2 % (c) The GRASP/AQUA Project, Glasgow University, 1993-1998
4 \section[Specialise]{Stamping out overloading, and (optionally) polymorphism}
7 module Specialise ( specProgram ) where
9 #include "HsVersions.h"
11 import CmdLineOpts ( DynFlags, DynFlag(..) )
12 import Id ( Id, idName, idType, mkUserLocal,
13 idSpecialisation, modifyIdInfo
15 import IdInfo ( zapSpecPragInfo )
19 import Type ( Type, mkTyVarTy, splitSigmaTy,
20 tyVarsOfTypes, tyVarsOfTheta,
23 import Subst ( Subst, mkSubst, substTy, mkSubst, substBndrs, extendSubstList, mkInScopeSet,
24 substId, substAndCloneId, substAndCloneIds, lookupIdSubst, substInScope
29 import CoreUtils ( applyTypeToArgs )
30 import CoreUnfold ( certainlyWillInline )
31 import CoreFVs ( exprFreeVars, exprsFreeVars )
32 import CoreLint ( showPass, endPass )
33 import PprCore ( pprCoreRules )
34 import Rules ( addIdSpecialisations, lookupRule )
36 import UniqSupply ( UniqSupply,
37 UniqSM, initUs_, thenUs, thenUs_, returnUs, getUniqueUs,
40 import Name ( nameOccName, mkSpecOcc, getSrcLoc )
42 import Maybes ( catMaybes, maybeToBool )
43 import ErrUtils ( dumpIfSet_dyn )
45 import List ( partition )
46 import Util ( zipEqual, zipWithEqual )
53 %************************************************************************
55 \subsection[notes-Specialise]{Implementation notes [SLPJ, Aug 18 1993]}
57 %************************************************************************
59 These notes describe how we implement specialisation to eliminate
62 The specialisation pass works on Core
63 syntax, complete with all the explicit dictionary application,
64 abstraction and construction as added by the type checker. The
65 existing type checker remains largely as it is.
67 One important thought: the {\em types} passed to an overloaded
68 function, and the {\em dictionaries} passed are mutually redundant.
69 If the same function is applied to the same type(s) then it is sure to
70 be applied to the same dictionary(s)---or rather to the same {\em
71 values}. (The arguments might look different but they will evaluate
74 Second important thought: we know that we can make progress by
75 treating dictionary arguments as static and worth specialising on. So
76 we can do without binding-time analysis, and instead specialise on
77 dictionary arguments and no others.
86 and suppose f is overloaded.
88 STEP 1: CALL-INSTANCE COLLECTION
90 We traverse <body>, accumulating all applications of f to types and
93 (Might there be partial applications, to just some of its types and
94 dictionaries? In principle yes, but in practice the type checker only
95 builds applications of f to all its types and dictionaries, so partial
96 applications could only arise as a result of transformation, and even
97 then I think it's unlikely. In any case, we simply don't accumulate such
98 partial applications.)
103 So now we have a collection of calls to f:
107 Notice that f may take several type arguments. To avoid ambiguity, we
108 say that f is called at type t1/t2 and t3/t4.
110 We take equivalence classes using equality of the *types* (ignoring
111 the dictionary args, which as mentioned previously are redundant).
113 STEP 3: SPECIALISATION
115 For each equivalence class, choose a representative (f t1 t2 d1 d2),
116 and create a local instance of f, defined thus:
118 f@t1/t2 = <f_rhs> t1 t2 d1 d2
120 f_rhs presumably has some big lambdas and dictionary lambdas, so lots
121 of simplification will now result. However we don't actually *do* that
122 simplification. Rather, we leave it for the simplifier to do. If we
123 *did* do it, though, we'd get more call instances from the specialised
124 RHS. We can work out what they are by instantiating the call-instance
125 set from f's RHS with the types t1, t2.
127 Add this new id to f's IdInfo, to record that f has a specialised version.
129 Before doing any of this, check that f's IdInfo doesn't already
130 tell us about an existing instance of f at the required type/s.
131 (This might happen if specialisation was applied more than once, or
132 it might arise from user SPECIALIZE pragmas.)
136 Wait a minute! What if f is recursive? Then we can't just plug in
137 its right-hand side, can we?
139 But it's ok. The type checker *always* creates non-recursive definitions
140 for overloaded recursive functions. For example:
142 f x = f (x+x) -- Yes I know its silly
146 f a (d::Num a) = let p = +.sel a d
148 letrec fl (y::a) = fl (p y y)
152 We still have recusion for non-overloaded functions which we
153 speciailise, but the recursive call should get specialised to the
154 same recursive version.
160 All this is crystal clear when the function is applied to *constant
161 types*; that is, types which have no type variables inside. But what if
162 it is applied to non-constant types? Suppose we find a call of f at type
163 t1/t2. There are two possibilities:
165 (a) The free type variables of t1, t2 are in scope at the definition point
166 of f. In this case there's no problem, we proceed just as before. A common
167 example is as follows. Here's the Haskell:
172 After typechecking we have
174 g a (d::Num a) (y::a) = let f b (d'::Num b) (x::b) = +.sel b d' x x
175 in +.sel a d (f a d y) (f a d y)
177 Notice that the call to f is at type type "a"; a non-constant type.
178 Both calls to f are at the same type, so we can specialise to give:
180 g a (d::Num a) (y::a) = let f@a (x::a) = +.sel a d x x
181 in +.sel a d (f@a y) (f@a y)
184 (b) The other case is when the type variables in the instance types
185 are *not* in scope at the definition point of f. The example we are
186 working with above is a good case. There are two instances of (+.sel a d),
187 but "a" is not in scope at the definition of +.sel. Can we do anything?
