- ghc-commits - Haskell.org

[Git][ghc/ghc][wip/T24603] 6 commits: Add name for -Wunusable-unpack-pragmas
by Serge S. Gulin (＠gulin.serge) 21 Apr '25

21 Apr '25

Serge S. Gulin pushed to branch wip/T24603 at Glasgow Haskell Compiler / GHC Commits: e2f2f9d0 by Vladislav Zavialov at 2025-04-20T10:53:39-04:00 Add name for -Wunusable-unpack-pragmas This warning had no name or flag and was triggered unconditionally. Now it is part of -Wdefault. In GHC.Tc.TyCl.tcTyClGroupsPass's strict mode, we now have to force-enable this warning to ensure that detection of flawed groups continues to work even if the user disables the warning with the -Wno-unusable-unpack-pragmas option. Test case: T3990c Also, the misnamed BackpackUnpackAbstractType is now called UnusableUnpackPragma. - - - - - 6caa6508 by Adam Gundry at 2025-04-20T10:54:22-04:00 Fix specialisation of incoherent instances (fixes #25883) GHC normally assumes that class constraints are canonical, meaning that the specialiser is allowed to replace one dictionary argument with another provided that they have the same type. The `-fno-specialise-incoherents` flag alters INCOHERENT instance definitions so that they will prevent specialisation in some cases, by inserting `nospec`. This commit fixes a bug in 7124e4ad76d98f1fc246ada4fd7bf64413ff2f2e, which treated some INCOHERENT instance matches as if `-fno-specialise-incoherents` was in effect, thereby unnecessarily preventing specialisation. In addition it updates the relevant `Note [Rules for instance lookup]` and adds a new `Note [Canonicity for incoherent matches]`. - - - - - 0426fd6c by Adam Gundry at 2025-04-20T10:54:23-04:00 Add regression test for #23429 - - - - - eec96527 by Adam Gundry at 2025-04-20T10:54:23-04:00 user's guide: update specification of overlapping/incoherent instances The description of the instance resolution algorithm in the user's guide was slightly out of date, because it mentioned in-scope given constraints only at the end, whereas the implementation checks for their presence before any of the other steps. This also adds a warning to the user's guide about the impact of incoherent instances on specialisation, and more clearly documents some of the other effects of `-XIncoherentInstances`. - - - - - a00eeaec by Matthew Craven at 2025-04-20T10:55:03-04:00 Fix bytecode generation for `tagToEnum# <LITERAL>` Fixes #25975. - - - - - dfb4c44c by Serge S. Gulin at 2025-04-21T10:24:53+03:00 Support for ARM64 Windows (LLVM-enabled) (fixes #24603) submodule Co-authored-by: Cheng Shao <terrorjack(a)type.dance> Co-authored-by: Dmitrii Egorov <egorov.d.i(a)icloud.com> Co-authored-by: Andrei Borzenkov <root(a)sandwitch.dev> - - - - - 72 changed files: - .gitlab-ci.yml - .gitlab/ci.sh - .gitlab/generate-ci/gen_ci.hs - .gitlab/hello.hs - .gitlab/jobs.yaml - compiler/CodeGen.Platform.h - compiler/GHC/CmmToAsm/AArch64/CodeGen.hs - compiler/GHC/CmmToAsm/AArch64/Instr.hs - compiler/GHC/CmmToAsm/AArch64/Ppr.hs - compiler/GHC/CmmToAsm/Reg/Linear/AArch64.hs - compiler/GHC/Core/InstEnv.hs - compiler/GHC/Driver/Flags.hs - compiler/GHC/Driver/Session.hs - compiler/GHC/Platform/Regs.hs - compiler/GHC/StgToByteCode.hs - compiler/GHC/Tc/Errors/Ppr.hs - compiler/GHC/Tc/Errors/Types.hs - compiler/GHC/Tc/Solver/Dict.hs - compiler/GHC/Tc/TyCl.hs - compiler/GHC/Tc/Utils/Monad.hs - compiler/GHC/Types/Error/Codes.hs - docs/users_guide/9.14.1-notes.rst - docs/users_guide/exts/instances.rst - docs/users_guide/using-warnings.rst - hadrian/src/Oracles/Setting.hs - hadrian/src/Rules/BinaryDist.hs - libraries/Cabal - libraries/Win32 - libraries/base/src/System/CPUTime/Windows.hsc - libraries/base/tests/perf/encodingAllocations.hs - libraries/directory - libraries/ghc-internal/jsbits/errno.js - libraries/haskeline - libraries/process - libraries/unix - llvm-targets - m4/fp_cc_supports_target.m4 - m4/fptools_set_platform_vars.m4 - m4/ghc_tables_next_to_code.m4 - rts/StgCRun.c - rts/linker/PEi386.c - rts/win32/veh_excn.c - + testsuite/tests/bytecode/T25975.hs - + testsuite/tests/bytecode/T25975.stdout - testsuite/tests/bytecode/all.T - testsuite/tests/simplCore/should_compile/Makefile - testsuite/tests/simplCore/should_compile/T23307c.stderr - + testsuite/tests/simplCore/should_compile/T25883.hs - + testsuite/tests/simplCore/should_compile/T25883.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883b.hs - + testsuite/tests/simplCore/should_compile/T25883b.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883c.hs - + testsuite/tests/simplCore/should_compile/T25883c.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883d.hs - + testsuite/tests/simplCore/should_compile/T25883d.stderr - + testsuite/tests/simplCore/should_compile/T25883d_import.hs - + testsuite/tests/simplCore/should_compile/T3990c.hs - + testsuite/tests/simplCore/should_compile/T3990c.stdout - testsuite/tests/simplCore/should_compile/all.T - testsuite/tests/simplCore/should_fail/T25672.stderr - + testsuite/tests/simplCore/should_run/T23429.hs - + testsuite/tests/simplCore/should_run/T23429.stdout - testsuite/tests/simplCore/should_run/all.T - testsuite/tests/typecheck/should_compile/T7050.stderr - testsuite/tests/typecheck/should_fail/T3966.stderr - + testsuite/tests/typecheck/should_fail/T3966b.hs - + testsuite/tests/typecheck/should_fail/T3966b.stderr - testsuite/tests/typecheck/should_fail/all.T - testsuite/tests/unboxedsums/unpack_sums_5.stderr - utils/ghc-toolchain/exe/Main.hs - utils/hsc2hs - utils/llvm-targets/gen-data-layout.sh The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/b6424e486e4f95ba09bed32cd20bca… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/b6424e486e4f95ba09bed32cd20bca… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/spj-apporv-Oct24] ignore ds warnings originating from gen locations
by Apoorv Ingle (＠ani) 20 Apr '25

20 Apr '25

Apoorv Ingle pushed to branch wip/spj-apporv-Oct24 at Glasgow Haskell Compiler / GHC Commits: ec95eca2 by Apoorv Ingle at 2025-04-20T20:58:34-05:00 ignore ds warnings originating from gen locations - - - - - 1 changed file: - compiler/GHC/HsToCore/Pmc.hs Changes: ===================================== compiler/GHC/HsToCore/Pmc.hs ===================================== @@ -51,7 +51,7 @@ import GHC.HsToCore.Pmc.Utils import GHC.HsToCore.Pmc.Desugar import GHC.HsToCore.Pmc.Check import GHC.HsToCore.Pmc.Solver -import GHC.Types.Basic (Origin(..), isDoExpansionGenerated) +import GHC.Types.Basic (Origin(..)) import GHC.Core import GHC.Driver.DynFlags import GHC.Hs @@ -68,7 +68,7 @@ import GHC.HsToCore.Monad import GHC.Data.Bag import GHC.Data.OrdList -import Control.Monad (when, unless, forM_) +import Control.Monad (when, forM_) import qualified Data.Semigroup as Semi import Data.List.NonEmpty ( NonEmpty(..) ) import qualified Data.List.NonEmpty as NE @@ -189,12 +189,11 @@ pmcMatches origin ctxt vars matches = {-# SCC "pmcMatches" #-} do Just matches -> do matches <- {-# SCC "desugarMatches" #-} noCheckDs $ desugarMatches vars matches - tracePm "desugared matches" (ppr matches) + -- tracePm "desugared matches" (ppr matches) result <- {-# SCC "checkMatchGroup" #-} unCA (checkMatchGroup matches) missing tracePm "}: " (ppr (cr_uncov result)) - unless (isDoExpansionGenerated origin) -- Do expansion generated code shouldn't emit overlapping warnings - ({-# SCC "formatReportWarnings" #-} + ({-# SCC "formatReportWarnings" #-} formatReportWarnings ReportMatchGroup ctxt vars result) return (NE.toList (ldiMatchGroup (cr_ret result))) @@ -585,18 +584,18 @@ reportWarnings dflags report_mode (DsMatchContext kind loc) vars approx = precision == Approximate when (approx && (exists_u || exists_i)) $ - putSrcSpanDs loc (diagnosticDs (DsMaxPmCheckModelsReached (maxPmCheckModels dflags))) + maybePutSrcSpanDs loc (diagnosticDs (DsMaxPmCheckModelsReached (maxPmCheckModels dflags))) when exists_b $ forM_ redundant_bangs $ \(SrcInfo (L l q)) -> - putSrcSpanDs l (diagnosticDs (DsRedundantBangPatterns kind q)) + maybePutSrcSpanDs l (diagnosticDs (DsRedundantBangPatterns kind q)) when exists_r $ forM_ redundant_rhss $ \(SrcInfo (L l q)) -> - putSrcSpanDs l (diagnosticDs (DsOverlappingPatterns kind q)) + maybePutSrcSpanDs l (diagnosticDs (DsOverlappingPatterns kind q)) when exists_i $ forM_ inaccessible_rhss $ \(SrcInfo (L l q)) -> - putSrcSpanDs l (diagnosticDs (DsInaccessibleRhs kind q)) + maybePutSrcSpanDs l (diagnosticDs (DsInaccessibleRhs kind q)) when exists_u $ - putSrcSpanDs loc (diagnosticDs (DsNonExhaustivePatterns kind check_type maxPatterns vars unc_examples)) + maybePutSrcSpanDs loc (diagnosticDs (DsNonExhaustivePatterns kind check_type maxPatterns vars unc_examples)) where flag_i = overlapping dflags kind flag_u = exhaustive dflags kind @@ -608,6 +607,15 @@ reportWarnings dflags report_mode (DsMatchContext kind loc) vars maxPatterns = maxUncoveredPatterns dflags + -- we only want to report warnings for source constructs and not compiler generated constructors + -- c.f. Typeable1 + maybePutSrcSpanDs :: SrcSpan -> DsM () -> DsM () + maybePutSrcSpanDs l thing_inside = + if isGeneratedSrcSpan l + then return () + else putSrcSpanDs l thing_inside + + getNFirstUncovered :: GenerateInhabitingPatternsMode -> [Id] -> Int -> Nablas -> DsM [Nabla] getNFirstUncovered mode vars n (MkNablas nablas) = go n (bagToList nablas) where View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/ec95eca2eaee4b7f1240b24270b9c0a… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/ec95eca2eaee4b7f1240b24270b9c0a… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/T23109a] Wibble SetLevels
by Simon Peyton Jones (＠simonpj) 20 Apr '25

