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 ; expose_all = dopt Opt_ExposeAllUnfoldings dflags
302 ; th = dopt Opt_TemplateHaskell dflags
304 ; showPass dflags "Tidy Core"
306 ; let { implicit_binds = getImplicitBinds type_env }
308 ; (unfold_env, tidy_occ_env)
309 <- chooseExternalIds hsc_env mod omit_prags expose_all
310 binds implicit_binds imp_rules
312 ; let { ext_rules = findExternalRules omit_prags binds imp_rules unfold_env }
313 -- Glom together imp_rules and rules currently attached to binders
314 -- Then pick just the ones we need to expose
315 -- See Note [Which rules to expose]
317 ; let { (tidy_env, tidy_binds)
318 = tidyTopBinds hsc_env unfold_env tidy_occ_env binds }
320 ; let { export_set = availsToNameSet exports
321 ; final_ids = [ id | id <- bindersOfBinds tidy_binds,
322 isExternalName (idName id)]
323 ; tidy_type_env = tidyTypeEnv omit_prags th export_set
325 ; tidy_insts = tidyInstances (lookup_dfun tidy_type_env) insts
326 -- A DFunId will have a binding in tidy_binds, and so
327 -- will now be in final_env, replete with IdInfo
328 -- Its name will be unchanged since it was born, but
329 -- we want Global, IdInfo-rich (or not) DFunId in the
332 ; tidy_rules = tidyRules tidy_env ext_rules
333 -- You might worry that the tidy_env contains IdInfo-rich stuff
334 -- and indeed it does, but if omit_prags is on, ext_rules is
337 ; tidy_vect_info = tidyVectInfo tidy_env vect_info
339 -- See Note [Injecting implicit bindings]
340 ; all_tidy_binds = implicit_binds ++ tidy_binds
342 ; alg_tycons = filter isAlgTyCon (typeEnvTyCons type_env)
345 ; endPass dflags "Tidy Core" Opt_D_dump_simpl all_tidy_binds tidy_rules
346 ; let dir_imp_mods = moduleEnvKeys dir_imps
348 ; return (CgGuts { cg_module = mod,
349 cg_tycons = alg_tycons,
350 cg_binds = all_tidy_binds,
351 cg_dir_imps = dir_imp_mods,
352 cg_foreign = foreign_stubs,
353 cg_dep_pkgs = dep_pkgs deps,
354 cg_hpc_info = hpc_info,
355 cg_modBreaks = modBreaks },
357 ModDetails { md_types = tidy_type_env,
358 md_rules = tidy_rules,
359 md_insts = tidy_insts,
360 md_vect_info = tidy_vect_info,
361 md_fam_insts = fam_insts,
362 md_exports = exports,
363 md_anns = anns -- are already tidy
367 lookup_dfun :: TypeEnv -> Var -> Id
368 lookup_dfun type_env dfun_id
369 = case lookupTypeEnv type_env (idName dfun_id) of
370 Just (AnId dfun_id') -> dfun_id'
371 _other -> pprPanic "lookup_dfun" (ppr dfun_id)
373 --------------------------
374 tidyTypeEnv :: Bool -- Compiling without -O, so omit prags
375 -> Bool -- Template Haskell is on
376 -> NameSet -> TypeEnv -> [Id] -> TypeEnv
378 -- The competed type environment is gotten from
379 -- Dropping any wired-in things, and then
380 -- a) keeping the types and classes
381 -- b) removing all Ids,
382 -- c) adding Ids with correct IdInfo, including unfoldings,
383 -- gotten from the bindings
384 -- From (c) we keep only those Ids with External names;
385 -- the CoreTidy pass makes sure these are all and only
386 -- the externally-accessible ones
387 -- This truncates the type environment to include only the
388 -- exported Ids and things needed from them, which saves space
390 tidyTypeEnv omit_prags th exports type_env final_ids
391 = let type_env1 = filterNameEnv keep_it type_env
392 type_env2 = extendTypeEnvWithIds type_env1 final_ids
393 type_env3 | omit_prags = mapNameEnv (trimThing th exports) type_env2
394 | otherwise = type_env2
398 -- We keep GlobalIds, because they won't appear
399 -- in the bindings from which final_ids are derived!
400 -- (The bindings bind LocalIds.)
