- ghc-commits - Haskell.org

[Git][ghc/ghc] Deleted branch wip/fix-hpc-bytecode-modname
by Cheng Shao (＠TerrorJack) 28 May '26

28 May '26

Cheng Shao deleted branch wip/fix-hpc-bytecode-modname at Glasgow Haskell Compiler / GHC -- You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/supersven/libDir-setting] Update docs
by Sven Tennie (＠supersven) 28 May '26

28 May '26

Sven Tennie pushed to branch wip/supersven/libDir-setting at Glasgow Haskell Compiler / GHC Commits: f99d858c by Sven Tennie at 2026-05-28T07:31:13+00:00 Update docs - - - - - 1 changed file: - compiler/GHC/Settings.hs Changes: ===================================== compiler/GHC/Settings.hs ===================================== @@ -191,7 +191,7 @@ data FileSettings = FileSettings -- (e.g. settings). -- See Note [topdir: How GHC finds its files] in `GHC.SysTools.BaseDir` , fileSettings_globalPackageDatabase :: FilePath - -- ^ Path to the global package database + -- ^ Path to the global package database, relative to `libDir` , fileSettings_libDir :: FilePath -- ^ Directory containing GHC's library packages and the global package -- database. Defaults to 'fileSettings_topDir' but can differ in inplace View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/f99d858cc211ca2d6577d8bf6a53e7b… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/f99d858cc211ca2d6577d8bf6a53e7b… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/marge_bot_batch_merge_job] 4 commits: Fixes for black holes
by Marge Bot (＠marge-bot) 28 May '26

28 May '26

Marge Bot pushed to branch wip/marge_bot_batch_merge_job at Glasgow Haskell Compiler / GHC Commits: 99cb9c39 by Luite Stegeman at 2026-05-28T03:10:49-04:00 Fixes for black holes - suspend duplicate work for eager black holes - detect eager black holes in checkBlockingQueues - don't overwrite existing black holes even if they're not in an eager blackhole frame - don't deadlock on self when thunk is already blackholed Fixes #26936 - - - - - 80264eb1 by Wolfgang Jeltsch at 2026-05-28T03:10:49-04:00 Allow `downsweep` to use nodes of an existing module graph To this end, `downsweep` has not been able to use the nodes of a module graph obtained from a previous downsweeping round. In some GHC API applications, downsweeping is performed somewhat incrementally and therefore could profit from reusing such existing results. This contribution makes this possible. Resolves #27054. Co-authored-by: Matthew Pickering <matthewtpickering(a)gmail.com> - - - - - bfea8114 by Simon Jakobi at 2026-05-28T03:10:51-04:00 compiler: use nubOrd from containers Address #27103 by replacing GHC.Utils.Misc.ordNub[On] with Data.Containers.ListUtils.nubOrd[On]. Note that nubOrd suffers from a small inefficiency, a fix for which will be included in the next containers release: https://github.com/haskell/containers/issues/1202 - - - - - 179e22a9 by Simon Jakobi at 2026-05-28T03:10:52-04:00 Add regression test for T11226 Closes #11226. - - - - - 35 changed files: - + changelog.d/fix-blackhole-handling - + changelog.d/module-graph-reuse-in-downsweep - compiler/GHC/CmmToAsm/BlockLayout.hs - compiler/GHC/Driver/Backpack.hs - compiler/GHC/Driver/Downsweep.hs - compiler/GHC/Driver/Make.hs - compiler/GHC/Driver/Session/Units.hs - compiler/GHC/HsToCore/Usage.hs - compiler/GHC/Linker/Unit.hs - compiler/GHC/Rename/Pat.hs - compiler/GHC/Unit/Info.hs - compiler/GHC/Unit/State.hs - compiler/GHC/Utils/Misc.hs - ghc/GHCi/UI.hs - rts/Messages.c - rts/ThreadPaused.c - rts/Threads.c - rts/Updates.h - rts/include/rts/storage/ClosureMacros.h - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/A.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/B.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/C.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/D.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/X.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/Y.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/Z.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.stdout - testsuite/tests/ghc-api/downsweep/OldModLocation.hs - testsuite/tests/ghc-api/downsweep/PartialDownsweep.hs - testsuite/tests/ghc-api/downsweep/all.T - testsuite/tests/ghc-api/fixed-nodes/InterfaceModuleGraph.hs - + testsuite/tests/perf/should_run/T11226.hs - + testsuite/tests/perf/should_run/T11226.stdout - testsuite/tests/perf/should_run/all.T The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/9ba3e13a1e8a3f90c1ca0897058641… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/9ba3e13a1e8a3f90c1ca0897058641… 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: Trim the continuation in mkDupableContWithDmds
by Marge Bot (＠marge-bot) 28 May '26

