summaryrefslogtreecommitdiff
path: root/compiler
Commit message (Collapse)AuthorAgeFilesLines
...
* Use a more efficient printer for code generation (#21853)Krzysztof Gogolewski2022-11-1135-698/+1213
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The changes in `GHC.Utils.Outputable` are the bulk of the patch and drive the rest. The types `HLine` and `HDoc` in Outputable can be used instead of `SDoc` and support printing directly to a handle with `bPutHDoc`. See Note [SDoc versus HDoc] and Note [HLine versus HDoc]. The classes `IsLine` and `IsDoc` are used to make the existing code polymorphic over `HLine`/`HDoc` and `SDoc`. This is done for X86, PPC, AArch64, DWARF and dependencies (printing module names, labels etc.). Co-authored-by: Alexis King <lexi.lambda@gmail.com> Metric Decrease: CoOpt_Read ManyAlternatives ManyConstructors T10421 T12425 T12707 T13035 T13056 T13253 T13379 T18140 T18282 T18698a T18698b T1969 T20049 T21839c T21839r T3064 T3294 T4801 T5321FD T5321Fun T5631 T6048 T783 T9198 T9233
* Add a fast path for data constructor workersSimon Peyton Jones2022-11-113-18/+55
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | See Note [Fast path for data constructors] in GHC.Core.Opt.Simplify.Iteration This bypasses lots of expensive logic, in the special case of applications of data constructors. It is a surprisingly worthwhile improvement, as you can see in the figures below. Metrics: compile_time/bytes allocated ------------------------------------------------ CoOpt_Read(normal) -2.0% CoOpt_Singletons(normal) -2.0% ManyConstructors(normal) -1.3% T10421(normal) -1.9% GOOD T10421a(normal) -1.5% T10858(normal) -1.6% T11545(normal) -1.7% T12234(optasm) -1.3% T12425(optasm) -1.9% GOOD T13035(normal) -1.0% GOOD T13056(optasm) -1.8% T13253(normal) -3.3% GOOD T15164(normal) -1.7% T15304(normal) -3.4% T15630(normal) -2.8% T16577(normal) -4.3% GOOD T17096(normal) -1.1% T17516(normal) -3.1% T18282(normal) -1.9% T18304(normal) -1.2% T18698a(normal) -1.2% GOOD T18698b(normal) -1.5% GOOD T18923(normal) -1.3% T1969(normal) -1.3% GOOD T19695(normal) -4.4% GOOD T21839c(normal) -2.7% GOOD T21839r(normal) -2.7% GOOD T4801(normal) -3.8% GOOD T5642(normal) -3.1% GOOD T6048(optasm) -2.5% GOOD T9020(optasm) -2.7% GOOD T9630(normal) -2.1% GOOD T9961(normal) -11.7% GOOD WWRec(normal) -1.0% geo. mean -1.1% minimum -11.7% maximum +0.1% Metric Decrease: T10421 T12425 T13035 T13253 T16577 T18698a T18698b T1969 T19695 T21839c T21839r T4801 T5642 T6048 T9020 T9630 T9961
* driver: Fix -fdefer-diagnostics flagMatthew Pickering2022-11-111-2/+2
| | | | | | | | | The `withDeferredDiagnostics` wrapper wasn't doing anything because the session it was modifying wasn't used in hsc_env. Therefore the fix is simple, just push the `getSession` call into the scope of `withDeferredDiagnostics`. Fixes #22391
* compiler: wasm32 NCGCheng Shao2022-11-117-0/+2722
| | | | This patch adds the wasm32 NCG.
* compiler: annotate CmmFileEmbed with blob lengthCheng Shao2022-11-116-12/+12
| | | | | This patch adds the blob length field to CmmFileEmbed. The wasm32 NCG needs to know the precise size of each data segment.
* compiler: enforce cmm switch planning for wasm32Cheng Shao2022-11-111-1/+2
| | | | | | This patch forcibly enable Cmm switch planning for wasm32, since otherwise the switch tables we generate may exceed the br_table maximum allowed size.
* driver: pass -Wa,--no-type-check for wasm32 when runAsPhaseCheng Shao2022-11-111-0/+31
| | | | | This patch passes -Wa,--no-type-check for wasm32 when compiling assembly. See the added note for more detailed explanation.
* compiler: allow big arith for wasm32Cheng Shao2022-11-111-1/+1
| | | | | This patch enables Cmm big arithmetic on wasm32, since 64-bit arithmetic can be efficiently lowered to wasm32 opcodes.
