Simplify.lhs 82.4 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
module Simplify ( simplTopBinds, simplExpr ) where
8

9
#include "HsVersions.h"
10

simonpj@microsoft.com's avatar
Wibble  
simonpj@microsoft.com committed
11
import DynFlags
12
import SimplMonad
Ian Lynagh's avatar
Ian Lynagh committed
13 14
import Type hiding      ( substTy, extendTvSubst )
import SimplEnv
15
import SimplUtils
16 17
import Literal		( mkStringLit )
import MkId		( rUNTIME_ERROR_ID )
18
import Id
19
import Var
20 21
import IdInfo
import Coercion
Ian Lynagh's avatar
Ian Lynagh committed
22 23
import FamInstEnv       ( topNormaliseType )
import DataCon          ( dataConRepStrictness, dataConUnivTyVars )
24
import CoreSyn
Ian Lynagh's avatar
Ian Lynagh committed
25 26 27
import NewDemand        ( isStrictDmd )
import PprCore          ( pprParendExpr, pprCoreExpr )
import CoreUnfold       ( mkUnfolding, callSiteInline, CallCtxt(..) )
28
import CoreUtils
Ian Lynagh's avatar
Ian Lynagh committed
29 30 31 32 33 34 35 36 37
import Rules            ( lookupRule )
import BasicTypes       ( isMarkedStrict )
import CostCentre       ( currentCCS )
import TysPrim          ( realWorldStatePrimTy )
import PrelInfo         ( realWorldPrimId )
import BasicTypes       ( TopLevelFlag(..), isTopLevel,
                          RecFlag(..), isNonRuleLoopBreaker )
import Maybes           ( orElse )
import Data.List        ( mapAccumL )
38
import Outputable
39
import MonadUtils
40
import FastString
41 42 43
\end{code}


44 45
The guts of the simplifier is in this module, but the driver loop for
the simplifier is in SimplCore.lhs.
46 47


48
-----------------------------------------
Ian Lynagh's avatar
Ian Lynagh committed
49
        *** IMPORTANT NOTE ***
50 51 52 53 54 55
-----------------------------------------
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.


56
-----------------------------------------
Ian Lynagh's avatar
Ian Lynagh committed
57
        *** IMPORTANT NOTE ***
58 59 60 61 62 63 64 65 66 67
-----------------------------------------
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).



-----------------------------------------
Ian Lynagh's avatar
Ian Lynagh committed
68
        ORGANISATION OF FUNCTIONS
69 70 71 72 73 74
-----------------------------------------
simplTopBinds
  - simplify all top-level binders
  - for NonRec, call simplRecOrTopPair
  - for Rec,    call simplRecBind

Ian Lynagh's avatar
Ian Lynagh committed
75 76 77

        ------------------------------
simplExpr (applied lambda)      ==> simplNonRecBind
78 79 80
simplExpr (Let (NonRec ...) ..) ==> simplNonRecBind
simplExpr (Let (Rec ...)    ..) ==> simplify binders; simplRecBind

Ian Lynagh's avatar
Ian Lynagh committed
81 82
        ------------------------------
simplRecBind    [binders already simplfied]
83 84 85 86
  - use simplRecOrTopPair on each pair in turn

simplRecOrTopPair [binder already simplified]
  Used for: recursive bindings (top level and nested)
Ian Lynagh's avatar
Ian Lynagh committed
87 88
            top-level non-recursive bindings
  Returns:
89 90 91 92 93
  - check for PreInlineUnconditionally
  - simplLazyBind

simplNonRecBind
  Used for: non-top-level non-recursive bindings
Ian Lynagh's avatar
Ian Lynagh committed
94 95 96
            beta reductions (which amount to the same thing)
  Because it can deal with strict arts, it takes a
        "thing-inside" and returns an expression
97 98 99 100

  - check for PreInlineUnconditionally
  - simplify binder, including its IdInfo
  - if strict binding
Ian Lynagh's avatar
Ian Lynagh committed
101 102 103
        simplStrictArg
        mkAtomicArgs
        completeNonRecX
104
    else
Ian Lynagh's avatar
Ian Lynagh committed
105 106
        simplLazyBind
        addFloats
107

Ian Lynagh's avatar
Ian Lynagh committed
108
simplNonRecX:   [given a *simplified* RHS, but an *unsimplified* binder]
109 110 111 112
  Used for: binding case-binder and constr args in a known-constructor case
  - check for PreInLineUnconditionally
  - simplify binder
  - completeNonRecX
Ian Lynagh's avatar
Ian Lynagh committed
113 114 115

        ------------------------------
simplLazyBind:  [binder already simplified, RHS not]
116
  Used for: recursive bindings (top level and nested)
Ian Lynagh's avatar
Ian Lynagh committed
117 118 119
            top-level non-recursive bindings
            non-top-level, but *lazy* non-recursive bindings
        [must not be strict or unboxed]
120
  Returns floats + an augmented environment, not an expression
Ian Lynagh's avatar
Ian Lynagh committed
121 122
  - substituteIdInfo and add result to in-scope
        [so that rules are available in rec rhs]
123 124 125
  - simplify rhs
  - mkAtomicArgs
  - float if exposes constructor or PAP
126
  - completeBind
127 128