188 Yes, we can "common them up", a sort of limited common sub-expression deal.
191 g a (d::Num a) (y::a) = let +.sel@a = +.sel a d
192 f@a (x::a) = +.sel@a x x
193 in +.sel@a (f@a y) (f@a y)
195 This can save work, and can't be spotted by the type checker, because
196 the two instances of +.sel weren't originally at the same type.
200 * There are quite a few variations here. For example, the defn of
201 +.sel could be floated ouside the \y, to attempt to gain laziness.
202 It certainly mustn't be floated outside the \d because the d has to
205 * We don't want to inline f_rhs in this case, because
206 that will duplicate code. Just commoning up the call is the point.
208 * Nothing gets added to +.sel's IdInfo.
210 * Don't bother unless the equivalence class has more than one item!
212 Not clear whether this is all worth it. It is of course OK to
213 simply discard call-instances when passing a big lambda.
215 Polymorphism 2 -- Overloading
217 Consider a function whose most general type is
219 f :: forall a b. Ord a => [a] -> b -> b
221 There is really no point in making a version of g at Int/Int and another
222 at Int/Bool, because it's only instancing the type variable "a" which
223 buys us any efficiency. Since g is completely polymorphic in b there
224 ain't much point in making separate versions of g for the different
227 That suggests that we should identify which of g's type variables
228 are constrained (like "a") and which are unconstrained (like "b").
229 Then when taking equivalence classes in STEP 2, we ignore the type args
230 corresponding to unconstrained type variable. In STEP 3 we make
231 polymorphic versions. Thus:
233 f@t1/ = /\b -> <f_rhs> t1 b d1 d2
242 f a (d::Num a) = let g = ...
244 ...(let d1::Ord a = Num.Ord.sel a d in g a d1)...
246 Here, g is only called at one type, but the dictionary isn't in scope at the
247 definition point for g. Usually the type checker would build a
248 definition for d1 which enclosed g, but the transformation system
249 might have moved d1's defn inward. Solution: float dictionary bindings
250 outwards along with call instances.
254 f x = let g p q = p==q
260 Before specialisation, leaving out type abstractions we have
262 f df x = let g :: Eq a => a -> a -> Bool
264 h :: Num a => a -> a -> (a, Bool)
265 h dh r s = let deq = eqFromNum dh
266 in (+ dh r s, g deq r s)
270 After specialising h we get a specialised version of h, like this:
272 h' r s = let deq = eqFromNum df
273 in (+ df r s, g deq r s)
275 But we can't naively make an instance for g from this, because deq is not in scope
276 at the defn of g. Instead, we have to float out the (new) defn of deq
277 to widen its scope. Notice that this floating can't be done in advance -- it only
278 shows up when specialisation is done.
280 User SPECIALIZE pragmas
281 ~~~~~~~~~~~~~~~~~~~~~~~
282 Specialisation pragmas can be digested by the type checker, and implemented
283 by adding extra definitions along with that of f, in the same way as before
285 f@t1/t2 = <f_rhs> t1 t2 d1 d2
287 Indeed the pragmas *have* to be dealt with by the type checker, because
288 only it knows how to build the dictionaries d1 and d2! For example
290 g :: Ord a => [a] -> [a]
291 {-# SPECIALIZE f :: [Tree Int] -> [Tree Int] #-}
293 Here, the specialised version of g is an application of g's rhs to the
294 Ord dictionary for (Tree Int), which only the type checker can conjure
295 up. There might not even *be* one, if (Tree Int) is not an instance of
296 Ord! (All the other specialision has suitable dictionaries to hand
299 Problem. The type checker doesn't have to hand a convenient <f_rhs>, because
300 it is buried in a complex (as-yet-un-desugared) binding group.
303 f@t1/t2 = f* t1 t2 d1 d2
305 where f* is the Id f with an IdInfo which says "inline me regardless!".
306 Indeed all the specialisation could be done in this way.
307 That in turn means that the simplifier has to be prepared to inline absolutely
308 any in-scope let-bound thing.
311 Again, the pragma should permit polymorphism in unconstrained variables:
313 h :: Ord a => [a] -> b -> b
314 {-# SPECIALIZE h :: [Int] -> b -> b #-}
316 We *insist* that all overloaded type variables are specialised to ground types,
317 (and hence there can be no context inside a SPECIALIZE pragma).
318 We *permit* unconstrained type variables to be specialised to
320 - or left as a polymorphic type variable
321 but nothing in between. So
323 {-# SPECIALIZE h :: [Int] -> [c] -> [c] #-}
325 is *illegal*. (It can be handled, but it adds complication, and gains the
329 SPECIALISING INSTANCE DECLARATIONS
330 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
333 instance Foo a => Foo [a] where
335 {-# SPECIALIZE instance Foo [Int] #-}
337 The original instance decl creates a dictionary-function
340 dfun.Foo.List :: forall a. Foo a -> Foo [a]
342 The SPECIALIZE pragma just makes a specialised copy, just as for
343 ordinary function definitions:
345 dfun.Foo.List@Int :: Foo [Int]
346 dfun.Foo.List@Int = dfun.Foo.List Int dFooInt
348 The information about what instance of the dfun exist gets added to
349 the dfun's IdInfo in the same way as a user-defined function too.
