summaryrefslogtreecommitdiff
path: root/compiler
Commit message (Collapse)AuthorAgeFilesLines
* Eta expand un-saturated primopsBen Gamari2020-05-295-48/+87
| | | | | | | | | | | Now since we no longer try to predict CAFfyness we have no need for the solution to #16846. Eta expanding unsaturated primop applications is conceptually simpler, especially in the presence of levity polymorphism. This essentially reverts cac8dc9f51e31e4c0a6cd9bc302f7e1bc7c03beb, as suggested in #18079. Closes #18079.
* Allow simplification through runRW#Ben Gamari2020-05-296-40/+231
| | | | | | | | | | | | | | | | | | | | | | | | Because runRW# inlines so late, we were previously able to do very little simplification across it. For instance, given even a simple program like case runRW# (\s -> let n = I# 42# in n) of I# n# -> f n# we previously had no way to avoid the allocation of the I#. This patch allows the simplifier to push strict contexts into the continuation of a runRW# application, as explained in in Note [Simplification of runRW#] in GHC.CoreToStg.Prep. Fixes #15127. Metric Increase: T9961 Metric Decrease: ManyConstructors Co-Authored-By: Simon Peyton-Jone <simonpj@microsoft.com>
* Do not float join points in exprIsConApp_maybeSimon Peyton Jones2020-05-291-0/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | | | We hvae been making exprIsConApp_maybe cleverer in recent times: commit b78cc64e923716ac0512c299f42d4d0012306c05 Date: Thu Nov 15 17:14:31 2018 +0100 Make constructor wrappers inline only during the final phase commit 7833cf407d1f608bebb1d38bb99d3035d8d735e6 Date: Thu Jan 24 17:58:50 2019 +0100 Look through newtype wrappers (Trac #16254) commit c25b135ff5b9c69a90df0ccf51b04952c2dc6ee1 Date: Thu Feb 21 12:03:22 2019 +0000 Fix exprIsConApp_maybe But alas there was still a bug, now immortalised in Note [Don't float join points] in SimpleOpt. It's quite hard to trigger because it requires a dead join point, but it came up when compiling Cabal Cabal.Distribution.Fields.Lexer.hs, when working on !3113. Happily, the fix is extremly easy. Finding the bug was not so easy.
* Make Lint check return type of a join pointSimon Peyton Jones2020-05-291-7/+31
| | | | | | | | | Consider join x = rhs in body It's important that the type of 'rhs' is the same as the type of 'body', but Lint wasn't checking that invariant. Now it does! This was exposed by investigation into !3113.
* CoreToStg: Add Outputable ArgInfo instanceBen Gamari2020-05-291-1/+7
|
* Fix "build/elem" RULE.Andreas Klebinger2020-05-292-23/+53
| | | | | | | | | | | | | | | | | | | | | | | | An redundant constraint prevented the rule from matching. Fixing this allows a call to elem on a known list to be translated into a series of equality checks, and eventually a simple case expression. Surprisingly this seems to regress elem for strings. To avoid this we now also allow foldrCString to inline and add an UTF8 variant. This results in elem being compiled to a tight non-allocating loop over the primitive string literal which performs a linear search. In the process this commit adds UTF8 variants for some of the functions in GHC.CString. This is required to make this work for both ASCII and UTF8 strings. There are also small tweaks to the CString related rules. We now allow ourselfes the luxury to compare the folding function via eqExpr, which helps to ensure the rule fires before we inline foldrCString*. Together with a few changes to allow matching on both the UTF8 and ASCII variants of the CString functions.
* Rip out CmmStackInfo(updfr_space)Ben Gamari2020-05-285-14/+6
| | | | | As noted in #18232, this field is currently completely unused and moreover doesn't have a clear meaning.
* Ticky-ticky: Record DataCon name in ticker nameBen Gamari2020-05-282-4/+7
| | | | | This makes it significantly easier to spot the nature of allocations regressions and comes at a reasonably low cost.