Ian Lynagh's avatar
Ian Lynagh committed
129
completeNonRecX:        [binder and rhs both simplified]
130
  - if the the thing needs case binding (unlifted and not ok-for-spec)
Ian Lynagh's avatar
Ian Lynagh committed
131
        build a Case
132
   else
Ian Lynagh's avatar
Ian Lynagh committed
133 134
        completeBind
        addFloats
135

Ian Lynagh's avatar
Ian Lynagh committed
136 137
completeBind:   [given a simplified RHS]
        [used for both rec and non-rec bindings, top level and not]
138 139 140 141 142 143 144 145
  - try PostInlineUnconditionally
  - add unfolding [this is the only place we add an unfolding]
  - add arity



Right hand sides and arguments
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ian Lynagh's avatar
Ian Lynagh committed
146 147 148
In many ways we want to treat
        (a) the right hand side of a let(rec), and
        (b) a function argument
149 150 151
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.
Ian Lynagh's avatar
Ian Lynagh committed
152 153 154

        f (g x, h x)
        g (+ x)
155 156 157 158

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

Ian Lynagh's avatar
Ian Lynagh committed
159 160
        f (let { a = g x; b = h x } in (a,b))
        g (\y. + x y)
161 162 163

On the other hand if we see the let-defns

Ian Lynagh's avatar
Ian Lynagh committed
164 165
        p = (g x, h x)
        q = + x
166 167

then we *do* want to ANF-ise and eta-expand, so that p and q
Ian Lynagh's avatar
Ian Lynagh committed
168
can be safely inlined.
169 170 171 172 173

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

Ian Lynagh's avatar
Ian Lynagh committed
174
        r = let x = e in (x,x)
175 176 177 178 179 180 181 182 183 184 185 186 187 188

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

Ian Lynagh's avatar
Ian Lynagh committed
189
        case e of (a,b) -> \x -> case a of (p,q) -> \y -> r
190 191 192 193 194

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.
195 196


197
%************************************************************************
Ian Lynagh's avatar
Ian Lynagh committed
198
%*                                                                      *
199
\subsection{Bindings}
Ian Lynagh's avatar
Ian Lynagh committed
200
%*                                                                      *
201 202 203
%************************************************************************

\begin{code}
204
simplTopBinds :: SimplEnv -> [InBind] -> SimplM [OutBind]
205

Ian Lynagh's avatar
Ian Lynagh committed
206
simplTopBinds env0 binds0
Ian Lynagh's avatar
Ian Lynagh committed
207 208 209 210
  = 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.
Ian Lynagh's avatar
Ian Lynagh committed
211
        ; env1 <- simplRecBndrs env0 (bindersOfBinds binds0)
Ian Lynagh's avatar
Ian Lynagh committed
212 213 214
        ; dflags <- getDOptsSmpl
        ; let dump_flag = dopt Opt_D_dump_inlinings dflags ||
                          dopt Opt_D_dump_rule_firings dflags
Ian Lynagh's avatar
Ian Lynagh committed
215
        ; env2 <- simpl_binds dump_flag env1 binds0
Ian Lynagh's avatar
Ian Lynagh committed
216
        ; freeTick SimplifierDone
Ian Lynagh's avatar
Ian Lynagh committed
217
        ; return (getFloats env2) }
218
  where
Ian Lynagh's avatar
Ian Lynagh committed
219 220 221 222 223 224
        -- We need to track the zapped top-level binders, because
        -- they should have their fragile IdInfo zapped (notably occurrence info)
        -- That's why we run down binds and bndrs' simultaneously.
        --
        -- The dump-flag emits a trace for each top-level binding, which
        -- helps to locate the tracing for inlining and rule firing
225
    simpl_binds :: Bool -> SimplEnv -> [InBind] -> SimplM SimplEnv
Ian Lynagh's avatar
Ian Lynagh committed
226 227
    simpl_binds _    env []           = return env
    simpl_binds dump env (bind:binds) = do { env' <- trace_bind dump bind $
Ian Lynagh's avatar
Ian Lynagh committed
228 229
                                                     simpl_bind env bind
                                           ; simpl_binds dump env' binds }
230

Ian Lynagh's avatar
Ian Lynagh committed
231 232
    trace_bind True  bind = pprTrace "SimplBind" (ppr (bindersOf bind))
    trace_bind False _    = \x -> x
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
233

234 235
    simpl_bind env (Rec pairs)  = simplRecBind      env  TopLevel pairs
    simpl_bind env (NonRec b r) = simplRecOrTopPair env' TopLevel b b' r
Ian Lynagh's avatar
Ian Lynagh committed
236 237
        where
          (env', b') = addBndrRules env b (lookupRecBndr env b)
238 239 240 241
\end{code}


