2 % (c) The AQUA Project, Glasgow University, 1993-1998
4 \section[Simplify]{The main module of the simplifier}
7 module Simplify ( simplTopBinds, simplExpr ) where
9 #include "HsVersions.h"
13 import Type hiding ( substTy, extendTvSubst )
16 import Literal ( mkStringLit )
17 import MkId ( rUNTIME_ERROR_ID )
22 import FamInstEnv ( topNormaliseType )
23 import DataCon ( dataConRepStrictness, dataConUnivTyVars )
25 import NewDemand ( isStrictDmd )
26 import PprCore ( pprParendExpr, pprCoreExpr )
27 import CoreUnfold ( mkUnfolding, callSiteInline, CallCtxt(..) )
29 import Rules ( lookupRule )
30 import BasicTypes ( isMarkedStrict )
31 import CostCentre ( currentCCS )
32 import TysPrim ( realWorldStatePrimTy )
33 import PrelInfo ( realWorldPrimId )
34 import BasicTypes ( TopLevelFlag(..), isTopLevel,
35 RecFlag(..), isNonRuleLoopBreaker )
36 import Maybes ( orElse )
37 import Data.List ( mapAccumL )
44 The guts of the simplifier is in this module, but the driver loop for
45 the simplifier is in SimplCore.lhs.
48 -----------------------------------------
49 *** IMPORTANT NOTE ***
50 -----------------------------------------
51 The simplifier used to guarantee that the output had no shadowing, but
52 it does not do so any more. (Actually, it never did!) The reason is
53 documented with simplifyArgs.
56 -----------------------------------------
57 *** IMPORTANT NOTE ***
58 -----------------------------------------
59 Many parts of the simplifier return a bunch of "floats" as well as an
60 expression. This is wrapped as a datatype SimplUtils.FloatsWith.
62 All "floats" are let-binds, not case-binds, but some non-rec lets may
63 be unlifted (with RHS ok-for-speculation).
67 -----------------------------------------
68 ORGANISATION OF FUNCTIONS
69 -----------------------------------------
71 - simplify all top-level binders
72 - for NonRec, call simplRecOrTopPair
73 - for Rec, call simplRecBind
76 ------------------------------
77 simplExpr (applied lambda) ==> simplNonRecBind
78 simplExpr (Let (NonRec ...) ..) ==> simplNonRecBind
79 simplExpr (Let (Rec ...) ..) ==> simplify binders; simplRecBind
81 ------------------------------
82 simplRecBind [binders already simplfied]
83 - use simplRecOrTopPair on each pair in turn
85 simplRecOrTopPair [binder already simplified]
86 Used for: recursive bindings (top level and nested)
87 top-level non-recursive bindings
89 - check for PreInlineUnconditionally
93 Used for: non-top-level non-recursive bindings
94 beta reductions (which amount to the same thing)
95 Because it can deal with strict arts, it takes a
96 "thing-inside" and returns an expression
98 - check for PreInlineUnconditionally
99 - simplify binder, including its IdInfo
108 simplNonRecX: [given a *simplified* RHS, but an *unsimplified* binder]
109 Used for: binding case-binder and constr args in a known-constructor case
110 - check for PreInLineUnconditionally
114 ------------------------------
115 simplLazyBind: [binder already simplified, RHS not]
116 Used for: recursive bindings (top level and nested)
117 top-level non-recursive bindings
118 non-top-level, but *lazy* non-recursive bindings
119 [must not be strict or unboxed]
120 Returns floats + an augmented environment, not an expression
121 - substituteIdInfo and add result to in-scope
122 [so that rules are available in rec rhs]
125 - float if exposes constructor or PAP
129 completeNonRecX: [binder and rhs both simplified]
130 - if the the thing needs case binding (unlifted and not ok-for-spec)
136 completeBind: [given a simplified RHS]
137 [used for both rec and non-rec bindings, top level and not]
138 - try PostInlineUnconditionally
139 - add unfolding [this is the only place we add an unfolding]
144 Right hand sides and arguments
145 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
146 In many ways we want to treat
147 (a) the right hand side of a let(rec), and
148 (b) a function argument
149 in the same way. But not always! In particular, we would
150 like to leave these arguments exactly as they are, so they
151 will match a RULE more easily.
156 It's harder to make the rule match if we ANF-ise the constructor,
157 or eta-expand the PAP:
159 f (let { a = g x; b = h x } in (a,b))
162 On the other hand if we see the let-defns
167 then we *do* want to ANF-ise and eta-expand, so that p and q
168 can be safely inlined.
170 Even floating lets out is a bit dubious. For let RHS's we float lets
171 out if that exposes a value, so that the value can be inlined more vigorously.
174 r = let x = e in (x,x)
176 Here, if we float the let out we'll expose a nice constructor. We did experiments
177 that showed this to be a generally good thing. But it was a bad thing to float
178 lets out unconditionally, because that meant they got allocated more often.
180 For function arguments, there's less reason to expose a constructor (it won't
181 get inlined). Just possibly it might make a rule match, but I'm pretty skeptical.
182 So for the moment we don't float lets out of function arguments either.
187 For eta expansion, we want to catch things like
189 case e of (a,b) -> \x -> case a of (p,q) -> \y -> r
191 If the \x was on the RHS of a let, we'd eta expand to bring the two
192 lambdas together. And in general that's a good thing to do. Perhaps
193 we should eta expand wherever we find a (value) lambda? Then the eta
194 expansion at a let RHS can concentrate solely on the PAP case.
197 %************************************************************************
199 \subsection{Bindings}
201 %************************************************************************
204 simplTopBinds :: SimplEnv -> [InBind] -> SimplM [OutBind]
206 simplTopBinds env0 binds0
207 = do { -- Put all the top-level binders into scope at the start
208 -- so that if a transformation rule has unexpectedly brought
209 -- anything into scope, then we don't get a complaint about that.
210 -- It's rather as if the top-level binders were imported.
211 ; env1 <- simplRecBndrs env0 (bindersOfBinds binds0)
212 ; dflags <- getDOptsSmpl
213 ; let dump_flag = dopt Opt_D_dump_inlinings dflags ||
214 dopt Opt_D_dump_rule_firings dflags
215 ; env2 <- simpl_binds dump_flag env1 binds0
216 ; freeTick SimplifierDone
217 ; return (getFloats env2) }
219 -- We need to track the zapped top-level binders, because
220 -- they should have their fragile IdInfo zapped (notably occurrence info)
221 -- That's why we run down binds and bndrs' simultaneously.
223 -- The dump-flag emits a trace for each top-level binding, which
224 -- helps to locate the tracing for inlining and rule firing
225 simpl_binds :: Bool -> SimplEnv -> [InBind] -> SimplM SimplEnv
226 simpl_binds _ env [] = return env
227 simpl_binds dump env (bind:binds) = do { env' <- trace_bind dump bind $
229 ; simpl_binds dump env' binds }
231 trace_bind True bind = pprTrace "SimplBind" (ppr (bindersOf bind))
232 trace_bind False _ = \x -> x
234 simpl_bind env (Rec pairs) = simplRecBind env TopLevel pairs
235 simpl_bind env (NonRec b r) = simplRecOrTopPair env' TopLevel b b' r
237 (env', b') = addBndrRules env b (lookupRecBndr env b)
241 %************************************************************************
243 \subsection{Lazy bindings}
245 %************************************************************************
247 simplRecBind is used for
248 * recursive bindings only
251 simplRecBind :: SimplEnv -> TopLevelFlag
254 simplRecBind env0 top_lvl pairs0
255 = do { let (env_with_info, triples) = mapAccumL add_rules env0 pairs0
256 ; env1 <- go (zapFloats env_with_info) triples
257 ; return (env0 `addRecFloats` env1) }
258 -- addFloats adds the floats from env1,
259 -- *and* updates env0 with the in-scope set from env1
261 add_rules :: SimplEnv -> (InBndr,InExpr) -> (SimplEnv, (InBndr, OutBndr, InExpr))
262 -- Add the (substituted) rules to the binder
263 add_rules env (bndr, rhs) = (env', (bndr, bndr', rhs))
265 (env', bndr') = addBndrRules env bndr (lookupRecBndr env bndr)
267 go env [] = return env
269 go env ((old_bndr, new_bndr, rhs) : pairs)
270 = do { env' <- simplRecOrTopPair env top_lvl old_bndr new_bndr rhs
274 simplOrTopPair is used for
275 * recursive bindings (whether top level or not)
276 * top-level non-recursive bindings
278 It assumes the binder has already been simplified, but not its IdInfo.
281 simplRecOrTopPair :: SimplEnv
283 -> InId -> OutBndr -> InExpr -- Binder and rhs
284 -> SimplM SimplEnv -- Returns an env that includes the binding
286 simplRecOrTopPair env top_lvl old_bndr new_bndr rhs
287 | preInlineUnconditionally env top_lvl old_bndr rhs -- Check for unconditional inline
288 = do { tick (PreInlineUnconditionally old_bndr)
289 ; return (extendIdSubst env old_bndr (mkContEx env rhs)) }
292 = simplLazyBind env top_lvl Recursive old_bndr new_bndr rhs env
293 -- May not actually be recursive, but it doesn't matter
297 simplLazyBind is used for
298 * [simplRecOrTopPair] recursive bindings (whether top level or not)
299 * [simplRecOrTopPair] top-level non-recursive bindings
300 * [simplNonRecE] non-top-level *lazy* non-recursive bindings
303 1. It assumes that the binder is *already* simplified,
304 and is in scope, and its IdInfo too, except unfolding
306 2. It assumes that the binder type is lifted.
308 3. It does not check for pre-inline-unconditionallly;
309 that should have been done already.
