TidyPgm.hs 61.9 KB
Newer Older
Austin Seipp's avatar
Austin Seipp committed
1 2
{-
(c) The GRASP/AQUA Project, Glasgow University, 1992-1998
Simon Marlow's avatar
Simon Marlow committed
3

4
\section{Tidying up Core}
Austin Seipp's avatar
Austin Seipp committed
5
-}
6

7
{-# LANGUAGE CPP, ViewPatterns #-}
8

9 10 11
module TidyPgm (
       mkBootModDetailsTc, tidyProgram, globaliseAndTidyId
   ) where
12 13 14

#include "HsVersions.h"

Simon Marlow's avatar
Simon Marlow committed
15
import TcRnTypes
Simon Marlow's avatar
Simon Marlow committed
16
import DynFlags
17
import CoreSyn
Simon Marlow's avatar
Simon Marlow committed
18 19 20
import CoreUnfold
import CoreFVs
import CoreTidy
21
import CoreMonad
22
import CorePrep
23
import CoreUtils        (rhsIsStatic)
24
import CoreStats        (coreBindsStats, CoreStats(..))
25
import CoreLint
26
import Literal
27
import Rules
28 29
import PatSyn
import ConLike
30
import CoreArity        ( exprArity, exprBotStrictness_maybe )
31
import StaticPtrTable
32 33
import VarEnv
import VarSet
34
import Var
Simon Marlow's avatar
Simon Marlow committed
35
import Id
36
import MkId             ( mkDictSelRhs )
Simon Marlow's avatar
Simon Marlow committed
37 38
import IdInfo
import InstEnv
39
import FamInstEnv
Simon Peyton Jones's avatar
Simon Peyton Jones committed
40
import Type             ( tidyTopType )
Joachim Breitner's avatar
Joachim Breitner committed
41
import Demand           ( appIsBottom, isTopSig, isBottomingSig )
Simon Marlow's avatar
Simon Marlow committed
42
import BasicTypes
43
import Name hiding (varName)
Simon Marlow's avatar
Simon Marlow committed
44 45
import NameSet
import NameEnv
46
import NameCache
47 48
import Avail
import IfaceEnv
49 50
import TcEnv
import TcRnMonad
Simon Marlow's avatar
Simon Marlow committed
51 52
import DataCon
import TyCon
53
import Class
54
import Module
55
import Packages( isDllName )
56
import HscTypes
Simon Marlow's avatar
Simon Marlow committed
57 58
import Maybes
import UniqSupply
59
import ErrUtils (Severity(..))
60
import Outputable
niteria's avatar
niteria committed
61
import UniqDFM
62
import SrcLoc
63
import qualified ErrUtils as Err
Simon Marlow's avatar
Simon Marlow committed
64

65
import Control.Monad
Ian Lynagh's avatar
Ian Lynagh committed
66
import Data.Function
67
import Data.List        ( sortBy )
68
import Data.IORef       ( atomicModifyIORef' )
69

Austin Seipp's avatar
Austin Seipp committed
70
{-
71 72 73
Constructing the TypeEnv, Instances, Rules, VectInfo from which the
ModIface is constructed, and which goes on to subsequent modules in
--make mode.
74 75 76 77 78 79 80 81

Most of the interface file is obtained simply by serialising the
TypeEnv.  One important consequence is that if the *interface file*
has pragma info if and only if the final TypeEnv does. This is not so
important for *this* module, but it's essential for ghc --make:
subsequent compilations must not see (e.g.) the arity if the interface
file does not contain arity If they do, they'll exploit the arity;
then the arity might change, but the iface file doesn't change =>
82
recompilation does not happen => disaster.
83 84 85 86 87 88

For data types, the final TypeEnv will have a TyThing for the TyCon,
plus one for each DataCon; the interface file will contain just one
data type declaration, but it is de-serialised back into a collection
of TyThings.

Austin Seipp's avatar
Austin Seipp committed
89 90
************************************************************************
*                                                                      *
91
                Plan A: simpleTidyPgm
Austin Seipp's avatar
Austin Seipp committed
92 93
*                                                                      *
************************************************************************
94 95


96
Plan A: mkBootModDetails: omit pragmas, make interfaces small
97 98 99
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* Ignore the bindings

100 101
* Drop all WiredIn things from the TypeEnv
        (we never want them in interface files)
102 103

* Retain all TyCons and Classes in the TypeEnv, to avoid
104 105
        having to find which ones are mentioned in the
        types of exported Ids
106 107

* Trim off the constructors of non-exported TyCons, both
108
        from the TyCon and from the TypeEnv
109 110 111

* Drop non-exported Ids from the TypeEnv

112
* Tidy the types of the DFunIds of Instances,
113 114 115 116 117 118 119 120
  make them into GlobalIds, (they already have External Names)
  and add them to the TypeEnv

* Tidy the types of the (exported) Ids in the TypeEnv,
  make them into GlobalIds (they already have External Names)

* Drop rules altogether

121
* Tidy the bindings, to ensure that the Caf and Arity
122
  information is correct for each top-level binder; the
123 124
  code generator needs it. And to ensure that local names have
  distinct OccNames in case of object-file splitting
125