352 Automatic instance decl specialisation?
353 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
354 Can instance decls be specialised automatically? It's tricky.
355 We could collect call-instance information for each dfun, but
356 then when we specialised their bodies we'd get new call-instances
357 for ordinary functions; and when we specialised their bodies, we might get
358 new call-instances of the dfuns, and so on. This all arises because of
359 the unrestricted mutual recursion between instance decls and value decls.
361 Still, there's no actual problem; it just means that we may not do all
362 the specialisation we could theoretically do.
364 Furthermore, instance decls are usually exported and used non-locally,
365 so we'll want to compile enough to get those specialisations done.
367 Lastly, there's no such thing as a local instance decl, so we can
368 survive solely by spitting out *usage* information, and then reading that
369 back in as a pragma when next compiling the file. So for now,
370 we only specialise instance decls in response to pragmas.
373 SPITTING OUT USAGE INFORMATION
374 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
376 To spit out usage information we need to traverse the code collecting
377 call-instance information for all imported (non-prelude?) functions
378 and data types. Then we equivalence-class it and spit it out.
380 This is done at the top-level when all the call instances which escape
381 must be for imported functions and data types.
383 *** Not currently done ***
386 Partial specialisation by pragmas
387 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
388 What about partial specialisation:
390 k :: (Ord a, Eq b) => [a] -> b -> b -> [a]
391 {-# SPECIALIZE k :: Eq b => [Int] -> b -> b -> [a] #-}
395 {-# SPECIALIZE k :: Eq b => [Int] -> [b] -> [b] -> [a] #-}
397 Seems quite reasonable. Similar things could be done with instance decls:
399 instance (Foo a, Foo b) => Foo (a,b) where
401 {-# SPECIALIZE instance Foo a => Foo (a,Int) #-}
402 {-# SPECIALIZE instance Foo b => Foo (Int,b) #-}
404 Ho hum. Things are complex enough without this. I pass.
407 Requirements for the simplifer
408 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
409 The simplifier has to be able to take advantage of the specialisation.
411 * When the simplifier finds an application of a polymorphic f, it looks in
412 f's IdInfo in case there is a suitable instance to call instead. This converts
414 f t1 t2 d1 d2 ===> f_t1_t2
416 Note that the dictionaries get eaten up too!
418 * Dictionary selection operations on constant dictionaries must be
421 +.sel Int d ===> +Int
423 The obvious way to do this is in the same way as other specialised
424 calls: +.sel has inside it some IdInfo which tells that if it's applied
425 to the type Int then it should eat a dictionary and transform to +Int.
427 In short, dictionary selectors need IdInfo inside them for constant
430 * Exactly the same applies if a superclass dictionary is being
433 Eq.sel Int d ===> dEqInt
435 * Something similar applies to dictionary construction too. Suppose
436 dfun.Eq.List is the function taking a dictionary for (Eq a) to
437 one for (Eq [a]). Then we want
439 dfun.Eq.List Int d ===> dEq.List_Int
441 Where does the Eq [Int] dictionary come from? It is built in
442 response to a SPECIALIZE pragma on the Eq [a] instance decl.
444 In short, dfun Ids need IdInfo with a specialisation for each
445 constant instance of their instance declaration.
447 All this uses a single mechanism: the SpecEnv inside an Id
450 What does the specialisation IdInfo look like?
451 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
453 The SpecEnv of an Id maps a list of types (the template) to an expression
457 For example, if f has this SpecInfo:
459 [Int, a] -> \d:Ord Int. f' a
461 it means that we can replace the call
463 f Int t ===> (\d. f' t)
465 This chucks one dictionary away and proceeds with the
466 specialised version of f, namely f'.
469 What can't be done this way?
470 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
471 There is no way, post-typechecker, to get a dictionary for (say)
472 Eq a from a dictionary for Eq [a]. So if we find
476 we can't transform to
481 eqList :: (a->a->Bool) -> [a] -> [a] -> Bool
483 Of course, we currently have no way to automatically derive
484 eqList, nor to connect it to the Eq [a] instance decl, but you
485 can imagine that it might somehow be possible. Taking advantage
486 of this is permanently ruled out.
488 Still, this is no great hardship, because we intend to eliminate
489 overloading altogether anyway!
493 A note about non-tyvar dictionaries
494 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
495 Some Ids have types like
497 forall a,b,c. Eq a -> Ord [a] -> tau
499 This seems curious at first, because we usually only have dictionary
500 args whose types are of the form (C a) where a is a type variable.
501 But this doesn't hold for the functions arising from instance decls,
502 which sometimes get arguements with types of form (C (T a)) for some
505 Should we specialise wrt this compound-type dictionary? We used to say
507 "This is a heuristic judgement, as indeed is the fact that we
508 specialise wrt only dictionaries. We choose *not* to specialise
509 wrt compound dictionaries because at the moment the only place
510 they show up is in instance decls, where they are simply plugged
511 into a returned dictionary. So nothing is gained by specialising
514 But it is simpler and more uniform to specialise wrt these dicts too;
515 and in future GHC is likely to support full fledged type signatures
517 f ;: Eq [(a,b)] => ...
