2 % (c) The GRASP/AQUA Project, Glasgow University, 1992-1998
4 \section[SpecConstr]{Specialise over constructors}
7 -- The above warning supression flag is a temporary kludge.
8 -- While working on this module you are encouraged to remove it and fix
9 -- any warnings in the module. See
10 -- http://hackage.haskell.org/trac/ghc/wiki/Commentary/CodingStyle#Warnings
14 specConstrProgram, SpecConstrAnnotation(..)
17 #include "HsVersions.h"
22 import CoreUnfold ( couldBeSmallEnoughToInline )
23 import CoreFVs ( exprsFreeVars )
25 import HscTypes ( ModGuts(..) )
26 import WwLib ( mkWorkerArgs )
27 import DataCon ( dataConTyCon, dataConRepArity, dataConUnivTyVars )
28 import TyCon ( TyCon )
29 import Literal ( literalType )
32 import Type hiding( substTy )
34 import MkId ( mkImpossibleExpr )
39 import DynFlags ( DynFlags(..) )
40 import StaticFlags ( opt_PprStyle_Debug )
41 import StaticFlags ( opt_SpecInlineJoinPoints )
42 import BasicTypes ( Activation(..) )
43 import Maybes ( orElse, catMaybes, isJust, isNothing )
45 import DmdAnal ( both )
46 import Serialized ( deserializeWithData )
52 import qualified LazyUniqFM as L
54 import Control.Monad ( zipWithM )
56 #if __GLASGOW_HASKELL__ > 609
57 import Data.Data ( Data, Typeable )
59 import Data.Generics ( Data, Typeable )
63 -----------------------------------------------------
65 -----------------------------------------------------
70 drop n (x:xs) = drop (n-1) xs
72 After the first time round, we could pass n unboxed. This happens in
73 numerical code too. Here's what it looks like in Core:
75 drop n xs = case xs of
80 _ -> drop (I# (n# -# 1#)) xs
82 Notice that the recursive call has an explicit constructor as argument.
83 Noticing this, we can make a specialised version of drop
85 RULE: drop (I# n#) xs ==> drop' n# xs
87 drop' n# xs = let n = I# n# in ...orig RHS...
89 Now the simplifier will apply the specialisation in the rhs of drop', giving
91 drop' n# xs = case xs of
95 _ -> drop (n# -# 1#) xs
99 We'd also like to catch cases where a parameter is carried along unchanged,
100 but evaluated each time round the loop:
102 f i n = if i>0 || i>n then i else f (i*2) n
104 Here f isn't strict in n, but we'd like to avoid evaluating it each iteration.
105 In Core, by the time we've w/wd (f is strict in i) we get
107 f i# n = case i# ># 0 of
109 True -> case n of n' { I# n# ->
112 True -> f (i# *# 2#) n'
114 At the call to f, we see that the argument, n is know to be (I# n#),
115 and n is evaluated elsewhere in the body of f, so we can play the same
121 We must be careful not to allocate the same constructor twice. Consider
122 f p = (...(case p of (a,b) -> e)...p...,
123 ...let t = (r,s) in ...t...(f t)...)
124 At the recursive call to f, we can see that t is a pair. But we do NOT want
125 to make a specialised copy:
126 f' a b = let p = (a,b) in (..., ...)
127 because now t is allocated by the caller, then r and s are passed to the
128 recursive call, which allocates the (r,s) pair again.
131 (a) the argument p is used in other than a case-scrutinsation way.
132 (b) the argument to the call is not a 'fresh' tuple; you have to
133 look into its unfolding to see that it's a tuple
135 Hence the "OR" part of Note [Good arguments] below.
137 ALTERNATIVE 2: pass both boxed and unboxed versions. This no longer saves
138 allocation, but does perhaps save evals. In the RULE we'd have
141 f (I# x#) = f' (I# x#) x#
143 If at the call site the (I# x) was an unfolding, then we'd have to
144 rely on CSE to eliminate the duplicate allocation.... This alternative
145 doesn't look attractive enough to pursue.
147 ALTERNATIVE 3: ignore the reboxing problem. The trouble is that
148 the conservative reboxing story prevents many useful functions from being
149 specialised. Example:
150 foo :: Maybe Int -> Int -> Int
152 foo x@(Just m) n = foo x (n-m)
153 Here the use of 'x' will clearly not require boxing in the specialised function.
155 The strictness analyser has the same problem, in fact. Example:
157 If we pass just 'a' and 'b' to the worker, it might need to rebox the
158 pair to create (a,b). A more sophisticated analysis might figure out
159 precisely the cases in which this could happen, but the strictness
160 analyser does no such analysis; it just passes 'a' and 'b', and hopes
163 So my current choice is to make SpecConstr similarly aggressive, and
164 ignore the bad potential of reboxing.
167 Note [Good arguments]
168 ~~~~~~~~~~~~~~~~~~~~~
171 * A self-recursive function. Ignore mutual recursion for now,
172 because it's less common, and the code is simpler for self-recursion.
176 a) At a recursive call, one or more parameters is an explicit
177 constructor application
179 That same parameter is scrutinised by a case somewhere in
180 the RHS of the function
184 b) At a recursive call, one or more parameters has an unfolding
185 that is an explicit constructor application
187 That same parameter is scrutinised by a case somewhere in
188 the RHS of the function
190 Those are the only uses of the parameter (see Note [Reboxing])
193 What to abstract over
194 ~~~~~~~~~~~~~~~~~~~~~
195 There's a bit of a complication with type arguments. If the call
198 f p = ...f ((:) [a] x xs)...
200 then our specialised function look like
202 f_spec x xs = let p = (:) [a] x xs in ....as before....
204 This only makes sense if either
205 a) the type variable 'a' is in scope at the top of f, or
206 b) the type variable 'a' is an argument to f (and hence fs)
208 Actually, (a) may hold for value arguments too, in which case
209 we may not want to pass them. Supose 'x' is in scope at f's
210 defn, but xs is not. Then we'd like
212 f_spec xs = let p = (:) [a] x xs in ....as before....
214 Similarly (b) may hold too. If x is already an argument at the
215 call, no need to pass it again.
217 Finally, if 'a' is not in scope at the call site, we could abstract
218 it as we do the term variables:
220 f_spec a x xs = let p = (:) [a] x xs in ...as before...
222 So the grand plan is:
224 * abstract the call site to a constructor-only pattern
225 e.g. C x (D (f p) (g q)) ==> C s1 (D s2 s3)
227 * Find the free variables of the abstracted pattern
229 * Pass these variables, less any that are in scope at
230 the fn defn. But see Note [Shadowing] below.
233 NOTICE that we only abstract over variables that are not in scope,
234 so we're in no danger of shadowing variables used in "higher up"
240 In this pass we gather up usage information that may mention variables
241 that are bound between the usage site and the definition site; or (more
242 seriously) may be bound to something different at the definition site.
245 f x = letrec g y v = let x = ...
248 Since 'x' is in scope at the call site, we may make a rewrite rule that
250 RULE forall a,b. g (a,b) x = ...
251 But this rule will never match, because it's really a different 'x' at
252 the call site -- and that difference will be manifest by the time the
253 simplifier gets to it. [A worry: the simplifier doesn't *guarantee*
254 no-shadowing, so perhaps it may not be distinct?]
256 Anyway, the rule isn't actually wrong, it's just not useful. One possibility
257 is to run deShadowBinds before running SpecConstr, but instead we run the
258 simplifier. That gives the simplest possible program for SpecConstr to
259 chew on; and it virtually guarantees no shadowing.
