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, exprBotStrictness_maybe )
23 import Class ( classAllSelIds )
32 import Name hiding (varName)
40 import Packages( isDllName )
45 import FastBool hiding ( fastOr )
49 import Control.Monad ( when )
50 import Data.List ( sortBy )
51 import Data.IORef ( IORef, readIORef, writeIORef )
55 Constructing the TypeEnv, Instances, Rules, VectInfo from which the
56 ModIface is constructed, and which goes on to subsequent modules in
59 Most of the interface file is obtained simply by serialising the
60 TypeEnv. One important consequence is that if the *interface file*
61 has pragma info if and only if the final TypeEnv does. This is not so
62 important for *this* module, but it's essential for ghc --make:
63 subsequent compilations must not see (e.g.) the arity if the interface
64 file does not contain arity If they do, they'll exploit the arity;
65 then the arity might change, but the iface file doesn't change =>
66 recompilation does not happen => disaster.
68 For data types, the final TypeEnv will have a TyThing for the TyCon,
69 plus one for each DataCon; the interface file will contain just one
70 data type declaration, but it is de-serialised back into a collection
73 %************************************************************************
77 %************************************************************************
80 Plan A: mkBootModDetails: omit pragmas, make interfaces small
81 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
84 * Drop all WiredIn things from the TypeEnv
85 (we never want them in interface files)
87 * Retain all TyCons and Classes in the TypeEnv, to avoid
88 having to find which ones are mentioned in the
91 * Trim off the constructors of non-exported TyCons, both
92 from the TyCon and from the TypeEnv
94 * Drop non-exported Ids from the TypeEnv
96 * Tidy the types of the DFunIds of Instances,
97 make them into GlobalIds, (they already have External Names)
98 and add them to the TypeEnv
100 * Tidy the types of the (exported) Ids in the TypeEnv,
101 make them into GlobalIds (they already have External Names)
103 * Drop rules altogether
105 * Tidy the bindings, to ensure that the Caf and Arity
106 information is correct for each top-level binder; the
107 code generator needs it. And to ensure that local names have
108 distinct OccNames in case of object-file splitting
111 -- This is Plan A: make a small type env when typechecking only,
112 -- or when compiling a hs-boot file, or simply when not using -O
114 -- We don't look at the bindings at all -- there aren't any
117 mkBootModDetailsTc :: HscEnv -> TcGblEnv -> IO ModDetails
118 mkBootModDetailsTc hsc_env
119 TcGblEnv{ tcg_exports = exports,
120 tcg_type_env = type_env,
122 tcg_fam_insts = fam_insts
124 = mkBootModDetails hsc_env exports type_env insts fam_insts
126 mkBootModDetailsDs :: HscEnv -> ModGuts -> IO ModDetails
127 mkBootModDetailsDs hsc_env
128 ModGuts{ mg_exports = exports,
131 mg_fam_insts = fam_insts
133 = mkBootModDetails hsc_env exports type_env insts fam_insts
135 mkBootModDetails :: HscEnv -> [AvailInfo] -> NameEnv TyThing
136 -> [Instance] -> [FamInstEnv.FamInst] -> IO ModDetails
137 mkBootModDetails hsc_env exports type_env insts fam_insts
138 = do { let dflags = hsc_dflags hsc_env
139 ; showPass dflags CoreTidy
141 ; let { insts' = tidyInstances globaliseAndTidyId insts
142 ; dfun_ids = map instanceDFunId insts'
143 ; type_env1 = tidyBootTypeEnv (availsToNameSet exports) type_env
144 ; type_env' = extendTypeEnvWithIds type_env1 dfun_ids
146 ; return (ModDetails { md_types = type_env'
148 , md_fam_insts = fam_insts
151 , md_exports = exports
152 , md_vect_info = noVectInfo
157 tidyBootTypeEnv :: NameSet -> TypeEnv -> TypeEnv
158 tidyBootTypeEnv exports type_env
159 = tidyTypeEnv True False exports type_env final_ids
161 -- Find the LocalIds in the type env that are exported
162 -- Make them into GlobalIds, and tidy their types
164 -- It's very important to remove the non-exported ones
165 -- because we don't tidy the OccNames, and if we don't remove
166 -- the non-exported ones we'll get many things with the
167 -- same name in the interface file, giving chaos.
168 final_ids = [ globaliseAndTidyId id
169 | id <- typeEnvIds type_env
173 -- default methods have their export flag set, but everything
174 -- else doesn't (yet), because this is pre-desugaring, so we
176 keep_it id = isExportedId id || idName id `elemNameSet` exports
180 globaliseAndTidyId :: Id -> Id
181 -- Takes an LocalId with an External Name,
182 -- makes it into a GlobalId
183 -- * unchanged Name (might be Internal or External)
184 -- * unchanged details
185 -- * VanillaIdInfo (makes a conservative assumption about Caf-hood)
186 globaliseAndTidyId id
187 = Id.setIdType (globaliseId id) tidy_type
189 tidy_type = tidyTopType (idType id)
193 %************************************************************************
195 Plan B: tidy bindings, make TypeEnv full of IdInfo
197 %************************************************************************
199 Plan B: include pragmas, make interfaces
200 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
201 * Figure out which Ids are externally visible
203 * Tidy the bindings, externalising appropriate Ids
205 * Drop all Ids from the TypeEnv, and add all the External Ids from
206 the bindings. (This adds their IdInfo to the TypeEnv; and adds
207 floated-out Ids that weren't even in the TypeEnv before.)