126 127
* If this an hsig file, drop the instances altogether too (they'll
  get pulled in by the implicit module import.
Austin Seipp's avatar
Austin Seipp committed
128
-}
129

130 131
-- This is Plan A: make a small type env when typechecking only,
-- or when compiling a hs-boot file, or simply when not using -O
132 133 134
--
-- We don't look at the bindings at all -- there aren't any
-- for hs-boot files
135

Simon Marlow's avatar
Simon Marlow committed
136
mkBootModDetailsTc :: HscEnv -> TcGblEnv -> IO ModDetails
137
mkBootModDetailsTc hsc_env
Simon Marlow's avatar
Simon Marlow committed
138
        TcGblEnv{ tcg_exports   = exports,
139 140
                  tcg_type_env  = type_env, -- just for the Ids
                  tcg_tcs       = tcs,
141
                  tcg_patsyns   = pat_syns,
Simon Marlow's avatar
Simon Marlow committed
142
                  tcg_insts     = insts,
143 144
                  tcg_fam_insts = fam_insts,
                  tcg_mod       = this_mod
Simon Marlow's avatar
Simon Marlow committed
145
                }
146 147 148 149 150 151
  = -- This timing isn't terribly useful since the result isn't forced, but
    -- the message is useful to locating oneself in the compilation process.
    Err.withTiming (pure dflags)
                   (text "CoreTidy"<+>brackets (ppr this_mod))
                   (const ()) $
    do  { let { insts'     = map (tidyClsInstDFun globaliseAndTidyId) insts
Simon Peyton Jones's avatar
Simon Peyton Jones committed
152
              ; pat_syns'  = map (tidyPatSynIds   globaliseAndTidyId) pat_syns
153
              ; type_env1  = mkBootTypeEnv (availsToNameSet exports)
154
                                           (typeEnvIds type_env) tcs fam_insts
155
              ; type_env2  = extendTypeEnvWithPatSyns pat_syns' type_env1
156
              ; dfun_ids   = map instanceDFunId insts'
157
              ; type_env'  = extendTypeEnvWithIds type_env2 dfun_ids
158 159 160 161 162 163 164
              }
        ; return (ModDetails { md_types     = type_env'
                             , md_insts     = insts'
                             , md_fam_insts = fam_insts
                             , md_rules     = []
                             , md_anns      = []
                             , md_exports   = exports
chak@cse.unsw.edu.au.'s avatar
chak@cse.unsw.edu.au. committed
165
                             , md_vect_info = noVectInfo
166
                             })
167
        }
168
  where
169
    dflags = hsc_dflags hsc_env
170

171
mkBootTypeEnv :: NameSet -> [Id] -> [TyCon] -> [FamInst] -> TypeEnv
dreixel's avatar
dreixel committed
172
mkBootTypeEnv exports ids tcs fam_insts
173
  = tidyTypeEnv True $
dreixel's avatar
dreixel committed
174
       typeEnvFromEntities final_ids tcs fam_insts
175
  where
176
        -- Find the LocalIds in the type env that are exported
177 178 179 180 181 182
        -- Make them into GlobalIds, and tidy their types
        --
        -- It's very important to remove the non-exported ones
        -- because we don't tidy the OccNames, and if we don't remove
        -- the non-exported ones we'll get many things with the
        -- same name in the interface file, giving chaos.
183 184 185 186
        --
        -- Do make sure that we keep Ids that are already Global.
        -- When typechecking an .hs-boot file, the Ids come through as
        -- GlobalIds.
187 188 189
    final_ids = [ (if isLocalId id then globaliseAndTidyId id
                                   else id)
                        `setIdUnfolding` BootUnfolding
190 191
                | id <- ids
                , keep_it id ]
192 193 194 195 196

        -- default methods have their export flag set, but everything
        -- else doesn't (yet), because this is pre-desugaring, so we
        -- must test both.
    keep_it id = isExportedId id || idName id `elemNameSet` exports
197 198


199 200

globaliseAndTidyId :: Id -> Id
201 202
-- Takes an LocalId with an External Name,
-- makes it into a GlobalId
203 204 205
--     * unchanged Name (might be Internal or External)
--     * unchanged details
--     * VanillaIdInfo (makes a conservative assumption about Caf-hood)
206
globaliseAndTidyId id
207 208 209
  = Id.setIdType (globaliseId id) tidy_type
  where
    tidy_type = tidyTopType (idType id)
210

Austin Seipp's avatar
Austin Seipp committed
211 212 213
{-
************************************************************************
*                                                                      *
214
        Plan B: tidy bindings, make TypeEnv full of IdInfo
Austin Seipp's avatar
Austin Seipp committed
215 216
*                                                                      *
************************************************************************
217

218
Plan B: include pragmas, make interfaces
219 220 221 222
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* Figure out which Ids are externally visible

* Tidy the bindings, externalising appropriate Ids
223

224
* Drop all Ids from the TypeEnv, and add all the External Ids from
225 226
  the bindings.  (This adds their IdInfo to the TypeEnv; and adds
  floated-out Ids that weren't even in the TypeEnv before.)
227 228 229

