2 % (c) The GRASP/AQUA Project, Glasgow University, 1992-1998
4 \section{Tidying up Core}
7 module TidyPgm( mkBootModDetailsDs, mkBootModDetailsTc,
8 tidyProgram, globaliseAndTidyId ) where
10 #include "HsVersions.h"
22 import CoreArity ( exprArity )
23 import Class ( classSelIds )
46 import FastBool hiding ( fastOr )
48 import Data.List ( partition )
49 import Data.Maybe ( isJust )
50 import Data.IORef ( IORef, readIORef, writeIORef )
54 Constructing the TypeEnv, Instances, Rules from which the ModIface is
55 constructed, and which goes on to subsequent modules in --make mode.
57 Most of the interface file is obtained simply by serialising the
58 TypeEnv. One important consequence is that if the *interface file*
59 has pragma info if and only if the final TypeEnv does. This is not so
60 important for *this* module, but it's essential for ghc --make:
61 subsequent compilations must not see (e.g.) the arity if the interface
62 file does not contain arity If they do, they'll exploit the arity;
63 then the arity might change, but the iface file doesn't change =>
64 recompilation does not happen => disaster.
66 For data types, the final TypeEnv will have a TyThing for the TyCon,
67 plus one for each DataCon; the interface file will contain just one
68 data type declaration, but it is de-serialised back into a collection
71 %************************************************************************
75 %************************************************************************
78 Plan A: mkBootModDetails: omit pragmas, make interfaces small
79 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
82 * Drop all WiredIn things from the TypeEnv
83 (we never want them in interface files)
85 * Retain all TyCons and Classes in the TypeEnv, to avoid
86 having to find which ones are mentioned in the
89 * Trim off the constructors of non-exported TyCons, both
90 from the TyCon and from the TypeEnv
92 * Drop non-exported Ids from the TypeEnv
94 * Tidy the types of the DFunIds of Instances,
95 make them into GlobalIds, (they already have External Names)
96 and add them to the TypeEnv
98 * Tidy the types of the (exported) Ids in the TypeEnv,
99 make them into GlobalIds (they already have External Names)
101 * Drop rules altogether
103 * Tidy the bindings, to ensure that the Caf and Arity
104 information is correct for each top-level binder; the
105 code generator needs it. And to ensure that local names have
106 distinct OccNames in case of object-file splitting
109 -- This is Plan A: make a small type env when typechecking only,
110 -- or when compiling a hs-boot file, or simply when not using -O
112 -- We don't look at the bindings at all -- there aren't any
115 mkBootModDetailsTc :: HscEnv -> TcGblEnv -> IO ModDetails
116 mkBootModDetailsTc hsc_env
117 TcGblEnv{ tcg_exports = exports,
118 tcg_type_env = type_env,
120 tcg_fam_insts = fam_insts
122 = mkBootModDetails hsc_env exports type_env insts fam_insts
124 mkBootModDetailsDs :: HscEnv -> ModGuts -> IO ModDetails
125 mkBootModDetailsDs hsc_env
126 ModGuts{ mg_exports = exports,
129 mg_fam_insts = fam_insts
131 = mkBootModDetails hsc_env exports type_env insts fam_insts
133 mkBootModDetails :: HscEnv -> [AvailInfo] -> NameEnv TyThing
134 -> [Instance] -> [FamInstEnv.FamInst] -> IO ModDetails
135 mkBootModDetails hsc_env exports type_env insts fam_insts
136 = do { let dflags = hsc_dflags hsc_env
137 ; showPass dflags "Tidy [hoot] type env"
139 ; let { insts' = tidyInstances globaliseAndTidyId insts
140 ; dfun_ids = map instanceDFunId insts'
141 ; type_env1 = tidyBootTypeEnv (availsToNameSet exports) type_env
142 ; type_env' = extendTypeEnvWithIds type_env1 dfun_ids
144 ; return (ModDetails { md_types = type_env'
146 , md_fam_insts = fam_insts
149 , md_exports = exports
150 , md_vect_info = noVectInfo
155 tidyBootTypeEnv :: NameSet -> TypeEnv -> TypeEnv
156 tidyBootTypeEnv exports type_env
157 = tidyTypeEnv True False exports type_env final_ids
159 -- Find the LocalIds in the type env that are exported
160 -- Make them into GlobalIds, and tidy their types
162 -- It's very important to remove the non-exported ones
163 -- because we don't tidy the OccNames, and if we don't remove
164 -- the non-exported ones we'll get many things with the
165 -- same name in the interface file, giving chaos.