20 Apr '25

Simon Peyton Jones pushed to branch wip/T23109a at Glasgow Haskell Compiler / GHC Commits: 6bee78af by Simon Peyton Jones at 2025-04-20T23:02:40+01:00 Wibble SetLevels - - - - - 1 changed file: - compiler/GHC/Core/Opt/SetLevels.hs Changes: ===================================== compiler/GHC/Core/Opt/SetLevels.hs ===================================== @@ -708,9 +708,15 @@ lvlMFE env strict_ctxt ann_expr -- See Note [Floating to the top] saves_alloc = isTopLvl dest_lvl + && (escapes_value_lam || floatConsts env) + -- Always float allocation out of a value lambda + -- if it gets to top level + && (not strict_ctxt || is_hnf || is_bot_lam) +{- && ( (floatConsts env && (not strict_ctxt || is_hnf)) -- (FT1) and (FT2) || (is_bot_lam && escapes_value_lam)) -- (FT3) +-} hasFreeJoin :: LevelEnv -> DVarSet -> Bool -- Has a free join point which is not being floated to top level. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/6bee78af5db225c26bb21254bc9cc9f… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/6bee78af5db225c26bb21254bc9cc9f… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/T25965] 11 commits: Refactor Handling of Multiple Default Declarations
by Simon Peyton Jones (＠simonpj) 20 Apr '25

20 Apr '25

Simon Peyton Jones pushed to branch wip/T25965 at Glasgow Haskell Compiler / GHC Commits: e0f3ff11 by Patrick at 2025-04-17T04:31:12-04:00 Refactor Handling of Multiple Default Declarations Fixes: #25912, #25914, #25934 Previously, GHC discarded all loaded defaults (tcg_default) when local defaults were encountered during typechecking. According to the exportable-named-default proposal (sections 2.4.2 and 2.4.3), local defaults should be merged into tcg_default, retaining any defaults already present while overriding where necessary. Key Changes: * Introduce DefaultProvenance to track the origin of default declarations (local, imported, or built-in), replacing the original cd_module in ClassDefaults with cd_provenance :: DefaultProvenance. * Rename tcDefaults to tcDefaultDecls, limiting its responsibility to only converting renamed class defaults into ClassDefaults. * Add extendDefaultEnvWithLocalDefaults to merge local defaults into the environment, with proper duplication checks: - Duplicate local defaults for a class trigger an error. - Local defaults override imported or built-in defaults. * Update and add related notes: Note [Builtin class defaults], Note [DefaultProvenance]. * Add regression tests: T25912, T25914, T25934. Thanks sam and simon for the help on this patch. Co-authored-by: sheaf <sam.derbyshire(a)gmail.com> - - - - - 386f1854 by Teo Camarasu at 2025-04-17T04:31:55-04:00 template-haskell: Remove `addrToByteArrayName` and `addrToByteArray` These were part of the implementation of the `Lift ByteArray` instance and were errornously exported because this module lacked an explicit export list. They have no usages on Hackage. Resolves #24782 - - - - - b96e2f77 by Sylvain Henry at 2025-04-18T20:46:33-04:00 RTS: remove target info and fix host info (#24058) The RTS isn't a compiler, hence it doesn't have a target and we remove the reported target info displayed by "+RTS --info". We also fix the host info displayed by "+RTS --info": the host of the RTS is the RTS-building compiler's target, not the compiler's host (wrong when doing cross-compilation). - - - - - 6d9965f4 by Sylvain Henry at 2025-04-18T20:46:33-04:00 RTS: remove build info As per the discussion in !13967, there is no reason to tag the RTS with information about the build platform. - - - - - d52e9b3f by Vladislav Zavialov at 2025-04-18T20:47:15-04:00 Diagnostics: remove the KindMismatch constructor (#25957) The KindMismatch constructor was only used as an intermediate representation in pretty-printing. Its removal addresses a problem detected by the "codes" test case: [GHC-89223] is untested (constructor = KindMismatch) In a concious deviation from the usual procedure, the error code GHC-89223 is removed entirely rather than marked as Outdated. The reason is that it never was user-facing in the first place. - - - - - e2f2f9d0 by Vladislav Zavialov at 2025-04-20T10:53:39-04:00 Add name for -Wunusable-unpack-pragmas This warning had no name or flag and was triggered unconditionally. Now it is part of -Wdefault. In GHC.Tc.TyCl.tcTyClGroupsPass's strict mode, we now have to force-enable this warning to ensure that detection of flawed groups continues to work even if the user disables the warning with the -Wno-unusable-unpack-pragmas option. Test case: T3990c Also, the misnamed BackpackUnpackAbstractType is now called UnusableUnpackPragma. - - - - - 6caa6508 by Adam Gundry at 2025-04-20T10:54:22-04:00 Fix specialisation of incoherent instances (fixes #25883) GHC normally assumes that class constraints are canonical, meaning that the specialiser is allowed to replace one dictionary argument with another provided that they have the same type. The `-fno-specialise-incoherents` flag alters INCOHERENT instance definitions so that they will prevent specialisation in some cases, by inserting `nospec`. This commit fixes a bug in 7124e4ad76d98f1fc246ada4fd7bf64413ff2f2e, which treated some INCOHERENT instance matches as if `-fno-specialise-incoherents` was in effect, thereby unnecessarily preventing specialisation. In addition it updates the relevant `Note [Rules for instance lookup]` and adds a new `Note [Canonicity for incoherent matches]`. - - - - - 0426fd6c by Adam Gundry at 2025-04-20T10:54:23-04:00 Add regression test for #23429 - - - - - eec96527 by Adam Gundry at 2025-04-20T10:54:23-04:00 user's guide: update specification of overlapping/incoherent instances The description of the instance resolution algorithm in the user's guide was slightly out of date, because it mentioned in-scope given constraints only at the end, whereas the implementation checks for their presence before any of the other steps. This also adds a warning to the user's guide about the impact of incoherent instances on specialisation, and more clearly documents some of the other effects of `-XIncoherentInstances`. - - - - - a00eeaec by Matthew Craven at 2025-04-20T10:55:03-04:00 Fix bytecode generation for `tagToEnum# <LITERAL>` Fixes #25975. - - - - - 5c976307 by Simon Peyton Jones at 2025-04-20T22:39:35+01:00 Fix infelicities in the Specialiser On the way to #23109 (unary classes) I discovered some infelicities (or maybe tiny bugs, I forget) in the type-class specialiser. I also tripped over #25965, an outright bug in the rule matcher Specifically: * Refactor: I enhanced `wantCallsFor`, whih previously always said `True`, to discard calls of class-ops, data constructors etc. This is a bit more efficient; and it means we don't need to worry about filtering them out later. * Fix: I tidied up some tricky logic that eliminated redundant specialisations. It wasn't working correctly. See the expanded Note [Specialisations already covered], and (MP3) in Note [Specialising polymorphic dictionaries]. See also the new top-level `alreadyCovered` function, which now goes via `GHC.Core.Rules.ruleLhsIsMoreSpecific` I also added a useful Note [The (CI-KEY) invariant] * Fix #25965: fixed a tricky bug in the `go_fam_fam` in `GHC.Core.Unify.uVarOrFam`, which allows matching to succeed without binding all type varibles. I enhanced Note [Apartness and type families] some more * #25703. This ticket "just works" with -fpolymorphic-specialisation; but I was surprised that it worked! In this MR I added documentation to Note [Interesting dictionary arguments] to explain; and tests to ensure it stays fixed. - - - - - 73 changed files: - compiler/GHC/Core/InstEnv.hs - compiler/GHC/Core/Opt/Specialise.hs - compiler/GHC/Core/Rules.hs - compiler/GHC/Core/Unify.hs - compiler/GHC/Driver/Flags.hs - compiler/GHC/Driver/Session.hs - compiler/GHC/IfaceToCore.hs - compiler/GHC/StgToByteCode.hs - compiler/GHC/Tc/Errors/Ppr.hs - compiler/GHC/Tc/Errors/Types.hs - compiler/GHC/Tc/Gen/Default.hs - compiler/GHC/Tc/Gen/Export.hs - compiler/GHC/Tc/Module.hs - compiler/GHC/Tc/Solver/Dict.hs - compiler/GHC/Tc/Solver/Equality.hs - compiler/GHC/Tc/TyCl.hs - compiler/GHC/Tc/Utils/Env.hs - compiler/GHC/Tc/Utils/Monad.hs - compiler/GHC/Types/Basic.hs - compiler/GHC/Types/DefaultEnv.hs - compiler/GHC/Types/Error/Codes.hs - configure.ac - docs/users_guide/9.14.1-notes.rst - docs/users_guide/exts/instances.rst - docs/users_guide/using-warnings.rst - hadrian/src/Settings/Packages.hs - libraries/template-haskell/Language/Haskell/TH/Syntax.hs - libraries/template-haskell/changelog.md - rts/RtsUtils.c - testsuite/ghc-config/ghc-config.hs - + testsuite/tests/bytecode/T25975.hs - + testsuite/tests/bytecode/T25975.stdout - testsuite/tests/bytecode/all.T - + testsuite/tests/default/T25912.hs - + testsuite/tests/default/T25912.stdout - + testsuite/tests/default/T25912_helper.hs - + testsuite/tests/default/T25914.hs - + testsuite/tests/default/T25934.hs - testsuite/tests/default/all.T - testsuite/tests/default/default-fail03.stderr - testsuite/tests/diagnostic-codes/codes.stdout - testsuite/tests/interface-stability/template-haskell-exports.stdout - testsuite/tests/linters/notes.stdout - testsuite/tests/module/mod58.stderr - testsuite/tests/simplCore/should_compile/Makefile - testsuite/tests/simplCore/should_compile/T23307c.stderr - + testsuite/tests/simplCore/should_compile/T25703.hs - + testsuite/tests/simplCore/should_compile/T25703.stderr - + testsuite/tests/simplCore/should_compile/T25703a.hs - + testsuite/tests/simplCore/should_compile/T25703a.stderr - + testsuite/tests/simplCore/should_compile/T25883.hs - + testsuite/tests/simplCore/should_compile/T25883.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883b.hs - + testsuite/tests/simplCore/should_compile/T25883b.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883c.hs - + testsuite/tests/simplCore/should_compile/T25883c.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883d.hs - + testsuite/tests/simplCore/should_compile/T25883d.stderr - + testsuite/tests/simplCore/should_compile/T25883d_import.hs - + testsuite/tests/simplCore/should_compile/T25965.hs - + testsuite/tests/simplCore/should_compile/T3990c.hs - + testsuite/tests/simplCore/should_compile/T3990c.stdout - testsuite/tests/simplCore/should_compile/all.T - testsuite/tests/simplCore/should_fail/T25672.stderr - + testsuite/tests/simplCore/should_run/T23429.hs - + testsuite/tests/simplCore/should_run/T23429.stdout - testsuite/tests/simplCore/should_run/all.T - testsuite/tests/typecheck/should_compile/T7050.stderr - testsuite/tests/typecheck/should_fail/T3966.stderr - + testsuite/tests/typecheck/should_fail/T3966b.hs - + testsuite/tests/typecheck/should_fail/T3966b.stderr - testsuite/tests/typecheck/should_fail/all.T - testsuite/tests/unboxedsums/unpack_sums_5.stderr The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/3909c6899e6fb7fd95f60c73cc25f4… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/3909c6899e6fb7fd95f60c73cc25f4… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][master] Fix bytecode generation for `tagToEnum# <LITERAL>`
by Marge Bot (＠marge-bot) 20 Apr '25