Step 1: Figure out external Ids
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
230 231
Note [choosing external names]

Simon Marlow's avatar
Simon Marlow committed
232 233
See also the section "Interface stability" in the
RecompilationAvoidance commentary:
234
  http://ghc.haskell.org/trac/ghc/wiki/Commentary/Compiler/RecompilationAvoidance
Simon Marlow's avatar
Simon Marlow committed
235

236 237 238
First we figure out which Ids are "external" Ids.  An
"external" Id is one that is visible from outside the compilation
unit.  These are
239
  a) the user exported ones
240 241
  b) the ones bound to static forms
  c) ones mentioned in the unfoldings, workers,
242 243
     rules of externally-visible ones ,
     or vectorised versions of externally-visible ones
244

245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269
While figuring out which Ids are external, we pick a "tidy" OccName
for each one.  That is, we make its OccName distinct from the other
external OccNames in this module, so that in interface files and
object code we can refer to it unambiguously by its OccName.  The
OccName for each binder is prefixed by the name of the exported Id
that references it; e.g. if "f" references "x" in its unfolding, then
"x" is renamed to "f_x".  This helps distinguish the different "x"s
from each other, and means that if "f" is later removed, things that
depend on the other "x"s will not need to be recompiled.  Of course,
if there are multiple "f_x"s, then we have to disambiguate somehow; we
use "f_x0", "f_x1" etc.

As far as possible we should assign names in a deterministic fashion.
Each time this module is compiled with the same options, we should end
up with the same set of external names with the same types.  That is,
the ABI hash in the interface should not change.  This turns out to be
quite tricky, since the order of the bindings going into the tidy
phase is already non-deterministic, as it is based on the ordering of
Uniques, which are assigned unpredictably.

To name things in a stable way, we do a depth-first-search of the
bindings, starting from the exports sorted by name.  This way, as long
as the bindings themselves are deterministic (they sometimes aren't!),
the order in which they are presented to the tidying phase does not
affect the names we assign.
270 271 272 273 274 275