166 final_ids = [ globaliseAndTidyId id
167 | id <- typeEnvIds type_env
171 -- default methods have their export flag set, but everything
172 -- else doesn't (yet), because this is pre-desugaring, so we
174 keep_it id = isExportedId id || idName id `elemNameSet` exports
178 globaliseAndTidyId :: Id -> Id
179 -- Takes an LocalId with an External Name,
180 -- makes it into a GlobalId
181 -- * unchanged Name (might be Internal or External)
182 -- * unchanged details
183 -- * VanillaIdInfo (makes a conservative assumption about Caf-hood)
184 globaliseAndTidyId id
185 = Id.setIdType (globaliseId id) tidy_type
187 tidy_type = tidyTopType (idType id)
191 %************************************************************************
193 Plan B: tidy bindings, make TypeEnv full of IdInfo
195 %************************************************************************
197 Plan B: include pragmas, make interfaces
198 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
199 * Figure out which Ids are externally visible
201 * Tidy the bindings, externalising appropriate Ids
203 * Drop all Ids from the TypeEnv, and add all the External Ids from
204 the bindings. (This adds their IdInfo to the TypeEnv; and adds
205 floated-out Ids that weren't even in the TypeEnv before.)
207 Step 1: Figure out external Ids
208 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
209 First we figure out which Ids are "external" Ids. An
210 "external" Id is one that is visible from outside the compilation
212 a) the user exported ones
213 b) ones mentioned in the unfoldings, workers,
214 or rules of externally-visible ones
215 This exercise takes a sweep of the bindings bottom to top. Actually,
216 in Step 2 we're also going to need to know which Ids should be
217 exported with their unfoldings, so we produce not an IdSet but an
221 Step 2: Tidy the program
222 ~~~~~~~~~~~~~~~~~~~~~~~~
223 Next we traverse the bindings top to bottom. For each *top-level*
226 1. Make it into a GlobalId; its IdDetails becomes VanillaGlobal,
227 reflecting the fact that from now on we regard it as a global,
230 2. Give it a system-wide Unique.
231 [Even non-exported things need system-wide Uniques because the
232 byte-code generator builds a single Name->BCO symbol table.]
234 We use the NameCache kept in the HscEnv as the
235 source of such system-wide uniques.
237 For external Ids, use the original-name cache in the NameCache
238 to ensure that the unique assigned is the same as the Id had
239 in any previous compilation run.
241 3. If it's an external Id, make it have a External Name, otherwise
242 make it have an Internal Name.
243 This is used by the code generator to decide whether
244 to make the label externally visible
246 4. Give external Ids a "tidy" OccName. This means
247 we can print them in interface files without confusing
248 "x" (unique 5) with "x" (unique 10).
250 5. Give it its UTTERLY FINAL IdInfo; in ptic,
251 * its unfolding, if it should have one
253 * its arity, computed from the number of visible lambdas
255 * its CAF info, computed from what is free in its RHS
258 Finally, substitute these new top-level binders consistently
259 throughout, including in unfoldings. We also tidy binders in
260 RHSs, so that they print nicely in interfaces.
263 tidyProgram :: HscEnv -> ModGuts -> IO (CgGuts, ModDetails)
264 tidyProgram hsc_env (ModGuts { mg_module = mod, mg_exports = exports,
266 mg_insts = insts, mg_fam_insts = fam_insts,
268 mg_rules = imp_rules,
269 mg_vect_info = vect_info,
270 mg_dir_imps = dir_imps,
273 mg_foreign = foreign_stubs,
274 mg_hpc_info = hpc_info,
275 mg_modBreaks = modBreaks })
277 = do { let dflags = hsc_dflags hsc_env
278 ; showPass dflags "Tidy Core"
280 ; let { omit_prags = dopt Opt_OmitInterfacePragmas dflags
281 ; th = dopt Opt_TemplateHaskell dflags
282 ; ext_ids = findExternalIds omit_prags binds
285 | otherwise = findExternalRules binds imp_rules ext_ids
286 -- findExternalRules filters imp_rules to avoid binders that
287 -- aren't externally visible; but the externally-visible binders
288 -- are computed (by findExternalIds) assuming that all orphan
289 -- rules are exported (they get their Exported flag set in the desugarer)
290 -- So in fact we may export more than we need.
291 -- (It's a sort of mutual recursion.)
