Simplify.lhs 69.4 KB
Newer Older
1
%
2
% (c) The AQUA Project, Glasgow University, 1993-1998
3 4 5 6
%
\section[Simplify]{The main module of the simplifier}

\begin{code}
7
{-# OPTIONS -w #-}
8 9 10
-- The above warning supression flag is a temporary kludge.
-- While working on this module you are encouraged to remove it and fix
-- any warnings in the module. See
Ian Lynagh's avatar
Ian Lynagh committed
11
--     http://hackage.haskell.org/trac/ghc/wiki/Commentary/CodingStyle#Warnings
12 13
-- for details

14
module Simplify ( simplTopBinds, simplExpr ) where
15

16
#include "HsVersions.h"
17

simonpj@microsoft.com's avatar
Wibble  
simonpj@microsoft.com committed
18
import DynFlags
19
import SimplMonad
20
import Type hiding	( substTy, extendTvSubst )
21
import SimplEnv	
22 23
import SimplUtils
import Id
24
import Var
25 26
import IdInfo
import Coercion
27
import FamInstEnv	( topNormaliseType )
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
28
import DataCon		( dataConRepStrictness, dataConUnivTyVars )
29
import CoreSyn
30
import NewDemand	( isStrictDmd )
31
import PprCore		( pprParendExpr, pprCoreExpr )
32
import CoreUnfold	( mkUnfolding, callSiteInline )
33
import CoreUtils
34
import Rules		( lookupRule )
35
import BasicTypes	( isMarkedStrict )
36
import CostCentre	( currentCCS )
37
import TysPrim		( realWorldStatePrimTy )
38
import PrelInfo		( realWorldPrimId )
39
import BasicTypes	( TopLevelFlag(..), isTopLevel, 
40
			  RecFlag(..), isNonRuleLoopBreaker )
41
import Maybes		( orElse )
42
import Outputable
43
import Util
44 45 46
\end{code}


47 48
The guts of the simplifier is in this module, but the driver loop for
the simplifier is in SimplCore.lhs.
49 50


51 52 53 54 55 56 57 58
-----------------------------------------
	*** IMPORTANT NOTE ***
-----------------------------------------
The simplifier used to guarantee that the output had no shadowing, but
it does not do so any more.   (Actually, it never did!)  The reason is
documented with simplifyArgs.


59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
-----------------------------------------
	*** IMPORTANT NOTE ***
-----------------------------------------
Many parts of the simplifier return a bunch of "floats" as well as an
expression. This is wrapped as a datatype SimplUtils.FloatsWith.

All "floats" are let-binds, not case-binds, but some non-rec lets may
be unlifted (with RHS ok-for-speculation).



-----------------------------------------
	ORGANISATION OF FUNCTIONS
-----------------------------------------
simplTopBinds
  - simplify all top-level binders
  - for NonRec, call simplRecOrTopPair
  - for Rec,    call simplRecBind

	
	------------------------------
simplExpr (applied lambda)	==> simplNonRecBind
simplExpr (Let (NonRec ...) ..) ==> simplNonRecBind
simplExpr (Let (Rec ...)    ..) ==> simplify binders; simplRecBind

	------------------------------
simplRecBind	[binders already simplfied]
  - use simplRecOrTopPair on each pair in turn

simplRecOrTopPair [binder already simplified]
  Used for: recursive bindings (top level and nested)
	    top-level non-recursive bindings
  Returns: 
  - check for PreInlineUnconditionally
  - simplLazyBind

simplNonRecBind
  Used for: non-top-level non-recursive bindings
	    beta reductions (which amount to the same thing)
  Because it can deal with strict arts, it takes a 
	"thing-inside" and returns an expression

  - check for PreInlineUnconditionally
  - simplify binder, including its IdInfo
  - if strict binding
	simplStrictArg
	mkAtomicArgs
	completeNonRecX
    else
	simplLazyBind
	addFloats

simplNonRecX:	[given a *simplified* RHS, but an *unsimplified* binder]
  Used for: binding case-binder and constr args in a known-constructor case
  - check for PreInLineUnconditionally
  - simplify binder
  - completeNonRecX
 
	------------------------------
simplLazyBind:	[binder already simplified, RHS not]
  Used for: recursive bindings (top level and nested)
	    top-level non-recursive bindings
	    non-top-level, but *lazy* non-recursive bindings
	[must not be strict or unboxed]
  Returns floats + an augmented environment, not an expression
  - substituteIdInfo and add result to in-scope 
	[so that rules are available in rec rhs]
  - simplify rhs
  - mkAtomicArgs
  - float if exposes constructor or PAP
129
  - completeBind
130 131 132 133 134 135


completeNonRecX:	[binder and rhs both simplified]
  - if the the thing needs case binding (unlifted and not ok-for-spec)
	build a Case
   else
136
	completeBind
137 138
	addFloats

139
completeBind: 	[given a simplified RHS]
140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
	[used for both rec and non-rec bindings, top level and not]
  - try PostInlineUnconditionally
  - add unfolding [this is the only place we add an unfolding]
  - add arity



Right hand sides and arguments
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In many ways we want to treat 
	(a) the right hand side of a let(rec), and 
	(b) a function argument
in the same way.  But not always!  In particular, we would
like to leave these arguments exactly as they are, so they
will match a RULE more easily.
	
	f (g x, h x)	
	g (+ x)

It's harder to make the rule match if we ANF-ise the constructor,
or eta-expand the PAP:

	f (let { a = g x; b = h x } in (a,b))
	g (\y. + x y)

On the other hand if we see the let-defns

	p = (g x, h x)
	q = + x

then we *do* want to ANF-ise and eta-expand, so that p and q
can be safely inlined.   

Even floating lets out is a bit dubious.  For let RHS's we float lets
out if that exposes a value, so that the value can be inlined more vigorously.
For example

	r = let x = e in (x,x)

Here, if we float the let out we'll expose a nice constructor. We did experiments
that showed this to be a generally good thing.  But it was a bad thing to float
lets out unconditionally, because that meant they got allocated more often.

For function arguments, there's less reason to expose a constructor (it won't
get inlined).  Just possibly it might make a rule match, but I'm pretty skeptical.
So for the moment we don't float lets out of function arguments either.