28 May '26

Marge Bot pushed to branch wip/marge_bot_batch_merge_job at Glasgow Haskell Compiler / GHC Commits: 4a645683 by Simon Peyton Jones at 2026-05-27T21:41:59-04:00 Trim the continuation in mkDupableContWithDmds When there are no remaining argument demands, it means the application is bottoming. In this case, we can trim the continuation to avoid the panic that was observed in #27261. See Note [Trimming the continuation for bottoming functions] in GHC.Core.Opt.Simplify.Iteration. - - - - - 8ab506ff by Cheng Shao at 2026-05-27T21:42:47-04:00 ghci: fix module name string lifetime in hs_hpc_module invocation This patch makes hpcAddModule pass a properly malloced module name string to hs_hpc_module, instead of using useAsCString which causes use-after-free of module name string. Fixes #27297. Co-authored-by: Codex <codex(a)openai.com> - - - - - b0233814 by sheaf at 2026-05-27T21:43:31-04:00 Relax acceptance threshold for T10421 As seen in #27289, the 1% acceptance threshold for this text was overly narrow, resulting in spurious test failures. This commit widens the acceptance threshold to 2%. Fixes #27289. - - - - - d83f887f by Wolfgang Jeltsch at 2026-05-27T22:17:21-04:00 Allow `downsweep` to use nodes of an existing module graph To this end, `downsweep` has not been able to use the nodes of a module graph obtained from a previous downsweeping round. In some GHC API applications, downsweeping is performed somewhat incrementally and therefore could profit from reusing such existing results. This contribution makes this possible. Resolves #27054. Co-authored-by: Matthew Pickering <matthewtpickering(a)gmail.com> - - - - - 0264f626 by Simon Jakobi at 2026-05-27T22:17:22-04:00 compiler: use nubOrd from containers Address #27103 by replacing GHC.Utils.Misc.ordNub[On] with Data.Containers.ListUtils.nubOrd[On]. Note that nubOrd suffers from a small inefficiency, a fix for which will be included in the next containers release: https://github.com/haskell/containers/issues/1202 - - - - - 9ba3e13a by Simon Jakobi at 2026-05-27T22:17:23-04:00 Add regression test for T11226 Closes #11226. - - - - - 40 changed files: - + changelog.d/T27261 - + changelog.d/module-graph-reuse-in-downsweep - compiler/GHC/CmmToAsm/BlockLayout.hs - compiler/GHC/Core/Opt/Simplify/Iteration.hs - compiler/GHC/Core/Opt/Simplify/Utils.hs - compiler/GHC/Driver/Backpack.hs - compiler/GHC/Driver/Downsweep.hs - compiler/GHC/Driver/Make.hs - compiler/GHC/Driver/Session/Units.hs - compiler/GHC/HsToCore/Usage.hs - compiler/GHC/Linker/Unit.hs - compiler/GHC/Rename/Pat.hs - compiler/GHC/Unit/Info.hs - compiler/GHC/Unit/State.hs - compiler/GHC/Utils/Misc.hs - ghc/GHCi/UI.hs - + libraries/ghc-boot/GHC/Data/ShortByteString.hs - libraries/ghc-boot/ghc-boot.cabal.in - libraries/ghci/GHCi/Coverage.hs - libraries/ghci/GHCi/Run.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/A.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/B.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/C.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/D.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/X.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/Y.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.modules/Z.hs - + testsuite/tests/ghc-api/downsweep/IncrementalDownsweep.stdout - testsuite/tests/ghc-api/downsweep/OldModLocation.hs - testsuite/tests/ghc-api/downsweep/PartialDownsweep.hs - testsuite/tests/ghc-api/downsweep/all.T - testsuite/tests/ghc-api/fixed-nodes/InterfaceModuleGraph.hs - testsuite/tests/perf/compiler/all.T - + testsuite/tests/perf/should_run/T11226.hs - + testsuite/tests/perf/should_run/T11226.stdout - testsuite/tests/perf/should_run/all.T - + testsuite/tests/simplCore/should_compile/T27261.hs - + testsuite/tests/simplCore/should_compile/T27261_aux.hs - testsuite/tests/simplCore/should_compile/all.T The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f716714b9cc801ec40b946f1164db0… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f716714b9cc801ec40b946f1164db0… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][master] Relax acceptance threshold for T10421
by Marge Bot (＠marge-bot) 28 May '26

28 May '26

Marge Bot pushed to branch master at Glasgow Haskell Compiler / GHC Commits: b0233814 by sheaf at 2026-05-27T21:43:31-04:00 Relax acceptance threshold for T10421 As seen in #27289, the 1% acceptance threshold for this text was overly narrow, resulting in spurious test failures. This commit widens the acceptance threshold to 2%. Fixes #27289. - - - - - 1 changed file: - testsuite/tests/perf/compiler/all.T Changes: ===================================== testsuite/tests/perf/compiler/all.T ===================================== @@ -678,7 +678,7 @@ test ('T18140', ['-v0 -O']) test('T10421', [ only_ways(['normal']), - collect_compiler_runtime(1) + collect_compiler_runtime(2) # 1% tolerance was too small (#27289) ], multimod_compile, ['T10421', '-v0 -O']) View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/b0233814d63c2802a521dfc7dae08b1… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/b0233814d63c2802a521dfc7dae08b1… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][master] ghci: fix module name string lifetime in hs_hpc_module invocation
by Marge Bot (＠marge-bot) 28 May '26