209 Step 1: Figure out external Ids
210 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
211 Note [choosing external names]
213 See also the section "Interface stability" in the
214 RecompilationAvoidance commentary:
215 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/RecompilationAvoidance
217 First we figure out which Ids are "external" Ids. An
218 "external" Id is one that is visible from outside the compilation
220 a) the user exported ones
221 b) ones mentioned in the unfoldings, workers,
222 or rules of externally-visible ones
224 While figuring out which Ids are external, we pick a "tidy" OccName
225 for each one. That is, we make its OccName distinct from the other
226 external OccNames in this module, so that in interface files and
227 object code we can refer to it unambiguously by its OccName. The
228 OccName for each binder is prefixed by the name of the exported Id
229 that references it; e.g. if "f" references "x" in its unfolding, then
230 "x" is renamed to "f_x". This helps distinguish the different "x"s
231 from each other, and means that if "f" is later removed, things that
232 depend on the other "x"s will not need to be recompiled. Of course,
233 if there are multiple "f_x"s, then we have to disambiguate somehow; we
234 use "f_x0", "f_x1" etc.
236 As far as possible we should assign names in a deterministic fashion.
237 Each time this module is compiled with the same options, we should end
238 up with the same set of external names with the same types. That is,
239 the ABI hash in the interface should not change. This turns out to be
240 quite tricky, since the order of the bindings going into the tidy
241 phase is already non-deterministic, as it is based on the ordering of
242 Uniques, which are assigned unpredictably.
244 To name things in a stable way, we do a depth-first-search of the
245 bindings, starting from the exports sorted by name. This way, as long
246 as the bindings themselves are deterministic (they sometimes aren't!),
247 the order in which they are presented to the tidying phase does not
248 affect the names we assign.
250 Step 2: Tidy the program
251 ~~~~~~~~~~~~~~~~~~~~~~~~
252 Next we traverse the bindings top to bottom. For each *top-level*
255 1. Make it into a GlobalId; its IdDetails becomes VanillaGlobal,
256 reflecting the fact that from now on we regard it as a global,
259 2. Give it a system-wide Unique.
260 [Even non-exported things need system-wide Uniques because the
261 byte-code generator builds a single Name->BCO symbol table.]
263 We use the NameCache kept in the HscEnv as the
264 source of such system-wide uniques.
266 For external Ids, use the original-name cache in the NameCache
267 to ensure that the unique assigned is the same as the Id had
268 in any previous compilation run.
270 3. Rename top-level Ids according to the names we chose in step 1.
271 If it's an external Id, make it have a External Name, otherwise
272 make it have an Internal Name. This is used by the code generator
273 to decide whether to make the label externally visible
275 4. Give it its UTTERLY FINAL IdInfo; in ptic,
276 * its unfolding, if it should have one
278 * its arity, computed from the number of visible lambdas
280 * its CAF info, computed from what is free in its RHS
283 Finally, substitute these new top-level binders consistently
284 throughout, including in unfoldings. We also tidy binders in
285 RHSs, so that they print nicely in interfaces.
288 tidyProgram :: HscEnv -> ModGuts -> IO (CgGuts, ModDetails)
289 tidyProgram hsc_env (ModGuts { mg_module = mod, mg_exports = exports,
291 mg_insts = insts, mg_fam_insts = fam_insts,
293 mg_rules = imp_rules,
294 mg_vect_info = vect_info,
297 mg_foreign = foreign_stubs,
298 mg_hpc_info = hpc_info,
299 mg_modBreaks = modBreaks })
301 = do { let { dflags = hsc_dflags hsc_env
302 ; omit_prags = dopt Opt_OmitInterfacePragmas dflags
303 ; expose_all = dopt Opt_ExposeAllUnfoldings dflags
304 ; th = xopt Opt_TemplateHaskell dflags
306 ; showPass dflags CoreTidy
308 ; let { implicit_binds = getImplicitBinds type_env }
310 ; (unfold_env, tidy_occ_env)
311 <- chooseExternalIds hsc_env mod omit_prags expose_all
312 binds implicit_binds imp_rules
314 ; let { ext_rules = findExternalRules omit_prags binds imp_rules unfold_env }
315 -- Glom together imp_rules and rules currently attached to binders
316 -- Then pick just the ones we need to expose
317 -- See Note [Which rules to expose]
319 ; let { (tidy_env, tidy_binds)
320 = tidyTopBinds hsc_env unfold_env tidy_occ_env binds }
322 ; let { export_set = availsToNameSet exports
323 ; final_ids = [ id | id <- bindersOfBinds tidy_binds,
324 isExternalName (idName id)]
325 ; tidy_type_env = tidyTypeEnv omit_prags th export_set
327 ; tidy_insts = tidyInstances (lookup_dfun tidy_type_env) insts
328 -- A DFunId will have a binding in tidy_binds, and so
329 -- will now be in final_env, replete with IdInfo
330 -- Its name will be unchanged since it was born, but