Eta expansion
~~~~~~~~~~~~~~
For eta expansion, we want to catch things like

	case e of (a,b) -> \x -> case a of (p,q) -> \y -> r

If the \x was on the RHS of a let, we'd eta expand to bring the two
lambdas together.  And in general that's a good thing to do.  Perhaps
we should eta expand wherever we find a (value) lambda?  Then the eta
expansion at a let RHS can concentrate solely on the PAP case.
198 199


200 201 202 203 204 205 206
%************************************************************************
%*									*
\subsection{Bindings}
%*									*
%************************************************************************

\begin{code}
207
simplTopBinds :: SimplEnv -> [InBind] -> SimplM [OutBind]
208

209
simplTopBinds env binds
210 211 212 213 214 215
  = do	{ 	-- Put all the top-level binders into scope at the start
		-- so that if a transformation rule has unexpectedly brought
		-- anything into scope, then we don't get a complaint about that.
		-- It's rather as if the top-level binders were imported.
	; env <- simplRecBndrs env (bindersOfBinds binds)
	; dflags <- getDOptsSmpl
216 217
	; let dump_flag = dopt Opt_D_dump_inlinings dflags || 
			  dopt Opt_D_dump_rule_firings dflags
218 219 220
	; env' <- simpl_binds dump_flag env binds
	; freeTick SimplifierDone
	; return (getFloats env') }
221
  where
222 223
	-- We need to track the zapped top-level binders, because
	-- they should have their fragile IdInfo zapped (notably occurrence info)
224
	-- That's why we run down binds and bndrs' simultaneously.
225 226 227
	--
	-- The dump-flag emits a trace for each top-level binding, which
	-- helps to locate the tracing for inlining and rule firing
228 229 230 231 232
    simpl_binds :: Bool -> SimplEnv -> [InBind] -> SimplM SimplEnv
    simpl_binds dump env []	      = return env
    simpl_binds dump env (bind:binds) = do { env' <- trace dump bind $
						     simpl_bind env bind
					   ; simpl_binds dump env' binds }
233

234 235
    trace True  bind = pprTrace "SimplBind" (ppr (bindersOf bind))
    trace False bind = \x -> x
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
236

237 238
    simpl_bind env (NonRec b r) = simplRecOrTopPair env TopLevel b r
    simpl_bind env (Rec pairs)  = simplRecBind      env TopLevel pairs
239 240 241 242 243 244 245 246 247 248 249 250 251 252
\end{code}


%************************************************************************
%*									*
\subsection{Lazy bindings}
%*									*
%************************************************************************

simplRecBind is used for
	* recursive bindings only

\begin{code}
simplRecBind :: SimplEnv -> TopLevelFlag
253 254 255 256 257 258 259
	     -> [(InId, InExpr)]
	     -> SimplM SimplEnv
simplRecBind env top_lvl pairs
  = do	{ env' <- go (zapFloats env) pairs
	; return (env `addRecFloats` env') }
	-- addFloats adds the floats from env', 
	-- *and* updates env with the in-scope set from env'
260
  where
261
    go env [] = return env
262
	
263 264 265
    go env ((bndr, rhs) : pairs)
	= do { env <- simplRecOrTopPair env top_lvl bndr rhs
	     ; go env pairs }
266 267
\end{code}

268
simplOrTopPair is used for
269 270 271 272 273 274 275 276
	* recursive bindings (whether top level or not)
	* top-level non-recursive bindings

It assumes the binder has already been simplified, but not its IdInfo.

\begin{code}
simplRecOrTopPair :: SimplEnv
	     	  -> TopLevelFlag
277 278
	     	  -> InId -> InExpr	-- Binder and rhs
	     	  -> SimplM SimplEnv	-- Returns an env that includes the binding
279

280
simplRecOrTopPair env top_lvl bndr rhs
281
  | preInlineUnconditionally env top_lvl bndr rhs  	-- Check for unconditional inline
282 283
  = do	{ tick (PreInlineUnconditionally bndr)
	; return (extendIdSubst env bndr (mkContEx env rhs)) }
284 285

  | otherwise
286 287 288
  = do	{ let bndr' = lookupRecBndr env bndr
	      (env', bndr'') = addLetIdInfo env bndr bndr'
	; simplLazyBind env' top_lvl Recursive bndr bndr'' rhs env' }
289 290 291 292 293
	-- May not actually be recursive, but it doesn't matter
\end{code}


simplLazyBind is used for
294 295 296
  * [simplRecOrTopPair] recursive bindings (whether top level or not)
  * [simplRecOrTopPair] top-level non-recursive bindings
  * [simplNonRecE]	non-top-level *lazy* non-recursive bindings
297 298 299

Nota bene:
    1. It assumes that the binder is *already* simplified, 
300
       and is in scope, and its IdInfo too, except unfolding
301 302 303 304 305 306 307 308 309 310

    2. It assumes that the binder type is lifted.

    3. It does not check for pre-inline-unconditionallly;
       that should have been done already.

\begin{code}
simplLazyBind :: SimplEnv
	      -> TopLevelFlag -> RecFlag
	      -> InId -> OutId		-- Binder, both pre-and post simpl
311
					-- The OutId has IdInfo, except arity, unfolding
312
	      -> InExpr -> SimplEnv 	-- The RHS and its environment
313
	      -> SimplM SimplEnv
314

315
simplLazyBind env top_lvl is_rec bndr bndr1 rhs rhs_se
316 317 318 319 320
  = do	{ let	rhs_env     = rhs_se `setInScope` env
		(tvs, body) = collectTyBinders rhs
	; (body_env, tvs') <- simplBinders rhs_env tvs
		-- See Note [Floating and type abstraction]
		-- in SimplUtils
321

322
  	-- Simplify the RHS; note the mkRhsStop, which tells 
323
	-- the simplifier that this is the RHS of a let.
324 325 326
	; let rhs_cont = mkRhsStop (applyTys (idType bndr1) (mkTyVarTys tvs'))
	; (body_env1, body1) <- simplExprF body_env body rhs_cont

327
	-- ANF-ise a constructor or PAP rhs
328
	; (body_env2, body2) <- prepareRhs body_env1 body1
329

330 331 332 333 334
	; (env', rhs')
	    <-	if not (doFloatFromRhs top_lvl is_rec False body2 body_env2)
		then				-- No floating, just wrap up!
		     do	{ rhs' <- mkLam tvs' (wrapFloats body_env2 body2)
			; return (env, rhs') }
335

336 337 338 339 340 341
		else if null tvs then 		-- Simple floating
		     do	{ tick LetFloatFromLet
			; return (addFloats env body_env2, body2) }

		else  				-- Do type-abstraction first
		     do	{ tick LetFloatFromLet
342
			; (poly_binds, body3) <- abstractFloats tvs' body_env2 body2
343 344 345 346 347
			; rhs' <- mkLam tvs' body3
			; return (extendFloats env poly_binds, rhs') }

	; completeBind env' top_lvl bndr bndr1 rhs' }
\end{code}
348 349 350 351 352 353 354 355 356 357 358 359 360

A specialised variant of simplNonRec used when the RHS is already simplified, 
notably in knownCon.  It uses case-binding where necessary.

\begin{code}
simplNonRecX :: SimplEnv
	     -> InId 		-- Old binder
	     -> OutExpr		-- Simplified RHS
	     -> SimplM SimplEnv

simplNonRecX env bndr new_rhs
  = do	{ (env, bndr') <- simplBinder env bndr
	; completeNonRecX env NotTopLevel NonRecursive
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
361
			  (isStrictId bndr) bndr bndr' new_rhs }
362 363 364 365 366 367 368 369 370 371

completeNonRecX :: SimplEnv
		-> TopLevelFlag -> RecFlag -> Bool
	        -> InId 		-- Old binder
		-> OutId		-- New binder
	     	-> OutExpr		-- Simplified RHS
	     	-> SimplM SimplEnv

completeNonRecX env top_lvl is_rec is_strict old_bndr new_bndr new_rhs
  = do 	{ (env1, rhs1) <- prepareRhs (zapFloats env) new_rhs
372 373 374 375 376
	; (env2, rhs2) <- 
		if doFloatFromRhs top_lvl is_rec is_strict rhs1 env1
		then do	{ tick LetFloatFromLet
			; return (addFloats env env1, rhs1) }	-- Add the floats to the main env
		else return (env, wrapFloats env1 rhs1)		-- Wrap the floats around the RHS
377 378 379 380 381 382 383 384 385 386 387
	; completeBind env2 NotTopLevel old_bndr new_bndr rhs2 }
\end{code}

{- No, no, no!  Do not try preInlineUnconditionally in completeNonRecX
   Doing so risks exponential behaviour, because new_rhs has been simplified once already
   In the cases described by the folowing commment, postInlineUnconditionally will 
   catch many of the relevant cases.
  	-- This happens; for example, the case_bndr during case of
	-- known constructor:  case (a,b) of x { (p,q) -> ... }
	-- Here x isn't mentioned in the RHS, so we don't want to
	-- create the (dead) let-binding  let x = (a,b) in ...
388
	--
389 390 391
	-- Similarly, single occurrences can be inlined vigourously
	-- e.g.  case (f x, g y) of (a,b) -> ....
	-- If a,b occur once we can avoid constructing the let binding for them.
392

393 394 395 396 397 398 399
   Furthermore in the case-binding case preInlineUnconditionally risks extra thunks
	-- Consider 	case I# (quotInt# x y) of 
	--		  I# v -> let w = J# v in ...
	-- If we gaily inline (quotInt# x y) for v, we end up building an
	-- extra thunk:
	--		  let w = J# (quotInt# x y) in ...
	-- because quotInt# can fail.
400

401 402 403 404
  | preInlineUnconditionally env NotTopLevel bndr new_rhs
  = thing_inside (extendIdSubst env bndr (DoneEx new_rhs))
-}

405
----------------------------------
406 407 408 409 410 411 412 413 414
prepareRhs takes a putative RHS, checks whether it's a PAP or
constructor application and, if so, converts it to ANF, so that the 
resulting thing can be inlined more easily.  Thus
	x = (f a, g b)
becomes
	t1 = f a
	t2 = g b
	x = (t1,t2)

415 416 417 418 419 420
We also want to deal well cases like this
	v = (f e1 `cast` co) e2
Here we want to make e1,e2 trivial and get
	x1 = e1; x2 = e2; v = (f x1 `cast` co) v2
That's what the 'go' loop in prepareRhs does

421 422 423
\begin{code}
prepareRhs :: SimplEnv -> OutExpr -> SimplM (SimplEnv, OutExpr)
-- Adds new floats to the env iff that allows us to return a good RHS
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
424
prepareRhs env (Cast rhs co)	-- Note [Float coercions]
425 426 427 428
  = do	{ (env', rhs') <- makeTrivial env rhs
	; return (env', Cast rhs' co) }

prepareRhs env rhs
429 430
  = do	{ (is_val, env', rhs') <- go 0 env rhs 
	; return (env', rhs') }
431
  where
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
    go n_val_args env (Cast rhs co)
	= do { (is_val, env', rhs') <- go n_val_args env rhs
	     ; return (is_val, env', Cast rhs' co) }
    go n_val_args env (App fun (Type ty))
	= do { (is_val, env', rhs') <- go n_val_args env fun
	     ; return (is_val, env', App rhs' (Type ty)) }
    go n_val_args env (App fun arg)
	= do { (is_val, env', fun') <- go (n_val_args+1) env fun
	     ; case is_val of
		True -> do { (env'', arg') <- makeTrivial env' arg
			   ; return (True, env'', App fun' arg') }
		False -> return (False, env, App fun arg) }
    go n_val_args env (Var fun)
	= return (is_val, env, Var fun)
	where
	  is_val = n_val_args > 0	-- There is at least one arg
					-- ...and the fun a constructor or PAP
		 && (isDataConWorkId fun || n_val_args < idArity fun)
    go n_val_args env other
	= return (False, env, other)
452 453
\end{code}

454

455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488
Note [Float coercions]
~~~~~~~~~~~~~~~~~~~~~~
When we find the binding
	x = e `cast` co
we'd like to transform it to
	x' = e
	x = x `cast` co		-- A trivial binding
There's a chance that e will be a constructor application or function, or something
like that, so moving the coerion to the usage site may well cancel the coersions
and lead to further optimisation.  Example:

     data family T a :: *
     data instance T Int = T Int

     foo :: Int -> Int -> Int
     foo m n = ...
        where
          x = T m
          go 0 = 0
          go n = case x of { T m -> go (n-m) }
		-- This case should optimise


\begin{code}
makeTrivial :: SimplEnv -> OutExpr -> SimplM (SimplEnv, OutExpr)
-- Binds the expression to a variable, if it's not trivial, returning the variable
makeTrivial env expr
  | exprIsTrivial expr
  = return (env, expr)
  | otherwise		-- See Note [Take care] below
  = do 	{ var <- newId FSLIT("a") (exprType expr)
	; env <- completeNonRecX env NotTopLevel NonRecursive 
				 False var var expr
	; return (env, substExpr env (Var var)) }
489
\end{code}
490 491


492 493 494 495 496 497
%************************************************************************
%*									*
\subsection{Completing a lazy binding}
%*									*
%************************************************************************

498 499 500 501 502
completeBind
  * deals only with Ids, not TyVars
  * takes an already-simplified binder and RHS
  * is used for both recursive and non-recursive bindings
  * is used for both top-level and non-top-level bindings
503 504 505 506 507 508 509 510

It does the following:
  - tries discarding a dead binding
  - tries PostInlineUnconditionally
  - add unfolding [this is the only place we add an unfolding]
  - add arity

It does *not* attempt to do let-to-case.  Why?  Because it is used for
511 512
  - top-level bindings (when let-to-case is impossible) 
  - many situations where the "rhs" is known to be a WHNF
513 514
		(so let-to-case is inappropriate).