%************************************************************************
Ian Lynagh's avatar
Ian Lynagh committed
242
%*                                                                      *
243
\subsection{Lazy bindings}
Ian Lynagh's avatar
Ian Lynagh committed
244
%*                                                                      *
245 246 247
%************************************************************************

simplRecBind is used for
Ian Lynagh's avatar
Ian Lynagh committed
248
        * recursive bindings only
249 250 251

\begin{code}
simplRecBind :: SimplEnv -> TopLevelFlag
Ian Lynagh's avatar
Ian Lynagh committed
252 253
             -> [(InId, InExpr)]
             -> SimplM SimplEnv
Ian Lynagh's avatar
Ian Lynagh committed
254 255 256 257 258 259
simplRecBind env0 top_lvl pairs0
  = do  { let (env_with_info, triples) = mapAccumL add_rules env0 pairs0
        ; env1 <- go (zapFloats env_with_info) triples
        ; return (env0 `addRecFloats` env1) }
        -- addFloats adds the floats from env1,
        -- *and* updates env0 with the in-scope set from env1
260
  where
261
    add_rules :: SimplEnv -> (InBndr,InExpr) -> (SimplEnv, (InBndr, OutBndr, InExpr))
Ian Lynagh's avatar
Ian Lynagh committed
262
        -- Add the (substituted) rules to the binder
263
    add_rules env (bndr, rhs) = (env', (bndr, bndr', rhs))
Ian Lynagh's avatar
Ian Lynagh committed
264 265
        where
          (env', bndr') = addBndrRules env bndr (lookupRecBndr env bndr)
266

267
    go env [] = return env
Ian Lynagh's avatar
Ian Lynagh committed
268

269
    go env ((old_bndr, new_bndr, rhs) : pairs)
Ian Lynagh's avatar
Ian Lynagh committed
270 271
        = do { env' <- simplRecOrTopPair env top_lvl old_bndr new_bndr rhs
             ; go env' pairs }
272 273
\end{code}

274
simplOrTopPair is used for
Ian Lynagh's avatar
Ian Lynagh committed
275 276
        * recursive bindings (whether top level or not)
        * top-level non-recursive bindings
277 278 279 280 281

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

\begin{code}
simplRecOrTopPair :: SimplEnv
Ian Lynagh's avatar
Ian Lynagh committed
282 283 284
                  -> TopLevelFlag
                  -> InId -> OutBndr -> InExpr  -- Binder and rhs
                  -> SimplM SimplEnv    -- Returns an env that includes the binding
285

286
simplRecOrTopPair env top_lvl old_bndr new_bndr rhs
Ian Lynagh's avatar
Ian Lynagh committed
287 288 289
  | preInlineUnconditionally env top_lvl old_bndr rhs   -- Check for unconditional inline
  = do  { tick (PreInlineUnconditionally old_bndr)
        ; return (extendIdSubst env old_bndr (mkContEx env rhs)) }
290 291

  | otherwise
Ian Lynagh's avatar
Ian Lynagh committed
292 293
  = simplLazyBind env top_lvl Recursive old_bndr new_bndr rhs env
        -- May not actually be recursive, but it doesn't matter
294 295 296 297
\end{code}


simplLazyBind is used for
298 299
  * [simplRecOrTopPair] recursive bindings (whether top level or not)
  * [simplRecOrTopPair] top-level non-recursive bindings
Ian Lynagh's avatar
Ian Lynagh committed
300
  * [simplNonRecE]      non-top-level *lazy* non-recursive bindings
301 302

Nota bene:
Ian Lynagh's avatar
Ian Lynagh committed
303
    1. It assumes that the binder is *already* simplified,
304
       and is in scope, and its IdInfo too, except unfolding
305 306 307 308 309 310 311 312

    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
Ian Lynagh's avatar
Ian Lynagh committed
313 314 315 316 317
              -> TopLevelFlag -> RecFlag
              -> InId -> OutId          -- Binder, both pre-and post simpl
                                        -- The OutId has IdInfo, except arity, unfolding
              -> InExpr -> SimplEnv     -- The RHS and its environment
              -> SimplM SimplEnv
318

319
simplLazyBind env top_lvl is_rec bndr bndr1 rhs rhs_se
Ian Lynagh's avatar
Ian Lynagh committed
320
  = do  { let   rhs_env     = rhs_se `setInScope` env
321 322 323 324 325 326 327 328 329 330
		(tvs, body) = case collectTyBinders rhs of
			        (tvs, body) | not_lam body -> (tvs,body)
					    | otherwise	   -> ([], rhs)
		not_lam (Lam _ _) = False
		not_lam _	  = True
			-- Do not do the "abstract tyyvar" thing if there's
			-- a lambda inside, becuase it defeats eta-reduction
			--    f = /\a. \x. g a x  
			-- should eta-reduce

Ian Lynagh's avatar
Ian Lynagh committed
331
        ; (body_env, tvs') <- simplBinders rhs_env tvs
332
                -- See Note [Floating and type abstraction] in SimplUtils
Ian Lynagh's avatar
Ian Lynagh committed
333

334 335
        -- Simplify the RHS
        ; (body_env1, body1) <- simplExprF body_env body mkBoringStop
Ian Lynagh's avatar
Ian Lynagh committed
336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353