312 simplLazyBind :: SimplEnv
313 -> TopLevelFlag -> RecFlag
314 -> InId -> OutId -- Binder, both pre-and post simpl
315 -- The OutId has IdInfo, except arity, unfolding
316 -> InExpr -> SimplEnv -- The RHS and its environment
319 simplLazyBind env top_lvl is_rec bndr bndr1 rhs rhs_se
320 = do { let rhs_env = rhs_se `setInScope` env
321 (tvs, body) = case collectTyBinders rhs of
322 (tvs, body) | not_lam body -> (tvs,body)
323 | otherwise -> ([], rhs)
324 not_lam (Lam _ _) = False
326 -- Do not do the "abstract tyyvar" thing if there's
327 -- a lambda inside, becuase it defeats eta-reduction
328 -- f = /\a. \x. g a x
331 ; (body_env, tvs') <- simplBinders rhs_env tvs
332 -- See Note [Floating and type abstraction] in SimplUtils
335 ; (body_env1, body1) <- simplExprF body_env body mkBoringStop
337 -- ANF-ise a constructor or PAP rhs
338 ; (body_env2, body2) <- prepareRhs body_env1 body1
341 <- if not (doFloatFromRhs top_lvl is_rec False body2 body_env2)
342 then -- No floating, just wrap up!
343 do { rhs' <- mkLam tvs' (wrapFloats body_env2 body2)
344 ; return (env, rhs') }
346 else if null tvs then -- Simple floating
347 do { tick LetFloatFromLet
348 ; return (addFloats env body_env2, body2) }
350 else -- Do type-abstraction first
351 do { tick LetFloatFromLet
352 ; (poly_binds, body3) <- abstractFloats tvs' body_env2 body2
353 ; rhs' <- mkLam tvs' body3
354 ; env' <- foldlM add_poly_bind env poly_binds
355 ; return (env', rhs') }
357 ; completeBind env' top_lvl bndr bndr1 rhs' }
359 add_poly_bind env (NonRec poly_id rhs)
360 = completeBind env top_lvl poly_id poly_id rhs
361 -- completeBind adds the new binding in the
362 -- proper way (ie complete with unfolding etc),
363 -- and extends the in-scope set
364 add_poly_bind env bind@(Rec _)
365 = return (extendFloats env bind)
366 -- Hack: letrecs are more awkward, so we extend "by steam"
367 -- without adding unfoldings etc. At worst this leads to
368 -- more simplifier iterations
371 A specialised variant of simplNonRec used when the RHS is already simplified,
372 notably in knownCon. It uses case-binding where necessary.
375 simplNonRecX :: SimplEnv
376 -> InId -- Old binder
377 -> OutExpr -- Simplified RHS
380 simplNonRecX env bndr new_rhs
381 = do { (env', bndr') <- simplBinder env bndr
382 ; completeNonRecX env' (isStrictId bndr) bndr bndr' new_rhs }
384 completeNonRecX :: SimplEnv
386 -> InId -- Old binder
387 -> OutId -- New binder
388 -> OutExpr -- Simplified RHS
391 completeNonRecX env is_strict old_bndr new_bndr new_rhs
392 = do { (env1, rhs1) <- prepareRhs (zapFloats env) new_rhs
394 if doFloatFromRhs NotTopLevel NonRecursive is_strict rhs1 env1
395 then do { tick LetFloatFromLet
396 ; return (addFloats env env1, rhs1) } -- Add the floats to the main env
397 else return (env, wrapFloats env1 rhs1) -- Wrap the floats around the RHS
398 ; completeBind env2 NotTopLevel old_bndr new_bndr rhs2 }
401 {- No, no, no! Do not try preInlineUnconditionally in completeNonRecX
402 Doing so risks exponential behaviour, because new_rhs has been simplified once already
403 In the cases described by the folowing commment, postInlineUnconditionally will
404 catch many of the relevant cases.
405 -- This happens; for example, the case_bndr during case of
406 -- known constructor: case (a,b) of x { (p,q) -> ... }
407 -- Here x isn't mentioned in the RHS, so we don't want to
408 -- create the (dead) let-binding let x = (a,b) in ...
410 -- Similarly, single occurrences can be inlined vigourously
411 -- e.g. case (f x, g y) of (a,b) -> ....
412 -- If a,b occur once we can avoid constructing the let binding for them.
414 Furthermore in the case-binding case preInlineUnconditionally risks extra thunks
415 -- Consider case I# (quotInt# x y) of
416 -- I# v -> let w = J# v in ...
417 -- If we gaily inline (quotInt# x y) for v, we end up building an
419 -- let w = J# (quotInt# x y) in ...
420 -- because quotInt# can fail.
422 | preInlineUnconditionally env NotTopLevel bndr new_rhs
423 = thing_inside (extendIdSubst env bndr (DoneEx new_rhs))
426 ----------------------------------
427 prepareRhs takes a putative RHS, checks whether it's a PAP or
428 constructor application and, if so, converts it to ANF, so that the
429 resulting thing can be inlined more easily. Thus
436 We also want to deal well cases like this
437 v = (f e1 `cast` co) e2
438 Here we want to make e1,e2 trivial and get
439 x1 = e1; x2 = e2; v = (f x1 `cast` co) v2
440 That's what the 'go' loop in prepareRhs does
443 prepareRhs :: SimplEnv -> OutExpr -> SimplM (SimplEnv, OutExpr)
444 -- Adds new floats to the env iff that allows us to return a good RHS
445 prepareRhs env (Cast rhs co) -- Note [Float coercions]
446 | (ty1, _ty2) <- coercionKind co -- Do *not* do this if rhs has an unlifted type
447 , not (isUnLiftedType ty1) -- see Note [Float coercions (unlifted)]
448 = do { (env', rhs') <- makeTrivial env rhs
449 ; return (env', Cast rhs' co) }
452 = do { (_is_val, env1, rhs1) <- go 0 env0 rhs0
453 ; return (env1, rhs1) }
455 go n_val_args env (Cast rhs co)
456 = do { (is_val, env', rhs') <- go n_val_args env rhs
457 ; return (is_val, env', Cast rhs' co) }
458 go n_val_args env (App fun (Type ty))
459 = do { (is_val, env', rhs') <- go n_val_args env fun
460 ; return (is_val, env', App rhs' (Type ty)) }
461 go n_val_args env (App fun arg)
462 = do { (is_val, env', fun') <- go (n_val_args+1) env fun
464 True -> do { (env'', arg') <- makeTrivial env' arg
465 ; return (True, env'', App fun' arg') }
466 False -> return (False, env, App fun arg) }
467 go n_val_args env (Var fun)
468 = return (is_val, env, Var fun)
470 is_val = n_val_args > 0 -- There is at least one arg
471 -- ...and the fun a constructor or PAP
472 && (isDataConWorkId fun || n_val_args < idArity fun)
474 = return (False, env, other)
478 Note [Float coercions]
479 ~~~~~~~~~~~~~~~~~~~~~~
480 When we find the binding
482 we'd like to transform it to
484 x = x `cast` co -- A trivial binding
485 There's a chance that e will be a constructor application or function, or something
486 like that, so moving the coerion to the usage site may well cancel the coersions
487 and lead to further optimisation. Example:
490 data instance T Int = T Int
492 foo :: Int -> Int -> Int
497 go n = case x of { T m -> go (n-m) }
498 -- This case should optimise
500 Note [Float coercions (unlifted)]
501 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
502 BUT don't do [Float coercions] if 'e' has an unlifted type.
505 foo :: Int = (error (# Int,Int #) "urk")
506 `cast` CoUnsafe (# Int,Int #) Int
508 If do the makeTrivial thing to the error call, we'll get
509 foo = case error (# Int,Int #) "urk" of v -> v `cast` ...
510 But 'v' isn't in scope!
512 These strange casts can happen as a result of case-of-case
513 bar = case (case x of { T -> (# 2,3 #); F -> error "urk" }) of
518 makeTrivial :: SimplEnv -> OutExpr -> SimplM (SimplEnv, OutExpr)
519 -- Binds the expression to a variable, if it's not trivial, returning the variable
523 | otherwise -- See Note [Take care] below
524 = do { var <- newId (fsLit "a") (exprType expr)
525 ; env' <- completeNonRecX env False var var expr
526 ; return (env', substExpr env' (Var var)) }
530 %************************************************************************
532 \subsection{Completing a lazy binding}
534 %************************************************************************
537 * deals only with Ids, not TyVars
538 * takes an already-simplified binder and RHS
539 * is used for both recursive and non-recursive bindings
540 * is used for both top-level and non-top-level bindings
542 It does the following:
543 - tries discarding a dead binding
544 - tries PostInlineUnconditionally
545 - add unfolding [this is the only place we add an unfolding]
548 It does *not* attempt to do let-to-case. Why? Because it is used for
549 - top-level bindings (when let-to-case is impossible)
550 - many situations where the "rhs" is known to be a WHNF
551 (so let-to-case is inappropriate).
553 Nor does it do the atomic-argument thing
556 completeBind :: SimplEnv
557 -> TopLevelFlag -- Flag stuck into unfolding
558 -> InId -- Old binder
559 -> OutId -> OutExpr -- New binder and RHS
561 -- completeBind may choose to do its work
562 -- * by extending the substitution (e.g. let x = y in ...)
563 -- * or by adding to the floats in the envt
565 completeBind env top_lvl old_bndr new_bndr new_rhs
566 | postInlineUnconditionally env top_lvl new_bndr occ_info new_rhs unfolding
567 -- Inline and discard the binding
568 = do { tick (PostInlineUnconditionally old_bndr)
569 ; -- pprTrace "postInlineUnconditionally" (ppr old_bndr <+> ppr new_bndr <+> ppr new_rhs) $
570 return (extendIdSubst env old_bndr (DoneEx new_rhs)) }
571 -- Use the substitution to make quite, quite sure that the
572 -- substitution will happen, since we are going to discard the binding
577 new_bndr_info = idInfo new_bndr `setArityInfo` exprArity new_rhs
580 -- Add the unfolding *only* for non-loop-breakers
581 -- Making loop breakers not have an unfolding at all
582 -- means that we can avoid tests in exprIsConApp, for example.