294 ; (tidy_env, tidy_binds) <- tidyTopBinds hsc_env mod type_env ext_ids
297 ; let { export_set = availsToNameSet exports
298 ; final_ids = [ id | id <- bindersOfBinds tidy_binds,
299 isExternalName (idName id)]
300 ; tidy_type_env = tidyTypeEnv omit_prags th export_set
302 ; tidy_insts = tidyInstances (lookup_dfun tidy_type_env) insts
303 -- A DFunId will have a binding in tidy_binds, and so
304 -- will now be in final_env, replete with IdInfo
305 -- Its name will be unchanged since it was born, but
306 -- we want Global, IdInfo-rich (or not) DFunId in the
309 ; tidy_rules = tidyRules tidy_env ext_rules
310 -- You might worry that the tidy_env contains IdInfo-rich stuff
311 -- and indeed it does, but if omit_prags is on, ext_rules is
314 -- See Note [Injecting implicit bindings]
315 ; implicit_binds = getImplicitBinds type_env
316 ; all_tidy_binds = implicit_binds ++ tidy_binds
318 ; alg_tycons = filter isAlgTyCon (typeEnvTyCons type_env)
321 ; endPass dflags "Tidy Core" Opt_D_dump_simpl all_tidy_binds
322 ; dumpIfSet_core dflags Opt_D_dump_simpl
324 (pprRules tidy_rules)
326 ; let dir_imp_mods = moduleEnvKeys dir_imps
328 ; return (CgGuts { cg_module = mod,
329 cg_tycons = alg_tycons,
330 cg_binds = all_tidy_binds,
331 cg_dir_imps = dir_imp_mods,
332 cg_foreign = foreign_stubs,
333 cg_dep_pkgs = dep_pkgs deps,
334 cg_hpc_info = hpc_info,
335 cg_modBreaks = modBreaks },
337 ModDetails { md_types = tidy_type_env,
338 md_rules = tidy_rules,
339 md_insts = tidy_insts,
340 md_fam_insts = fam_insts,
341 md_exports = exports,
342 md_anns = anns, -- are already tidy
343 md_vect_info = vect_info --
347 lookup_dfun :: TypeEnv -> Var -> Id
348 lookup_dfun type_env dfun_id
349 = case lookupTypeEnv type_env (idName dfun_id) of
350 Just (AnId dfun_id') -> dfun_id'
351 _other -> pprPanic "lookup_dfun" (ppr dfun_id)
353 --------------------------
354 tidyTypeEnv :: Bool -- Compiling without -O, so omit prags
355 -> Bool -- Template Haskell is on
356 -> NameSet -> TypeEnv -> [Id] -> TypeEnv
358 -- The competed type environment is gotten from
359 -- Dropping any wired-in things, and then
360 -- a) keeping the types and classes
361 -- b) removing all Ids,
362 -- c) adding Ids with correct IdInfo, including unfoldings,
363 -- gotten from the bindings
364 -- From (c) we keep only those Ids with External names;
365 -- the CoreTidy pass makes sure these are all and only
366 -- the externally-accessible ones
367 -- This truncates the type environment to include only the
368 -- exported Ids and things needed from them, which saves space
370 tidyTypeEnv omit_prags th exports type_env final_ids
371 = let type_env1 = filterNameEnv keep_it type_env
372 type_env2 = extendTypeEnvWithIds type_env1 final_ids
373 type_env3 | omit_prags = mapNameEnv (trimThing th exports) type_env2
374 | otherwise = type_env2
378 -- We keep GlobalIds, because they won't appear
379 -- in the bindings from which final_ids are derived!
380 -- (The bindings bind LocalIds.)
381 keep_it thing | isWiredInThing thing = False
382 keep_it (AnId id) = isGlobalId id -- Keep GlobalIds (e.g. class ops)
383 keep_it _other = True -- Keep all TyCons, DataCons, and Classes
385 --------------------------
386 isWiredInThing :: TyThing -> Bool
387 isWiredInThing thing = isWiredInName (getName thing)
389 --------------------------
390 trimThing :: Bool -> NameSet -> TyThing -> TyThing
391 -- Trim off inessentials, for boot files and no -O
392 trimThing th exports (ATyCon tc)
393 | not th && not (mustExposeTyCon exports tc)
394 = ATyCon (makeTyConAbstract tc) -- Note [Trimming and Template Haskell]
396 trimThing _th _exports (AnId id)
397 | not (isImplicitId id)
398 = AnId (id `setIdInfo` vanillaIdInfo)
400 trimThing _th _exports other_thing
404 {- Note [Trimming and Template Haskell]
405 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
406 Consider (Trac #2386) this
407 module M(T, makeOne) where
409 makeOne = [| Yay "Yep" |]
