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"
21 import CoreArity ( exprArity, exprBotStrictness_maybe )
22 import Class ( classSelIds )
31 import Name hiding (varName)
44 import FastBool hiding ( fastOr )
47 import Data.List ( sortBy )
48 import Data.IORef ( IORef, readIORef, writeIORef )
52 Constructing the TypeEnv, Instances, Rules, VectInfo from which the
53 ModIface is constructed, and which goes on to subsequent modules in
56 Most of the interface file is obtained simply by serialising the
57 TypeEnv. One important consequence is that if the *interface file*
58 has pragma info if and only if the final TypeEnv does. This is not so
59 important for *this* module, but it's essential for ghc --make:
60 subsequent compilations must not see (e.g.) the arity if the interface
61 file does not contain arity If they do, they'll exploit the arity;
62 then the arity might change, but the iface file doesn't change =>
63 recompilation does not happen => disaster.
65 For data types, the final TypeEnv will have a TyThing for the TyCon,
66 plus one for each DataCon; the interface file will contain just one
67 data type declaration, but it is de-serialised back into a collection
70 %************************************************************************
74 %************************************************************************
77 Plan A: mkBootModDetails: omit pragmas, make interfaces small
78 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
81 * Drop all WiredIn things from the TypeEnv
82 (we never want them in interface files)
84 * Retain all TyCons and Classes in the TypeEnv, to avoid
85 having to find which ones are mentioned in the
88 * Trim off the constructors of non-exported TyCons, both
89 from the TyCon and from the TypeEnv
91 * Drop non-exported Ids from the TypeEnv
93 * Tidy the types of the DFunIds of Instances,
94 make them into GlobalIds, (they already have External Names)
95 and add them to the TypeEnv
97 * Tidy the types of the (exported) Ids in the TypeEnv,
98 make them into GlobalIds (they already have External Names)
100 * Drop rules altogether
102 * Tidy the bindings, to ensure that the Caf and Arity
103 information is correct for each top-level binder; the
104 code generator needs it. And to ensure that local names have
105 distinct OccNames in case of object-file splitting
108 -- This is Plan A: make a small type env when typechecking only,
109 -- or when compiling a hs-boot file, or simply when not using -O
111 -- We don't look at the bindings at all -- there aren't any
114 mkBootModDetailsTc :: HscEnv -> TcGblEnv -> IO ModDetails
115 mkBootModDetailsTc hsc_env
116 TcGblEnv{ tcg_exports = exports,
117 tcg_type_env = type_env,
119 tcg_fam_insts = fam_insts
121 = mkBootModDetails hsc_env exports type_env insts fam_insts
123 mkBootModDetailsDs :: HscEnv -> ModGuts -> IO ModDetails
124 mkBootModDetailsDs hsc_env
125 ModGuts{ mg_exports = exports,
128 mg_fam_insts = fam_insts
130 = mkBootModDetails hsc_env exports type_env insts fam_insts
132 mkBootModDetails :: HscEnv -> [AvailInfo] -> NameEnv TyThing
133 -> [Instance] -> [FamInstEnv.FamInst] -> IO ModDetails
134 mkBootModDetails hsc_env exports type_env insts fam_insts
135 = do { let dflags = hsc_dflags hsc_env
136 ; showPass dflags "Tidy [hoot] type env"
138 ; let { insts' = tidyInstances globaliseAndTidyId insts
139 ; dfun_ids = map instanceDFunId insts'
140 ; type_env1 = tidyBootTypeEnv (availsToNameSet exports) type_env
141 ; type_env' = extendTypeEnvWithIds type_env1 dfun_ids
143 ; return (ModDetails { md_types = type_env'
145 , md_fam_insts = fam_insts
148 , md_exports = exports
149 , md_vect_info = noVectInfo
154 tidyBootTypeEnv :: NameSet -> TypeEnv -> TypeEnv
155 tidyBootTypeEnv exports type_env
156 = tidyTypeEnv True False exports type_env final_ids
158 -- Find the LocalIds in the type env that are exported
159 -- Make them into GlobalIds, and tidy their types
161 -- It's very important to remove the non-exported ones
162 -- because we don't tidy the OccNames, and if we don't remove
163 -- the non-exported ones we'll get many things with the
164 -- same name in the interface file, giving chaos.
165 final_ids = [ globaliseAndTidyId id
166 | id <- typeEnvIds type_env
170 -- default methods have their export flag set, but everything
171 -- else doesn't (yet), because this is pre-desugaring, so we
173 keep_it id = isExportedId id || idName id `elemNameSet` exports
177 globaliseAndTidyId :: Id -> Id
178 -- Takes an LocalId with an External Name,
179 -- makes it into a GlobalId
180 -- * unchanged Name (might be Internal or External)
181 -- * unchanged details
182 -- * VanillaIdInfo (makes a conservative assumption about Caf-hood)
183 globaliseAndTidyId id
184 = Id.setIdType (globaliseId id) tidy_type
186 tidy_type = tidyTopType (idType id)
190 %************************************************************************
192 Plan B: tidy bindings, make TypeEnv full of IdInfo
194 %************************************************************************
196 Plan B: include pragmas, make interfaces
197 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
198 * Figure out which Ids are externally visible
200 * Tidy the bindings, externalising appropriate Ids
202 * Drop all Ids from the TypeEnv, and add all the External Ids from
203 the bindings. (This adds their IdInfo to the TypeEnv; and adds
204 floated-out Ids that weren't even in the TypeEnv before.)
