2 % (c) The GRASP/AQUA Project, Glasgow University, 1993-1998
4 \section[Specialise]{Stamping out overloading, and (optionally) polymorphism}
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
13 module Specialise ( specProgram ) where
15 #include "HsVersions.h"
17 import DynFlags ( DynFlags, DynFlag(..) )
18 import Id ( Id, idName, idType, mkUserLocal, idCoreRules,
19 idInlinePragma, setInlinePragma, setIdUnfolding,
21 import TcType ( Type, mkTyVarTy, tcSplitSigmaTy,
22 tyVarsOfTypes, tyVarsOfTheta, isClassPred,
23 tcCmpType, isUnLiftedType
25 import CoreSubst ( Subst, mkEmptySubst, extendTvSubstList, lookupIdSubst,
26 substBndr, substBndrs, substTy, substInScope,
27 cloneIdBndr, cloneIdBndrs, cloneRecIdBndrs,
30 import CoreUnfold ( mkUnfolding )
31 import SimplUtils ( interestingArg )
37 import CoreUtils ( exprIsTrivial, applyTypeToArgs, mkPiTypes )
38 import CoreFVs ( exprFreeVars, exprsFreeVars, idFreeVars )
39 import CoreLint ( showPass, endPass )
40 import UniqSupply ( UniqSupply,
45 import MkId ( voidArgId, realWorldPrimId )
47 import Maybes ( catMaybes, isJust )
48 import ErrUtils ( dumpIfSet_dyn )
56 %************************************************************************
58 \subsection[notes-Specialise]{Implementation notes [SLPJ, Aug 18 1993]}
60 %************************************************************************
62 These notes describe how we implement specialisation to eliminate
65 The specialisation pass works on Core
66 syntax, complete with all the explicit dictionary application,
67 abstraction and construction as added by the type checker. The
68 existing type checker remains largely as it is.
70 One important thought: the {\em types} passed to an overloaded
71 function, and the {\em dictionaries} passed are mutually redundant.
72 If the same function is applied to the same type(s) then it is sure to
73 be applied to the same dictionary(s)---or rather to the same {\em
74 values}. (The arguments might look different but they will evaluate
77 Second important thought: we know that we can make progress by
78 treating dictionary arguments as static and worth specialising on. So
79 we can do without binding-time analysis, and instead specialise on
80 dictionary arguments and no others.
89 and suppose f is overloaded.
91 STEP 1: CALL-INSTANCE COLLECTION
93 We traverse <body>, accumulating all applications of f to types and
96 (Might there be partial applications, to just some of its types and
97 dictionaries? In principle yes, but in practice the type checker only
98 builds applications of f to all its types and dictionaries, so partial
99 applications could only arise as a result of transformation, and even
100 then I think it's unlikely. In any case, we simply don't accumulate such
101 partial applications.)
106 So now we have a collection of calls to f:
110 Notice that f may take several type arguments. To avoid ambiguity, we
111 say that f is called at type t1/t2 and t3/t4.
113 We take equivalence classes using equality of the *types* (ignoring
114 the dictionary args, which as mentioned previously are redundant).
116 STEP 3: SPECIALISATION
118 For each equivalence class, choose a representative (f t1 t2 d1 d2),
119 and create a local instance of f, defined thus:
121 f@t1/t2 = <f_rhs> t1 t2 d1 d2
123 f_rhs presumably has some big lambdas and dictionary lambdas, so lots
124 of simplification will now result. However we don't actually *do* that
125 simplification. Rather, we leave it for the simplifier to do. If we
126 *did* do it, though, we'd get more call instances from the specialised
127 RHS. We can work out what they are by instantiating the call-instance
128 set from f's RHS with the types t1, t2.
130 Add this new id to f's IdInfo, to record that f has a specialised version.
132 Before doing any of this, check that f's IdInfo doesn't already
133 tell us about an existing instance of f at the required type/s.
134 (This might happen if specialisation was applied more than once, or
135 it might arise from user SPECIALIZE pragmas.)
139 Wait a minute! What if f is recursive? Then we can't just plug in
140 its right-hand side, can we?
142 But it's ok. The type checker *always* creates non-recursive definitions
143 for overloaded recursive functions. For example:
145 f x = f (x+x) -- Yes I know its silly
149 f a (d::Num a) = let p = +.sel a d
151 letrec fl (y::a) = fl (p y y)
155 We still have recusion for non-overloaded functions which we
156 speciailise, but the recursive call should get specialised to the
157 same recursive version.
163 All this is crystal clear when the function is applied to *constant
164 types*; that is, types which have no type variables inside. But what if
165 it is applied to non-constant types? Suppose we find a call of f at type
166 t1/t2. There are two possibilities:
168 (a) The free type variables of t1, t2 are in scope at the definition point
169 of f. In this case there's no problem, we proceed just as before. A common
170 example is as follows. Here's the Haskell:
175 After typechecking we have
177 g a (d::Num a) (y::a) = let f b (d'::Num b) (x::b) = +.sel b d' x x
178 in +.sel a d (f a d y) (f a d y)
180 Notice that the call to f is at type type "a"; a non-constant type.
181 Both calls to f are at the same type, so we can specialise to give:
183 g a (d::Num a) (y::a) = let f@a (x::a) = +.sel a d x x
184 in +.sel a d (f@a y) (f@a y)
187 (b) The other case is when the type variables in the instance types
188 are *not* in scope at the definition point of f. The example we are
189 working with above is a good case. There are two instances of (+.sel a d),
190 but "a" is not in scope at the definition of +.sel. Can we do anything?
191 Yes, we can "common them up", a sort of limited common sub-expression deal.
194 g a (d::Num a) (y::a) = let +.sel@a = +.sel a d
195 f@a (x::a) = +.sel@a x x
196 in +.sel@a (f@a y) (f@a y)
198 This can save work, and can't be spotted by the type checker, because
199 the two instances of +.sel weren't originally at the same type.
203 * There are quite a few variations here. For example, the defn of
204 +.sel could be floated ouside the \y, to attempt to gain laziness.
205 It certainly mustn't be floated outside the \d because the d has to
208 * We don't want to inline f_rhs in this case, because
209 that will duplicate code. Just commoning up the call is the point.
211 * Nothing gets added to +.sel's IdInfo.
213 * Don't bother unless the equivalence class has more than one item!
215 Not clear whether this is all worth it. It is of course OK to
216 simply discard call-instances when passing a big lambda.
218 Polymorphism 2 -- Overloading
220 Consider a function whose most general type is
222 f :: forall a b. Ord a => [a] -> b -> b
224 There is really no point in making a version of g at Int/Int and another
225 at Int/Bool, because it's only instancing the type variable "a" which
226 buys us any efficiency. Since g is completely polymorphic in b there
227 ain't much point in making separate versions of g for the different
230 That suggests that we should identify which of g's type variables
231 are constrained (like "a") and which are unconstrained (like "b").