261 Note [Specialising for constant parameters]
262 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
263 This one is about specialising on a *constant* (but not necessarily
264 constructor) argument
266 foo :: Int -> (Int -> Int) -> Int
268 foo m f = foo (f m) (+1)
272 lvl_rmV :: GHC.Base.Int -> GHC.Base.Int
274 \ (ds_dlk :: GHC.Base.Int) ->
275 case ds_dlk of wild_alH { GHC.Base.I# x_alG ->
276 GHC.Base.I# (GHC.Prim.+# x_alG 1)
278 T.$wfoo :: GHC.Prim.Int# -> (GHC.Base.Int -> GHC.Base.Int) ->
281 \ (ww_sme :: GHC.Prim.Int#) (w_smg :: GHC.Base.Int -> GHC.Base.Int) ->
282 case ww_sme of ds_Xlw {
284 case w_smg (GHC.Base.I# ds_Xlw) of w1_Xmo { GHC.Base.I# ww1_Xmz ->
285 T.$wfoo ww1_Xmz lvl_rmV
290 The recursive call has lvl_rmV as its argument, so we could create a specialised copy
291 with that argument baked in; that is, not passed at all. Now it can perhaps be inlined.
293 When is this worth it? Call the constant 'lvl'
294 - If 'lvl' has an unfolding that is a constructor, see if the corresponding
295 parameter is scrutinised anywhere in the body.
297 - If 'lvl' has an unfolding that is a inlinable function, see if the corresponding
298 parameter is applied (...to enough arguments...?)
300 Also do this is if the function has RULES?
304 Note [Specialising for lambda parameters]
305 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
306 foo :: Int -> (Int -> Int) -> Int
308 foo m f = foo (f m) (\n -> n-m)
310 This is subtly different from the previous one in that we get an
311 explicit lambda as the argument:
313 T.$wfoo :: GHC.Prim.Int# -> (GHC.Base.Int -> GHC.Base.Int) ->
316 \ (ww_sm8 :: GHC.Prim.Int#) (w_sma :: GHC.Base.Int -> GHC.Base.Int) ->
317 case ww_sm8 of ds_Xlr {
319 case w_sma (GHC.Base.I# ds_Xlr) of w1_Xmf { GHC.Base.I# ww1_Xmq ->
322 (\ (n_ad3 :: GHC.Base.Int) ->
323 case n_ad3 of wild_alB { GHC.Base.I# x_alA ->
324 GHC.Base.I# (GHC.Prim.-# x_alA ds_Xlr)
330 I wonder if SpecConstr couldn't be extended to handle this? After all,
331 lambda is a sort of constructor for functions and perhaps it already
332 has most of the necessary machinery?
334 Furthermore, there's an immediate win, because you don't need to allocate the lamda
335 at the call site; and if perchance it's called in the recursive call, then you
336 may avoid allocating it altogether. Just like for constructors.
338 Looks cool, but probably rare...but it might be easy to implement.
341 Note [SpecConstr for casts]
342 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
345 data instance T Int = T Int
350 go (T n) = go (T (n-1))
352 The recursive call ends up looking like
353 go (T (I# ...) `cast` g)
354 So we want to spot the construtor application inside the cast.
355 That's why we have the Cast case in argToPat
357 Note [Local recursive groups]
358 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
359 For a *local* recursive group, we can see all the calls to the
360 function, so we seed the specialisation loop from the calls in the
361 body, not from the calls in the RHS. Consider:
363 bar m n = foo n (n,n) (n,n) (n,n) (n,n)
367 | n > 3000 = case p of { (p1,p2) -> foo (n-1) (p2,p1) q r s }
368 | n > 2000 = case q of { (q1,q2) -> foo (n-1) p (q2,q1) r s }
369 | n > 1000 = case r of { (r1,r2) -> foo (n-1) p q (r2,r1) s }
370 | otherwise = case s of { (s1,s2) -> foo (n-1) p q r (s2,s1) }
372 If we start with the RHSs of 'foo', we get lots and lots of specialisations,
373 most of which are not needed. But if we start with the (single) call
374 in the rhs of 'bar' we get exactly one fully-specialised copy, and all
375 the recursive calls go to this fully-specialised copy. Indeed, the original
376 function is later collected as dead code. This is very important in
377 specialising the loops arising from stream fusion, for example in NDP where
378 we were getting literally hundreds of (mostly unused) specialisations of
381 Note [Do not specialise diverging functions]
382 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
383 Specialising a function that just diverges is a waste of code.
384 Furthermore, it broke GHC (simpl014) thus:
386 f = \x. case x of (a,b) -> f x
387 If we specialise f we get
388 f = \x. case x of (a,b) -> fspec a b
389 But fspec doesn't have decent strictnes info. As it happened,
390 (f x) :: IO t, so the state hack applied and we eta expanded fspec,
391 and hence f. But now f's strictness is less than its arity, which
394 -----------------------------------------------------
395 Stuff not yet handled
396 -----------------------------------------------------
398 Here are notes arising from Roman's work that I don't want to lose.
404 foo :: Int -> T Int -> Int
406 foo x t | even x = case t of { T n -> foo (x-n) t }
407 | otherwise = foo (x-1) t
409 SpecConstr does no specialisation, because the second recursive call
410 looks like a boxed use of the argument. A pity.
412 $wfoo_sFw :: GHC.Prim.Int# -> T.T GHC.Base.Int -> GHC.Prim.Int#
414 \ (ww_sFo [Just L] :: GHC.Prim.Int#) (w_sFq [Just L] :: T.T GHC.Base.Int) ->
415 case ww_sFo of ds_Xw6 [Just L] {
417 case GHC.Prim.remInt# ds_Xw6 2 of wild1_aEF [Dead Just A] {
418 __DEFAULT -> $wfoo_sFw (GHC.Prim.-# ds_Xw6 1) w_sFq;
420 case w_sFq of wild_Xy [Just L] { T.T n_ad5 [Just U(L)] ->
421 case n_ad5 of wild1_aET [Just A] { GHC.Base.I# y_aES [Just L] ->
422 $wfoo_sFw (GHC.Prim.-# ds_Xw6 y_aES) wild_Xy
428 data a :*: b = !a :*: !b
431 foo :: (Int :*: T Int) -> Int
433 foo (x :*: t) | even x = case t of { T n -> foo ((x-n) :*: t) }
434 | otherwise = foo ((x-1) :*: t)
436 Very similar to the previous one, except that the parameters are now in
437 a strict tuple. Before SpecConstr, we have
439 $wfoo_sG3 :: GHC.Prim.Int# -> T.T GHC.Base.Int -> GHC.Prim.Int#
441 \ (ww_sFU [Just L] :: GHC.Prim.Int#) (ww_sFW [Just L] :: T.T
443 case ww_sFU of ds_Xws [Just L] {
445 case GHC.Prim.remInt# ds_Xws 2 of wild1_aEZ [Dead Just A] {
447 case ww_sFW of tpl_B2 [Just L] { T.T a_sFo [Just A] ->
448 $wfoo_sG3 (GHC.Prim.-# ds_Xws 1) tpl_B2 -- $wfoo1
451 case ww_sFW of wild_XB [Just A] { T.T n_ad7 [Just S(L)] ->
452 case n_ad7 of wild1_aFd [Just L] { GHC.Base.I# y_aFc [Just L] ->
453 $wfoo_sG3 (GHC.Prim.-# ds_Xws y_aFc) wild_XB -- $wfoo2
457 We get two specialisations:
458 "SC:$wfoo1" [0] __forall {a_sFB :: GHC.Base.Int sc_sGC :: GHC.Prim.Int#}
459 Foo.$wfoo sc_sGC (Foo.T @ GHC.Base.Int a_sFB)
460 = Foo.$s$wfoo1 a_sFB sc_sGC ;
461 "SC:$wfoo2" [0] __forall {y_aFp :: GHC.Prim.Int# sc_sGC :: GHC.Prim.Int#}
462 Foo.$wfoo sc_sGC (Foo.T @ GHC.Base.Int (GHC.Base.I# y_aFp))
463 = Foo.$s$wfoo y_aFp sc_sGC ;
465 But perhaps the first one isn't good. After all, we know that tpl_B2 is
466 a T (I# x) really, because T is strict and Int has one constructor. (We can't
467 unbox the strict fields, becuase T is polymorphic!)