* DmdAnal: Recognise precise exceptions from case alternatives (#18086)Sebastian Graf2020-05-281-8/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Consider ```hs m :: IO () m = do putStrLn "foo" error "bar" ``` `m` (from #18086) always throws a (precise or imprecise) exception or diverges. Yet demand analysis infers `<L,A>` as demand signature instead of `<L,A>x` for it. That's because the demand analyser sees `putStrLn` occuring in a case scrutinee and decides that it has to `deferAfterPreciseException`, because `putStrLn` throws a precise exception on some control flow paths. This will mask the `botDiv` `Divergence`of the single case alt containing `error` to `topDiv`. Since `putStrLn` has `topDiv` itself, the final `Divergence` is `topDiv`. This is easily fixed: `deferAfterPreciseException` works by `lub`ing with the demand type of a virtual case branch denoting the precise exceptional control flow. We used `nopDmdType` before, but we can be more precise and use `exnDmdType`, which is `nopDmdType` with `exnDiv`. Now the `Divergence` from the case alt will degrade `botDiv` to `exnDiv` instead of `topDiv`, which combines with the result from the scrutinee to `exnDiv`, and all is well. Fixes #18086.
* FloatOut: Only eta-expand dead-end RHS if arity will increase (#18231)Sebastian Graf2020-05-282-3/+7
| | | | | | | | Otherwise we risk turning trivial RHS into non-trivial RHS, introducing unnecessary bindings in the next Simplifier run, resulting in more churn. Fixes #18231.
* Fix #18071Xavier Denis2020-05-281-26/+53
| | | | | | | | Run the core linter on candidate instances to ensure they are well-kinded. Better handle quantified constraints by using a CtWanted to avoid having unsolved constraints thrown away at the end by the solver.
* Avoid unnecessary allocations due to tracing utilitiesBen Gamari2020-05-287-22/+80
| | | | | | | | | | | | | | | | | | | | While ticky-profiling the typechecker I noticed that hundreds of millions of SDocs are being allocated just in case -ddump-*-trace is enabled. This is awful. We avoid this by ensuring that the dump flag check is inlined into the call site, ensuring that the tracing document needn't be allocated unless it's actually needed. See Note [INLINE conditional tracing utilities] for details. Fixes #18168. Metric Decrease: T9961 haddock.Cabal haddock.base haddock.compiler
* GHC.Core.Unfold: Refactor traceInlineBen Gamari2020-05-281-9/+10
| | | | | This reduces duplication as well as fixes a bug wherein -dinlining-check would override -ddump-inlinings. Moreover, the new variant
* Make `identifier` parse unparenthesized `->` (#18060)Joshua Price2020-05-271-0/+2
|
* Enhance Note [About units] for BackpackSylvain Henry2020-05-261-234/+331
|
* Make WorkWrap.Lib.isWorkerSmallEnough aware of the old aritySebastian Graf2020-05-263-8/+12
| | | | | | | | | | We should allow a wrapper with up to 82 parameters when the original function had 82 parameters to begin with. I verified that this made no difference on NoFib, but then again it doesn't use huge records... Fixes #18122.
* Add info about typeclass evidence to .hie filesZubin Duggal2020-05-265-218/+663
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | See `testsuite/tests/hiefile/should_run/HieQueries.hs` and `testsuite/tests/hiefile/should_run/HieQueries.stdout` for an example of this We add two new fields, `EvidenceVarBind` and `EvidenceVarUse` to the `ContextInfo` associated with an Identifier. These are associated with the appropriate identifiers for the evidence variables collected when we come across `HsWrappers`, `TcEvBinds` and `IPBinds` while traversing the AST. Instance dictionary and superclass selector dictionaries from `tcg_insts` and classes defined in `tcg_tcs` are also recorded in the AST as originating from their definition span This allows us to save a complete picture of the evidence constructed by the constraint solver, and will let us report this to the user, enabling features like going to the instance definition from the invocation of a class method(or any other method taking a constraint) and finding all usages of a particular instance. Additionally, - Mark NodeInfo with an origin so we can differentiate between bindings origininating in the source vs those in ghc - Along with typeclass evidence info, also include information on Implicit Parameters - Add a few utility functions to HieUtils in order to query the new info Updates haddock submodule
* Coverage: Factor out addMixEntryBen Gamari2020-05-251-26/+20
|
* Coverage: Don't produce ModBreaks if not HscInterpretedBen Gamari2020-05-251-6/+10
| | | | | emptyModBreaks contains a bottom and consequently it's important that we don't use it unless necessary.