206 Step 1: Figure out external Ids
207 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
208 Note [choosing external names]
210 See also the section "Interface stability" in the
211 RecompilationAvoidance commentary:
212 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/RecompilationAvoidance
214 First we figure out which Ids are "external" Ids. An
215 "external" Id is one that is visible from outside the compilation
217 a) the user exported ones
218 b) ones mentioned in the unfoldings, workers,
219 or rules of externally-visible ones
221 While figuring out which Ids are external, we pick a "tidy" OccName
222 for each one. That is, we make its OccName distinct from the other
223 external OccNames in this module, so that in interface files and
224 object code we can refer to it unambiguously by its OccName. The
225 OccName for each binder is prefixed by the name of the exported Id
226 that references it; e.g. if "f" references "x" in its unfolding, then
227 "x" is renamed to "f_x". This helps distinguish the different "x"s
228 from each other, and means that if "f" is later removed, things that
229 depend on the other "x"s will not need to be recompiled. Of course,
230 if there are multiple "f_x"s, then we have to disambiguate somehow; we
231 use "f_x0", "f_x1" etc.
233 As far as possible we should assign names in a deterministic fashion.
234 Each time this module is compiled with the same options, we should end
235 up with the same set of external names with the same types. That is,
236 the ABI hash in the interface should not change. This turns out to be
237 quite tricky, since the order of the bindings going into the tidy
238 phase is already non-deterministic, as it is based on the ordering of
239 Uniques, which are assigned unpredictably.
241 To name things in a stable way, we do a depth-first-search of the
242 bindings, starting from the exports sorted by name. This way, as long
243 as the bindings themselves are deterministic (they sometimes aren't!),
244 the order in which they are presented to the tidying phase does not
245 affect the names we assign.
247 Step 2: Tidy the program
248 ~~~~~~~~~~~~~~~~~~~~~~~~
249 Next we traverse the bindings top to bottom. For each *top-level*
252 1. Make it into a GlobalId; its IdDetails becomes VanillaGlobal,
253 reflecting the fact that from now on we regard it as a global,
256 2. Give it a system-wide Unique.
257 [Even non-exported things need system-wide Uniques because the
258 byte-code generator builds a single Name->BCO symbol table.]
260 We use the NameCache kept in the HscEnv as the
261 source of such system-wide uniques.
263 For external Ids, use the original-name cache in the NameCache
264 to ensure that the unique assigned is the same as the Id had
265 in any previous compilation run.
267 3. Rename top-level Ids according to the names we chose in step 1.
268 If it's an external Id, make it have a External Name, otherwise
269 make it have an Internal Name. This is used by the code generator
270 to decide whether to make the label externally visible
272 4. Give it its UTTERLY FINAL IdInfo; in ptic,
273 * its unfolding, if it should have one
275 * its arity, computed from the number of visible lambdas
277 * its CAF info, computed from what is free in its RHS
280 Finally, substitute these new top-level binders consistently
281 throughout, including in unfoldings. We also tidy binders in
282 RHSs, so that they print nicely in interfaces.
285 tidyProgram :: HscEnv -> ModGuts -> IO (CgGuts, ModDetails)
286 tidyProgram hsc_env (ModGuts { mg_module = mod, mg_exports = exports,
288 mg_insts = insts, mg_fam_insts = fam_insts,
290 mg_rules = imp_rules,
291 mg_vect_info = vect_info,
292 mg_dir_imps = dir_imps,
295 mg_foreign = foreign_stubs,
296 mg_hpc_info = hpc_info,
297 mg_modBreaks = modBreaks })
299 = do { let { dflags = hsc_dflags hsc_env
300 ; omit_prags = dopt Opt_OmitInterfacePragmas dflags
301 ; th = dopt Opt_TemplateHaskell dflags
303 ; showPass dflags "Tidy Core"
305 ; let { implicit_binds = getImplicitBinds type_env }
307 ; (unfold_env, tidy_occ_env)
308 <- chooseExternalIds hsc_env mod omit_prags binds implicit_binds imp_rules
310 ; let { ext_rules = findExternalRules omit_prags binds imp_rules unfold_env }
311 -- See Note [Which rules to expose]
313 ; let { (tidy_env, tidy_binds)
314 = tidyTopBinds hsc_env unfold_env tidy_occ_env binds }
316 ; let { export_set = availsToNameSet exports
317 ; final_ids = [ id | id <- bindersOfBinds tidy_binds,
318 isExternalName (idName id)]
319 ; tidy_type_env = tidyTypeEnv omit_prags th export_set
321 ; tidy_insts = tidyInstances (lookup_dfun tidy_type_env) insts
322 -- A DFunId will have a binding in tidy_binds, and so
323 -- will now be in final_env, replete with IdInfo
324 -- Its name will be unchanged since it was born, but
325 -- we want Global, IdInfo-rich (or not) DFunId in the
328 ; tidy_rules = tidyRules tidy_env ext_rules
329 -- You might worry that the tidy_env contains IdInfo-rich stuff
330 -- and indeed it does, but if omit_prags is on, ext_rules is
333 ; tidy_vect_info = tidyVectInfo tidy_env vect_info
335 -- See Note [Injecting implicit bindings]
336 ; all_tidy_binds = implicit_binds ++ tidy_binds