401 keep_it thing | isWiredInThing thing = False
402 keep_it (AnId id) = isGlobalId id -- Keep GlobalIds (e.g. class ops)
403 keep_it _other = True -- Keep all TyCons, DataCons, and Classes
405 --------------------------
406 isWiredInThing :: TyThing -> Bool
407 isWiredInThing thing = isWiredInName (getName thing)
409 --------------------------
410 trimThing :: Bool -> NameSet -> TyThing -> TyThing
411 -- Trim off inessentials, for boot files and no -O
412 trimThing th exports (ATyCon tc)
413 | not th && not (mustExposeTyCon exports tc)
414 = ATyCon (makeTyConAbstract tc) -- Note [Trimming and Template Haskell]
416 trimThing _th _exports (AnId id)
417 | not (isImplicitId id)
418 = AnId (id `setIdInfo` vanillaIdInfo)
420 trimThing _th _exports other_thing
424 {- Note [Trimming and Template Haskell]
425 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
426 Consider (Trac #2386) this
427 module M(T, makeOne) where
429 makeOne = [| Yay "Yep" |]
430 Notice that T is exported abstractly, but makeOne effectively exports it too!
431 A module that splices in $(makeOne) will then look for a declartion of Yay,
432 so it'd better be there. Hence, brutally but simply, we switch off type
433 constructor trimming if TH is enabled in this module. -}
436 mustExposeTyCon :: NameSet -- Exports
437 -> TyCon -- The tycon
438 -> Bool -- Can its rep be hidden?
439 -- We are compiling without -O, and thus trying to write as little as
440 -- possible into the interface file. But we must expose the details of
441 -- any data types whose constructors or fields are exported
442 mustExposeTyCon exports tc
443 | not (isAlgTyCon tc) -- Synonyms
445 | isEnumerationTyCon tc -- For an enumeration, exposing the constructors
446 = True -- won't lead to the need for further exposure
447 -- (This includes data types with no constructors.)
448 | isOpenTyCon tc -- Open type family
451 | otherwise -- Newtype, datatype
452 = any exported_con (tyConDataCons tc)
453 -- Expose rep if any datacon or field is exported
455 || (isNewTyCon tc && isFFITy (snd (newTyConRhs tc)))
456 -- Expose the rep for newtypes if the rep is an FFI type.
457 -- For a very annoying reason. 'Foreign import' is meant to
458 -- be able to look through newtypes transparently, but it
459 -- can only do that if it can "see" the newtype representation
461 exported_con con = any (`elemNameSet` exports)
462 (dataConName con : dataConFieldLabels con)
464 tidyInstances :: (DFunId -> DFunId) -> [Instance] -> [Instance]
465 tidyInstances tidy_dfun ispecs
468 tidy ispec = setInstanceDFunId ispec $
469 tidy_dfun (instanceDFunId ispec)
473 tidyVectInfo :: TidyEnv -> VectInfo -> VectInfo
474 tidyVectInfo (_, var_env) info@(VectInfo { vectInfoVar = vars
475 , vectInfoPADFun = pas
476 , vectInfoIso = isos })
477 = info { vectInfoVar = tidy_vars
478 , vectInfoPADFun = tidy_pas
479 , vectInfoIso = tidy_isos }
482 $ map tidy_var_mapping
485 tidy_pas = mapNameEnv tidy_snd_var pas
486 tidy_isos = mapNameEnv tidy_snd_var isos
488 tidy_var_mapping (from, to) = (from', (from', lookup_var to))
489 where from' = lookup_var from
490 tidy_snd_var (x, var) = (x, lookup_var var)
492 lookup_var var = lookupWithDefaultVarEnv var_env var var
496 %************************************************************************
500 %************************************************************************
502 Note [Injecting implicit bindings]
503 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
504 We inject the implict bindings right at the end, in CoreTidy.
505 Some of these bindings, notably record selectors, are not
506 constructed in an optimised form. E.g. record selector for
507 data T = MkT { x :: {-# UNPACK #-} !Int }
508 Then the unfolding looks like
509 x = \t. case t of MkT x1 -> let x = I# x1 in x
510 This generates bad code unless it's first simplified a bit. That is
511 why CoreUnfold.mkImplicitUnfolding uses simleExprOpt to do a bit of
512 optimisation first. (Only matters when the selector is used curried;
513 eg map x ys.) See Trac #2070.
515 [Oct 09: in fact, record selectors are no longer implicit Ids at all,
516 because we really do want to optimise them properly. They are treated
517 much like any other Id. But doing "light" optimisation on an implicit
518 Id still makes sense.]
520 At one time I tried injecting the implicit bindings *early*, at the
521 beginning of SimplCore. But that gave rise to real difficulty,
522 becuase GlobalIds are supposed to have *fixed* IdInfo, but the
523 simplifier and other core-to-core passes mess with IdInfo all the
524 time. The straw that broke the camels back was when a class selector
525 got the wrong arity -- ie the simplifier gave it arity 2, whereas
526 importing modules were expecting it to have arity 1 (Trac #2844).