232 Then when taking equivalence classes in STEP 2, we ignore the type args
233 corresponding to unconstrained type variable. In STEP 3 we make
234 polymorphic versions. Thus:
236 f@t1/ = /\b -> <f_rhs> t1 b d1 d2
245 f a (d::Num a) = let g = ...
247 ...(let d1::Ord a = Num.Ord.sel a d in g a d1)...
249 Here, g is only called at one type, but the dictionary isn't in scope at the
250 definition point for g. Usually the type checker would build a
251 definition for d1 which enclosed g, but the transformation system
252 might have moved d1's defn inward. Solution: float dictionary bindings
253 outwards along with call instances.
257 f x = let g p q = p==q
263 Before specialisation, leaving out type abstractions we have
265 f df x = let g :: Eq a => a -> a -> Bool
267 h :: Num a => a -> a -> (a, Bool)
268 h dh r s = let deq = eqFromNum dh
269 in (+ dh r s, g deq r s)
273 After specialising h we get a specialised version of h, like this:
275 h' r s = let deq = eqFromNum df
276 in (+ df r s, g deq r s)
278 But we can't naively make an instance for g from this, because deq is not in scope
279 at the defn of g. Instead, we have to float out the (new) defn of deq
280 to widen its scope. Notice that this floating can't be done in advance -- it only
281 shows up when specialisation is done.
283 User SPECIALIZE pragmas
284 ~~~~~~~~~~~~~~~~~~~~~~~
285 Specialisation pragmas can be digested by the type checker, and implemented
286 by adding extra definitions along with that of f, in the same way as before
288 f@t1/t2 = <f_rhs> t1 t2 d1 d2
290 Indeed the pragmas *have* to be dealt with by the type checker, because
291 only it knows how to build the dictionaries d1 and d2! For example
293 g :: Ord a => [a] -> [a]
294 {-# SPECIALIZE f :: [Tree Int] -> [Tree Int] #-}
296 Here, the specialised version of g is an application of g's rhs to the
297 Ord dictionary for (Tree Int), which only the type checker can conjure
298 up. There might not even *be* one, if (Tree Int) is not an instance of
299 Ord! (All the other specialision has suitable dictionaries to hand
302 Problem. The type checker doesn't have to hand a convenient <f_rhs>, because
303 it is buried in a complex (as-yet-un-desugared) binding group.
306 f@t1/t2 = f* t1 t2 d1 d2
308 where f* is the Id f with an IdInfo which says "inline me regardless!".
309 Indeed all the specialisation could be done in this way.
310 That in turn means that the simplifier has to be prepared to inline absolutely
311 any in-scope let-bound thing.
314 Again, the pragma should permit polymorphism in unconstrained variables:
316 h :: Ord a => [a] -> b -> b
317 {-# SPECIALIZE h :: [Int] -> b -> b #-}
319 We *insist* that all overloaded type variables are specialised to ground types,
320 (and hence there can be no context inside a SPECIALIZE pragma).
321 We *permit* unconstrained type variables to be specialised to
323 - or left as a polymorphic type variable
324 but nothing in between. So
326 {-# SPECIALIZE h :: [Int] -> [c] -> [c] #-}
328 is *illegal*. (It can be handled, but it adds complication, and gains the
332 SPECIALISING INSTANCE DECLARATIONS
333 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
336 instance Foo a => Foo [a] where
338 {-# SPECIALIZE instance Foo [Int] #-}
340 The original instance decl creates a dictionary-function
343 dfun.Foo.List :: forall a. Foo a -> Foo [a]
345 The SPECIALIZE pragma just makes a specialised copy, just as for
346 ordinary function definitions:
348 dfun.Foo.List@Int :: Foo [Int]
349 dfun.Foo.List@Int = dfun.Foo.List Int dFooInt
351 The information about what instance of the dfun exist gets added to
352 the dfun's IdInfo in the same way as a user-defined function too.
355 Automatic instance decl specialisation?
356 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
357 Can instance decls be specialised automatically? It's tricky.
358 We could collect call-instance information for each dfun, but
359 then when we specialised their bodies we'd get new call-instances
360 for ordinary functions; and when we specialised their bodies, we might get
361 new call-instances of the dfuns, and so on. This all arises because of
362 the unrestricted mutual recursion between instance decls and value decls.
364 Still, there's no actual problem; it just means that we may not do all
365 the specialisation we could theoretically do.
367 Furthermore, instance decls are usually exported and used non-locally,
368 so we'll want to compile enough to get those specialisations done.
370 Lastly, there's no such thing as a local instance decl, so we can
371 survive solely by spitting out *usage* information, and then reading that
372 back in as a pragma when next compiling the file. So for now,
373 we only specialise instance decls in response to pragmas.
376 SPITTING OUT USAGE INFORMATION
377 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
379 To spit out usage information we need to traverse the code collecting
380 call-instance information for all imported (non-prelude?) functions
381 and data types. Then we equivalence-class it and spit it out.
383 This is done at the top-level when all the call instances which escape
384 must be for imported functions and data types.
386 *** Not currently done ***
389 Partial specialisation by pragmas
390 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
391 What about partial specialisation:
393 k :: (Ord a, Eq b) => [a] -> b -> b -> [a]
394 {-# SPECIALIZE k :: Eq b => [Int] -> b -> b -> [a] #-}
398 {-# SPECIALIZE k :: Eq b => [Int] -> [b] -> [b] -> [a] #-}
400 Seems quite reasonable. Similar things could be done with instance decls:
402 instance (Foo a, Foo b) => Foo (a,b) where
404 {-# SPECIALIZE instance Foo a => Foo (a,Int) #-}
405 {-# SPECIALIZE instance Foo b => Foo (Int,b) #-}
407 Ho hum. Things are complex enough without this. I pass.
410 Requirements for the simplifer
411 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
412 The simplifier has to be able to take advantage of the specialisation.
414 * When the simplifier finds an application of a polymorphic f, it looks in
415 f's IdInfo in case there is a suitable instance to call instead. This converts
417 f t1 t2 d1 d2 ===> f_t1_t2
419 Note that the dictionaries get eaten up too!
421 * Dictionary selection operations on constant dictionaries must be
424 +.sel Int d ===> +Int
426 The obvious way to do this is in the same way as other specialised
427 calls: +.sel has inside it some IdInfo which tells that if it's applied
428 to the type Int then it should eat a dictionary and transform to +Int.
430 In short, dictionary selectors need IdInfo inside them for constant
433 * Exactly the same applies if a superclass dictionary is being
436 Eq.sel Int d ===> dEqInt
438 * Something similar applies to dictionary construction too. Suppose
439 dfun.Eq.List is the function taking a dictionary for (Eq a) to
440 one for (Eq [a]). Then we want
442 dfun.Eq.List Int d ===> dEq.List_Int
444 Where does the Eq [Int] dictionary come from? It is built in
445 response to a SPECIALIZE pragma on the Eq [a] instance decl.