410 Notice that T is exported abstractly, but makeOne effectively exports it too!
411 A module that splices in $(makeOne) will then look for a declartion of Yay,
412 so it'd better be there. Hence, brutally but simply, we switch off type
413 constructor trimming if TH is enabled in this module. -}
416 mustExposeTyCon :: NameSet -- Exports
417 -> TyCon -- The tycon
418 -> Bool -- Can its rep be hidden?
419 -- We are compiling without -O, and thus trying to write as little as
420 -- possible into the interface file. But we must expose the details of
421 -- any data types whose constructors or fields are exported
422 mustExposeTyCon exports tc
423 | not (isAlgTyCon tc) -- Synonyms
425 | isEnumerationTyCon tc -- For an enumeration, exposing the constructors
426 = True -- won't lead to the need for further exposure
427 -- (This includes data types with no constructors.)
428 | isOpenTyCon tc -- Open type family
431 | otherwise -- Newtype, datatype
432 = any exported_con (tyConDataCons tc)
433 -- Expose rep if any datacon or field is exported
435 || (isNewTyCon tc && isFFITy (snd (newTyConRhs tc)))
436 -- Expose the rep for newtypes if the rep is an FFI type.
437 -- For a very annoying reason. 'Foreign import' is meant to
438 -- be able to look through newtypes transparently, but it
439 -- can only do that if it can "see" the newtype representation
441 exported_con con = any (`elemNameSet` exports)
442 (dataConName con : dataConFieldLabels con)
444 tidyInstances :: (DFunId -> DFunId) -> [Instance] -> [Instance]
445 tidyInstances tidy_dfun ispecs
448 tidy ispec = setInstanceDFunId ispec $
449 tidy_dfun (instanceDFunId ispec)
453 %************************************************************************
457 %************************************************************************
459 Note [Injecting implicit bindings]
460 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
461 We inject the implict bindings right at the end, in CoreTidy.
462 Some of these bindings, notably record selectors, are not
463 constructed in an optimised form. E.g. record selector for
464 data T = MkT { x :: {-# UNPACK #-} !Int }
465 Then the unfolding looks like
466 x = \t. case t of MkT x1 -> let x = I# x1 in x
467 This generates bad code unless it's first simplified a bit. That is
468 why CoreUnfold.mkImplicitUnfolding uses simleExprOpt to do a bit of
469 optimisation first. (Only matters when the selector is used curried;
470 eg map x ys.) See Trac #2070.
472 At one time I tried injecting the implicit bindings *early*, at the
473 beginning of SimplCore. But that gave rise to real difficulty,
474 becuase GlobalIds are supposed to have *fixed* IdInfo, but the
475 simplifier and other core-to-core passes mess with IdInfo all the
476 time. The straw that broke the camels back was when a class selector
477 got the wrong arity -- ie the simplifier gave it arity 2, whereas
478 importing modules were expecting it to have arity 1 (Trac #2844).