583 -- This is important: if exprIsConApp says 'yes' for a recursive
584 -- thing, then we can get into an infinite loop
587 -- If the unfolding is a value, the demand info may
588 -- go pear-shaped, so we nuke it. Example:
590 -- case x of (p,q) -> h p q x
591 -- Here x is certainly demanded. But after we've nuked
592 -- the case, we'll get just
593 -- let x = (a,b) in h a b x
594 -- and now x is not demanded (I'm assuming h is lazy)
595 -- This really happens. Similarly
596 -- let f = \x -> e in ...f..f...
597 -- After inlining f at some of its call sites the original binding may
598 -- (for example) be no longer strictly demanded.
599 -- The solution here is a bit ad hoc...
600 info_w_unf = new_bndr_info `setUnfoldingInfo` unfolding
601 `setWorkerInfo` worker_info
603 final_info | omit_unfolding = new_bndr_info
604 | isEvaldUnfolding unfolding = zapDemandInfo info_w_unf `orElse` info_w_unf
605 | otherwise = info_w_unf
607 final_id = new_bndr `setIdInfo` final_info
609 -- These seqs forces the Id, and hence its IdInfo,
610 -- and hence any inner substitutions
612 -- pprTrace "Binding" (ppr final_id <+> ppr unfolding) $
613 return (addNonRec env final_id new_rhs)
614 -- The addNonRec adds it to the in-scope set too
616 unfolding = mkUnfolding (isTopLevel top_lvl) new_rhs
617 worker_info = substWorker env (workerInfo old_info)
618 omit_unfolding = isNonRuleLoopBreaker occ_info || not (activeInline env old_bndr)
619 old_info = idInfo old_bndr
620 occ_info = occInfo old_info
625 %************************************************************************
627 \subsection[Simplify-simplExpr]{The main function: simplExpr}
629 %************************************************************************
631 The reason for this OutExprStuff stuff is that we want to float *after*
632 simplifying a RHS, not before. If we do so naively we get quadratic
633 behaviour as things float out.
635 To see why it's important to do it after, consider this (real) example:
649 a -- Can't inline a this round, cos it appears twice
653 Each of the ==> steps is a round of simplification. We'd save a
654 whole round if we float first. This can cascade. Consider
659 let f = let d1 = ..d.. in \y -> e
663 in \x -> ...(\y ->e)...
665 Only in this second round can the \y be applied, and it
666 might do the same again.
670 simplExpr :: SimplEnv -> CoreExpr -> SimplM CoreExpr
671 simplExpr env expr = simplExprC env expr mkBoringStop
673 simplExprC :: SimplEnv -> CoreExpr -> SimplCont -> SimplM CoreExpr
674 -- Simplify an expression, given a continuation
675 simplExprC env expr cont
676 = -- pprTrace "simplExprC" (ppr expr $$ ppr cont {- $$ ppr (seIdSubst env) -} $$ ppr (seFloats env) ) $
677 do { (env', expr') <- simplExprF (zapFloats env) expr cont
678 ; -- pprTrace "simplExprC ret" (ppr expr $$ ppr expr') $
679 -- pprTrace "simplExprC ret3" (ppr (seInScope env')) $
680 -- pprTrace "simplExprC ret4" (ppr (seFloats env')) $
681 return (wrapFloats env' expr') }
683 --------------------------------------------------
684 simplExprF :: SimplEnv -> InExpr -> SimplCont
685 -> SimplM (SimplEnv, OutExpr)
687 simplExprF env e cont
688 = -- pprTrace "simplExprF" (ppr e $$ ppr cont $$ ppr (seTvSubst env) $$ ppr (seIdSubst env) {- $$ ppr (seFloats env) -} ) $
689 simplExprF' env e cont
691 simplExprF' :: SimplEnv -> InExpr -> SimplCont
692 -> SimplM (SimplEnv, OutExpr)
693 simplExprF' env (Var v) cont = simplVar env v cont
694 simplExprF' env (Lit lit) cont = rebuild env (Lit lit) cont
695 simplExprF' env (Note n expr) cont = simplNote env n expr cont
696 simplExprF' env (Cast body co) cont = simplCast env body co cont
697 simplExprF' env (App fun arg) cont = simplExprF env fun $
698 ApplyTo NoDup arg env cont
700 simplExprF' env expr@(Lam _ _) cont
701 = simplLam env (map zap bndrs) body cont
702 -- The main issue here is under-saturated lambdas
703 -- (\x1. \x2. e) arg1
704 -- Here x1 might have "occurs-once" occ-info, because occ-info
705 -- is computed assuming that a group of lambdas is applied
706 -- all at once. If there are too few args, we must zap the
709 n_args = countArgs cont
710 n_params = length bndrs
711 (bndrs, body) = collectBinders expr
712 zap | n_args >= n_params = \b -> b
713 | otherwise = \b -> if isTyVar b then b
715 -- NB: we count all the args incl type args
716 -- so we must count all the binders (incl type lambdas)
718 simplExprF' env (Type ty) cont
719 = ASSERT( contIsRhsOrArg cont )
720 do { ty' <- simplType env ty
721 ; rebuild env (Type ty') cont }
723 simplExprF' env (Case scrut bndr _ alts) cont
724 | not (switchIsOn (getSwitchChecker env) NoCaseOfCase)
725 = -- Simplify the scrutinee with a Select continuation
726 simplExprF env scrut (Select NoDup bndr alts env cont)
729 = -- If case-of-case is off, simply simplify the case expression
730 -- in a vanilla Stop context, and rebuild the result around it
731 do { case_expr' <- simplExprC env scrut case_cont
732 ; rebuild env case_expr' cont }
734 case_cont = Select NoDup bndr alts env mkBoringStop
736 simplExprF' env (Let (Rec pairs) body) cont
737 = do { env' <- simplRecBndrs env (map fst pairs)
738 -- NB: bndrs' don't have unfoldings or rules
739 -- We add them as we go down
741 ; env'' <- simplRecBind env' NotTopLevel pairs
742 ; simplExprF env'' body cont }
744 simplExprF' env (Let (NonRec bndr rhs) body) cont
745 = simplNonRecE env bndr (rhs, env) ([], body) cont
747 ---------------------------------
748 simplType :: SimplEnv -> InType -> SimplM OutType
749 -- Kept monadic just so we can do the seqType
751 = -- pprTrace "simplType" (ppr ty $$ ppr (seTvSubst env)) $
752 seqType new_ty `seq` return new_ty
754 new_ty = substTy env ty
758 %************************************************************************
760 \subsection{The main rebuilder}
762 %************************************************************************
765 rebuild :: SimplEnv -> OutExpr -> SimplCont -> SimplM (SimplEnv, OutExpr)
766 -- At this point the substitution in the SimplEnv should be irrelevant
767 -- only the in-scope set and floats should matter
768 rebuild env expr cont0
769 = -- pprTrace "rebuild" (ppr expr $$ ppr cont0 $$ ppr (seFloats env)) $
771 Stop {} -> return (env, expr)
772 CoerceIt co cont -> rebuild env (mkCoerce co expr) cont
773 Select _ bndr alts se cont -> rebuildCase (se `setFloats` env) expr bndr alts cont
774 StrictArg fun _ info cont -> rebuildCall env (fun `App` expr) info cont
775 StrictBind b bs body se cont -> do { env' <- simplNonRecX (se `setFloats` env) b expr
776 ; simplLam env' bs body cont }
777 ApplyTo _ arg se cont -> do { arg' <- simplExpr (se `setInScope` env) arg
778 ; rebuild env (App expr arg') cont }
782 %************************************************************************
786 %************************************************************************
789 simplCast :: SimplEnv -> InExpr -> Coercion -> SimplCont
790 -> SimplM (SimplEnv, OutExpr)
791 simplCast env body co0 cont0
792 = do { co1 <- simplType env co0
793 ; simplExprF env body (addCoerce co1 cont0) }
795 addCoerce co cont = add_coerce co (coercionKind co) cont
797 add_coerce _co (s1, k1) cont -- co :: ty~ty
798 | s1 `coreEqType` k1 = cont -- is a no-op
800 add_coerce co1 (s1, _k2) (CoerceIt co2 cont)
801 | (_l1, t1) <- coercionKind co2
802 -- coerce T1 S1 (coerce S1 K1 e)
805 -- coerce T1 K1 e, otherwise
807 -- For example, in the initial form of a worker
808 -- we may find (coerce T (coerce S (\x.e))) y
809 -- and we'd like it to simplify to e[y/x] in one round
811 , s1 `coreEqType` t1 = cont -- The coerces cancel out
812 | otherwise = CoerceIt (mkTransCoercion co1 co2) cont
814 add_coerce co (s1s2, _t1t2) (ApplyTo dup (Type arg_ty) arg_se cont)
815 -- (f `cast` g) ty ---> (f ty) `cast` (g @ ty)
816 -- This implements the PushT rule from the paper
817 | Just (tyvar,_) <- splitForAllTy_maybe s1s2
818 , not (isCoVar tyvar)
819 = ApplyTo dup (Type ty') (zapSubstEnv env) (addCoerce (mkInstCoercion co ty') cont)
821 ty' = substTy (arg_se `setInScope` env) arg_ty
823 -- ToDo: the PushC rule is not implemented at all
825 add_coerce co (s1s2, _t1t2) (ApplyTo dup arg arg_se cont)
826 | not (isTypeArg arg) -- This implements the Push rule from the paper
827 , isFunTy s1s2 -- t1t2 must be a function type, becuase it's applied
828 -- co : s1s2 :=: t1t2
829 -- (coerce (T1->T2) (S1->S2) F) E
831 -- coerce T2 S2 (F (coerce S1 T1 E))
833 -- t1t2 must be a function type, T1->T2, because it's applied
834 -- to something but s1s2 might conceivably not be
836 -- When we build the ApplyTo we can't mix the out-types
837 -- with the InExpr in the argument, so we simply substitute
838 -- to make it all consistent. It's a bit messy.
839 -- But it isn't a common case.