520 %************************************************************************
522 \subsubsection{The new specialiser}
524 %************************************************************************
526 Our basic game plan is this. For let(rec) bound function
527 f :: (C a, D c) => (a,b,c,d) -> Bool
529 * Find any specialised calls of f, (f ts ds), where
530 ts are the type arguments t1 .. t4, and
531 ds are the dictionary arguments d1 .. d2.
533 * Add a new definition for f1 (say):
535 f1 = /\ b d -> (..body of f..) t1 b t3 d d1 d2
537 Note that we abstract over the unconstrained type arguments.
541 [t1,b,t3,d] |-> \d1 d2 -> f1 b d
543 to the specialisations of f. This will be used by the
544 simplifier to replace calls
545 (f t1 t2 t3 t4) da db
547 (\d1 d1 -> f1 t2 t4) da db
549 All the stuff about how many dictionaries to discard, and what types
550 to apply the specialised function to, are handled by the fact that the
551 SpecEnv contains a template for the result of the specialisation.
553 We don't build *partial* specialisations for f. For example:
555 f :: Eq a => a -> a -> Bool
556 {-# SPECIALISE f :: (Eq b, Eq c) => (b,c) -> (b,c) -> Bool #-}
558 Here, little is gained by making a specialised copy of f.
559 There's a distinct danger that the specialised version would
560 first build a dictionary for (Eq b, Eq c), and then select the (==)
561 method from it! Even if it didn't, not a great deal is saved.
563 We do, however, generate polymorphic, but not overloaded, specialisations:
565 f :: Eq a => [a] -> b -> b -> b
566 {#- SPECIALISE f :: [Int] -> b -> b -> b #-}
568 Hence, the invariant is this:
570 *** no specialised version is overloaded ***
573 %************************************************************************
575 \subsubsection{The exported function}
577 %************************************************************************
580 specProgram :: DynFlags -> UniqSupply -> [CoreBind] -> IO [CoreBind]
581 specProgram dflags us binds
583 showPass dflags "Specialise"
585 let binds' = initSM us (go binds `thenSM` \ (binds', uds') ->
586 returnSM (dumpAllDictBinds uds' binds'))
588 endPass dflags "Specialise" Opt_D_dump_spec binds'
590 dumpIfSet_dyn dflags Opt_D_dump_rules "Top-level specialisations"
591 (vcat (map dump_specs (concat (map bindersOf binds'))))
595 -- We need to start with a Subst that knows all the things
596 -- that are in scope, so that the substitution engine doesn't
597 -- accidentally re-use a unique that's already in use
598 -- Easiest thing is to do it all at once, as if all the top-level
599 -- decls were mutually recursive
600 top_subst = mkSubst (mkInScopeSet (mkVarSet (bindersOfBinds binds))) emptySubstEnv
602 go [] = returnSM ([], emptyUDs)
603 go (bind:binds) = go binds `thenSM` \ (binds', uds) ->
604 specBind top_subst bind uds `thenSM` \ (bind', uds') ->
605 returnSM (bind' ++ binds', uds')
607 dump_specs var = pprCoreRules var (idSpecialisation var)
610 %************************************************************************
612 \subsubsection{@specExpr@: the main function}
614 %************************************************************************
617 specVar :: Subst -> Id -> CoreExpr
618 specVar subst v = case lookupIdSubst subst v of
622 specExpr :: Subst -> CoreExpr -> SpecM (CoreExpr, UsageDetails)
623 -- We carry a substitution down:
624 -- a) we must clone any binding that might flaot outwards,
625 -- to avoid name clashes
626 -- b) we carry a type substitution to use when analysing
627 -- the RHS of specialised bindings (no type-let!)
629 ---------------- First the easy cases --------------------
630 specExpr subst (Type ty) = returnSM (Type (substTy subst ty), emptyUDs)
631 specExpr subst (Var v) = returnSM (specVar subst v, emptyUDs)
632 specExpr subst (Lit lit) = returnSM (Lit lit, emptyUDs)
634 specExpr subst (Note note body)
635 = specExpr subst body `thenSM` \ (body', uds) ->
636 returnSM (Note (specNote subst note) body', uds)
639 ---------------- Applications might generate a call instance --------------------
640 specExpr subst expr@(App fun arg)
643 go (App fun arg) args = specExpr subst arg `thenSM` \ (arg', uds_arg) ->
644 go fun (arg':args) `thenSM` \ (fun', uds_app) ->
645 returnSM (App fun' arg', uds_arg `plusUDs` uds_app)
647 go (Var f) args = case specVar subst f of
648 Var f' -> returnSM (Var f', mkCallUDs subst f' args)
649 e' -> returnSM (e', emptyUDs) -- I don't expect this!