* Coverage: Make ccIndices strictBen Gamari2020-05-251-1/+1
| | | | This just seems like a good idea.
* Coverage: Make tickBoxCount strictBen Gamari2020-05-251-1/+1
| | | | This could otherwise easily cause a leak of (+) thunks.
* Coverage: Drop redundant ad-hoc boot module checkBen Gamari2020-05-251-2/+1
| | | | | | | | To determine whether the module is a boot module Coverage.addTicksToBinds was checking for a `boot` suffix in the module source filename. This is quite ad-hoc and shouldn't be necessary; the callsite in `deSugar` already checks that the module isn't a boot module.
* Make Unicode brackets opening/closing tokens (#18225)Joshua Price2020-05-241-4/+8
| | | | | | | The tokens `[|`, `|]`, `(|`, and `|)` are opening/closing tokens as described in GHC Proposal #229. This commit makes the unicode variants (`⟦`, `⟧`, `⦇`, and `⦈`) act the same as their ASCII counterparts.
* Remove unused hs-boot fileMatthew Pickering2020-05-242-8/+1
|
* Add orderingTyCon to wiredInTyCons (#18185)Ryan Scott2020-05-242-6/+6
| | | | | | | | | | `Ordering` needs to be wired in for use in the built-in `CmpNat` and `CmpSymbol` type families, but somehow it was never added to the list of `wiredInTyCons`, leading to the various oddities observed in #18185. Easily fixed by moving `orderingTyCon` from `basicKnownKeyNames` to `wiredInTyCons`. Fixes #18185.
* Rename GHC.Hs.Types into GHC.Hs.TypeSylvain Henry2020-05-2421-32/+32
| | | | See discussion in https://gitlab.haskell.org/ghc/ghc/issues/13009#note_268610
* Rename GHC.Core.Arity into GHC.Core.Opt.AritySylvain Henry2020-05-2422-33/+33
|
* Move Config module into GHC.SettingsSylvain Henry2020-05-2410-15/+14
|
* Fix #18145 and also avoid needless work with implicit varsJohn Ericson2020-05-232-78/+93
| | | | | | | | | | | | - `forAllOrNothing` now is monadic, so we can trace whether we bind an explicit `forall` or not. - #18145 arose because the free vars calculation was needlessly complex. It is now greatly simplified. - Replaced some other implicit var code with `filterFreeVarsToBind`. Co-authored-by: Ryan Scott <ryan.gl.scott@gmail.com>
* simplCore: Ignore ticks in rule templatesBen Gamari2020-05-232-2/+30
| | | | | | | | | | | | This fixes #17619, where a tick snuck in to the template of a rule, resulting in a panic during rule matching. The tick in question was introduced via post-inlining, as discussed in `Note [Simplifying rules]`. The solution we decided upon was to simply ignore ticks in the rule template, as discussed in `Note [Tick annotations in RULE matching]`. Fixes #18162. Fixes #17619.
* Implement cstringLength# and FinalPtrAndrew Martin2020-05-236-8/+39
| | | | | | | | | | | | | | | | | | | | This function and its accompanying rule resolve issue #5218. A future PR to the bytestring library will make the internal Data.ByteString.Internal.unsafePackAddress compute string length with cstringLength#. This will improve the status quo because it is eligible for constant folding. Additionally, introduce a new data constructor to ForeignPtrContents named FinalPtr. This additional data constructor, when used in the IsString instance for ByteString, leads to more Core-to-Core optimization opportunities, fewer runtime allocations, and smaller binaries. Also, this commit re-exports all the functions from GHC.CString (including cstringLength#) in GHC.Exts. It also adds a new test driver. This test driver is used to perform substring matches on Core that is dumped after all the simplifier passes. In this commit, it is used to check that constant folding of cstringLength# works.