20 Apr '25

Marge Bot pushed to branch master at Glasgow Haskell Compiler / GHC Commits: a00eeaec by Matthew Craven at 2025-04-20T10:55:03-04:00 Fix bytecode generation for `tagToEnum# <LITERAL>` Fixes #25975. - - - - - 4 changed files: - compiler/GHC/StgToByteCode.hs - + testsuite/tests/bytecode/T25975.hs - + testsuite/tests/bytecode/T25975.stdout - testsuite/tests/bytecode/all.T Changes: ===================================== compiler/GHC/StgToByteCode.hs ===================================== @@ -1801,10 +1801,14 @@ maybe_getCCallReturnRep fn_ty _ -> pprPanic "maybe_getCCallReturn: can't handle:" (pprType fn_ty) -maybe_is_tagToEnum_call :: CgStgExpr -> Maybe (Id, [Name]) +maybe_is_tagToEnum_call :: CgStgExpr -> Maybe (StgArg, [Name]) -- Detect and extract relevant info for the tagToEnum kludge. -maybe_is_tagToEnum_call (StgOpApp (StgPrimOp TagToEnumOp) [StgVarArg v] t) +maybe_is_tagToEnum_call (StgOpApp (StgPrimOp TagToEnumOp) args t) + | [v] <- args = Just (v, extract_constr_Names t) + | otherwise + = pprPanic "StgToByteCode: tagToEnum#" + $ text "Expected exactly one arg, but actual args are:" <+> ppr args where extract_constr_Names ty | rep_ty <- unwrapType ty @@ -1851,13 +1855,13 @@ implement_tagToId :: StackDepth -> Sequel -> BCEnv - -> Id + -> StgArg -> [Name] -> BcM BCInstrList -- See Note [Implementing tagToEnum#] implement_tagToId d s p arg names = assert (notNull names) $ - do (push_arg, arg_bytes) <- pushAtom d p (StgVarArg arg) + do (push_arg, arg_bytes) <- pushAtom d p arg labels <- getLabelsBc (strictGenericLength names) label_fail <- getLabelBc label_exit <- getLabelBc ===================================== testsuite/tests/bytecode/T25975.hs ===================================== @@ -0,0 +1,27 @@ +-- Tests bytecode generation for tagToEnum# applied to literals +{-# LANGUAGE MagicHash #-} +module Main (main) where + +import GHC.Exts + +f1 :: Int# -> Bool +{-# OPAQUE f1 #-} +f1 v = case v of + 4# -> tagToEnum# v + _ -> False + +f2 :: Int# -> Bool +{-# OPAQUE f2 #-} +f2 v = case v of + 5# -> tagToEnum# 6# + _ -> True + +f3 :: Ordering +f3 = tagToEnum# (noinline runRW# (\_ -> 1#)) + + +main :: IO () +main = do + print $ f1 2# + print $ f2 3# + print f3 ===================================== testsuite/tests/bytecode/T25975.stdout ===================================== @@ -0,0 +1,3 @@ +False +True +EQ ===================================== testsuite/tests/bytecode/all.T ===================================== @@ -1,3 +1,7 @@ ghci_dump_bcos = [only_ways(['ghci']), extra_run_opts('-dno-typeable-binds -dsuppress-uniques -ddump-bcos')] test('T23068', ghci_dump_bcos + [filter_stdout_lines(r'.*bitmap: .*')], ghci_script, ['T23068.script']) + +test('T25975', extra_ways(ghci_ways), compile_and_run, + # Some of the examples work more robustly with these flags + ['-fno-break-points -fno-full-laziness']) View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/a00eeaec8f0b98ec2b8c4630f359fde… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/a00eeaec8f0b98ec2b8c4630f359fde… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][master] 3 commits: Fix specialisation of incoherent instances (fixes #25883)
by Marge Bot (＠marge-bot) 20 Apr '25