447 In short, dfun Ids need IdInfo with a specialisation for each
448 constant instance of their instance declaration.
450 All this uses a single mechanism: the SpecEnv inside an Id
453 What does the specialisation IdInfo look like?
454 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
456 The SpecEnv of an Id maps a list of types (the template) to an expression
460 For example, if f has this SpecInfo:
462 [Int, a] -> \d:Ord Int. f' a
464 it means that we can replace the call
466 f Int t ===> (\d. f' t)
468 This chucks one dictionary away and proceeds with the
469 specialised version of f, namely f'.
472 What can't be done this way?
473 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
474 There is no way, post-typechecker, to get a dictionary for (say)
475 Eq a from a dictionary for Eq [a]. So if we find
479 we can't transform to
484 eqList :: (a->a->Bool) -> [a] -> [a] -> Bool
486 Of course, we currently have no way to automatically derive
487 eqList, nor to connect it to the Eq [a] instance decl, but you
488 can imagine that it might somehow be possible. Taking advantage
489 of this is permanently ruled out.
491 Still, this is no great hardship, because we intend to eliminate
492 overloading altogether anyway!
494 A note about non-tyvar dictionaries
495 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
496 Some Ids have types like
498 forall a,b,c. Eq a -> Ord [a] -> tau
500 This seems curious at first, because we usually only have dictionary
501 args whose types are of the form (C a) where a is a type variable.
502 But this doesn't hold for the functions arising from instance decls,
503 which sometimes get arguements with types of form (C (T a)) for some
506 Should we specialise wrt this compound-type dictionary? We used to say
508 "This is a heuristic judgement, as indeed is the fact that we
509 specialise wrt only dictionaries. We choose *not* to specialise
510 wrt compound dictionaries because at the moment the only place
511 they show up is in instance decls, where they are simply plugged
512 into a returned dictionary. So nothing is gained by specialising
515 But it is simpler and more uniform to specialise wrt these dicts too;
516 and in future GHC is likely to support full fledged type signatures
518 f :: Eq [(a,b)] => ...
521 %************************************************************************
523 \subsubsection{The new specialiser}
525 %************************************************************************
527 Our basic game plan is this. For let(rec) bound function
528 f :: (C a, D c) => (a,b,c,d) -> Bool
530 * Find any specialised calls of f, (f ts ds), where
531 ts are the type arguments t1 .. t4, and
532 ds are the dictionary arguments d1 .. d2.
534 * Add a new definition for f1 (say):
536 f1 = /\ b d -> (..body of f..) t1 b t3 d d1 d2
538 Note that we abstract over the unconstrained type arguments.
542 [t1,b,t3,d] |-> \d1 d2 -> f1 b d
544 to the specialisations of f. This will be used by the
545 simplifier to replace calls
546 (f t1 t2 t3 t4) da db
548 (\d1 d1 -> f1 t2 t4) da db
550 All the stuff about how many dictionaries to discard, and what types
551 to apply the specialised function to, are handled by the fact that the
552 SpecEnv contains a template for the result of the specialisation.
554 We don't build *partial* specialisations for f. For example:
556 f :: Eq a => a -> a -> Bool
557 {-# SPECIALISE f :: (Eq b, Eq c) => (b,c) -> (b,c) -> Bool #-}
559 Here, little is gained by making a specialised copy of f.
560 There's a distinct danger that the specialised version would
561 first build a dictionary for (Eq b, Eq c), and then select the (==)
562 method from it! Even if it didn't, not a great deal is saved.
564 We do, however, generate polymorphic, but not overloaded, specialisations:
566 f :: Eq a => [a] -> b -> b -> b
567 {#- SPECIALISE f :: [Int] -> b -> b -> b #-}
569 Hence, the invariant is this:
571 *** no specialised version is overloaded ***
574 %************************************************************************
576 \subsubsection{The exported function}
578 %************************************************************************
581 specProgram :: DynFlags -> UniqSupply -> [CoreBind] -> IO [CoreBind]
582 specProgram dflags us binds = do
584 showPass dflags "Specialise"
586 let binds' = initSM us (do (binds', uds') <- go binds
587 return (dumpAllDictBinds uds' binds'))
589 endPass dflags "Specialise" Opt_D_dump_spec binds'
591 dumpIfSet_dyn dflags Opt_D_dump_rules "Top-level specialisations"
592 (pprRulesForUser (rulesOfBinds binds'))
596 -- We need to start with a Subst that knows all the things
597 -- that are in scope, so that the substitution engine doesn't
598 -- accidentally re-use a unique that's already in use
599 -- Easiest thing is to do it all at once, as if all the top-level
600 -- decls were mutually recursive
601 top_subst = mkEmptySubst (mkInScopeSet (mkVarSet (bindersOfBinds binds)))
603 go [] = return ([], emptyUDs)
604 go (bind:binds) = do (binds', uds) <- go binds
605 (bind', uds') <- specBind top_subst bind uds
606 return (bind' ++ binds', uds')
609 %************************************************************************
611 \subsubsection{@specExpr@: the main function}
613 %************************************************************************
616 specVar :: Subst -> Id -> CoreExpr
617 specVar subst v = lookupIdSubst subst v
619 specExpr :: Subst -> CoreExpr -> SpecM (CoreExpr, UsageDetails)
620 -- We carry a substitution down:
621 -- a) we must clone any binding that might float outwards,
622 -- to avoid name clashes
623 -- b) we carry a type substitution to use when analysing
624 -- the RHS of specialised bindings (no type-let!)
626 ---------------- First the easy cases --------------------
627 specExpr subst (Type ty) = return (Type (substTy subst ty), emptyUDs)
628 specExpr subst (Var v) = return (specVar subst v, emptyUDs)
629 specExpr _ (Lit lit) = return (Lit lit, emptyUDs)
630 specExpr subst (Cast e co) = do
631 (e', uds) <- specExpr subst e
632 return ((Cast e' (substTy subst co)), uds)
633 specExpr subst (Note note body) = do
634 (body', uds) <- specExpr subst body
635 return (Note (specNote subst note) body', uds)
638 ---------------- Applications might generate a call instance --------------------
639 specExpr subst expr@(App {})
642 go (App fun arg) args = do (arg', uds_arg) <- specExpr subst arg
643 (fun', uds_app) <- go fun (arg':args)
644 return (App fun' arg', uds_arg `plusUDs` uds_app)
646 go (Var f) args = case specVar subst f of
647 Var f' -> return (Var f', mkCallUDs f' args)
648 e' -> return (e', emptyUDs) -- I don't expect this!
