Ben Gamari pushed to branch ghc-9.10 at Glasgow Haskell Compiler / GHC Commits: 5e26d065 by Ben Gamari at 2025-04-18T17:40:04-04:00 More release note cleanup - - - - - 2 changed files: - docs/users_guide/9.10.1-notes.rst - docs/users_guide/9.10.2-notes.rst Changes: ===================================== docs/users_guide/9.10.1-notes.rst ===================================== @@ -244,6 +244,9 @@ Compiler arguments. This may be preferred over :ghc-flag:`-fprof-late-overloaded` since it may reveal whether imported functions are called overloaded. +- The :ghc-flag:`-fkeep-auto-rules` flag was introduced, forcing GHC to keep auto generated + specialization rules. + JavaScript backend ~~~~~~~~~~~~~~~~~~ ===================================== docs/users_guide/9.10.2-notes.rst ===================================== @@ -11,11 +11,9 @@ for specific guidance on migrating programs to this release. Compiler ~~~~~~~~ -- Optimized lowering of BigNat# literals. ((:ghc-ticket:`23942`), !12179) +- Optimized lowering of ``BigNat#`` literals. (:ghc-ticket:`23942`) -- Fixed a bug where the jump shortcutting optimization could lead to unsoundness in pattern matching. ((:ghc-ticket:`24507`)) - -- Look for haskell symbols only in relevant libraries. ((:ghc-ticket:`23415`)) +- Look for Haskell symbols only in relevant libraries. (:ghc-ticket:`23415`) When dynamically loading a Haskell symbol (typical when running a splice or GHCi expression), before this change we would search for the symbol in @@ -24,100 +22,104 @@ Compiler many package dependencies) because the time to lookup the would be linear in the number of packages loaded. - In one measurement, this improved performance for executing a GHCi expression + In one measurement this improved performance for executing a GHCi expression from ~38 seconds down to ~2s. -- Fixed a bug that could cause a linker error when `TypeData` and `StrictData` - where used in combination. ((:ghc-ticket:`24620`)) +- Fixed a bug that could cause a linker error when ``TypeData`` and ``StrictData`` + where used in combination. (:ghc-ticket:`24620`) -- The flag :ghc-flag:`-fprof-late` when combined with `-prof` will no longer prevent top level - constructors from being statically allocated. +- Fixed an issue where :ghc-flag:`-Wmissing-home-modules` caused quadratic compile time increases. (:ghc-ticket:`24778`) - It used to be the case that we would add a cost centre for bindings like `foo = Just bar`. - This turned the binding into a caf that would allocate the constructor on first evaluation. +- Better type inference in certain situations. (:ghc-ticket:`24810`) - However without the cost centre `foo` can be allocated at compile time. This reduces code-bloat and - reduces overhead for short-running applications. +- Fixed a bug where the magic ``GHC.Magic.inline`` function sometimes failed to inline functions + if there were coercions involved. (:ghc-ticket:`24808`) - The tradeoff is that calling `whoCreated` on top level value definitions like `foo` will be less informative. +- Fixed a bug where the simplifier sometimes destroyed join points during float out. (:ghc-ticket:`24768`) -- IPE profiling - optimize the representation of some internal representations. Leading to smaller executables. ((:ghc-ticket:`24504`)) +- Fixed inability to re-export the ``Data.Tuple.MkSolo`` constructor (:ghc-ticket:`25182`) -- Fixed an issue where `-Wmissing-home-modules` caused quadratic compile time increases. ((:ghc-ticket:`24778`)) +- Fixed ``Language.Haskell.TH.mkName "FUN"`` (:ghc-ticket:`25174`) -- Fixed a crash when using `-prof` with `-forig-thunk-info`. ((:ghc-ticket:`24809`)) +- Fixed a bug that could case segfaults in GHCi when :ghc-flag:`-fworker-wrapper-cbv` was enabled. (:ghc-ticket:`24870`) -- Better type inference in certain situations. ((:ghc-ticket:`24810`)) +- Fixed a bug where the use of ``throw`` caused a loss of strictness (:ghc-ticket:`25066`). -- Fixed a bug where the magic `inline` function sometimes failed to inline functions - if there were coercions involved. ((:ghc-ticket:`24808`)) +- Fixed incorrect behaviour that could occur when using ``control0#/prompt#`` (:ghc-ticket:`25439`) by making `prompt#` lazy. -- Fixed a bug where the simplifier sometimes destroyed join points during float out. ((:ghc-ticket:`24768`)) +- Fixed a bug where GHC panicked during type checking. (:ghc-ticket:`25325`) -- PPC NCG: Fix sign hints in C calls. This bug caused segfaults when using FFI under some - circumstances. ((:ghc-ticket:`23034`)) +- Fix linearity for strict unit patterns in a local ``let`` bindings. (:ghc-ticket:`25428`) -- Fixed a bug that caused GHC to panic when using the aarch64 ncg and -fregs-graph - on certain programs. ((:ghc-ticket:`24941`)) +- Improve the performance of type family consistency checks. ((:ghc-ticket:`25554`)) -- The flag :ghc-flag:`-fkeep-auto-rules` that forces GHC to keep auto generated - specialization rules was added. It was actually added ghc-9.10.1 already but - mistakenly not mentioned in the 9.10.1 changelog. +- Fixed a bug where type checking failed to terminate. (:ghc-ticket:`25597`) -- Fixed re-exports of ``MkSolo`` (:ghc-ticket:`25182`) +- Added flags to control speculative evaluation: :ghc-flag:`-fspec-eval` and :ghc-flag:`-fspec-eval-dictfun`. This allows + users to work around cases where recent changes to speculative evaluation caused performance regressions. (:ghc-ticket:`25606`) -- Fixed the behavior of ``Language.Haskell.TH.mkName "FUN"`` (:ghc-ticket:`25174`) +Profiling +~~~~~~~~~ -- Various code generation improvements to the ARM backend. +- The flag :ghc-flag:`-fprof-late` when combined with :ghc-flag:`-prof` will no longer prevent top level + constructors from being statically allocated. -- Fixed a bug that could case segfaults in GHCi when `-fworker-wrapper-cbv` was enabled. ((:ghc-ticket:`24870`)) + It used to be the case that we would add a cost centre for bindings like ``foo = Just bar``. + This turned the binding into a constant applicative form that would allocate + the constructor on first evaluation. -- Fixed a correctness issues for Int64/Word64 on 32bit backends ((:ghc-ticket:`24893`)) + However without the cost-centre ``foo`` can be allocated at compile time. This reduces code-bloat and + reduces overhead for short-running applications. -- Various rts linker improvements: - - R_X86_64_PC64 relocation support on windows - - Use m32 allocator for sections when NEED_PLT - works around ((:ghc-ticket:`24432`)) - - AArch64 - Skip NONE relocations + The tradeoff is that calling ``GHC.Stack.whoCreated`` on top level value definitions like ``foo`` will be less informative. -- AArch64 ncg: Multiway branches are now lowered to jump tables for better performance. ((:ghc-ticket:`19912`)) +- Fixed a crash when using :ghc-flag:`-prof` with :ghc-flag:`-forig-thunk-info`. (:ghc-ticket:`24809`) -- Fixes related to better use of sse4 when using the llvm backend: - - Use +sse4.2 rather than +sse42, fixing ((:ghc-ticket:`25019`)), - - make SSE4.2 imply +popcnt, fixing ((:ghc-ticket:`25353`)). +- Optimize the representation of :ghc-flag:`info table provenance maps <-finfo-table-map>`, significantly reducing code size when this feature is in use. (:ghc-ticket:`24504`) -- Fixed a bug where the use of ``throw`` caused a loss of strictness ((:ghc-ticket:`25066`)). +Native code generator backend +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -- Fixed incorrect behaviour that could occur when using ``control0#/prompt#`` ((:ghc-ticket:`25439`)) by making `prompt#` lazy. +- Fixed a bug where the jump shortcutting optimization could lead to unsoundness in pattern matching. (:ghc-ticket:`24507`) -- Fixed a bug where ghc paniced during type checking. ((:ghc-ticket:`25325`)) +- PowerPC: Fix sign hints in C calls. This bug caused segfaults when using FFI under some + circumstances. (:ghc-ticket:`23034`) -- Fix linearity for strict unit patterns in a local let bindings. ((:ghc-ticket:`25428`)) +- AArch64: Various code generation performance improvements -- Improve the performance of type family consistency checks. ((:ghc-ticket:`25554`)) +- AArch64: Fixed a bug that caused GHC to panic when using :ghc-flag:`-fregs-graph` + on certain programs. (:ghc-ticket:`24941`) -- Fixed a bug