515 516
Nor does it do the atomic-argument thing

517
\begin{code}
518 519 520 521 522 523 524 525 526 527
completeBind :: SimplEnv
	     -> TopLevelFlag		-- Flag stuck into unfolding
	     -> InId 			-- Old binder
	     -> OutId -> OutExpr	-- New binder and RHS
	     -> SimplM SimplEnv
-- completeBind may choose to do its work 
--	* by extending the substitution (e.g. let x = y in ...)
--	* or by adding to the floats in the envt

completeBind env top_lvl old_bndr new_bndr new_rhs
528
  | postInlineUnconditionally env top_lvl new_bndr occ_info new_rhs unfolding
529 530 531 532 533 534
		-- Inline and discard the binding
  = do	{ tick (PostInlineUnconditionally old_bndr)
	; -- pprTrace "postInlineUnconditionally" (ppr old_bndr <+> ppr new_bndr <+> ppr new_rhs) $
	  return (extendIdSubst env old_bndr (DoneEx new_rhs)) }
	-- Use the substitution to make quite, quite sure that the
	-- substitution will happen, since we are going to discard the binding
535 536 537

  |  otherwise
  = let
538
	-- 	Arity info
539 540
  	new_bndr_info = idInfo new_bndr `setArityInfo` exprArity new_rhs

541
	-- 	Unfolding info
542 543 544 545 546
	-- Add the unfolding *only* for non-loop-breakers
	-- Making loop breakers not have an unfolding at all 
	-- means that we can avoid tests in exprIsConApp, for example.
	-- This is important: if exprIsConApp says 'yes' for a recursive
	-- thing, then we can get into an infinite loop
547 548

	-- 	Demand info
549 550 551 552 553 554 555 556 557 558
	-- If the unfolding is a value, the demand info may
	-- go pear-shaped, so we nuke it.  Example:
	--	let x = (a,b) in
	--	case x of (p,q) -> h p q x
	-- Here x is certainly demanded. But after we've nuked
	-- the case, we'll get just
	--	let x = (a,b) in h a b x
	-- and now x is not demanded (I'm assuming h is lazy)
	-- This really happens.  Similarly
	--	let f = \x -> e in ...f..f...
559
	-- After inlining f at some of its call sites the original binding may
560 561 562 563 564 565 566 567
	-- (for example) be no longer strictly demanded.
	-- The solution here is a bit ad hoc...
 	info_w_unf = new_bndr_info `setUnfoldingInfo` unfolding
        final_info | loop_breaker		= new_bndr_info
		   | isEvaldUnfolding unfolding = zapDemandInfo info_w_unf `orElse` info_w_unf
		   | otherwise			= info_w_unf

	final_id = new_bndr `setIdInfo` final_info
568 569 570 571
    in
		-- These seqs forces the Id, and hence its IdInfo,
		-- and hence any inner substitutions
    final_id					`seq`
572
    -- pprTrace "Binding" (ppr final_id <+> ppr unfolding) $
573
    return (addNonRec env final_id new_rhs)
574
  where 
575
    unfolding    = mkUnfolding (isTopLevel top_lvl) new_rhs
576
    loop_breaker = isNonRuleLoopBreaker occ_info
577 578
    old_info     = idInfo old_bndr
    occ_info     = occInfo old_info
SamB's avatar
SamB committed
579
\end{code}
580 581 582



583 584 585 586 587 588
%************************************************************************
%*									*
\subsection[Simplify-simplExpr]{The main function: simplExpr}
%*									*
%************************************************************************

589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626
The reason for this OutExprStuff stuff is that we want to float *after*
simplifying a RHS, not before.  If we do so naively we get quadratic
behaviour as things float out.

To see why it's important to do it after, consider this (real) example:

	let t = f x
	in fst t
==>
	let t = let a = e1
		    b = e2
		in (a,b)
	in fst t
==>
	let a = e1
	    b = e2
	    t = (a,b)
	in
	a	-- Can't inline a this round, cos it appears twice
==>
	e1

Each of the ==> steps is a round of simplification.  We'd save a
whole round if we float first.  This can cascade.  Consider

	let f = g d
	in \x -> ...f...
==>
	let f = let d1 = ..d.. in \y -> e
	in \x -> ...f...
==>
	let d1 = ..d..
	in \x -> ...(\y ->e)...

Only in this second round can the \y be applied, and it 
might do the same again.


627
\begin{code}
628
simplExpr :: SimplEnv -> CoreExpr -> SimplM CoreExpr
629
simplExpr env expr = simplExprC env expr (mkBoringStop expr_ty')
630
		   where
631
		     expr_ty' = substTy env (exprType expr)
632
	-- The type in the Stop continuation, expr_ty', is usually not used
633
	-- It's only needed when discarding continuations after finding
634 635
	-- a function that returns bottom.
	-- Hence the lazy substitution
636

637

638 639 640
simplExprC :: SimplEnv -> CoreExpr -> SimplCont -> SimplM CoreExpr
	-- Simplify an expression, given a continuation
simplExprC env expr cont 
641 642 643 644 645 646 647 648 649 650 651
  = -- pprTrace "simplExprC" (ppr expr $$ ppr cont {- $$ ppr (seIdSubst env) -} $$ ppr (seFloats env) ) $
    do	{ (env', expr') <- simplExprF (zapFloats env) expr cont
	; -- pprTrace "simplExprC ret" (ppr expr $$ ppr expr') $
	  -- pprTrace "simplExprC ret3" (ppr (seInScope env')) $
	  -- pprTrace "simplExprC ret4" (ppr (seFloats env')) $
          return (wrapFloats env' expr') }

--------------------------------------------------
simplExprF :: SimplEnv -> InExpr -> SimplCont
	   -> SimplM (SimplEnv, OutExpr)

652 653 654
simplExprF env e cont 
  = -- pprTrace "simplExprF" (ppr e $$ ppr cont $$ ppr (seTvSubst env) $$ ppr (seIdSubst env) {- $$ ppr (seFloats env) -} ) $
    simplExprF' env e cont
655
    				     
656
simplExprF' env (Var v)	       cont = simplVar env v cont
657 658 659 660
simplExprF' env (Lit lit)      cont = rebuild env (Lit lit) cont
simplExprF' env (Note n expr)  cont = simplNote env n expr cont
simplExprF' env (Cast body co) cont = simplCast env body co cont
simplExprF' env (App fun arg)  cont = simplExprF env fun $
661
				      ApplyTo NoDup arg env cont
662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679

simplExprF' env expr@(Lam _ _) cont 
  = simplLam env (map zap bndrs) body cont
	-- The main issue here is under-saturated lambdas
	--   (\x1. \x2. e) arg1
	-- Here x1 might have "occurs-once" occ-info, because occ-info
	-- is computed assuming that a group of lambdas is applied
	-- all at once.  If there are too few args, we must zap the 
	-- occ-info.
  where
    n_args   = countArgs cont
    n_params = length bndrs
    (bndrs, body) = collectBinders expr
    zap | n_args >= n_params = \b -> b	
	| otherwise	     = \b -> if isTyVar b then b
				     else zapLamIdInfo b
	-- NB: we count all the args incl type args
	-- so we must count all the binders (incl type lambdas)
680

681
simplExprF' env (Type ty) cont
682
  = ASSERT( contIsRhsOrArg cont )
683 684
    do	{ ty' <- simplType env ty
	; rebuild env (Type ty') cont }
685