20 Apr '25

Marge Bot pushed to branch master at Glasgow Haskell Compiler / GHC Commits: 6caa6508 by Adam Gundry at 2025-04-20T10:54:22-04:00 Fix specialisation of incoherent instances (fixes #25883) GHC normally assumes that class constraints are canonical, meaning that the specialiser is allowed to replace one dictionary argument with another provided that they have the same type. The `-fno-specialise-incoherents` flag alters INCOHERENT instance definitions so that they will prevent specialisation in some cases, by inserting `nospec`. This commit fixes a bug in 7124e4ad76d98f1fc246ada4fd7bf64413ff2f2e, which treated some INCOHERENT instance matches as if `-fno-specialise-incoherents` was in effect, thereby unnecessarily preventing specialisation. In addition it updates the relevant `Note [Rules for instance lookup]` and adds a new `Note [Canonicity for incoherent matches]`. - - - - - 0426fd6c by Adam Gundry at 2025-04-20T10:54:23-04:00 Add regression test for #23429 - - - - - eec96527 by Adam Gundry at 2025-04-20T10:54:23-04:00 user's guide: update specification of overlapping/incoherent instances The description of the instance resolution algorithm in the user's guide was slightly out of date, because it mentioned in-scope given constraints only at the end, whereas the implementation checks for their presence before any of the other steps. This also adds a warning to the user's guide about the impact of incoherent instances on specialisation, and more clearly documents some of the other effects of `-XIncoherentInstances`. - - - - - 16 changed files: - compiler/GHC/Core/InstEnv.hs - compiler/GHC/Tc/Solver/Dict.hs - docs/users_guide/exts/instances.rst - + testsuite/tests/simplCore/should_compile/T25883.hs - + testsuite/tests/simplCore/should_compile/T25883.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883b.hs - + testsuite/tests/simplCore/should_compile/T25883b.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883c.hs - + testsuite/tests/simplCore/should_compile/T25883c.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883d.hs - + testsuite/tests/simplCore/should_compile/T25883d.stderr - + testsuite/tests/simplCore/should_compile/T25883d_import.hs - testsuite/tests/simplCore/should_compile/all.T - + testsuite/tests/simplCore/should_run/T23429.hs - + testsuite/tests/simplCore/should_run/T23429.stdout - testsuite/tests/simplCore/should_run/all.T Changes: ===================================== compiler/GHC/Core/InstEnv.hs ===================================== @@ -599,7 +599,7 @@ These functions implement the carefully-written rules in the user manual section on "overlapping instances". At risk of duplication, here are the rules. If the rules change, change this text and the user manual simultaneously. The link may be this: -http://www.haskell.org/ghc/docs/latest/html/users_guide/glasgow_exts.html#instance-overlap +https://downloads.haskell.org/ghc/latest/docs/users_guide/exts/instances.html#instance-overlap The willingness to be overlapped or incoherent is a property of the instance declaration itself, controlled by its `OverlapMode`, as follows @@ -627,14 +627,22 @@ of the target constraint (C ty1 .. tyn). The search works like this. (IL0) If there are any local Givens that match (potentially unifying any metavariables, even untouchable ones) the target constraint, - the search fails. See Note [Instance and Given overlap] in - GHC.Tc.Solver.Dict. - -(IL1) Find all instances `I` that *match* the target constraint; that is, the target - constraint is a substitution instance of `I`. These instance declarations are - the /candidates/. - -(IL2) If there are no candidates, the search fails. + the search fails unless -XIncoherentInstances is enabled. See + Note [Instance and Given overlap] in GHC.Tc.Solver.Dict. This is + implemented by the first guard in matchClassInst. + +(IL1) Find `all_matches` and `all_unifs` in `lookupInstEnv`: + - all_matches: all instances `I` that *match* the target constraint (that + is, the target constraint is a substitution instance of `I`). These + instance declarations are the /candidates/. + - all_unifs: all non-incoherent instances that *unify with but do not match* + the target constraint. These are not candidates, but might match later if + the target constraint is furhter instantiated. See + `data PotentialUnifiers` for more precise details. + +(IL2) If there are no candidates, the search fails + (lookupInstEnv returns no final_matches). The PotentialUnifiers are returned + by lookupInstEnv for use in error message generation (mkDictErr). (IL3) Eliminate any candidate `IX` for which there is another candidate `IY` such that both of the following hold: @@ -644,29 +652,39 @@ of the target constraint (C ty1 .. tyn). The search works like this. "either/or" design, rather than a "both/and" design, allow a client to deliberately override an instance from a library, without requiring a change to the library.) - This is done by `pruneOverlappingMatches` -(IL4) If all the remaining candidates are *incoherent*, the search succeeds, - returning an arbitrary surviving candidate. + In addition, provided there is at least one candidate, eliminate any other + candidates that are *incoherent*. (In particular, if all remaining candidates + are incoherent, all except an arbitrarily chosen one will be eliminated.) + + This is implemented by `pruneOverlappedMatches`, producing final_matches in + lookupInstEnv. See Note [Instance overlap and guards] and + Note [Incoherent instances]. + +(IL4) If exactly one *incoherent* candidate remains, the search succeeds. + (By the previous step, there cannot be more than one incoherent candidate + remaining.) - If any coherent or non-canonical incoherent unifiers were discarded, - return NoUnifiers EvNonCanonical; if only canonical incoherent unifiers - were discarded, return NoUnifiers EvCanonical + In this case, lookupInstEnv returns the successful match, and it returns + NoUnifiers as the final_unifs, which amounts to skipping the following + steps. -(IL5) If more than one non-*incoherent* candidate remains, the search - fails. Otherwise there is exactly one non-*incoherent* - candidate; call it the "prime candidate". +(IL5) If more than one candidate remains, the search fails. (We have already + eliminated the incoherent candidates, and we have no way to select + between non-incoherent candidates.) -(IL6) Now find all instances that unify with the target constraint, - but do not match it. Such non-candidate instances might match - when the target constraint is further instantiated. +(IL6) Otherwise there is exactly one candidate remaining. The all_unifs + computed at step (IL1) are returned from lookupInstEnv as final_unifs. - If any are *coherent* (not incoherent) return them - as PotentialUnifiers. + If there are no potential unifiers, the search succeeds (in matchInstEnv). + If there is at least one (non-incoherent) potential unifier, matchInstEnv + returns a NotSure result and refrains from committing to the instance. + + Incoherent instances are not returned as part of the potential unifiers. This + affects error messages: they will not be listed as "potentially matching instances" + in an "Overlapping instances" or "Ambiguous type variable" error. + See also Note [Recording coherence information in `PotentialUnifiers`]. - If all are *incoherent* (OverlapFlag = Incoherent or NonCanonical) - return (NoUnifiers nc), where nc is EvNonCanonical if any of the discarded - unifiers are NonCanonical. Notice that these rules are not influenced by flag settings in the client module, where the instances are *used*. These rules make it @@ -894,8 +912,8 @@ instances, i.e. with `-fno-specialise-incoherents`. To avoid this incoherence breaking the specialiser, -* We label as "non-canonical" the dictionary constructed by a (potentially) - incoherent use of an ClsInst whose `OverlapFlag` is `NonCanonical`. +* We label as "non-canonical" any dictionary constructed by a (potentially) + incoherent use of an ClsInst. * We do not specialise a function if there is a non-canonical dictionary in the /transistive dependencies/ of its dictionary @@ -922,7 +940,9 @@ So `d2` is incoherent, and hence (transitively) so is `d1`. Here are the moving parts: * GHC.Core.InstEnv.lookupInstEnv tells if any incoherent unifiers were discarded - in step (IL6) of the instance lookup. + in step (IL4) or (IL6) of the instance lookup: see + Note [Recording coherence information in `PotentialUnifiers`] and + Note [Canonicity for incoherent matches]. * That info is recorded in the `cir_is_coherent` field of `OneInst`, and thence transferred to the `ep_is_coherent` field of the `EvBind` for the dictionary. @@ -930,6 +950,47 @@ Here are the moving parts: * In the desugarer we exploit this info: see Note [Desugaring non-canonical evidence] in GHC.HsToCore.Expr. See also Note [nospecId magic] in GHC.Types.Id.Make. + + +Note [Canonicity for incoherent matches] +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +When the selected instance is INCOHERENT at step (IL4) of +Note [Rules for instance lookup], we ignore all unifiers, +whether or not they are marked with INCOHERENT pragmas. +This is implemented by returning NoUnifiers in final_unifs. +NoUnifiers takes an argument indicating whether the match was canonical +as described in Note [Coherence and specialisation: overview] and +Note [Recording coherence information in `PotentialUnifiers`]. + +To determine whether an incoherent match was canonical, we look *only* +at the OverlapFlag of the instance being matched. For example: + + class C a + instance {-# INCOHERENT #-} C a -- (1) + instance C Int -- (2) + + [W] C tau + +Here we match instance (1) and discard instance (2). If (1) is Incoherent +(under -fspecialise-incoherents), it is important that we treat the match +as EvCanonical so that we do not block specialisation (see #25883). + +What about the following situation: + + instance {-# INCOHERENT #-} C a -- (1), in a module with -fspecialise-incoherents (Incoherent) + instance {-# INCOHERENT #-} C Int -- (2), in a module with -fno-specialise-incoherents (NonCanonical) + + [W] C tau + +Again we match instance (1) and discard instance (2). It is not obvious +whether Incoherent or NonCanonical should "win" here, but it seems more +consistent with the previous example to look only at the flag on instance (1). + +What about if the only instance that can match is marked as NonCanonical? +In this case are no unifiers at all, so all_unifs = NoUnifiers EvCanonical. +It is not obvious what -fno-specialise-incoherents should do here, but +currently it returns NoUnifiers EvCanonical. + -} type DFunInstType = Maybe Type @@ -1070,8 +1131,8 @@ data PotentialUnifiers ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ When we find a matching instance, there might be other instances that could potentially unify with the goal. For `INCOHERENT` instances, we -don't care (see steps IL4 and IL6 in Note [Rules for instance -lookup]). But if we have potentially unifying coherent instance, we +don't care (see step (IL6) in Note [Rules for instance lookup]). +But if we have potentially unifying coherent instance, we report these `OneOrMoreUnifiers` so that `matchInstEnv` can go down the `NotSure` route. @@ -1081,6 +1142,21 @@ solution is canonical or not (see Note [Coherence and specialisation: overview] for why we care at all). So when the set of potential unifiers is empty, we record in `NoUnifiers` if the one solution is `Canonical`. + +For example, suppose we have: + + class C x y + instance C a Bool -- (1) + instance {-# INCOHERENT #-} C Int a -- (2) + + [W] C x Bool + +Here instance (1) matches the Wanted, and since instance (2) is INCOHERENT +we want to succeed with the match rather than getting stick at step (IL6). +But if -fno-specialise-incoherents was enabled for (2), the specialiser is +not permitted to specialise this dictionary later, so lookupInstEnv reports +the PotentialUnifiers as NoUnifiers EvNonCanonical. + -} instance Outputable CanonicalEvidence where @@ -1101,10 +1177,18 @@ getCoherentUnifiers :: PotentialUnifiers -> [ClsInst] getCoherentUnifiers NoUnifiers{} = [] getCoherentUnifiers (OneOrMoreUnifiers cls) = NE.toList cls +-- | Are there no *coherent* unifiers? nullUnifiers :: PotentialUnifiers -> Bool nullUnifiers NoUnifiers{} = True nullUnifiers _ = False +-- | Are there any unifiers, ignoring those marked Incoherent (but including any +-- marked NonCanonical)? +someUnifiers :: PotentialUnifiers -> Bool +someUnifiers (NoUnifiers EvCanonical) = False +someUnifiers _ = True + + instEnvMatchesAndUnifiers :: InstEnv -- InstEnv to look in -> VisibleOrphanModules -- But filter against this @@ -1209,10 +1293,13 @@ lookupInstEnv check_overlap_safe tys = (final_matches, final_unifs, unsafe_overlapped) where + -- (IL1): Find all instances that match the target constraint (home_matches, home_unifs) = instEnvMatchesAndUnifiers home_ie vis_mods cls tys (pkg_matches, pkg_unifs) = instEnvMatchesAndUnifiers pkg_ie vis_mods cls tys all_matches = home_matches <> pkg_matches all_unifs = home_unifs <> pkg_unifs + + -- (IL3): Eliminate candidates that are overlapped or incoherent final_matches = pruneOverlappedMatches all_matches -- Even if the unifs is non-empty (an error situation) -- we still prune the matches, so that the error message isn't @@ -1230,10 +1317,13 @@ lookupInstEnv check_overlap_safe (m:ms) | isIncoherent (fst m) -- Incoherent match, so discard all unifiers, but -- keep track of dropping coherent or non-canonical ones + -- if the match is non-canonical. + -- See Note [Canonicity for incoherent matches] -> assertPpr (null ms) (ppr final_matches) $ - case all_unifs of - OneOrMoreUnifiers{} -> NoUnifiers EvNonCanonical - NoUnifiers{} -> all_unifs + NoUnifiers $ + if isNonCanonical (fst m) && someUnifiers all_unifs + then EvNonCanonical + else EvCanonical _ -> all_unifs -- Note [Safe Haskell isSafeOverlap] ===================================== compiler/GHC/Tc/Solver/Dict.hs ===================================== @@ -926,7 +926,8 @@ matchClassInst dflags inerts clas tys loc -- First check whether there is an in-scope Given that could -- match this constraint. In that case, do not use any instance -- whether top level, or local quantified constraints. --- See Note [Instance and Given overlap] +-- See Note [Instance and Given overlap] and see +-- (IL0) in Note [Rules for instance lookup] in GHC.Core.InstEnv | not (xopt LangExt.IncoherentInstances dflags) , not (isCTupleClass clas) -- It is always safe to unpack constraint tuples ===================================== docs/users_guide/exts/instances.rst ===================================== @@ -417,7 +417,7 @@ Overlapping instances :status: Deprecated Deprecated extension to weaken checks intended to ensure instance resolution - termination. + termination. Use ``OVERLAPPING``, ``OVERLAPPABLE`` or ``OVERLAPS`` pragmas instead. .. extension:: IncoherentInstances :shortdesc: Allow definitions of instances that may result in incoherence. @@ -429,7 +429,9 @@ Overlapping instances :status: Deprecated Deprecated extension to weaken checks intended to ensure instance resolution - termination. + termination. Use ``INCOHERENT`` pragmas instead. Also permits classes to + have non-nominal roles, and affects the instance resolution algorithm for + in-scope given constraints. In general, as discussed in :ref:`instance-resolution`, *GHC requires that it be unambiguous which instance declaration should be used to @@ -473,11 +475,15 @@ Now suppose that, in some client module, we are searching for an instance of the *target constraint* ``(C ty1 .. tyn)``. The search works like this: +- If there are any in-scope given constraints that might match the target + constraint (after unifying any metavariables), and + :extension:`IncoherentInstances` is not enabled, the search fails. + - Find all instances :math:`I` that *match* the target constraint; that is, the target constraint is a substitution instance of :math:`I`. These instance declarations are the *candidates*. -- If no candidates remain, the search fails +- If there are no candidates, the search fails. - Eliminate any candidate :math:`IX` for which there is another candidate :math:`IY` such that both of the following hold: @@ -498,7 +504,7 @@ like this: - Otherwise there is exactly one non-incoherent candidate; call it the "prime candidate". -- Now find all instances, or in-scope given constraints, that *unify* with +- Now find all instances that *unify* with the target constraint, but do not *match* it. Such non-candidate instances might match when the target constraint is further instantiated. If all of them are @@ -596,8 +602,8 @@ declaration, thus: :: (You need :extension:`FlexibleContexts` to do this.) -In the unification check in the final bullet, GHC also uses the -"in-scope given constraints". Consider for example :: +As an example of the "in-scope given constraints" in the first bullet, +consider :: instance C a Int @@ -609,7 +615,7 @@ top-level instance, because a particular call of ``g`` might instantiate both ``b`` and ``c`` to the same type, which would allow the constraint to be solved in a different way. This latter restriction is principally to make the constraint-solver complete. -(Interested folk can read ``Note [Instance and Given overlap]`` in ``TcInteract``.) +(Interested folk can read ``Note [Instance and Given overlap]`` in ``GHC.Tc.Solver.Dict``.) It is easy to avoid: in a type signature avoid a constraint that matches a top-level instance. The flag :ghc-flag:`-Wsimplifiable-class-constraints` warns about such signatures. @@ -660,6 +666,22 @@ matches a top-level instance. The flag :ghc-flag:`-Wsimplifiable-class-constrai to reject module ``Help`` on the grounds that a later instance declaration might overlap the local one.) +.. warning:: + GHC's optimiser (in particular, the :ghc-flag:`-fspecialise` option) + assumes that type-classes are coherent, and hence it may replace + any type-class dictionary argument with another dictionary of the same + type. + + This may cause unexpected results if incoherence occurs due to incoherent + or overlapping instances, and there is an observable difference between the + instances (see :ghc-ticket:`22448` and :ghc-ticket:`24924` for examples). + + The :ghc-flag:`-fno-specialise-incoherents <-fspecialise-incoherents>` will + inhibit specialisation in the presence of some incoherent instance matches, + which may help avoid this issue at the cost of runtime performance. + Alternatively, specialisation can be disabled entirely with + :ghc-flag:`-fno-specialise <-fspecialise>`. + .. _instance-sigs: Instance signatures: type signatures in instance declarations ===================================== testsuite/tests/simplCore/should_compile/T25883.hs ===================================== @@ -0,0 +1,20 @@ +-- By default -fspecialise-incoherents is in effect, so the call to m in f +-- should get specialised even though there is another potential instance. + +{-# LANGUAGE UndecidableInstances #-} +module T25833 (y) where + +class C a where + m :: a -> a + +instance {-# INCOHERENT #-} Num a => C a where + m = (* 3) + +instance C () where + m = id + +f :: Num a => a -> a +f = m + +y :: Int +y = f 2 ===================================== testsuite/tests/simplCore/should_compile/T25883.substr-simpl ===================================== @@ -0,0 +1 @@ +y = I# 6# \ No newline at end of file ===================================== testsuite/tests/simplCore/should_compile/T25883b.hs ===================================== @@ -0,0 +1,21 @@ +-- Under -fno-specialise-incoherents, the call to m in f should not be +-- specialised, because there is another possible (though unused) instance. + +{-# OPTIONS_GHC -fno-specialise-incoherents #-} +{-# LANGUAGE UndecidableInstances #-} +module T25833b (y) where + +class C a where + m :: a -> a + +instance {-# INCOHERENT #-} Num a => C a where + m = (* 3) + +instance C () where + m = id + +f :: Num a => a -> a +f = m + +y :: Int +y = f 2 ===================================== testsuite/tests/simplCore/should_compile/T25883b.substr-simpl ===================================== @@ -0,0 +1 @@ +y = nospec m \ No newline at end of file ===================================== testsuite/tests/simplCore/should_compile/T25883c.hs ===================================== @@ -0,0 +1,18 @@ +-- Here -fno-specialise-incoherents is in effect, but f refers unambiguously to +-- a single instance (because there are no others), so it should be specialised. + +{-# OPTIONS_GHC -fno-specialise-incoherents #-} +{-# LANGUAGE UndecidableInstances #-} +module T25833c (y) where + +class C a where + m :: a -> a + +instance {-# INCOHERENT #-} Num a => C a where + m = (* 3) + +f :: Num a => a -> a +f = m + +y :: Int +y = f 2 ===================================== testsuite/tests/simplCore/should_compile/T25883c.substr-simpl ===================================== @@ -0,0 +1 @@ +y = I# 6# \ No newline at end of file ===================================== testsuite/tests/simplCore/should_compile/T25883d.hs ===================================== @@ -0,0 +1,14 @@ +{-# OPTIONS_GHC -fno-specialise-incoherents #-} + +module T25883d (y) where + +import T25883d_import + +instance {-# INCOHERENT #-} C () where + m = id + +f :: Num a => a -> a +f = m + +y :: Int +y = f 2 \ No newline at end of file ===================================== testsuite/tests/simplCore/should_compile/T25883d.stderr ===================================== @@ -0,0 +1 @@ +y = I# 6# ===================================== testsuite/tests/simplCore/should_compile/T25883d_import.hs ===================================== @@ -0,0 +1,11 @@ +-- This module defines an instance with -fspecialise-incoherents in effect, +-- then it will be imported by a module that uses -fno-specialise-incoherents. + +{-# LANGUAGE UndecidableInstances #-} +module T25883d_import where + +class C a where + m :: a -> a + +instance {-# INCOHERENT #-} Num a => C a where + m = (* 3) ===================================== testsuite/tests/simplCore/should_compile/all.T ===================================== @@ -536,3 +536,8 @@ test('T25197', [req_th, extra_files(["T25197_TH.hs"]), only_ways(['optasm'])], m test('T25389', normal, compile, ['-O -ddump-simpl -dsuppress-uniques -dno-typeable-binds']) test('T24359a', normal, compile, ['-O -ddump-rules']) test('T25713', [grep_errmsg('W:::')], compile, ['-O -ddump-simpl']) + +test('T25883', normal, compile_grep_core, ['']) +test('T25883b', normal, compile_grep_core, ['']) +test('T25883c', normal, compile_grep_core, ['']) +test('T25883d', [extra_files(['T25883d_import.hs'])], multimod_compile_filter, ['T25883d', '-O -ddump-simpl -dno-typeable-binds -dsuppress-all -dsuppress-uniques', r'grep -e "y ="']) ===================================== testsuite/tests/simplCore/should_run/T23429.hs ===================================== @@ -0,0 +1,39 @@ +{-# OPTIONS_GHC -fno-specialise-incoherents #-} +{-# LANGUAGE MonoLocalBinds #-} +class C a where + op :: a -> String + +instance {-# OVERLAPPABLE #-} C a where + op _ = "C a" + {-# NOINLINE op #-} + +instance {-# INCOHERENT #-} C (Maybe a) where + op _ = "C (Maybe a)" + {-# NOINLINE op #-} + +instance {-# INCOHERENT #-} C (Maybe ()) where + op _ = "C (Maybe ())" + {-# NOINLINE op #-} + +-- | Inhibit inlining, but keep specialize-ability +large :: a -> a +large x = x +{-# NOINLINE large #-} + +bar :: C a => a -> String +bar x = large (large (large (large (large (large (large (large (large (large (large (large (large (large (op x)))))))))))))) + +gen :: a -> String -- No C a constraint, has to choose the incoherent generic instance +gen = bar + +specMaybe :: Maybe a -> String -- C () constraint is resolved to the specialized instance for Maybe a +specMaybe = bar + +specMaybeUnit :: Maybe () -> String -- C () constraint is resolved to the specialized instance for Maybe () +specMaybeUnit = bar + +main :: IO () +main = do + putStrLn $ "gen () == " <> gen (Just ()) + putStrLn $ "specMaybe () == " <> specMaybe (Just ()) + putStrLn $ "specMaybeUnit () == " <> specMaybeUnit (Just ()) ===================================== testsuite/tests/simplCore/should_run/T23429.stdout ===================================== @@ -0,0 +1,3 @@ +gen () == C a +specMaybe () == C (Maybe a) +specMaybeUnit () == C (Maybe ()) ===================================== testsuite/tests/simplCore/should_run/all.T ===================================== @@ -119,3 +119,4 @@ test('T24725', normal, compile_and_run, ['-O -dcore-lint']) test('T25096', normal, compile_and_run, ['-O -dcore-lint']) test('AppIsHNF', normal, compile_and_run, ['-O']) test('T24359b', normal, compile_and_run, ['-O']) +test('T23429', normal, compile_and_run, ['-O']) View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/e2f2f9d08274c90e775a35dc5b776f… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/e2f2f9d08274c90e775a35dc5b776f… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][master] Add name for -Wunusable-unpack-pragmas
by Marge Bot (＠marge-bot) 20 Apr '25