* driver: avoid -Wl,--no-as-needed for wasm32Cheng Shao2022-11-111-1/+1
| | | | | | The driver used to pass -Wl,--no-as-needed for LLD linking. This is actually only supported for ELF targets, and must be avoided when linking for wasm32.
* compiler: add util functions for UniqFM and UniqMapCheng Shao2022-11-112-1/+41
| | | | | | This patch adds addToUFM_L (backed by insertLookupWithKey), addToUniqMap_L and intersectUniqMap_C. These UniqFM/UniqMap util functions are used by the wasm32 NCG.
* Add register mapping for wasm32Cheng Shao2022-11-114-1/+19
| | | | | This patch adds register mapping logic for wasm32. See Note [Register mapping on WebAssembly] in wasm32 NCG for more description.
* Add support for the wasm32-wasi target tupleCheng Shao2022-11-116-0/+12
| | | | | | This patch adds the wasm32-wasi tuple support to various places in the tree: autoconf, hadrian, ghc-boot and also the compiler. The codegen logic will come in subsequent commits.
* add new modules for reducibility and WebAssembly translationNorman Ramsey2022-11-115-0/+898
|
* add the two key graph modules from Martin Erwig's FGLNorman Ramsey2022-11-114-0/+1022
| | | | | | | | | | | | | | | | | | | | | | | | | | Martin Erwig's FGL (Functional Graph Library) provides an "inductive" representation of graphs. A general graph has labeled nodes and labeled edges. The key operation on a graph is to decompose it by removing one node, together with the edges that connect the node to the rest of the graph. There is also an inverse composition operation. The decomposition and composition operations make this representation of graphs exceptionally well suited to implement graph algorithms in which the graph is continually changing, as alluded to in #21259. This commit adds `GHC.Data.Graph.Inductive.Graph`, which defines the interface, and `GHC.Data.Graph.Inductive.PatriciaTree`, which provides an implementation. Both modules are taken from `fgl-5.7.0.3` on Hackage, with these changes: - Copyright and license text have been copied into the files themselves, not stored separately. - Some calls to `error` have been replaced with calls to `panic`. - Conditional-compilation support for older versions of GHC, `containers`, and `base` has been removed.
* Fix Cmm symbol kindCheng Shao2022-11-111-3/+6
|
* Boxity: Handle argument budget of unboxed tuples correctly (#21737)Sebastian Graf2022-11-102-50/+174
| | | | | | | Now Budget roughly tracks the combined width of all arguments after unarisation. See the changes to `Note [Worker argument budgets]`. Fixes #21737.
* WorkWrap: Unboxing unboxed tuples is not always useful (#22388)Sebastian Graf2022-11-101-12/+28
| | | | | | See Note [Unboxing through unboxed tuples]. Fixes #22388.
* Fix DsUselessSpecialiseForClassMethodSelector msgwip/T21851Simon Peyton Jones2022-11-101-1/+1
| | | | | The error message for DsUselessSpecialiseForClassMethodSelector was just wrong (a typo in some earlier work); trivial fix
* Make indexError work betterSimon Peyton Jones2022-11-101-0/+3
| | | | | | | | | | | | | | | The problem here is described at some length in Note [Boxity for bottoming functions] and Note [Reboxed crud for bottoming calls] in GHC.Core.Opt.DmdAnal. This patch adds a SPECIALISE pragma for indexError, which makes it much less vulnerable to the problem described in these Notes. (This came up in another line of work, where a small change made indexError do reboxing (in nofib/spectral/simple/table_sort) that didn't happen before my change. I've opened #22404 to document the fagility.