479 It's much safer just to inject them right at the end, after tidying.
483 getImplicitBinds :: TypeEnv -> [CoreBind]
484 getImplicitBinds type_env
485 = map get_defn (concatMap implicit_ids (typeEnvElts type_env))
487 implicit_ids (ATyCon tc) = mapCatMaybes dataConWrapId_maybe (tyConDataCons tc)
488 implicit_ids (AClass cls) = classSelIds cls
491 get_defn :: Id -> CoreBind
492 get_defn id = NonRec id (unfoldingTemplate (idUnfolding id))
496 %************************************************************************
498 \subsection{Step 1: finding externals}
500 %************************************************************************
503 findExternalIds :: Bool
505 -> IdEnv Bool -- In domain => external
506 -- Range = True <=> show unfolding
507 -- Step 1 from the notes above
508 findExternalIds omit_prags binds
510 = mkVarEnv [ (id,False) | id <- bindersOfBinds binds, isExportedId id ]
513 = foldr find emptyVarEnv binds
515 find (NonRec id rhs) needed
516 | need_id needed id = addExternal (id,rhs) needed
518 find (Rec prs) needed = find_prs prs needed
520 -- For a recursive group we have to look for a fixed point
522 | null needed_prs = needed
523 | otherwise = find_prs other_prs new_needed
525 (needed_prs, other_prs) = partition (need_pr needed) prs
526 new_needed = foldr addExternal needed needed_prs
528 -- The 'needed' set contains the Ids that are needed by earlier
529 -- interface file emissions. If the Id isn't in this set, and isn't
530 -- exported, there's no need to emit anything
531 need_id needed_set id = id `elemVarEnv` needed_set || isExportedId id
532 need_pr needed_set (id,_) = need_id needed_set id
534 addExternal :: (Id,CoreExpr) -> IdEnv Bool -> IdEnv Bool
535 -- The Id is needed; extend the needed set
536 -- with it and its dependents (free vars etc)
537 addExternal (id,rhs) needed
538 = extendVarEnv (foldVarSet add_occ needed new_needed_ids)
541 add_occ id needed | id `elemVarEnv` needed = needed
542 | otherwise = extendVarEnv needed id False
543 -- "False" because we don't know we need the Id's unfolding
544 -- Don't override existing bindings; we might have already set it to True
546 new_needed_ids = worker_ids `unionVarSet`
547 unfold_ids `unionVarSet`
551 dont_inline = isNeverActive (inlinePragInfo idinfo)
552 loop_breaker = isNonRuleLoopBreaker (occInfo idinfo)
553 bottoming_fn = isBottomingSig (newStrictnessInfo idinfo `orElse` topSig)
554 spec_ids = specInfoFreeVars (specInfo idinfo)
555 worker_info = workerInfo idinfo
557 -- Stuff to do with the Id's unfolding
558 -- The simplifier has put an up-to-date unfolding
559 -- in the IdInfo, but the RHS will do just as well
560 unfolding = unfoldingInfo idinfo
561 rhs_is_small = not (neverUnfold unfolding)
563 -- We leave the unfolding there even if there is a worker
564 -- In GHCI the unfolding is used by importers
565 -- When writing an interface file, we omit the unfolding
566 -- if there is a worker
567 show_unfold = not bottoming_fn && -- Not necessary
570 rhs_is_small -- Small enough
572 unfold_ids | show_unfold = exprSomeFreeVars isLocalId rhs
573 | otherwise = emptyVarSet
575 worker_ids = case worker_info of
576 HasWorker work_id _ -> unitVarSet work_id
577 _otherwise -> emptyVarSet
582 findExternalRules :: [CoreBind]
583 -> [CoreRule] -- Non-local rules (i.e. ones for imported fns)
584 -> IdEnv a -- Ids that are exported, so we need their rules
586 -- The complete rules are gotten by combining
587 -- a) the non-local rules
588 -- b) rules embedded in the top-level Ids
589 findExternalRules binds non_local_rules ext_ids
590 = filter (not . internal_rule) (non_local_rules ++ local_rules)
593 | id <- bindersOfBinds binds,
594 id `elemVarEnv` ext_ids,
595 rule <- idCoreRules id
599 = any internal_id (varSetElems (ruleLhsFreeIds rule))
600 -- Don't export a rule whose LHS mentions a locally-defined
601 -- Id that is completely internal (i.e. not visible to an
604 internal_id id = not (id `elemVarEnv` ext_ids)
609 %************************************************************************
611 \subsection{Step 2: top-level tidying}
613 %************************************************************************
617 -- TopTidyEnv: when tidying we need to know
618 -- * nc_var: The NameCache, containing a unique supply and any pre-ordained Names.
619 -- These may have arisen because the
620 -- renamer read in an interface file mentioning M.$wf, say,
621 -- and assigned it unique r77. If, on this compilation, we've
622 -- invented an Id whose name is $wf (but with a different unique)
623 -- we want to rename it to have unique r77, so that we can do easy
624 -- comparisons with stuff from the interface file
626 -- * occ_env: The TidyOccEnv, which tells us which local occurrences
629 -- * subst_env: A Var->Var mapping that substitutes the new Var for the old
631 tidyTopBinds :: HscEnv
634 -> IdEnv Bool -- Domain = Ids that should be external
635 -- True <=> their unfolding is external too
637 -> IO (TidyEnv, [CoreBind])
639 tidyTopBinds hsc_env mod type_env ext_ids binds
640 = tidy init_env binds
642 nc_var = hsc_NC hsc_env
644 -- We also make sure to avoid any exported binders. Consider
645 -- f{-u1-} = 1 -- Local decl
647 -- f{-u2-} = 2 -- Exported decl
649 -- The second exported decl must 'get' the name 'f', so we
650 -- have to put 'f' in the avoids list before we get to the first
651 -- decl. tidyTopId then does a no-op on exported binders.