841 -- Example of use: Trac #995
842 = ApplyTo dup new_arg (zapSubstEnv env) (addCoerce co2 cont)
844 -- we split coercion t1->t2 :=: s1->s2 into t1 :=: s1 and
845 -- t2 :=: s2 with left and right on the curried form:
846 -- (->) t1 t2 :=: (->) s1 s2
847 [co1, co2] = decomposeCo 2 co
848 new_arg = mkCoerce (mkSymCoercion co1) arg'
849 arg' = substExpr (arg_se `setInScope` env) arg
851 add_coerce co _ cont = CoerceIt co cont
855 %************************************************************************
859 %************************************************************************
862 simplLam :: SimplEnv -> [InId] -> InExpr -> SimplCont
863 -> SimplM (SimplEnv, OutExpr)
865 simplLam env [] body cont = simplExprF env body cont
867 -- Type-beta reduction
868 simplLam env (bndr:bndrs) body (ApplyTo _ (Type ty_arg) arg_se cont)
869 = ASSERT( isTyVar bndr )
870 do { tick (BetaReduction bndr)
871 ; ty_arg' <- simplType (arg_se `setInScope` env) ty_arg
872 ; simplLam (extendTvSubst env bndr ty_arg') bndrs body cont }
874 -- Ordinary beta reduction
875 simplLam env (bndr:bndrs) body (ApplyTo _ arg arg_se cont)
876 = do { tick (BetaReduction bndr)
877 ; simplNonRecE env bndr (arg, arg_se) (bndrs, body) cont }
879 -- Not enough args, so there are real lambdas left to put in the result
880 simplLam env bndrs body cont
881 = do { (env', bndrs') <- simplLamBndrs env bndrs
882 ; body' <- simplExpr env' body
883 ; new_lam <- mkLam bndrs' body'
884 ; rebuild env' new_lam cont }
887 simplNonRecE :: SimplEnv
888 -> InId -- The binder
889 -> (InExpr, SimplEnv) -- Rhs of binding (or arg of lambda)
890 -> ([InBndr], InExpr) -- Body of the let/lambda
893 -> SimplM (SimplEnv, OutExpr)
895 -- simplNonRecE is used for
896 -- * non-top-level non-recursive lets in expressions
899 -- It deals with strict bindings, via the StrictBind continuation,
900 -- which may abort the whole process
902 -- The "body" of the binding comes as a pair of ([InId],InExpr)
903 -- representing a lambda; so we recurse back to simplLam
904 -- Why? Because of the binder-occ-info-zapping done before
905 -- the call to simplLam in simplExprF (Lam ...)
907 -- First deal with type lets: let a = Type ty in b
908 simplNonRecE env bndr (Type ty_arg, rhs_se) (bndrs, body) cont
909 = do { ty_arg' <- simplType (rhs_se `setInScope` env) ty_arg
910 ; simplLam (extendTvSubst env bndr ty_arg') bndrs body cont }
912 simplNonRecE env bndr (rhs, rhs_se) (bndrs, body) cont
913 | preInlineUnconditionally env NotTopLevel bndr rhs
914 = do { tick (PreInlineUnconditionally bndr)
915 ; simplLam (extendIdSubst env bndr (mkContEx rhs_se rhs)) bndrs body cont }
918 = do { simplExprF (rhs_se `setFloats` env) rhs
919 (StrictBind bndr bndrs body env cont) }
922 = do { (env1, bndr1) <- simplNonRecBndr env bndr
923 ; let (env2, bndr2) = addBndrRules env1 bndr bndr1
924 ; env3 <- simplLazyBind env2 NotTopLevel NonRecursive bndr bndr2 rhs rhs_se
925 ; simplLam env3 bndrs body cont }
929 %************************************************************************
933 %************************************************************************
936 -- Hack alert: we only distinguish subsumed cost centre stacks for the
937 -- purposes of inlining. All other CCCSs are mapped to currentCCS.
938 simplNote :: SimplEnv -> Note -> CoreExpr -> SimplCont
939 -> SimplM (SimplEnv, OutExpr)
940 simplNote env (SCC cc) e cont
941 = do { e' <- simplExpr (setEnclosingCC env currentCCS) e
942 ; rebuild env (mkSCC cc e') cont }
944 -- See notes with SimplMonad.inlineMode
945 simplNote env InlineMe e cont
946 | Just (inside, outside) <- splitInlineCont cont -- Boring boring continuation; see notes above
947 = do { -- Don't inline inside an INLINE expression
948 e' <- simplExprC (setMode inlineMode env) e inside
949 ; rebuild env (mkInlineMe e') outside }
951 | otherwise -- Dissolve the InlineMe note if there's
952 -- an interesting context of any kind to combine with
953 -- (even a type application -- anything except Stop)
954 = simplExprF env e cont
956 simplNote env (CoreNote s) e cont = do
957 e' <- simplExpr env e
958 rebuild env (Note (CoreNote s) e') cont
962 %************************************************************************
964 \subsection{Dealing with calls}
966 %************************************************************************
969 simplVar :: SimplEnv -> Id -> SimplCont -> SimplM (SimplEnv, OutExpr)
970 simplVar env var cont
971 = case substId env var of
972 DoneEx e -> simplExprF (zapSubstEnv env) e cont
973 ContEx tvs ids e -> simplExprF (setSubstEnv env tvs ids) e cont
974 DoneId var1 -> completeCall (zapSubstEnv env) var1 cont
975 -- Note [zapSubstEnv]
976 -- The template is already simplified, so don't re-substitute.
977 -- This is VITAL. Consider
979 -- let y = \z -> ...x... in
981 -- We'll clone the inner \x, adding x->x' in the id_subst
982 -- Then when we inline y, we must *not* replace x by x' in
983 -- the inlined copy!!
985 ---------------------------------------------------------
986 -- Dealing with a call site
988 completeCall :: SimplEnv -> Id -> SimplCont -> SimplM (SimplEnv, OutExpr)
989 completeCall env var cont
990 = do { dflags <- getDOptsSmpl
991 ; let (args,call_cont) = contArgs cont
992 -- The args are OutExprs, obtained by *lazily* substituting
993 -- in the args found in cont. These args are only examined
994 -- to limited depth (unless a rule fires). But we must do
995 -- the substitution; rule matching on un-simplified args would
998 ------------- First try rules ----------------
999 -- Do this before trying inlining. Some functions have
1000 -- rules *and* are strict; in this case, we don't want to
1001 -- inline the wrapper of the non-specialised thing; better
1002 -- to call the specialised thing instead.
1004 -- We used to use the black-listing mechanism to ensure that inlining of
1005 -- the wrapper didn't occur for things that have specialisations till a
1006 -- later phase, so but now we just try RULES first
1008 -- Note [Rules for recursive functions]
1009 -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1010 -- You might think that we shouldn't apply rules for a loop breaker:
1011 -- doing so might give rise to an infinite loop, because a RULE is
1012 -- rather like an extra equation for the function:
1013 -- RULE: f (g x) y = x+y
1016 -- But it's too drastic to disable rules for loop breakers.
1017 -- Even the foldr/build rule would be disabled, because foldr
1018 -- is recursive, and hence a loop breaker:
1019 -- foldr k z (build g) = g k z
1020 -- So it's up to the programmer: rules can cause divergence
1022 ; let in_scope = getInScope env
1023 maybe_rule = case activeRule dflags env of
1024 Nothing -> Nothing -- No rules apply
1025 Just act_fn -> lookupRule act_fn in_scope
1027 ; case maybe_rule of {
1028 Just (rule, rule_rhs) -> do
1029 tick (RuleFired (ru_name rule))
1030 (if dopt Opt_D_dump_rule_firings dflags then
1031 pprTrace "Rule fired" (vcat [
1032 text "Rule:" <+> ftext (ru_name rule),
1033 text "Before:" <+> ppr var <+> sep (map pprParendExpr args),
1034 text "After: " <+> pprCoreExpr rule_rhs,
1035 text "Cont: " <+> ppr call_cont])
1038 simplExprF env rule_rhs (dropArgs (ruleArity rule) cont)
1039 -- The ruleArity says how many args the rule consumed
1041 ; Nothing -> do -- No rules
1043 ------------- Next try inlining ----------------
1044 { let arg_infos = [interestingArg arg | arg <- args, isValArg arg]
1045 n_val_args = length arg_infos
1046 interesting_cont = interestingCallContext call_cont
1047 active_inline = activeInline env var
1048 maybe_inline = callSiteInline dflags active_inline var
1049 (null args) arg_infos interesting_cont
1050 ; case maybe_inline of {
1051 Just unfolding -- There is an inlining!
1052 -> do { tick (UnfoldingDone var)
1053 ; (if dopt Opt_D_dump_inlinings dflags then
1054 pprTrace ("Inlining done" ++ showSDoc (ppr var)) (vcat [
1055 text "Before:" <+> ppr var <+> sep (map pprParendExpr args),
1056 text "Inlined fn: " <+> nest 2 (ppr unfolding),
1057 text "Cont: " <+> ppr call_cont])
1060 simplExprF env unfolding cont }
1062 ; Nothing -> -- No inlining!
1064 ------------- No inlining! ----------------
1065 -- Next, look for rules or specialisations that match
1067 rebuildCall env (Var var)
1068 (mkArgInfo var n_val_args call_cont) cont
1071 rebuildCall :: SimplEnv
1072 -> OutExpr -- Function
1075 -> SimplM (SimplEnv, OutExpr)
1076 rebuildCall env fun (ArgInfo { ai_strs = [] }) cont
1077 -- When we run out of strictness args, it means
1078 -- that the call is definitely bottom; see SimplUtils.mkArgInfo
1079 -- Then we want to discard the entire strict continuation. E.g.
1080 -- * case (error "hello") of { ... }
1081 -- * (error "Hello") arg
1082 -- * f (error "Hello") where f is strict
1084 -- Then, especially in the first of these cases, we'd like to discard
1085 -- the continuation, leaving just the bottoming expression. But the
1086 -- type might not be right, so we may have to add a coerce.