686
simplExprF' env (Case scrut bndr case_ty alts) cont
687 688 689
  | not (switchIsOn (getSwitchChecker env) NoCaseOfCase)
  = 	-- Simplify the scrutinee with a Select continuation
    simplExprF env scrut (Select NoDup bndr alts env cont)
690

691 692
  | otherwise
  = 	-- If case-of-case is off, simply simplify the case expression
693
	-- in a vanilla Stop context, and rebuild the result around it
694 695
    do	{ case_expr' <- simplExprC env scrut case_cont
	; rebuild env case_expr' cont }
696
  where
697
    case_cont = Select NoDup bndr alts env (mkBoringStop case_ty')
698
    case_ty'  = substTy env case_ty	-- c.f. defn of simplExpr
699

700 701 702 703
simplExprF' env (Let (Rec pairs) body) cont
  = do	{ env <- simplRecBndrs env (map fst pairs)
		-- NB: bndrs' don't have unfoldings or rules
		-- We add them as we go down
704

705 706
	; env <- simplRecBind env NotTopLevel pairs
	; simplExprF env body cont }
707

708 709
simplExprF' env (Let (NonRec bndr rhs) body) cont
  = simplNonRecE env bndr (rhs, env) ([], body) cont
710 711

---------------------------------
712 713 714
simplType :: SimplEnv -> InType -> SimplM OutType
	-- Kept monadic just so we can do the seqType
simplType env ty
715 716
  = -- pprTrace "simplType" (ppr ty $$ ppr (seTvSubst env)) $
    seqType new_ty   `seq`   returnSmpl new_ty
717
  where
718
    new_ty = substTy env ty
719 720 721
\end{code}


722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745
%************************************************************************
%*									*
\subsection{The main rebuilder}
%*									*
%************************************************************************

\begin{code}
rebuild :: SimplEnv -> OutExpr -> SimplCont -> SimplM (SimplEnv, OutExpr)
-- At this point the substitution in the SimplEnv should be irrelevant
-- only the in-scope set and floats should matter
rebuild env expr cont
  = -- pprTrace "rebuild" (ppr expr $$ ppr cont $$ ppr (seFloats env)) $
    case cont of
      Stop {}		      	   -> return (env, expr)
      CoerceIt co cont	      	   -> rebuild env (mkCoerce co expr) cont
      Select _ bndr alts se cont   -> rebuildCase (se `setFloats` env) expr bndr alts cont
      StrictArg fun ty info cont   -> rebuildCall env (fun `App` expr) (funResultTy ty) info cont
      StrictBind b bs body se cont -> do { env' <- simplNonRecX (se `setFloats` env) b expr
					 ; simplLam env' bs body cont }
      ApplyTo _ arg se cont	   -> do { arg' <- simplExpr (se `setInScope` env) arg
				         ; rebuild env (App expr arg') cont }
\end{code}


746 747 748 749 750 751 752
%************************************************************************
%*									*
\subsection{Lambdas}
%*									*
%************************************************************************

\begin{code}
753 754
simplCast :: SimplEnv -> InExpr -> Coercion -> SimplCont
	  -> SimplM (SimplEnv, OutExpr)
755
simplCast env body co cont
756 757 758
  = do	{ co' <- simplType env co
	; simplExprF env body (addCoerce co' cont) }
  where
759 760
       addCoerce co cont = add_coerce co (coercionKind co) cont

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
761 762 763
       add_coerce co (s1, k1) cont 	-- co :: ty~ty
         | s1 `coreEqType` k1 = cont	-- is a no-op

764 765
       add_coerce co1 (s1, k2) (CoerceIt co2 cont)
         | (l1, t1) <- coercionKind co2
766 767 768 769 770 771 772 773 774 775 776 777
                -- 	coerce T1 S1 (coerce S1 K1 e)
		-- ==>
		--	e, 			if T1=K1
		--	coerce T1 K1 e,		otherwise
		--
		-- For example, in the initial form of a worker
		-- we may find 	(coerce T (coerce S (\x.e))) y
		-- and we'd like it to simplify to e[y/x] in one round 
		-- of simplification
         , s1 `coreEqType` t1  = cont		 -- The coerces cancel out  
         | otherwise           = CoerceIt (mkTransCoercion co1 co2) cont
    
778 779 780 781 782 783 784 785 786 787 788
       add_coerce co (s1s2, t1t2) (ApplyTo dup (Type arg_ty) arg_se cont)
		-- (f `cast` g) ty  --->   (f ty) `cast` (g @ ty)
		-- This implements the PushT rule from the paper
	 | Just (tyvar,_) <- splitForAllTy_maybe s1s2
	 , not (isCoVar tyvar)
	 = ApplyTo dup (Type ty') (zapSubstEnv env) (addCoerce (mkInstCoercion co ty') cont)
	 where
	   ty' = substTy arg_se arg_ty

	-- ToDo: the PushC rule is not implemented at all

789
       add_coerce co (s1s2, t1t2) (ApplyTo dup arg arg_se cont)
790
         | not (isTypeArg arg)  -- This implements the Push rule from the paper
791
         , isFunTy s1s2	  -- t1t2 must be a function type, becuase it's applied
792 793 794 795 796 797 798 799 800 801 802 803
                -- co : s1s2 :=: t1t2
		--	(coerce (T1->T2) (S1->S2) F) E
		-- ===> 
		--	coerce T2 S2 (F (coerce S1 T1 E))
		--
		-- t1t2 must be a function type, T1->T2, because it's applied
		-- to something but s1s2 might conceivably not be
		--
		-- When we build the ApplyTo we can't mix the out-types
		-- with the InExpr in the argument, so we simply substitute
		-- to make it all consistent.  It's a bit messy.
		-- But it isn't a common case.
804 805
		--
		-- Example of use: Trac #995
806
         = ApplyTo dup new_arg (zapSubstEnv env) (addCoerce co2 cont)
807 808 809 810 811
         where
           -- we split coercion t1->t2 :=: s1->s2 into t1 :=: s1 and 
           -- t2 :=: s2 with left and right on the curried form: 
           --    (->) t1 t2 :=: (->) s1 s2
           [co1, co2] = decomposeCo 2 co
812
           new_arg    = mkCoerce (mkSymCoercion co1) arg'
813 814
	   arg'       = substExpr arg_se arg

815
       add_coerce co _ cont = CoerceIt co cont
816 817
\end{code}

818

819 820 821 822 823
%************************************************************************
%*									*
\subsection{Lambdas}
%*									*
%************************************************************************
824 825

\begin{code}
826 827 828 829
simplLam :: SimplEnv -> [InId] -> InExpr -> SimplCont
	 -> SimplM (SimplEnv, OutExpr)

simplLam env [] body cont = simplExprF env body cont
830 831

      	-- Type-beta reduction
832 833 834 835 836
simplLam env (bndr:bndrs) body (ApplyTo _ (Type ty_arg) arg_se cont)
  = ASSERT( isTyVar bndr )
    do	{ tick (BetaReduction bndr)
	; ty_arg' <- simplType (arg_se `setInScope` env) ty_arg
	; simplLam (extendTvSubst env bndr ty_arg') bndrs body cont }
837 838