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

        ; (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') }

                else if null tvs then           -- Simple floating
                     do { tick LetFloatFromLet
                        ; return (addFloats env body_env2, body2) }

                else                            -- Do type-abstraction first
                     do { tick LetFloatFromLet
                        ; (poly_binds, body3) <- abstractFloats tvs' body_env2 body2
                        ; rhs' <- mkLam tvs' body3
354 355
                        ; env' <- foldlM add_poly_bind env poly_binds
                        ; return (env', rhs') }
Ian Lynagh's avatar
Ian Lynagh committed
356 357

        ; completeBind env' top_lvl bndr bndr1 rhs' }
358 359 360 361 362 363 364 365 366 367 368
  where
    add_poly_bind env (NonRec poly_id rhs)
	= completeBind env top_lvl poly_id poly_id rhs
		-- completeBind adds the new binding in the
		-- proper way (ie complete with unfolding etc),
		-- and extends the in-scope set
    add_poly_bind env bind@(Rec _)
	= return (extendFloats env bind)
		-- Hack: letrecs are more awkward, so we extend "by steam"
		-- without adding unfoldings etc.  At worst this leads to
		-- more simplifier iterations
369
\end{code}
370

Ian Lynagh's avatar
Ian Lynagh committed
371
A specialised variant of simplNonRec used when the RHS is already simplified,
372 373 374 375
notably in knownCon.  It uses case-binding where necessary.

\begin{code}
simplNonRecX :: SimplEnv
Ian Lynagh's avatar
Ian Lynagh committed
376 377 378
             -> InId            -- Old binder
             -> OutExpr         -- Simplified RHS
             -> SimplM SimplEnv
379 380

simplNonRecX env bndr new_rhs
Ian Lynagh's avatar
Ian Lynagh committed
381
  = do  { (env', bndr') <- simplBinder env bndr
382
        ; completeNonRecX env' (isStrictId bndr) bndr bndr' new_rhs }
383 384

completeNonRecX :: SimplEnv
385
                -> Bool
Ian Lynagh's avatar
Ian Lynagh committed
386 387 388 389
                -> InId                 -- Old binder
                -> OutId                -- New binder
                -> OutExpr              -- Simplified RHS
                -> SimplM SimplEnv
390

391
completeNonRecX env is_strict old_bndr new_bndr new_rhs
Ian Lynagh's avatar
Ian Lynagh committed
392 393
  = do  { (env1, rhs1) <- prepareRhs (zapFloats env) new_rhs
        ; (env2, rhs2) <-
394
                if doFloatFromRhs NotTopLevel NonRecursive is_strict rhs1 env1
Ian Lynagh's avatar
Ian Lynagh committed
395 396 397 398
                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
        ; completeBind env2 NotTopLevel old_bndr new_bndr rhs2 }
399 400 401 402
\end{code}

{- No, no, no!  Do not try preInlineUnconditionally in completeNonRecX
   Doing so risks exponential behaviour, because new_rhs has been simplified once already
Ian Lynagh's avatar
Ian Lynagh committed
403
   In the cases described by the folowing commment, postInlineUnconditionally will
404
   catch many of the relevant cases.
Ian Lynagh's avatar
Ian Lynagh committed
405 406 407 408 409 410 411 412
        -- 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 ...
        --
        -- 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.
413

414
   Furthermore in the case-binding case preInlineUnconditionally risks extra thunks
Ian Lynagh's avatar
Ian Lynagh committed
415 416 417 418 419 420
        -- 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.
421

422 423 424 425
  | preInlineUnconditionally env NotTopLevel bndr new_rhs
  = thing_inside (extendIdSubst env bndr (DoneEx new_rhs))
-}

426
----------------------------------
427
prepareRhs takes a putative RHS, checks whether it's a PAP or
Ian Lynagh's avatar
Ian Lynagh committed
428
constructor application and, if so, converts it to ANF, so that the
429
resulting thing can be inlined more easily.  Thus
Ian Lynagh's avatar
Ian Lynagh committed
430
        x = (f a, g b)
431
becomes
Ian Lynagh's avatar
Ian Lynagh committed
432 433 434
        t1 = f a
        t2 = g b
        x = (t1,t2)
435

436
We also want to deal well cases like this
Ian Lynagh's avatar
Ian Lynagh committed
437
        v = (f e1 `cast` co) e2
438
Here we want to make e1,e2 trivial and get
Ian Lynagh's avatar
Ian Lynagh committed
439
        x1 = e1; x2 = e2; v = (f x1 `cast` co) v2
440 441
That's what the 'go' loop in prepareRhs does

442 443 444
\begin{code}
prepareRhs :: SimplEnv -> OutExpr -> SimplM (SimplEnv, OutExpr)
-- Adds new floats to the env iff that allows us to return a good RHS
Ian Lynagh's avatar
Ian Lynagh committed
445
prepareRhs env (Cast rhs co)    -- Note [Float coercions]
Ian Lynagh's avatar
Ian Lynagh committed
446
  | (ty1, _ty2) <- coercionKind co       -- Do *not* do this if rhs has an unlifted type
Ian Lynagh's avatar
Ian Lynagh committed
447 448 449
  , not (isUnLiftedType ty1)            -- see Note [Float coercions (unlifted)]
  = do  { (env', rhs') <- makeTrivial env rhs
        ; return (env', Cast rhs' co) }
450