1087 | not (contIsTrivial cont) -- Only do this if there is a non-trivial
1088 = return (env, mk_coerce fun) -- contination to discard, else we do it
1089 where -- again and again!
1090 fun_ty = exprType fun
1091 cont_ty = contResultType env fun_ty cont
1092 co = mkUnsafeCoercion fun_ty cont_ty
1093 mk_coerce expr | cont_ty `coreEqType` fun_ty = expr
1094 | otherwise = mkCoerce co expr
1096 rebuildCall env fun info (ApplyTo _ (Type arg_ty) se cont)
1097 = do { ty' <- simplType (se `setInScope` env) arg_ty
1098 ; rebuildCall env (fun `App` Type ty') info cont }
1101 (ArgInfo { ai_rules = has_rules, ai_strs = str:strs, ai_discs = disc:discs })
1102 (ApplyTo _ arg arg_se cont)
1103 | str -- Strict argument
1104 = -- pprTrace "Strict Arg" (ppr arg $$ ppr (seIdSubst env) $$ ppr (seInScope env)) $
1105 simplExprF (arg_se `setFloats` env) arg
1106 (StrictArg fun cci arg_info' cont)
1109 | otherwise -- Lazy argument
1110 -- DO NOT float anything outside, hence simplExprC
1111 -- There is no benefit (unlike in a let-binding), and we'd
1112 -- have to be very careful about bogus strictness through
1113 -- floating a demanded let.
1114 = do { arg' <- simplExprC (arg_se `setInScope` env) arg
1116 ; rebuildCall env (fun `App` arg') arg_info' cont }
1118 arg_info' = ArgInfo { ai_rules = has_rules, ai_strs = strs, ai_discs = discs }
1119 cci | has_rules || disc > 0 = ArgCtxt has_rules disc -- Be keener here
1120 | otherwise = BoringCtxt -- Nothing interesting
1122 rebuildCall env fun _ cont
1123 = rebuild env fun cont
1128 This part of the simplifier may break the no-shadowing invariant
1130 f (...(\a -> e)...) (case y of (a,b) -> e')
1131 where f is strict in its second arg
1132 If we simplify the innermost one first we get (...(\a -> e)...)
1133 Simplifying the second arg makes us float the case out, so we end up with
1134 case y of (a,b) -> f (...(\a -> e)...) e'
1135 So the output does not have the no-shadowing invariant. However, there is
1136 no danger of getting name-capture, because when the first arg was simplified
1137 we used an in-scope set that at least mentioned all the variables free in its
1138 static environment, and that is enough.
1140 We can't just do innermost first, or we'd end up with a dual problem:
1141 case x of (a,b) -> f e (...(\a -> e')...)
1143 I spent hours trying to recover the no-shadowing invariant, but I just could
1144 not think of an elegant way to do it. The simplifier is already knee-deep in
1145 continuations. We have to keep the right in-scope set around; AND we have
1146 to get the effect that finding (error "foo") in a strict arg position will
1147 discard the entire application and replace it with (error "foo"). Getting
1148 all this at once is TOO HARD!
1150 %************************************************************************
1152 Rebuilding a cse expression
1154 %************************************************************************
1156 Blob of helper functions for the "case-of-something-else" situation.
1159 ---------------------------------------------------------
1160 -- Eliminate the case if possible
1162 rebuildCase :: SimplEnv
1163 -> OutExpr -- Scrutinee
1164 -> InId -- Case binder
1165 -> [InAlt] -- Alternatives (inceasing order)
1167 -> SimplM (SimplEnv, OutExpr)
1169 --------------------------------------------------
1170 -- 1. Eliminate the case if there's a known constructor
1171 --------------------------------------------------
1173 rebuildCase env scrut case_bndr alts cont
1174 | Just (con,args) <- exprIsConApp_maybe scrut
1175 -- Works when the scrutinee is a variable with a known unfolding
1176 -- as well as when it's an explicit constructor application
1177 = knownCon env scrut (DataAlt con) args case_bndr alts cont
1179 | Lit lit <- scrut -- No need for same treatment as constructors
1180 -- because literals are inlined more vigorously
1181 = knownCon env scrut (LitAlt lit) [] case_bndr alts cont
1184 --------------------------------------------------
1185 -- 2. Eliminate the case if scrutinee is evaluated
1186 --------------------------------------------------
1188 rebuildCase env scrut case_bndr [(_, bndrs, rhs)] cont
1189 -- See if we can get rid of the case altogether
1190 -- See the extensive notes on case-elimination above
1191 -- mkCase made sure that if all the alternatives are equal,
1192 -- then there is now only one (DEFAULT) rhs
1193 | all isDeadBinder bndrs -- bndrs are [InId]
1195 -- Check that the scrutinee can be let-bound instead of case-bound
1196 , exprOkForSpeculation scrut
1197 -- OK not to evaluate it
1198 -- This includes things like (==# a# b#)::Bool
1199 -- so that we simplify
1200 -- case ==# a# b# of { True -> x; False -> x }
1203 -- This particular example shows up in default methods for
1204 -- comparision operations (e.g. in (>=) for Int.Int32)
1205 || exprIsHNF scrut -- It's already evaluated
1206 || var_demanded_later scrut -- It'll be demanded later
1208 -- || not opt_SimplPedanticBottoms) -- Or we don't care!
1209 -- We used to allow improving termination by discarding cases, unless -fpedantic-bottoms was on,
1210 -- but that breaks badly for the dataToTag# primop, which relies on a case to evaluate
1211 -- its argument: case x of { y -> dataToTag# y }
1212 -- Here we must *not* discard the case, because dataToTag# just fetches the tag from
1213 -- the info pointer. So we'll be pedantic all the time, and see if that gives any
1215 -- Also we don't want to discard 'seq's
1216 = do { tick (CaseElim case_bndr)
1217 ; env' <- simplNonRecX env case_bndr scrut
1218 ; simplExprF env' rhs cont }
1220 -- The case binder is going to be evaluated later,
1221 -- and the scrutinee is a simple variable
1222 var_demanded_later (Var v) = isStrictDmd (idNewDemandInfo case_bndr)
1223 && not (isTickBoxOp v)
1224 -- ugly hack; covering this case is what
1225 -- exprOkForSpeculation was intended for.
1226 var_demanded_later _ = False
1229 --------------------------------------------------
1230 -- 3. Catch-all case
1231 --------------------------------------------------
1233 rebuildCase env scrut case_bndr alts cont
1234 = do { -- Prepare the continuation;
1235 -- The new subst_env is in place
1236 (env', dup_cont, nodup_cont) <- prepareCaseCont env alts cont
1238 -- Simplify the alternatives
1239 ; (scrut', case_bndr', alts') <- simplAlts env' scrut case_bndr alts dup_cont
1241 -- Check for empty alternatives
1242 ; if null alts' then
1243 -- This isn't strictly an error, although it is unusual.
1244 -- It's possible that the simplifer might "see" that
1245 -- an inner case has no accessible alternatives before
1246 -- it "sees" that the entire branch of an outer case is
1247 -- inaccessible. So we simply put an error case here instead.
1248 pprTrace "mkCase: null alts" (ppr case_bndr <+> ppr scrut) $
1249 let res_ty' = contResultType env' (substTy env' (coreAltsType alts)) dup_cont
1250 lit = Lit (mkStringLit "Impossible alternative")
1251 in return (env', mkApps (Var rUNTIME_ERROR_ID) [Type res_ty', lit])
1254 { case_expr <- mkCase scrut' case_bndr' alts'
1256 -- Notice that rebuild gets the in-scope set from env, not alt_env
1257 -- The case binder *not* scope over the whole returned case-expression
1258 ; rebuild env' case_expr nodup_cont } }
1261 simplCaseBinder checks whether the scrutinee is a variable, v. If so,
1262 try to eliminate uses of v in the RHSs in favour of case_bndr; that
1263 way, there's a chance that v will now only be used once, and hence
1266 Note [no-case-of-case]
1267 ~~~~~~~~~~~~~~~~~~~~~~
1268 We *used* to suppress the binder-swap in case expressoins when
1269 -fno-case-of-case is on. Old remarks:
1270 "This happens in the first simplifier pass,
1271 and enhances full laziness. Here's the bad case:
1272 f = \ y -> ...(case x of I# v -> ...(case x of ...) ... )
1273 If we eliminate the inner case, we trap it inside the I# v -> arm,
1274 which might prevent some full laziness happening. I've seen this
1275 in action in spectral/cichelli/Prog.hs:
1276 [(m,n) | m <- [1..max], n <- [1..max]]
1277 Hence the check for NoCaseOfCase."
1278 However, now the full-laziness pass itself reverses the binder-swap, so this
1279 check is no longer necessary.
1281 Note [Suppressing the case binder-swap]
1282 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1283 There is another situation when it might make sense to suppress the
1284 case-expression binde-swap. If we have
1286 case x of w1 { DEFAULT -> case x of w2 { A -> e1; B -> e2 }
1287 ...other cases .... }
1289 We'll perform the binder-swap for the outer case, giving
1291 case x of w1 { DEFAULT -> case w1 of w2 { A -> e1; B -> e2 }
1292 ...other cases .... }
1294 But there is no point in doing it for the inner case, because w1 can't
1295 be inlined anyway. Furthermore, doing the case-swapping involves
1296 zapping w2's occurrence info (see paragraphs that follow), and that
1297 forces us to bind w2 when doing case merging. So we get
1299 case x of w1 { A -> let w2 = w1 in e1
1300 B -> let w2 = w1 in e2
1301 ...other cases .... }
1303 This is plain silly in the common case where w2 is dead.
1305 Even so, I can't see a good way to implement this idea. I tried
1306 not doing the binder-swap if the scrutinee was already evaluated
1307 but that failed big-time:
1311 case v of w { MkT x ->
1312 case x of x1 { I# y1 ->
1313 case x of x2 { I# y2 -> ...