527 It's much safer just to inject them right at the end, after tidying.
529 Oh: two other reasons for injecting them late:
531 - If implicit Ids are already in the bindings when we start TidyPgm,
532 we'd have to be careful not to treat them as external Ids (in
533 the sense of findExternalIds); else the Ids mentioned in *their*
534 RHSs will be treated as external and you get an interface file
536 but nothing refererring to a18 (because the implicit Id is the
537 one that does, and implicit Ids don't appear in interface files).
539 - More seriously, the tidied type-envt will include the implicit
540 Id replete with a18 in its unfolding; but we won't take account
541 of a18 when computing a fingerprint for the class; result chaos.
543 There is one sort of implicit binding that is injected still later,
544 namely those for data constructor workers. Reason (I think): it's
545 really just a code generation trick.... binding itself makes no sense.
546 See CorePrep Note [Data constructor workers].
549 getImplicitBinds :: TypeEnv -> [CoreBind]
550 getImplicitBinds type_env
551 = map get_defn (concatMap implicit_ids (typeEnvElts type_env))
553 implicit_ids (ATyCon tc) = mapCatMaybes dataConWrapId_maybe (tyConDataCons tc)
554 implicit_ids (AClass cls) = classSelIds cls
557 get_defn :: Id -> CoreBind
558 get_defn id = NonRec id (unfoldingTemplate (realIdUnfolding id))
562 %************************************************************************
564 \subsection{Step 1: finding externals}
566 %************************************************************************
568 Sete Note [choosing external names].
571 type UnfoldEnv = IdEnv (Name{-new name-}, Bool {-show unfolding-})
572 -- Maps each top-level Id to its new Name (the Id is tidied in step 2)
573 -- The Unique is unchanged. If the new Name is external, it will be
574 -- visible in the interface file.
576 -- Bool => expose unfolding or not.
578 chooseExternalIds :: HscEnv
584 -> IO (UnfoldEnv, TidyOccEnv)
585 -- Step 1 from the notes above
587 chooseExternalIds hsc_env mod omit_prags expose_all binds implicit_binds imp_id_rules
588 = do { (unfold_env1,occ_env1) <- search init_work_list emptyVarEnv init_occ_env
589 ; let internal_ids = filter (not . (`elemVarEnv` unfold_env1)) binders
590 ; tidy_internal internal_ids unfold_env1 occ_env1 }
592 nc_var = hsc_NC hsc_env
594 -- init_ext_ids is the intial list of Ids that should be
595 -- externalised. It serves as the starting point for finding a
596 -- deterministic, tidy, renaming for all external Ids in this
599 -- It is sorted, so that it has adeterministic order (i.e. it's the
600 -- same list every time this module is compiled), in contrast to the
601 -- bindings, which are ordered non-deterministically.
602 init_work_list = zip init_ext_ids init_ext_ids
603 init_ext_ids = sortBy (compare `on` getOccName) $
604 filter is_external binders
606 -- An Id should be external if either (a) it is exported or
607 -- (b) it appears in the RHS of a local rule for an imported Id.
608 -- See Note [Which rules to expose]
609 is_external id = isExportedId id || id `elemVarSet` rule_rhs_vars
610 rule_rhs_vars = foldr (unionVarSet . ruleRhsFreeVars) emptyVarSet imp_id_rules
612 binders = bindersOfBinds binds
613 implicit_binders = bindersOfBinds implicit_binds
614 binder_set = mkVarSet binders
616 avoids = [getOccName name | bndr <- binders ++ implicit_binders,
617 let name = idName bndr,
618 isExternalName name ]
619 -- In computing our "avoids" list, we must include
621 -- all things with global names (assigned once and for
622 -- all by the renamer)
623 -- since their names are "taken".
624 -- The type environment is a convenient source of such things.
625 -- In particular, the set of binders doesn't include
626 -- implicit Ids at this stage.
628 -- We also make sure to avoid any exported binders. Consider
629 -- f{-u1-} = 1 -- Local decl
631 -- f{-u2-} = 2 -- Exported decl
633 -- The second exported decl must 'get' the name 'f', so we
634 -- have to put 'f' in the avoids list before we get to the first
635 -- decl. tidyTopId then does a no-op on exported binders.