28 May '26

Marge Bot pushed to branch master at Glasgow Haskell Compiler / GHC Commits: 8ab506ff by Cheng Shao at 2026-05-27T21:42:47-04:00 ghci: fix module name string lifetime in hs_hpc_module invocation This patch makes hpcAddModule pass a properly malloced module name string to hs_hpc_module, instead of using useAsCString which causes use-after-free of module name string. Fixes #27297. Co-authored-by: Codex <codex(a)openai.com> - - - - - 4 changed files: - + libraries/ghc-boot/GHC/Data/ShortByteString.hs - libraries/ghc-boot/ghc-boot.cabal.in - libraries/ghci/GHCi/Coverage.hs - libraries/ghci/GHCi/Run.hs Changes: ===================================== libraries/ghc-boot/GHC/Data/ShortByteString.hs ===================================== @@ -0,0 +1,17 @@ +module GHC.Data.ShortByteString + ( newCStringFromSBS + ) where + +import Prelude + +import qualified Data.ByteString.Short as SBS +import Foreign +import Foreign.C + +newCStringFromSBS :: SBS.ShortByteString -> IO CString +newCStringFromSBS sbs = + SBS.useAsCStringLen sbs $ \(src, len) -> do + dst <- mallocBytes (len + 1) + copyBytes dst src len + pokeByteOff dst len (0 :: Word8) + pure dst ===================================== libraries/ghc-boot/ghc-boot.cabal.in ===================================== @@ -51,6 +51,7 @@ Library exposed-modules: GHC.BaseDir + GHC.Data.ShortByteString GHC.Data.ShortText GHC.Data.SizedSeq GHC.Data.SmallArray ===================================== libraries/ghci/GHCi/Coverage.hs ===================================== @@ -9,9 +9,9 @@ import Prelude -- See note [Why do we import Prelude here?] import Control.Exception import Data.ByteString.Short (ShortByteString) -import qualified Data.ByteString.Short as SBS import Data.Word import Foreign +import GHC.Data.ShortByteString import GHC.Foreign (CString) import GHC.Utils.Encoding.UTF8 (utf8DecodeShortByteString) import GHCi.ObjLink (lookupSymbol) @@ -31,17 +31,19 @@ hpcAddModule :: -- ^ Name of the ticks array found in the c-stub. IO () hpcAddModule modlName ticks hash tickboxes = do - SBS.useAsCString modlName $ \modlNameLiteral -> do - -- we need to find the reference to the ticks array. - lookupSymbol tickboxes >>= \ case - Nothing -> do - -- the symbol is not found, this is a bug! - throwIO $ ErrorCall $ "hpcAddModule: failed to find symbol " <> utf8DecodeShortByteString tickboxes - Just tickBoxRef -> do - -- Calling 'hs_hpc_module' multiple times is safe, it will add the module only once. - hpc_register_module modlNameLiteral (fromIntegral ticks) (fromIntegral hash) (castPtr tickBoxRef) - -- calling 'hpc_startup' multiple times is safe, it will only be initialised once. - hpc_startup + -- we need to find the reference to the ticks array. + lookupSymbol tickboxes >>= \ case + Nothing -> do + -- the symbol is not found, this is a bug! + throwIO $ ErrorCall $ "hpcAddModule: failed to find symbol " <> utf8DecodeShortByteString tickboxes + Just tickBoxRef -> do + -- hs_hpc_module stores the module name pointer in the RTS hash table + -- until exitHpc, so pass a malloced C string. + modlNameLiteral <- newCStringFromSBS modlName + -- Calling 'hs_hpc_module' multiple times is safe, it will add the module only once. + hpc_register_module modlNameLiteral (fromIntegral ticks) (fromIntegral hash) (castPtr tickBoxRef) + -- calling 'hpc_startup' multiple times is safe, it will only be initialised once. + hpc_startup foreign import ccall unsafe "hs_hpc_module" hpc_register_module :: CString -> Word32 -> Word32 -> Ptr Word64 -> IO () ===================================== libraries/ghci/GHCi/Run.hs ===================================== @@ -37,6 +37,9 @@ import Control.Monad import Data.ByteString (ByteString) import qualified Data.ByteString.Short.Internal as BS import qualified Data.ByteString.Unsafe as B +#if defined(PROFILING) +import GHC.Data.ShortByteString +#endif import GHC.Exts import qualified GHC.Exts.Heap as Heap import GHC.Stack @@ -447,13 +450,6 @@ mkCostCentres mod ccs = do c_srcspan <- newCStringFromSBS srcspan toRemotePtr <$> c_mkCostCentre c_name c_module c_srcspan - newCStringFromSBS sbs = do - let len = BS.length sbs - buf <- mallocBytes $ len + 1 - BS.copyToPtr sbs 0 buf (fromIntegral len) - pokeByteOff buf len (0 :: Word8) - pure buf - foreign import ccall unsafe "mkCostCentre" c_mkCostCentre :: Ptr CChar -> Ptr CChar -> Ptr CChar -> IO (Ptr CostCentre) #else View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/8ab506ff3219560b20e1d9040e79337… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/8ab506ff3219560b20e1d9040e79337… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][master] Trim the continuation in mkDupableContWithDmds
by Marge Bot (＠marge-bot) 28 May '26