331 -- we want Global, IdInfo-rich (or not) DFunId in the
334 ; tidy_rules = tidyRules tidy_env ext_rules
335 -- You might worry that the tidy_env contains IdInfo-rich stuff
336 -- and indeed it does, but if omit_prags is on, ext_rules is
339 ; tidy_vect_info = tidyVectInfo tidy_env vect_info
341 -- See Note [Injecting implicit bindings]
342 ; all_tidy_binds = implicit_binds ++ tidy_binds
344 ; alg_tycons = filter isAlgTyCon (typeEnvTyCons type_env)
347 ; endPass dflags CoreTidy all_tidy_binds tidy_rules
349 -- If the endPass didn't print the rules, but ddump-rules is on, print now
350 ; dumpIfSet (dopt Opt_D_dump_rules dflags
351 && (not (dopt Opt_D_dump_simpl dflags)))
353 (ptext (sLit "rules"))
354 (pprRulesForUser tidy_rules)
356 -- Print one-line size info
357 ; let cs = coreBindsStats tidy_binds
358 ; when (dopt Opt_D_dump_core_stats dflags)
359 (printDump (ptext (sLit "Tidy size (terms,types,coercions)")
360 <+> ppr (moduleName mod) <> colon
363 <+> int (cs_co cs) ))
365 ; return (CgGuts { cg_module = mod,
366 cg_tycons = alg_tycons,
367 cg_binds = all_tidy_binds,
368 cg_foreign = foreign_stubs,
369 cg_dep_pkgs = dep_pkgs deps,
370 cg_hpc_info = hpc_info,
371 cg_modBreaks = modBreaks },
373 ModDetails { md_types = tidy_type_env,
374 md_rules = tidy_rules,
375 md_insts = tidy_insts,
376 md_vect_info = tidy_vect_info,
377 md_fam_insts = fam_insts,
378 md_exports = exports,
379 md_anns = anns -- are already tidy
383 lookup_dfun :: TypeEnv -> Var -> Id
384 lookup_dfun type_env dfun_id
385 = case lookupTypeEnv type_env (idName dfun_id) of
386 Just (AnId dfun_id') -> dfun_id'
387 _other -> pprPanic "lookup_dfun" (ppr dfun_id)
389 --------------------------
390 tidyTypeEnv :: Bool -- Compiling without -O, so omit prags
391 -> Bool -- Template Haskell is on
392 -> NameSet -> TypeEnv -> [Id] -> TypeEnv
394 -- The competed type environment is gotten from
395 -- Dropping any wired-in things, and then
396 -- a) keeping the types and classes
397 -- b) removing all Ids,
398 -- c) adding Ids with correct IdInfo, including unfoldings,
399 -- gotten from the bindings
400 -- From (c) we keep only those Ids with External names;
401 -- the CoreTidy pass makes sure these are all and only
402 -- the externally-accessible ones
403 -- This truncates the type environment to include only the
404 -- exported Ids and things needed from them, which saves space
406 tidyTypeEnv omit_prags th exports type_env final_ids
407 = let type_env1 = filterNameEnv keep_it type_env
408 type_env2 = extendTypeEnvWithIds type_env1 final_ids
409 type_env3 | omit_prags = mapNameEnv (trimThing th exports) type_env2
410 | otherwise = type_env2
414 -- We keep GlobalIds, because they won't appear
415 -- in the bindings from which final_ids are derived!
416 -- (The bindings bind LocalIds.)
417 keep_it thing | isWiredInThing thing = False
418 keep_it (AnId id) = isGlobalId id -- Keep GlobalIds (e.g. class ops)
419 keep_it _other = True -- Keep all TyCons, DataCons, and Classes
421 --------------------------
422 isWiredInThing :: TyThing -> Bool
423 isWiredInThing thing = isWiredInName (getName thing)
425 --------------------------
426 trimThing :: Bool -> NameSet -> TyThing -> TyThing
427 -- Trim off inessentials, for boot files and no -O
428 trimThing th exports (ATyCon tc)
429 | not th && not (mustExposeTyCon exports tc)
430 = ATyCon (makeTyConAbstract tc) -- Note [Trimming and Template Haskell]
432 trimThing _th _exports (AnId id)
433 | not (isImplicitId id)
434 = AnId (id `setIdInfo` vanillaIdInfo)
436 trimThing _th _exports other_thing
440 {- Note [Trimming and Template Haskell]
441 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
442 Consider (Trac #2386) this
443 module M(T, makeOne) where
445 makeOne = [| Yay "Yep" |]
446 Notice that T is exported abstractly, but makeOne effectively exports it too!
447 A module that splices in $(makeOne) will then look for a declartion of Yay,
448 so it'd better be there. Hence, brutally but simply, we switch off type
449 constructor trimming if TH is enabled in this module. -}
452 mustExposeTyCon :: NameSet -- Exports
453 -> TyCon -- The tycon
454 -> Bool -- Can its rep be hidden?
455 -- We are compiling without -O, and thus trying to write as little as
456 -- possible into the interface file. But we must expose the details of
457 -- any data types whose constructors or fields are exported
458 mustExposeTyCon exports tc
459 | not (isAlgTyCon tc) -- Synonyms
461 | isEnumerationTyCon tc -- For an enumeration, exposing the constructors
462 = True -- won't lead to the need for further exposure
463 -- (This includes data types with no constructors.)
464 | isFamilyTyCon tc -- Open type family
467 | otherwise -- Newtype, datatype
468 = any exported_con (tyConDataCons tc)
469 -- Expose rep if any datacon or field is exported
471 || (isNewTyCon tc && isFFITy (snd (newTyConRhs tc)))
472 -- Expose the rep for newtypes if the rep is an FFI type.