338 ; alg_tycons = filter isAlgTyCon (typeEnvTyCons type_env)
341 ; endPass dflags "Tidy Core" Opt_D_dump_simpl all_tidy_binds tidy_rules
342 ; let dir_imp_mods = moduleEnvKeys dir_imps
344 ; return (CgGuts { cg_module = mod,
345 cg_tycons = alg_tycons,
346 cg_binds = all_tidy_binds,
347 cg_dir_imps = dir_imp_mods,
348 cg_foreign = foreign_stubs,
349 cg_dep_pkgs = dep_pkgs deps,
350 cg_hpc_info = hpc_info,
351 cg_modBreaks = modBreaks },
353 ModDetails { md_types = tidy_type_env,
354 md_rules = tidy_rules,
355 md_insts = tidy_insts,
356 md_vect_info = tidy_vect_info, md_fam_insts = fam_insts,
357 md_exports = exports,
358 md_anns = anns -- are already tidy
362 lookup_dfun :: TypeEnv -> Var -> Id
363 lookup_dfun type_env dfun_id
364 = case lookupTypeEnv type_env (idName dfun_id) of
365 Just (AnId dfun_id') -> dfun_id'
366 _other -> pprPanic "lookup_dfun" (ppr dfun_id)
368 --------------------------
369 tidyTypeEnv :: Bool -- Compiling without -O, so omit prags
370 -> Bool -- Template Haskell is on
371 -> NameSet -> TypeEnv -> [Id] -> TypeEnv
373 -- The competed type environment is gotten from
374 -- Dropping any wired-in things, and then
375 -- a) keeping the types and classes
376 -- b) removing all Ids,
377 -- c) adding Ids with correct IdInfo, including unfoldings,
378 -- gotten from the bindings
379 -- From (c) we keep only those Ids with External names;
380 -- the CoreTidy pass makes sure these are all and only
381 -- the externally-accessible ones
382 -- This truncates the type environment to include only the
383 -- exported Ids and things needed from them, which saves space
385 tidyTypeEnv omit_prags th exports type_env final_ids
386 = let type_env1 = filterNameEnv keep_it type_env
387 type_env2 = extendTypeEnvWithIds type_env1 final_ids
388 type_env3 | omit_prags = mapNameEnv (trimThing th exports) type_env2
389 | otherwise = type_env2
393 -- We keep GlobalIds, because they won't appear
394 -- in the bindings from which final_ids are derived!
395 -- (The bindings bind LocalIds.)
396 keep_it thing | isWiredInThing thing = False
397 keep_it (AnId id) = isGlobalId id -- Keep GlobalIds (e.g. class ops)
398 keep_it _other = True -- Keep all TyCons, DataCons, and Classes
400 --------------------------
401 isWiredInThing :: TyThing -> Bool
402 isWiredInThing thing = isWiredInName (getName thing)
404 --------------------------
405 trimThing :: Bool -> NameSet -> TyThing -> TyThing
406 -- Trim off inessentials, for boot files and no -O
407 trimThing th exports (ATyCon tc)
408 | not th && not (mustExposeTyCon exports tc)
409 = ATyCon (makeTyConAbstract tc) -- Note [Trimming and Template Haskell]
411 trimThing _th _exports (AnId id)
412 | not (isImplicitId id)
413 = AnId (id `setIdInfo` vanillaIdInfo)
415 trimThing _th _exports other_thing
419 {- Note [Trimming and Template Haskell]
420 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
421 Consider (Trac #2386) this
422 module M(T, makeOne) where
424 makeOne = [| Yay "Yep" |]
425 Notice that T is exported abstractly, but makeOne effectively exports it too!
426 A module that splices in $(makeOne) will then look for a declartion of Yay,
427 so it'd better be there. Hence, brutally but simply, we switch off type
428 constructor trimming if TH is enabled in this module. -}
431 mustExposeTyCon :: NameSet -- Exports
432 -> TyCon -- The tycon
433 -> Bool -- Can its rep be hidden?
434 -- We are compiling without -O, and thus trying to write as little as
435 -- possible into the interface file. But we must expose the details of
436 -- any data types whose constructors or fields are exported
437 mustExposeTyCon exports tc
438 | not (isAlgTyCon tc) -- Synonyms
440 | isEnumerationTyCon tc -- For an enumeration, exposing the constructors
441 = True -- won't lead to the need for further exposure
442 -- (This includes data types with no constructors.)
443 | isOpenTyCon tc -- Open type family
446 | otherwise -- Newtype, datatype
447 = any exported_con (tyConDataCons tc)
448 -- Expose rep if any datacon or field is exported
450 || (isNewTyCon tc && isFFITy (snd (newTyConRhs tc)))
451 -- Expose the rep for newtypes if the rep is an FFI type.
452 -- For a very annoying reason. 'Foreign import' is meant to
453 -- be able to look through newtypes transparently, but it
454 -- can only do that if it can "see" the newtype representation
456 exported_con con = any (`elemNameSet` exports)
457 (dataConName con : dataConFieldLabels con)
459 tidyInstances :: (DFunId -> DFunId) -> [Instance] -> [Instance]
460 tidyInstances tidy_dfun ispecs
463 tidy ispec = setInstanceDFunId ispec $
464 tidy_dfun (instanceDFunId ispec)
468 tidyVectInfo :: TidyEnv -> VectInfo -> VectInfo
469 tidyVectInfo (_, var_env) info@(VectInfo { vectInfoVar = vars
470 , vectInfoPADFun = pas
471 , vectInfoIso = isos })
472 = info { vectInfoVar = tidy_vars
473 , vectInfoPADFun = tidy_pas
474 , vectInfoIso = tidy_isos }
477 $ map tidy_var_mapping
480 tidy_pas = mapNameEnv tidy_snd_var pas
481 tidy_isos = mapNameEnv tidy_snd_var isos
483 tidy_var_mapping (from, to) = (from', (from', lookup_var to))
484 where from' = lookup_var from
485 tidy_snd_var (x, var) = (x, lookup_var var)