652 init_env = (initTidyOccEnv avoids, emptyVarEnv)
653 avoids = [getOccName name | bndr <- typeEnvIds type_env,
654 let name = idName bndr,
656 -- In computing our "avoids" list, we must include
658 -- all things with global names (assigned once and for
659 -- all by the renamer)
660 -- since their names are "taken".
661 -- The type environment is a convenient source of such things.
663 this_pkg = thisPackage (hsc_dflags hsc_env)
665 tidy env [] = return (env, [])
666 tidy env (b:bs) = do { (env1, b') <- tidyTopBind this_pkg mod nc_var ext_ids env b
667 ; (env2, bs') <- tidy env1 bs
668 ; return (env2, b':bs') }
670 ------------------------
671 tidyTopBind :: PackageId
673 -> IORef NameCache -- For allocating new unique names
674 -> IdEnv Bool -- Domain = Ids that should be external
675 -- True <=> their unfolding is external too
676 -> TidyEnv -> CoreBind
677 -> IO (TidyEnv, CoreBind)
679 tidyTopBind this_pkg mod nc_var ext_ids (occ_env1,subst1) (NonRec bndr rhs)
680 = do { (occ_env2, name') <- tidyTopName mod nc_var ext_ids occ_env1 bndr
681 ; let { (bndr', rhs') = tidyTopPair ext_ids tidy_env2 caf_info name' (bndr, rhs)
682 ; subst2 = extendVarEnv subst1 bndr bndr'
683 ; tidy_env2 = (occ_env2, subst2) }
684 ; return (tidy_env2, NonRec bndr' rhs') }
686 caf_info = hasCafRefs this_pkg subst1 (idArity bndr) rhs
688 tidyTopBind this_pkg mod nc_var ext_ids (occ_env1,subst1) (Rec prs)
689 = do { (occ_env2, names') <- tidyTopNames mod nc_var ext_ids occ_env1 bndrs
690 ; let { prs' = zipWith (tidyTopPair ext_ids tidy_env2 caf_info)
692 ; subst2 = extendVarEnvList subst1 (bndrs `zip` map fst prs')
693 ; tidy_env2 = (occ_env2, subst2) }
694 ; return (tidy_env2, Rec prs') }
698 -- the CafInfo for a recursive group says whether *any* rhs in
699 -- the group may refer indirectly to a CAF (because then, they all do).
701 | or [ mayHaveCafRefs (hasCafRefs this_pkg subst1 (idArity bndr) rhs)
702 | (bndr,rhs) <- prs ] = MayHaveCafRefs
703 | otherwise = NoCafRefs
705 --------------------------------------------------------------------
707 -- This is where we set names to local/global based on whether they really are
708 -- externally visible (see comment at the top of this module). If the name
709 -- was previously local, we have to give it a unique occurrence name if
710 -- we intend to externalise it.
711 tidyTopNames :: Module -> IORef NameCache -> VarEnv Bool -> TidyOccEnv
712 -> [Id] -> IO (TidyOccEnv, [Name])
713 tidyTopNames _mod _nc_var _ext_ids occ_env [] = return (occ_env, [])
714 tidyTopNames mod nc_var ext_ids occ_env (id:ids)
715 = do { (occ_env1, name) <- tidyTopName mod nc_var ext_ids occ_env id
716 ; (occ_env2, names) <- tidyTopNames mod nc_var ext_ids occ_env1 ids
717 ; return (occ_env2, name:names) }
719 tidyTopName :: Module -> IORef NameCache -> VarEnv Bool -> TidyOccEnv
720 -> Id -> IO (TidyOccEnv, Name)
721 tidyTopName mod nc_var ext_ids occ_env id
722 | global && internal = return (occ_env, localiseName name)
724 | global && external = return (occ_env, name)
725 -- Global names are assumed to have been allocated by the renamer,
726 -- so they already have the "right" unique
727 -- And it's a system-wide unique too
729 -- Now we get to the real reason that all this is in the IO Monad:
730 -- we have to update the name cache in a nice atomic fashion
732 | local && internal = do { nc <- readIORef nc_var
733 ; let (nc', new_local_name) = mk_new_local nc
734 ; writeIORef nc_var nc'
735 ; return (occ_env', new_local_name) }
736 -- Even local, internal names must get a unique occurrence, because
737 -- if we do -split-objs we externalise the name later, in the code generator
739 -- Similarly, we must make sure it has a system-wide Unique, because
740 -- the byte-code generator builds a system-wide Name->BCO symbol table
742 | local && external = do { nc <- readIORef nc_var
743 ; let (nc', new_external_name) = mk_new_external nc
744 ; writeIORef nc_var nc'
745 ; return (occ_env', new_external_name) }
747 | otherwise = panic "tidyTopName"
750 external = id `elemVarEnv` ext_ids
751 global = isExternalName name
753 internal = not external
754 loc = nameSrcSpan name
756 (occ_env', occ') = tidyOccName occ_env (nameOccName name)
758 mk_new_local nc = (nc { nsUniqs = us2 }, mkInternalName uniq occ' loc)
760 (us1, us2) = splitUniqSupply (nsUniqs nc)
761 uniq = uniqFromSupply us1
763 mk_new_external nc = allocateGlobalBinder nc mod occ' loc
764 -- If we want to externalise a currently-local name, check
765 -- whether we have already assigned a unique for it.