1315 Notice that because MkT is strict, x is marked "evaluated". But to
1316 eliminate the last case, we must either make sure that x (as well as
1317 x1) has unfolding MkT y1. THe straightforward thing to do is to do
1318 the binder-swap. So this whole note is a no-op.
1322 If we replace the scrutinee, v, by tbe case binder, then we have to nuke
1323 any occurrence info (eg IAmDead) in the case binder, because the
1324 case-binder now effectively occurs whenever v does. AND we have to do
1325 the same for the pattern-bound variables! Example:
1327 (case x of { (a,b) -> a }) (case x of { (p,q) -> q })
1329 Here, b and p are dead. But when we move the argment inside the first
1330 case RHS, and eliminate the second case, we get
1332 case x of { (a,b) -> a b }
1334 Urk! b is alive! Reason: the scrutinee was a variable, and case elimination
1337 Indeed, this can happen anytime the case binder isn't dead:
1338 case <any> of x { (a,b) ->
1339 case x of { (p,q) -> p } }
1340 Here (a,b) both look dead, but come alive after the inner case is eliminated.
1341 The point is that we bring into the envt a binding
1343 after the outer case, and that makes (a,b) alive. At least we do unless
1344 the case binder is guaranteed dead.
1348 Consider case (v `cast` co) of x { I# ->
1349 ... (case (v `cast` co) of {...}) ...
1350 We'd like to eliminate the inner case. We can get this neatly by
1351 arranging that inside the outer case we add the unfolding
1352 v |-> x `cast` (sym co)
1353 to v. Then we should inline v at the inner case, cancel the casts, and away we go
1355 Note [Improving seq]
1358 type family F :: * -> *
1359 type instance F Int = Int
1361 ... case e of x { DEFAULT -> rhs } ...
1363 where x::F Int. Then we'd like to rewrite (F Int) to Int, getting
1365 case e `cast` co of x'::Int
1366 I# x# -> let x = x' `cast` sym co
1369 so that 'rhs' can take advantage of the form of x'. Notice that Note
1370 [Case of cast] may then apply to the result.
1372 This showed up in Roman's experiments. Example:
1373 foo :: F Int -> Int -> Int
1374 foo t n = t `seq` bar n
1377 bar n = bar (n - case t of TI i -> i)
1378 Here we'd like to avoid repeated evaluating t inside the loop, by
1379 taking advantage of the `seq`.
1381 At one point I did transformation in LiberateCase, but it's more robust here.
1382 (Otherwise, there's a danger that we'll simply drop the 'seq' altogether, before
1383 LiberateCase gets to see it.)
1385 Note [Case elimination]
1386 ~~~~~~~~~~~~~~~~~~~~~~~
1387 The case-elimination transformation discards redundant case expressions.
1388 Start with a simple situation:
1390 case x# of ===> e[x#/y#]
1393 (when x#, y# are of primitive type, of course). We can't (in general)
1394 do this for algebraic cases, because we might turn bottom into
1397 The code in SimplUtils.prepareAlts has the effect of generalise this
1398 idea to look for a case where we're scrutinising a variable, and we
1399 know that only the default case can match. For example:
1403 DEFAULT -> ...(case x of
1407 Here the inner case is first trimmed to have only one alternative, the
1408 DEFAULT, after which it's an instance of the previous case. This
1409 really only shows up in eliminating error-checking code.
1411 We also make sure that we deal with this very common case:
1416 Here we are using the case as a strict let; if x is used only once
1417 then we want to inline it. We have to be careful that this doesn't
1418 make the program terminate when it would have diverged before, so we
1420 - e is already evaluated (it may so if e is a variable)
1421 - x is used strictly, or
1423 Lastly, the code in SimplUtils.mkCase combines identical RHSs. So
1425 case e of ===> case e of DEFAULT -> r
1429 Now again the case may be elminated by the CaseElim transformation.
1432 Further notes about case elimination
1433 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1434 Consider: test :: Integer -> IO ()
1437 Turns out that this compiles to:
1440 eta1 :: State# RealWorld ->
1441 case PrelNum.< eta PrelNum.zeroInteger of wild { __DEFAULT ->
1443 (PrelNum.jtos eta ($w[] @ Char))
1445 of wild1 { (# new_s, a4 #) -> PrelIO.lvl23 new_s }}
1447 Notice the strange '<' which has no effect at all. This is a funny one.
1448 It started like this:
1450 f x y = if x < 0 then jtos x
1451 else if y==0 then "" else jtos x
1453 At a particular call site we have (f v 1). So we inline to get
1455 if v < 0 then jtos x
1456 else if 1==0 then "" else jtos x
1458 Now simplify the 1==0 conditional:
1460 if v<0 then jtos v else jtos v
1462 Now common-up the two branches of the case:
1464 case (v<0) of DEFAULT -> jtos v
1466 Why don't we drop the case? Because it's strict in v. It's technically
1467 wrong to drop even unnecessary evaluations, and in practice they
1468 may be a result of 'seq' so we *definitely* don't want to drop those.
1469 I don't really know how to improve this situation.
1473 simplCaseBinder :: SimplEnv -> OutExpr -> OutId -> [InAlt]
1474 -> SimplM (SimplEnv, OutExpr, OutId)
1475 simplCaseBinder env0 scrut0 case_bndr0 alts
1476 = do { (env1, case_bndr1) <- simplBinder env0 case_bndr0
1478 ; fam_envs <- getFamEnvs
1479 ; (env2, scrut2, case_bndr2) <- improve_seq fam_envs env1 scrut0
1480 case_bndr0 case_bndr1 alts
1481 -- Note [Improving seq]
1483 ; let (env3, case_bndr3) = improve_case_bndr env2 scrut2 case_bndr2
1484 -- Note [Case of cast]
1486 ; return (env3, scrut2, case_bndr3) }
1489 improve_seq fam_envs env scrut case_bndr case_bndr1 [(DEFAULT,_,_)]
1490 | Just (co, ty2) <- topNormaliseType fam_envs (idType case_bndr1)
1491 = do { case_bndr2 <- newId (fsLit "nt") ty2
1492 ; let rhs = DoneEx (Var case_bndr2 `Cast` mkSymCoercion co)
1493 env2 = extendIdSubst env case_bndr rhs
1494 ; return (env2, scrut `Cast` co, case_bndr2) }
1496 improve_seq _ env scrut _ case_bndr1 _
1497 = return (env, scrut, case_bndr1)
1500 improve_case_bndr env scrut case_bndr
1501 -- See Note [no-case-of-case]
1502 -- | switchIsOn (getSwitchChecker env) NoCaseOfCase
1503 -- = (env, case_bndr)
1505 | otherwise -- Failed try; see Note [Suppressing the case binder-swap]
1506 -- not (isEvaldUnfolding (idUnfolding v))
1508 Var v -> (modifyInScope env1 v case_bndr', case_bndr')
1509 -- Note about using modifyInScope for v here
1510 -- We could extend the substitution instead, but it would be
1511 -- a hack because then the substitution wouldn't be idempotent
1512 -- any more (v is an OutId). And this does just as well.
1514 Cast (Var v) co -> (addBinderUnfolding env1 v rhs, case_bndr')
1516 rhs = Cast (Var case_bndr') (mkSymCoercion co)
1518 _ -> (env, case_bndr)
1520 case_bndr' = zapOccInfo case_bndr
1521 env1 = modifyInScope env case_bndr case_bndr'
1524 zapOccInfo :: InId -> InId -- See Note [zapOccInfo]
1525 zapOccInfo b = b `setIdOccInfo` NoOccInfo
1529 simplAlts does two things:
1531 1. Eliminate alternatives that cannot match, including the
1532 DEFAULT alternative.
1534 2. If the DEFAULT alternative can match only one possible constructor,
1535 then make that constructor explicit.
1537 case e of x { DEFAULT -> rhs }
1539 case e of x { (a,b) -> rhs }
1540 where the type is a single constructor type. This gives better code
1541 when rhs also scrutinises x or e.
1543 Here "cannot match" includes knowledge from GADTs
1545 It's a good idea do do this stuff before simplifying the alternatives, to
1546 avoid simplifying alternatives we know can't happen, and to come up with
1547 the list of constructors that are handled, to put into the IdInfo of the
1548 case binder, for use when simplifying the alternatives.
1550 Eliminating the default alternative in (1) isn't so obvious, but it can
1553 data Colour = Red | Green | Blue
1562 DEFAULT -> [ case y of ... ]
1564 If we inline h into f, the default case of the inlined h can't happen.
1565 If we don't notice this, we may end up filtering out *all* the cases
1566 of the inner case y, which give us nowhere to go!
1570 simplAlts :: SimplEnv
1572 -> InId -- Case binder
1573 -> [InAlt] -- Non-empty
1575 -> SimplM (OutExpr, OutId, [OutAlt]) -- Includes the continuation
1576 -- Like simplExpr, this just returns the simplified alternatives;
1577 -- it not return an environment
1579 simplAlts env scrut case_bndr alts cont'
1580 = -- pprTrace "simplAlts" (ppr alts $$ ppr (seIdSubst env)) $
1581 do { let alt_env = zapFloats env
1582 ; (alt_env', scrut', case_bndr') <- simplCaseBinder alt_env scrut case_bndr alts
1584 ; (imposs_deflt_cons, in_alts) <- prepareAlts alt_env' scrut case_bndr' alts
1586 ; alts' <- mapM (simplAlt alt_env' imposs_deflt_cons case_bndr' cont') in_alts
1587 ; return (scrut', case_bndr', alts') }
1589 ------------------------------------
1590 simplAlt :: SimplEnv
1591 -> [AltCon] -- These constructors can't be present when
1592 -- matching the DEFAULT alternative
1593 -> OutId -- The case binder
1598 simplAlt env imposs_deflt_cons case_bndr' cont' (DEFAULT, bndrs, rhs)
1599 = ASSERT( null bndrs )
1600 do { let env' = addBinderOtherCon env case_bndr' imposs_deflt_cons
1601 -- Record the constructors that the case-binder *can't* be.