636 init_occ_env = initTidyOccEnv avoids
639 search :: [(Id,Id)] -- The work-list: (external id, referrring id)
640 -- Make a tidy, external Name for the external id,
641 -- add it to the UnfoldEnv, and do the same for the
642 -- transitive closure of Ids it refers to
643 -- The referring id is used to generate a tidy
644 --- name for the external id
645 -> UnfoldEnv -- id -> (new Name, show_unfold)
646 -> TidyOccEnv -- occ env for choosing new Names
647 -> IO (UnfoldEnv, TidyOccEnv)
649 search [] unfold_env occ_env = return (unfold_env, occ_env)
651 search ((idocc,referrer) : rest) unfold_env occ_env
652 | idocc `elemVarEnv` unfold_env = search rest unfold_env occ_env
654 (occ_env', name') <- tidyTopName mod nc_var (Just referrer) occ_env idocc
656 (new_ids, show_unfold)
657 | omit_prags = ([], False)
658 | otherwise = addExternal expose_all refined_id
660 -- 'idocc' is an *occurrence*, but we need to see the
661 -- unfolding in the *definition*; so look up in binder_set
662 refined_id = case lookupVarSet binder_set idocc of
664 Nothing -> WARN( True, ppr idocc ) idocc
666 unfold_env' = extendVarEnv unfold_env idocc (name',show_unfold)
667 referrer' | isExportedId refined_id = refined_id
668 | otherwise = referrer
670 search (zip new_ids (repeat referrer') ++ rest) unfold_env' occ_env'
672 tidy_internal :: [Id] -> UnfoldEnv -> TidyOccEnv
673 -> IO (UnfoldEnv, TidyOccEnv)
674 tidy_internal [] unfold_env occ_env = return (unfold_env,occ_env)
675 tidy_internal (id:ids) unfold_env occ_env = do
676 (occ_env', name') <- tidyTopName mod nc_var Nothing occ_env id
677 let unfold_env' = extendVarEnv unfold_env id (name',False)
678 tidy_internal ids unfold_env' occ_env'
680 addExternal :: Bool -> Id -> ([Id],Bool)
681 addExternal expose_all id = (new_needed_ids, show_unfold)
683 new_needed_ids = unfold_ids ++
684 filter (\id -> isLocalId id &&
685 not (id `elemVarSet` unfold_set))
686 (varSetElems spec_ids) -- XXX non-det ordering
689 never_active = isNeverActive (inlinePragmaActivation (inlinePragInfo idinfo))
690 loop_breaker = isNonRuleLoopBreaker (occInfo idinfo)
691 bottoming_fn = isBottomingSig (strictnessInfo idinfo `orElse` topSig)
692 spec_ids = specInfoFreeVars (specInfo idinfo)
694 -- Stuff to do with the Id's unfolding
695 -- We leave the unfolding there even if there is a worker
696 -- In GHCI the unfolding is used by importers
697 show_unfold = isJust mb_unfold_ids
698 (unfold_set, unfold_ids) = mb_unfold_ids `orElse` (emptyVarSet, [])
700 mb_unfold_ids :: Maybe (IdSet, [Id]) -- Nothing => don't unfold
701 mb_unfold_ids = case unfoldingInfo idinfo of
702 CoreUnfolding { uf_tmpl = unf_rhs, uf_src = src, uf_guidance = guide }
703 | show_unfolding src guide
704 -> Just (exprFvsInOrder unf_rhs)
705 DFunUnfolding _ ops -> Just (exprsFvsInOrder ops)
708 show_unfolding unf_source unf_guidance
709 = expose_all -- 'expose_all' says to expose all
710 -- unfoldings willy-nilly
712 || isInlineRuleSource unf_source -- Always expose things whose
713 -- source is an inline rule
715 || not (bottoming_fn -- No need to inline bottom functions
716 || never_active -- Or ones that say not to
717 || loop_breaker -- Or that are loop breakers
718 || neverUnfoldGuidance unf_guidance)
720 -- We want a deterministic free-variable list. exprFreeVars gives us
721 -- a VarSet, which is in a non-deterministic order when converted to a
722 -- list. Hence, here we define a free-variable finder that returns
723 -- the free variables in the order that they are encountered.