487 lookup_var var = lookupWithDefaultVarEnv var_env var var
491 %************************************************************************
495 %************************************************************************
497 Note [Injecting implicit bindings]
498 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
499 We inject the implict bindings right at the end, in CoreTidy.
500 Some of these bindings, notably record selectors, are not
501 constructed in an optimised form. E.g. record selector for
502 data T = MkT { x :: {-# UNPACK #-} !Int }
503 Then the unfolding looks like
504 x = \t. case t of MkT x1 -> let x = I# x1 in x
505 This generates bad code unless it's first simplified a bit. That is
506 why CoreUnfold.mkImplicitUnfolding uses simleExprOpt to do a bit of
507 optimisation first. (Only matters when the selector is used curried;
508 eg map x ys.) See Trac #2070.
510 [Oct 09: in fact, record selectors are no longer implicit Ids at all,
511 because we really do want to optimise them properly. They are treated
512 much like any other Id. But doing "light" optimisation on an implicit
513 Id still makes sense.]
515 At one time I tried injecting the implicit bindings *early*, at the
516 beginning of SimplCore. But that gave rise to real difficulty,
517 becuase GlobalIds are supposed to have *fixed* IdInfo, but the
518 simplifier and other core-to-core passes mess with IdInfo all the
519 time. The straw that broke the camels back was when a class selector
520 got the wrong arity -- ie the simplifier gave it arity 2, whereas
521 importing modules were expecting it to have arity 1 (Trac #2844).
522 It's much safer just to inject them right at the end, after tidying.
524 Oh: two other reasons for injecting them late:
526 - If implicit Ids are already in the bindings when we start TidyPgm,
527 we'd have to be careful not to treat them as external Ids (in
528 the sense of findExternalIds); else the Ids mentioned in *their*
529 RHSs will be treated as external and you get an interface file
531 but nothing refererring to a18 (because the implicit Id is the
532 one that does, and implicit Ids don't appear in interface files).
534 - More seriously, the tidied type-envt will include the implicit
535 Id replete with a18 in its unfolding; but we won't take account
536 of a18 when computing a fingerprint for the class; result chaos.
538 There is one sort of implicit binding that is injected still later,
539 namely those for data constructor workers. Reason (I think): it's
540 really just a code generation trick.... binding itself makes no sense.
541 See CorePrep Note [Data constructor workers].
544 getImplicitBinds :: TypeEnv -> [CoreBind]
545 getImplicitBinds type_env
546 = map get_defn (concatMap implicit_ids (typeEnvElts type_env))
548 implicit_ids (ATyCon tc) = mapCatMaybes dataConWrapId_maybe (tyConDataCons tc)
549 implicit_ids (AClass cls) = classSelIds cls
552 get_defn :: Id -> CoreBind
553 get_defn id = NonRec id (unfoldingTemplate (idUnfolding id))
557 %************************************************************************
559 \subsection{Step 1: finding externals}
561 %************************************************************************
563 Sete Note [choosing external names].
566 type UnfoldEnv = IdEnv (Name{-new name-}, Bool {-show unfolding-})
567 -- Maps each top-level Id to its new Name (the Id is tidied in step 2)
568 -- The Unique is unchanged. If the new Name is external, it will be
569 -- visible in the interface file.
571 -- Bool => expose unfolding or not.
573 chooseExternalIds :: HscEnv
579 -> IO (UnfoldEnv, TidyOccEnv)
580 -- Step 1 from the notes above
582 chooseExternalIds hsc_env mod omit_prags binds implicit_binds imp_id_rules
583 = do { (unfold_env1,occ_env1) <- search init_work_list emptyVarEnv init_occ_env
584 ; let internal_ids = filter (not . (`elemVarEnv` unfold_env1)) binders
585 ; tidy_internal internal_ids unfold_env1 occ_env1 }
587 nc_var = hsc_NC hsc_env
589 -- init_ext_ids is the intial list of Ids that should be
590 -- externalised. It serves as the starting point for finding a
591 -- deterministic, tidy, renaming for all external Ids in this
594 -- It is sorted, so that it has adeterministic order (i.e. it's the
595 -- same list every time this module is compiled), in contrast to the
596 -- bindings, which are ordered non-deterministically.
597 init_work_list = zip init_ext_ids init_ext_ids
598 init_ext_ids = sortBy (compare `on` getOccName) $
599 filter is_external binders
601 -- An Id should be external if either (a) it is exported or
602 -- (b) it appears in the RHS of a local rule for an imported Id.
603 -- See Note [Which rules to expose]
604 is_external id = isExportedId id || id `elemVarSet` rule_rhs_vars
605 rule_rhs_vars = foldr (unionVarSet . ruleRhsFreeVars) emptyVarSet imp_id_rules
607 binders = bindersOfBinds binds
608 implicit_binders = bindersOfBinds implicit_binds
609 binder_set = mkVarSet binders
611 avoids = [getOccName name | bndr <- binders ++ implicit_binders,
612 let name = idName bndr,
613 isExternalName name ]
614 -- In computing our "avoids" list, we must include
616 -- all things with global names (assigned once and for
617 -- all by the renamer)
618 -- since their names are "taken".