	-- Ordinary beta reduction
839 840 841
simplLam env (bndr:bndrs) body (ApplyTo _ arg arg_se cont)
  = do	{ tick (BetaReduction bndr)	
	; simplNonRecE env bndr (arg, arg_se) (bndrs, body) cont }
842

843
	-- Not enough args, so there are real lambdas left to put in the result
844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
simplLam env bndrs body cont
  = do	{ (env, bndrs') <- simplLamBndrs env bndrs
	; body' <- simplExpr env body
	; new_lam <- mkLam bndrs' body'
	; rebuild env new_lam cont }

------------------
simplNonRecE :: SimplEnv 
	     -> InId 			-- The binder
	     -> (InExpr, SimplEnv)	-- Rhs of binding (or arg of lambda)
	     -> ([InId], InExpr)	-- Body of the let/lambda
					--	\xs.e
	     -> SimplCont
	     -> SimplM (SimplEnv, OutExpr)

-- simplNonRecE is used for
--  * non-top-level non-recursive lets in expressions
--  * beta reduction
--
-- It deals with strict bindings, via the StrictBind continuation,
-- which may abort the whole process
--
-- The "body" of the binding comes as a pair of ([InId],InExpr)
-- representing a lambda; so we recurse back to simplLam
-- Why?  Because of the binder-occ-info-zapping done before 
-- 	 the call to simplLam in simplExprF (Lam ...)

simplNonRecE env bndr (rhs, rhs_se) (bndrs, body) cont
  | preInlineUnconditionally env NotTopLevel bndr rhs
  = do	{ tick (PreInlineUnconditionally bndr)
	; simplLam (extendIdSubst env bndr (mkContEx rhs_se rhs)) bndrs body cont }

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
876
  | isStrictId bndr
877 878 879 880
  = do	{ simplExprF (rhs_se `setFloats` env) rhs 
		     (StrictBind bndr bndrs body env cont) }

  | otherwise
881
  = do	{ (env, bndr') <- simplNonRecBndr env bndr
882 883
	; env <- simplLazyBind env NotTopLevel NonRecursive bndr bndr' rhs rhs_se
	; simplLam env bndrs body cont }
884 885
\end{code}

886

887 888 889 890 891 892
%************************************************************************
%*									*
\subsection{Notes}
%*									*
%************************************************************************

sof's avatar
sof committed
893
\begin{code}
894 895
-- Hack alert: we only distinguish subsumed cost centre stacks for the 
-- purposes of inlining.  All other CCCSs are mapped to currentCCS.
896
simplNote env (SCC cc) e cont
897 898
  = do 	{ e' <- simplExpr (setEnclosingCC env currentCCS) e
	; rebuild env (mkSCC cc e') cont }
899 900 901

-- See notes with SimplMonad.inlineMode
simplNote env InlineMe e cont
902
  | Just (inside, outside) <- splitInlineCont cont  -- Boring boring continuation; see notes above
903
  = do	{ 			-- Don't inline inside an INLINE expression
904 905
	  e' <- simplExprC (setMode inlineMode env) e inside
	; rebuild env (mkInlineMe e') outside }
906 907 908 909

  | otherwise  	-- Dissolve the InlineMe note if there's
		-- an interesting context of any kind to combine with
		-- (even a type application -- anything except Stop)
910
  = simplExprF env e cont
911 912

simplNote env (CoreNote s) e cont
913 914
  = simplExpr env e    `thenSmpl` \ e' ->
    rebuild env (Note (CoreNote s) e') cont
915 916 917
\end{code}


918 919
%************************************************************************
%*									*
920
\subsection{Dealing with calls}
921 922
%*									*
%************************************************************************
923

924
\begin{code}
925
simplVar env var cont
926 927 928
  = case substId env var of
	DoneEx e	 -> simplExprF (zapSubstEnv env) e cont
	ContEx tvs ids e -> simplExprF (setSubstEnv env tvs ids) e cont
929
	DoneId var1      -> completeCall (zapSubstEnv env) var1 cont
930
		-- Note [zapSubstEnv]
931 932 933 934 935 936 937 938
		-- The template is already simplified, so don't re-substitute.
		-- This is VITAL.  Consider
		--	let x = e in
		--	let y = \z -> ...x... in
		--	\ x -> ...y...
		-- We'll clone the inner \x, adding x->x' in the id_subst
		-- Then when we inline y, we must *not* replace x by x' in
		-- the inlined copy!!
939

940
---------------------------------------------------------
941
--	Dealing with a call site
942

943
completeCall env var cont
944 945 946 947 948 949 950 951 952 953 954 955
  = do	{ dflags <- getDOptsSmpl
	; let	(args,call_cont) = contArgs cont
		-- The args are OutExprs, obtained by *lazily* substituting
		-- in the args found in cont.  These args are only examined
		-- to limited depth (unless a rule fires).  But we must do
		-- the substitution; rule matching on un-simplified args would
		-- be bogus

	------------- First try rules ----------------
	-- Do this before trying inlining.  Some functions have 
	-- rules *and* are strict; in this case, we don't want to 
	-- inline the wrapper of the non-specialised thing; better
956
	-- to call the specialised thing instead.
957
	--
958 959 960
	-- We used to use the black-listing mechanism to ensure that inlining of 
	-- the wrapper didn't occur for things that have specialisations till a 
	-- later phase, so but now we just try RULES first
961
	--
962 963
	-- Note [Self-recursive rules]
	-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~
964 965 966 967 968 969 970 971 972 973 974
	-- You might think that we shouldn't apply rules for a loop breaker: 
	-- doing so might give rise to an infinite loop, because a RULE is
	-- rather like an extra equation for the function:
	--	RULE:		f (g x) y = x+y
	--	Eqn:		f a     y = a-y
	--
	-- But it's too drastic to disable rules for loop breakers.  
	-- Even the foldr/build rule would be disabled, because foldr 
	-- is recursive, and hence a loop breaker:
	--	foldr k z (build g) = g k z
	-- So it's up to the programmer: rules can cause divergence
975
	; rules <- getRules
976
	; let	in_scope   = getInScope env
977
		maybe_rule = case activeRule dflags env of
978 979 980 981 982 983
				Nothing     -> Nothing	-- No rules apply
				Just act_fn -> lookupRule act_fn in_scope 
							  rules var args 
	; case maybe_rule of {
	    Just (rule, rule_rhs) -> 
		tick (RuleFired (ru_name rule))			`thenSmpl_`
984
		(if dopt Opt_D_dump_rule_firings dflags then
985
		   pprTrace "Rule fired" (vcat [
986
			text "Rule:" <+> ftext (ru_name rule),
987
			text "Before:" <+> ppr var <+> sep (map pprParendExpr args),
988 989
			text "After: " <+> pprCoreExpr rule_rhs,
			text "Cont:  " <+> ppr call_cont])
990 991
		 else
			id)		$
992 993
		simplExprF env rule_rhs (dropArgs (ruleArity rule) cont)
		-- The ruleArity says how many args the rule consumed
994
	