Step 2: Tidy the program
~~~~~~~~~~~~~~~~~~~~~~~~
Next we traverse the bindings top to bottom.  For each *top-level*
binder

276 277
 1. Make it into a GlobalId; its IdDetails becomes VanillaGlobal,
    reflecting the fact that from now on we regard it as a global,
278
    not local, Id
279 280 281 282 283

 2. Give it a system-wide Unique.
    [Even non-exported things need system-wide Uniques because the
    byte-code generator builds a single Name->BCO symbol table.]

284
    We use the NameCache kept in the HscEnv as the
285 286
    source of such system-wide uniques.

287
    For external Ids, use the original-name cache in the NameCache
288
    to ensure that the unique assigned is the same as the Id had
289
    in any previous compilation run.
290 291 292 293 294 295

 3. Rename top-level Ids according to the names we chose in step 1.
    If it's an external Id, make it have a External Name, otherwise
    make it have an Internal Name.  This is used by the code generator
    to decide whether to make the label externally visible

296 297 298 299 300 301
 4. Give it its UTTERLY FINAL IdInfo; in ptic,
        * its unfolding, if it should have one

        * its arity, computed from the number of visible lambdas

        * its CAF info, computed from what is free in its RHS
302 303 304 305 306


Finally, substitute these new top-level binders consistently
throughout, including in unfoldings.  We also tidy binders in
RHSs, so that they print nicely in interfaces.
Austin Seipp's avatar
Austin Seipp committed
307
-}
308

309
tidyProgram :: HscEnv -> ModGuts -> IO (CgGuts, ModDetails)
310 311
tidyProgram hsc_env  (ModGuts { mg_module    = mod
                              , mg_exports   = exports
312
                              , mg_rdr_env   = rdr_env
313
                              , mg_tcs       = tcs
314
                              , mg_insts     = cls_insts
315 316
                              , mg_fam_insts = fam_insts
                              , mg_binds     = binds
cactus's avatar
cactus committed
317
                              , mg_patsyns   = patsyns
318 319 320
                              , mg_rules     = imp_rules
                              , mg_vect_info = vect_info
                              , mg_anns      = anns
321
                              , mg_deps      = deps
322 323
                              , mg_foreign   = foreign_stubs
                              , mg_hpc_info  = hpc_info
324
                              , mg_modBreaks = modBreaks
325 326
                              })

327 328 329 330
  = Err.withTiming (pure dflags)
                   (text "CoreTidy"<+>brackets (ppr mod))
                   (const ()) $
    do  { let { omit_prags = gopt Opt_OmitInterfacePragmas dflags
ian@well-typed.com's avatar
ian@well-typed.com committed
331
              ; expose_all = gopt Opt_ExposeAllUnfoldings  dflags
332
              ; print_unqual = mkPrintUnqualified dflags rdr_env
333 334
              }

dreixel's avatar
dreixel committed
335
        ; let { type_env = typeEnvFromEntities [] tcs fam_insts
336 337 338 339 340

              ; implicit_binds
                  = concatMap getClassImplicitBinds (typeEnvClasses type_env) ++
                    concatMap getTyConImplicitBinds (typeEnvTyCons type_env)
              }
341

342
        ; (unfold_env, tidy_occ_env)
343
              <- chooseExternalIds hsc_env mod omit_prags expose_all
344
                                   binds implicit_binds imp_rules (vectInfoVar vect_info)
Austin Seipp's avatar
Austin Seipp committed
345
        ; let { (trimmed_binds, trimmed_rules)
346
                    = findExternalRules omit_prags binds imp_rules unfold_env }
347

348
        ; (tidy_env, tidy_binds)
349
                 <- tidyTopBinds hsc_env mod unfold_env tidy_occ_env trimmed_binds
350

351
        ; let { final_ids  = [ id | id <- bindersOfBinds tidy_binds,
352
                                    isExternalName (idName id)]
353
              ; type_env1  = extendTypeEnvWithIds type_env final_ids
354

Simon Peyton Jones's avatar
Simon Peyton Jones committed
355
              ; tidy_cls_insts = map (tidyClsInstDFun (tidyVarOcc tidy_env)) cls_insts
356 357 358
                -- A DFunId will have a binding in tidy_binds, and so will now be in
                -- tidy_type_env, replete with IdInfo.  Its name will be unchanged since
                -- it was born, but we want Global, IdInfo-rich (or not) DFunId in the
359
                -- tidy_cls_insts.  Similarly the Ids inside a PatSyn.
360

361
              ; tidy_rules = tidyRules tidy_env trimmed_rules
362 363 364
                -- You might worry that the tidy_env contains IdInfo-rich stuff
                -- and indeed it does, but if omit_prags is on, ext_rules is
                -- empty
365

366 367
              ; tidy_vect_info = tidyVectInfo tidy_env vect_info

368 369 370 371
                -- Tidy the Ids inside each PatSyn, very similarly to DFunIds
                -- and then override the PatSyns in the type_env with the new tidy ones
                -- This is really the only reason we keep mg_patsyns at all; otherwise
                -- they could just stay in type_env
Simon Peyton Jones's avatar
Simon Peyton Jones committed
372
              ; tidy_patsyns = map (tidyPatSynIds (tidyVarOcc tidy_env)) patsyns
373
              ; type_env2    = extendTypeEnvWithPatSyns tidy_patsyns type_env1
374 375

              ; tidy_type_env = tidyTypeEnv omit_prags type_env2
376 377 378 379 380 381
              }
          -- See Note [Grand plan for static forms] in StaticPtrTable.
        ; (spt_init_code, tidy_binds') <-
             sptCreateStaticBinds hsc_env mod tidy_binds
        ; let { -- See Note [Injecting implicit bindings]