649 go other _ = specExpr subst other
651 ---------------- Lambda/case require dumping of usage details --------------------
652 specExpr subst e@(Lam _ _) = do
653 (body', uds) <- specExpr subst' body
654 let (filtered_uds, body'') = dumpUDs bndrs' uds body'
655 return (mkLams bndrs' body'', filtered_uds)
657 (bndrs, body) = collectBinders e
658 (subst', bndrs') = substBndrs subst bndrs
659 -- More efficient to collect a group of binders together all at once
660 -- and we don't want to split a lambda group with dumped bindings
662 specExpr subst (Case scrut case_bndr ty alts) = do
663 (scrut', uds_scrut) <- specExpr subst scrut
664 (alts', uds_alts) <- mapAndCombineSM spec_alt alts
665 return (Case scrut' case_bndr' (substTy subst ty) alts', uds_scrut `plusUDs` uds_alts)
667 (subst_alt, case_bndr') = substBndr subst case_bndr
668 -- No need to clone case binder; it can't float like a let(rec)
670 spec_alt (con, args, rhs) = do
671 (rhs', uds) <- specExpr subst_rhs rhs
672 let (uds', rhs'') = dumpUDs args uds rhs'
673 return ((con, args', rhs''), uds')
675 (subst_rhs, args') = substBndrs subst_alt args
677 ---------------- Finally, let is the interesting case --------------------
678 specExpr subst (Let bind body) = do
680 (rhs_subst, body_subst, bind') <- cloneBindSM subst bind
682 -- Deal with the body
683 (body', body_uds) <- specExpr body_subst body
685 -- Deal with the bindings
686 (binds', uds) <- specBind rhs_subst bind' body_uds
689 return (foldr Let body' binds', uds)
691 -- Must apply the type substitution to coerceions
692 specNote :: Subst -> Note -> Note
693 specNote _ note = note
696 %************************************************************************
698 \subsubsection{Dealing with a binding}
700 %************************************************************************
703 specBind :: Subst -- Use this for RHSs
705 -> UsageDetails -- Info on how the scope of the binding
706 -> SpecM ([CoreBind], -- New bindings
707 UsageDetails) -- And info to pass upstream
709 specBind rhs_subst bind body_uds
710 = do { (bind', bind_uds) <- specBindItself rhs_subst bind (calls body_uds)
711 ; return (finishSpecBind bind' bind_uds body_uds) }
713 finishSpecBind :: CoreBind -> UsageDetails -> UsageDetails -> ([CoreBind], UsageDetails)
715 (MkUD { dict_binds = rhs_dbs, calls = rhs_calls, ud_fvs = rhs_fvs })
716 (MkUD { dict_binds = body_dbs, calls = body_calls, ud_fvs = body_fvs })
717 | not (mkVarSet bndrs `intersectsVarSet` all_fvs)
718 -- Common case 1: the bound variables are not
719 -- mentioned in the dictionary bindings
720 = ([bind], MkUD { dict_binds = body_dbs `unionBags` rhs_dbs
721 -- It's important that the `unionBags` is this way round,
722 -- because body_uds may bind dictionaries that are
723 -- used in the calls passed to specDefn. So the
724 -- dictionary bindings in rhs_uds may mention
725 -- dictionaries bound in body_uds.
727 , ud_fvs = all_fvs })
729 | case bind of { NonRec {} -> True; Rec {} -> False }
730 -- Common case 2: no specialisation happened, and binding
731 -- is non-recursive. But the binding may be
732 -- mentioned in body_dbs, so we should put it first
733 = ([], MkUD { dict_binds = rhs_dbs `unionBags` ((bind, b_fvs) `consBag` body_dbs)
735 , ud_fvs = all_fvs `unionVarSet` b_fvs })
737 | otherwise -- General case: make a huge Rec (sigh)
738 = ([], MkUD { dict_binds = unitBag (Rec all_db_prs, all_db_fvs)
740 , ud_fvs = all_fvs `unionVarSet` b_fvs })
742 all_fvs = rhs_fvs `unionVarSet` body_fvs
743 all_calls = zapCalls bndrs (rhs_calls `unionCalls` body_calls)
745 bndrs = bindersOf bind
746 b_fvs = bind_fvs bind
748 (all_db_prs, all_db_fvs) = add (bind, b_fvs) $
749 foldrBag add ([], emptyVarSet) $
750 rhs_dbs `unionBags` body_dbs
751 add (NonRec b r, b_fvs) (prs, fvs) = ((b,r) : prs, b_fvs `unionVarSet` fvs)
752 add (Rec b_prs, b_fvs) (prs, fvs) = (b_prs ++ prs, b_fvs `unionVarSet` fvs)
754 ---------------------------
755 specBindItself :: Subst -> CoreBind -> CallDetails -> SpecM (CoreBind, UsageDetails)
757 -- specBindItself deals with the RHS, specialising it according
758 -- to the calls found in the body (if any)
759 specBindItself rhs_subst (NonRec fn rhs) call_info
760 = do { (rhs', rhs_uds) <- specExpr rhs_subst rhs -- Do RHS of original fn
761 ; (fn', spec_defns, spec_uds) <- specDefn rhs_subst call_info fn rhs
762 ; if null spec_defns then
763 return (NonRec fn rhs', rhs_uds)
765 return (Rec ((fn',rhs') : spec_defns), rhs_uds `plusUDs` spec_uds) }
766 -- bndr' mentions the spec_defns in its SpecEnv
767 -- Not sure why we couln't just put the spec_defns first
769 specBindItself rhs_subst (Rec pairs) call_info
770 -- Note [Specialising a recursive group]
771 = do { let (bndrs,rhss) = unzip pairs
772 ; (rhss', rhs_uds) <- mapAndCombineSM (specExpr rhs_subst) rhss
773 ; let all_calls = call_info `unionCalls` calls rhs_uds
774 ; (bndrs1, spec_defns1, spec_uds1) <- specDefns rhs_subst all_calls pairs
776 ; if null spec_defns1 then -- Common case: no specialisation
777 return (Rec (bndrs `zip` rhss'), rhs_uds)
778 else do -- Specialisation occurred; do it again
779 { (bndrs2, spec_defns2, spec_uds2) <-
780 -- pprTrace "specB" (ppr bndrs $$ ppr rhs_uds) $
781 specDefns rhs_subst (calls spec_uds1) (bndrs1 `zip` rhss)
783 ; let all_defns = spec_defns1 ++ spec_defns2 ++ zip bndrs2 rhss'
785 ; return (Rec all_defns, rhs_uds `plusUDs` spec_uds1 `plusUDs` spec_uds2) } }
788 ---------------------------
790 -> CallDetails -- Info on how it is used in its scope
791 -> [(Id,CoreExpr)] -- The things being bound and their un-processed RHS
792 -> SpecM ([Id], -- Original Ids with RULES added
793 [(Id,CoreExpr)], -- Extra, specialised bindings
794 UsageDetails) -- Stuff to fling upwards from the specialised versions
796 -- Specialise a list of bindings (the contents of a Rec), but flowing usages
797 -- upwards binding by binding. Example: { f = ...g ...; g = ...f .... }
798 -- Then if the input CallDetails has a specialised call for 'g', whose specialisation
799 -- in turn generates a specialised call for 'f', we catch that in this one sweep.