473 -- For a very annoying reason. 'Foreign import' is meant to
474 -- be able to look through newtypes transparently, but it
475 -- can only do that if it can "see" the newtype representation
477 exported_con con = any (`elemNameSet` exports)
478 (dataConName con : dataConFieldLabels con)
480 tidyInstances :: (DFunId -> DFunId) -> [Instance] -> [Instance]
481 tidyInstances tidy_dfun ispecs
484 tidy ispec = setInstanceDFunId ispec $
485 tidy_dfun (instanceDFunId ispec)
489 tidyVectInfo :: TidyEnv -> VectInfo -> VectInfo
490 tidyVectInfo (_, var_env) info@(VectInfo { vectInfoVar = vars
491 , vectInfoPADFun = pas
492 , vectInfoIso = isos })
493 = info { vectInfoVar = tidy_vars
494 , vectInfoPADFun = tidy_pas
495 , vectInfoIso = tidy_isos }
498 $ map tidy_var_mapping
501 tidy_pas = mapNameEnv tidy_snd_var pas
502 tidy_isos = mapNameEnv tidy_snd_var isos
504 tidy_var_mapping (from, to) = (from', (from', lookup_var to))
505 where from' = lookup_var from
506 tidy_snd_var (x, var) = (x, lookup_var var)
508 lookup_var var = lookupWithDefaultVarEnv var_env var var
512 %************************************************************************
516 %************************************************************************
518 Note [Injecting implicit bindings]
519 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
520 We inject the implict bindings right at the end, in CoreTidy.
521 Some of these bindings, notably record selectors, are not
522 constructed in an optimised form. E.g. record selector for
523 data T = MkT { x :: {-# UNPACK #-} !Int }
524 Then the unfolding looks like
525 x = \t. case t of MkT x1 -> let x = I# x1 in x
526 This generates bad code unless it's first simplified a bit. That is
527 why CoreUnfold.mkImplicitUnfolding uses simleExprOpt to do a bit of
528 optimisation first. (Only matters when the selector is used curried;
529 eg map x ys.) See Trac #2070.
531 [Oct 09: in fact, record selectors are no longer implicit Ids at all,
532 because we really do want to optimise them properly. They are treated
533 much like any other Id. But doing "light" optimisation on an implicit
534 Id still makes sense.]
536 At one time I tried injecting the implicit bindings *early*, at the
537 beginning of SimplCore. But that gave rise to real difficulty,
538 becuase GlobalIds are supposed to have *fixed* IdInfo, but the
539 simplifier and other core-to-core passes mess with IdInfo all the
540 time. The straw that broke the camels back was when a class selector
541 got the wrong arity -- ie the simplifier gave it arity 2, whereas
542 importing modules were expecting it to have arity 1 (Trac #2844).
543 It's much safer just to inject them right at the end, after tidying.
545 Oh: two other reasons for injecting them late:
547 - If implicit Ids are already in the bindings when we start TidyPgm,
548 we'd have to be careful not to treat them as external Ids (in
549 the sense of findExternalIds); else the Ids mentioned in *their*
550 RHSs will be treated as external and you get an interface file
552 but nothing refererring to a18 (because the implicit Id is the
553 one that does, and implicit Ids don't appear in interface files).
555 - More seriously, the tidied type-envt will include the implicit
556 Id replete with a18 in its unfolding; but we won't take account
557 of a18 when computing a fingerprint for the class; result chaos.
559 There is one sort of implicit binding that is injected still later,
560 namely those for data constructor workers. Reason (I think): it's
561 really just a code generation trick.... binding itself makes no sense.
562 See CorePrep Note [Data constructor workers].
565 getImplicitBinds :: TypeEnv -> [CoreBind]
566 getImplicitBinds type_env
567 = map get_defn (concatMap implicit_ids (typeEnvElts type_env))
569 implicit_ids (ATyCon tc) = mapCatMaybes dataConWrapId_maybe (tyConDataCons tc)
570 implicit_ids (AClass cls) = classAllSelIds cls
573 get_defn :: Id -> CoreBind
574 get_defn id = NonRec id (unfoldingTemplate (realIdUnfolding id))
578 %************************************************************************
580 \subsection{Step 1: finding externals}
582 %************************************************************************
584 Sete Note [choosing external names].
587 type UnfoldEnv = IdEnv (Name{-new name-}, Bool {-show unfolding-})
588 -- Maps each top-level Id to its new Name (the Id is tidied in step 2)
589 -- The Unique is unchanged. If the new Name is external, it will be
590 -- visible in the interface file.
592 -- Bool => expose unfolding or not.
594 chooseExternalIds :: HscEnv
600 -> IO (UnfoldEnv, TidyOccEnv)
601 -- Step 1 from the notes above
603 chooseExternalIds hsc_env mod omit_prags expose_all binds implicit_binds imp_id_rules
604 = do { (unfold_env1,occ_env1) <- search init_work_list emptyVarEnv init_occ_env
605 ; let internal_ids = filter (not . (`elemVarEnv` unfold_env1)) binders
606 ; tidy_internal internal_ids unfold_env1 occ_env1 }
608 nc_var = hsc_NC hsc_env
610 -- init_ext_ids is the intial list of Ids that should be
611 -- externalised. It serves as the starting point for finding a
612 -- deterministic, tidy, renaming for all external Ids in this
615 -- It is sorted, so that it has adeterministic order (i.e. it's the
616 -- same list every time this module is compiled), in contrast to the
617 -- bindings, which are ordered non-deterministically.
618 init_work_list = zip init_ext_ids init_ext_ids
619 init_ext_ids = sortBy (compare `on` getOccName) $
620 filter is_external binders
622 -- An Id should be external if either (a) it is exported or
623 -- (b) it appears in the RHS of a local rule for an imported Id.