725 -- Note [choosing external names]
727 exprFvsInOrder :: CoreExpr -> (VarSet, [Id])
728 exprFvsInOrder e = run (dffvExpr e)
730 exprsFvsInOrder :: [CoreExpr] -> (VarSet, [Id])
731 exprsFvsInOrder es = run (mapM_ dffvExpr es)
733 run :: DFFV () -> (VarSet, [Id])
734 run (DFFV m) = case m emptyVarSet [] of
735 (set,ids,_) -> (set,ids)
737 newtype DFFV a = DFFV (VarSet -> [Var] -> (VarSet,[Var],a))
739 instance Monad DFFV where
740 return a = DFFV $ \set ids -> (set, ids, a)
741 (DFFV m) >>= k = DFFV $ \set ids ->
743 (set',ids',a) -> case k a of
744 DFFV f -> f set' ids'
746 insert :: Var -> DFFV ()
747 insert v = DFFV $ \ set ids -> case () of
748 _ | v `elemVarSet` set -> (set,ids,())
749 | otherwise -> (extendVarSet set v, v:ids, ())
751 dffvExpr :: CoreExpr -> DFFV ()
752 dffvExpr e = go emptyVarSet e
754 go scope e = case e of
755 Var v | isLocalId v && not (v `elemVarSet` scope) -> insert v
756 App e1 e2 -> do go scope e1; go scope e2
757 Lam v e -> go (extendVarSet scope v) e
758 Note _ e -> go scope e
759 Cast e _ -> go scope e
760 Let (NonRec x r) e -> do go scope r; go (extendVarSet scope x) e
761 Let (Rec prs) e -> do let scope' = extendVarSetList scope (map fst prs)
762 mapM_ (go scope') (map snd prs)
764 Case e b _ as -> do go scope e
765 mapM_ (go_alt (extendVarSet scope b)) as
768 go_alt scope (_,xs,r) = go (extendVarSetList scope xs) r
772 --------------------------------------------------------------------
774 -- This is where we set names to local/global based on whether they really are
775 -- externally visible (see comment at the top of this module). If the name
776 -- was previously local, we have to give it a unique occurrence name if
777 -- we intend to externalise it.
780 tidyTopName :: Module -> IORef NameCache -> Maybe Id -> TidyOccEnv
781 -> Id -> IO (TidyOccEnv, Name)
782 tidyTopName mod nc_var maybe_ref occ_env id
783 | global && internal = return (occ_env, localiseName name)
785 | global && external = return (occ_env, name)
786 -- Global names are assumed to have been allocated by the renamer,
787 -- so they already have the "right" unique
788 -- And it's a system-wide unique too
790 -- Now we get to the real reason that all this is in the IO Monad:
791 -- we have to update the name cache in a nice atomic fashion
793 | local && internal = do { nc <- readIORef nc_var
794 ; let (nc', new_local_name) = mk_new_local nc
795 ; writeIORef nc_var nc'
796 ; return (occ_env', new_local_name) }
797 -- Even local, internal names must get a unique occurrence, because
798 -- if we do -split-objs we externalise the name later, in the code generator
800 -- Similarly, we must make sure it has a system-wide Unique, because
801 -- the byte-code generator builds a system-wide Name->BCO symbol table
803 | local && external = do { nc <- readIORef nc_var
804 ; let (nc', new_external_name) = mk_new_external nc
805 ; writeIORef nc_var nc'
806 ; return (occ_env', new_external_name) }
808 | otherwise = panic "tidyTopName"
811 external = isJust maybe_ref
812 global = isExternalName name
814 internal = not external
815 loc = nameSrcSpan name
817 old_occ = nameOccName name
819 | Just ref <- maybe_ref, ref /= id =
820 mkOccName (occNameSpace old_occ) $
822 ref_str = occNameString (getOccName ref)
823 occ_str = occNameString old_occ
827 -- workers: the worker for a function already
828 -- includes the occname for its parent, so there's
829 -- no need to prepend the referrer.
830 _other | isSystemName name -> ref_str
831 | otherwise -> ref_str ++ '_' : occ_str
832 -- If this name was system-generated, then don't bother
833 -- to retain its OccName, just use the referrer. These
834 -- system-generated names will become "f1", "f2", etc. for
836 | otherwise = old_occ
838 (occ_env', occ') = tidyOccName occ_env new_occ
840 mk_new_local nc = (nc { nsUniqs = us2 }, mkInternalName uniq occ' loc)
842 (us1, us2) = splitUniqSupply (nsUniqs nc)
843 uniq = uniqFromSupply us1
845 mk_new_external nc = allocateGlobalBinder nc mod occ' loc
846 -- If we want to externalise a currently-local name, check
847 -- whether we have already assigned a unique for it.