                all_tidy_binds = implicit_binds ++ tidy_binds'
382 383

              -- Get the TyCons to generate code for.  Careful!  We must use
384 385 386 387 388 389 390
              -- the untidied TypeEnv here, because we need
              --  (a) implicit TyCons arising from types and classes defined
              --      in this module
              --  (b) wired-in TyCons, which are normally removed from the
              --      TypeEnv we put in the ModDetails
              --  (c) Constructors even if they are not exported (the
              --      tidied TypeEnv has trimmed these away)
391 392
              ; alg_tycons = filter isAlgTyCon (typeEnvTyCons type_env)
              }
393

394
        ; endPassIO hsc_env print_unqual CoreTidy all_tidy_binds tidy_rules
395

396 397
          -- If the endPass didn't print the rules, but ddump-rules is
          -- on, print now
398
        ; unless (dopt Opt_D_dump_simpl dflags) $
399
            Err.dumpIfSet_dyn dflags Opt_D_dump_rules
400
              (showSDoc dflags (ppr CoreTidy <+> text "rules"))
401
              (pprRulesForUser tidy_rules)
402

403 404
          -- Print one-line size info
        ; let cs = coreBindsStats tidy_binds
405
        ; when (dopt Opt_D_dump_core_stats dflags)
406
               (log_action dflags dflags NoReason SevDump noSrcSpan defaultDumpStyle
407
                          (text "Tidy size (terms,types,coercions)"
408 409 410
                           <+> ppr (moduleName mod) <> colon
                           <+> int (cs_tm cs)
                           <+> int (cs_ty cs)
411 412
                           <+> int (cs_co cs) ))

413 414 415
        ; return (CgGuts { cg_module   = mod,
                           cg_tycons   = alg_tycons,
                           cg_binds    = all_tidy_binds,
416
                           cg_foreign  = foreign_stubs `appendStubC`
417
                                         spt_init_code,
418 419
                           cg_dep_pkgs = map fst $ dep_pkgs deps,
                           cg_hpc_info = hpc_info,
420
                           cg_modBreaks = modBreaks },
421

422
                   ModDetails { md_types     = tidy_type_env,
423
                                md_rules     = tidy_rules,
424
                                md_insts     = tidy_cls_insts,
425 426
                                md_vect_info = tidy_vect_info,
                                md_fam_insts = fam_insts,
427 428
                                md_exports   = exports,
                                md_anns      = anns      -- are already tidy
chak@cse.unsw.edu.au.'s avatar
chak@cse.unsw.edu.au. committed
429
                              })
430
        }
431 432
  where
    dflags = hsc_dflags hsc_env
433

434
tidyTypeEnv :: Bool       -- Compiling without -O, so omit prags
435
            -> TypeEnv -> TypeEnv
436

437
-- The competed type environment is gotten from
438 439
--      a) the types and classes defined here (plus implicit things)
--      b) adding Ids with correct IdInfo, including unfoldings,
440
--              gotten from the bindings
441
-- From (b) we keep only those Ids with External names;
442 443 444
--          the CoreTidy pass makes sure these are all and only
--          the externally-accessible ones
-- This truncates the type environment to include only the
445
-- exported Ids and things needed from them, which saves space
446 447
--
-- See Note [Don't attempt to trim data types]
448

449
tidyTypeEnv omit_prags type_env
450 451 452
 = let
        type_env1 = filterNameEnv (not . isWiredInName . getName) type_env
          -- (1) remove wired-in things
453
        type_env2 | omit_prags = mapNameEnv trimThing type_env1
454 455 456 457
                  | otherwise  = type_env1
          -- (2) trimmed if necessary
    in
    type_env2
458 459

--------------------------
460
trimThing :: TyThing -> TyThing
461
-- Trim off inessentials, for boot files and no -O
462
trimThing (AnId id)
463
   | not (isImplicitId id)
464
   = AnId (id `setIdInfo` vanillaIdInfo)
465

466
trimThing other_thing
467
  = other_thing
468 469 470 471

extendTypeEnvWithPatSyns :: [PatSyn] -> TypeEnv -> TypeEnv
extendTypeEnvWithPatSyns tidy_patsyns type_env
  = extendTypeEnvList type_env [AConLike (PatSynCon ps) | ps <- tidy_patsyns ]
472

473
tidyVectInfo :: TidyEnv -> VectInfo -> VectInfo
474
tidyVectInfo (_, var_env) info@(VectInfo { vectInfoVar          = vars
475
                                         , vectInfoParallelVars = parallelVars
476 477
                                         })
  = info { vectInfoVar          = tidy_vars
478
         , vectInfoParallelVars = tidy_parallelVars
479
         }
480
  where
481 482
      -- we only export mappings whose domain and co-domain is exported (otherwise, the iface is
      -- inconsistent)
niteria's avatar
niteria committed
483 484 485 486 487 488 489 490
    tidy_vars = mkDVarEnv [ (tidy_var, (tidy_var, tidy_var_v))
                          | (var, var_v) <- eltsUDFM vars
                          , let tidy_var   = lookup_var var
                                tidy_var_v = lookup_var var_v
                          , isExternalId tidy_var   && isExportedId tidy_var
                          , isExternalId tidy_var_v && isExportedId tidy_var_v
                          , isDataConWorkId var || not (isImplicitId var)
                          ]
491

niteria's avatar
niteria committed
492 493 494 495 496 497
    tidy_parallelVars = mkDVarSet
                          [ tidy_var
                          | var <- dVarSetElems parallelVars
                          , let tidy_var = lookup_var var
                          , isExternalId tidy_var && isExportedId tidy_var
                          ]
498

499
    lookup_var var = lookupWithDefaultVarEnv var_env var var