Ian Lynagh's avatar
Ian Lynagh committed
451 452 453
prepareRhs env0 rhs0
  = do  { (_is_val, env1, rhs1) <- go 0 env0 rhs0
        ; return (env1, rhs1) }
454
  where
455
    go n_val_args env (Cast rhs co)
Ian Lynagh's avatar
Ian Lynagh committed
456 457
        = do { (is_val, env', rhs') <- go n_val_args env rhs
             ; return (is_val, env', Cast rhs' co) }
458
    go n_val_args env (App fun (Type ty))
Ian Lynagh's avatar
Ian Lynagh committed
459 460
        = do { (is_val, env', rhs') <- go n_val_args env fun
             ; return (is_val, env', App rhs' (Type ty)) }
461
    go n_val_args env (App fun arg)
Ian Lynagh's avatar
Ian Lynagh committed
462 463 464 465 466
        = 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) }
467
    go n_val_args env (Var fun)
Ian Lynagh's avatar
Ian Lynagh committed
468 469 470 471 472
        = 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)
Ian Lynagh's avatar
Ian Lynagh committed
473
    go _ env other
Ian Lynagh's avatar
Ian Lynagh committed
474
        = return (False, env, other)
475 476
\end{code}

477

478 479 480
Note [Float coercions]
~~~~~~~~~~~~~~~~~~~~~~
When we find the binding
Ian Lynagh's avatar
Ian Lynagh committed
481
        x = e `cast` co
482
we'd like to transform it to
Ian Lynagh's avatar
Ian Lynagh committed
483 484
        x' = e
        x = x `cast` co         -- A trivial binding
485 486 487 488 489 490 491 492 493 494 495 496 497
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) }
Ian Lynagh's avatar
Ian Lynagh committed
498
                -- This case should optimise
499

500 501
Note [Float coercions (unlifted)]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ian Lynagh's avatar
Ian Lynagh committed
502
BUT don't do [Float coercions] if 'e' has an unlifted type.
503 504
This *can* happen:

Ian Lynagh's avatar
Ian Lynagh committed
505 506
     foo :: Int = (error (# Int,Int #) "urk")
                  `cast` CoUnsafe (# Int,Int #) Int
507 508 509

If do the makeTrivial thing to the error call, we'll get
    foo = case error (# Int,Int #) "urk" of v -> v `cast` ...
Ian Lynagh's avatar
Ian Lynagh committed
510
But 'v' isn't in scope!
511 512

These strange casts can happen as a result of case-of-case
Ian Lynagh's avatar
Ian Lynagh committed
513 514
        bar = case (case x of { T -> (# 2,3 #); F -> error "urk" }) of
                (# p,q #) -> p+q
515

516 517 518 519 520 521 522

\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)
Ian Lynagh's avatar
Ian Lynagh committed
523
  | otherwise           -- See Note [Take care] below
Ian Lynagh's avatar
Ian Lynagh committed
524
  = do  { var <- newId (fsLit "a") (exprType expr)
525
        ; env' <- completeNonRecX env False var var expr
Ian Lynagh's avatar
Ian Lynagh committed
526
        ; return (env', substExpr env' (Var var)) }
527
\end{code}
528 529


530
%************************************************************************
Ian Lynagh's avatar
Ian Lynagh committed
531
%*                                                                      *
532
\subsection{Completing a lazy binding}
Ian Lynagh's avatar
Ian Lynagh committed
533
%*                                                                      *
534 535
%************************************************************************

536 537 538 539 540
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
541 542 543 544 545 546 547 548

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
Ian Lynagh's avatar
Ian Lynagh committed
549
  - top-level bindings (when let-to-case is impossible)
550
  - many situations where the "rhs" is known to be a WHNF
Ian Lynagh's avatar
Ian Lynagh committed
551
                (so let-to-case is inappropriate).
552

553 554
Nor does it do the atomic-argument thing

555
\begin{code}
556
completeBind :: SimplEnv
Ian Lynagh's avatar
Ian Lynagh committed
557 558 559 560 561 562 563
             -> 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
564 565

completeBind env top_lvl old_bndr new_bndr new_rhs
566
  | postInlineUnconditionally env top_lvl new_bndr occ_info new_rhs unfolding
Ian Lynagh's avatar
Ian Lynagh committed
567 568 569 570 571 572
                -- 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
573 574 575

  |  otherwise
  = let
Ian Lynagh's avatar
Ian Lynagh committed
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602
        --      Arity info
        new_bndr_info = idInfo new_bndr `setArityInfo` exprArity new_rhs

        --      Unfolding info
        -- 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

        --      Demand info
        -- 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...
        -- After inlining f at some of its call sites the original binding may
        -- (for example) be no longer strictly demanded.
        -- The solution here is a bit ad hoc...