624 -- See Note [Which rules to expose]
625 is_external id = isExportedId id || id `elemVarSet` rule_rhs_vars
626 rule_rhs_vars = foldr (unionVarSet . ruleRhsFreeVars) emptyVarSet imp_id_rules
628 binders = bindersOfBinds binds
629 implicit_binders = bindersOfBinds implicit_binds
630 binder_set = mkVarSet binders
632 avoids = [getOccName name | bndr <- binders ++ implicit_binders,
633 let name = idName bndr,
634 isExternalName name ]
635 -- In computing our "avoids" list, we must include
637 -- all things with global names (assigned once and for
638 -- all by the renamer)
639 -- since their names are "taken".
640 -- The type environment is a convenient source of such things.
641 -- In particular, the set of binders doesn't include
642 -- implicit Ids at this stage.
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_occ_env = initTidyOccEnv avoids
655 search :: [(Id,Id)] -- The work-list: (external id, referrring id)
656 -- Make a tidy, external Name for the external id,
657 -- add it to the UnfoldEnv, and do the same for the
658 -- transitive closure of Ids it refers to
659 -- The referring id is used to generate a tidy
660 --- name for the external id
661 -> UnfoldEnv -- id -> (new Name, show_unfold)
662 -> TidyOccEnv -- occ env for choosing new Names
663 -> IO (UnfoldEnv, TidyOccEnv)
665 search [] unfold_env occ_env = return (unfold_env, occ_env)
667 search ((idocc,referrer) : rest) unfold_env occ_env
668 | idocc `elemVarEnv` unfold_env = search rest unfold_env occ_env
670 (occ_env', name') <- tidyTopName mod nc_var (Just referrer) occ_env idocc
672 (new_ids, show_unfold)
673 | omit_prags = ([], False)
674 | otherwise = addExternal expose_all refined_id
676 -- 'idocc' is an *occurrence*, but we need to see the
677 -- unfolding in the *definition*; so look up in binder_set
678 refined_id = case lookupVarSet binder_set idocc of
680 Nothing -> WARN( True, ppr idocc ) idocc
682 unfold_env' = extendVarEnv unfold_env idocc (name',show_unfold)
683 referrer' | isExportedId refined_id = refined_id
684 | otherwise = referrer
686 search (zip new_ids (repeat referrer') ++ rest) unfold_env' occ_env'
688 tidy_internal :: [Id] -> UnfoldEnv -> TidyOccEnv
689 -> IO (UnfoldEnv, TidyOccEnv)
690 tidy_internal [] unfold_env occ_env = return (unfold_env,occ_env)
691 tidy_internal (id:ids) unfold_env occ_env = do
692 (occ_env', name') <- tidyTopName mod nc_var Nothing occ_env id
693 let unfold_env' = extendVarEnv unfold_env id (name',False)
694 tidy_internal ids unfold_env' occ_env'
696 addExternal :: Bool -> Id -> ([Id],Bool)
697 addExternal expose_all id = (new_needed_ids, show_unfold)
699 new_needed_ids = unfold_ids ++
700 filter (\id -> isLocalId id &&
701 not (id `elemVarSet` unfold_set))
702 (varSetElems spec_ids) -- XXX non-det ordering
705 never_active = isNeverActive (inlinePragmaActivation (inlinePragInfo idinfo))
706 loop_breaker = isNonRuleLoopBreaker (occInfo idinfo)
707 bottoming_fn = isBottomingSig (strictnessInfo idinfo `orElse` topSig)
708 spec_ids = specInfoFreeVars (specInfo idinfo)
710 -- Stuff to do with the Id's unfolding
711 -- We leave the unfolding there even if there is a worker
712 -- In GHCI the unfolding is used by importers
713 show_unfold = isJust mb_unfold_ids
714 (unfold_set, unfold_ids) = mb_unfold_ids `orElse` (emptyVarSet, [])
716 mb_unfold_ids :: Maybe (IdSet, [Id]) -- Nothing => don't unfold
717 mb_unfold_ids = case unfoldingInfo idinfo of
718 CoreUnfolding { uf_tmpl = unf_rhs, uf_src = src, uf_guidance = guide }
719 | show_unfolding src guide
720 -> Just (unf_ext_ids src unf_rhs)
721 DFunUnfolding _ _ ops -> Just (exprsFvsInOrder (dfunArgExprs ops))
724 unf_ext_ids (InlineWrapper v) _ = (unitVarSet v, [v])
725 unf_ext_ids _ unf_rhs = exprFvsInOrder unf_rhs
726 -- For a wrapper, externalise the wrapper id rather than the
727 -- fvs of the rhs. The two usually come down to the same thing
728 -- but I've seen cases where we had a wrapper id $w but a
729 -- rhs where $w had been inlined; see Trac #3922
731 show_unfolding unf_source unf_guidance
732 = expose_all -- 'expose_all' says to expose all
733 -- unfoldings willy-nilly
735 || isStableSource unf_source -- Always expose things whose
736 -- source is an inline rule
738 || not (bottoming_fn -- No need to inline bottom functions
739 || never_active -- Or ones that say not to
740 || loop_breaker -- Or that are loop breakers
741 || neverUnfoldGuidance unf_guidance)
743 -- We want a deterministic free-variable list. exprFreeVars gives us
744 -- a VarSet, which is in a non-deterministic order when converted to a
745 -- list. Hence, here we define a free-variable finder that returns
746 -- the free variables in the order that they are encountered.