* Fire RULES in the SpecialiserSimon Peyton Jones2022-11-1011-234/+367
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | The Specialiser has, for some time, fires class-op RULES in the specialiser itself: see Note [Specialisation modulo dictionary selectors] This MR beefs it up a bit, so that it fires /all/ RULES in the specialiser, not just class-op rules. See Note [Fire rules in the specialiser] The result is a bit more specialisation; see test simplCore/should_compile/T21851_2 This pushed me into a bit of refactoring. I made a new data types GHC.Core.Rules.RuleEnv, which combines - the several source of rules (local, home-package, external) - the orphan-module dependencies in a single record for `getRules` to consult. That drove a bunch of follow-on refactoring, including allowing me to remove cr_visible_orphan_mods from the CoreReader data type. I moved some of the RuleBase/RuleEnv stuff into GHC.Core.Rule. The reorganisation in the Simplifier improve compile times a bit (geom mean -0.1%), but T9961 is an outlier Metric Decrease: T9961
* Use TcRnDiagnostic in GHC.Tc.TyCl.Instance (#20117)Giles Anderson2022-11-095-49/+159
| | | | | | | | | | | | The following `TcRnDiagnostic` messages have been introduced: TcRnWarnUnsatisfiedMinimalDefinition TcRnMisplacedInstSig TcRnBadBootFamInstDeclErr TcRnIllegalFamilyInstance TcRnAssocInClassErr TcRnBadFamInstDecl TcRnNotOpenFamily
* Fix TypeData issues (fixes #22315 and #22332)Ross Paterson2022-11-0814-74/+157
| | | | | | | | | | | | | | | | | | | | | | | | There were two bugs here: 1. Treating type-level constructors as PromotedDataCon doesn't always work, in particular because constructors promoted via DataKinds are called both T and 'T. (Tests T22332a, T22332b, T22315a, T22315b) Fix: guard these cases with isDataKindsPromotedDataCon. 2. Type-level constructors were sent to the code generator, producing things like constructor wrappers. (Tests T22332a, T22332b) Fix: test for them in isDataTyCon. Other changes: * changed the marking of "type data" DataCon's as suggested by SPJ. * added a test TDGADT for a type-level GADT. * comment tweaks * change tcIfaceTyCon to ignore IfaceTyConInfo, so that IfaceTyConInfo is used only for pretty printing, not for typechecking. (SPJ)
* Define `Infinite` list and use where appropriate.M Farkas-Dyck2022-11-0816-88/+282
| | | | | | | | Also add perf test for infinite list fusion. In particular, in `GHC.Core`, often we deal with infinite lists of roles. Also in a few locations we deal with infinite lists of names. Thanks to simonpj for helping to write the Note [Fusion for `Infinite` lists].
* Bump Win32 submodule to 2.13.4.0Matthew Pickering2022-11-051-1/+1
| | | | Fixes #22098
* Minor refactor around FastStringsKrzysztof Gogolewski2022-11-0533-72/+87
| | | | | | | Pass FastStrings to functions directly, to make sure the rule for fsLit "literal" fires. Remove SDoc indirection in GHCi.UI.Tags and GHC.Unit.Module.Graph.
* Export pprTrace and friends from GHC.Prelude.Andreas Klebinger2022-11-0366-126/+143
| | | | | Introduces GHC.Prelude.Basic which can be used in modules which are a dependency of the ppr code.
* Clarify status of bindings in WholeCoreBindingsMatthew Pickering2022-11-032-5/+5
| | | | | | | | | Gergo points out that these bindings are tidied, rather than prepd as the variable claims. Therefore we update the name of the variable to reflect reality and add a comment to the data type to try to erase any future confusion. Fixes #22307
* Expose UnitEnvGraphKey for user-codeFendor2022-11-011-0/+1
|
* Add accurate skolem info when quantifyingSimon Peyton Jones2022-11-014-62/+74
| | | | | | | Ticket #22379 revealed that skolemiseQuantifiedTyVar was dropping the passed-in skol_info on the floor when it encountered a SkolemTv. Bad! Several TyCons thereby share a single SkolemInfo on their binders, which lead to bogus error reports.
* ThToHs: fix overzealous parenthesizationVladislav Zavialov2022-11-011-10/+6
| | | | | | | | | | | | | | | | | | Before this patch, when converting from TH.Exp to LHsExpr GhcPs, the compiler inserted more parentheses than required: ((f a) (b + c)) d This was happening because the LHS of the function application was parenthesized as if it was the RHS. Now we use funPrec and appPrec appropriately and produce sensibly parenthesized expressions: f a (b + c) d I also took the opportunity to remove the special case for LamE, which was not special at all and simply duplicated code.