Austin Seipp's avatar
Austin Seipp committed
500

501 502
    -- We need to make sure that all names getting into the iface version of 'VectInfo' are
    -- external; otherwise, 'MkIface' will bomb out.
503
    isExternalId = isExternalName . idName
504

Austin Seipp's avatar
Austin Seipp committed
505
{-
506 507 508 509
Note [Don't attempt to trim data types]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
For some time GHC tried to avoid exporting the data constructors
of a data type if it wasn't strictly necessary to do so; see Trac #835.
Austin Seipp's avatar
Austin Seipp committed
510
But "strictly necessary" accumulated a longer and longer list
Gabor Greif's avatar
typos  
Gabor Greif committed
511
of exceptions, and finally I gave up the battle:
512 513 514 515 516 517

    commit 9a20e540754fc2af74c2e7392f2786a81d8d5f11
    Author: Simon Peyton Jones <simonpj@microsoft.com>
    Date:   Thu Dec 6 16:03:16 2012 +0000

    Stop attempting to "trim" data types in interface files
Austin Seipp's avatar
Austin Seipp committed
518

519 520 521 522
    Without -O, we previously tried to make interface files smaller
    by not including the data constructors of data types.  But
    there are a lot of exceptions, notably when Template Haskell is
    involved or, more recently, DataKinds.
Austin Seipp's avatar
Austin Seipp committed
523

524 525 526
    However Trac #7445 shows that even without TemplateHaskell, using
    the Data class and invoking Language.Haskell.TH.Quote.dataToExpQ
    is enough to require us to expose the data constructors.
Austin Seipp's avatar
Austin Seipp committed
527

528 529 530 531 532
    So I've given up on this "optimisation" -- it's probably not
    important anyway.  Now I'm simply not attempting to trim off
    the data constructors.  The gain in simplicity is worth the
    modest cost in interface file growth, which is limited to the
    bits reqd to describe those data constructors.
533

Austin Seipp's avatar
Austin Seipp committed
534 535
************************************************************************
*                                                                      *
536
        Implicit bindings
Austin Seipp's avatar
Austin Seipp committed
537 538
*                                                                      *
************************************************************************
539 540 541

Note [Injecting implicit bindings]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Edward Z. Yang's avatar
Edward Z. Yang committed
542
We inject the implicit bindings right at the end, in CoreTidy.
543 544
Some of these bindings, notably record selectors, are not
constructed in an optimised form.  E.g. record selector for
545
        data T = MkT { x :: {-# UNPACK #-} !Int }
546
Then the unfolding looks like
547
        x = \t. case t of MkT x1 -> let x = I# x1 in x
548
This generates bad code unless it's first simplified a bit.  That is
549
why CoreUnfold.mkImplicitUnfolding uses simpleOptExpr to do a bit of
550 551 552
optimisation first.  (Only matters when the selector is used curried;
eg map x ys.)  See Trac #2070.

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
553 554 555 556 557
[Oct 09: in fact, record selectors are no longer implicit Ids at all,
because we really do want to optimise them properly. They are treated
much like any other Id.  But doing "light" optimisation on an implicit
Id still makes sense.]

558 559
At one time I tried injecting the implicit bindings *early*, at the
beginning of SimplCore.  But that gave rise to real difficulty,
Gabor Greif's avatar
typos  
Gabor Greif committed
560
because GlobalIds are supposed to have *fixed* IdInfo, but the
561 562 563 564 565 566
simplifier and other core-to-core passes mess with IdInfo all the
time.  The straw that broke the camels back was when a class selector
got the wrong arity -- ie the simplifier gave it arity 2, whereas
importing modules were expecting it to have arity 1 (Trac #2844).
It's much safer just to inject them right at the end, after tidying.

567
Oh: two other reasons for injecting them late:
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
568

569 570
  - If implicit Ids are already in the bindings when we start TidyPgm,
    we'd have to be careful not to treat them as external Ids (in
571
    the sense of chooseExternalIds); else the Ids mentioned in *their*
572
    RHSs will be treated as external and you get an interface file
573
    saying      a18 = <blah>
574
    but nothing referring to a18 (because the implicit Id is the
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
575
    one that does, and implicit Ids don't appear in interface files).
576 577 578 579

  - More seriously, the tidied type-envt will include the implicit
    Id replete with a18 in its unfolding; but we won't take account
    of a18 when computing a fingerprint for the class; result chaos.
580

simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
581 582 583
There is one sort of implicit binding that is injected still later,
namely those for data constructor workers. Reason (I think): it's
really just a code generation trick.... binding itself makes no sense.
584
See Note [Data constructor workers] in CorePrep.
Austin Seipp's avatar
Austin Seipp committed
585
-}
586

587
getTyConImplicitBinds :: TyCon -> [CoreBind]
Icelandjack's avatar
Icelandjack committed
588
getTyConImplicitBinds tc = map get_defn (mapMaybe dataConWrapId_maybe (tyConDataCons tc))
589 590

getClassImplicitBinds :: Class -> [CoreBind]
591 592 593
getClassImplicitBinds cls
  = [ NonRec op (mkDictSelRhs cls val_index)
    | (op, val_index) <- classAllSelIds cls `zip` [0..] ]
594 595 596

get_defn :: Id -> CoreBind
get_defn id = NonRec id (unfoldingTemplate (realIdUnfolding id))
597

Austin Seipp's avatar