28 May '26

Marge Bot pushed to branch master at Glasgow Haskell Compiler / GHC Commits: 4a645683 by Simon Peyton Jones at 2026-05-27T21:41:59-04:00 Trim the continuation in mkDupableContWithDmds When there are no remaining argument demands, it means the application is bottoming. In this case, we can trim the continuation to avoid the panic that was observed in #27261. See Note [Trimming the continuation for bottoming functions] in GHC.Core.Opt.Simplify.Iteration. - - - - - 6 changed files: - + changelog.d/T27261 - compiler/GHC/Core/Opt/Simplify/Iteration.hs - compiler/GHC/Core/Opt/Simplify/Utils.hs - + testsuite/tests/simplCore/should_compile/T27261.hs - + testsuite/tests/simplCore/should_compile/T27261_aux.hs - testsuite/tests/simplCore/should_compile/all.T Changes: ===================================== changelog.d/T27261 ===================================== @@ -0,0 +1,10 @@ +section: compiler +issues: #27261 +mrs: !16084 +synopsis: + Avoid a crash in ``mkDupableContWithDmds`` when given empty demands +description: + The case of an empty list of remaining argument demands is now explicitly + handled by trimming the simplifier continuation, to avoid a compiler crash + of the form ``Non-exhaustive patterns in dmd : cont_dmds`` or ``expectNonEmpty`` + in ``mkDupableContWithDmds``. ===================================== compiler/GHC/Core/Opt/Simplify/Iteration.hs ===================================== @@ -62,6 +62,7 @@ import GHC.Types.Var ( isTyCoVar ) import GHC.Builtin.Types.Prim( realWorldStatePrimTy ) import GHC.Builtin.Names( runRWKey, seqHashKey ) +import qualified GHC.Data.List.Infinite as Inf import GHC.Data.Maybe ( isNothing, orElse, mapMaybe ) import GHC.Data.FastString import GHC.Unit.Module ( moduleName ) @@ -2444,24 +2445,9 @@ rebuildCall env arg_info _cont ---------- Bottoming applications -------------- rebuildCall env (ArgInfo { ai_fun = fun, ai_args = rev_args, ai_dmds = [] }) cont - -- When we run out of strictness args, it means - -- that the call is definitely bottom; see GHC.Core.Opt.Simplify.Utils.mkArgInfo - -- Then we want to discard the entire strict continuation. E.g. - -- * case (error "hello") of { ... } - -- * (error "Hello") arg - -- * f (error "Hello") where f is strict - -- etc - -- Then, especially in the first of these cases, we'd like to discard - -- the continuation, leaving just the bottoming expression. But the - -- type might not be right, so we may have to add a coerce. - | not (contIsTrivial cont) -- Only do this if there is a non-trivial - -- continuation to discard, else we do it - -- again and again! - = seqType cont_ty `seq` -- See Note [Avoiding space leaks in OutType] - return (emptyFloats env, castBottomExpr res cont_ty) - where - res = argInfoExpr fun rev_args - cont_ty = contResultType cont + -- When we run out of demands, it means that the call is definitely bottom. + -- See (TC2) in Note [Trimming the continuation for bottoming functions] + = rebuild env (argInfoExpr fun rev_args) (mkBottomCont cont) ---------- Simplify type applications -------------- rebuildCall env info (ApplyToTy { sc_arg_ty = arg_ty, sc_hole_ty = hole_ty, sc_cont = cont }) @@ -4045,6 +4031,41 @@ When we have then we can just duplicate those alts because the A and C cases will disappear immediately. This is more direct than creating join points and inlining them away. See #4930. + +Note [Trimming the continuation for bottoming functions] +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +Suppose + f :: Int -> Int -> Int + f x = error "urk" + + foo = f 3 4 + +f's demand signature say "after one arg I return bottom". We can drop +the remaining arguments, thus + + foo = case f 3 of {} + +This trimming can also be done with other continuations: + * case (error "hello") of { ... } + * f (error "Hello") where f is strict + etc + +We implement the trimming in three parts: + +(TC1) In `mkArgInfo`, for a bottoming function, we make a list of `RemainingArgDmds` + with a finite list of elements (in the example above, just one). + + For comparison, note that, for non-bottoming functions, the `RemainingArgDmds` + always finishes with an infinite list of `topDmd`. + +(TC2) In `rebuildCall`, when we run out of `RemainingArgDmds` we discard the + remaining continuation. + + After discarding the continuation, the types might not match, in which case + we leave behind a (case <hole> of {}) wrapper. See the call to `mkBottomCont`. + +(TC3) In `mkDupableContWithDmds`, we similarly discard the continuation when + we run out of `RemainingArgDmds`. -} -------------------- @@ -4079,10 +4100,10 @@ mkDupableCont env cont = mkDupableContWithDmds (zapSubstEnv env) (repeat topDmd) cont mkDupableContWithDmds - :: SimplEnvIS -> [Demand] -- Demands on arguments; always infinite + :: SimplEnvIS -> RemainingArgDmds -> SimplCont -> SimplM ( SimplFloats, SimplCont) -mkDupableContWithDmds env _ cont +mkDupableContWithDmds env remaining_dmds cont -- Check the invariant | assertPpr (checkSimplEnvIS env) (pprBadSimplEnvIS env) False = pprPanic "mkDupableContWithDmds" empty @@ -4090,6 +4111,13 @@ mkDupableContWithDmds env _ cont | contIsDupable cont = return (emptyFloats env, cont) + -- No more demands => function is definitely bottom + -- => simply trim the continuation + -- c.f. the null-demands case in `rebuildCall` + -- See (TC3) in Note [Trimming the continuation for bottoming functions] + | null remaining_dmds + = return (emptyFloats env, mkBottomCont cont) + mkDupableContWithDmds _ _ (Stop {}) = panic "mkDupableCont" -- Handled by previous eqn mkDupableContWithDmds env dmds (CastIt { sc_co = co, sc_opt = opt, sc_cont = cont }) @@ -4134,7 +4162,8 @@ mkDupableContWithDmds env _ , thumbsUpPlanA cont = -- Use Plan A of Note [Duplicating StrictArg] -- pprTrace "Using plan A" (ppr (ai_fun fun) $$ text "args" <+> ppr (ai_args fun) $$ text "cont" <+> ppr cont) $ - do { let _ :| dmds = expectNonEmpty $ ai_dmds fun + do { let _ :| dmds = expectNonEmpty (ai_dmds fun) -- See Invariant of StrictArg; + -- ai_dmds is never empty ; (floats1, cont') <- mkDupableContWithDmds env dmds cont -- Use the demands from the function to add the right -- demand info on any bindings we make for further args @@ -4180,7 +4209,10 @@ mkDupableContWithDmds env dmds -- let a = ...arg... -- in [...hole...] a -- NB: sc_dup /= OkToDup; that is caught earlier by contIsDupable - do { let dmd:|cont_dmds = expectNonEmpty dmds + do { let dmd:|cont_dmds = + -- We took care to handle an empty demand list at the start, + -- ensuring this call to 'expectNonEmpty' does not panic (#27261). + expectNonEmpty dmds ; (floats1, cont') <- mkDupableContWithDmds