20 Apr '25

Marge Bot pushed to branch master at Glasgow Haskell Compiler / GHC Commits: e2f2f9d0 by Vladislav Zavialov at 2025-04-20T10:53:39-04:00 Add name for -Wunusable-unpack-pragmas This warning had no name or flag and was triggered unconditionally. Now it is part of -Wdefault. In GHC.Tc.TyCl.tcTyClGroupsPass's strict mode, we now have to force-enable this warning to ensure that detection of flawed groups continues to work even if the user disables the warning with the -Wno-unusable-unpack-pragmas option. Test case: T3990c Also, the misnamed BackpackUnpackAbstractType is now called UnusableUnpackPragma. - - - - - 21 changed files: - compiler/GHC/Driver/Flags.hs - compiler/GHC/Driver/Session.hs - compiler/GHC/Tc/Errors/Ppr.hs - compiler/GHC/Tc/Errors/Types.hs - compiler/GHC/Tc/TyCl.hs - compiler/GHC/Tc/Utils/Monad.hs - compiler/GHC/Types/Error/Codes.hs - docs/users_guide/9.14.1-notes.rst - docs/users_guide/using-warnings.rst - testsuite/tests/simplCore/should_compile/Makefile - testsuite/tests/simplCore/should_compile/T23307c.stderr - + testsuite/tests/simplCore/should_compile/T3990c.hs - + testsuite/tests/simplCore/should_compile/T3990c.stdout - testsuite/tests/simplCore/should_compile/all.T - testsuite/tests/simplCore/should_fail/T25672.stderr - testsuite/tests/typecheck/should_compile/T7050.stderr - testsuite/tests/typecheck/should_fail/T3966.stderr - + testsuite/tests/typecheck/should_fail/T3966b.hs - + testsuite/tests/typecheck/should_fail/T3966b.stderr - testsuite/tests/typecheck/should_fail/all.T - testsuite/tests/unboxedsums/unpack_sums_5.stderr Changes: ===================================== compiler/GHC/Driver/Flags.hs ===================================== @@ -1098,6 +1098,7 @@ data WarningFlag = -- ^ @since 9.14, scheduled to be removed in 9.18 -- -- See Note [Quantifying over equalities in RULES] in GHC.Tc.Gen.Sig + | Opt_WarnUnusableUnpackPragmas -- Since 9.14 deriving (Eq, Ord, Show, Enum, Bounded) -- | Return the names of a WarningFlag @@ -1217,6 +1218,7 @@ warnFlagNames wflag = case wflag of Opt_WarnUselessSpecialisations -> "useless-specialisations" :| ["useless-specializations"] Opt_WarnDeprecatedPragmas -> "deprecated-pragmas" :| [] Opt_WarnRuleLhsEqualities -> "rule-lhs-equalities" :| [] + Opt_WarnUnusableUnpackPragmas -> "unusable-unpack-pragmas" :| [] -- ----------------------------------------------------------------------------- -- Standard sets of warning options @@ -1362,7 +1364,8 @@ standardWarnings -- see Note [Documenting warning flags] Opt_WarnViewPatternSignatures, Opt_WarnUselessSpecialisations, Opt_WarnDeprecatedPragmas, - Opt_WarnRuleLhsEqualities + Opt_WarnRuleLhsEqualities, + Opt_WarnUnusableUnpackPragmas ] -- | Things you get with @-W@. ===================================== compiler/GHC/Driver/Session.hs ===================================== @@ -2385,6 +2385,7 @@ wWarningFlagsDeps = [minBound..maxBound] >>= \x -> case x of Opt_WarnUselessSpecialisations -> warnSpec x Opt_WarnDeprecatedPragmas -> warnSpec x Opt_WarnRuleLhsEqualities -> warnSpec x + Opt_WarnUnusableUnpackPragmas -> warnSpec x warningGroupsDeps :: [(Deprecation, FlagSpec WarningGroup)] warningGroupsDeps = map mk warningGroups ===================================== compiler/GHC/Tc/Errors/Ppr.hs ===================================== @@ -2521,6 +2521,8 @@ instance Diagnostic TcRnMessage where -> WarningWithFlag Opt_WarnIncompleteRecordSelectors TcRnBadFieldAnnotation _ _ LazyFieldsDisabled -> ErrorWithoutFlag + TcRnBadFieldAnnotation _ _ UnusableUnpackPragma + -> WarningWithFlag Opt_WarnUnusableUnpackPragmas TcRnBadFieldAnnotation{} -> WarningWithoutFlag TcRnSuperclassCycle{} @@ -5814,7 +5816,7 @@ pprBadFieldAnnotationReason = \case text "Lazy field annotations (~) are disabled" UnpackWithoutStrictness -> text "UNPACK pragma lacks '!'" - BackpackUnpackAbstractType -> + UnusableUnpackPragma -> text "Ignoring unusable UNPACK pragma" pprSuperclassCycleDetail :: SuperclassCycleDetail -> SDoc ===================================== compiler/GHC/Tc/Errors/Types.hs ===================================== @@ -6275,13 +6275,28 @@ data BadFieldAnnotationReason where T14761a, T7562 -} UnpackWithoutStrictness :: BadFieldAnnotationReason - {-| An UNPACK pragma was applied to an abstract type in an indefinite package - in Backpack. + {-| An UNPACK pragma is unusable. + + A possible reason for this warning is that the UNPACK pragma was applied to + one of the following: + + * a function type @a -> b@ + * a recursive use of the data type being defined + * a sum type that cannot be unpacked, see @Note [UNPACK for sum types]@ + * a type/data family application with no matching instance in the environment + + However, it is deliberately /not/ emitted if: + + * the failure occurs in an indefinite package in Backpack + * the pragma is usable, but unpacking is disabled by @-O0@ Test cases: - unpack_sums_5, T3966, T7050 + unpack_sums_5, T3966, T7050, T25672, T23307c + + Negative test cases (must not trigger this warning): + T3990 -} - BackpackUnpackAbstractType :: BadFieldAnnotationReason + UnusableUnpackPragma :: BadFieldAnnotationReason deriving (Generic) data SuperclassCycle = ===================================== compiler/GHC/Tc/TyCl.hs ===================================== @@ -322,7 +322,7 @@ splice to separate the module and force the desired order of kind-checking: data D1 = MkD1 !(F Int) -- now (F Int) surely gets unpacked The current version of GHC is more predictable. Neither the (Complex Double) nor -the (F Int) example gets unpacking, the type/data instance is put into a +the (F Int) example gets unpacking unless the type/data instance is put into a separate HsGroup, either with $(return []) or by placing it in another module altogether. This is a direct result of placing instances after the other SCCs, as described in Note [Put instances at the end] in GHC.Rename.Module @@ -449,11 +449,20 @@ tcTyClGroupsPass all_gs thing_inside = go True ttcgs_zero mempty nilOL all_gs -- pass, the current group's lexical dependencies must have been -- satisfied by the preceding groups; no need for the ready check, -- this avoids some lookups in tcg_env + + -- See Note [Expedient use of diagnostics in tcTyClGroupsPass] + set_opts action + | strict = setWOptM Opt_WarnUnusableUnpackPragmas action + | otherwise = action + validate _ msgs _ + | strict = not (unpackErrorsFound msgs) + | otherwise = True + if not ready then return on_blocked else - tryTcDiscardingErrs' (\_ msgs _ -> not (strict && unpackErrorsFound msgs)) + tryTcDiscardingErrs' validate (return on_flawed) (return on_failed) - (on_success <$> tcTyClGroup g) + (on_success <$> set_opts (tcTyClGroup g)) data TcTyClGroupsStats = TcTyClGroupsStats @@ -479,15 +488,36 @@ instance Outputable TcTyClGroupsStats where , text "n_failed =" <+> ppr (ttcgs_n_failed stats) , text "n_flawed =" <+> ppr (ttcgs_n_flawed stats) ] +-- See Note [Expedient use of diagnostics in tcTyClGroupsPass] unpackErrorsFound :: Messages TcRnMessage -> Bool unpackErrorsFound = any is_unpack_error where is_unpack_error :: TcRnMessage -> Bool is_unpack_error (TcRnMessageWithInfo _ (TcRnMessageDetailed _ msg)) = is_unpack_error msg is_unpack_error (TcRnWithHsDocContext _ msg) = is_unpack_error msg - is_unpack_error (TcRnBadFieldAnnotation _ _ BackpackUnpackAbstractType) = True + is_unpack_error (TcRnBadFieldAnnotation _ _ UnusableUnpackPragma) = True is_unpack_error _ = False +{- Note [Expedient use of diagnostics in tcTyClGroupsPass] +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +In tcTyClGroupsPass.go with strict=True, we want to skip "flawed" groups, i.e. +groups with unusable unpack pragmas, as explained in Note [Retrying TyClGroups]. +To detect these unusable {-# UNPACK #-} pragmas, we currently piggy-back on the +diagnostics infrastructure: + + 1. (setWOptM Opt_WarnUnusableUnpackPragmas) to enable the warning. + The warning is on by default, but the user may have disabled it with + -Wno-unusable-unpack-pragmas, in which case we need to turn it back on. + + 2. (unpackErrorsFound msgs) to check if UnusableUnpackPragma is one of the + collected diagnostics. This is somewhat unpleasant because of the need to + recurse into TcRnMessageWithInfo and TcRnWithHsDocContext. + +Arguably, this is not a principled solution, because diagnostics are meant for +the user and here we inspect them to determine the order of type-checking. The +only reason for the current setup is that it was the easy thing to do. +-} + isReadyTyClGroup :: TcGblEnv -> TyClGroup GhcRn -> Bool isReadyTyClGroup tcg_env TyClGroup{group_ext = deps} = nameSetAll (\n -> n `elemNameEnv` tcg_type_env tcg_env) deps @@ -5123,7 +5153,7 @@ checkValidDataCon dflags existential_ok tc con -- warn in this case (it gives users the wrong idea about whether -- or not UNPACK on abstract types is supported; it is!) , isHomeUnitDefinite (hsc_home_unit hsc_env) - = addDiagnosticTc (bad_bang n BackpackUnpackAbstractType) + = addDiagnosticTc (bad_bang n UnusableUnpackPragma) | otherwise = return () ===================================== compiler/GHC/Tc/Utils/Monad.hs ===================================== @@ -20,7 +20,8 @@ module GHC.Tc.Utils.Monad( updTopFlags, getEnvs, setEnvs, updEnvs, restoreEnvs, xoptM, doptM, goptM, woptM, - setXOptM, unsetXOptM, unsetGOptM, unsetWOptM, + setXOptM, setWOptM, + unsetXOptM, unsetGOptM, unsetWOptM, whenDOptM, whenGOptM, whenWOptM, whenXOptM, unlessXOptM, getGhcMode, @@ -579,6 +580,9 @@ woptM flag = wopt flag <$> getDynFlags setXOptM :: LangExt.Extension -> TcRnIf gbl lcl a -> TcRnIf gbl lcl a setXOptM flag = updTopFlags (\dflags -> xopt_set dflags flag) +setWOptM :: WarningFlag -> TcRnIf gbl lcl a -> TcRnIf gbl lcl a +setWOptM flag = updTopFlags (\dflags -> wopt_set dflags flag) + unsetXOptM :: LangExt.Extension -> TcRnIf gbl lcl a -> TcRnIf gbl lcl a unsetXOptM flag = updTopFlags (\dflags -> xopt_unset dflags flag) ===================================== compiler/GHC/Types/Error/Codes.hs ===================================== @@ -711,7 +711,7 @@ type family GhcDiagnosticCode c = n | n -> c where -- TcRnBadFieldAnnotation/BadFieldAnnotationReason GhcDiagnosticCode "LazyFieldsDisabled" = 81601 GhcDiagnosticCode "UnpackWithoutStrictness" = 10107 - GhcDiagnosticCode "BackpackUnpackAbstractType" = 40091 + GhcDiagnosticCode "UnusableUnpackPragma" = 40091 -- TcRnRoleValidationFailed/RoleInferenceFailedReason GhcDiagnosticCode "TyVarRoleMismatch" = 22221 ===================================== docs/users_guide/9.14.1-notes.rst ===================================== @@ -25,17 +25,6 @@ Language This deprecation is controlled by the newly introduced ``-Wdeprecated-pragmas`` flag in ``-Wdefault``. -* A new flag, ``-Wuseless-specialisations``, controls warnings emitted when GHC - determines that a SPECIALISE pragma would have no effect. - -* A new flag, ``-Wrule-lhs-equalities``, controls warnings emitted for RULES - whose left-hand side attempts to quantify over equality constraints that - previous GHC versions accepted quantifying over. GHC will now drop such RULES, - emitting a warning message controlled by this flag. - - This warning is intended to give visibility to the fact that the RULES that - previous GHC versions generated in such circumstances could never fire. - * ``-Wincomplete-record-selectors`` is now part of `-Wall`, as specified by `GHC Proposal 516: add warning for incomplete record selectors <https://github.com/ghc-proposals/ghc-proposals/blob/master/proposals/0516-i…>`_. Hence, if a library is compiled with ``-Werror``, compilation may now fail. Solution: fix the library. @@ -93,6 +82,25 @@ Compiler :ghc-ticket:`20875`, :ghc-ticket:`21172`, :ghc-ticket:`22257`, :ghc-ticket:`25238`, :ghc-ticket:`25834`. +- A new flag, :ghc-flag:`-Wuseless-specialisations`, controls warnings emitted when GHC + determines that a SPECIALISE pragma would have no effect. + +- A new flag, :ghc-flag:`-Wrule-lhs-equalities`, controls warnings emitted for RULES + whose left-hand side attempts to quantify over equality constraints that + previous GHC versions accepted quantifying over. GHC will now drop such RULES, + emitting a warning message controlled by this flag. + + This warning is intended to give visibility to the fact that the RULES that + previous GHC versions generated in such circumstances could never fire. + +- A new flag, :ghc-flag:`-Wunusable-unpack-pragmas`, controls warnings emitted + when GHC is unable to unpack a data constructor field annotated by the + ``{-# UNPACK #-}`` pragma. + + Previous GHC versions issued this warning unconditionally. Now it is possible + to disable it with ``-Wno-unusable-unpack-pragmas`` or turn it into an error + with ``-Werror=unusable-unpack-pragmas``. + GHCi ~~~~ ===================================== docs/users_guide/using-warnings.rst ===================================== @@ -84,6 +84,7 @@ as ``-Wno-...`` for every individual warning in the group. * :ghc-flag:`-Wnoncanonical-monad-instances` * :ghc-flag:`-Wdata-kinds-tc` * :ghc-flag:`-Wimplicit-rhs-quantification` + * :ghc-flag:`-Wunusable-unpack-pragmas` .. ghc-flag:: -W :shortdesc: enable normal warnings @@ -2646,6 +2647,40 @@ of ``-W(no-)*``. To make the code forwards-compatible and silence the warning, users are advised to add parentheses manually. +.. ghc-flag:: -Wunusable-unpack-pragmas + :shortdesc: warn when an ``{-# UNPACK #-}`` pragma is unusable + :type: dynamic + :reverse: -Wno-unusable-unpack-pragmas + + :since: 9.14.1 + :default: on + + Warn on unusable ``{-# UNPACK #-}`` pragmas in data type declarations. + Examples:: + + data T = MkT {-# UNPACK #-} !(Int -> Bool) + + data G where + MkG :: {-# UNPACK #-} !G -> G + + type family F a where {} + data R a = MkR { fld :: {-# UNPACK #-} !(F a) } + + A possible reason for this warning is that the ``{-# UNPACK #-}`` pragma was + applied to one of the following: + + * a function type ``a -> b`` + * a recursive use of the data type being defined + * a sum type that cannot be unpacked + * a type/data family application with no matching instance in the environment + + However, it is deliberately **not** emitted if: + + * the failure occurs in an indefinite package in Backpack + * the pragma is usable, but unpacking is disabled by :ghc-flag:`-O0` + +---- + If you're feeling really paranoid, the :ghc-flag:`-dcore-lint` option is a good choice. It turns on heavyweight intra-pass sanity-checking within GHC. (It checks GHC's sanity, not yours.) ===================================== testsuite/tests/simplCore/should_compile/Makefile ===================================== @@ -43,6 +43,11 @@ T3990b: '$(TEST_HC)' $(TEST_HC_OPTS) -O -c -ddump-simpl T3990b.hs | grep 'test_case' # Grep output should show an unpacked constructor +T3990c: + $(RM) -f T3990c.o T3990c.hi + '$(TEST_HC)' $(TEST_HC_OPTS) -O -c -ddump-simpl T3990c.hs | grep 'test_case' + # Grep output should show an unpacked constructor + T8848: $(RM) -f T8848.o T8848.hi '$(TEST_HC)' $(TEST_HC_OPTS) -O -c -ddump-rule-firings T8848.hs | grep 'SPEC map2' ===================================== testsuite/tests/simplCore/should_compile/T23307c.stderr ===================================== @@ -1,5 +1,5 @@ - -T23307c.hs:7:13: warning: [GHC-40091] +T23307c.hs:7:13: warning: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault)] • Ignoring unusable UNPACK pragma on the first argument of ‘LCon’ • In the definition of data constructor ‘LCon’ In the data type declaration for ‘Loop’ + ===================================== testsuite/tests/simplCore/should_compile/T3990c.hs ===================================== @@ -0,0 +1,18 @@ +{-# OPTIONS -Wno-unusable-unpack-pragmas #-} + -- The warning is disabled, but this should not affect unpacking + +{-# LANGUAGE TypeFamilies #-} +module T3990c where + +type family F a + +data D1 = MkD1 {-# UNPACK #-} !(F Int) + -- This should actually get unpacked + +data D2 = MkD2 {-# UNPACK #-} !Int + {-# UNPACK #-} !Int + +type instance F Int = D2 + +test_case :: D1 +test_case = MkD1 (MkD2 1 1) ===================================== testsuite/tests/simplCore/should_compile/T3990c.stdout ===================================== @@ -0,0 +1,2 @@ +test_case :: D1 +test_case = T3990c.MkD1 1# 1# ===================================== testsuite/tests/simplCore/should_compile/all.T ===================================== @@ -211,6 +211,7 @@ test('T11742', normal, compile, ['-O2']) test('T11644', normal, compile, ['-O2']) test('T3990', normal, makefile_test, ['T3990']) test('T3990b', normal, makefile_test, ['T3990b']) +test('T3990c', normal, makefile_test, ['T3990c']) test('T12076', [], multimod_compile, ['T12076', '-v0']) test('T12076lit', normal, compile, ['-O']) ===================================== testsuite/tests/simplCore/should_fail/T25672.stderr ===================================== @@ -1,4 +1,4 @@ -T25672.hs:12:7: warning: [GHC-40091] +T25672.hs:12:7: warning: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault)] • Ignoring unusable UNPACK pragma on the first argument of ‘WrapIntOrWord’ • In the definition of data constructor ‘WrapIntOrWord’ ===================================== testsuite/tests/typecheck/should_compile/T7050.stderr ===================================== @@ -1,5 +1,5 @@ - -T7050.hs:3:14: warning: [GHC-40091] +T7050.hs:3:14: warning: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault)] • Ignoring unusable UNPACK pragma on the first argument of ‘Foo’ • In the definition of data constructor ‘Foo’ In the data type declaration for ‘Foo’ + ===================================== testsuite/tests/typecheck/should_fail/T3966.stderr ===================================== @@ -1,5 +1,5 @@ - -T3966.hs:5:16: error: [GHC-40091] [-Werror] +T3966.hs:5:16: error: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault), Werror=unusable-unpack-pragmas] • Ignoring unusable UNPACK pragma on the first argument of ‘Foo’ • In the definition of data constructor ‘Foo’ In the data type declaration for ‘Foo’ + ===================================== testsuite/tests/typecheck/should_fail/T3966b.hs ===================================== @@ -0,0 +1,12 @@ +{-# OPTIONS -Werror=unusable-unpack-pragmas #-} +{-# LANGUAGE TypeFamilies #-} + +module T3966b where + +data T = MkT {-# UNPACK #-} !(Int -> Bool) + +data G where + MkG :: {-# UNPACK #-} !G -> G + +type family F a where {} +data R a = MkR { fld :: {-# UNPACK #-} !(F a) } ===================================== testsuite/tests/typecheck/should_fail/T3966b.stderr ===================================== @@ -0,0 +1,15 @@ +T3966b.hs:6:10: error: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault), Werror=unusable-unpack-pragmas] + • Ignoring unusable UNPACK pragma on the first argument of ‘MkT’ + • In the definition of data constructor ‘MkT’ + In the data type declaration for ‘T’ + +T3966b.hs:9:3: error: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault), Werror=unusable-unpack-pragmas] + • Ignoring unusable UNPACK pragma on the first argument of ‘MkG’ + • In the definition of data constructor ‘MkG’ + In the data type declaration for ‘G’ + +T3966b.hs:12:12: error: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault), Werror=unusable-unpack-pragmas] + • Ignoring unusable UNPACK pragma on the first argument of ‘MkR’ + • In the definition of data constructor ‘MkR’ + In the data type declaration for ‘R’ + ===================================== testsuite/tests/typecheck/should_fail/all.T ===================================== @@ -213,6 +213,7 @@ test('T3613', normal, compile_fail, ['']) test('fd-loop', normal, compile_fail, ['']) test('T3950', normal, compile_fail, ['']) test('T3966', normal, compile_fail, ['']) +test('T3966b', normal, compile_fail, ['']) test('IPFail', normal, compile_fail, ['']) test('T3468', [], multimod_compile_fail, ['T3468', '-v0']) ===================================== testsuite/tests/unboxedsums/unpack_sums_5.stderr ===================================== @@ -1,10 +1,10 @@ - -unpack_sums_5.hs:4:22: warning: [GHC-40091] +unpack_sums_5.hs:4:22: warning: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault)] • Ignoring unusable UNPACK pragma on the first argument of ‘JustT’ • In the definition of data constructor ‘JustT’ In the data type declaration for ‘SMaybeT’ -unpack_sums_5.hs:7:10: warning: [GHC-40091] +unpack_sums_5.hs:7:10: warning: [GHC-40091] [-Wunusable-unpack-pragmas (in -Wdefault)] • Ignoring unusable UNPACK pragma on the first argument of ‘MkT’ • In the definition of data constructor ‘MkT’ In the data type declaration for ‘T’ + View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/e2f2f9d08274c90e775a35dc5b776f7… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/e2f2f9d08274c90e775a35dc5b776f7… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/marge_bot_batch_merge_job] 5 commits: Add name for -Wunusable-unpack-pragmas
by Marge Bot (＠marge-bot) 20 Apr '25