848 -- If so, use it; if not, extend the table.
849 -- All this is done by allcoateGlobalBinder.
850 -- This is needed when *re*-compiling a module in GHCi; we must
851 -- use the same name for externally-visible things as we did before.
855 findExternalRules :: Bool -- Omit pragmas
857 -> [CoreRule] -- Local rules for imported fns
858 -> UnfoldEnv -- Ids that are exported, so we need their rules
860 -- The complete rules are gotten by combining
861 -- a) local rules for imported Ids
862 -- b) rules embedded in the top-level Ids
863 findExternalRules omit_prags binds imp_id_rules unfold_env
865 | otherwise = filterOut internal_rule (imp_id_rules ++ local_rules)
868 | id <- bindersOfBinds binds,
870 rule <- idCoreRules id
874 = any (not . external_id) (varSetElems (ruleLhsFreeIds rule))
875 -- Don't export a rule whose LHS mentions a locally-defined
876 -- Id that is completely internal (i.e. not visible to an
880 | Just (name,_) <- lookupVarEnv unfold_env id = isExternalName name
884 Note [Which rules to expose]
885 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
886 findExternalRules filters imp_rules to avoid binders that
887 aren't externally visible; but the externally-visible binders
888 are computed (by findExternalIds) assuming that all orphan
889 rules are externalised (see init_ext_ids in function
890 'search'). So in fact we may export more than we need.
891 (It's a sort of mutual recursion.)
893 %************************************************************************
895 \subsection{Step 2: top-level tidying}
897 %************************************************************************
901 -- TopTidyEnv: when tidying we need to know
902 -- * nc_var: The NameCache, containing a unique supply and any pre-ordained Names.
903 -- These may have arisen because the
904 -- renamer read in an interface file mentioning M.$wf, say,
905 -- and assigned it unique r77. If, on this compilation, we've
906 -- invented an Id whose name is $wf (but with a different unique)
907 -- we want to rename it to have unique r77, so that we can do easy
908 -- comparisons with stuff from the interface file
910 -- * occ_env: The TidyOccEnv, which tells us which local occurrences
913 -- * subst_env: A Var->Var mapping that substitutes the new Var for the old
915 tidyTopBinds :: HscEnv
919 -> (TidyEnv, [CoreBind])
921 tidyTopBinds hsc_env unfold_env init_occ_env binds
922 = tidy init_env binds
924 init_env = (init_occ_env, emptyVarEnv)
926 this_pkg = thisPackage (hsc_dflags hsc_env)
928 tidy env [] = (env, [])
929 tidy env (b:bs) = let (env1, b') = tidyTopBind this_pkg unfold_env env b
930 (env2, bs') = tidy env1 bs
934 ------------------------
935 tidyTopBind :: PackageId
939 -> (TidyEnv, CoreBind)
941 tidyTopBind this_pkg unfold_env (occ_env,subst1) (NonRec bndr rhs)
942 = (tidy_env2, NonRec bndr' rhs')
944 Just (name',show_unfold) = lookupVarEnv unfold_env bndr
945 caf_info = hasCafRefs this_pkg subst1 (idArity bndr) rhs
946 (bndr', rhs') = tidyTopPair show_unfold tidy_env2 caf_info name' (bndr, rhs)
947 subst2 = extendVarEnv subst1 bndr bndr'
948 tidy_env2 = (occ_env, subst2)
950 tidyTopBind this_pkg unfold_env (occ_env,subst1) (Rec prs)
951 = (tidy_env2, Rec prs')
953 prs' = [ tidyTopPair show_unfold tidy_env2 caf_info name' (id,rhs)
955 let (name',show_unfold) =
956 expectJust "tidyTopBind" $ lookupVarEnv unfold_env id
959 subst2 = extendVarEnvList subst1 (bndrs `zip` map fst prs')
960 tidy_env2 = (occ_env, subst2)
964 -- the CafInfo for a recursive group says whether *any* rhs in
965 -- the group may refer indirectly to a CAF (because then, they all do).
967 | or [ mayHaveCafRefs (hasCafRefs this_pkg subst1 (idArity bndr) rhs)
968 | (bndr,rhs) <- prs ] = MayHaveCafRefs
969 | otherwise = NoCafRefs
971 -----------------------------------------------------------
972 tidyTopPair :: Bool -- show unfolding
973 -> TidyEnv -- The TidyEnv is used to tidy the IdInfo
974 -- It is knot-tied: don't look at it!