469 %************************************************************************
471 \subsection{Annotations}
473 %************************************************************************
475 Annotating a type with NoSpecConstr will make SpecConstr not specialise
476 for arguments of that type.
479 data SpecConstrAnnotation = NoSpecConstr deriving( Data, Typeable )
482 %************************************************************************
484 \subsection{Top level wrapper stuff}
486 %************************************************************************
489 specConstrProgram :: ModGuts -> CoreM ModGuts
490 specConstrProgram guts
492 dflags <- getDynFlags
493 us <- getUniqueSupplyM
494 annos <- deserializeAnnotations deserializeWithData
495 let binds' = fst $ initUs us (go (initScEnv dflags annos) (mg_binds guts))
496 return (guts { mg_binds = binds' })
499 go env (bind:binds) = do (env', bind') <- scTopBind env bind
500 binds' <- go env' binds
501 return (bind' : binds')
505 %************************************************************************
507 \subsection{Environment: goes downwards}
509 %************************************************************************
512 data ScEnv = SCE { sc_size :: Maybe Int, -- Size threshold
513 sc_count :: Maybe Int, -- Max # of specialisations for any one fn
515 sc_subst :: Subst, -- Current substitution
516 -- Maps InIds to OutExprs
518 sc_how_bound :: HowBoundEnv,
519 -- Binds interesting non-top-level variables
520 -- Domain is OutVars (*after* applying the substitution)
523 -- Domain is OutIds (*after* applying the substitution)
524 -- Used even for top-level bindings (but not imported ones)
526 sc_annotations :: L.UniqFM SpecConstrAnnotation
529 ---------------------
530 -- As we go, we apply a substitution (sc_subst) to the current term
531 type InExpr = CoreExpr -- _Before_ applying the subst
533 type OutExpr = CoreExpr -- _After_ applying the subst
537 ---------------------
538 type HowBoundEnv = VarEnv HowBound -- Domain is OutVars
540 ---------------------
541 type ValueEnv = IdEnv Value -- Domain is OutIds
542 data Value = ConVal AltCon [CoreArg] -- _Saturated_ constructors
543 | LambdaVal -- Inlinable lambdas or PAPs
545 instance Outputable Value where
546 ppr (ConVal con args) = ppr con <+> interpp'SP args
547 ppr LambdaVal = ptext (sLit "<Lambda>")
549 ---------------------
550 initScEnv :: DynFlags -> L.UniqFM [SpecConstrAnnotation] -> ScEnv
551 initScEnv dflags annos
552 = SCE { sc_size = specConstrThreshold dflags,
553 sc_count = specConstrCount dflags,
554 sc_subst = emptySubst,
555 sc_how_bound = emptyVarEnv,
556 sc_vals = emptyVarEnv,
557 sc_annotations = L.mapUFM head $ L.filterUFM (not . null) annos }
559 data HowBound = RecFun -- These are the recursive functions for which
560 -- we seek interesting call patterns
562 | RecArg -- These are those functions' arguments, or their sub-components;
563 -- we gather occurrence information for these
565 instance Outputable HowBound where
566 ppr RecFun = text "RecFun"
567 ppr RecArg = text "RecArg"
569 lookupHowBound :: ScEnv -> Id -> Maybe HowBound
570 lookupHowBound env id = lookupVarEnv (sc_how_bound env) id
572 scSubstId :: ScEnv -> Id -> CoreExpr
573 scSubstId env v = lookupIdSubst (sc_subst env) v
575 scSubstTy :: ScEnv -> Type -> Type
576 scSubstTy env ty = substTy (sc_subst env) ty
578 zapScSubst :: ScEnv -> ScEnv
579 zapScSubst env = env { sc_subst = zapSubstEnv (sc_subst env) }
581 extendScInScope :: ScEnv -> [Var] -> ScEnv
582 -- Bring the quantified variables into scope
583 extendScInScope env qvars = env { sc_subst = extendInScopeList (sc_subst env) qvars }
585 -- Extend the substitution
586 extendScSubst :: ScEnv -> Var -> OutExpr -> ScEnv
587 extendScSubst env var expr = env { sc_subst = extendSubst (sc_subst env) var expr }
589 extendScSubstList :: ScEnv -> [(Var,OutExpr)] -> ScEnv
590 extendScSubstList env prs = env { sc_subst = extendSubstList (sc_subst env) prs }
592 extendHowBound :: ScEnv -> [Var] -> HowBound -> ScEnv
593 extendHowBound env bndrs how_bound
594 = env { sc_how_bound = extendVarEnvList (sc_how_bound env)
595 [(bndr,how_bound) | bndr <- bndrs] }
597 extendBndrsWith :: HowBound -> ScEnv -> [Var] -> (ScEnv, [Var])
598 extendBndrsWith how_bound env bndrs
599 = (env { sc_subst = subst', sc_how_bound = hb_env' }, bndrs')
601 (subst', bndrs') = substBndrs (sc_subst env) bndrs
602 hb_env' = sc_how_bound env `extendVarEnvList`
603 [(bndr,how_bound) | bndr <- bndrs']
605 extendBndrWith :: HowBound -> ScEnv -> Var -> (ScEnv, Var)
606 extendBndrWith how_bound env bndr
607 = (env { sc_subst = subst', sc_how_bound = hb_env' }, bndr')
609 (subst', bndr') = substBndr (sc_subst env) bndr
610 hb_env' = extendVarEnv (sc_how_bound env) bndr' how_bound
612 extendRecBndrs :: ScEnv -> [Var] -> (ScEnv, [Var])
613 extendRecBndrs env bndrs = (env { sc_subst = subst' }, bndrs')
615 (subst', bndrs') = substRecBndrs (sc_subst env) bndrs
617 extendBndr :: ScEnv -> Var -> (ScEnv, Var)
618 extendBndr env bndr = (env { sc_subst = subst' }, bndr')
620 (subst', bndr') = substBndr (sc_subst env) bndr
622 extendValEnv :: ScEnv -> Id -> Maybe Value -> ScEnv
623 extendValEnv env _ Nothing = env
624 extendValEnv env id (Just cv) = env { sc_vals = extendVarEnv (sc_vals env) id cv }
626 extendCaseBndrs :: ScEnv -> Id -> AltCon -> [Var] -> (ScEnv, [Var])
630 -- we want to bind b, to (C x y)
631 -- NB1: Extends only the sc_vals part of the envt
632 -- NB2: Kill the dead-ness info on the pattern binders x,y, since
633 -- they are potentially made alive by the [b -> C x y] binding
634 extendCaseBndrs env case_bndr con alt_bndrs