* Move isDynLinkName into GHC.Types.NameSylvain Henry2020-05-214-38/+36
| | | | It doesn't belong into GHC.Unit.State
* Don't variable-length encode magic iface constant.Andreas Klebinger2020-05-212-66/+80
| | | | | | | | | | | | | | | We changed to use variable length encodings for many types by default, including Word32. This makes sense for numbers but not when Word32 is meant to represent four bytes. I added a FixedLengthEncoding newtype to Binary who's instances interpret their argument as a collection of bytes instead of a number. We then use this when writing/reading magic numbers to the iface file. I also took the libery to remove the dummy iface field. This fixes #18180.
* Fix wording in primops documentation to reflect the correct reasoning:Alexey Kuleshevich2020-05-211-5/+5
| | | | | | | | | * Besides resizing functions, shrinking ones also mutate the size of a mutable array and because of those two `sizeofMutabeByteArray` and `sizeofSmallMutableArray` are now deprecated * Change reference in documentation to the newer functions `getSizeof*` instead of `sizeof*` for shrinking functions * Fix incorrect mention of "byte" instead of "small"
* Fix spelling mistakes and typosbuggymcbugfix2020-05-212-3/+3
|
* Use braces with do in `SplicePat` case for consistencyJohn Ericson2020-05-211-2/+2
|
* Tiny cleaup eta-reduce away a function argumentJohn Ericson2020-05-211-2/+2
| | | | In GHC, not in the code being compiled!
* Put `PatEnv` first in `GHC.Tc.Gen.Pat.Checker`John Ericson2020-05-211-62/+65
|
* More judiciously panic in `ts_pat`John Ericson2020-05-211-3/+3
|
* Use `Checker` for `tc_lpat` and `tc_lpats`John Ericson2020-05-211-32/+25
|
* Use `Checker` for `tc_pat`John Ericson2020-05-211-56/+58
|
* llvmGen: Consider Relocatable read-only data as not constantReferences: #18137Tuan Le2020-05-213-13/+34
|
* NCG: Codelayout: Distinguish conditional and other branches.Andreas Klebinger2020-05-211-3/+54
| | | | | | | | | | | | | | In #18053 we ended up with a suboptimal code layout because the code layout algorithm didn't distinguish between conditional and unconditional control flow. We can completely eliminate unconditional control flow instructions by placing blocks next to each other, not so much for conditionals. In terms of implementation we simply give conditional branches less weight before computing the layout. Fixes #18053
* Refactor linear reg alloc to remember past assignments.Andreas Klebinger2020-05-2114-44/+128
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When assigning registers we now first try registers we assigned to in the past, instead of picking the "first" one. This is in extremely helpful when dealing with loops for which variables are dead for part of the loop. This is important for patterns like this: foo = arg1 loop: use(foo) ... foo = getVal() goto loop; There we: * assign foo to the register of arg1. * use foo, it's dead after this use as it's overwritten after. * do other things. * look for a register to put foo in. If we pick an arbitrary one it might differ from the register the start of the loop expect's foo to be in. To fix this we simply look for past register assignments for the given variable. If we find one and the register is free we use that register. This reduces the need for fixup blocks which match the register assignment between blocks. In the example above between the end and the head of the loop. This patch also moves branch weight estimation ahead of register allocation and adds a flag to control it (cmm-static-pred). * It means the linear allocator is more likely to assign the hotter code paths first. * If it assign these first we are: + Less likely to spill on the hot path. + Less likely to introduce fixup blocks on the hot path. These two measure combined are surprisingly effective. Based on nofib we get in the mean: * -0.9% instructions executed * -0.1% reads/writes * -0.2% code size. * -0.1% compiler allocations. * -0.9% compile time. * -0.8% runtime. Most of the benefits are simply a result of removing redundant moves and spills. Reduced compiler allocations likely are the result of less code being generated. (The added lookup is mostly non-allocating).
* Lint should say when it is checking a ruleBen Price2020-05-211-1/+5
| | | | | | | | | | | | | | | | It is rather confusing that when lint finds an error in a rule attached to a binder, it reports the error as in the RHS, not the rule: ... In the RHS of foo We add a clarifying line: ... In the RHS of foo In a rule attached to foo The implication that the rule lives inside the RHS is a bit odd, but this niggle is already present for unfoldings, whose pattern we are following.