800 -- But not vice versa (it's a fixpoint problem).
802 specDefns _subst _call_info []
803 = return ([], [], emptyUDs)
804 specDefns subst call_info ((bndr,rhs):pairs)
805 = do { (bndrs', spec_defns, spec_uds) <- specDefns subst call_info pairs
806 ; let all_calls = call_info `unionCalls` calls spec_uds
807 ; (bndr', spec_defns1, spec_uds1) <- specDefn subst all_calls bndr rhs
808 ; return (bndr' : bndrs',
809 spec_defns1 ++ spec_defns,
810 spec_uds1 `plusUDs` spec_uds) }
812 ---------------------------
814 -> CallDetails -- Info on how it is used in its scope
815 -> Id -> CoreExpr -- The thing being bound and its un-processed RHS
816 -> SpecM (Id, -- Original Id with added RULES
817 [(Id,CoreExpr)], -- Extra, specialised bindings
818 UsageDetails) -- Stuff to fling upwards from the specialised versions
820 specDefn subst calls fn rhs
821 -- The first case is the interesting one
822 | rhs_tyvars `lengthIs` n_tyvars -- Rhs of fn's defn has right number of big lambdas
823 && rhs_ids `lengthAtLeast` n_dicts -- and enough dict args
824 && notNull calls_for_me -- And there are some calls to specialise
826 -- && not (certainlyWillInline (idUnfolding fn)) -- And it's not small
827 -- See Note [Inline specialisation] for why we do not
828 -- switch off specialisation for inline functions
830 = do { -- Make a specialised version for each call in calls_for_me
831 stuff <- mapM spec_call calls_for_me
832 ; let (spec_defns, spec_uds, spec_rules) = unzip3 (catMaybes stuff)
833 fn' = addIdSpecialisations fn spec_rules
834 ; return (fn', spec_defns, plusUDList spec_uds) }
836 | otherwise -- No calls or RHS doesn't fit our preconceptions
837 = WARN( notNull calls_for_me, ptext (sLit "Missed specialisation opportunity for") <+> ppr fn )
838 -- Note [Specialisation shape]
839 return (fn, [], emptyUDs)
843 (tyvars, theta, _) = tcSplitSigmaTy fn_type
844 n_tyvars = length tyvars
845 n_dicts = length theta
846 inline_prag = idInlinePragma fn
848 -- It's important that we "see past" any INLINE pragma
849 -- else we'll fail to specialise an INLINE thing
850 (inline_rhs, rhs_inside) = dropInline rhs
851 (rhs_tyvars, rhs_ids, rhs_body) = collectTyAndValBinders rhs_inside
853 rhs_dict_ids = take n_dicts rhs_ids
854 body = mkLams (drop n_dicts rhs_ids) rhs_body
855 -- Glue back on the non-dict lambdas
857 calls_for_me = case lookupFM calls fn of
859 Just cs -> fmToList cs
861 already_covered :: [CoreExpr] -> Bool
862 already_covered args -- Note [Specialisations already covered]
863 = isJust (lookupRule (const True) (substInScope subst)
864 fn args (idCoreRules fn))
866 mk_ty_args :: [Maybe Type] -> [CoreExpr]
867 mk_ty_args call_ts = zipWithEqual "spec_call" mk_ty_arg rhs_tyvars call_ts
869 mk_ty_arg rhs_tyvar Nothing = Type (mkTyVarTy rhs_tyvar)
870 mk_ty_arg _ (Just ty) = Type ty
872 ----------------------------------------------------------
873 -- Specialise to one particular call pattern
874 spec_call :: (CallKey, ([DictExpr], VarSet)) -- Call instance
875 -> SpecM (Maybe ((Id,CoreExpr), -- Specialised definition
876 UsageDetails, -- Usage details from specialised body
877 CoreRule)) -- Info for the Id's SpecEnv
878 spec_call (CallKey call_ts, (call_ds, _))
879 = ASSERT( call_ts `lengthIs` n_tyvars && call_ds `lengthIs` n_dicts )
881 -- Suppose f's defn is f = /\ a b c -> \ d1 d2 -> rhs
882 -- Supppose the call is for f [Just t1, Nothing, Just t3] [dx1, dx2]
884 -- Construct the new binding
885 -- f1 = SUBST[a->t1,c->t3, d1->d1', d2->d2'] (/\ b -> rhs)
886 -- PLUS the usage-details
887 -- { d1' = dx1; d2' = dx2 }
888 -- where d1', d2' are cloned versions of d1,d2, with the type substitution
889 -- applied. These auxiliary bindings just avoid duplication of dx1, dx2
891 -- Note that the substitution is applied to the whole thing.
892 -- This is convenient, but just slightly fragile. Notably:
893 -- * There had better be no name clashes in a/b/c
895 -- poly_tyvars = [b] in the example above
896 -- spec_tyvars = [a,c]
897 -- ty_args = [t1,b,t3]
898 poly_tyvars = [tv | (tv, Nothing) <- rhs_tyvars `zip` call_ts]
899 spec_tv_binds = [(tv,ty) | (tv, Just ty) <- rhs_tyvars `zip` call_ts]
900 spec_ty_args = map snd spec_tv_binds
901 ty_args = mk_ty_args call_ts
902 rhs_subst = extendTvSubstList subst spec_tv_binds
904 ; (rhs_subst1, inst_dict_ids) <- cloneDictBndrs rhs_subst rhs_dict_ids
905 -- Clone rhs_dicts, including instantiating their types
907 ; let (rhs_subst2, dx_binds) = bindAuxiliaryDicts rhs_subst1 $
908 (my_zipEqual rhs_dict_ids inst_dict_ids call_ds)
909 inst_args = ty_args ++ map Var inst_dict_ids
911 ; if already_covered inst_args then
914 { -- Figure out the type of the specialised function
915 let body_ty = applyTypeToArgs rhs fn_type inst_args
916 (lam_args, app_args) -- Add a dummy argument if body_ty is unlifted
917 | isUnLiftedType body_ty -- C.f. WwLib.mkWorkerArgs
918 = (poly_tyvars ++ [voidArgId], poly_tyvars ++ [realWorldPrimId])
919 | otherwise = (poly_tyvars, poly_tyvars)
920 spec_id_ty = mkPiTypes lam_args body_ty
922 ; spec_f <- newSpecIdSM fn spec_id_ty
923 ; (spec_rhs, rhs_uds) <- specExpr rhs_subst2 (mkLams lam_args body)
925 -- The rule to put in the function's specialisation is:
926 -- forall b, d1',d2'. f t1 b t3 d1' d2' = f1 b
927 rule_name = mkFastString ("SPEC " ++ showSDoc (ppr fn <+> ppr spec_ty_args))
928 spec_env_rule = mkLocalRule
930 inline_prag -- Note [Auto-specialisation and RULES]
932 (poly_tyvars ++ inst_dict_ids)
934 (mkVarApps (Var spec_f) app_args)
936 -- Add the { d1' = dx1; d2' = dx2 } usage stuff
937 final_uds = foldr addDictBind rhs_uds dx_binds
939 spec_pr | inline_rhs = (spec_f `setInlinePragma` inline_prag, Note InlineMe spec_rhs)
940 | otherwise = (spec_f, spec_rhs)
942 ; return (Just (spec_pr, final_uds, spec_env_rule)) } }
945 | debugIsOn && not (equalLength xs ys && equalLength ys zs)
946 = pprPanic "my_zipEqual" (vcat [ ppr xs, ppr ys
947 , ppr fn <+> ppr call_ts
948 , ppr (idType fn), ppr theta
949 , ppr n_dicts, ppr rhs_dict_ids
951 | otherwise = zip3 xs ys zs
955 -> [(DictId,DictId,CoreExpr)] -- (orig_dict, inst_dict, dx)
956 -> (Subst, -- Substitute for all orig_dicts
957 [(DictId, CoreExpr)]) -- Auxiliary bindings
958 -- Bind any dictionary arguments to fresh names, to preserve sharing
959 -- Substitution already substitutes orig_dict -> inst_dict
960 bindAuxiliaryDicts subst triples = go subst [] triples
962 go subst binds [] = (subst, binds)
963 go subst binds ((d, dx_id, dx) : pairs)
964 | exprIsTrivial dx = go (extendIdSubst subst d dx) binds pairs
965 -- No auxiliary binding necessary
966 | otherwise = go subst_w_unf ((dx_id,dx) : binds) pairs
968 dx_id1 = dx_id `setIdUnfolding` mkUnfolding False dx
969 subst_w_unf = extendIdSubst subst d (Var dx_id1)
970 -- Important! We're going to substitute dx_id1 for d
971 -- and we want it to look "interesting", else we won't gather *any*
972 -- consequential calls. E.g.