977 -> (Id, CoreExpr) -- Binder and RHS before tidying
979 -- This function is the heart of Step 2
980 -- The rec_tidy_env is the one to use for the IdInfo
981 -- It's necessary because when we are dealing with a recursive
982 -- group, a variable late in the group might be mentioned
983 -- in the IdInfo of one early in the group
985 tidyTopPair show_unfold rhs_tidy_env caf_info name' (bndr, rhs)
988 bndr1 = mkGlobalId details name' ty' idinfo'
989 details = idDetails bndr -- Preserve the IdDetails
990 ty' = tidyTopType (idType bndr)
991 rhs1 = tidyExpr rhs_tidy_env rhs
992 idinfo' = tidyTopIdInfo rhs_tidy_env name' rhs rhs1 (idInfo bndr)
995 -- tidyTopIdInfo creates the final IdInfo for top-level
996 -- binders. There are two delicate pieces:
998 -- * Arity. After CoreTidy, this arity must not change any more.
999 -- Indeed, CorePrep must eta expand where necessary to make
1000 -- the manifest arity equal to the claimed arity.
1002 -- * CAF info. This must also remain valid through to code generation.
1003 -- We add the info here so that it propagates to all
1004 -- occurrences of the binders in RHSs, and hence to occurrences in
1005 -- unfoldings, which are inside Ids imported by GHCi. Ditto RULES.
1006 -- CoreToStg makes use of this when constructing SRTs.
1007 tidyTopIdInfo :: TidyEnv -> Name -> CoreExpr -> CoreExpr
1008 -> IdInfo -> Bool -> CafInfo -> IdInfo
1009 tidyTopIdInfo rhs_tidy_env name orig_rhs tidy_rhs idinfo show_unfold caf_info
1010 | not is_external -- For internal Ids (not externally visible)
1011 = vanillaIdInfo -- we only need enough info for code generation
1012 -- Arity and strictness info are enough;
1013 -- c.f. CoreTidy.tidyLetBndr
1014 `setCafInfo` caf_info
1015 `setArityInfo` arity
1016 `setStrictnessInfo` final_sig
1018 | otherwise -- Externally-visible Ids get the whole lot
1020 `setCafInfo` caf_info
1021 `setArityInfo` arity
1022 `setStrictnessInfo` final_sig
1023 `setOccInfo` robust_occ_info
1024 `setInlinePragInfo` (inlinePragInfo idinfo)
1025 `setUnfoldingInfo` unfold_info
1026 -- NB: we throw away the Rules
1027 -- They have already been extracted by findExternalRules
1029 is_external = isExternalName name
1031 --------- OccInfo ------------
1032 robust_occ_info = zapFragileOcc (occInfo idinfo)
1033 -- It's important to keep loop-breaker information
1034 -- when we are doing -fexpose-all-unfoldings
1036 --------- Strictness ------------
1037 final_sig | Just sig <- strictnessInfo idinfo
1038 = WARN( _bottom_hidden sig, ppr name ) Just sig
1039 | Just (_, sig) <- mb_bot_str = Just sig
1040 | otherwise = Nothing
1042 -- If the cheap-and-cheerful bottom analyser can see that
1043 -- the RHS is bottom, it should jolly well be exposed
1044 _bottom_hidden id_sig = case mb_bot_str of
1046 Just (arity, _) -> not (appIsBottom id_sig arity)
1048 mb_bot_str = exprBotStrictness_maybe orig_rhs
1050 --------- Unfolding ------------
1051 unf_info = unfoldingInfo idinfo
1052 unfold_info | show_unfold = tidyUnfolding rhs_tidy_env tidy_rhs final_sig unf_info
1053 | otherwise = noUnfolding
1054 -- NB: do *not* expose the worker if show_unfold is off,
1055 -- because that means this thing is a loop breaker or
1056 -- marked NOINLINE or something like that
1057 -- This is important: if you expose the worker for a loop-breaker
1058 -- then you can make the simplifier go into an infinite loop, because
1059 -- in effect the unfolding is exposed. See Trac #1709
1061 -- You might think that if show_unfold is False, then the thing should
1062 -- not be w/w'd in the first place. But a legitimate reason is this:
1063 -- the function returns bottom
1064 -- In this case, show_unfold will be false (we don't expose unfoldings
1065 -- for bottoming functions), but we might still have a worker/wrapper
1066 -- split (see Note [Worker-wrapper for bottoming functions] in WorkWrap.lhs
1068 --------- Arity ------------
1069 -- Usually the Id will have an accurate arity on it, because
1070 -- the simplifier has just run, but not always.