650 go other args = specExpr subst other
652 ---------------- Lambda/case require dumping of usage details --------------------
653 specExpr subst e@(Lam _ _)
654 = specExpr subst' body `thenSM` \ (body', uds) ->
656 (filtered_uds, body'') = dumpUDs bndrs' uds body'
658 returnSM (mkLams bndrs' body'', filtered_uds)
660 (bndrs, body) = collectBinders e
661 (subst', bndrs') = substBndrs subst bndrs
662 -- More efficient to collect a group of binders together all at once
663 -- and we don't want to split a lambda group with dumped bindings
665 specExpr subst (Case scrut case_bndr alts)
666 = specExpr subst scrut `thenSM` \ (scrut', uds_scrut) ->
667 mapAndCombineSM spec_alt alts `thenSM` \ (alts', uds_alts) ->
668 returnSM (Case scrut' case_bndr' alts', uds_scrut `plusUDs` uds_alts)
670 (subst_alt, case_bndr') = substId subst case_bndr
671 -- No need to clone case binder; it can't float like a let(rec)
673 spec_alt (con, args, rhs)
674 = specExpr subst_rhs rhs `thenSM` \ (rhs', uds) ->
676 (uds', rhs'') = dumpUDs args uds rhs'
678 returnSM ((con, args', rhs''), uds')
680 (subst_rhs, args') = substBndrs subst_alt args
682 ---------------- Finally, let is the interesting case --------------------
683 specExpr subst (Let bind body)
685 cloneBindSM subst bind `thenSM` \ (rhs_subst, body_subst, bind') ->
687 -- Deal with the body
688 specExpr body_subst body `thenSM` \ (body', body_uds) ->
690 -- Deal with the bindings
691 specBind rhs_subst bind' body_uds `thenSM` \ (binds', uds) ->
694 returnSM (foldr Let body' binds', uds)
696 -- Must apply the type substitution to coerceions
697 specNote subst (Coerce t1 t2) = Coerce (substTy subst t1) (substTy subst t2)
698 specNote subst note = note
701 %************************************************************************
703 \subsubsection{Dealing with a binding}
705 %************************************************************************
708 specBind :: Subst -- Use this for RHSs
710 -> UsageDetails -- Info on how the scope of the binding
711 -> SpecM ([CoreBind], -- New bindings
712 UsageDetails) -- And info to pass upstream
714 specBind rhs_subst bind body_uds
715 = specBindItself rhs_subst bind (calls body_uds) `thenSM` \ (bind', bind_uds) ->
717 bndrs = bindersOf bind
718 all_uds = zapCalls bndrs (body_uds `plusUDs` bind_uds)
719 -- It's important that the `plusUDs` is this way round,
720 -- because body_uds may bind dictionaries that are
721 -- used in the calls passed to specDefn. So the
722 -- dictionary bindings in bind_uds may mention
723 -- dictionaries bound in body_uds.
725 case splitUDs bndrs all_uds of
727 (_, ([],[])) -- This binding doesn't bind anything needed
728 -- in the UDs, so put the binding here
729 -- This is the case for most non-dict bindings, except
730 -- for the few that are mentioned in a dict binding
731 -- that is floating upwards in body_uds
732 -> returnSM ([bind'], all_uds)
734 (float_uds, (dict_binds, calls)) -- This binding is needed in the UDs, so float it out
735 -> returnSM ([], float_uds `plusUDs` mkBigUD bind' dict_binds calls)
738 -- A truly gruesome function
739 mkBigUD bind@(NonRec _ _) dbs calls
740 = -- Common case: non-recursive and no specialisations
741 -- (if there were any specialistions it would have been made recursive)
742 MkUD { dict_binds = listToBag (mkDB bind : dbs),
743 calls = listToCallDetails calls }
745 mkBigUD bind dbs calls
747 MkUD { dict_binds = unitBag (mkDB (Rec (bind_prs bind ++ dbsToPairs dbs))),
749 calls = listToCallDetails calls }
751 bind_prs (NonRec b r) = [(b,r)]
752 bind_prs (Rec prs) = prs
755 dbsToPairs ((bind,_):dbs) = bind_prs bind ++ dbsToPairs dbs
757 -- specBindItself deals with the RHS, specialising it according
758 -- to the calls found in the body (if any)
759 specBindItself rhs_subst (NonRec bndr rhs) call_info
760 = specDefn rhs_subst call_info (bndr,rhs) `thenSM` \ ((bndr',rhs'), spec_defns, spec_uds) ->
762 new_bind | null spec_defns = NonRec bndr' rhs'
763 | otherwise = Rec ((bndr',rhs'):spec_defns)
764 -- bndr' mentions the spec_defns in its SpecEnv
765 -- Not sure why we couln't just put the spec_defns first
767 returnSM (new_bind, spec_uds)
769 specBindItself rhs_subst (Rec pairs) call_info
770 = mapSM (specDefn rhs_subst call_info) pairs `thenSM` \ stuff ->
772 (pairs', spec_defns_s, spec_uds_s) = unzip3 stuff
773 spec_defns = concat spec_defns_s
774 spec_uds = plusUDList spec_uds_s
775 new_bind = Rec (spec_defns ++ pairs')
777 returnSM (new_bind, spec_uds)
780 specDefn :: Subst -- Subst to use for RHS
781 -> CallDetails -- Info on how it is used in its scope
782 -> (Id, CoreExpr) -- The thing being bound and its un-processed RHS
783 -> SpecM ((Id, CoreExpr), -- The thing and its processed RHS
784 -- the Id may now have specialisations attached
785 [(Id,CoreExpr)], -- Extra, specialised bindings
786 UsageDetails -- Stuff to fling upwards from the RHS and its
787 ) -- specialised versions
789 specDefn subst calls (fn, rhs)
790 -- The first case is the interesting one
791 | n_tyvars == length rhs_tyvars -- Rhs of fn's defn has right number of big lambdas
792 && n_dicts <= length rhs_bndrs -- and enough dict args
793 && not (null calls_for_me) -- And there are some calls to specialise
794 && not (certainlyWillInline fn) -- And it's not small
795 -- If it's small, it's better just to inline
796 -- it than to construct lots of specialisations
797 = -- Specialise the body of the function
798 specExpr subst rhs `thenSM` \ (rhs', rhs_uds) ->
800 -- Make a specialised version for each call in calls_for_me
801 mapSM spec_call calls_for_me `thenSM` \ stuff ->
803 (spec_defns, spec_uds, spec_env_stuff) = unzip3 stuff
805 fn' = addIdSpecialisations zapped_fn spec_env_stuff
807 returnSM ((fn',rhs'),
809 rhs_uds `plusUDs` plusUDList spec_uds)
811 | otherwise -- No calls or RHS doesn't fit our preconceptions
812 = specExpr subst rhs `thenSM` \ (rhs', rhs_uds) ->
813 returnSM ((zapped_fn, rhs'), [], rhs_uds)
816 zapped_fn = modifyIdInfo zapSpecPragInfo fn
817 -- If the fn is a SpecPragmaId, make it discardable
818 -- It's role as a holder for a call instance is o'er
819 -- But it might be alive for some other reason by now.