619 -- The type environment is a convenient source of such things.
620 -- In particular, the set of binders doesn't include
621 -- implicit Ids at this stage.
623 -- We also make sure to avoid any exported binders. Consider
624 -- f{-u1-} = 1 -- Local decl
626 -- f{-u2-} = 2 -- Exported decl
628 -- The second exported decl must 'get' the name 'f', so we
629 -- have to put 'f' in the avoids list before we get to the first
630 -- decl. tidyTopId then does a no-op on exported binders.
631 init_occ_env = initTidyOccEnv avoids
634 search :: [(Id,Id)] -- The work-list: (external id, referrring id)
635 -- Make a tidy, external Name for the external id,
636 -- add it to the UnfoldEnv, and do the same for the
637 -- transitive closure of Ids it refers to
638 -- The referring id is used to generate a tidy
639 --- name for the external id
640 -> UnfoldEnv -- id -> (new Name, show_unfold)
641 -> TidyOccEnv -- occ env for choosing new Names
642 -> IO (UnfoldEnv, TidyOccEnv)
644 search [] unfold_env occ_env = return (unfold_env, occ_env)
646 search ((idocc,referrer) : rest) unfold_env occ_env
647 | idocc `elemVarEnv` unfold_env = search rest unfold_env occ_env
649 (occ_env', name') <- tidyTopName mod nc_var (Just referrer) occ_env idocc
651 (new_ids, show_unfold)
652 | omit_prags = ([], False)
653 | otherwise = addExternal refined_id
655 -- 'idocc' is an *occurrence*, but we need to see the
656 -- unfolding in the *definition*; so look up in binder_set
657 refined_id = case lookupVarSet binder_set idocc of
659 Nothing -> WARN( True, ppr idocc ) idocc
661 unfold_env' = extendVarEnv unfold_env idocc (name',show_unfold)
662 referrer' | isExportedId refined_id = refined_id
663 | otherwise = referrer
665 search (zip new_ids (repeat referrer') ++ rest) unfold_env' occ_env'
667 tidy_internal :: [Id] -> UnfoldEnv -> TidyOccEnv
668 -> IO (UnfoldEnv, TidyOccEnv)
669 tidy_internal [] unfold_env occ_env = return (unfold_env,occ_env)
670 tidy_internal (id:ids) unfold_env occ_env = do
671 (occ_env', name') <- tidyTopName mod nc_var Nothing occ_env id
672 let unfold_env' = extendVarEnv unfold_env id (name',False)
673 tidy_internal ids unfold_env' occ_env'
675 addExternal :: Id -> ([Id],Bool)
676 addExternal id = (new_needed_ids, show_unfold)
678 new_needed_ids = unfold_ids ++
679 filter (\id -> isLocalId id &&
680 not (id `elemVarSet` unfold_set))
681 (varSetElems spec_ids) -- XXX non-det ordering
684 dont_inline = isNeverActive (inlinePragmaActivation (inlinePragInfo idinfo))
685 loop_breaker = isNonRuleLoopBreaker (occInfo idinfo)
686 bottoming_fn = isBottomingSig (newStrictnessInfo idinfo `orElse` topSig)
687 spec_ids = specInfoFreeVars (specInfo idinfo)
689 -- Stuff to do with the Id's unfolding
690 -- We leave the unfolding there even if there is a worker
691 -- In GHCI the unfolding is used by importers
692 show_unfold = isJust mb_unfold_ids
693 (unfold_set, unfold_ids) = mb_unfold_ids `orElse` (emptyVarSet, [])
695 mb_unfold_ids :: Maybe (IdSet, [Id]) -- Nothing => don't unfold
696 mb_unfold_ids = case unfoldingInfo idinfo of
697 CoreUnfolding { uf_tmpl = unf_rhs, uf_guidance = guide }
698 | not bottoming_fn -- Not necessary
701 , not (neverUnfoldGuidance guide)
702 -> Just (exprFvsInOrder unf_rhs)
703 DFunUnfolding _ ops -> Just (exprsFvsInOrder ops)
706 -- We want a deterministic free-variable list. exprFreeVars gives us
707 -- a VarSet, which is in a non-deterministic order when converted to a
708 -- list. Hence, here we define a free-variable finder that returns
709 -- the free variables in the order that they are encountered.