635 | isDeadBinder case_bndr
638 = (env1, map zap alt_bndrs)
639 -- NB: We used to bind v too, if scrut = (Var v); but
640 -- the simplifer has already done this so it seems
641 -- redundant to do so here
643 -- Var v -> extendValEnv env1 v cval
646 zap v | isTyVar v = v -- See NB2 above
647 | otherwise = zapIdOccInfo v
648 env1 = extendValEnv env case_bndr cval
651 LitAlt {} -> Just (ConVal con [])
652 DataAlt {} -> Just (ConVal con vanilla_args)
654 vanilla_args = map Type (tyConAppArgs (idType case_bndr)) ++
655 varsToCoreExprs alt_bndrs
657 ignoreTyCon :: ScEnv -> TyCon -> Bool
658 ignoreTyCon env tycon
659 = case L.lookupUFM (sc_annotations env) tycon of
660 Just NoSpecConstr -> True
663 ignoreType :: ScEnv -> Type -> Bool
665 = case splitTyConApp_maybe ty of
666 Just (tycon, _) -> ignoreTyCon env tycon
669 ignoreAltCon :: ScEnv -> AltCon -> Bool
670 ignoreAltCon env (DataAlt dc) = ignoreTyCon env (dataConTyCon dc)
671 ignoreAltCon env (LitAlt lit) = ignoreType env (literalType lit)
672 ignoreAltCon _ DEFAULT = True
676 %************************************************************************
678 \subsection{Usage information: flows upwards}
680 %************************************************************************
685 scu_calls :: CallEnv, -- Calls
686 -- The functions are a subset of the
687 -- RecFuns in the ScEnv
689 scu_occs :: !(IdEnv ArgOcc) -- Information on argument occurrences
690 } -- The domain is OutIds
692 type CallEnv = IdEnv [Call]
693 type Call = (ValueEnv, [CoreArg])
694 -- The arguments of the call, together with the
695 -- env giving the constructor bindings at the call site
698 nullUsage = SCU { scu_calls = emptyVarEnv, scu_occs = emptyVarEnv }
700 combineCalls :: CallEnv -> CallEnv -> CallEnv
701 combineCalls = plusVarEnv_C (++)
703 combineUsage :: ScUsage -> ScUsage -> ScUsage
704 combineUsage u1 u2 = SCU { scu_calls = combineCalls (scu_calls u1) (scu_calls u2),
705 scu_occs = plusVarEnv_C combineOcc (scu_occs u1) (scu_occs u2) }
707 combineUsages :: [ScUsage] -> ScUsage
708 combineUsages [] = nullUsage
709 combineUsages us = foldr1 combineUsage us
711 lookupOcc :: ScUsage -> OutVar -> (ScUsage, ArgOcc)
712 lookupOcc (SCU { scu_calls = sc_calls, scu_occs = sc_occs }) bndr
713 = (SCU {scu_calls = sc_calls, scu_occs = delVarEnv sc_occs bndr},
714 lookupVarEnv sc_occs bndr `orElse` NoOcc)
716 lookupOccs :: ScUsage -> [OutVar] -> (ScUsage, [ArgOcc])
717 lookupOccs (SCU { scu_calls = sc_calls, scu_occs = sc_occs }) bndrs
718 = (SCU {scu_calls = sc_calls, scu_occs = delVarEnvList sc_occs bndrs},
719 [lookupVarEnv sc_occs b `orElse` NoOcc | b <- bndrs])
721 data ArgOcc = NoOcc -- Doesn't occur at all; or a type argument
722 | UnkOcc -- Used in some unknown way
724 | ScrutOcc (UniqFM [ArgOcc]) -- See Note [ScrutOcc]
726 | BothOcc -- Definitely taken apart, *and* perhaps used in some other way
730 An occurrence of ScrutOcc indicates that the thing, or a `cast` version of the thing,
731 is *only* taken apart or applied.
733 Functions, literal: ScrutOcc emptyUFM
734 Data constructors: ScrutOcc subs,
736 where (subs :: UniqFM [ArgOcc]) gives usage of the *pattern-bound* components,
737 The domain of the UniqFM is the Unique of the data constructor
739 The [ArgOcc] is the occurrences of the *pattern-bound* components
740 of the data structure. E.g.
741 data T a = forall b. MkT a b (b->a)
742 A pattern binds b, x::a, y::b, z::b->a, but not 'a'!
746 instance Outputable ArgOcc where
747 ppr (ScrutOcc xs) = ptext (sLit "scrut-occ") <> ppr xs
748 ppr UnkOcc = ptext (sLit "unk-occ")
749 ppr BothOcc = ptext (sLit "both-occ")
750 ppr NoOcc = ptext (sLit "no-occ")
752 -- Experimentally, this vesion of combineOcc makes ScrutOcc "win", so
753 -- that if the thing is scrutinised anywhere then we get to see that
754 -- in the overall result, even if it's also used in a boxed way
755 -- This might be too agressive; see Note [Reboxing] Alternative 3
756 combineOcc :: ArgOcc -> ArgOcc -> ArgOcc
757 combineOcc NoOcc occ = occ
758 combineOcc occ NoOcc = occ
759 combineOcc (ScrutOcc xs) (ScrutOcc ys) = ScrutOcc (plusUFM_C combineOccs xs ys)
760 combineOcc _occ (ScrutOcc ys) = ScrutOcc ys
761 combineOcc (ScrutOcc xs) _occ = ScrutOcc xs
762 combineOcc UnkOcc UnkOcc = UnkOcc
763 combineOcc _ _ = BothOcc
765 combineOccs :: [ArgOcc] -> [ArgOcc] -> [ArgOcc]
766 combineOccs xs ys = zipWithEqual "combineOccs" combineOcc xs ys
768 setScrutOcc :: ScEnv -> ScUsage -> OutExpr -> ArgOcc -> ScUsage
769 -- _Overwrite_ the occurrence info for the scrutinee, if the scrutinee
770 -- is a variable, and an interesting variable
771 setScrutOcc env usg (Cast e _) occ = setScrutOcc env usg e occ
772 setScrutOcc env usg (Note _ e) occ = setScrutOcc env usg e occ
773 setScrutOcc env usg (Var v) occ
774 | Just RecArg <- lookupHowBound env v = usg { scu_occs = extendVarEnv (scu_occs usg) v occ }
776 setScrutOcc _env usg _other _occ -- Catch-all
779 conArgOccs :: ArgOcc -> AltCon -> [ArgOcc]
780 -- Find usage of components of data con; returns [UnkOcc...] if unknown
781 -- See Note [ScrutOcc] for the extra UnkOccs in the vanilla datacon case
783 conArgOccs (ScrutOcc fm) (DataAlt dc)
784 | Just pat_arg_occs <- lookupUFM fm dc
785 = [UnkOcc | _ <- dataConUnivTyVars dc] ++ pat_arg_occs
787 conArgOccs _other _con = repeat UnkOcc
790 %************************************************************************
792 \subsection{The main recursive function}
794 %************************************************************************
796 The main recursive function gathers up usage information, and
797 creates specialised versions of functions.