974 -- If we specialise f for a call (f (dfun dNumInt)), we'll get
975 -- a consequent call (g d') with an auxiliary definition
977 -- We want that consequent call to look interesting
980 Note [Specialising a recursive group]
981 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
983 let rec { f x = ...g x'...
984 ; g y = ...f y'.... }
986 Here we specialise 'f' at Char; but that is very likely to lead to
987 a specialisation of 'g' at Char. We must do the latter, else the
988 whole point of specialisation is lost.
990 But we do not want to keep iterating to a fixpoint, because in the
991 presence of polymorphic recursion we might generate an infinite number
994 So we use the following heuristic:
995 * Arrange the rec block in dependency order, so far as possible
996 (the occurrence analyser already does this)
998 * Specialise it much like a sequence of lets
1000 * Then go through the block a second time, feeding call-info from
1001 the RHSs back in the bottom, as it were
1003 In effect, the ordering maxmimises the effectiveness of each sweep,
1004 and we do just two sweeps. This should catch almost every case of
1005 monomorphic recursion -- the exception could be a very knotted-up
1006 recursion with multiple cycles tied up together.
1008 This plan is implemented in the Rec case of specBindItself.
1010 Note [Specialisations already covered]
1011 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1012 We obviously don't want to generate two specialisations for the same
1013 argument pattern. There are two wrinkles
1015 1. We do the already-covered test in specDefn, not when we generate
1016 the CallInfo in mkCallUDs. We used to test in the latter place, but
1017 we now iterate the specialiser somewhat, and the Id at the call site
1018 might therefore not have all the RULES that we can see in specDefn
1020 2. What about two specialisations where the second is an *instance*
1021 of the first? If the more specific one shows up first, we'll generate
1022 specialisations for both. If the *less* specific one shows up first,
1023 we *don't* currently generate a specialisation for the more specific
1024 one. (See the call to lookupRule in already_covered.) Reasons:
1025 (a) lookupRule doesn't say which matches are exact (bad reason)
1026 (b) if the earlier specialisation is user-provided, it's
1027 far from clear that we should auto-specialise further
1029 Note [Auto-specialisation and RULES]
1030 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1032 g :: Num a => a -> a
1035 f :: (Int -> Int) -> Int
1037 {-# RULE f g = 0 #-}
1039 Suppose that auto-specialisation makes a specialised version of
1040 g::Int->Int That version won't appear in the LHS of the RULE for f.
1041 So if the specialisation rule fires too early, the rule for f may
1044 It might be possible to add new rules, to "complete" the rewrite system.
1046 RULE forall d. g Int d = g_spec
1050 But that's a bit complicated. For now we ask the programmer's help,
1051 by *copying the INLINE activation pragma* to the auto-specialised rule.
1052 So if g says {-# NOINLINE[2] g #-}, then the auto-spec rule will also
1053 not be active until phase 2.
1056 Note [Specialisation shape]
1057 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
1058 We only specialise a function if it has visible top-level lambdas
1059 corresponding to its overloading. E.g. if
1060 f :: forall a. Eq a => ....
1061 then its body must look like
1064 Reason: when specialising the body for a call (f ty dexp), we want to
1065 substitute dexp for d, and pick up specialised calls in the body of f.
1067 This doesn't always work. One example I came across was this:
1068 newtype Gen a = MkGen{ unGen :: Int -> a }
1070 choose :: Eq a => a -> Gen a
1071 choose n = MkGen (\r -> n)
1073 oneof = choose (1::Int)
1075 It's a silly exapmle, but we get
1076 choose = /\a. g `cast` co
1077 where choose doesn't have any dict arguments. Thus far I have not
1078 tried to fix this (wait till there's a real example).
1081 Note [Inline specialisations]
1082 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1083 We transfer to the specialised function any INLINE stuff from the
1084 original. This means (a) the Activation in the IdInfo, and (b) any
1085 InlineMe on the RHS.
1087 This is a change (Jun06). Previously the idea is that the point of
1088 inlining was precisely to specialise the function at its call site,
1089 and that's not so important for the specialised copies. But
1090 *pragma-directed* specialisation now takes place in the
1091 typechecker/desugarer, with manually specified INLINEs. The
1092 specialiation here is automatic. It'd be very odd if a function
1093 marked INLINE was specialised (because of some local use), and then
1094 forever after (including importing modules) the specialised version
1095 wasn't INLINEd. After all, the programmer said INLINE!
1097 You might wonder why we don't just not specialise INLINE functions.
1098 It's because even INLINE functions are sometimes not inlined, when
1099 they aren't applied to interesting arguments. But perhaps the type
1100 arguments alone are enough to specialise (even though the args are too
1101 boring to trigger inlining), and it's certainly better to call the
1102 specialised version.