711 -- Note [choosing external names]
713 exprFvsInOrder :: CoreExpr -> (VarSet, [Id])
714 exprFvsInOrder e = run (dffvExpr e)
716 exprsFvsInOrder :: [CoreExpr] -> (VarSet, [Id])
717 exprsFvsInOrder es = run (mapM_ dffvExpr es)
719 run :: DFFV () -> (VarSet, [Id])
720 run (DFFV m) = case m emptyVarSet [] of
721 (set,ids,_) -> (set,ids)
723 newtype DFFV a = DFFV (VarSet -> [Var] -> (VarSet,[Var],a))
725 instance Monad DFFV where
726 return a = DFFV $ \set ids -> (set, ids, a)
727 (DFFV m) >>= k = DFFV $ \set ids ->
729 (set',ids',a) -> case k a of
730 DFFV f -> f set' ids'
732 insert :: Var -> DFFV ()
733 insert v = DFFV $ \ set ids -> case () of
734 _ | v `elemVarSet` set -> (set,ids,())
735 | otherwise -> (extendVarSet set v, v:ids, ())
737 dffvExpr :: CoreExpr -> DFFV ()
738 dffvExpr e = go emptyVarSet e
740 go scope e = case e of
741 Var v | isLocalId v && not (v `elemVarSet` scope) -> insert v
742 App e1 e2 -> do go scope e1; go scope e2
743 Lam v e -> go (extendVarSet scope v) e
744 Note _ e -> go scope e
745 Cast e _ -> go scope e
746 Let (NonRec x r) e -> do go scope r; go (extendVarSet scope x) e
747 Let (Rec prs) e -> do let scope' = extendVarSetList scope (map fst prs)
748 mapM_ (go scope') (map snd prs)
750 Case e b _ as -> do go scope e
751 mapM_ (go_alt (extendVarSet scope b)) as
754 go_alt scope (_,xs,r) = go (extendVarSetList scope xs) r
758 --------------------------------------------------------------------
760 -- This is where we set names to local/global based on whether they really are
761 -- externally visible (see comment at the top of this module). If the name
762 -- was previously local, we have to give it a unique occurrence name if
763 -- we intend to externalise it.
766 tidyTopName :: Module -> IORef NameCache -> Maybe Id -> TidyOccEnv
767 -> Id -> IO (TidyOccEnv, Name)
768 tidyTopName mod nc_var maybe_ref occ_env id
769 | global && internal = return (occ_env, localiseName name)
771 | global && external = return (occ_env, name)
772 -- Global names are assumed to have been allocated by the renamer,
773 -- so they already have the "right" unique
774 -- And it's a system-wide unique too
776 -- Now we get to the real reason that all this is in the IO Monad:
777 -- we have to update the name cache in a nice atomic fashion
779 | local && internal = do { nc <- readIORef nc_var
780 ; let (nc', new_local_name) = mk_new_local nc
781 ; writeIORef nc_var nc'
782 ; return (occ_env', new_local_name) }
783 -- Even local, internal names must get a unique occurrence, because
784 -- if we do -split-objs we externalise the name later, in the code generator
786 -- Similarly, we must make sure it has a system-wide Unique, because
787 -- the byte-code generator builds a system-wide Name->BCO symbol table
789 | local && external = do { nc <- readIORef nc_var
790 ; let (nc', new_external_name) = mk_new_external nc
791 ; writeIORef nc_var nc'
792 ; return (occ_env', new_external_name) }
794 | otherwise = panic "tidyTopName"
797 external = isJust maybe_ref
798 global = isExternalName name
800 internal = not external
801 loc = nameSrcSpan name
803 old_occ = nameOccName name
805 | Just ref <- maybe_ref, ref /= id =
806 mkOccName (occNameSpace old_occ) $
808 ref_str = occNameString (getOccName ref)
809 occ_str = occNameString old_occ
813 -- workers: the worker for a function already
814 -- includes the occname for its parent, so there's
815 -- no need to prepend the referrer.
816 _other | isSystemName name -> ref_str
817 | otherwise -> ref_str ++ '_' : occ_str
818 -- If this name was system-generated, then don't bother
819 -- to retain its OccName, just use the referrer. These
820 -- system-generated names will become "f1", "f2", etc. for
822 | otherwise = old_occ
824 (occ_env', occ') = tidyOccName occ_env new_occ
826 mk_new_local nc = (nc { nsUniqs = us2 }, mkInternalName uniq occ' loc)
828 (us1, us2) = splitUniqSupply (nsUniqs nc)
829 uniq = uniqFromSupply us1
831 mk_new_external nc = allocateGlobalBinder nc mod occ' loc
832 -- If we want to externalise a currently-local name, check
833 -- whether we have already assigned a unique for it.
834 -- If so, use it; if not, extend the table.
835 -- All this is done by allcoateGlobalBinder.
836 -- This is needed when *re*-compiling a module in GHCi; we must
837 -- use the same name for externally-visible things as we did before.
841 findExternalRules :: Bool -- Omit pragmas
843 -> [CoreRule] -- Local rules for imported fns
844 -> UnfoldEnv -- Ids that are exported, so we need their rules
846 -- The complete rules are gotten by combining
847 -- a) local rules for imported Ids
848 -- b) rules embedded in the top-level Ids
849 findExternalRules omit_prags binds imp_id_rules unfold_env
851 | otherwise = filterOut internal_rule (imp_id_rules ++ local_rules)
854 | id <- bindersOfBinds binds,
856 rule <- idCoreRules id
860 = any (not . external_id) (varSetElems (ruleLhsFreeIds rule))
861 -- Don't export a rule whose LHS mentions a locally-defined
862 -- Id that is completely internal (i.e. not visible to an
866 | Just (name,_) <- lookupVarEnv unfold_env id = isExternalName name
870 Note [Which rules to expose]
871 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
872 findExternalRules filters imp_rules to avoid binders that
873 aren't externally visible; but the externally-visible binders
874 are computed (by findExternalIds) assuming that all orphan
875 rules are externalised (see init_ext_ids in function
876 'search'). So in fact we may export more than we need.
877 (It's a sort of mutual recursion.)
879 %************************************************************************
881 \subsection{Step 2: top-level tidying}
883 %************************************************************************
887 -- TopTidyEnv: when tidying we need to know
888 -- * nc_var: The NameCache, containing a unique supply and any pre-ordained Names.
