508 isVolatileReg :: MagicId -> Bool
509 isVolatileReg any = True