995 996 997 998 999 1000 1001 1002 1003 1004
	  ; Nothing -> do	-- No rules

	------------- Next try inlining ----------------
	{ let	arg_infos = [interestingArg arg | arg <- args, isValArg arg]
		n_val_args = length arg_infos
	      	interesting_cont = interestingCallContext (notNull args)
						  	  (notNull arg_infos)
						  	  call_cont
	 	active_inline = activeInline env var
		maybe_inline  = callSiteInline dflags active_inline
1005
				       var arg_infos interesting_cont
1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
	; case maybe_inline of {
	    Just unfolding  	-- There is an inlining!
	      ->  do { tick (UnfoldingDone var)
		     ; (if dopt Opt_D_dump_inlinings dflags then
			   pprTrace "Inlining done" (vcat [
				text "Before:" <+> ppr var <+> sep (map pprParendExpr args),
				text "Inlined fn: " <+> nest 2 (ppr unfolding),
				text "Cont:  " <+> ppr call_cont])
			 else
				id)
		       simplExprF env unfolding cont }

	    ; Nothing -> 		-- No inlining!

	------------- No inlining! ----------------
	-- Next, look for rules or specialisations that match
	--
	rebuildCall env (Var var) (idType var) 
		    (mkArgInfo var n_val_args call_cont) cont
    }}}}
1026

1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
rebuildCall :: SimplEnv
	    -> OutExpr -> OutType	-- Function and its type
	    -> (Bool, [Bool])		-- See SimplUtils.mkArgInfo
	    -> SimplCont
	    -> SimplM (SimplEnv, OutExpr)
rebuildCall env fun fun_ty (has_rules, []) cont
  -- When we run out of strictness args, it means
  -- that the call is definitely bottom; see SimplUtils.mkArgInfo
  -- Then we want to discard the entire strict continuation.  E.g.
  --	* case (error "hello") of { ... }
  --	* (error "Hello") arg
  --	* f (error "Hello") where f is strict
  --	etc
  -- Then, especially in the first of these cases, we'd like to discard
  -- the continuation, leaving just the bottoming expression.  But the
  -- type might not be right, so we may have to add a coerce.
1043
  | not (contIsTrivial cont)	 -- Only do this if there is a non-trivial
1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069
  = return (env, mk_coerce fun)  -- contination to discard, else we do it
  where				 -- again and again!
    cont_ty = contResultType cont
    co      = mkUnsafeCoercion fun_ty cont_ty
    mk_coerce expr | cont_ty `coreEqType` fun_ty = fun
		   | otherwise = mkCoerce co fun

rebuildCall env fun fun_ty info (ApplyTo _ (Type arg_ty) se cont)
  = do	{ ty' <- simplType (se `setInScope` env) arg_ty
	; rebuildCall env (fun `App` Type ty') (applyTy fun_ty ty') info cont }

rebuildCall env fun fun_ty (has_rules, str:strs) (ApplyTo _ arg arg_se cont)
  | str || isStrictType arg_ty		-- Strict argument
  = -- pprTrace "Strict Arg" (ppr arg $$ ppr (seIdSubst env) $$ ppr (seInScope env)) $
    simplExprF (arg_se `setFloats` env) arg
	       (StrictArg fun fun_ty (has_rules, strs) cont)
		-- Note [Shadowing]

  | otherwise				-- Lazy argument
	-- DO NOT float anything outside, hence simplExprC
	-- There is no benefit (unlike in a let-binding), and we'd
	-- have to be very careful about bogus strictness through 
	-- floating a demanded let.
  = do	{ arg' <- simplExprC (arg_se `setInScope` env) arg
			     (mkLazyArgStop arg_ty has_rules)
	; rebuildCall env (fun `App` arg') res_ty (has_rules, strs) cont }
1070
  where
1071
    (arg_ty, res_ty) = splitFunTy fun_ty
1072

1073 1074
rebuildCall env fun fun_ty info cont
  = rebuild env fun cont
1075
\end{code}
1076