env cont_dmds cont ; let env' = env `setInScopeFromF` floats1 ; arg' <- simplArg env' Nothing hole_ty se arg arg_mco @@ -4251,7 +4283,7 @@ mkDupableStrictBind env arg_bndr join_rhs res_ty ; let arg_info = ArgInfo { ai_fun = join_bndr , ai_rules = [], ai_args = [] , ai_encl = False, ai_dmds = repeat topDmd - , ai_discs = repeat 0 } + , ai_discs = Inf.repeat 0 } ; return ( addJoinFloats (emptyFloats env) $ unitJoinFloat $ NonRec join_bndr $ ===================================== compiler/GHC/Core/Opt/Simplify/Utils.hs ===================================== @@ -25,13 +25,13 @@ module GHC.Core.Opt.Simplify.Utils ( StaticEnv(..), isSimplified, contIsStop, contIsDupable, contResultType, contHoleType, contHoleScaling, - contIsTrivial, contArgs, contIsRhs, + contIsTrivial, contArgs, contIsRhs, mkBottomCont, hasArgs, countArgs, contOutArgs, dropContArgs, mkBoringStop, mkRhsStop, mkLazyArgStop, interestingCallContext, -- ArgInfo - ArgInfo(..), ArgSpec(..), mkArgInfo, + ArgInfo(..), ArgSpec(..), RemainingArgDmds, mkArgInfo, addValArgTo, addTyArgTo, argInfoExpr, argSpecArg, pushOutArgs, pushArgSpecs, @@ -54,8 +54,10 @@ import GHC.Core.Opt.Stats ( Tick(..) ) import qualified GHC.Core.Subst import GHC.Core.Ppr import GHC.Core.TyCo.Ppr ( pprParendType ) +import GHC.Core.TyCo.Compare ( eqTypeIgnoringMultiplicity ) import GHC.Core.FVs import GHC.Core.Utils +import GHC.Core.Make( mkWildValBinder ) import GHC.Core.Opt.Arity import GHC.Core.Unfold import GHC.Core.Unfold.Make @@ -75,6 +77,8 @@ import GHC.Types.Var.Set import GHC.Types.Basic import GHC.Types.Name.Env +import GHC.Data.List.Infinite ( Infinite(..) ) +import qualified GHC.Data.List.Infinite as Inf import GHC.Data.OrdList ( isNilOL ) import GHC.Data.FastString ( fsLit ) @@ -205,10 +209,10 @@ data SimplCont | StrictArg -- (StrictArg (f e1 ..en) K)[e] = K[ f e1 .. en e ] { sc_dup :: DupFlag - , sc_fun :: ArgInfo -- Specifies f, e1..en, Whether f has rules, etc + , sc_fun :: ArgInfo -- Specifies f, e1..en, whether f has rules, etc -- plus demands and discount flags for *this* arg -- and further args - -- So ai_dmds and ai_discs are never empty + -- Invariant: ai_dmds and ai_discs are never empty , sc_fun_ty :: OutType -- Type of the function (f e1 .. en), -- presumably (arg_ty -> res_ty) -- where res_ty is expected by sc_cont @@ -348,32 +352,41 @@ doesn't matter because we'll never compute them all. data ArgInfo = ArgInfo { - ai_fun :: OutId, -- The function - ai_args :: [ArgSpec], -- ...applied to these args (which are in *reverse* order) + ai_fun :: OutId, -- ^ The function + ai_args :: [ArgSpec], -- ^ ...applied to these args (which are in *reverse* order) -- NB: all these argumennts are already simplified - ai_rules :: [CoreRule], -- Rules for this function - ai_encl :: Bool, -- Flag saying whether this function - -- or an enclosing one has rules (recursively) - -- True => be keener to inline in all args + ai_rules :: [CoreRule], -- ^ Rules for this function + ai_encl :: Bool, + -- ^ Flag saying whether this function or an enclosing one has rules + -- (recursively) + -- + -- @True@ means: be keener to inline in all args - ai_dmds :: [Demand], -- Demands on remaining value arguments (beyond ai_args) - -- Usually infinite, but if it is finite it guarantees - -- that the function diverges after being given - -- that number of args + ai_dmds :: RemainingArgDmds, + -- ^ Demands on remaining value arguments (beyond 'ai_args') - ai_discs :: [Int] -- Discounts for remaining value arguments (beyond ai_args) - -- non-zero => be keener to inline - -- Always infinite + ai_discs :: Infinite Int + -- ^ Discounts for remaining value arguments (beyond 'ai_args') + -- + -- A non-zero value means: be keener to inline } -data ArgSpec - = ValArg { as_dmd :: Demand -- Demand placed on this argument - , as_arg :: OutExpr -- Apply to this (coercion or value); c.f. ApplyToVal - , as_hole_ty :: OutType } -- Type of the function (presumably t1 -> t2) +-- | 'RemainingArgDmds' gives the demands on any remaining value arguments. +-- +-- It is usually infinite (with 'topDmd's in the tail), but if it is finite it +-- guarantees that the function diverges after being applied to that number +-- of arguments. +type RemainingArgDmds = [Demand] - | TyArg { as_arg_ty :: OutType -- Apply to this type; c.f. ApplyToTy - , as_hole_ty :: OutType } -- Type of the function (presumably forall a. blah) +data ArgSpec + -- | A value argument + = ValArg { as_dmd :: Demand -- ^ Demand placed on this argument + , as_arg :: OutExpr -- ^ Apply to this (coercion or value); c.f. 'ApplyToVal' + , as_hole_ty :: OutType } -- ^ Type of the function (presumably @t1 -> t2@ for 'ValArg' or @forall a. blah@ for 'TyArg') + -- | A type argument + | TyArg { as_arg_ty :: OutType -- ^ Apply to this type; c.f. 'ApplyToTy' + , as_hole_ty :: OutType } -- ^ Type of the function (presumably @t1 -> t2@ for 'ValArg' or @forall a. blah@ for 'TyArg') instance Outputable ArgInfo where ppr (ArgInfo { ai_fun = fun, ai_args = args, ai_dmds = dmds, ai_rules = rules }) @@ -389,7 +402,7 @@ instance Outputable ArgSpec where addValArgTo :: ArgInfo -> OutExpr -> OutType -> ArgInfo addValArgTo ai arg hole_ty - | ArgInfo { ai_dmds = dmd:dmds, ai_discs = _:discs } <- ai + | ArgInfo { ai_dmds = dmd:dmds, ai_discs = Inf _ discs } <- ai -- Pop the top demand and and discounts off , let arg_spec = ValArg { as_arg = arg, as_hole_ty = hole_ty, as_dmd = dmd } = ai { ai_args = arg_spec : ai_args ai @@ -492,12 +505,23 @@ contIsDupable (TickIt _ k) = contIsDupable k contIsTrivial :: SimplCont -> Bool contIsTrivial (Stop {}) = True contIsTrivial (ApplyToTy { sc_cont = k }) = contIsTrivial k --- This one doesn't look right. A value application is not trivial --- contIsTrivial (ApplyToVal { sc_arg = Coercion _, sc_cont = k }) = contIsTrivial k contIsTrivial (CastIt { sc_cont = k }) = contIsTrivial k contIsTrivial _ = False ------------------- +contStop :: SimplCont -> SimplCont +-- ^ Get the 'Stop' at the tail of the continuation +-- +-- Always returns a continuation of form @(Stop ...)