766 -- If so, use it; if not, extend the table.
767 -- All this is done by allcoateGlobalBinder.
768 -- This is needed when *re*-compiling a module in GHCi; we must
769 -- use the same name for externally-visible things as we did before.
772 -----------------------------------------------------------
773 tidyTopPair :: VarEnv Bool
774 -> TidyEnv -- The TidyEnv is used to tidy the IdInfo
775 -- It is knot-tied: don't look at it!
778 -> (Id, CoreExpr) -- Binder and RHS before tidying
780 -- This function is the heart of Step 2
781 -- The rec_tidy_env is the one to use for the IdInfo
782 -- It's necessary because when we are dealing with a recursive
783 -- group, a variable late in the group might be mentioned
784 -- in the IdInfo of one early in the group
786 tidyTopPair ext_ids rhs_tidy_env caf_info name' (bndr, rhs)
789 bndr' = mkGlobalId details name' ty' idinfo'
790 details = idDetails bndr -- Preserve the IdDetails
791 ty' = tidyTopType (idType bndr)
792 rhs' = tidyExpr rhs_tidy_env rhs
794 idinfo' = tidyTopIdInfo (isJust maybe_external)
795 idinfo unfold_info worker_info
798 -- Expose an unfolding if ext_ids tells us to
799 -- Remember that ext_ids maps an Id to a Bool:
800 -- True to show the unfolding, False to hide it
801 maybe_external = lookupVarEnv ext_ids bndr
802 show_unfold = maybe_external `orElse` False
803 unfold_info | show_unfold = mkTopUnfolding rhs'
804 | otherwise = noUnfolding
805 worker_info = tidyWorker rhs_tidy_env show_unfold (workerInfo idinfo)
807 -- Usually the Id will have an accurate arity on it, because
808 -- the simplifier has just run, but not always.
809 -- One case I found was when the last thing the simplifier
810 -- did was to let-bind a non-atomic argument and then float
811 -- it to the top level. So it seems more robust just to
813 arity = exprArity rhs
816 -- tidyTopIdInfo creates the final IdInfo for top-level
817 -- binders. There are two delicate pieces:
819 -- * Arity. After CoreTidy, this arity must not change any more.
820 -- Indeed, CorePrep must eta expand where necessary to make
821 -- the manifest arity equal to the claimed arity.
823 -- * CAF info. This must also remain valid through to code generation.
824 -- We add the info here so that it propagates to all
825 -- occurrences of the binders in RHSs, and hence to occurrences in
826 -- unfoldings, which are inside Ids imported by GHCi. Ditto RULES.
827 -- CoreToStg makes use of this when constructing SRTs.