800 scExpr, scExpr' :: ScEnv -> CoreExpr -> UniqSM (ScUsage, CoreExpr)
801 -- The unique supply is needed when we invent
802 -- a new name for the specialised function and its args
804 scExpr env e = scExpr' env e
807 scExpr' env (Var v) = case scSubstId env v of
808 Var v' -> return (varUsage env v' UnkOcc, Var v')
809 e' -> scExpr (zapScSubst env) e'
811 scExpr' env (Type t) = return (nullUsage, Type (scSubstTy env t))
812 scExpr' _ e@(Lit {}) = return (nullUsage, e)
813 scExpr' env (Note n e) = do (usg,e') <- scExpr env e
814 return (usg, Note n e')
815 scExpr' env (Cast e co) = do (usg, e') <- scExpr env e
816 return (usg, Cast e' (scSubstTy env co))
817 scExpr' env e@(App _ _) = scApp env (collectArgs e)
818 scExpr' env (Lam b e) = do let (env', b') = extendBndr env b
819 (usg, e') <- scExpr env' e
820 return (usg, Lam b' e')
822 scExpr' env (Case scrut b ty alts)
823 = do { (scrut_usg, scrut') <- scExpr env scrut
824 ; case isValue (sc_vals env) scrut' of
825 Just (ConVal con args) -> sc_con_app con args scrut'
826 _other -> sc_vanilla scrut_usg scrut'
829 sc_con_app con args scrut' -- Known constructor; simplify
830 = do { let (_, bs, rhs) = findAlt con alts
831 `orElse` (DEFAULT, [], mkImpossibleExpr (coreAltsType alts))
832 alt_env' = extendScSubstList env ((b,scrut') : bs `zip` trimConArgs con args)
833 ; scExpr alt_env' rhs }
835 sc_vanilla scrut_usg scrut' -- Normal case
836 = do { let (alt_env,b') = extendBndrWith RecArg env b
837 -- Record RecArg for the components
839 ; (alt_usgs, alt_occs, alts')
840 <- mapAndUnzip3M (sc_alt alt_env scrut' b') alts
842 ; let (alt_usg, b_occ) = lookupOcc (combineUsages alt_usgs) b'
843 scrut_occ = foldr combineOcc b_occ alt_occs
844 scrut_usg' = setScrutOcc env scrut_usg scrut' scrut_occ
845 -- The combined usage of the scrutinee is given
846 -- by scrut_occ, which is passed to scScrut, which
847 -- in turn treats a bare-variable scrutinee specially
849 ; return (alt_usg `combineUsage` scrut_usg',
850 Case scrut' b' (scSubstTy env ty) alts') }
852 sc_alt env _scrut' b' (con,bs,rhs)
853 = do { let (env1, bs1) = extendBndrsWith RecArg env bs
854 (env2, bs2) = extendCaseBndrs env1 b' con bs1
855 ; (usg,rhs') <- scExpr env2 rhs
856 ; let (usg', arg_occs) = lookupOccs usg bs2
857 scrut_occ = case con of
858 DataAlt dc -> ScrutOcc (unitUFM dc arg_occs)
859 _ -> ScrutOcc emptyUFM
860 ; return (usg', scrut_occ, (con, bs2, rhs')) }
862 scExpr' env (Let (NonRec bndr rhs) body)
863 | isTyVar bndr -- Type-lets may be created by doBeta
864 = scExpr' (extendScSubst env bndr rhs) body
866 = do { let (body_env, bndr') = extendBndr env bndr
867 ; (rhs_usg, (_, args', rhs_body', _)) <- scRecRhs env (bndr',rhs)
868 ; let rhs' = mkLams args' rhs_body'
870 ; if not opt_SpecInlineJoinPoints || null args' || isEmptyVarEnv (scu_calls rhs_usg) then do
872 let body_env2 = extendValEnv body_env bndr' (isValue (sc_vals env) rhs')
873 -- Record if the RHS is a value
874 ; (body_usg, body') <- scExpr body_env2 body
875 ; return (body_usg `combineUsage` rhs_usg, Let (NonRec bndr' rhs') body') }
876 else -- For now, just brutally inline the join point
877 do { let body_env2 = extendScSubst env bndr rhs'
878 ; scExpr body_env2 body } }
882 do { -- Join-point case
883 let body_env2 = extendHowBound body_env [bndr'] RecFun
884 -- If the RHS of this 'let' contains calls
885 -- to recursive functions that we're trying
886 -- to specialise, then treat this let too
887 -- as one to specialise
888 ; (body_usg, body') <- scExpr body_env2 body
890 ; (spec_usg, _, specs) <- specialise env (scu_calls body_usg) ([], rhs_info)
892 ; return (body_usg { scu_calls = scu_calls body_usg `delVarEnv` bndr' }
893 `combineUsage` rhs_usg `combineUsage` spec_usg,
894 mkLets [NonRec b r | (b,r) <- specInfoBinds rhs_info specs] body')
898 -- A *local* recursive group: see Note [Local recursive groups]
899 scExpr' env (Let (Rec prs) body)
900 = do { let (bndrs,rhss) = unzip prs
901 (rhs_env1,bndrs') = extendRecBndrs env bndrs
902 rhs_env2 = extendHowBound rhs_env1 bndrs' RecFun
904 ; (rhs_usgs, rhs_infos) <- mapAndUnzipM (scRecRhs rhs_env2) (bndrs' `zip` rhss)
905 ; (body_usg, body') <- scExpr rhs_env2 body
907 -- NB: start specLoop from body_usg
908 ; (spec_usg, specs) <- specLoop rhs_env2 (scu_calls body_usg) rhs_infos nullUsage
909 [SI [] 0 (Just usg) | usg <- rhs_usgs]
911 ; let all_usg = spec_usg `combineUsage` body_usg
912 bind' = Rec (concat (zipWith specInfoBinds rhs_infos specs))
914 ; return (all_usg { scu_calls = scu_calls all_usg `delVarEnvList` bndrs' },
917 -----------------------------------
918 scApp :: ScEnv -> (InExpr, [InExpr]) -> UniqSM (ScUsage, CoreExpr)
920 scApp env (Var fn, args) -- Function is a variable
921 = ASSERT( not (null args) )
922 do { args_w_usgs <- mapM (scExpr env) args
923 ; let (arg_usgs, args') = unzip args_w_usgs
924 arg_usg = combineUsages arg_usgs
925 ; case scSubstId env fn of
926 fn'@(Lam {}) -> scExpr (zapScSubst env) (doBeta fn' args')
927 -- Do beta-reduction and try again
929 Var fn' -> return (arg_usg `combineUsage` fn_usg, mkApps (Var fn') args')
931 fn_usg = case lookupHowBound env fn' of
932 Just RecFun -> SCU { scu_calls = unitVarEnv fn' [(sc_vals env, args')],
933 scu_occs = emptyVarEnv }
934 Just RecArg -> SCU { scu_calls = emptyVarEnv,
935 scu_occs = unitVarEnv fn' (ScrutOcc emptyUFM) }
939 other_fn' -> return (arg_usg, mkApps other_fn' args') }
940 -- NB: doing this ignores any usage info from the substituted
941 -- function, but I don't think that matters. If it does
944 doBeta :: OutExpr -> [OutExpr] -> OutExpr
945 -- ToDo: adjust for System IF
946 doBeta (Lam bndr body) (arg : args) = Let (NonRec bndr arg) (doBeta body args)
947 doBeta fn args = mkApps fn args
949 -- The function is almost always a variable, but not always.