1602 ; rhs' <- simplExprC env' rhs cont'
1603 ; return (DEFAULT, [], rhs') }
1605 simplAlt env _ case_bndr' cont' (LitAlt lit, bndrs, rhs)
1606 = ASSERT( null bndrs )
1607 do { let env' = addBinderUnfolding env case_bndr' (Lit lit)
1608 ; rhs' <- simplExprC env' rhs cont'
1609 ; return (LitAlt lit, [], rhs') }
1611 simplAlt env _ case_bndr' cont' (DataAlt con, vs, rhs)
1612 = do { -- Deal with the pattern-bound variables
1613 -- Mark the ones that are in ! positions in the
1614 -- data constructor as certainly-evaluated.
1615 -- NB: simplLamBinders preserves this eval info
1616 let vs_with_evals = add_evals (dataConRepStrictness con)
1617 ; (env', vs') <- simplLamBndrs env vs_with_evals
1619 -- Bind the case-binder to (con args)
1620 ; let inst_tys' = tyConAppArgs (idType case_bndr')
1621 con_args = map Type inst_tys' ++ varsToCoreExprs vs'
1622 env'' = addBinderUnfolding env' case_bndr'
1623 (mkConApp con con_args)
1625 ; rhs' <- simplExprC env'' rhs cont'
1626 ; return (DataAlt con, vs', rhs') }
1628 -- add_evals records the evaluated-ness of the bound variables of
1629 -- a case pattern. This is *important*. Consider
1630 -- data T = T !Int !Int
1632 -- case x of { T a b -> T (a+1) b }
1634 -- We really must record that b is already evaluated so that we don't
1635 -- go and re-evaluate it when constructing the result.
1636 -- See Note [Data-con worker strictness] in MkId.lhs
1641 go (v:vs') strs | isTyVar v = v : go vs' strs
1642 go (v:vs') (str:strs)
1643 | isMarkedStrict str = evald_v : go vs' strs
1644 | otherwise = zapped_v : go vs' strs
1646 zapped_v = zap_occ_info v
1647 evald_v = zapped_v `setIdUnfolding` evaldUnfolding
1648 go _ _ = pprPanic "cat_evals" (ppr con $$ ppr vs $$ ppr the_strs)
1650 -- zap_occ_info: if the case binder is alive, then we add the unfolding
1652 -- to the envt; so vs are now very much alive
1653 -- Note [Aug06] I can't see why this actually matters, but it's neater
1654 -- case e of t { (a,b) -> ...(case t of (p,q) -> p)... }
1655 -- ==> case e of t { (a,b) -> ...(a)... }
1656 -- Look, Ma, a is alive now.
1657 zap_occ_info | isDeadBinder case_bndr' = \ident -> ident
1658 | otherwise = zapOccInfo
1660 addBinderUnfolding :: SimplEnv -> Id -> CoreExpr -> SimplEnv
1661 addBinderUnfolding env bndr rhs
1662 = modifyInScope env bndr (bndr `setIdUnfolding` mkUnfolding False rhs)
1664 addBinderOtherCon :: SimplEnv -> Id -> [AltCon] -> SimplEnv
1665 addBinderOtherCon env bndr cons
1666 = modifyInScope env bndr (bndr `setIdUnfolding` mkOtherCon cons)
1670 %************************************************************************
1672 \subsection{Known constructor}
1674 %************************************************************************
1676 We are a bit careful with occurrence info. Here's an example
1678 (\x* -> case x of (a*, b) -> f a) (h v, e)
1680 where the * means "occurs once". This effectively becomes
1681 case (h v, e) of (a*, b) -> f a)
1683 let a* = h v; b = e in f a
1687 All this should happen in one sweep.
1690 knownCon :: SimplEnv -> OutExpr -> AltCon
1691 -> [OutExpr] -- Args *including* the universal args
1692 -> InId -> [InAlt] -> SimplCont
1693 -> SimplM (SimplEnv, OutExpr)
1695 knownCon env scrut con args bndr alts cont
1696 = do { tick (KnownBranch bndr)
1697 ; knownAlt env scrut args bndr (findAlt con alts) cont }
1699 knownAlt :: SimplEnv -> OutExpr -> [OutExpr]
1700 -> InId -> (AltCon, [CoreBndr], InExpr) -> SimplCont
1701 -> SimplM (SimplEnv, OutExpr)
1702 knownAlt env scrut _ bndr (DEFAULT, bs, rhs) cont
1704 do { env' <- simplNonRecX env bndr scrut
1705 -- This might give rise to a binding with non-atomic args
1706 -- like x = Node (f x) (g x)
1707 -- but simplNonRecX will atomic-ify it
1708 ; simplExprF env' rhs cont }
1710 knownAlt env scrut _ bndr (LitAlt _, bs, rhs) cont
1712 do { env' <- simplNonRecX env bndr scrut
1713 ; simplExprF env' rhs cont }
1715 knownAlt env scrut the_args bndr (DataAlt dc, bs, rhs) cont
1716 = do { let dead_bndr = isDeadBinder bndr -- bndr is an InId
1717 n_drop_tys = length (dataConUnivTyVars dc)
1718 ; env' <- bind_args env dead_bndr bs (drop n_drop_tys the_args)
1720 -- It's useful to bind bndr to scrut, rather than to a fresh
1721 -- binding x = Con arg1 .. argn
1722 -- because very often the scrut is a variable, so we avoid
1723 -- creating, and then subsequently eliminating, a let-binding
1724 -- BUT, if scrut is a not a variable, we must be careful
1725 -- about duplicating the arg redexes; in that case, make
1726 -- a new con-app from the args
1727 bndr_rhs = case scrut of
1730 con_app = mkConApp dc (take n_drop_tys the_args ++ con_args)
1731 con_args = [substExpr env' (varToCoreExpr b) | b <- bs]
1732 -- args are aready OutExprs, but bs are InIds
1734 ; env'' <- simplNonRecX env' bndr bndr_rhs
1735 ; -- pprTrace "knownCon2" (ppr bs $$ ppr rhs $$ ppr (seIdSubst env'')) $
1736 simplExprF env'' rhs cont }
1739 bind_args env' _ [] _ = return env'
1741 bind_args env' dead_bndr (b:bs') (Type ty : args)
1742 = ASSERT( isTyVar b )
1743 bind_args (extendTvSubst env' b ty) dead_bndr bs' args
1745 bind_args env' dead_bndr (b:bs') (arg : args)
1747 do { let b' = if dead_bndr then b else zapOccInfo b
1748 -- Note that the binder might be "dead", because it doesn't
1749 -- occur in the RHS; and simplNonRecX may therefore discard
1750 -- it via postInlineUnconditionally.
1751 -- Nevertheless we must keep it if the case-binder is alive,
1752 -- because it may be used in the con_app. See Note [zapOccInfo]
1753 ; env'' <- simplNonRecX env' b' arg
1754 ; bind_args env'' dead_bndr bs' args }
1757 pprPanic "bind_args" $ ppr dc $$ ppr bs $$ ppr the_args $$
1758 text "scrut:" <+> ppr scrut
1762 %************************************************************************
1764 \subsection{Duplicating continuations}
1766 %************************************************************************
1769 prepareCaseCont :: SimplEnv
1770 -> [InAlt] -> SimplCont
1771 -> SimplM (SimplEnv, SimplCont,SimplCont)
1772 -- Return a duplicatable continuation, a non-duplicable part
1773 -- plus some extra bindings (that scope over the entire
1776 -- No need to make it duplicatable if there's only one alternative
1777 prepareCaseCont env [_] cont = return (env, cont, mkBoringStop)
1778 prepareCaseCont env _ cont = mkDupableCont env cont
1782 mkDupableCont :: SimplEnv -> SimplCont
1783 -> SimplM (SimplEnv, SimplCont, SimplCont)
1785 mkDupableCont env cont
1786 | contIsDupable cont
1787 = return (env, cont, mkBoringStop)
1789 mkDupableCont _ (Stop {}) = panic "mkDupableCont" -- Handled by previous eqn
1791 mkDupableCont env (CoerceIt ty cont)
1792 = do { (env', dup, nodup) <- mkDupableCont env cont
1793 ; return (env', CoerceIt ty dup, nodup) }
1795 mkDupableCont env cont@(StrictBind {})
1796 = return (env, mkBoringStop, cont)
1797 -- See Note [Duplicating strict continuations]
1799 mkDupableCont env cont@(StrictArg {})
1800 = return (env, mkBoringStop, cont)
1801 -- See Note [Duplicating strict continuations]
1803 mkDupableCont env (ApplyTo _ arg se cont)
1804 = -- e.g. [...hole...] (...arg...)
1806 -- let a = ...arg...
1807 -- in [...hole...] a
1808 do { (env', dup_cont, nodup_cont) <- mkDupableCont env cont
1809 ; arg' <- simplExpr (se `setInScope` env') arg
1810 ; (env'', arg'') <- makeTrivial env' arg'
1811 ; let app_cont = ApplyTo OkToDup arg'' (zapSubstEnv env') dup_cont
1812 ; return (env'', app_cont, nodup_cont) }
1814 mkDupableCont env cont@(Select _ _ [(_, bs, _rhs)] _ _)
1815 -- See Note [Single-alternative case]
1816 -- | not (exprIsDupable rhs && contIsDupable case_cont)
1817 -- | not (isDeadBinder case_bndr)
1818 | all isDeadBinder bs -- InIds
1819 = return (env, mkBoringStop, cont)
1821 mkDupableCont env (Select _ case_bndr alts se cont)
1822 = -- e.g. (case [...hole...] of { pi -> ei })
1824 -- let ji = \xij -> ei
1825 -- in case [...hole...] of { pi -> ji xij }
1826 do { tick (CaseOfCase case_bndr)
1827 ; (env', dup_cont, nodup_cont) <- mkDupableCont env cont
1828 -- NB: call mkDupableCont here, *not* prepareCaseCont
1829 -- We must make a duplicable continuation, whereas prepareCaseCont
1830 -- doesn't when there is a single case branch
1832 ; let alt_env = se `setInScope` env'
1833 ; (alt_env', case_bndr') <- simplBinder alt_env case_bndr
1834 ; alts' <- mapM (simplAlt alt_env' [] case_bndr' dup_cont) alts
1835 -- Safe to say that there are no handled-cons for the DEFAULT case
1836 -- NB: simplBinder does not zap deadness occ-info, so
1837 -- a dead case_bndr' will still advertise its deadness
1838 -- This is really important because in
1839 -- case e of b { (# p,q #) -> ... }
1840 -- b is always dead, and indeed we are not allowed to bind b to (# p,q #),
1841 -- which might happen if e was an explicit unboxed pair and b wasn't marked dead.