* Typo: rename -fwrite-if-simplfied-core to -fwrite-if-simplified-coreKrzysztof Gogolewski2022-11-013-6/+6
|
* Drop a kludge for binutils<2.17, which is now over 10 years old.M Farkas-Dyck2022-11-012-39/+2
|
* Minor SDoc-related cleanupKrzysztof Gogolewski2022-10-2821-106/+107
| | | | | | | | | | | * Rename pprCLabel to pprCLabelStyle, and use the name pprCLabel for a function using CStyle (analogous to pprAsmLabel) * Move LabelStyle to the CLabel module, it no longer needs to be in Outputable. * Move calls to 'text' right next to literals, to make sure the text/str rule is triggered. * Remove FastString/String roundtrip in Tc.Deriv.Generate * Introduce showSDocForUser', which abstracts over a pattern in GHCi.UI
* Add missing dict binds to specialiserSimon Peyton Jones2022-10-272-14/+13
| | | | | | | | | | I had forgotten to add the auxiliary dict bindings to the /unfolding/ of a specialised function. This caused #22358, which reports failures when compiling Hackage packages fixed-vector indexed-traversable Regression test T22357 is snarfed from indexed-traversable
* Remove source location information from interface filesOwen Shepherd2022-10-278-72/+88
| | | | | | | | | | | | | This change aims to minimize source location information leaking into interface files, which makes ABI hashes dependent on the build location. The `Binary (Located a)` instance has been removed completely. It seems that the HIE interface still needs the ability to serialize SrcSpans, but by wrapping the instances, it should be a lot more difficult to inadvertently add source location information.
* Use TcRnVDQInTermType in noNestedForallsContextsErr (#20115)Vladislav Zavialov2022-10-265-20/+26
| | | | | | | | | | | | | | | | | | | | | When faced with VDQ in the type of a term, GHC generates the following error message: Illegal visible, dependent quantification in the type of a term (GHC does not yet support this) Prior to this patch, there were two ways this message could have been generated and represented: 1. with the dedicated constructor TcRnVDQInTermType (see check_type in GHC.Tc.Validity) 2. with the transitional constructor TcRnUnknownMessage (see noNestedForallsContextsErr in GHC.Rename.Utils) Not only this led to duplication of code generating the final SDoc, it also made it tricky to track the origin of the error message. This patch fixes the problem by using TcRnVDQInTermType exclusively.
* Introduce TcRnWithHsDocContext (#22346)Vladislav Zavialov2022-10-267-28/+45
| | | | | | | | | | | | | | | | | Before this patch, GHC used withHsDocContext to attach an HsDocContext to an error message: addErr $ mkTcRnUnknownMessage $ mkPlainError noHints (withHsDocContext ctxt msg) The problem with this approach is that it only works with TcRnUnknownMessage. But could we attach an HsDocContext to a structured error message in a generic way? This patch solves the problem by introducing a new constructor to TcRnMessage: data TcRnMessage where ... TcRnWithHsDocContext :: !HsDocContext -> !TcRnMessage -> TcRnMessage ...
* Broaden the in-scope sets for liftEnvSubst and composeTCvSubstRyan Scott2022-10-262-3/+13
| | | | | | | | | | | | | | | | | This patch fixes two distinct (but closely related) buglets that were uncovered in #22235: * `liftEnvSubst` used an empty in-scope set, which was not wide enough to cover the variables in the range of the substitution. This patch fixes this by populating the in-scope set from the free variables in the range of the substitution. * `composeTCvSubst` applied the first substitution argument to the range of the second substitution argument, but the first substitution's in-scope set was not wide enough to cover the range of the second substutition. We similarly fix this issue in this patch by widening the first substitution's in-scope set before applying it. Fixes #22235.
* Cleanup String/FastString conversionsKrzysztof Gogolewski2022-10-2516-98/+66
| | | | | | | Remove unused mkPtrString and isUnderscoreFS. We no longer use mkPtrString since 1d03d8bef96. Remove unnecessary conversions between FastString and String and back.
* Add GHC.SysTools.Cpp moduleSylvain Henry2022-10-254-156/+246
| | | | Move doCpp out of the driver to be able to use it in the upcoming JS backend.
* Enable popcount rewrite rule when cross-compilingSylvain Henry2022-10-251-1/+1
| | | | | The comment applies only when host's word size < target's word size. So we can relax the guard.
* Numeric exceptions: replace FFI calls with primopsSylvain Henry2022-10-254-38/+31
| | | | | | | | | | | | | ghc-bignum needs a way to raise numerical exceptions defined in base package. At the time we used FFI calls into primops defined in the RTS. These FFI calls had to be wrapped into hacky bottoming functions because "foreign import prim" syntax doesn't support giving a bottoming demand to the foreign call (cf #16929). These hacky wrapper functions trip up the JavaScript backend (#21078) because they are polymorphic in their return type. This commit replaces them with primops very similar to raise# but raising predefined exceptions.