@. +contStop stop@(Stop {}) = stop +contStop (CastIt { sc_cont = k }) = contStop k +contStop (StrictBind { sc_cont = k }) = contStop k +contStop (StrictArg { sc_cont = k }) = contStop k +contStop (Select { sc_cont = k }) = contStop k +contStop (ApplyToTy { sc_cont = k }) = contStop k +contStop (ApplyToVal { sc_cont = k }) = contStop k +contStop (TickIt _ k) = contStop k + contResultType :: SimplCont -> OutType contResultType (Stop ty _ _) = ty contResultType (CastIt { sc_cont = k }) = contResultType k @@ -651,6 +675,35 @@ contEvalContext bndrs cont = go cont -- Perhaps reconstruct the demand on the scrutinee by looking at field -- and case binder dmds, see addCaseBndrDmd. No priority right now. +------------------- +mkBottomCont ::SimplCont -> SimplCont +-- ^ Given a continuation `cont`, return a `cont` /of the same type/, +-- looking like @(case \<hole\> of {})@. +-- +-- This is used when we are going to fill in the @<hole>@ with bottom. +-- See (TC2,3) in Note [Trimming the continuation for bottoming functions] +-- +-- Don't bother to trim, making a @case <hole> of {}@, if we have only +-- an essentially-trivial continuation; e.g. @(<hole> \@ty |> co)@. +mkBottomCont cont = go cont + where + go k@(Stop {}) = k + go (TickIt t k') = TickIt t (go k') + go k@(CastIt { sc_cont = k' }) = k { sc_cont = go k' } + go k@(ApplyToTy { sc_cont = k' }) = k { sc_cont = go k' } + go k@(Select { sc_alts = [], sc_cont = Stop {} }) = k -- Optimisation only + go k | Stop res_ty _ _ <- stop_cont + , hole_ty `eqTypeIgnoringMultiplicity` res_ty + = stop_cont + | otherwise + = Select { sc_alts = [] + , sc_bndr = mkWildValBinder OneTy hole_ty + , sc_env = Simplified OkDup + , sc_cont = stop_cont } + where + hole_ty = contHoleType k + stop_cont = contStop k + ------------------- mkArgInfo :: SimplEnv -> Id -> [CoreRule] -> SimplCont -> ArgInfo mkArgInfo env fun rules_for_fun cont @@ -672,16 +725,17 @@ mkArgInfo env fun rules_for_fun cont fun_has_rules = not (null rules_for_fun) - vanilla_discounts, arg_discounts :: [Int] - vanilla_discounts = repeat 0 + vanilla_discounts, arg_discounts :: Infinite Int + vanilla_discounts = Inf.repeat 0 arg_discounts = case idUnfolding fun of CoreUnfolding {uf_guidance = UnfIfGoodArgs {ug_args = discounts}} - -> discounts ++ vanilla_discounts + -> discounts Inf.++ vanilla_discounts _ -> vanilla_discounts - vanilla_dmds, arg_dmds :: [Demand] + vanilla_dmds :: RemainingArgDmds vanilla_dmds = repeat topDmd + arg_dmds :: RemainingArgDmds arg_dmds | not (seInline env) = vanilla_dmds -- See Note [Do not expose strictness if sm_inline=False] @@ -689,26 +743,22 @@ mkArgInfo env fun rules_for_fun cont = -- add_type_str fun_ty $ case splitDmdSig (idDmdSig fun) of (demands, result_info) - | not (demands `lengthExceeds` n_val_args) - -> -- Enough args, use the strictness given. - -- For bottoming functions we used to pretend that the arg - -- is lazy, so that we don't treat the arg as an - -- interesting context. This avoids substituting - -- top-level bindings for (say) strings into - -- calls to error. But now we are more careful about - -- inlining lone variables, so its ok - -- (see GHC.Core.Op.Simplify.Utils.analyseCont) - if isDeadEndDiv result_info then - demands -- Finite => result is bottom - else - demands ++ vanilla_dmds + | not (demands `lengthExceeds` n_val_args) + -> remaining_dmds -- Enough args, use the strictness given. | otherwise -> warnPprTrace True "More demands than arity" (ppr fun <+> ppr (idArity fun) <+> ppr n_val_args <+> ppr demands) $ vanilla_dmds -- Not enough args, or no strictness - add_type_strictness :: Type -> [Demand] -> [Demand] - -- If the function arg types are strict, record that in the 'strictness bits' + where + remaining_dmds :: RemainingArgDmds + -- isDeadEndDiv: if remaining_dmds is finite, result is bottom + -- See (TC1) in Note [Trimming the continuation for bottoming functions] + remaining_dmds | isDeadEndDiv result_info = demands + | otherwise = demands ++ vanilla_dmds + + add_type_strictness :: Type -> RemainingArgDmds -> RemainingArgDmds + -- If the function arg /types/ are strict, record that in the RemainingArgDmds -- No need to instantiate because unboxed types (which dominate the strict -- types) can't instantiate type variables. -- add_type_strictness is done repeatedly (for each call); @@ -915,16 +965,16 @@ the incentive to disappear when we inline `f`! lazyArgContext :: ArgInfo -> CallCtxt -- Use this for lazy arguments lazyArgContext (ArgInfo { ai_encl = encl_rules, ai_discs = discs }) - | encl_rules = RuleArgCtxt - | disc:_ <- discs, disc > 0 = DiscArgCtxt -- Be keener here - | otherwise = BoringCtxt -- Nothing interesting + | encl_rules = RuleArgCtxt + | Inf disc _ <- discs, disc > 0 = DiscArgCtxt -- Be keener here + | otherwise = BoringCtxt -- Nothing interesting strictArgContext :: ArgInfo -> CallCtxt strictArgContext (ArgInfo { ai_encl = encl_rules, ai_discs = discs }) -- Use this for strict arguments - | encl_rules = RuleArgCtxt - | disc:_ <- discs, disc > 0 = DiscArgCtxt -- Be keener here - | otherwise = RhsCtxt NonRecursive + | encl_rules = RuleArgCtxt + | Inf disc _ <- discs, disc > 0 = DiscArgCtxt -- Be keener here + | otherwise = RhsCtxt NonRecursive -- Why RhsCtxt? if we see f (g x), and f is strict, we -- want to be a bit more eager to inline g, because it may -- expose an eval (on x perhaps) that can be eliminated or ===================================== testsuite/tests/simplCore/should_compile/T27261.hs ===================================== @@ -0,0 +1,17 @@ +{-# OPTIONS_GHC -fno-full-laziness #-} + +module T27261 (foo) where + +import T27261_aux (myError) + +foo :: [String] -> (() -> Int) -> Int +foo cs = + \ k -> ( case bar of + Just str -> let cs2 = case cs of { [] -> cs; _ -> "stack entry" : cs } + in myError cs2 str + Nothing -> \ c -> c () ) + ( \ _ -> k () ) + +bar :: Maybe String +bar = Nothing +{-# NOINLINE bar #-} ===================================== testsuite/tests/simplCore/should_compile/T27261_aux.hs ===================================== @@ -0,0 +1,7 @@ +{-# LANGUAGE BangPatterns #-} + +module T27261_aux (myError) where + +myError :: [String] -> String -> a +myError !_ _ = undefined +{-# NOINLINE myError #-} ===================================== testsuite/tests/simplCore/should_compile/all.T ===================================== @@ -601,3 +601,4 @@ test('T25718a', normal, compile, ['-O -ddump-simpl -dsuppress-uniques -dsuppress test('T25718b', normal, compile, ['-O -ddump-simpl -dsuppress-uniques -dsuppress-all -dno-typeable-binds']) test('T25718c', normal, compile, ['-O -ddump-simpl -dsuppress-uniques -dsuppress-all -dno-typeable-binds']) test('T19166', normal, compile, ['-O -ddump-simpl -dsuppress-uniques -dsuppress-all -dno-typeable-binds']) +test('T27261', [extra_files(['T27261_aux.hs'])], multimod_compile, ['T27261', '-v0 -O']) View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/4a645683ee0bd4421a88cd6ec49b40c… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/4a645683ee0bd4421a88cd6ec49b40c… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/az/ghc-10.1-exactprint-update-wip] Added an intro section
by Simon Peyton Jones (＠simonpj) 28 May '26