889 -- These may have arisen because the
890 -- renamer read in an interface file mentioning M.$wf, say,
891 -- and assigned it unique r77. If, on this compilation, we've
892 -- invented an Id whose name is $wf (but with a different unique)
893 -- we want to rename it to have unique r77, so that we can do easy
894 -- comparisons with stuff from the interface file
896 -- * occ_env: The TidyOccEnv, which tells us which local occurrences
899 -- * subst_env: A Var->Var mapping that substitutes the new Var for the old
901 tidyTopBinds :: HscEnv
905 -> (TidyEnv, [CoreBind])
907 tidyTopBinds hsc_env unfold_env init_occ_env binds
908 = tidy init_env binds
910 init_env = (init_occ_env, emptyVarEnv)
912 this_pkg = thisPackage (hsc_dflags hsc_env)
914 tidy env [] = (env, [])
915 tidy env (b:bs) = let (env1, b') = tidyTopBind this_pkg unfold_env env b
916 (env2, bs') = tidy env1 bs
920 ------------------------
921 tidyTopBind :: PackageId
925 -> (TidyEnv, CoreBind)
927 tidyTopBind this_pkg unfold_env (occ_env,subst1) (NonRec bndr rhs)
928 = (tidy_env2, NonRec bndr' rhs')
930 Just (name',show_unfold) = lookupVarEnv unfold_env bndr
931 caf_info = hasCafRefs this_pkg subst1 (idArity bndr) rhs
932 (bndr', rhs') = tidyTopPair show_unfold tidy_env2 caf_info name' (bndr, rhs)
933 subst2 = extendVarEnv subst1 bndr bndr'
934 tidy_env2 = (occ_env, subst2)
936 tidyTopBind this_pkg unfold_env (occ_env,subst1) (Rec prs)
937 = (tidy_env2, Rec prs')
939 prs' = [ tidyTopPair show_unfold tidy_env2 caf_info name' (id,rhs)
941 let (name',show_unfold) =
942 expectJust "tidyTopBind" $ lookupVarEnv unfold_env id
945 subst2 = extendVarEnvList subst1 (bndrs `zip` map fst prs')
946 tidy_env2 = (occ_env, subst2)
950 -- the CafInfo for a recursive group says whether *any* rhs in
951 -- the group may refer indirectly to a CAF (because then, they all do).
953 | or [ mayHaveCafRefs (hasCafRefs this_pkg subst1 (idArity bndr) rhs)
954 | (bndr,rhs) <- prs ] = MayHaveCafRefs
955 | otherwise = NoCafRefs
957 -----------------------------------------------------------
958 tidyTopPair :: Bool -- show unfolding
959 -> TidyEnv -- The TidyEnv is used to tidy the IdInfo
960 -- It is knot-tied: don't look at it!
963 -> (Id, CoreExpr) -- Binder and RHS before tidying
965 -- This function is the heart of Step 2
966 -- The rec_tidy_env is the one to use for the IdInfo
967 -- It's necessary because when we are dealing with a recursive
968 -- group, a variable late in the group might be mentioned
969 -- in the IdInfo of one early in the group
971 tidyTopPair show_unfold rhs_tidy_env caf_info name' (bndr, rhs)
972 = WARN( not _bottom_exposed, ppr bndr1 )
975 -- If the cheap-and-cheerful bottom analyser can see that
976 -- the RHS is bottom, it should jolly well be exposed
977 _bottom_exposed = case exprBotStrictness_maybe rhs of
979 Just (arity, _) -> appIsBottom str arity
981 str = newStrictnessInfo idinfo `orElse` topSig
983 bndr1 = mkGlobalId details name' ty' idinfo'
984 details = idDetails bndr -- Preserve the IdDetails
985 ty' = tidyTopType (idType bndr)
986 rhs1 = tidyExpr rhs_tidy_env rhs
988 idinfo' = tidyTopIdInfo (isExternalName name')
992 unfold_info | show_unfold = tidyUnfolding rhs_tidy_env rhs1 (unfoldingInfo idinfo)
993 | otherwise = noUnfolding
994 -- NB: do *not* expose the worker if show_unfold is off,
995 -- because that means this thing is a loop breaker or
996 -- marked NOINLINE or something like that
997 -- This is important: if you expose the worker for a loop-breaker
998 -- then you can make the simplifier go into an infinite loop, because
999 -- in effect the unfolding is exposed. See Trac #1709
1001 -- You might think that if show_unfold is False, then the thing should
1002 -- not be w/w'd in the first place. But a legitimate reason is this:
1003 -- the function returns bottom
1004 -- In this case, show_unfold will be false (we don't expose unfoldings
1005 -- for bottoming functions), but we might still have a worker/wrapper
1006 -- split (see Note [Worker-wrapper for bottoming functions] in WorkWrap.lhs
1008 -- Usually the Id will have an accurate arity on it, because
1009 -- the simplifier has just run, but not always.
1010 -- One case I found was when the last thing the simplifier
1011 -- did was to let-bind a non-atomic argument and then float
1012 -- it to the top level. So it seems more robust just to
1014 arity = exprArity rhs
1017 -- tidyTopIdInfo creates the final IdInfo for top-level
1018 -- binders. There are two delicate pieces:
1020 -- * Arity. After CoreTidy, this arity must not change any more.
1021 -- Indeed, CorePrep must eta expand where necessary to make
1022 -- the manifest arity equal to the claimed arity.
1024 -- * CAF info. This must also remain valid through to code generation.