1071 -- One case I found was when the last thing the simplifier
1072 -- did was to let-bind a non-atomic argument and then float
1073 -- it to the top level. So it seems more robust just to
1075 arity = exprArity orig_rhs
1079 ------------ Unfolding --------------
1080 tidyUnfolding :: TidyEnv -> CoreExpr -> Maybe StrictSig -> Unfolding -> Unfolding
1081 tidyUnfolding tidy_env _ _ (DFunUnfolding con ids)
1082 = DFunUnfolding con (map (tidyExpr tidy_env) ids)
1083 tidyUnfolding tidy_env tidy_rhs strict_sig
1084 unf@(CoreUnfolding { uf_tmpl = unf_rhs, uf_src = src })
1085 | isInlineRuleSource src
1086 = unf { uf_tmpl = tidyExpr tidy_env unf_rhs, -- Preserves OccInfo
1087 uf_src = tidyInl tidy_env src }
1089 = mkTopUnfolding is_bot tidy_rhs
1091 is_bot = case strict_sig of
1092 Just sig -> isBottomingSig sig
1095 tidyUnfolding _ _ _ unf = unf
1097 tidyInl :: TidyEnv -> UnfoldingSource -> UnfoldingSource
1098 tidyInl tidy_env (InlineWrapper w) = InlineWrapper (tidyVarOcc tidy_env w)
1099 tidyInl _ inl_info = inl_info
1102 %************************************************************************
1104 \subsection{Figuring out CafInfo for an expression}
1106 %************************************************************************
1108 hasCafRefs decides whether a top-level closure can point into the dynamic heap.
1109 We mark such things as `MayHaveCafRefs' because this information is
1110 used to decide whether a particular closure needs to be referenced
1113 There are two reasons for setting MayHaveCafRefs:
1114 a) The RHS is a CAF: a top-level updatable thunk.
1115 b) The RHS refers to something that MayHaveCafRefs
1117 Possible improvement: In an effort to keep the number of CAFs (and
1118 hence the size of the SRTs) down, we could also look at the expression and
1119 decide whether it requires a small bounded amount of heap, so we can ignore
1120 it as a CAF. In these cases however, we would need to use an additional
1121 CAF list to keep track of non-collectable CAFs.
1124 hasCafRefs :: PackageId -> VarEnv Var -> Arity -> CoreExpr -> CafInfo
1125 hasCafRefs this_pkg p arity expr
1126 | is_caf || mentions_cafs
1128 | otherwise = NoCafRefs
1130 mentions_cafs = isFastTrue (cafRefs p expr)
1131 is_caf = not (arity > 0 || rhsIsStatic this_pkg expr)
1133 -- NB. we pass in the arity of the expression, which is expected
1134 -- to be calculated by exprArity. This is because exprArity
1135 -- knows how much eta expansion is going to be done by
1136 -- CorePrep later on, and we don't want to duplicate that
1137 -- knowledge in rhsIsStatic below.
1139 cafRefs :: VarEnv Id -> Expr a -> FastBool
1141 -- imported Ids first:
1142 | not (isLocalId id) = fastBool (mayHaveCafRefs (idCafInfo id))
1143 -- now Ids local to this module:
1145 case lookupVarEnv p id of
1146 Just id' -> fastBool (mayHaveCafRefs (idCafInfo id'))
1147 Nothing -> fastBool False
1149 cafRefs _ (Lit _) = fastBool False
1150 cafRefs p (App f a) = fastOr (cafRefs p f) (cafRefs p) a
1151 cafRefs p (Lam _ e) = cafRefs p e
1152 cafRefs p (Let b e) = fastOr (cafRefss p (rhssOfBind b)) (cafRefs p) e
1153 cafRefs p (Case e _bndr _ alts) = fastOr (cafRefs p e) (cafRefss p) (rhssOfAlts alts)
1154 cafRefs p (Note _n e) = cafRefs p e
1155 cafRefs p (Cast e _co) = cafRefs p e
1156 cafRefs _ (Type _) = fastBool False
1158 cafRefss :: VarEnv Id -> [Expr a] -> FastBool
1159 cafRefss _ [] = fastBool False
1160 cafRefss p (e:es) = fastOr (cafRefs p e) (cafRefss p) es
1162 fastOr :: FastBool -> (a -> FastBool) -> a -> FastBool
1163 -- hack for lazy-or over FastBool.
1164 fastOr a f x = fastBool (isFastTrue a || isFastTrue (f x))