1842 -- In the new alts we build, we have the new case binder, so it must retain
1844 -- NB: we don't use alt_env further; it has the substEnv for
1845 -- the alternatives, and we don't want that
1847 ; (env'', alts'') <- mkDupableAlts env' case_bndr' alts'
1848 ; return (env'', -- Note [Duplicated env]
1849 Select OkToDup case_bndr' alts'' (zapSubstEnv env'') mkBoringStop,
1853 mkDupableAlts :: SimplEnv -> OutId -> [InAlt]
1854 -> SimplM (SimplEnv, [InAlt])
1855 -- Absorbs the continuation into the new alternatives
1857 mkDupableAlts env case_bndr' the_alts
1860 go env0 [] = return (env0, [])
1862 = do { (env1, alt') <- mkDupableAlt env0 case_bndr' alt
1863 ; (env2, alts') <- go env1 alts
1864 ; return (env2, alt' : alts' ) }
1866 mkDupableAlt :: SimplEnv -> OutId -> (AltCon, [CoreBndr], CoreExpr)
1867 -> SimplM (SimplEnv, (AltCon, [CoreBndr], CoreExpr))
1868 mkDupableAlt env case_bndr' (con, bndrs', rhs')
1869 | exprIsDupable rhs' -- Note [Small alternative rhs]
1870 = return (env, (con, bndrs', rhs'))
1872 = do { let rhs_ty' = exprType rhs'
1873 used_bndrs' = filter abstract_over (case_bndr' : bndrs')
1875 | isTyVar bndr = True -- Abstract over all type variables just in case
1876 | otherwise = not (isDeadBinder bndr)
1877 -- The deadness info on the new Ids is preserved by simplBinders
1879 ; (final_bndrs', final_args) -- Note [Join point abstraction]
1880 <- if (any isId used_bndrs')
1881 then return (used_bndrs', varsToCoreExprs used_bndrs')
1882 else do { rw_id <- newId (fsLit "w") realWorldStatePrimTy
1883 ; return ([rw_id], [Var realWorldPrimId]) }
1885 ; join_bndr <- newId (fsLit "$j") (mkPiTypes final_bndrs' rhs_ty')
1886 -- Note [Funky mkPiTypes]
1888 ; let -- We make the lambdas into one-shot-lambdas. The
1889 -- join point is sure to be applied at most once, and doing so
1890 -- prevents the body of the join point being floated out by
1891 -- the full laziness pass
1892 really_final_bndrs = map one_shot final_bndrs'
1893 one_shot v | isId v = setOneShotLambda v
1895 join_rhs = mkLams really_final_bndrs rhs'
1896 join_call = mkApps (Var join_bndr) final_args
1898 ; return (addNonRec env join_bndr join_rhs, (con, bndrs', join_call)) }
1899 -- See Note [Duplicated env]
1902 Note [Duplicated env]
1903 ~~~~~~~~~~~~~~~~~~~~~
1904 Some of the alternatives are simplified, but have not been turned into a join point
1905 So they *must* have an zapped subst-env. So we can't use completeNonRecX to
1906 bind the join point, because it might to do PostInlineUnconditionally, and
1907 we'd lose that when zapping the subst-env. We could have a per-alt subst-env,
1908 but zapping it (as we do in mkDupableCont, the Select case) is safe, and
1909 at worst delays the join-point inlining.
1911 Note [Small alterantive rhs]
1912 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1913 It is worth checking for a small RHS because otherwise we
1914 get extra let bindings that may cause an extra iteration of the simplifier to
1915 inline back in place. Quite often the rhs is just a variable or constructor.
1916 The Ord instance of Maybe in PrelMaybe.lhs, for example, took several extra
1917 iterations because the version with the let bindings looked big, and so wasn't
1918 inlined, but after the join points had been inlined it looked smaller, and so
1921 NB: we have to check the size of rhs', not rhs.
1922 Duplicating a small InAlt might invalidate occurrence information
1923 However, if it *is* dupable, we return the *un* simplified alternative,
1924 because otherwise we'd need to pair it up with an empty subst-env....
1925 but we only have one env shared between all the alts.
1926 (Remember we must zap the subst-env before re-simplifying something).
1927 Rather than do this we simply agree to re-simplify the original (small) thing later.
1929 Note [Funky mkPiTypes]
1930 ~~~~~~~~~~~~~~~~~~~~~~
1931 Notice the funky mkPiTypes. If the contructor has existentials
1932 it's possible that the join point will be abstracted over
1933 type varaibles as well as term variables.
1934 Example: Suppose we have
1935 data T = forall t. C [t]
1937 case (case e of ...) of
1939 We get the join point
1940 let j :: forall t. [t] -> ...
1941 j = /\t \xs::[t] -> rhs
1943 case (case e of ...) of
1944 C t xs::[t] -> j t xs
1946 Note [Join point abstaction]
1947 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1948 If we try to lift a primitive-typed something out
1949 for let-binding-purposes, we will *caseify* it (!),
1950 with potentially-disastrous strictness results. So
1951 instead we turn it into a function: \v -> e
1952 where v::State# RealWorld#. The value passed to this function
1953 is realworld#, which generates (almost) no code.
1955 There's a slight infelicity here: we pass the overall
1956 case_bndr to all the join points if it's used in *any* RHS,
1957 because we don't know its usage in each RHS separately
1959 We used to say "&& isUnLiftedType rhs_ty'" here, but now
1960 we make the join point into a function whenever used_bndrs'
1961 is empty. This makes the join-point more CPR friendly.
1962 Consider: let j = if .. then I# 3 else I# 4
1963 in case .. of { A -> j; B -> j; C -> ... }
1965 Now CPR doesn't w/w j because it's a thunk, so
1966 that means that the enclosing function can't w/w either,
1967 which is a lose. Here's the example that happened in practice:
1968 kgmod :: Int -> Int -> Int
1969 kgmod x y = if x > 0 && y < 0 || x < 0 && y > 0
1973 I have seen a case alternative like this:
1975 It's a bit silly to add the realWorld dummy arg in this case, making
1978 (the \v alone is enough to make CPR happy) but I think it's rare
1980 Note [Duplicating strict continuations]
1981 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1982 Do *not* duplicate StrictBind and StritArg continuations. We gain
1983 nothing by propagating them into the expressions, and we do lose a
1984 lot. Here's an example:
1985 && (case x of { T -> F; F -> T }) E
1986 Now, && is strict so we end up simplifying the case with
1987 an ArgOf continuation. If we let-bind it, we get
1989 let $j = \v -> && v E
1990 in simplExpr (case x of { T -> F; F -> T })
1992 And after simplifying more we get
1994 let $j = \v -> && v E
1995 in case x of { T -> $j F; F -> $j T }
1996 Which is a Very Bad Thing
1998 The desire not to duplicate is the entire reason that
1999 mkDupableCont returns a pair of continuations.
2001 The original plan had:
2002 e.g. (...strict-fn...) [...hole...]
2004 let $j = \a -> ...strict-fn...
2007 Note [Single-alternative cases]
2008 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2009 This case is just like the ArgOf case. Here's an example:
2013 case (case x of I# x' ->
2015 True -> I# (negate# x')
2016 False -> I# x') of y {
2018 Because the (case x) has only one alternative, we'll transform to
2020 case (case x' <# 0# of
2021 True -> I# (negate# x')
2022 False -> I# x') of y {
2024 But now we do *NOT* want to make a join point etc, giving
2026 let $j = \y -> MkT y
2028 True -> $j (I# (negate# x'))
2030 In this case the $j will inline again, but suppose there was a big
2031 strict computation enclosing the orginal call to MkT. Then, it won't
2032 "see" the MkT any more, because it's big and won't get duplicated.
2033 And, what is worse, nothing was gained by the case-of-case transform.
2035 When should use this case of mkDupableCont?
2036 However, matching on *any* single-alternative case is a *disaster*;
2037 e.g. case (case ....) of (a,b) -> (# a,b #)
2038 We must push the outer case into the inner one!
2041 * Match [(DEFAULT,_,_)], but in the common case of Int,
2042 the alternative-filling-in code turned the outer case into
2043 case (...) of y { I# _ -> MkT y }
2045 * Match on single alternative plus (not (isDeadBinder case_bndr))
2046 Rationale: pushing the case inwards won't eliminate the construction.
2047 But there's a risk of
2048 case (...) of y { (a,b) -> let z=(a,b) in ... }
2049 Now y looks dead, but it'll come alive again. Still, this
2050 seems like the best option at the moment.
2052 * Match on single alternative plus (all (isDeadBinder bndrs))
2053 Rationale: this is essentially seq.
2055 * Match when the rhs is *not* duplicable, and hence would lead to a
2056 join point. This catches the disaster-case above. We can test
2057 the *un-simplified* rhs, which is fine. It might get bigger or
2058 smaller after simplification; if it gets smaller, this case might
2059 fire next time round. NB also that we must test contIsDupable
2060 case_cont *btoo, because case_cont might be big!
2062 HOWEVER: I found that this version doesn't work well, because
2063 we can get let x = case (...) of { small } in ...case x...
2064 When x is inlined into its full context, we find that it was a bad
2065 idea to have pushed the outer case inside the (...) case.