20 Apr '25

Marge Bot pushed to branch wip/marge_bot_batch_merge_job at Glasgow Haskell Compiler / GHC Commits: bed38773 by Vladislav Zavialov at 2025-04-20T06:23:15-04:00 Add name for -Wunusable-unpack-pragmas This warning had no name or flag and was triggered unconditionally. Now it is part of -Wdefault. In GHC.Tc.TyCl.tcTyClGroupsPass's strict mode, we now have to force-enable this warning to ensure that detection of flawed groups continues to work even if the user disables the warning with the -Wno-unusable-unpack-pragmas option. Test case: T3990c Also, the misnamed BackpackUnpackAbstractType is now called UnusableUnpackPragma. - - - - - 47aaf40c by Adam Gundry at 2025-04-20T06:23:17-04:00 Fix specialisation of incoherent instances (fixes #25883) GHC normally assumes that class constraints are canonical, meaning that the specialiser is allowed to replace one dictionary argument with another provided that they have the same type. The `-fno-specialise-incoherents` flag alters INCOHERENT instance definitions so that they will prevent specialisation in some cases, by inserting `nospec`. This commit fixes a bug in 7124e4ad76d98f1fc246ada4fd7bf64413ff2f2e, which treated some INCOHERENT instance matches as if `-fno-specialise-incoherents` was in effect, thereby unnecessarily preventing specialisation. In addition it updates the relevant `Note [Rules for instance lookup]` and adds a new `Note [Canonicity for incoherent matches]`. - - - - - e8878442 by Adam Gundry at 2025-04-20T06:23:18-04:00 Add regression test for #23429 - - - - - b8f2a43e by Adam Gundry at 2025-04-20T06:23:18-04:00 user's guide: update specification of overlapping/incoherent instances The description of the instance resolution algorithm in the user's guide was slightly out of date, because it mentioned in-scope given constraints only at the end, whereas the implementation checks for their presence before any of the other steps. This also adds a warning to the user's guide about the impact of incoherent instances on specialisation, and more clearly documents some of the other effects of `-XIncoherentInstances`. - - - - - aece199c by Matthew Craven at 2025-04-20T06:23:18-04:00 Fix bytecode generation for `tagToEnum# <LITERAL>` Fixes #25975. - - - - - 40 changed files: - compiler/GHC/Core/InstEnv.hs - compiler/GHC/Driver/Flags.hs - compiler/GHC/Driver/Session.hs - compiler/GHC/StgToByteCode.hs - compiler/GHC/Tc/Errors/Ppr.hs - compiler/GHC/Tc/Errors/Types.hs - compiler/GHC/Tc/Solver/Dict.hs - compiler/GHC/Tc/TyCl.hs - compiler/GHC/Tc/Utils/Monad.hs - compiler/GHC/Types/Error/Codes.hs - docs/users_guide/9.14.1-notes.rst - docs/users_guide/exts/instances.rst - docs/users_guide/using-warnings.rst - + testsuite/tests/bytecode/T25975.hs - + testsuite/tests/bytecode/T25975.stdout - testsuite/tests/bytecode/all.T - testsuite/tests/simplCore/should_compile/Makefile - testsuite/tests/simplCore/should_compile/T23307c.stderr - + testsuite/tests/simplCore/should_compile/T25883.hs - + testsuite/tests/simplCore/should_compile/T25883.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883b.hs - + testsuite/tests/simplCore/should_compile/T25883b.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883c.hs - + testsuite/tests/simplCore/should_compile/T25883c.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883d.hs - + testsuite/tests/simplCore/should_compile/T25883d.stderr - + testsuite/tests/simplCore/should_compile/T25883d_import.hs - + testsuite/tests/simplCore/should_compile/T3990c.hs - + testsuite/tests/simplCore/should_compile/T3990c.stdout - testsuite/tests/simplCore/should_compile/all.T - testsuite/tests/simplCore/should_fail/T25672.stderr - + testsuite/tests/simplCore/should_run/T23429.hs - + testsuite/tests/simplCore/should_run/T23429.stdout - testsuite/tests/simplCore/should_run/all.T - testsuite/tests/typecheck/should_compile/T7050.stderr - testsuite/tests/typecheck/should_fail/T3966.stderr - + testsuite/tests/typecheck/should_fail/T3966b.hs - + testsuite/tests/typecheck/should_fail/T3966b.stderr - testsuite/tests/typecheck/should_fail/all.T - testsuite/tests/unboxedsums/unpack_sums_5.stderr The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f0206e27434bfdd8151666d39417cb… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f0206e27434bfdd8151666d39417cb… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/T24603] 4 commits: RTS: remove target info and fix host info (#24058)
by Serge S. Gulin (＠gulin.serge) 20 Apr '25