Austin Seipp committed
598 599 600
{-
************************************************************************
*                                                                      *
601
\subsection{Step 1: finding externals}
Austin Seipp's avatar
Austin Seipp committed
602 603
*                                                                      *
************************************************************************
604

605
See Note [Choosing external names].
Austin Seipp's avatar
Austin Seipp committed
606
-}
607 608

type UnfoldEnv  = IdEnv (Name{-new name-}, Bool {-show unfolding-})
609 610
  -- Maps each top-level Id to its new Name (the Id is tidied in step 2)
  -- The Unique is unchanged.  If the new Name is external, it will be
611
  -- visible in the interface file.
Simon Marlow's avatar
Simon Marlow committed
612 613
  --
  -- Bool => expose unfolding or not.
614 615 616

chooseExternalIds :: HscEnv
                  -> Module
617
                  -> Bool -> Bool
618
                  -> [CoreBind]
619 620
                  -> [CoreBind]
                  -> [CoreRule]
niteria's avatar
niteria committed
621
                  -> DVarEnv (Var, Var)
622
                  -> IO (UnfoldEnv, TidyOccEnv)
623
                  -- Step 1 from the notes above
624

625
chooseExternalIds hsc_env mod omit_prags expose_all binds implicit_binds imp_id_rules vect_vars
626 627 628
  = do { (unfold_env1,occ_env1) <- search init_work_list emptyVarEnv init_occ_env
       ; let internal_ids = filter (not . (`elemVarEnv` unfold_env1)) binders
       ; tidy_internal internal_ids unfold_env1 occ_env1 }
629
 where
630
  nc_var = hsc_NC hsc_env
631

632 633
  -- init_ext_ids is the intial list of Ids that should be
  -- externalised.  It serves as the starting point for finding a
634 635
  -- deterministic, tidy, renaming for all external Ids in this
  -- module.
636
  --
637 638 639 640
  -- It is sorted, so that it has adeterministic order (i.e. it's the
  -- same list every time this module is compiled), in contrast to the
  -- bindings, which are ordered non-deterministically.
  init_work_list = zip init_ext_ids init_ext_ids
641
  init_ext_ids   = sortBy (compare `on` getOccName) $ filter is_external binders
642

643 644
  -- An Id should be external if either (a) it is exported,
  -- (b) it appears in the RHS of a local rule for an imported Id, or
645
  -- (c) it is the vectorised version of an imported Id.
646
  -- See Note [Which rules to expose]
647 648
  is_external id = isExportedId id || id `elemVarSet` rule_rhs_vars
                 || id `elemVarSet` vect_var_vs
649

650
  rule_rhs_vars  = mapUnionVarSet ruleRhsFreeVars imp_id_rules
niteria's avatar
niteria committed
651
  vect_var_vs    = mkVarSet [var_v | (var, var_v) <- eltsUDFM vect_vars, isGlobalId var]
652

653
  binders          = map fst $ flattenBinds binds
654
  implicit_binders = bindersOfBinds implicit_binds
655
  binder_set       = mkVarSet binders
656

657
  avoids   = [getOccName name | bndr <- binders ++ implicit_binders,
658
                                let name = idName bndr,
659
                                isExternalName name ]
660 661 662 663 664 665
                -- In computing our "avoids" list, we must include
                --      all implicit Ids
                --      all things with global names (assigned once and for
                --                                      all by the renamer)
                -- since their names are "taken".
                -- The type environment is a convenient source of such things.
666 667
                -- In particular, the set of binders doesn't include
                -- implicit Ids at this stage.
668

669 670 671 672 673 674 675 676
        -- We also make sure to avoid any exported binders.  Consider
        --      f{-u1-} = 1     -- Local decl
        --      ...
        --      f{-u2-} = 2     -- Exported decl
        --
        -- The second exported decl must 'get' the name 'f', so we
        -- have to put 'f' in the avoids list before we get to the first
        -- decl.  tidyTopId then does a no-op on exported binders.
677 678 679
  init_occ_env = initTidyOccEnv avoids


680
  search :: [(Id,Id)]    -- The work-list: (external id, referrring id)
681
                         -- Make a tidy, external Name for the external id,
682 683
                         --   add it to the UnfoldEnv, and do the same for the
                         --   transitive closure of Ids it refers to
684 685
                         -- The referring id is used to generate a tidy
                         ---  name for the external id
686 687 688 689 690 691
         -> UnfoldEnv    -- id -> (new Name, show_unfold)
         -> TidyOccEnv   -- occ env for choosing new Names
         -> IO (UnfoldEnv, TidyOccEnv)

  search [] unfold_env occ_env = return (unfold_env, occ_env)

692 693
  search ((idocc,referrer) : rest) unfold_env occ_env
    | idocc `elemVarEnv` unfold_env = search rest unfold_env occ_env
694
    | otherwise = do
695
      (occ_env', name') <- tidyTopName mod nc_var (Just referrer) occ_env idocc
696
      let
697 698
          (new_ids, show_unfold)
                | omit_prags = ([], False)
699
                | otherwise  = addExternal expose_all refined_id
700

701
                -- add vectorised version if any exists
niteria's avatar
niteria committed
702
          new_ids' = new_ids ++ maybeToList (fmap snd $ lookupDVarEnv vect_vars idocc)
Austin Seipp's avatar
Austin Seipp committed
703

704 705
                -- 'idocc' is an *occurrence*, but we need to see the
                -- unfolding in the *definition*; so look up in binder_set
706 707 708 709 710 711 712
          refined_id = case lookupVarSet binder_set idocc of
                         Just id -> id