1104 A case in point is dictionary functions, which are current marked
1105 INLINE, but which are worth specialising.
1108 dropInline :: CoreExpr -> (Bool, CoreExpr)
1109 dropInline (Note InlineMe rhs) = (True, rhs)
1110 dropInline rhs = (False, rhs)
1113 %************************************************************************
1115 \subsubsection{UsageDetails and suchlike}
1117 %************************************************************************
1122 dict_binds :: !(Bag DictBind),
1123 -- Floated dictionary bindings
1124 -- The order is important;
1125 -- in ds1 `union` ds2, bindings in ds2 can depend on those in ds1
1126 -- (Remember, Bags preserve order in GHC.)
1128 calls :: !CallDetails,
1130 ud_fvs :: !VarSet -- A superset of the variables mentioned in
1131 -- either dict_binds or calls
1134 instance Outputable UsageDetails where
1135 ppr (MkUD { dict_binds = dbs, calls = calls, ud_fvs = fvs })
1136 = ptext (sLit "MkUD") <+> braces (sep (punctuate comma
1137 [ptext (sLit "binds") <+> equals <+> ppr dbs,
1138 ptext (sLit "calls") <+> equals <+> ppr calls,
1139 ptext (sLit "fvs") <+> equals <+> ppr fvs]))
1141 type DictBind = (CoreBind, VarSet)
1142 -- The set is the free vars of the binding
1143 -- both tyvars and dicts
1145 type DictExpr = CoreExpr
1147 emptyUDs :: UsageDetails
1148 emptyUDs = MkUD { dict_binds = emptyBag, calls = emptyFM, ud_fvs = emptyVarSet }
1150 ------------------------------------------------------------
1151 type CallDetails = FiniteMap Id CallInfo
1152 newtype CallKey = CallKey [Maybe Type] -- Nothing => unconstrained type argument
1154 -- CallInfo uses a FiniteMap, thereby ensuring that
1155 -- we record only one call instance for any key
1157 -- The list of types and dictionaries is guaranteed to
1158 -- match the type of f
1159 type CallInfo = FiniteMap CallKey ([DictExpr], VarSet)
1160 -- Range is dict args and the vars of the whole
1161 -- call (including tyvars)
1162 -- [*not* include the main id itself, of course]
1164 instance Outputable CallKey where
1165 ppr (CallKey ts) = ppr ts
1167 -- Type isn't an instance of Ord, so that we can control which
1168 -- instance we use. That's tiresome here. Oh well
1169 instance Eq CallKey where
1170 k1 == k2 = case k1 `compare` k2 of { EQ -> True; _ -> False }
1172 instance Ord CallKey where
1173 compare (CallKey k1) (CallKey k2) = cmpList cmp k1 k2
1175 cmp Nothing Nothing = EQ
1176 cmp Nothing (Just _) = LT
1177 cmp (Just _) Nothing = GT
1178 cmp (Just t1) (Just t2) = tcCmpType t1 t2
1180 unionCalls :: CallDetails -> CallDetails -> CallDetails
1181 unionCalls c1 c2 = plusFM_C plusFM c1 c2
1183 singleCall :: Id -> [Maybe Type] -> [DictExpr] -> UsageDetails
1184 singleCall id tys dicts
1185 = MkUD {dict_binds = emptyBag,
1186 calls = unitFM id (unitFM (CallKey tys) (dicts, call_fvs)),
1189 call_fvs = exprsFreeVars dicts `unionVarSet` tys_fvs
1190 tys_fvs = tyVarsOfTypes (catMaybes tys)
1191 -- The type args (tys) are guaranteed to be part of the dictionary
1192 -- types, because they are just the constrained types,
1193 -- and the dictionary is therefore sure to be bound
1194 -- inside the binding for any type variables free in the type;
1195 -- hence it's safe to neglect tyvars free in tys when making
1196 -- the free-var set for this call
1197 -- BUT I don't trust this reasoning; play safe and include tys_fvs
1199 -- We don't include the 'id' itself.
1201 mkCallUDs :: Id -> [CoreExpr] -> UsageDetails
1203 | not (isLocalId f) -- Imported from elsewhere
1204 || null theta -- Not overloaded
1205 || not (all isClassPred theta)
1206 -- Only specialise if all overloading is on class params.
1207 -- In ptic, with implicit params, the type args
1208 -- *don't* say what the value of the implicit param is!
1209 || not (spec_tys `lengthIs` n_tyvars)
1210 || not ( dicts `lengthIs` n_dicts)
1211 || not (any interestingArg dicts) -- Note [Interesting dictionary arguments]
1212 -- See also Note [Specialisations already covered]
1213 = -- pprTrace "mkCallUDs: discarding" (vcat [ppr f, ppr args, ppr n_tyvars, ppr n_dicts, ppr (map interestingArg dicts)])
1214 emptyUDs -- Not overloaded, or no specialisation wanted
1217 = -- pprTrace "mkCallUDs: keeping" (vcat [ppr f, ppr args, ppr n_tyvars, ppr n_dicts, ppr (map interestingArg dicts)])
1218 singleCall f spec_tys dicts
1220 (tyvars, theta, _) = tcSplitSigmaTy (idType f)
1221 constrained_tyvars = tyVarsOfTheta theta
1222 n_tyvars = length tyvars
1223 n_dicts = length theta
1225 spec_tys = [mk_spec_ty tv ty | (tv, Type ty) <- tyvars `zip` args]
1226 dicts = [dict_expr | (_, dict_expr) <- theta `zip` (drop n_tyvars args)]
1229 | tyvar `elemVarSet` constrained_tyvars = Just ty
1230 | otherwise = Nothing
1233 Note [Interesting dictionary arguments]
1234 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1236 \a.\d:Eq a. let f = ... in ...(f d)...
1237 There really is not much point in specialising f wrt the dictionary d,
1238 because the code for the specialised f is not improved at all, because
1239 d is lambda-bound. We simply get junk specialisations.
1241 We re-use the function SimplUtils.interestingArg function to determine
1242 what sort of dictionary arguments have *some* information in them.