where type checking failed to terminate. ((:ghc-ticket:`25597`)) +- AArch64: Multiway branches are now lowered to jump tables for better performance. (:ghc-ticket:`19912`) -- Fixed a bug where ``__GLASGOW_HASKELL_LLVM__`` was not defined even if `-fllvm` was set. ((:ghc-ticket:`25606`)) +- Fixed a correctness issues for ``Int64``/``Word64`` on 32-bit backends (:ghc-ticket:`24893`) + +LLVM backend +~~~~~~~~~~~~ + +- Fixes related to better use of SSE4: + - Use ``+sse4.2`` rather than ``+sse42``, fixing (:ghc-ticket:`25019`), + - make SSE4.2 imply ``+popcnt``, fixing (:ghc-ticket:`25353`). + +- Fixed a bug where ``__GLASGOW_HASKELL_LLVM__`` was not defined even if :ghc-flag:`-fllvm` was set. (:ghc-ticket:`25606`) -- Added flags to control speculative evaluation: ``-fspec-eval`` and ``-fspec-eval-dictfun``. This allows - users to work around cases where recent changes to speculative evaluation caused performance regressions. ((:ghc-ticket:`25606`)) JavaScript backend ~~~~~~~~~~~~~~~~~~ -- Changes for better support of the Google Closure Compiler ((:ghc-ticket:`24602`), !12330) +- Changes for better support of the Google Closure Compiler (:ghc-ticket:`24602`) -- JS Backend: support rubbish static literals +- Add support for rubbish static literals -- Fix h$withCStringOnHeap helper ((:ghc-ticket:`25288`)) +- Fix ``h$withCStringOnHeap`` helper (:ghc-ticket:`25288`) WebAssembly backend ~~~~~~~~~~~~~~~~~~~ -- wasm: use scheduler.postTask() for context switch when available +- Use ``scheduler.postTask()`` for context switch when available -- The in tree gmp library now is configured to produce faster and smaller code given - the current constraints of the wasm backend. (!12592) +- The in-tree gmp library now is configured to produce faster and smaller code given + the current constraints of the wasm backend. GHCi ~~~~ @@ -144,20 +146,29 @@ Runtime system Users who have fine-tuned the :rts-flag:`-F ⟨factor⟩`, :rts-flag:`-Fd ⟨factor⟩`, or :rts-flag:`-O ⟨size⟩` flags, and use the non-moving GC, should see if adjustments are needed in light of this change. -- Reduce fragmentation incurred by the nonmoving GC's segment allocator. In one application this reduced resident set size by 26%. See :ghc-ticket:`24150`. +- Reduce fragmentation incurred by the non-moving GC's segment allocator. In one application this reduced resident + set size by 26%. See :ghc-ticket:`24150`. -- Emit warning when -M < -H. +- Emit warning when :ghc-flag:`-M` is less than :ghc-flag:`-H`. - Fix a bug ((:ghc-ticket:`24672`)) causing programms on FreeBSD to occasionally terminate with the error: ``internal error: Ticker: read(timerfd) failed with Operation not supported and returned -1`` -- Fixed a bug ((:ghc-ticket:`25503`)) that sometimes caused the rts to terminate with ``internal error: alloc_adjustor_chunk: failed to allocate`` on windows. +- Fixed a bug ((:ghc-ticket:`25503`)) that sometimes caused the runtime to terminate with + ``internal error: alloc_adjustor_chunk: failed to allocate`` on windows. This primarily affect programs making use of the ``foreign import "wrapper"`` functionality. -- Fixed a bug in compacting gc that could result in program termination with ``update_fwd_large: unknown/strange object 38`` ((:ghc-ticket:`24791`)). +- Fixed a bug in compacting gc that could result in program termination with + ``update_fwd_large: unknown/strange object 38`` (:ghc-ticket:`24791`). - Fixed a bug where calling ``setNumCapabilities`` with a argument larger than 256 caused crashes. (:ghc-ticket:`25560`) +- Various runtime system linker improvements: + - ``R_X86_64_PC64`` relocation support on windows + - Use m32 allocator for sections when ``NEED_PLT`` - works around ((:ghc-ticket:`24432`)) + - AArch64 - Skip ``NONE`` relocations + + Build system and packaging ~~~~~~~~~~~~~~~~~~~~~~~~~~ @@ -167,9 +178,9 @@ Build system and packaging (From https://git.savannah.gnu.org/cgit/config.git/) - GHC now distinguishes more between the Cmm and C preprocessor. The new flags - `-pgmCmmP` and `-optCmmP` can be used to control + :ghc-flag:`-pgmCmmP` and :ghc-flag:`-optCmmP` can be used to control the Cmm