1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
Note [Shadowing]
~~~~~~~~~~~~~~~~
This part of the simplifier may break the no-shadowing invariant
Consider
	f (...(\a -> e)...) (case y of (a,b) -> e')
where f is strict in its second arg
If we simplify the innermost one first we get (...(\a -> e)...)
Simplifying the second arg makes us float the case out, so we end up with
	case y of (a,b) -> f (...(\a -> e)...) e'
So the output does not have the no-shadowing invariant.  However, there is
no danger of getting name-capture, because when the first arg was simplified
we used an in-scope set that at least mentioned all the variables free in its
static environment, and that is enough.

We can't just do innermost first, or we'd end up with a dual problem:
	case x of (a,b) -> f e (...(\a -> e')...)

I spent hours trying to recover the no-shadowing invariant, but I just could
not think of an elegant way to do it.  The simplifier is already knee-deep in
continuations.  We have to keep the right in-scope set around; AND we have
to get the effect that finding (error "foo") in a strict arg position will
discard the entire application and replace it with (error "foo").  Getting
all this at once is TOO HARD!
1100

1101 1102
%************************************************************************
%*									*
1103
		Rebuilding a cse expression
1104 1105
%*									*
%************************************************************************
1106

1107 1108 1109
Blob of helper functions for the "case-of-something-else" situation.

\begin{code}
1110
---------------------------------------------------------
1111
-- 	Eliminate the case if possible
1112

1113 1114 1115
rebuildCase :: SimplEnv
	    -> OutExpr		-- Scrutinee
	    -> InId		-- Case binder
1116
	    -> [InAlt]		-- Alternatives (inceasing order)
1117
	    -> SimplCont
1118
	    -> SimplM (SimplEnv, OutExpr)
1119

1120 1121 1122 1123
--------------------------------------------------
--	1. Eliminate the case if there's a known constructor
--------------------------------------------------

1124 1125 1126 1127
rebuildCase env scrut case_bndr alts cont
  | Just (con,args) <- exprIsConApp_maybe scrut	
	-- Works when the scrutinee is a variable with a known unfolding
	-- as well as when it's an explicit constructor application
1128
  = knownCon env scrut (DataAlt con) args case_bndr alts cont
1129

1130 1131
  | Lit lit <- scrut	-- No need for same treatment as constructors
			-- because literals are inlined more vigorously
1132
  = knownCon env scrut (LitAlt lit) [] case_bndr alts cont
1133

1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173

--------------------------------------------------
--	2. Eliminate the case if scrutinee is evaluated
--------------------------------------------------

rebuildCase env scrut case_bndr [(con,bndrs,rhs)] cont
  -- See if we can get rid of the case altogether
  -- See the extensive notes on case-elimination above
  -- mkCase made sure that if all the alternatives are equal, 
  -- then there is now only one (DEFAULT) rhs
 | all isDeadBinder bndrs	-- bndrs are [InId]

	-- Check that the scrutinee can be let-bound instead of case-bound
 , exprOkForSpeculation scrut
		-- OK not to evaluate it
		-- This includes things like (==# a# b#)::Bool
		-- so that we simplify 
		-- 	case ==# a# b# of { True -> x; False -> x }
		-- to just
		--	x
		-- This particular example shows up in default methods for
		-- comparision operations (e.g. in (>=) for Int.Int32)
	|| exprIsHNF scrut			-- It's already evaluated
	|| var_demanded_later scrut		-- It'll be demanded later

--      || not opt_SimplPedanticBottoms)	-- Or we don't care!
--	We used to allow improving termination by discarding cases, unless -fpedantic-bottoms was on,
-- 	but that breaks badly for the dataToTag# primop, which relies on a case to evaluate
-- 	its argument:  case x of { y -> dataToTag# y }
--	Here we must *not* discard the case, because dataToTag# just fetches the tag from
--	the info pointer.  So we'll be pedantic all the time, and see if that gives any
-- 	other problems
--	Also we don't want to discard 'seq's
  = do	{ tick (CaseElim case_bndr)
	; env <- simplNonRecX env case_bndr scrut
	; simplExprF env rhs cont }
  where
	-- The case binder is going to be evaluated later, 
	-- and the scrutinee is a simple variable
    var_demanded_later (Var v) = isStrictDmd (idNewDemandInfo case_bndr)
1174 1175 1176
    		       	         && not (isTickBoxOp v)	
				    -- ugly hack; covering this case is what 
				    -- exprOkForSpeculation was intended for.
1177 1178 1179 1180 1181 1182 1183 1184
    var_demanded_later other   = False


--------------------------------------------------
--	3. Catch-all case
--------------------------------------------------

rebuildCase env scrut case_bndr alts cont
1185 1186 1187
  = do	{ 	-- Prepare the continuation;
		-- The new subst_env is in place
	  (env, dup_cont, nodup_cont) <- prepareCaseCont env alts cont
1188

1189
	-- Simplify the alternatives
1190
	; (scrut', case_bndr', alts') <- simplAlts env scrut case_bndr alts dup_cont
1191
	; let res_ty' = contResultType dup_cont
1192
	; case_expr <- mkCase scrut' case_bndr' res_ty' alts'
sof's avatar
sof committed
1193

1194 1195
	-- Notice that rebuildDone returns the in-scope set from env, not alt_env
	-- The case binder *not* scope over the whole returned case-expression
1196
	; rebuild env case_expr nodup_cont }
1197
\end{code}
1198

1199 1200 1201 1202 1203
simplCaseBinder checks whether the scrutinee is a variable, v.  If so,
try to eliminate uses of v in the RHSs in favour of case_bndr; that
way, there's a chance that v will now only be used once, and hence
inlined.

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
1204 1205
Note [no-case-of-case]
~~~~~~~~~~~~~~~~~~~~~~
1206 1207 1208
There is a time we *don't* want to do that, namely when
-fno-case-of-case is on.  This happens in the first simplifier pass,
and enhances full laziness.  Here's the bad case:
1209 1210 1211 1212 1213
	f = \ y -> ...(case x of I# v -> ...(case x of ...) ... )
If we eliminate the inner case, we trap it inside the I# v -> arm,
which might prevent some full laziness happening.  I've seen this
in action in spectral/cichelli/Prog.hs:
	 [(m,n) | m <- [1..max], n <- [1..max]]
1214 1215
Hence the check for NoCaseOfCase.

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
1216 1217 1218 1219
Note [Suppressing the case binder-swap]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
There is another situation when it might make sense to suppress the
case-expression binde-swap. If we have
1220 1221 1222 1223 1224 1225 1226 1227 1228

    case x of w1 { DEFAULT -> case x of w2 { A -> e1; B -> e2 }
	           ...other cases .... }

We'll perform the binder-swap for the outer case, giving

    case x of w1 { DEFAULT -> case w1 of w2 { A -> e1; B -> e2 } 
	           ...other cases .... }

1229 1230 1231 1232
But there is no point in doing it for the inner case, because w1 can't
be inlined anyway.  Furthermore, doing the case-swapping involves
zapping w2's occurrence info (see paragraphs that follow), and that
forces us to bind w2 when doing case merging.  So we get
1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244

    case x of w1 { A -> let w2 = w1 in e1
		   B -> let w2 = w1 in e2
	           ...other cases .... }

This is plain silly in the common case where w2 is dead.

Even so, I can't see a good way to implement this idea.  I tried
not doing the binder-swap if the scrutinee was already evaluated
but that failed big-time:

	data T = MkT !Int
1245

1246 1247 1248
	case v of w  { MkT x ->
	case x of x1 { I# y1 ->
	case x of x2 { I# y2 -> ...
1249

1250 1251 1252 1253 1254
Notice that because MkT is strict, x is marked "evaluated".  But to
eliminate the last case, we must either make sure that x (as well as
x1) has unfolding MkT y1.  THe straightforward thing to do is to do
the binder-swap.  So this whole note is a no-op.

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
1255 1256
Note [zapOccInfo]
~~~~~~~~~~~~~~~~~
1257 1258 1259 1260
If we replace the scrutinee, v, by tbe case binder, then we have to nuke
any occurrence info (eg IAmDead) in the case binder, because the
case-binder now effectively occurs whenever v does.  AND we have to do
the same for the pattern-bound variables!  Example:
1261

1262
	(case x of { (a,b) -> a }) (case x of { (p,q) -> q })
1263

1264 1265
Here, b and p are dead.  But when we move the argment inside the first
case RHS, and eliminate the second case, we get
1266

1267
	case x of { (a,b) -> a b }
1268

1269
Urk! b is alive!  Reason: the scrutinee was a variable, and case elimination
1270 1271 1272 1273 1274 1275 1276 1277 1278 1279
happened.  

Indeed, this can happen anytime the case binder isn't dead:
	case <any> of x { (a,b) -> 
        case x of { (p,q) -> p } }
Here (a,b) both look dead, but come alive after the inner case is eliminated.
The point is that we bring into the envt a binding
	let x = (a,b) 
after the outer case, and that makes (a,b) alive.  At least we do unless
the case binder is guaranteed dead.
1280

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
1281 1282 1283 1284 1285 1286 1287 1288 1289
Note [Case of cast]
~~~~~~~~~~~~~~~~~~~
Consider 	case (v `cast` co) of x { I# ->
		... (case (v `cast` co) of {...}) ...
We'd like to eliminate the inner case.  We can get this neatly by 
arranging that inside the outer case we add the unfolding
	v |-> x `cast` (sym co)
to v.  Then we should inline v at the inner case, cancel the casts, and away we go
	
1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308