20 Apr '25

Serge S. Gulin pushed to branch wip/T24603 at Glasgow Haskell Compiler / GHC Commits: b96e2f77 by Sylvain Henry at 2025-04-18T20:46:33-04:00 RTS: remove target info and fix host info (#24058) The RTS isn't a compiler, hence it doesn't have a target and we remove the reported target info displayed by "+RTS --info". We also fix the host info displayed by "+RTS --info": the host of the RTS is the RTS-building compiler's target, not the compiler's host (wrong when doing cross-compilation). - - - - - 6d9965f4 by Sylvain Henry at 2025-04-18T20:46:33-04:00 RTS: remove build info As per the discussion in !13967, there is no reason to tag the RTS with information about the build platform. - - - - - d52e9b3f by Vladislav Zavialov at 2025-04-18T20:47:15-04:00 Diagnostics: remove the KindMismatch constructor (#25957) The KindMismatch constructor was only used as an intermediate representation in pretty-printing. Its removal addresses a problem detected by the "codes" test case: [GHC-89223] is untested (constructor = KindMismatch) In a concious deviation from the usual procedure, the error code GHC-89223 is removed entirely rather than marked as Outdated. The reason is that it never was user-facing in the first place. - - - - - b6424e48 by Serge S. Gulin at 2025-04-20T09:31:47+03:00 Support for ARM64 Windows (LLVM-enabled) (fixes #24603) submodule Co-authored-by: Cheng Shao <terrorjack(a)type.dance> Co-authored-by: Dmitrii Egorov <egorov.d.i(a)icloud.com> Co-authored-by: Andrei Borzenkov <root(a)sandwitch.dev> - - - - - 40 changed files: - .gitlab-ci.yml - .gitlab/ci.sh - .gitlab/generate-ci/gen_ci.hs - .gitlab/hello.hs - .gitlab/jobs.yaml - compiler/CodeGen.Platform.h - compiler/GHC/CmmToAsm/AArch64/CodeGen.hs - compiler/GHC/CmmToAsm/AArch64/Instr.hs - compiler/GHC/CmmToAsm/AArch64/Ppr.hs - compiler/GHC/CmmToAsm/Reg/Linear/AArch64.hs - compiler/GHC/Driver/Session.hs - compiler/GHC/Platform/Regs.hs - compiler/GHC/Tc/Errors/Ppr.hs - compiler/GHC/Tc/Errors/Types.hs - compiler/GHC/Types/Error/Codes.hs - configure.ac - hadrian/src/Oracles/Setting.hs - hadrian/src/Rules/BinaryDist.hs - hadrian/src/Settings/Packages.hs - libraries/Cabal - libraries/Win32 - libraries/base/src/System/CPUTime/Windows.hsc - libraries/base/tests/perf/encodingAllocations.hs - libraries/directory - libraries/haskeline - libraries/process - libraries/unix - llvm-targets - m4/fp_cc_supports_target.m4 - m4/fptools_set_platform_vars.m4 - m4/ghc_tables_next_to_code.m4 - rts/RtsUtils.c - rts/StgCRun.c - rts/linker/PEi386.c - rts/win32/veh_excn.c - testsuite/ghc-config/ghc-config.hs - testsuite/tests/diagnostic-codes/codes.stdout - utils/ghc-toolchain/exe/Main.hs - utils/hsc2hs - utils/llvm-targets/gen-data-layout.sh The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/aca81c371f72ad910c1b6501a0631d… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/aca81c371f72ad910c1b6501a0631d… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/marge_bot_batch_merge_job] 6 commits: Support for ARM64 Windows (LLVM-enabled) (fixes #24603)
by Marge Bot (＠marge-bot) 20 Apr '25

20 Apr '25

Marge Bot pushed to branch wip/marge_bot_batch_merge_job at Glasgow Haskell Compiler / GHC Commits: cb5fcc24 by Serge S. Gulin at 2025-04-20T00:11:47-04:00 Support for ARM64 Windows (LLVM-enabled) (fixes #24603) submodule Co-authored-by: Cheng Shao <terrorjack(a)type.dance> Co-authored-by: Dmitrii Egorov <egorov.d.i(a)icloud.com> Co-authored-by: Andrei Borzenkov <root(a)sandwitch.dev> - - - - - 90c1e7a6 by Vladislav Zavialov at 2025-04-20T00:11:48-04:00 Add name for -Wunusable-unpack-pragmas This warning had no name or flag and was triggered unconditionally. Now it is part of -Wdefault. In GHC.Tc.TyCl.tcTyClGroupsPass's strict mode, we now have to force-enable this warning to ensure that detection of flawed groups continues to work even if the user disables the warning with the -Wno-unusable-unpack-pragmas option. Test case: T3990c Also, the misnamed BackpackUnpackAbstractType is now called UnusableUnpackPragma. - - - - - b975fdfe by Adam Gundry at 2025-04-20T00:11:49-04:00 Fix specialisation of incoherent instances (fixes #25883) GHC normally assumes that class constraints are canonical, meaning that the specialiser is allowed to replace one dictionary argument with another provided that they have the same type. The `-fno-specialise-incoherents` flag alters INCOHERENT instance definitions so that they will prevent specialisation in some cases, by inserting `nospec`. This commit fixes a bug in 7124e4ad76d98f1fc246ada4fd7bf64413ff2f2e, which treated some INCOHERENT instance matches as if `-fno-specialise-incoherents` was in effect, thereby unnecessarily preventing specialisation. In addition it updates the relevant `Note [Rules for instance lookup]` and adds a new `Note [Canonicity for incoherent matches]`. - - - - - 3fbf3433 by Adam Gundry at 2025-04-20T00:11:49-04:00 Add regression test for #23429 - - - - - 06784c1d by Adam Gundry at 2025-04-20T00:11:49-04:00 user's guide: update specification of overlapping/incoherent instances The description of the instance resolution algorithm in the user's guide was slightly out of date, because it mentioned in-scope given constraints only at the end, whereas the implementation checks for their presence before any of the other steps. This also adds a warning to the user's guide about the impact of incoherent instances on specialisation, and more clearly documents some of the other effects of `-XIncoherentInstances`. - - - - - f0206e27 by Matthew Craven at 2025-04-20T00:11:50-04:00 Fix bytecode generation for `tagToEnum# <LITERAL>` Fixes #25975. - - - - - 71 changed files: - .gitlab-ci.yml - .gitlab/ci.sh - .gitlab/generate-ci/gen_ci.hs - .gitlab/hello.hs - .gitlab/jobs.yaml - compiler/CodeGen.Platform.h - compiler/GHC/CmmToAsm/AArch64/CodeGen.hs - compiler/GHC/CmmToAsm/AArch64/Instr.hs - compiler/GHC/CmmToAsm/AArch64/Ppr.hs - compiler/GHC/CmmToAsm/Reg/Linear/AArch64.hs - compiler/GHC/Core/InstEnv.hs - compiler/GHC/Driver/Flags.hs - compiler/GHC/Driver/Session.hs - compiler/GHC/Platform/Regs.hs - compiler/GHC/StgToByteCode.hs - compiler/GHC/Tc/Errors/Ppr.hs - compiler/GHC/Tc/Errors/Types.hs - compiler/GHC/Tc/Solver/Dict.hs - compiler/GHC/Tc/TyCl.hs - compiler/GHC/Tc/Utils/Monad.hs - compiler/GHC/Types/Error/Codes.hs - docs/users_guide/9.14.1-notes.rst - docs/users_guide/exts/instances.rst - docs/users_guide/using-warnings.rst - hadrian/src/Oracles/Setting.hs - hadrian/src/Rules/BinaryDist.hs - libraries/Cabal - libraries/Win32 - libraries/base/src/System/CPUTime/Windows.hsc - libraries/base/tests/perf/encodingAllocations.hs - libraries/directory - libraries/haskeline - libraries/process - libraries/unix - llvm-targets - m4/fp_cc_supports_target.m4 - m4/fptools_set_platform_vars.m4 - m4/ghc_tables_next_to_code.m4 - rts/StgCRun.c - rts/linker/PEi386.c - rts/win32/veh_excn.c - + testsuite/tests/bytecode/T25975.hs - + testsuite/tests/bytecode/T25975.stdout - testsuite/tests/bytecode/all.T - testsuite/tests/simplCore/should_compile/Makefile - testsuite/tests/simplCore/should_compile/T23307c.stderr - + testsuite/tests/simplCore/should_compile/T25883.hs - + testsuite/tests/simplCore/should_compile/T25883.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883b.hs - + testsuite/tests/simplCore/should_compile/T25883b.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883c.hs - + testsuite/tests/simplCore/should_compile/T25883c.substr-simpl - + testsuite/tests/simplCore/should_compile/T25883d.hs - + testsuite/tests/simplCore/should_compile/T25883d.stderr - + testsuite/tests/simplCore/should_compile/T25883d_import.hs - + testsuite/tests/simplCore/should_compile/T3990c.hs - + testsuite/tests/simplCore/should_compile/T3990c.stdout - testsuite/tests/simplCore/should_compile/all.T - testsuite/tests/simplCore/should_fail/T25672.stderr - + testsuite/tests/simplCore/should_run/T23429.hs - + testsuite/tests/simplCore/should_run/T23429.stdout - testsuite/tests/simplCore/should_run/all.T - testsuite/tests/typecheck/should_compile/T7050.stderr - testsuite/tests/typecheck/should_fail/T3966.stderr - + testsuite/tests/typecheck/should_fail/T3966b.hs - + testsuite/tests/typecheck/should_fail/T3966b.stderr - testsuite/tests/typecheck/should_fail/all.T - testsuite/tests/unboxedsums/unpack_sums_5.stderr - utils/ghc-toolchain/exe/Main.hs - utils/hsc2hs - utils/llvm-targets/gen-data-layout.sh The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/21f0194fd38c233844c4621e50d12a… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/21f0194fd38c233844c4621e50d12a… You're receiving this email because of your account on gitlab.haskell.org.

1 0