828 tidyTopIdInfo :: Bool -> IdInfo -> Unfolding
829 -> WorkerInfo -> ArityInfo -> CafInfo
831 tidyTopIdInfo is_external idinfo unfold_info worker_info arity caf_info
832 | not is_external -- For internal Ids (not externally visible)
833 = vanillaIdInfo -- we only need enough info for code generation
834 -- Arity and strictness info are enough;
835 -- c.f. CoreTidy.tidyLetBndr
836 `setCafInfo` caf_info
838 `setAllStrictnessInfo` newStrictnessInfo idinfo
840 | otherwise -- Externally-visible Ids get the whole lot
842 `setCafInfo` caf_info
844 `setAllStrictnessInfo` newStrictnessInfo idinfo
845 `setInlinePragInfo` inlinePragInfo idinfo
846 `setUnfoldingInfo` unfold_info
847 `setWorkerInfo` worker_info
848 -- NB: we throw away the Rules
849 -- They have already been extracted by findExternalRules
853 ------------ Worker --------------
854 tidyWorker :: TidyEnv -> Bool -> WorkerInfo -> WorkerInfo
855 tidyWorker _tidy_env _show_unfold NoWorker
857 tidyWorker tidy_env show_unfold (HasWorker work_id wrap_arity)
858 | show_unfold = HasWorker (tidyVarOcc tidy_env work_id) wrap_arity
859 | otherwise = NoWorker
860 -- NB: do *not* expose the worker if show_unfold is off,
861 -- because that means this thing is a loop breaker or
862 -- marked NOINLINE or something like that
863 -- This is important: if you expose the worker for a loop-breaker
864 -- then you can make the simplifier go into an infinite loop, because
865 -- in effect the unfolding is exposed. See Trac #1709
867 -- You might think that if show_unfold is False, then the thing should
868 -- not be w/w'd in the first place. But a legitimate reason is this:
869 -- the function returns bottom
870 -- In this case, show_unfold will be false (we don't expose unfoldings
871 -- for bottoming functions), but we might still have a worker/wrapper
872 -- split (see Note [Worker-wrapper for bottoming functions] in WorkWrap.lhs
875 %************************************************************************
877 \subsection{Figuring out CafInfo for an expression}
879 %************************************************************************
881 hasCafRefs decides whether a top-level closure can point into the dynamic heap.
882 We mark such things as `MayHaveCafRefs' because this information is
883 used to decide whether a particular closure needs to be referenced
886 There are two reasons for setting MayHaveCafRefs:
887 a) The RHS is a CAF: a top-level updatable thunk.
888 b) The RHS refers to something that MayHaveCafRefs
890 Possible improvement: In an effort to keep the number of CAFs (and
891 hence the size of the SRTs) down, we could also look at the expression and
892 decide whether it requires a small bounded amount of heap, so we can ignore
893 it as a CAF. In these cases however, we would need to use an additional
894 CAF list to keep track of non-collectable CAFs.
897 hasCafRefs :: PackageId -> VarEnv Var -> Arity -> CoreExpr -> CafInfo
898 hasCafRefs this_pkg p arity expr
899 | is_caf || mentions_cafs
901 | otherwise = NoCafRefs
903 mentions_cafs = isFastTrue (cafRefs p expr)
904 is_caf = not (arity > 0 || rhsIsStatic this_pkg expr)
906 -- NB. we pass in the arity of the expression, which is expected
907 -- to be calculated by exprArity. This is because exprArity
908 -- knows how much eta expansion is going to be done by
909 -- CorePrep later on, and we don't want to duplicate that
910 -- knowledge in rhsIsStatic below.
912 cafRefs :: VarEnv Id -> Expr a -> FastBool
914 -- imported Ids first:
915 | not (isLocalId id) = fastBool (mayHaveCafRefs (idCafInfo id))
916 -- now Ids local to this module:
918 case lookupVarEnv p id of
919 Just id' -> fastBool (mayHaveCafRefs (idCafInfo id'))
920 Nothing -> fastBool False
922 cafRefs _ (Lit _) = fastBool False
923 cafRefs p (App f a) = fastOr (cafRefs p f) (cafRefs p) a
924 cafRefs p (Lam _ e) = cafRefs p e
925 cafRefs p (Let b e) = fastOr (cafRefss p (rhssOfBind b)) (cafRefs p) e
926 cafRefs p (Case e _bndr _ alts) = fastOr (cafRefs p e) (cafRefss p) (rhssOfAlts alts)
927 cafRefs p (Note _n e) = cafRefs p e
928 cafRefs p (Cast e _co) = cafRefs p e
929 cafRefs _ (Type _) = fastBool False
931 cafRefss :: VarEnv Id -> [Expr a] -> FastBool
932 cafRefss _ [] = fastBool False
933 cafRefss p (e:es) = fastOr (cafRefs p e) (cafRefss p) es
935 fastOr :: FastBool -> (a -> FastBool) -> a -> FastBool
936 -- hack for lazy-or over FastBool.
937 fastOr a f x = fastBool (isFastTrue a || isFastTrue (f x))