* Make the specialiser handle polymorphic specialisationSimon Peyton Jones2022-10-254-65/+261
| | | | | | | | | | | | | | | | | | | | | | | | | | | Ticket #13873 unexpectedly showed that a SPECIALISE pragma made a program run (a lot) slower, because less specialisation took place overall. It turned out that the specialiser was missing opportunities because of quantified type variables. It was quite easy to fix. The story is given in Note [Specialising polymorphic dictionaries] Two other minor fixes in the specialiser * There is no benefit in specialising data constructor /wrappers/. (They can appear overloaded because they are given a dictionary to store in the constructor.) Small guard in canSpecImport. * There was a buglet in the UnspecArg case of specHeader, in the case where there is a dead binder. We need a LitRubbish filler for the specUnfolding stuff. I expanded Note [Drop dead args from specialisations] to explain. There is a 4% increase in compile time for T15164, because we generate more specialised code. This seems OK. Metric Increase: T15164
* Fix binder-swap bugSimon Peyton Jones2022-10-255-104/+162
| | | | | | | | | | | | | | | | | | | | | | | This patch fixes #21229 properly, by avoiding doing a binder-swap on dictionary Ids. This is pretty subtle, and explained in Note [Care with binder-swap on dictionaries]. Test is already in simplCore/should_run/T21229 This allows us to restore a feature to the specialiser that we had to revert: see Note [Specialising polymorphic dictionaries]. (This is done in a separate patch.) I also modularised things, using a new function scrutBinderSwap_maybe in all the places where we are (effectively) doing a binder-swap, notably * Simplify.Iteration.addAltUnfoldings * SpecConstr.extendCaseBndrs In Simplify.Iteration.addAltUnfoldings I also eliminated a guard Many <- idMult case_bndr because we concluded, in #22123, that it was doing no good.
* Improve stg lint for unboxed sums.Andreas Klebinger2022-10-241-11/+39
| | | | | | | It now properly lints cases where sums end up distributed over multiple args after unarise. Fixes #22026.
* Convert Diagnostics in GHC.Tc.Gen.Splice (#20116)Aaron Allen2022-10-247-183/+660
| | | | | | | Replaces uses of `TcRnUnknownMessage` in `GHC.Tc.Gen.Splice` with structured diagnostics. closes #20116
* Class layout info (#19623)Vladislav Zavialov2022-10-2317-85/+116
| | | | Updates the haddock submodule.
* Introduce a standard thunk for allocating stringsÖmer Sinan Ağacan2022-10-227-36/+121
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Currently for a top-level closure in the form hey = unpackCString# x we generate code like this: Main.hey_entry() // [R1] { info_tbls: [(c2T4, label: Main.hey_info rep: HeapRep static { Thunk } srt: Nothing)] stack_info: arg_space: 8 updfr_space: Just 8 } {offset c2T4: // global _rqm::P64 = R1; if ((Sp + 8) - 24 < SpLim) (likely: False) goto c2T5; else goto c2T6; c2T5: // global R1 = _rqm::P64; call (stg_gc_enter_1)(R1) args: 8, res: 0, upd: 8; c2T6: // global (_c2T1::I64) = call "ccall" arg hints: [PtrHint, PtrHint] result hints: [PtrHint] newCAF(BaseReg, _rqm::P64); if (_c2T1::I64 == 0) goto c2T3; else goto c2T2; c2T3: // global call (I64[_rqm::P64])() args: 8, res: 0, upd: 8; c2T2: // global I64[Sp - 16] = stg_bh_upd_frame_info; I64[Sp - 8] = _c2T1::I64; R2 = hey1_r2Gg_bytes; Sp = Sp - 16; call GHC.CString.unpackCString#_info(R2) args: 24, res: 0, upd: 24; } } This code is generated for every string literal. Only difference between top-level closures like this is the argument for the bytes of the string (hey1_r2Gg_bytes in the code above). With this patch we introduce a standard thunk in the RTS, called stg_MK_STRING_info, that does what `unpackCString# x` does, except it gets the bytes address from the payload. Using this, for the closure above, we generate this: Main.hey_closure" { Main.hey_closure: const stg_MK_STRING_info; const 0; // padding for indirectee const 0; // static link const 0; // saved info const hey1_r1Gg_bytes; // the payload } This is much smaller in code. Metric Decrease: T10421 T11195 T12150 T12425 T16577 T18282 T18698a T18698b Co-Authored By: Ben Gamari <ben@well-typed.com>
* Delete `eqExpr`, since GHC 9.4 has been released.M Farkas-Dyck2022-10-213-13/+5
|
* Scrub some partiality in `GHC.Core.Opt.Simplify.Utils`.M Farkas-Dyck2022-10-211-8/+9
|
View Lorry Controller Status