822 (tyvars, theta, _) = splitSigmaTy fn_type
823 n_tyvars = length tyvars
824 n_dicts = length theta
826 (rhs_tyvars, rhs_ids, rhs_body) = collectTyAndValBinders rhs
827 rhs_dicts = take n_dicts rhs_ids
828 rhs_bndrs = rhs_tyvars ++ rhs_dicts
829 body = mkLams (drop n_dicts rhs_ids) rhs_body
830 -- Glue back on the non-dict lambdas
832 calls_for_me = case lookupFM calls fn of
834 Just cs -> fmToList cs
836 ----------------------------------------------------------
837 -- Specialise to one particular call pattern
838 spec_call :: ([Maybe Type], ([DictExpr], VarSet)) -- Call instance
839 -> SpecM ((Id,CoreExpr), -- Specialised definition
840 UsageDetails, -- Usage details from specialised body
841 ([CoreBndr], [CoreExpr], CoreExpr)) -- Info for the Id's SpecEnv
842 spec_call (call_ts, (call_ds, call_fvs))
843 = ASSERT( length call_ts == n_tyvars && length call_ds == n_dicts )
844 -- Calls are only recorded for properly-saturated applications
846 -- Suppose f's defn is f = /\ a b c d -> \ d1 d2 -> rhs
847 -- Supppose the call is for f [Just t1, Nothing, Just t3, Nothing] [dx1, dx2]
849 -- Construct the new binding
850 -- f1 = SUBST[a->t1,c->t3, d1->d1', d2->d2'] (/\ b d -> rhs)
851 -- PLUS the usage-details
852 -- { d1' = dx1; d2' = dx2 }
853 -- where d1', d2' are cloned versions of d1,d2, with the type substitution applied.
855 -- Note that the substitution is applied to the whole thing.
856 -- This is convenient, but just slightly fragile. Notably:
857 -- * There had better be no name clashes in a/b/c/d
860 -- poly_tyvars = [b,d] in the example above
861 -- spec_tyvars = [a,c]
862 -- ty_args = [t1,b,t3,d]
863 poly_tyvars = [tv | (tv, Nothing) <- rhs_tyvars `zip` call_ts]
864 spec_tyvars = [tv | (tv, Just _) <- rhs_tyvars `zip` call_ts]
865 ty_args = zipWithEqual "spec_call" mk_ty_arg rhs_tyvars call_ts
867 mk_ty_arg rhs_tyvar Nothing = Type (mkTyVarTy rhs_tyvar)
868 mk_ty_arg rhs_tyvar (Just ty) = Type ty
869 rhs_subst = extendSubstList subst spec_tyvars [DoneTy ty | Just ty <- call_ts]
871 cloneBinders rhs_subst rhs_dicts `thenSM` \ (rhs_subst', rhs_dicts') ->
873 inst_args = ty_args ++ map Var rhs_dicts'
875 -- Figure out the type of the specialised function
876 spec_id_ty = mkForAllTys poly_tyvars (applyTypeToArgs rhs fn_type inst_args)
878 newIdSM fn spec_id_ty `thenSM` \ spec_f ->
879 specExpr rhs_subst' (mkLams poly_tyvars body) `thenSM` \ (spec_rhs, rhs_uds) ->
881 -- The rule to put in the function's specialisation is:
882 -- forall b,d, d1',d2'. f t1 b t3 d d1' d2' = f1 b d
883 spec_env_rule = (poly_tyvars ++ rhs_dicts',
885 mkTyApps (Var spec_f) (map mkTyVarTy poly_tyvars))
887 -- Add the { d1' = dx1; d2' = dx2 } usage stuff
888 final_uds = foldr addDictBind rhs_uds (my_zipEqual "spec_call" rhs_dicts' call_ds)
890 returnSM ((spec_f, spec_rhs),
895 my_zipEqual doc xs ys
896 | length xs /= length ys = pprPanic "my_zipEqual" (ppr xs $$ ppr ys $$ (ppr fn <+> ppr call_ts) $$ ppr rhs)
897 | otherwise = zipEqual doc xs ys
900 %************************************************************************
902 \subsubsection{UsageDetails and suchlike}
904 %************************************************************************
909 dict_binds :: !(Bag DictBind),
910 -- Floated dictionary bindings
911 -- The order is important;
912 -- in ds1 `union` ds2, bindings in ds2 can depend on those in ds1
913 -- (Remember, Bags preserve order in GHC.)