950 -- In particular, if this pass follows float-in,
951 -- which it may, we can get
952 -- (let f = ...f... in f) arg1 arg2
953 scApp env (other_fn, args)
954 = do { (fn_usg, fn') <- scExpr env other_fn
955 ; (arg_usgs, args') <- mapAndUnzipM (scExpr env) args
956 ; return (combineUsages arg_usgs `combineUsage` fn_usg, mkApps fn' args') }
958 ----------------------
959 scTopBind :: ScEnv -> CoreBind -> UniqSM (ScEnv, CoreBind)
960 scTopBind env (Rec prs)
961 | Just threshold <- sc_size env
962 , not (all (couldBeSmallEnoughToInline threshold) rhss)
964 = do { let (rhs_env,bndrs') = extendRecBndrs env bndrs
965 ; (_, rhss') <- mapAndUnzipM (scExpr rhs_env) rhss
966 ; return (rhs_env, Rec (bndrs' `zip` rhss')) }
967 | otherwise -- Do specialisation
968 = do { let (rhs_env1,bndrs') = extendRecBndrs env bndrs
969 rhs_env2 = extendHowBound rhs_env1 bndrs' RecFun
971 ; (rhs_usgs, rhs_infos) <- mapAndUnzipM (scRecRhs rhs_env2) (bndrs' `zip` rhss)
972 ; let rhs_usg = combineUsages rhs_usgs
974 ; (_, specs) <- specLoop rhs_env2 (scu_calls rhs_usg) rhs_infos nullUsage
975 [SI [] 0 Nothing | _ <- bndrs]
977 ; return (rhs_env1, -- For the body of the letrec, delete the RecFun business
978 Rec (concat (zipWith specInfoBinds rhs_infos specs))) }
980 (bndrs,rhss) = unzip prs
982 scTopBind env (NonRec bndr rhs)
983 = do { (_, rhs') <- scExpr env rhs
984 ; let (env1, bndr') = extendBndr env bndr
985 env2 = extendValEnv env1 bndr' (isValue (sc_vals env) rhs')
986 ; return (env2, NonRec bndr' rhs') }
988 ----------------------
989 scRecRhs :: ScEnv -> (OutId, InExpr) -> UniqSM (ScUsage, RhsInfo)
990 scRecRhs env (bndr,rhs)
991 = do { let (arg_bndrs,body) = collectBinders rhs
992 (body_env, arg_bndrs') = extendBndrsWith RecArg env arg_bndrs
993 ; (body_usg, body') <- scExpr body_env body
994 ; let (rhs_usg, arg_occs) = lookupOccs body_usg arg_bndrs'
995 ; return (rhs_usg, (bndr, arg_bndrs', body', arg_occs)) }
997 -- The arg_occs says how the visible,
998 -- lambda-bound binders of the RHS are used
999 -- (including the TyVar binders)
1000 -- Two pats are the same if they match both ways
1002 ----------------------
1003 specInfoBinds :: RhsInfo -> SpecInfo -> [(Id,CoreExpr)]
1004 specInfoBinds (fn, args, body, _) (SI specs _ _)
1005 = [(id,rhs) | OS _ _ id rhs <- specs] ++
1006 [(fn `addIdSpecialisations` rules, mkLams args body)]
1008 rules = [r | OS _ r _ _ <- specs]
1010 ----------------------
1011 varUsage :: ScEnv -> OutVar -> ArgOcc -> ScUsage
1013 | Just RecArg <- lookupHowBound env v = SCU { scu_calls = emptyVarEnv
1014 , scu_occs = unitVarEnv v use }
1015 | otherwise = nullUsage
1019 %************************************************************************
1021 The specialiser itself
1023 %************************************************************************
1026 type RhsInfo = (OutId, [OutVar], OutExpr, [ArgOcc])
1027 -- Info about the *original* RHS of a binding we are specialising
1028 -- Original binding f = \xs.body
1029 -- Plus info about usage of arguments
1031 data SpecInfo = SI [OneSpec] -- The specialisations we have generated
1032 Int -- Length of specs; used for numbering them
1033 (Maybe ScUsage) -- Nothing => we have generated specialisations
1034 -- from calls in the *original* RHS
1035 -- Just cs => we haven't, and this is the usage
1036 -- of the original RHS
1038 -- One specialisation: Rule plus definition
1039 data OneSpec = OS CallPat -- Call pattern that generated this specialisation
1040 CoreRule -- Rule connecting original id with the specialisation
1041 OutId OutExpr -- Spec id + its rhs
1047 -> ScUsage -> [SpecInfo] -- One per binder; acccumulating parameter
1048 -> UniqSM (ScUsage, [SpecInfo]) -- ...ditto...
1049 specLoop env all_calls rhs_infos usg_so_far specs_so_far
1050 = do { specs_w_usg <- zipWithM (specialise env all_calls) rhs_infos specs_so_far
1051 ; let (new_usg_s, all_specs) = unzip specs_w_usg
1052 new_usg = combineUsages new_usg_s
1053 new_calls = scu_calls new_usg
1054 all_usg = usg_so_far `combineUsage` new_usg
1055 ; if isEmptyVarEnv new_calls then
1056 return (all_usg, all_specs)
1058 specLoop env new_calls rhs_infos all_usg all_specs }
1062 -> CallEnv -- Info on calls
1064 -> SpecInfo -- Original RHS plus patterns dealt with
1065 -> UniqSM (ScUsage, SpecInfo) -- New specialised versions and their usage
1067 -- Note: the rhs here is the optimised version of the original rhs
1068 -- So when we make a specialised copy of the RHS, we're starting
1069 -- from an RHS whose nested functions have been optimised already.
1071 specialise env bind_calls (fn, arg_bndrs, body, arg_occs)
1072 spec_info@(SI specs spec_count mb_unspec)
1073 | not (isBottomingId fn) -- Note [Do not specialise diverging functions]
1074 , notNull arg_bndrs -- Only specialise functions
1075 , Just all_calls <- lookupVarEnv bind_calls fn
1076 = do { (boring_call, pats) <- callsToPats env specs arg_occs all_calls
1077 -- ; pprTrace "specialise" (vcat [ppr fn <+> ppr arg_occs,
1078 -- text "calls" <+> ppr all_calls,
1079 -- text "good pats" <+> ppr pats]) $
1082 -- Bale out if too many specialisations
1083 -- Rather a hacky way to do so, but it'll do for now
1084 ; let spec_count' = length pats + spec_count
1085 ; case sc_count env of
1086 Just max | spec_count' > max
1087 -> WARN( True, msg ) return (nullUsage, spec_info)
1089 msg = vcat [ sep [ ptext (sLit "SpecConstr: specialisation of") <+> quotes (ppr fn)
1090 , nest 2 (ptext (sLit "limited by bound of")) <+> int max ]
1091 , ptext (sLit "Use -fspec-constr-count=n to set the bound")
1093 extra | not opt_PprStyle_Debug = ptext (sLit "Use -dppr-debug to see specialisations")
1094 | otherwise = ptext (sLit "Specialisations:") <+> ppr (pats ++ [p | OS p _ _ _ <- specs])
1096 _normal_case -> do {
1098 (spec_usgs, new_specs) <- mapAndUnzipM (spec_one env fn arg_bndrs body)
1099 (pats `zip` [spec_count..])
1101 ; let spec_usg = combineUsages spec_usgs
1102 (new_usg, mb_unspec')
1104 Just rhs_usg | boring_call -> (spec_usg `combineUsage` rhs_usg, Nothing)
1105 _ -> (spec_usg, mb_unspec)
1107 ; return (new_usg, SI (new_specs ++ specs) spec_count' mb_unspec') } }
1109 = return (nullUsage, spec_info) -- The boring case
1112 ---------------------
1114 -> OutId -- Function
1115 -> [Var] -- Lambda-binders of RHS; should match patterns
1116 -> CoreExpr -- Body of the original function
1118 -> UniqSM (ScUsage, OneSpec) -- Rule and binding
1120 -- spec_one creates a specialised copy of the function, together
1121 -- with a rule for using it. I'm very proud of how short this
1122 -- function is, considering what it does :-).
1128 f = /\b \y::[(a,b)] -> ....f (b,c) ((:) (a,(b,c)) (x,v) (h w))...