        info_w_unf = new_bndr_info `setUnfoldingInfo` unfolding
                                   `setWorkerInfo`    worker_info

603
        final_info | omit_unfolding             = new_bndr_info
Ian Lynagh's avatar
Ian Lynagh committed
604 605 606 607
                   | isEvaldUnfolding unfolding = zapDemandInfo info_w_unf `orElse` info_w_unf
                   | otherwise                  = info_w_unf

        final_id = new_bndr `setIdInfo` final_info
608
    in
Ian Lynagh's avatar
Ian Lynagh committed
609 610 611
                -- These seqs forces the Id, and hence its IdInfo,
                -- and hence any inner substitutions
    final_id                                    `seq`
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
Ian Lynagh's avatar
Ian Lynagh committed
615
  where
616 617 618 619 620
    unfolding      = mkUnfolding (isTopLevel top_lvl) new_rhs
    worker_info    = substWorker env (workerInfo old_info)
    omit_unfolding = isNonRuleLoopBreaker occ_info || not (activeInline env old_bndr)
    old_info       = idInfo old_bndr
    occ_info       = occInfo old_info
SamB's avatar
SamB committed
621
\end{code}
622 623 624



625
%************************************************************************
Ian Lynagh's avatar
Ian Lynagh committed
626
%*                                                                      *
627
\subsection[Simplify-simplExpr]{The main function: simplExpr}
Ian Lynagh's avatar
Ian Lynagh committed
628
%*                                                                      *
629 630
%************************************************************************

631 632 633 634 635 636
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:

Ian Lynagh's avatar
Ian Lynagh committed
637 638
        let t = f x
        in fst t
639
==>
Ian Lynagh's avatar
Ian Lynagh committed
640 641 642 643
        let t = let a = e1
                    b = e2
                in (a,b)
        in fst t
644
==>
Ian Lynagh's avatar
Ian Lynagh committed
645 646 647 648 649
        let a = e1
            b = e2
            t = (a,b)
        in
        a       -- Can't inline a this round, cos it appears twice
650
==>
Ian Lynagh's avatar
Ian Lynagh committed
651
        e1
652 653 654 655

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

Ian Lynagh's avatar
Ian Lynagh committed
656 657
        let f = g d
        in \x -> ...f...
658
==>
Ian Lynagh's avatar
Ian Lynagh committed
659 660
        let f = let d1 = ..d.. in \y -> e
        in \x -> ...f...
661
==>
Ian Lynagh's avatar
Ian Lynagh committed
662 663
        let d1 = ..d..
        in \x -> ...(\y ->e)...
664

Ian Lynagh's avatar
Ian Lynagh committed
665
Only in this second round can the \y be applied, and it
666 667 668
might do the same again.


669
\begin{code}
670
simplExpr :: SimplEnv -> CoreExpr -> SimplM CoreExpr
671
simplExpr env expr = simplExprC env expr mkBoringStop
672

673
simplExprC :: SimplEnv -> CoreExpr -> SimplCont -> SimplM CoreExpr
Ian Lynagh's avatar
Ian Lynagh committed
674 675
        -- Simplify an expression, given a continuation
simplExprC env expr cont
676
  = -- pprTrace "simplExprC" (ppr expr $$ ppr cont {- $$ ppr (seIdSubst env) -} $$ ppr (seFloats env) ) $
Ian Lynagh's avatar
Ian Lynagh committed
677 678 679 680
    do  { (env', expr') <- simplExprF (zapFloats env) expr cont
        ; -- pprTrace "simplExprC ret" (ppr expr $$ ppr expr') $
          -- pprTrace "simplExprC ret3" (ppr (seInScope env')) $
          -- pprTrace "simplExprC ret4" (ppr (seFloats env')) $
681 682 683 684
          return (wrapFloats env' expr') }

--------------------------------------------------
simplExprF :: SimplEnv -> InExpr -> SimplCont
Ian Lynagh's avatar
Ian Lynagh committed
685
           -> SimplM (SimplEnv, OutExpr)
686

Ian Lynagh's avatar
Ian Lynagh committed
687
simplExprF env e cont
688 689
  = -- pprTrace "simplExprF" (ppr e $$ ppr cont $$ ppr (seTvSubst env) $$ ppr (seIdSubst env) {- $$ ppr (seFloats env) -} ) $
    simplExprF' env e cont
Ian Lynagh's avatar
Ian Lynagh committed
690

Ian Lynagh's avatar
Ian Lynagh committed
691 692
simplExprF' :: SimplEnv -> InExpr -> SimplCont
            -> SimplM (SimplEnv, OutExpr)
Ian Lynagh's avatar
Ian Lynagh committed
693
simplExprF' env (Var v)        cont = simplVar env v cont
694 695 696 697
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 $
Ian Lynagh's avatar
Ian Lynagh committed
698
                                      ApplyTo NoDup arg env cont
699

Ian Lynagh's avatar
Ian Lynagh committed
700
simplExprF' env expr@(Lam _ _) cont
701
  = simplLam env (map zap bndrs) body cont
Ian Lynagh's avatar
Ian Lynagh committed
702 703 704 705 706 707
        -- 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.
708 709 710 711
  where
    n_args   = countArgs cont
    n_params = length bndrs