1025 -- We add the info here so that it propagates to all
1026 -- occurrences of the binders in RHSs, and hence to occurrences in
1027 -- unfoldings, which are inside Ids imported by GHCi. Ditto RULES.
1028 -- CoreToStg makes use of this when constructing SRTs.
1029 tidyTopIdInfo :: Bool -> IdInfo -> Unfolding
1030 -> ArityInfo -> CafInfo
1032 tidyTopIdInfo is_external idinfo unfold_info arity caf_info
1033 | not is_external -- For internal Ids (not externally visible)
1034 = vanillaIdInfo -- we only need enough info for code generation
1035 -- Arity and strictness info are enough;
1036 -- c.f. CoreTidy.tidyLetBndr
1037 `setCafInfo` caf_info
1038 `setArityInfo` arity
1039 `setAllStrictnessInfo` newStrictnessInfo idinfo
1041 | otherwise -- Externally-visible Ids get the whole lot
1043 `setCafInfo` caf_info
1044 `setArityInfo` arity
1045 `setAllStrictnessInfo` newStrictnessInfo idinfo
1046 `setInlinePragInfo` inlinePragInfo idinfo
1047 `setUnfoldingInfo` unfold_info
1048 -- NB: we throw away the Rules
1049 -- They have already been extracted by findExternalRules
1053 ------------ Unfolding --------------
1054 tidyUnfolding :: TidyEnv -> CoreExpr -> Unfolding -> Unfolding
1055 tidyUnfolding tidy_env _ unf@(CoreUnfolding { uf_tmpl = rhs
1056 , uf_guidance = guide@(InlineRule {}) })
1057 = unf { uf_tmpl = tidyExpr tidy_env rhs, -- Preserves OccInfo
1058 uf_guidance = guide { ug_ir_info = tidyInl tidy_env (ug_ir_info guide) } }
1059 tidyUnfolding tidy_env _ (DFunUnfolding con ids)
1060 = DFunUnfolding con (map (tidyExpr tidy_env) ids)
1061 tidyUnfolding _ tidy_rhs (CoreUnfolding {})
1062 = mkTopUnfolding tidy_rhs
1063 tidyUnfolding _ _ unf = unf
1065 tidyInl :: TidyEnv -> InlineRuleInfo -> InlineRuleInfo
1066 tidyInl tidy_env (InlWrapper w) = InlWrapper (tidyVarOcc tidy_env w)
1067 tidyInl _ inl_info = inl_info
1070 %************************************************************************
1072 \subsection{Figuring out CafInfo for an expression}
1074 %************************************************************************
1076 hasCafRefs decides whether a top-level closure can point into the dynamic heap.
1077 We mark such things as `MayHaveCafRefs' because this information is
1078 used to decide whether a particular closure needs to be referenced
1081 There are two reasons for setting MayHaveCafRefs:
1082 a) The RHS is a CAF: a top-level updatable thunk.
1083 b) The RHS refers to something that MayHaveCafRefs
1085 Possible improvement: In an effort to keep the number of CAFs (and
1086 hence the size of the SRTs) down, we could also look at the expression and
1087 decide whether it requires a small bounded amount of heap, so we can ignore
1088 it as a CAF. In these cases however, we would need to use an additional
1089 CAF list to keep track of non-collectable CAFs.
1092 hasCafRefs :: PackageId -> VarEnv Var -> Arity -> CoreExpr -> CafInfo
1093 hasCafRefs this_pkg p arity expr
1094 | is_caf || mentions_cafs
1096 | otherwise = NoCafRefs
1098 mentions_cafs = isFastTrue (cafRefs p expr)
1099 is_caf = not (arity > 0 || rhsIsStatic this_pkg expr)
1101 -- NB. we pass in the arity of the expression, which is expected
1102 -- to be calculated by exprArity. This is because exprArity
1103 -- knows how much eta expansion is going to be done by
1104 -- CorePrep later on, and we don't want to duplicate that
1105 -- knowledge in rhsIsStatic below.
1107 cafRefs :: VarEnv Id -> Expr a -> FastBool
1109 -- imported Ids first:
1110 | not (isLocalId id) = fastBool (mayHaveCafRefs (idCafInfo id))
1111 -- now Ids local to this module:
1113 case lookupVarEnv p id of
1114 Just id' -> fastBool (mayHaveCafRefs (idCafInfo id'))
1115 Nothing -> fastBool False
1117 cafRefs _ (Lit _) = fastBool False
1118 cafRefs p (App f a) = fastOr (cafRefs p f) (cafRefs p) a
1119 cafRefs p (Lam _ e) = cafRefs p e
1120 cafRefs p (Let b e) = fastOr (cafRefss p (rhssOfBind b)) (cafRefs p) e
1121 cafRefs p (Case e _bndr _ alts) = fastOr (cafRefs p e) (cafRefss p) (rhssOfAlts alts)
1122 cafRefs p (Note _n e) = cafRefs p e
1123 cafRefs p (Cast e _co) = cafRefs p e
1124 cafRefs _ (Type _) = fastBool False
1126 cafRefss :: VarEnv Id -> [Expr a] -> FastBool
1127 cafRefss _ [] = fastBool False
1128 cafRefss p (e:es) = fastOr (cafRefs p e) (cafRefss p) es
1130 fastOr :: FastBool -> (a -> FastBool) -> a -> FastBool
1131 -- hack for lazy-or over FastBool.
1132 fastOr a f x = fastBool (isFastTrue a || isFastTrue (f x))