1129 [c::*, v::(b,c) are presumably bound by the (...) part]
1131 f_spec = /\ b c \ v::(b,c) hw::[(a,(b,c))] ->
1132 (...entire body of f...) [b -> (b,c),
1133 y -> ((:) (a,(b,c)) (x,v) hw)]
1135 RULE: forall b::* c::*, -- Note, *not* forall a, x
1139 f (b,c) ((:) (a,(b,c)) (x,v) hw) = f_spec b c v hw
1142 spec_one env fn arg_bndrs body (call_pat@(qvars, pats), rule_number)
1143 = do { -- Specialise the body
1144 let spec_env = extendScSubstList (extendScInScope env qvars)
1145 (arg_bndrs `zip` pats)
1146 ; (spec_usg, spec_body) <- scExpr spec_env body
1148 -- ; pprTrace "spec_one" (ppr fn <+> vcat [text "pats" <+> ppr pats,
1149 -- text "calls" <+> (ppr (scu_calls spec_usg))])
1152 -- And build the results
1153 ; spec_uniq <- getUniqueUs
1154 ; let (spec_lam_args, spec_call_args) = mkWorkerArgs qvars body_ty
1155 -- Usual w/w hack to avoid generating
1156 -- a spec_rhs of unlifted type and no args
1159 fn_loc = nameSrcSpan fn_name
1160 spec_occ = mkSpecOcc (nameOccName fn_name)
1161 rule_name = mkFastString ("SC:" ++ showSDoc (ppr fn <> int rule_number))
1162 spec_rhs = mkLams spec_lam_args spec_body
1163 spec_str = calcSpecStrictness fn spec_lam_args pats
1164 spec_id = mkUserLocal spec_occ spec_uniq (mkPiTypes spec_lam_args body_ty) fn_loc
1165 `setIdStrictness` spec_str -- See Note [Transfer strictness]
1166 `setIdArity` count isId spec_lam_args
1167 body_ty = exprType spec_body
1168 rule_rhs = mkVarApps (Var spec_id) spec_call_args
1169 rule = mkLocalRule rule_name specConstrActivation fn_name qvars pats rule_rhs
1170 ; return (spec_usg, OS call_pat rule spec_id spec_rhs) }
1172 calcSpecStrictness :: Id -- The original function
1173 -> [Var] -> [CoreExpr] -- Call pattern
1174 -> StrictSig -- Strictness of specialised thing
1175 -- See Note [Transfer strictness]
1176 calcSpecStrictness fn qvars pats
1177 = StrictSig (mkTopDmdType spec_dmds TopRes)
1179 spec_dmds = [ lookupVarEnv dmd_env qv `orElse` lazyDmd | qv <- qvars, isId qv ]
1180 StrictSig (DmdType _ dmds _) = idStrictness fn
1182 dmd_env = go emptyVarEnv dmds pats
1184 go env ds (Type {} : pats) = go env ds pats
1185 go env (d:ds) (pat : pats) = go (go_one env d pat) ds pats
1188 go_one env d (Var v) = extendVarEnv_C both env v d
1189 go_one env (Box d) e = go_one env d e
1190 go_one env (Eval (Prod ds)) e
1191 | (Var _, args) <- collectArgs e = go env ds args
1192 go_one env _ _ = env
1194 -- In which phase should the specialise-constructor rules be active?
1195 -- Originally I made them always-active, but Manuel found that
1196 -- this defeated some clever user-written rules. So Plan B
1197 -- is to make them active only in Phase 0; after all, currently,
1198 -- the specConstr transformation is only run after the simplifier
1199 -- has reached Phase 0. In general one would want it to be
1200 -- flag-controllable, but for now I'm leaving it baked in
1202 specConstrActivation :: Activation
1203 specConstrActivation = ActiveAfter 0 -- Baked in; see comments above
1206 Note [Transfer strictness]
1207 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1208 We must transfer strictness information from the original function to
1209 the specialised one. Suppose, for example
1212 and a RULE f (a:as) b = f_spec a as b
1214 Now we want f_spec to have strictess LLS, otherwise we'll use call-by-need
1215 when calling f_spec instead of call-by-value. And that can result in
1216 unbounded worsening in space (cf the classic foldl vs foldl')
1218 See Trac #3437 for a good example.
1220 The function calcSpecStrictness performs the calculation.
1223 %************************************************************************
1225 \subsection{Argument analysis}
1227 %************************************************************************
1229 This code deals with analysing call-site arguments to see whether
1230 they are constructor applications.
1234 type CallPat = ([Var], [CoreExpr]) -- Quantified variables and arguments
1237 callsToPats :: ScEnv -> [OneSpec] -> [ArgOcc] -> [Call] -> UniqSM (Bool, [CallPat])
1238 -- Result has no duplicate patterns,
1239 -- nor ones mentioned in done_pats
1240 -- Bool indicates that there was at least one boring pattern
1241 callsToPats env done_specs bndr_occs calls
1242 = do { mb_pats <- mapM (callToPats env bndr_occs) calls
1244 ; let good_pats :: [([Var], [CoreArg])]
1245 good_pats = catMaybes mb_pats
1246 done_pats = [p | OS p _ _ _ <- done_specs]
1247 is_done p = any (samePat p) done_pats
1249 ; return (any isNothing mb_pats,
1250 filterOut is_done (nubBy samePat good_pats)) }
1252 callToPats :: ScEnv -> [ArgOcc] -> Call -> UniqSM (Maybe CallPat)
1253 -- The [Var] is the variables to quantify over in the rule
1254 -- Type variables come first, since they may scope
1255 -- over the following term variables
1256 -- The [CoreExpr] are the argument patterns for the rule
1257 callToPats env bndr_occs (con_env, args)
1258 | length args < length bndr_occs -- Check saturated
1261 = do { let in_scope = substInScope (sc_subst env)
1262 ; prs <- argsToPats env in_scope con_env (args `zip` bndr_occs)
1263 ; let (interesting_s, pats) = unzip prs
1264 pat_fvs = varSetElems (exprsFreeVars pats)
1265 qvars = filterOut (`elemInScopeSet` in_scope) pat_fvs
1266 -- Quantify over variables that are not in sccpe
1268 -- See Note [Shadowing] at the top
1270 (tvs, ids) = partition isTyVar qvars
1272 -- Put the type variables first; the type of a term
1273 -- variable may mention a type variable
1275 ; -- pprTrace "callToPats" (ppr args $$ ppr prs $$ ppr bndr_occs) $
1277 then return (Just (qvars', pats))
1278 else return Nothing }
1280 -- argToPat takes an actual argument, and returns an abstracted
1281 -- version, consisting of just the "constructor skeleton" of the
1282 -- argument, with non-constructor sub-expression replaced by new
1283 -- placeholder variables. For example:
1284 -- C a (D (f x) (g y)) ==> C p1 (D p2 p3)
1287 -> InScopeSet -- What's in scope at the fn defn site
1288 -> ValueEnv -- ValueEnv at the call site
1289 -> CoreArg -- A call arg (or component thereof)
1291 -> UniqSM (Bool, CoreArg)
1292 -- Returns (interesting, pat),
1293 -- where pat is the pattern derived from the argument
1294 -- intersting=True if the pattern is non-trivial (not a variable or type)
1295 -- E.g. x:xs --> (True, x:xs)
1296 -- f xs --> (False, w) where w is a fresh wildcard
1297 -- (f xs, 'c') --> (True, (w, 'c')) where w is a fresh wildcard
1298 -- \x. x+y --> (True, \x. x+y)
1299 -- lvl7 --> (True, lvl7) if lvl7 is bound
1300 -- somewhere further out
1302 argToPat _env _in_scope _val_env arg@(Type {}) _arg_occ
1303 = return (False, arg)
1305 argToPat env in_scope val_env (Note _ arg) arg_occ
1306 = argToPat env in_scope val_env arg arg_occ
1307 -- Note [Notes in call patterns]
1308 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1309 -- Ignore Notes. In particular, we want to ignore any InlineMe notes
1310 -- Perhaps we should not ignore profiling notes, but I'm going to
1311 -- ride roughshod over them all for now.