748 -- Note [choosing external names]
750 exprFvsInOrder :: CoreExpr -> (VarSet, [Id])
751 exprFvsInOrder e = run (dffvExpr e)
753 exprsFvsInOrder :: [CoreExpr] -> (VarSet, [Id])
754 exprsFvsInOrder es = run (mapM_ dffvExpr es)
756 run :: DFFV () -> (VarSet, [Id])
757 run (DFFV m) = case m emptyVarSet [] of
758 (set,ids,_) -> (set,ids)
760 newtype DFFV a = DFFV (VarSet -> [Var] -> (VarSet,[Var],a))
762 instance Monad DFFV where
763 return a = DFFV $ \set ids -> (set, ids, a)
764 (DFFV m) >>= k = DFFV $ \set ids ->
766 (set',ids',a) -> case k a of
767 DFFV f -> f set' ids'
769 insert :: Var -> DFFV ()
770 insert v = DFFV $ \ set ids -> case () of
771 _ | v `elemVarSet` set -> (set,ids,())
772 | otherwise -> (extendVarSet set v, v:ids, ())
774 dffvExpr :: CoreExpr -> DFFV ()
775 dffvExpr e = go emptyVarSet e
777 go scope e = case e of
778 Var v | isLocalId v && not (v `elemVarSet` scope) -> insert v
779 App e1 e2 -> do go scope e1; go scope e2
780 Lam v e -> go (extendVarSet scope v) e
781 Note _ e -> go scope e
782 Cast e _ -> go scope e
783 Let (NonRec x r) e -> do go scope r; go (extendVarSet scope x) e
784 Let (Rec prs) e -> do let scope' = extendVarSetList scope (map fst prs)
785 mapM_ (go scope') (map snd prs)
787 Case e b _ as -> do go scope e
788 mapM_ (go_alt (extendVarSet scope b)) as
791 go_alt scope (_,xs,r) = go (extendVarSetList scope xs) r
795 --------------------------------------------------------------------
797 -- This is where we set names to local/global based on whether they really are
798 -- externally visible (see comment at the top of this module). If the name
799 -- was previously local, we have to give it a unique occurrence name if
800 -- we intend to externalise it.
803 tidyTopName :: Module -> IORef NameCache -> Maybe Id -> TidyOccEnv
804 -> Id -> IO (TidyOccEnv, Name)
805 tidyTopName mod nc_var maybe_ref occ_env id
806 | global && internal = return (occ_env, localiseName name)
808 | global && external = return (occ_env, name)
809 -- Global names are assumed to have been allocated by the renamer,
810 -- so they already have the "right" unique
811 -- And it's a system-wide unique too
813 -- Now we get to the real reason that all this is in the IO Monad:
814 -- we have to update the name cache in a nice atomic fashion
816 | local && internal = do { nc <- readIORef nc_var
817 ; let (nc', new_local_name) = mk_new_local nc
818 ; writeIORef nc_var nc'
819 ; return (occ_env', new_local_name) }
820 -- Even local, internal names must get a unique occurrence, because
821 -- if we do -split-objs we externalise the name later, in the code generator
823 -- Similarly, we must make sure it has a system-wide Unique, because
824 -- the byte-code generator builds a system-wide Name->BCO symbol table
826 | local && external = do { nc <- readIORef nc_var
827 ; let (nc', new_external_name) = mk_new_external nc
828 ; writeIORef nc_var nc'
829 ; return (occ_env', new_external_name) }
831 | otherwise = panic "tidyTopName"
834 external = isJust maybe_ref
835 global = isExternalName name
837 internal = not external
838 loc = nameSrcSpan name
840 old_occ = nameOccName name
842 | Just ref <- maybe_ref, ref /= id =
843 mkOccName (occNameSpace old_occ) $
845 ref_str = occNameString (getOccName ref)
846 occ_str = occNameString old_occ
850 -- workers: the worker for a function already
851 -- includes the occname for its parent, so there's
852 -- no need to prepend the referrer.
853 _other | isSystemName name -> ref_str
854 | otherwise -> ref_str ++ '_' : occ_str
855 -- If this name was system-generated, then don't bother
856 -- to retain its OccName, just use the referrer. These
857 -- system-generated names will become "f1", "f2", etc. for
859 | otherwise = old_occ
861 (occ_env', occ') = tidyOccName occ_env new_occ
863 mk_new_local nc = (nc { nsUniqs = us }, mkInternalName uniq occ' loc)
865 (uniq, us) = takeUniqFromSupply (nsUniqs nc)
867 mk_new_external nc = allocateGlobalBinder nc mod occ' loc
868 -- If we want to externalise a currently-local name, check
869 -- whether we have already assigned a unique for it.
870 -- If so, use it; if not, extend the table.
871 -- All this is done by allcoateGlobalBinder.
872 -- This is needed when *re*-compiling a module in GHCi; we must
873 -- use the same name for externally-visible things as we did before.
877 findExternalRules :: Bool -- Omit pragmas
879 -> [CoreRule] -- Local rules for imported fns
880 -> UnfoldEnv -- Ids that are exported, so we need their rules
882 -- The complete rules are gotten by combining
883 -- a) local rules for imported Ids
884 -- b) rules embedded in the top-level Ids
885 findExternalRules omit_prags binds imp_id_rules unfold_env
887 | otherwise = filterOut internal_rule (imp_id_rules ++ local_rules)
890 | id <- bindersOfBinds binds,
892 rule <- idCoreRules id
896 = any (not . external_id) (varSetElems (ruleLhsFreeIds rule))
897 -- Don't export a rule whose LHS mentions a locally-defined
898 -- Id that is completely internal (i.e. not visible to an
902 | Just (name,_) <- lookupVarEnv unfold_env id = isExternalName name
906 Note [Which rules to expose]
907 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
908 findExternalRules filters imp_rules to avoid binders that
909 aren't externally visible; but the externally-visible binders
910 are computed (by findExternalIds) assuming that all orphan
911 rules are externalised (see init_ext_ids in function
912 'search'). So in fact we may export more than we need.