915 calls :: !CallDetails
918 type DictBind = (CoreBind, VarSet)
919 -- The set is the free vars of the binding
920 -- both tyvars and dicts
922 type DictExpr = CoreExpr
924 emptyUDs = MkUD { dict_binds = emptyBag, calls = emptyFM }
926 type ProtoUsageDetails = ([DictBind],
927 [(Id, [Maybe Type], ([DictExpr], VarSet))]
930 ------------------------------------------------------------
931 type CallDetails = FiniteMap Id CallInfo
932 type CallInfo = FiniteMap [Maybe Type] -- Nothing => unconstrained type argument
933 ([DictExpr], VarSet) -- Dict args and the vars of the whole
934 -- call (including tyvars)
935 -- [*not* include the main id itself, of course]
936 -- The finite maps eliminate duplicates
937 -- The list of types and dictionaries is guaranteed to
938 -- match the type of f
940 unionCalls :: CallDetails -> CallDetails -> CallDetails
941 unionCalls c1 c2 = plusFM_C plusFM c1 c2
943 singleCall :: Id -> [Maybe Type] -> [DictExpr] -> CallDetails
944 singleCall id tys dicts
945 = unitFM id (unitFM tys (dicts, call_fvs))
947 call_fvs = exprsFreeVars dicts `unionVarSet` tys_fvs
948 tys_fvs = tyVarsOfTypes (catMaybes tys)
949 -- The type args (tys) are guaranteed to be part of the dictionary
950 -- types, because they are just the constrained types,
951 -- and the dictionary is therefore sure to be bound
952 -- inside the binding for any type variables free in the type;
953 -- hence it's safe to neglect tyvars free in tys when making
954 -- the free-var set for this call
955 -- BUT I don't trust this reasoning; play safe and include tys_fvs
957 -- We don't include the 'id' itself.
959 listToCallDetails calls
960 = foldr (unionCalls . mk_call) emptyFM calls
962 mk_call (id, tys, dicts_w_fvs) = unitFM id (unitFM tys dicts_w_fvs)
963 -- NB: the free vars of the call are provided
965 callDetailsToList calls = [ (id,tys,dicts)
966 | (id,fm) <- fmToList calls,
967 (tys,dicts) <- fmToList fm
970 mkCallUDs subst f args
972 || length spec_tys /= n_tyvars
973 || length dicts /= n_dicts
974 || maybeToBool (lookupRule (substInScope subst) f args)
975 -- There's already a rule covering this call. A typical case
976 -- is where there's an explicit user-provided rule. Then
977 -- we don't want to create a specialised version
978 -- of the function that overlaps.
979 = emptyUDs -- Not overloaded, or no specialisation wanted
982 = MkUD {dict_binds = emptyBag,
983 calls = singleCall f spec_tys dicts
986 (tyvars, theta, _) = splitSigmaTy (idType f)
987 constrained_tyvars = tyVarsOfTheta theta
988 n_tyvars = length tyvars
989 n_dicts = length theta
991 spec_tys = [mk_spec_ty tv ty | (tv, Type ty) <- tyvars `zip` args]
992 dicts = [dict_expr | (_, dict_expr) <- theta `zip` (drop n_tyvars args)]
994 mk_spec_ty tyvar ty | tyvar `elemVarSet` constrained_tyvars
999 ------------------------------------------------------------
1000 plusUDs :: UsageDetails -> UsageDetails -> UsageDetails
1001 plusUDs (MkUD {dict_binds = db1, calls = calls1})
1002 (MkUD {dict_binds = db2, calls = calls2})
1003 = MkUD {dict_binds = d, calls = c}
1005 d = db1 `unionBags` db2
1006 c = calls1 `unionCalls` calls2
1008 plusUDList = foldr plusUDs emptyUDs
1010 -- zapCalls deletes calls to ids from uds
1011 zapCalls ids uds = uds {calls = delListFromFM (calls uds) ids}
1013 mkDB bind = (bind, bind_fvs bind)
1015 bind_fvs (NonRec bndr rhs) = exprFreeVars rhs
1016 bind_fvs (Rec prs) = foldl delVarSet rhs_fvs bndrs
1019 rhs_fvs = unionVarSets [exprFreeVars rhs | (bndr,rhs) <- prs]
1021 addDictBind (dict,rhs) uds = uds { dict_binds = mkDB (NonRec dict rhs) `consBag` dict_binds uds }
1023 dumpAllDictBinds (MkUD {dict_binds = dbs}) binds
1024 = foldrBag add binds dbs
1026 add (bind,_) binds = bind : binds
1028 dumpUDs :: [CoreBndr]
1029 -> UsageDetails -> CoreExpr
1030 -> (UsageDetails, CoreExpr)
1031 dumpUDs bndrs uds body
1032 = (free_uds, foldr add_let body dict_binds)
1034 (free_uds, (dict_binds, _)) = splitUDs bndrs uds
1035 add_let (bind,_) body = Let bind body
1037 splitUDs :: [CoreBndr]
1039 -> (UsageDetails, -- These don't mention the binders
1040 ProtoUsageDetails) -- These do
1042 splitUDs bndrs uds@(MkUD {dict_binds = orig_dbs,
1043 calls = orig_calls})
1045 = if isEmptyBag dump_dbs && null dump_calls then
1046 -- Common case: binder doesn't affect floats
1050 -- Binders bind some of the fvs of the floats