28 May '26

Simon Peyton Jones pushed to branch wip/az/ghc-10.1-exactprint-update-wip at Glasgow Haskell Compiler / GHC Commits: 2d4151e9 by Simon Peyton Jones at 2026-05-28T00:10:26+01:00 Added an intro section - - - - - 1 changed file: - ExactPrint.md Changes: ===================================== ExactPrint.md ===================================== @@ -13,6 +13,67 @@ information directly into every AST node to allow the tree to be reprinted _exac original source, byte-for-byte (modulo tab expansion). When a tool modifies the AST it adjusts only the annotations it needs to change; all surrounding nodes reprint themselves unchanged. +## Background: the syntax tree + +The AST of the source program is represented using the data types defined in `Language.Haskell.Syntax.*`. +This data type uses the "Trees That Grow (TTG)" plan; +see [Implementing trees that grow](https://gitlab.haskell.org/ghc/ghc/-/wikis/implementing-trees-that-gr… + +For example, in (GHC-independent) `Language.Haskell.Syntax.Expr`: +``` +data HsExpr p + = ... many constructors including ... + | HsLet (XLet p) + (HsLocalBinds p) + (LHsExpr p) + +type family XLet p +type LHsExpr p = XRec p (HsExpr p) + +``` +Note that +* The `(XLet p)` field is the *extension field* of the `HsLet` data constructor, + where `XLet` is a type family. +* Almost every node is wrapped in an `XRec`, another type family. That makes it easy + for clients to attach arbitrary information to each node. + +GHC specialises this data a type in `GHC.Hs.*`, as follows: +``` +data Pass = Parsed | Renamed | Typechecked + +data GhcPass (c :: Pass) where + GhcPs :: GhcPass 'Parsed + GhcRn :: GhcPass 'Renamed + GhcTc :: GhcPass 'Typechecked + +type family XRec p a = r | r -> a +type instance XRec (GhcPass p) t = XRecGhc t + +-- (XRecGhc tree) wraps `tree` in a GHC-specific, +-- but pass-independent, source location +type XRecGhc t = GenLocated (Anno t) t + +data GenLocated l e = L l e +type family Anno t +``` +So, via `XRec`, every node `e :: t` in the GHC-specific version of HsSyn is wrapped in +a `L ann e`, where `ann :: Anno t` is the annotation on the node. + +Notice that there are two places we can hang information: +* (EXT) **Constructor-specific punctuation**: the extension field of each data constructor + can contain information that is specific to that constructor. Example: the location + of the keywords `let` and `in` for the `HsLet` construct. + +* (XREC) **Entire-node information**: the `Anno t` field that wraps almost every node in the + syntax tree can contain information that is needed for *every* node. Classic example: + the `SrcSpan` of the node. + +**NOTE**: currently there is information in (XREC) that more properly belongs in (EXT). +A refactoring project is under way to put this right. + + +## File layout + The subsystem lives primarily in `utils/check-exact/` and is exposed through three main layers: | Layer | Key operation | Purpose | View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/2d4151e92b1cce8fcb7140619d008ac… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/2d4151e92b1cce8fcb7140619d008ac… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/sjakobi/multi-caret] Add TcRnDuplicateDecls
by Simon Jakobi (＠sjakobi2) 28 May '26