913 (It's a sort of mutual recursion.)
915 %************************************************************************
917 \subsection{Step 2: top-level tidying}
919 %************************************************************************
923 -- TopTidyEnv: when tidying we need to know
924 -- * nc_var: The NameCache, containing a unique supply and any pre-ordained Names.
925 -- These may have arisen because the
926 -- renamer read in an interface file mentioning M.$wf, say,
927 -- and assigned it unique r77. If, on this compilation, we've
928 -- invented an Id whose name is $wf (but with a different unique)
929 -- we want to rename it to have unique r77, so that we can do easy
930 -- comparisons with stuff from the interface file
932 -- * occ_env: The TidyOccEnv, which tells us which local occurrences
935 -- * subst_env: A Var->Var mapping that substitutes the new Var for the old
937 tidyTopBinds :: HscEnv
941 -> (TidyEnv, [CoreBind])
943 tidyTopBinds hsc_env unfold_env init_occ_env binds
944 = tidy init_env binds
946 init_env = (init_occ_env, emptyVarEnv)
948 this_pkg = thisPackage (hsc_dflags hsc_env)
950 tidy env [] = (env, [])
951 tidy env (b:bs) = let (env1, b') = tidyTopBind this_pkg unfold_env env b
952 (env2, bs') = tidy env1 bs
956 ------------------------
957 tidyTopBind :: PackageId
961 -> (TidyEnv, CoreBind)
963 tidyTopBind this_pkg unfold_env (occ_env,subst1) (NonRec bndr rhs)
964 = (tidy_env2, NonRec bndr' rhs')
966 Just (name',show_unfold) = lookupVarEnv unfold_env bndr
967 caf_info = hasCafRefs this_pkg subst1 (idArity bndr) rhs
968 (bndr', rhs') = tidyTopPair show_unfold tidy_env2 caf_info name' (bndr, rhs)
969 subst2 = extendVarEnv subst1 bndr bndr'
970 tidy_env2 = (occ_env, subst2)
972 tidyTopBind this_pkg unfold_env (occ_env,subst1) (Rec prs)
973 = (tidy_env2, Rec prs')
975 prs' = [ tidyTopPair show_unfold tidy_env2 caf_info name' (id,rhs)
977 let (name',show_unfold) =
978 expectJust "tidyTopBind" $ lookupVarEnv unfold_env id
981 subst2 = extendVarEnvList subst1 (bndrs `zip` map fst prs')
982 tidy_env2 = (occ_env, subst2)
986 -- the CafInfo for a recursive group says whether *any* rhs in
987 -- the group may refer indirectly to a CAF (because then, they all do).
989 | or [ mayHaveCafRefs (hasCafRefs this_pkg subst1 (idArity bndr) rhs)
990 | (bndr,rhs) <- prs ] = MayHaveCafRefs
991 | otherwise = NoCafRefs
993 -----------------------------------------------------------
994 tidyTopPair :: Bool -- show unfolding
995 -> TidyEnv -- The TidyEnv is used to tidy the IdInfo
996 -- It is knot-tied: don't look at it!
999 -> (Id, CoreExpr) -- Binder and RHS before tidying
1001 -- This function is the heart of Step 2
1002 -- The rec_tidy_env is the one to use for the IdInfo
1003 -- It's necessary because when we are dealing with a recursive
1004 -- group, a variable late in the group might be mentioned
1005 -- in the IdInfo of one early in the group
1007 tidyTopPair show_unfold rhs_tidy_env caf_info name' (bndr, rhs)
1010 bndr1 = mkGlobalId details name' ty' idinfo'
1011 details = idDetails bndr -- Preserve the IdDetails
1012 ty' = tidyTopType (idType bndr)
1013 rhs1 = tidyExpr rhs_tidy_env rhs
1014 idinfo' = tidyTopIdInfo rhs_tidy_env name' rhs rhs1 (idInfo bndr)
1015 show_unfold caf_info
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 :: TidyEnv -> Name -> CoreExpr -> CoreExpr
1030 -> IdInfo -> Bool -> CafInfo -> IdInfo
1031 tidyTopIdInfo rhs_tidy_env name orig_rhs tidy_rhs idinfo show_unfold caf_info
1032 | not is_external -- For internal Ids (not externally visible)
1033 = vanillaIdInfo -- we only need enough info for code generation
1034 -- Arity and strictness info are enough;
1035 -- c.f. CoreTidy.tidyLetBndr
1036 `setCafInfo` caf_info
1037 `setArityInfo` arity
1038 `setStrictnessInfo` final_sig
1040 | otherwise -- Externally-visible Ids get the whole lot
1042 `setCafInfo` caf_info
1043 `setArityInfo` arity
1044 `setStrictnessInfo` final_sig
1045 `setOccInfo` robust_occ_info
1046 `setInlinePragInfo` (inlinePragInfo idinfo)
1047 `setUnfoldingInfo` unfold_info
1048 -- NB: we throw away the Rules
1049 -- They have already been extracted by findExternalRules
1051 is_external = isExternalName name
1053 --------- OccInfo ------------
1054 robust_occ_info = zapFragileOcc (occInfo idinfo)
1055 -- It's important to keep loop-breaker information
1056 -- when we are doing -fexpose-all-unfoldings
1058 --------- Strictness ------------
1059 final_sig | Just sig <- strictnessInfo idinfo
1060 = WARN( _bottom_hidden sig, ppr name ) Just sig