1246 plusUDs :: UsageDetails -> UsageDetails -> UsageDetails
1247 plusUDs (MkUD {dict_binds = db1, calls = calls1, ud_fvs = fvs1})
1248 (MkUD {dict_binds = db2, calls = calls2, ud_fvs = fvs2})
1249 = MkUD {dict_binds = d, calls = c, ud_fvs = fvs1 `unionVarSet` fvs2}
1251 d = db1 `unionBags` db2
1252 c = calls1 `unionCalls` calls2
1254 plusUDList :: [UsageDetails] -> UsageDetails
1255 plusUDList = foldr plusUDs emptyUDs
1257 -- zapCalls deletes calls to ids from uds
1258 zapCalls :: [Id] -> CallDetails -> CallDetails
1259 zapCalls ids calls = delListFromFM calls ids
1261 mkDB :: CoreBind -> DictBind
1262 mkDB bind = (bind, bind_fvs bind)
1264 bind_fvs :: CoreBind -> VarSet
1265 bind_fvs (NonRec bndr rhs) = pair_fvs (bndr,rhs)
1266 bind_fvs (Rec prs) = foldl delVarSet rhs_fvs bndrs
1269 rhs_fvs = unionVarSets (map pair_fvs prs)
1271 pair_fvs :: (Id, CoreExpr) -> VarSet
1272 pair_fvs (bndr, rhs) = exprFreeVars rhs `unionVarSet` idFreeVars bndr
1273 -- Don't forget variables mentioned in the
1274 -- rules of the bndr. C.f. OccAnal.addRuleUsage
1275 -- Also tyvars mentioned in its type; they may not appear in the RHS
1279 addDictBind :: (Id,CoreExpr) -> UsageDetails -> UsageDetails
1280 addDictBind (dict,rhs) uds
1281 = uds { dict_binds = db `consBag` dict_binds uds
1282 , ud_fvs = ud_fvs uds `unionVarSet` fvs }
1284 db@(_, fvs) = mkDB (NonRec dict rhs)
1286 dumpAllDictBinds :: UsageDetails -> [CoreBind] -> [CoreBind]
1287 dumpAllDictBinds (MkUD {dict_binds = dbs}) binds
1288 = foldrBag add binds dbs
1290 add (bind,_) binds = bind : binds
1292 dumpUDs :: [CoreBndr]
1293 -> UsageDetails -> CoreExpr
1294 -> (UsageDetails, CoreExpr)
1295 dumpUDs bndrs (MkUD { dict_binds = orig_dbs
1296 , calls = orig_calls
1297 , ud_fvs = fvs}) body
1298 = (new_uds, foldrBag add_let body dump_dbs)
1299 -- This may delete fewer variables
1300 -- than in priciple possible
1303 MkUD { dict_binds = free_dbs
1304 , calls = free_calls
1305 , ud_fvs = fvs `minusVarSet` bndr_set}
1307 bndr_set = mkVarSet bndrs
1308 add_let (bind,_) body = Let bind body
1310 (free_dbs, dump_dbs, dump_set)
1311 = foldlBag dump_db (emptyBag, emptyBag, bndr_set) orig_dbs
1312 -- Important that it's foldl not foldr;
1313 -- we're accumulating the set of dumped ids in dump_set
1315 free_calls = filterCalls dump_set orig_calls
1317 dump_db (free_dbs, dump_dbs, dump_idset) db@(bind, fvs)
1318 | dump_idset `intersectsVarSet` fvs -- Dump it
1319 = (free_dbs, dump_dbs `snocBag` db,
1320 extendVarSetList dump_idset (bindersOf bind))
1322 | otherwise -- Don't dump it
1323 = (free_dbs `snocBag` db, dump_dbs, dump_idset)
1325 filterCalls :: VarSet -> CallDetails -> CallDetails
1326 -- Remove any calls that mention the variables
1327 filterCalls bs calls
1328 = mapFM (\_ cs -> filter_calls cs) $
1329 filterFM (\k _ -> not (k `elemVarSet` bs)) calls
1331 filter_calls :: CallInfo -> CallInfo
1332 filter_calls = filterFM (\_ (_, fvs) -> not (fvs `intersectsVarSet` bs))
1336 %************************************************************************
1338 \subsubsection{Boring helper functions}
1340 %************************************************************************
1343 type SpecM a = UniqSM a
1345 initSM :: UniqSupply -> SpecM a -> a
1348 mapAndCombineSM :: (a -> SpecM (b, UsageDetails)) -> [a] -> SpecM ([b], UsageDetails)
1349 mapAndCombineSM _ [] = return ([], emptyUDs)
1350 mapAndCombineSM f (x:xs) = do (y, uds1) <- f x
1351 (ys, uds2) <- mapAndCombineSM f xs
1352 return (y:ys, uds1 `plusUDs` uds2)
1354 cloneBindSM :: Subst -> CoreBind -> SpecM (Subst, Subst, CoreBind)
1355 -- Clone the binders of the bind; return new bind with the cloned binders
1356 -- Return the substitution to use for RHSs, and the one to use for the body
1357 cloneBindSM subst (NonRec bndr rhs) = do
1358 us <- getUniqueSupplyM
1359 let (subst', bndr') = cloneIdBndr subst us bndr
1360 return (subst, subst', NonRec bndr' rhs)
1362 cloneBindSM subst (Rec pairs) = do
1363 us <- getUniqueSupplyM
1364 let (subst', bndrs') = cloneRecIdBndrs subst us (map fst pairs)
1365 return (subst', subst', Rec (bndrs' `zip` map snd pairs))
1367 cloneDictBndrs :: Subst -> [CoreBndr] -> SpecM (Subst, [CoreBndr])
1368 cloneDictBndrs subst bndrs
1369 = do { us <- getUniqueSupplyM
1370 ; return (cloneIdBndrs subst us bndrs) }
1372 newSpecIdSM :: Id -> Type -> SpecM Id
1373 -- Give the new Id a similar occurrence name to the old one
1374 newSpecIdSM old_id new_ty
1375 = do { uniq <- getUniqueM
1377 name = idName old_id
1378 new_occ = mkSpecOcc (nameOccName name)
1379 new_id = mkUserLocal new_occ uniq new_ty (getSrcSpan name)
1384 Old (but interesting) stuff about unboxed bindings
1385 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1387 What should we do when a value is specialised to a *strict* unboxed value?
1389 map_*_* f (x:xs) = let h = f x
1393 Could convert let to case:
1395 map_*_Int# f (x:xs) = case f x of h# ->
1399 This may be undesirable since it forces evaluation here, but the value
1400 may not be used in all branches of the body. In the general case this
1401 transformation is impossible since the mutual recursion in a letrec
1402 cannot be expressed as a case.
1404 There is also a problem with top-level unboxed values, since our
1405 implementation cannot handle unboxed values at the top level.
1407 Solution: Lift the binding of the unboxed value and extract it when it
1410 map_*_Int# f (x:xs) = let h = case (f x) of h# -> _Lift h#
1415 Now give it to the simplifier and the _Lifting will be optimised away.
1417 The benfit is that we have given the specialised "unboxed" values a
1418 very simplep lifted semantics and then leave it up to the simplifier to
1419 optimise it --- knowing that the overheads will be removed in nearly
1422 In particular, the value will only be evaluted in the branches of the
1423 program which use it, rather than being forced at the point where the
1424 value is bound. For example:
1426 filtermap_*_* p f (x:xs)
1433 filtermap_*_Int# p f (x:xs)
1434 = let h = case (f x) of h# -> _Lift h#
1437 True -> case h of _Lift h#
1441 The binding for h can still be inlined in the one branch and the
1442 _Lifting eliminated.
1445 Question: When won't the _Lifting be eliminated?
1447 Answer: When they at the top-level (where it is necessary) or when
1448 inlining would duplicate work (or possibly code depending on
1449 options). However, the _Lifting will still be eliminated if the
1450 strictness analyser deems the lifted binding strict.