    (bndrs, body) = collectBinders expr
Ian Lynagh's avatar
Ian Lynagh committed
712 713 714 715 716
    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)
717

718
simplExprF' env (Type ty) cont
719
  = ASSERT( contIsRhsOrArg cont )
Ian Lynagh's avatar
Ian Lynagh committed
720 721
    do  { ty' <- simplType env ty
        ; rebuild env (Type ty') cont }
722

723
simplExprF' env (Case scrut bndr _ alts) cont
724
  | not (switchIsOn (getSwitchChecker env) NoCaseOfCase)
Ian Lynagh's avatar
Ian Lynagh committed
725
  =     -- Simplify the scrutinee with a Select continuation
726
    simplExprF env scrut (Select NoDup bndr alts env cont)
727

728
  | otherwise
Ian Lynagh's avatar
Ian Lynagh committed
729 730 731 732
  =     -- If case-of-case is off, simply simplify the case expression
        -- in a vanilla Stop context, and rebuild the result around it
    do  { case_expr' <- simplExprC env scrut case_cont
        ; rebuild env case_expr' cont }
733
  where
734
    case_cont = Select NoDup bndr alts env mkBoringStop
735

736
simplExprF' env (Let (Rec pairs) body) cont
Ian Lynagh's avatar
Ian Lynagh committed
737
  = do  { env' <- simplRecBndrs env (map fst pairs)
Ian Lynagh's avatar
Ian Lynagh committed
738 739
                -- NB: bndrs' don't have unfoldings or rules
                -- We add them as we go down
740

Ian Lynagh's avatar
Ian Lynagh committed
741 742
        ; env'' <- simplRecBind env' NotTopLevel pairs
        ; simplExprF env'' body cont }
743

744 745
simplExprF' env (Let (NonRec bndr rhs) body) cont
  = simplNonRecE env bndr (rhs, env) ([], body) cont
746 747

---------------------------------
748
simplType :: SimplEnv -> InType -> SimplM OutType
Ian Lynagh's avatar
Ian Lynagh committed
749
        -- Kept monadic just so we can do the seqType
750
simplType env ty
751
  = -- pprTrace "simplType" (ppr ty $$ ppr (seTvSubst env)) $
752
    seqType new_ty   `seq`   return new_ty
753
  where
754
    new_ty = substTy env ty
755 756 757
\end{code}


758
%************************************************************************
Ian Lynagh's avatar
Ian Lynagh committed
759
%*                                                                      *
760
\subsection{The main rebuilder}
Ian Lynagh's avatar
Ian Lynagh committed
761
%*                                                                      *
762 763 764 765 766 767
%************************************************************************

\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
Ian Lynagh's avatar
Ian Lynagh committed
768 769 770
rebuild env expr cont0
  = -- pprTrace "rebuild" (ppr expr $$ ppr cont0 $$ ppr (seFloats env)) $
    case cont0 of
Ian Lynagh's avatar
Ian Lynagh committed
771 772
      Stop {}                      -> return (env, expr)
      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
Ian Lynagh's avatar
Ian Lynagh committed
776 777 778
                                         ; simplLam env' bs body cont }
      ApplyTo _ arg se cont        -> do { arg' <- simplExpr (se `setInScope` env) arg
                                         ; rebuild env (App expr arg') cont }
779 780 781
\end{code}


782
%************************************************************************
Ian Lynagh's avatar
Ian Lynagh committed
783
%*                                                                      *
784
\subsection{Lambdas}
Ian Lynagh's avatar
Ian Lynagh committed
785
%*                                                                      *
786 787 788
%************************************************************************

\begin{code}
789
simplCast :: SimplEnv -> InExpr -> Coercion -> SimplCont
Ian Lynagh's avatar
Ian Lynagh committed
790
          -> SimplM (SimplEnv, OutExpr)
Ian Lynagh's avatar
Ian Lynagh committed
791 792 793
simplCast env body co0 cont0
  = do  { co1 <- simplType env co0
        ; simplExprF env body (addCoerce co1 cont0) }
794
  where
795 796
       addCoerce co cont = add_coerce co (coercionKind co) cont

Ian Lynagh's avatar
Ian Lynagh committed
797
       add_coerce _co (s1, k1) cont     -- co :: ty~ty
Ian Lynagh's avatar
Ian Lynagh committed
798
         | s1 `coreEqType` k1 = cont    -- is a no-op
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
799

Ian Lynagh's avatar
Ian Lynagh committed
800 801
       add_coerce co1 (s1, _k2) (CoerceIt co2 cont)
         | (_l1, t1) <- coercionKind co2
Ian Lynagh's avatar
Ian Lynagh committed
802 803 804 805 806 807 808 809 810 811
                --      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
812
         | otherwise           = CoerceIt (mkTransCoercion co1 co2) cont
Ian Lynagh's avatar
Ian Lynagh committed
813

Ian Lynagh's avatar
Ian Lynagh committed
814
       add_coerce co (s1s2, _t1t2) (ApplyTo dup (Type arg_ty) arg_se cont)
Ian Lynagh's avatar
Ian Lynagh committed