1051 (MkUD {dict_binds = free_dbs,
1052 calls = listToCallDetails free_calls},
1053 (bagToList dump_dbs, dump_calls)
1057 bndr_set = mkVarSet bndrs
1059 (free_dbs, dump_dbs, dump_idset)
1060 = foldlBag dump_db (emptyBag, emptyBag, bndr_set) orig_dbs
1061 -- Important that it's foldl not foldr;
1062 -- we're accumulating the set of dumped ids in dump_set
1064 -- Filter out any calls that mention things that are being dumped
1065 orig_call_list = callDetailsToList orig_calls
1066 (dump_calls, free_calls) = partition captured orig_call_list
1067 captured (id,tys,(dicts, fvs)) = fvs `intersectsVarSet` dump_idset
1068 || id `elemVarSet` dump_idset
1070 dump_db (free_dbs, dump_dbs, dump_idset) db@(bind, fvs)
1071 | dump_idset `intersectsVarSet` fvs -- Dump it
1072 = (free_dbs, dump_dbs `snocBag` db,
1073 dump_idset `unionVarSet` mkVarSet (bindersOf bind))
1075 | otherwise -- Don't dump it
1076 = (free_dbs `snocBag` db, dump_dbs, dump_idset)
1080 %************************************************************************
1082 \subsubsection{Boring helper functions}
1084 %************************************************************************
1087 lookupId:: IdEnv Id -> Id -> Id
1088 lookupId env id = case lookupVarEnv env id of
1092 ----------------------------------------
1093 type SpecM a = UniqSM a
1097 getUniqSM = getUniqueUs
1101 mapAndCombineSM f [] = returnSM ([], emptyUDs)
1102 mapAndCombineSM f (x:xs) = f x `thenSM` \ (y, uds1) ->
1103 mapAndCombineSM f xs `thenSM` \ (ys, uds2) ->
1104 returnSM (y:ys, uds1 `plusUDs` uds2)
1106 cloneBindSM :: Subst -> CoreBind -> SpecM (Subst, Subst, CoreBind)
1107 -- Clone the binders of the bind; return new bind with the cloned binders
1108 -- Return the substitution to use for RHSs, and the one to use for the body
1109 cloneBindSM subst (NonRec bndr rhs)
1112 (subst', us', bndr') = substAndCloneId subst us bndr
1114 ((subst, subst', NonRec bndr' rhs), us')
1116 cloneBindSM subst (Rec pairs)
1119 (subst', us', bndrs') = substAndCloneIds subst us (map fst pairs)
1121 ((subst', subst', Rec (bndrs' `zip` map snd pairs)), us')
1123 cloneBinders subst bndrs
1126 (subst', us', bndrs') = substAndCloneIds subst us bndrs
1128 ((subst', bndrs'), us')
1130 newIdSM old_id new_ty
1131 = getUniqSM `thenSM` \ uniq ->
1133 -- Give the new Id a similar occurrence name to the old one
1134 name = idName old_id
1135 new_id = mkUserLocal (mkSpecOcc (nameOccName name)) uniq new_ty (getSrcLoc name)
1141 Old (but interesting) stuff about unboxed bindings
1142 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1144 What should we do when a value is specialised to a *strict* unboxed value?
1146 map_*_* f (x:xs) = let h = f x
1150 Could convert let to case:
1152 map_*_Int# f (x:xs) = case f x of h# ->
1156 This may be undesirable since it forces evaluation here, but the value
1157 may not be used in all branches of the body. In the general case this
1158 transformation is impossible since the mutual recursion in a letrec
1159 cannot be expressed as a case.
1161 There is also a problem with top-level unboxed values, since our
1162 implementation cannot handle unboxed values at the top level.
1164 Solution: Lift the binding of the unboxed value and extract it when it
1167 map_*_Int# f (x:xs) = let h = case (f x) of h# -> _Lift h#
1172 Now give it to the simplifier and the _Lifting will be optimised away.
1174 The benfit is that we have given the specialised "unboxed" values a
1175 very simplep lifted semantics and then leave it up to the simplifier to
1176 optimise it --- knowing that the overheads will be removed in nearly
1179 In particular, the value will only be evaluted in the branches of the
1180 program which use it, rather than being forced at the point where the
1181 value is bound. For example:
1183 filtermap_*_* p f (x:xs)
1190 filtermap_*_Int# p f (x:xs)
1191 = let h = case (f x) of h# -> _Lift h#
1194 True -> case h of _Lift h#
1198 The binding for h can still be inlined in the one branch and the
1199 _Lifting eliminated.
1202 Question: When won't the _Lifting be eliminated?
1204 Answer: When they at the top-level (where it is necessary) or when
1205 inlining would duplicate work (or possibly code depending on
1206 options). However, the _Lifting will still be eliminated if the
1207 strictness analyser deems the lifted binding strict.