* Explicit SpecificityGert-Jan Bottu2020-05-2157-644/+1052
| | | | | | | | | | | | | | | | | | | | Implementation for Ticket #16393. Explicit specificity allows users to manually create inferred type variables, by marking them with braces. This way, the user determines which variables can be instantiated through visible type application. The additional syntax is included in the parser, allowing users to write braces in type variable binders (type signatures, data constructors etc). This information is passed along through the renamer and verified in the type checker. The AST for type variable binders, data constructors, pattern synonyms, partial signatures and Template Haskell has been updated to include the specificity of type variables. Minor notes: - Bumps haddock submodule - Disables pattern match checking in GHC.Iface.Type with GHC 8.8
* Remove duplicate Note [When to print foralls] in GHC.Core.TyCo.Pprwip/duplicate-forall-printing-notesRyan Scott2020-05-171-19/+2
| | | | | | | | | | | There are two different Notes named `[When to print foralls]`. The most up-to-date one is in `GHC.Iface.Type`, but there is a second one in `GHC.Core.TyCo.Ppr`. The latter is less up-to-date, as it was written before GHC switched over to using ifaces to pretty-print types. I decided to just remove the latter and replace it with a reference to the former. [ci skip]
* GHC.Cmm.Opt: Handle MO_XX_ConvBen Gamari2020-05-151-0/+2
| | | | | | | | | This MachOp was introduced by 2c959a1894311e59cd2fd469c1967491c1e488f3 but a wildcard match in cmmMachOpFoldM hid the fact that it wasn't handled. Ideally we would eliminate the match but this appears to be a larger task. Fixes #18141.
* DmdAnal: Improve handling of precise exceptionsSebastian Graf2020-05-1522-421/+533
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This patch does two things: Fix possible unsoundness in what was called the "IO hack" and implement part 2.1 of the "fixing precise exceptions" plan in https://gitlab.haskell.org/ghc/ghc/wikis/fixing-precise-exceptions, which, in combination with !2956, supersedes !3014 and !2525. **IO hack** The "IO hack" (which is a fallback to preserve precise exceptions semantics and thus soundness, rather than some smart thing that increases precision) is called `exprMayThrowPreciseException` now. I came up with two testcases exemplifying possible unsoundness (if twisted enough) in the old approach: - `T13380d`: Demonstrating unsoundness of the "IO hack" when resorting to manual state token threading and direct use of primops. More details below. - `T13380e`: Demonstrating unsoundness of the "IO hack" when we have Nested CPR. Not currently relevant, as we don't have Nested CPR yet. - `T13380f`: Demonstrating unsoundness of the "IO hack" for safe FFI calls. Basically, the IO hack assumed that precise exceptions can only be thrown from a case scrutinee of type `(# State# RealWorld, _ #)`. I couldn't come up with a program using the `IO` abstraction that violates this assumption. But it's easy to do so via manual state token threading and direct use of primops, see `T13380d`. Also similar code might be generated by Nested CPR in the (hopefully not too) distant future, see `T13380e`. Hence, we now have a more careful test in `forcesRealWorld` that passes `T13380{d,e}` (and will hopefully be robust to Nested CPR). **Precise exceptions** In #13380 and #17676 we saw that we didn't preserve precise exception semantics in demand analysis. We fixed that with minimal changes in !2956, but that was terribly unprincipled. That unprincipledness resulted in a loss of precision, which is tracked by these new test cases: - `T13380b`: Regression in dead code elimination, because !2956 was too syntactic about `raiseIO#` - `T13380c`: No need to apply the "IO hack" when the IO action may not throw a precise exception (and the existing IO hack doesn't detect that) Fixing both issues in !3014 turned out to be too complicated and had the potential to regress in the future. Hence we decided to only fix `T13380b` and augment the `Divergence` lattice with a new middle-layer element, `ExnOrDiv`, which means either `Diverges` (, throws an imprecise exception) or throws a *precise* exception. See the wiki page on Step 2.1 for more implementational details: https://gitlab.haskell.org/ghc/ghc/wikis/fixing-precise-exceptions#dead-code-elimination-for-raiseio-with-isdeadenddiv-introducing-exnordiv-step-21
* IdInfo: Add reference to bitfield-packing ticketBen Gamari2020-05-141-1/+1
|
View Lorry Controller Status