preprocessor. C preprocessor flags will still be passed to the Cmm - preprocessor, expect for -g flags which have special handling. Fixes (:ghc-ticket:`24474`) + preprocessor, expect for :ghc-flag:`-g` flags which have special handling. Fixes (:ghc-ticket:`24474`) - Bindists: Fixed a bug in the Makefile which caused windows installs to duplicate files. ((:ghc-ticket:`24800`)) @@ -177,12 +188,12 @@ Build system and packaging - Fixed a bug where running ``Xelatex`` concurrently failed. (:ghc-ticket:`25564`) -- Fixed a bug where building ghc from source using ghc-9.8.4 failed with an error mentioning ``ghc_unique_counter64``. (:ghc-ticket:`25576`) +- Fixed a bug where building ghc from source using GHC 9.8.4 failed with an error mentioning ``ghc_unique_counter64`` (:ghc-ticket:`25576`) ``ghc`` library ~~~~~~~~~~~~~~~ -- Added some utility functions to `GHC.Data.SmallArray` +- Added some utility functions to ``GHC.Data.SmallArray`` ``ghc-heap`` library ~~~~~~~~~~~~~~~~~~~~ @@ -192,8 +203,9 @@ Build system and packaging ``ghc-experimental`` library ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -* Primops and other ghc extensions are now exported via ``GHC.PrimOps`` -* The library is now versioned according to the ghc version it shipped with. +- Primops and other GHC extensions are now exported via ``GHC.PrimOps`` + +- The library is now versioned according to the ghc version it shipped with. Included libraries @@ -202,3 +214,40 @@ Included libraries The package database provided with this distribution also contains a number of packages other than GHC itself. See the changelogs provided with these packages for further change information. + +.. ghc-package-list:: + + libraries/array/array.cabal: Dependency of ``ghc`` library + libraries/base/base.cabal: Core library + libraries/binary/binary.cabal: Dependency of ``ghc`` library + libraries/bytestring/bytestring.cabal: Dependency of ``ghc`` library + libraries/Cabal/Cabal/Cabal.cabal: Dependency of ``ghc-pkg`` utility + libraries/Cabal/Cabal-syntax/Cabal-syntax.cabal: Dependency of ``ghc-pkg`` utility + libraries/containers/containers/containers.cabal: Dependency of ``ghc`` library + libraries/deepseq/deepseq.cabal: Dependency of ``ghc`` library + libraries/directory/directory.cabal: Dependency of ``ghc`` library + libraries/exceptions/exceptions.cabal: Dependency of ``ghc`` and ``haskeline`` library + libraries/filepath/filepath.cabal: Dependency of ``ghc`` library + compiler/ghc.cabal: The compiler itself + libraries/ghci/ghci.cabal: The REPL interface + libraries/ghc-boot/ghc-boot.cabal: Internal compiler library + libraries/ghc-boot-th/ghc-boot-th.cabal: Internal compiler library + libraries/ghc-compact/ghc-compact.cabal: Core library + libraries/ghc-heap/ghc-heap.cabal: GHC heap-walking library + libraries/ghc-prim/ghc-prim.cabal: Core library + libraries/haskeline/haskeline.cabal: Dependency of ``ghci`` executable + libraries/hpc/hpc.cabal: Dependency of ``hpc`` executable + libraries/integer-gmp/integer-gmp.cabal: Core library + libraries/mtl/mtl.cabal: Dependency of ``Cabal`` library + libraries/parsec/parsec.cabal: Dependency of ``Cabal`` library + libraries/pretty/pretty.cabal: Dependency of ``ghc`` library + libraries/process/process.cabal: Dependency of ``ghc`` library + libraries/stm/stm.cabal: Dependency of ``haskeline`` library + libraries/template-haskell/template-haskell.cabal: Core library + libraries/terminfo/terminfo.cabal: Dependency of ``haskeline`` library + libraries/text/text.cabal: Dependency of ``Cabal`` library + libraries/time/time.cabal: Dependency of ``ghc`` library + libraries/transformers/transformers.cabal: Dependency of ``ghc`` library + libraries/unix/unix.cabal: Dependency of ``ghc`` library + libraries/Win32/Win32.cabal: Dependency of ``ghc`` library + libraries/xhtml/xhtml.cabal: Dependency of ``haddock`` executable View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/5e26d065451c849d266df6b7a0f02f60... -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/5e26d065451c849d266df6b7a0f02f60... You're receiving this email because of your account on gitlab.haskell.org.