                         Nothing -> WARN( True, ppr idocc ) idocc

          unfold_env' = extendVarEnv unfold_env idocc (name',show_unfold)
          referrer' | isExportedId refined_id = refined_id
                    | otherwise               = referrer
713
      --
714
      search (zip new_ids' (repeat referrer') ++ rest) unfold_env' occ_env'
715 716 717 718 719 720 721 722 723

  tidy_internal :: [Id] -> UnfoldEnv -> TidyOccEnv
                -> IO (UnfoldEnv, TidyOccEnv)
  tidy_internal []       unfold_env occ_env = return (unfold_env,occ_env)
  tidy_internal (id:ids) unfold_env occ_env = do
      (occ_env', name') <- tidyTopName mod nc_var Nothing occ_env id
      let unfold_env' = extendVarEnv unfold_env id (name',False)
      tidy_internal ids unfold_env' occ_env'

724
addExternal :: Bool -> Id -> ([Id], Bool)
725
addExternal expose_all id = (new_needed_ids, show_unfold)
726
  where
727
    new_needed_ids = bndrFvsInOrder show_unfold id
728
    idinfo         = idInfo id
729
    show_unfold    = show_unfolding (unfoldingInfo idinfo)
730
    never_active   = isNeverActive (inlinePragmaActivation (inlinePragInfo idinfo))
731
    loop_breaker   = isStrongLoopBreaker (occInfo idinfo)
732
    bottoming_fn   = isBottomingSig (strictnessInfo idinfo)
733

734 735 736
        -- Stuff to do with the Id's unfolding
        -- We leave the unfolding there even if there is a worker
        -- In GHCi the unfolding is used by importers
737 738

    show_unfolding (CoreUnfolding { uf_src = src, uf_guidance = guidance })
739 740
       =  expose_all         -- 'expose_all' says to expose all
                             -- unfoldings willy-nilly
741

742 743
       || isStableSource src     -- Always expose things whose
                                 -- source is an inline rule
744

745 746 747 748
       || not (bottoming_fn      -- No need to inline bottom functions
           || never_active       -- Or ones that say not to
           || loop_breaker       -- Or that are loop breakers
           || neverUnfoldGuidance guidance)
749 750
    show_unfolding (DFunUnfolding {}) = True
    show_unfolding _                  = False
751

Austin Seipp's avatar
Austin Seipp committed
752 753 754
{-
************************************************************************
*                                                                      *
755
               Deterministic free variables
Austin Seipp's avatar
Austin Seipp committed
756 757
*                                                                      *
************************************************************************
758

759 760 761 762
We want a deterministic free-variable list.  exprFreeVars gives us
a VarSet, which is in a non-deterministic order when converted to a
list.  Hence, here we define a free-variable finder that returns
the free variables in the order that they are encountered.
763

764
See Note [Choosing external names]
Austin Seipp's avatar
Austin Seipp committed
765
-}
766

767 768 769
bndrFvsInOrder :: Bool -> Id -> [Id]
bndrFvsInOrder show_unfold id
  = run (dffvLetBndr show_unfold id)
770

771 772 773 774
run :: DFFV () -> [Id]
run (DFFV m) = case m emptyVarSet (emptyVarSet, []) of
                 ((_,ids),_) -> ids

775 776
newtype DFFV a
  = DFFV (VarSet              -- Envt: non-top-level things that are in scope
777
                              -- we don't want to record these as free vars
778
      -> (VarSet, [Var])      -- Input State: (set, list) of free vars so far
779
      -> ((VarSet,[Var]),a))  -- Output state
780

Austin Seipp's avatar
Austin Seipp committed
781 782 783 784
instance Functor DFFV where
    fmap = liftM

instance Applicative DFFV where
785
    pure a = DFFV $ \_ st -> (st, a)
Austin Seipp's avatar
Austin Seipp committed
786 787
    (<*>) = ap

788
instance Monad DFFV where
789 790 791 792 793 794 795 796 797 798
  (DFFV m) >>= k = DFFV $ \env st ->
    case m env st of
       (st',a) -> case k a of
                     DFFV f -> f env st'

extendScope :: Var -> DFFV a -> DFFV a
extendScope v (DFFV f) = DFFV (\env st -> f (extendVarSet env v) st)

extendScopeList :: [Var] -> DFFV a -> DFFV a
extendScopeList vs (DFFV f) = DFFV (\env st -> f (extendVarSetList env vs) st)
799 800

insert :: Var -> DFFV ()
801 802
insert v = DFFV $ \ env (set, ids) ->
           let keep_me = isLocalId v &&
803
                         not (v `elemVarSet` env) &&
804 805
                           not (v `elemVarSet` set)
           in if keep_me
806 807 808
              then ((extendVarSet set v, v:ids), ())
              else ((set,                ids),   ())

809 810

dffvExpr :: CoreExpr -> DFFV ()
811
dffvExpr (Var v)              = insert v
812 813
dffvExpr (App e1 e2)          = dffvExpr e1 >> dffvExpr e2
dffvExpr (Lam v e)            = extendScope v (dffvExpr e)
814 815
dffvExpr (Tick (Breakpoint _ ids) e) = mapM_ insert ids >> dffvExpr e
dffvExpr (Tick _other e)    = dffvExpr e
816
dffvExpr (Cast e _)           = dffvExpr e
817 818 819 820 821 822 823 824 825 826
dffvExpr (Let (NonRec x r) e) = dffvBind (x,r) >> extendScope x (dffvExpr e)
dffvExpr (Let (Rec prs) e)    = extendScopeList (map fst prs) $
                                (mapM_ dffvBind prs >> dffvExpr e)
dffvExpr (Case e b _ as)      = dffvExpr e >> extendScope b (mapM_ dffvAlt as)
dffvExpr _other               = return ()

dffvAlt :: (t, [Var], CoreExpr) -> DFFV ()
dffvAlt (_,xs,r) = extendScopeList xs (dffvExpr r)

dffvBind :: (Id, CoreExpr) -> DFFV ()
827
dffvBind(x,r)
828 829
  | not (isId x) = dffvExpr r
  | otherwise    = dffvLetBndr False x >> dffvExpr r
830 831
                -- Pass False because we are doing the RHS right here
                -- If you say True you'll get *exponential* behaviour!
832 833

dffvLetBndr :: Bool -> Id -> DFFV ()
834 835 836 837 838 <