1312 --- See Note [Notes in RULE matching] in Rules
1314 argToPat env in_scope val_env (Let _ arg) arg_occ
1315 = argToPat env in_scope val_env arg arg_occ
1316 -- Look through let expressions
1317 -- e.g. f (let v = rhs in \y -> ...v...)
1318 -- Here we can specialise for f (\y -> ...)
1319 -- because the rule-matcher will look through the let.
1321 argToPat env in_scope val_env (Cast arg co) arg_occ
1322 | not (ignoreType env ty2)
1323 = do { (interesting, arg') <- argToPat env in_scope val_env arg arg_occ
1324 ; if not interesting then
1327 { -- Make a wild-card pattern for the coercion
1329 ; let co_name = mkSysTvName uniq (fsLit "sg")
1330 co_var = mkCoVar co_name (mkCoKind ty1 ty2)
1331 ; return (interesting, Cast arg' (mkTyVarTy co_var)) } }
1333 (ty1, ty2) = coercionKind co
1337 {- Disabling lambda specialisation for now
1338 It's fragile, and the spec_loop can be infinite
1339 argToPat in_scope val_env arg arg_occ
1341 = return (True, arg)
1343 is_value_lam (Lam v e) -- Spot a value lambda, even if
1344 | isId v = True -- it is inside a type lambda
1345 | otherwise = is_value_lam e
1346 is_value_lam other = False
1349 -- Check for a constructor application
1350 -- NB: this *precedes* the Var case, so that we catch nullary constrs
1351 argToPat env in_scope val_env arg arg_occ
1352 | Just (ConVal dc args) <- isValue val_env arg
1353 , not (ignoreAltCon env dc)
1355 ScrutOcc _ -> True -- Used only by case scrutinee
1356 BothOcc -> case arg of -- Used elsewhere
1357 App {} -> True -- see Note [Reboxing]
1359 _other -> False -- No point; the arg is not decomposed
1360 = do { args' <- argsToPats env in_scope val_env (args `zip` conArgOccs arg_occ dc)
1361 ; return (True, mk_con_app dc (map snd args')) }
1363 -- Check if the argument is a variable that
1364 -- is in scope at the function definition site
1365 -- It's worth specialising on this if
1366 -- (a) it's used in an interesting way in the body
1367 -- (b) we know what its value is
1368 argToPat env in_scope val_env (Var v) arg_occ
1369 | case arg_occ of { UnkOcc -> False; _other -> True }, -- (a)
1371 not (ignoreType env (varType v))
1372 = return (True, Var v)
1375 | isLocalId v = v `elemInScopeSet` in_scope
1376 && isJust (lookupVarEnv val_env v)
1377 -- Local variables have values in val_env
1378 | otherwise = isValueUnfolding (idUnfolding v)
1379 -- Imports have unfoldings
1381 -- I'm really not sure what this comment means
1382 -- And by not wild-carding we tend to get forall'd
1383 -- variables that are in soope, which in turn can
1384 -- expose the weakness in let-matching
1385 -- See Note [Matching lets] in Rules
1387 -- Check for a variable bound inside the function.
1388 -- Don't make a wild-card, because we may usefully share
1389 -- e.g. f a = let x = ... in f (x,x)
1390 -- NB: this case follows the lambda and con-app cases!!
1391 -- argToPat _in_scope _val_env (Var v) _arg_occ
1392 -- = return (False, Var v)
1393 -- SLPJ : disabling this to avoid proliferation of versions
1394 -- also works badly when thinking about seeding the loop
1395 -- from the body of the let
1396 -- f x y = letrec g z = ... in g (x,y)
1397 -- We don't want to specialise for that *particular* x,y
1399 -- The default case: make a wild-card
1400 argToPat _env _in_scope _val_env arg _arg_occ
1401 = wildCardPat (exprType arg)
1403 wildCardPat :: Type -> UniqSM (Bool, CoreArg)
1404 wildCardPat ty = do { uniq <- getUniqueUs
1405 ; let id = mkSysLocal (fsLit "sc") uniq ty
1406 ; return (False, Var id) }
1408 argsToPats :: ScEnv -> InScopeSet -> ValueEnv
1409 -> [(CoreArg, ArgOcc)]
1410 -> UniqSM [(Bool, CoreArg)]
1411 argsToPats env in_scope val_env args
1414 do_one (arg,occ) = argToPat env in_scope val_env arg occ
1419 isValue :: ValueEnv -> CoreExpr -> Maybe Value
1420 isValue _env (Lit lit)
1421 = Just (ConVal (LitAlt lit) [])
1424 | Just stuff <- lookupVarEnv env v
1425 = Just stuff -- You might think we could look in the idUnfolding here
1426 -- but that doesn't take account of which branch of a
1427 -- case we are in, which is the whole point
1429 | not (isLocalId v) && isCheapUnfolding unf
1430 = isValue env (unfoldingTemplate unf)
1433 -- However we do want to consult the unfolding
1434 -- as well, for let-bound constructors!
1436 isValue env (Lam b e)
1437 | isTyVar b = case isValue env e of
1438 Just _ -> Just LambdaVal
1440 | otherwise = Just LambdaVal
1442 isValue _env expr -- Maybe it's a constructor application
1443 | (Var fun, args) <- collectArgs expr
1444 = case isDataConWorkId_maybe fun of
1446 Just con | args `lengthAtLeast` dataConRepArity con
1447 -- Check saturated; might be > because the
1448 -- arity excludes type args
1449 -> Just (ConVal (DataAlt con) args)
1451 _other | valArgCount args < idArity fun
1452 -- Under-applied function
1453 -> Just LambdaVal -- Partial application
1457 isValue _env _expr = Nothing
1459 mk_con_app :: AltCon -> [CoreArg] -> CoreExpr
1460 mk_con_app (LitAlt lit) [] = Lit lit
1461 mk_con_app (DataAlt con) args = mkConApp con args
1462 mk_con_app _other _args = panic "SpecConstr.mk_con_app"
1464 samePat :: CallPat -> CallPat -> Bool
1465 samePat (vs1, as1) (vs2, as2)
1468 same (Var v1) (Var v2)
1469 | v1 `elem` vs1 = v2 `elem` vs2
1470 | v2 `elem` vs2 = False
1471 | otherwise = v1 == v2
1473 same (Lit l1) (Lit l2) = l1==l2
1474 same (App f1 a1) (App f2 a2) = same f1 f2 && same a1 a2
1476 same (Type {}) (Type {}) = True -- Note [Ignore type differences]
1477 same (Note _ e1) e2 = same e1 e2 -- Ignore casts and notes
1478 same (Cast e1 _) e2 = same e1 e2
1479 same e1 (Note _ e2) = same e1 e2
1480 same e1 (Cast e2 _) = same e1 e2
1482 same e1 e2 = WARN( bad e1 || bad e2, ppr e1 $$ ppr e2)
1483 False -- Let, lambda, case should not occur
1484 bad (Case {}) = True
1490 Note [Ignore type differences]
1491 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1492 We do not want to generate specialisations where the call patterns
1493 differ only in their type arguments! Not only is it utterly useless,
1494 but it also means that (with polymorphic recursion) we can generate
1495 an infinite number of specialisations. Example is Data.Sequence.adjustTree,