815 816 817 818 819 820
                -- (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
821
           ty' = substTy (arg_se `setInScope` env) arg_ty
822

Ian Lynagh's avatar
Ian Lynagh committed
823
        -- ToDo: the PushC rule is not implemented at all
824

Ian Lynagh's avatar
Ian Lynagh committed
825
       add_coerce co (s1s2, _t1t2) (ApplyTo dup arg arg_se cont)
826
         | not (isTypeArg arg)  -- This implements the Push rule from the paper
Ian Lynagh's avatar
Ian Lynagh committed
827
         , isFunTy s1s2   -- t1t2 must be a function type, becuase it's applied
828
                -- co : s1s2 :=: t1t2
Ian Lynagh's avatar
Ian Lynagh committed
829 830 831 832 833 834 835 836 837 838 839 840 841
                --      (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.
                --
                -- Example of use: Trac #995
842
         = ApplyTo dup new_arg (zapSubstEnv env) (addCoerce co2 cont)
843
         where
Ian Lynagh's avatar
Ian Lynagh committed
844 845
           -- we split coercion t1->t2 :=: s1->s2 into t1 :=: s1 and
           -- t2 :=: s2 with left and right on the curried form:
846 847
           --    (->) t1 t2 :=: (->) s1 s2
           [co1, co2] = decomposeCo 2 co
848
           new_arg    = mkCoerce (mkSymCoercion co1) arg'
849
           arg'       = substExpr (arg_se `setInScope` env) arg
850

851
       add_coerce co _ cont = CoerceIt co cont
852 853
\end{code}

854

855
%************************************************************************
Ian Lynagh's avatar
Ian Lynagh committed
856
%*                                                                      *
857
\subsection{Lambdas}
Ian Lynagh's avatar
Ian Lynagh committed
858
%*                                                                      *
859
%************************************************************************
860 861

\begin{code}
862
simplLam :: SimplEnv -> [InId] -> InExpr -> SimplCont
Ian Lynagh's avatar
Ian Lynagh committed
863
         -> SimplM (SimplEnv, OutExpr)
864 865

simplLam env [] body cont = simplExprF env body cont
866

Ian Lynagh's avatar
Ian Lynagh committed
867
        -- Type-beta reduction
868 869
simplLam env (bndr:bndrs) body (ApplyTo _ (Type ty_arg) arg_se cont)
  = ASSERT( isTyVar bndr )
Ian Lynagh's avatar
Ian Lynagh committed
870 871 872
    do  { tick (BetaReduction bndr)
        ; ty_arg' <- simplType (arg_se `setInScope` env) ty_arg
        ; simplLam (extendTvSubst env bndr ty_arg') bndrs body cont }
873

Ian Lynagh's avatar
Ian Lynagh committed
874
        -- Ordinary beta reduction
875
simplLam env (bndr:bndrs) body (ApplyTo _ arg arg_se cont)
Ian Lynagh's avatar
Ian Lynagh committed
876 877
  = do  { tick (BetaReduction bndr)
        ; simplNonRecE env bndr (arg, arg_se) (bndrs, body) cont }
878

Ian Lynagh's avatar
Ian Lynagh committed
879
        -- Not enough args, so there are real lambdas left to put in the result
880
simplLam env bndrs body cont
Ian Lynagh's avatar
Ian Lynagh committed
881 882
  = do  { (env', bndrs') <- simplLamBndrs env bndrs
        ; body' <- simplExpr env' body
Ian Lynagh's avatar
Ian Lynagh committed
883
        ; new_lam <- mkLam bndrs' body'
Ian Lynagh's avatar
Ian Lynagh committed
884
        ; rebuild env' new_lam cont }
885 886

------------------
Ian Lynagh's avatar
Ian Lynagh committed
887 888 889
simplNonRecE :: SimplEnv
             -> InId                    -- The binder
             -> (InExpr, SimplEnv)      -- Rhs of binding (or arg of lambda)
890
             -> ([InBndr], InExpr)      -- Body of the let/lambda
Ian Lynagh's avatar
Ian Lynagh committed
891 892 893
                                        --      \xs.e
             -> SimplCont
             -> SimplM (SimplEnv, OutExpr)
894 895 896 897 898 899 900 901 902 903

-- 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
Ian Lynagh's avatar
Ian Lynagh committed
904 905
-- Why?  Because of the binder-occ-info-zapping done before
--       the call to simplLam in simplExprF (Lam ...)
906

907 908 909 910 911
	-- First deal with type lets: let a = Type ty in b
simplNonRecE env bndr (Type ty_arg, rhs_se) (bndrs, body) cont
  = do	{ ty_arg' <- simplType (rhs_se `setInScope` env) ty_arg
	; simplLam (extendTvSubst env bndr ty_arg') bndrs body cont }

912 913
simplNonRecE env bndr (rhs, rhs_se) (bndrs, body) cont
  | preInlineUnconditionally env NotTopLevel bndr rhs
Ian Lynagh's avatar
Ian Lynagh committed
914 915
  = do  { tick (PreInlineUnconditionally bndr)
        ; simplLam (extendIdSubst env bndr (mkContEx rhs_se rhs)) bndrs body cont }
916

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
917
  | isStrictId bndr
Ian Lynagh's avatar
Ian Lynagh committed
918 919
  = do  { simplExprF (rhs_se `setFloats` env) rhs
                     (StrictBind bndr bndrs body env cont) }
920 921

  | otherwise