1061 | Just (_, sig) <- mb_bot_str = Just sig
1062 | otherwise = Nothing
1064 -- If the cheap-and-cheerful bottom analyser can see that
1065 -- the RHS is bottom, it should jolly well be exposed
1066 _bottom_hidden id_sig = case mb_bot_str of
1068 Just (arity, _) -> not (appIsBottom id_sig arity)
1070 mb_bot_str = exprBotStrictness_maybe orig_rhs
1072 --------- Unfolding ------------
1073 unf_info = unfoldingInfo idinfo
1074 unfold_info | show_unfold = tidyUnfolding rhs_tidy_env unf_info unf_from_rhs
1075 | otherwise = noUnfolding
1076 unf_from_rhs = mkTopUnfolding is_bot tidy_rhs
1077 is_bot = case final_sig of
1078 Just sig -> isBottomingSig sig
1080 -- NB: do *not* expose the worker if show_unfold is off,
1081 -- because that means this thing is a loop breaker or
1082 -- marked NOINLINE or something like that
1083 -- This is important: if you expose the worker for a loop-breaker
1084 -- then you can make the simplifier go into an infinite loop, because
1085 -- in effect the unfolding is exposed. See Trac #1709
1087 -- You might think that if show_unfold is False, then the thing should
1088 -- not be w/w'd in the first place. But a legitimate reason is this:
1089 -- the function returns bottom
1090 -- In this case, show_unfold will be false (we don't expose unfoldings
1091 -- for bottoming functions), but we might still have a worker/wrapper
1092 -- split (see Note [Worker-wrapper for bottoming functions] in WorkWrap.lhs
1094 --------- Arity ------------
1095 -- Usually the Id will have an accurate arity on it, because
1096 -- the simplifier has just run, but not always.
1097 -- One case I found was when the last thing the simplifier
1098 -- did was to let-bind a non-atomic argument and then float
1099 -- it to the top level. So it seems more robust just to
1101 arity = exprArity orig_rhs
1104 %************************************************************************
1106 \subsection{Figuring out CafInfo for an expression}
1108 %************************************************************************
1110 hasCafRefs decides whether a top-level closure can point into the dynamic heap.
1111 We mark such things as `MayHaveCafRefs' because this information is
1112 used to decide whether a particular closure needs to be referenced
1115 There are two reasons for setting MayHaveCafRefs:
1116 a) The RHS is a CAF: a top-level updatable thunk.
1117 b) The RHS refers to something that MayHaveCafRefs
1119 Possible improvement: In an effort to keep the number of CAFs (and
1120 hence the size of the SRTs) down, we could also look at the expression and
1121 decide whether it requires a small bounded amount of heap, so we can ignore
1122 it as a CAF. In these cases however, we would need to use an additional
1123 CAF list to keep track of non-collectable CAFs.
1126 hasCafRefs :: PackageId -> VarEnv Var -> Arity -> CoreExpr -> CafInfo
1127 hasCafRefs this_pkg p arity expr
1128 | is_caf || mentions_cafs = MayHaveCafRefs
1129 | otherwise = NoCafRefs
1131 mentions_cafs = isFastTrue (cafRefs p expr)
1132 is_dynamic_name = isDllName this_pkg
1133 is_caf = not (arity > 0 || rhsIsStatic is_dynamic_name expr)
1135 -- NB. we pass in the arity of the expression, which is expected
1136 -- to be calculated by exprArity. This is because exprArity
1137 -- knows how much eta expansion is going to be done by
1138 -- CorePrep later on, and we don't want to duplicate that
1139 -- knowledge in rhsIsStatic below.
1141 cafRefs :: VarEnv Id -> Expr a -> FastBool
1143 -- imported Ids first:
1144 | not (isLocalId id) = fastBool (mayHaveCafRefs (idCafInfo id))
1145 -- now Ids local to this module:
1147 case lookupVarEnv p id of
1148 Just id' -> fastBool (mayHaveCafRefs (idCafInfo id'))
1149 Nothing -> fastBool False
1151 cafRefs _ (Lit _) = fastBool False
1152 cafRefs p (App f a) = fastOr (cafRefs p f) (cafRefs p) a
1153 cafRefs p (Lam _ e) = cafRefs p e
1154 cafRefs p (Let b e) = fastOr (cafRefss p (rhssOfBind b)) (cafRefs p) e
1155 cafRefs p (Case e _bndr _ alts) = fastOr (cafRefs p e) (cafRefss p) (rhssOfAlts alts)
1156 cafRefs p (Note _n e) = cafRefs p e
1157 cafRefs p (Cast e _co) = cafRefs p e
1158 cafRefs _ (Type _) = fastBool False
1160 cafRefss :: VarEnv Id -> [Expr a] -> FastBool
1161 cafRefss _ [] = fastBool False
1162 cafRefss p (e:es) = fastOr (cafRefs p e) (cafRefss p) es
1164 fastOr :: FastBool -> (a -> FastBool) -> a -> FastBool
1165 -- hack for lazy-or over FastBool.
1166 fastOr a f x = fastBool (isFastTrue a || isFastTrue (f x))