28 May '26

Simon Jakobi pushed to branch wip/sjakobi/multi-caret at Glasgow Haskell Compiler / GHC Commits: 1f7ffaaa by Simon Jakobi at 2026-05-27T22:54:22+02:00 Add TcRnDuplicateDecls - - - - - 3 changed files: - compiler/GHC/Tc/Errors/Ppr.hs - testsuite/tests/rename/should_fail/T7164.stderr - testsuite/tests/rename/should_fail/all.T Changes: ===================================== compiler/GHC/Tc/Errors/Ppr.hs ===================================== @@ -3464,6 +3464,8 @@ instance Diagnostic TcRnMessage where -> diagnosticRelatedLocations msg' TcRnBindingNameConflict _ locs -> NE.tail locs + TcRnDuplicateDecls _ names + -> map nameSrcSpan (NE.init names) _ -> [] ===================================== testsuite/tests/rename/should_fail/T7164.stderr ===================================== @@ -1,5 +1,11 @@ - -T7164.hs:8:1: [GHC-29916] +T7164.hs:8:1: error: [GHC-29916] Multiple declarations of ‘derp’ Declared at: T7164.hs:5:5 T7164.hs:8:1 + | +5 | derp :: m a + | ^^^^^^^^^^^ + | +8 | derp = 123 + | ^^^^ + ===================================== testsuite/tests/rename/should_fail/all.T ===================================== @@ -91,7 +91,7 @@ test('T6148a', normal, compile_fail, ['']) test('T6148b', normal, compile_fail, ['']) test('T6148c', normal, compile_fail, ['']) test('T6148d', normal, compile_fail, ['']) -test('T7164', normal, compile_fail, ['']) +test('T7164', normal, compile_fail, ['-fdiagnostics-show-caret']) test('T7338', normal, compile_fail, ['']) test('T7338a', normal, compile_fail, ['']) test('T7454', normal, compile, ['']) View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/1f7ffaaadaa800daf07c937c690140b… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/1f7ffaaadaa800daf07c937c690140b… You're receiving this email because of your account on gitlab.haskell.org.

1 0

[Git][ghc/ghc][wip/mangoiv/ghc-9.12-bp] 3 commits: packaging: regenerate llvm-targets
by Magnus (＠MangoIV) 28 May '26

28 May '26

Magnus pushed to branch wip/mangoiv/ghc-9.12-bp at Glasgow Haskell Compiler / GHC Commits: 6b497eee by mangoiv at 2026-05-27T21:25:30+02:00 packaging: regenerate llvm-targets regenerate llvm-targets using utils/llvm-targets/gen-data-layout.sh and clang version 19.1.7 Target: x86_64-unknown-linux-gnu - - - - - e246c44d by Cheng Shao at 2026-05-27T21:25:30+02:00 ghc-internal: update to unicode 17.0.0 This commit updates the generated code in ghc-internal to match unicode 17.0.0. (cherry picked from commit 7077c9f76ebadedefd763078e7f7c42201b8a4b4) - - - - - 007b39e6 by Cheng Shao at 2026-05-27T21:25:30+02:00 wasm: ensure post-linker output is synchronous ESM This patch fixes wasm backend's post-linker output script to ensure it's synchronous ESM and doesn't use top-level await, which doesn't work in ServiceWorkers. Fixes #27257. (cherry picked from commit cccf45da01988801b0c97f7e7ed0160191980226) - - - - - 13 changed files: - + changelog.d/wasm-fix-serviceworker - libraries/base/tests/unicode002.stdout - libraries/base/tests/unicode003.stdout - libraries/ghc-internal/src/GHC/Internal/Unicode/Char/DerivedCoreProperties.hs - libraries/ghc-internal/src/GHC/Internal/Unicode/Char/UnicodeData/GeneralCategory.hs - libraries/ghc-internal/src/GHC/Internal/Unicode/Char/UnicodeData/SimpleLowerCaseMapping.hs - libraries/ghc-internal/src/GHC/Internal/Unicode/Char/UnicodeData/SimpleTitleCaseMapping.hs - libraries/ghc-internal/src/GHC/Internal/Unicode/Char/UnicodeData/SimpleUpperCaseMapping.hs - libraries/ghc-internal/src/GHC/Internal/Unicode/Version.hs - libraries/ghc-internal/tools/ucd2haskell/ucd.sh - libraries/ghc-internal/tools/ucd2haskell/unicode_version - llvm-targets - utils/jsffi/prelude.mjs The diff was not included because it is too large. View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f2e2876064cd043bc850fec297b1ca… -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f2e2876064cd043bc850fec297b1ca… You're receiving this email because of your account on gitlab.haskell.org.

1 0