- Jul 02, 2021
-
-
Fix #20066
-
I discovered that GHC.Core.Unify.bindTv was getting arity 2, rather than 3, in one of my builds. In HEAD it does get the right arity, but only because CallArity (just) manages to spot it. In my situation it (just) failed to discover this. Best to make it robust, which this patch does. See Note [INLINE pragmas and (>>)] in GHC.Utils.Monad. There a bunch of other modules that probably should have the same treatment: GHC.CmmToAsm.Reg.Linear.State GHC.Tc.Solver.Monad GHC.Tc.Solver.Rewrite GHC.Utils.Monad.State.Lazy GHC.Utils.Monad.State.Strict but doing so is not part of this patch
-
- Jul 01, 2021
-
-
The test case `print036` was marked `broken` by #9046. Issue #9046 is a duplicate of #12449. However the test case `T12449` contains several test that are similar to those in `print036`. Hence test case `print036` is redundant and can be deleted.
-
-
We should have fixed clangs mess now.
-
clang's cpp injects spaces prior to #!/.
-
-
by accepting the current state of metrics (and the NCG is new, so this seems prudent to do), we can require aarch64-linux (ncg) to build without permitting failure. Metric Increase: T13035 T13719 T14697 T1969 T9203 T9872a T9872b T9872c T9872d T9961 WWRec haddock.Cabal haddock.base parsing001 -
Signed-off-by:Emily Martins <emily.flakeheart@gmail.com>
-
Fixes #20042 Signed-off-by:
Hécate Moonlight <hecate@glitchbra.in>
-
Use DiagOpts for diagnostic options instead of directly querying DynFlags (#17957). Surprising performance improvements on CI: T4801(normal) ghc/alloc 313236344.0 306515216.0 -2.1% GOOD T9961(normal) ghc/alloc 384502736.0 380584384.0 -1.0% GOOD ManyAlternatives(normal) ghc/alloc 797356128.0 786644928.0 -1.3% ManyConstructors(normal) ghc/alloc 4389732432.0 4317740880.0 -1.6% T783(normal) ghc/alloc 408142680.0 402812176.0 -1.3% Metric Decrease: T4801 T9961 T783 ManyAlternatives ManyConstructors Bump haddock submodule
-
-
-
Avoid an other regression.
-
- Jun 29, 2021
-
-
Part way to #20013
-
-
In #19050, we identified several ways in which we could make more illegal states irrepresentable. This patch introduces a few representation changes around `Demand` and `Card` with a better and earlier-failing API exported through pattern synonyms. Specifically, 1. The old enum definition of `Card` led to severely bloated code of operations on it. I switched to a bit vector representation; much nicer overall IMO. See Note [Bit vector representation for Card]. Most of the gripes with the old representation were related to where which kind of `Card` was allowed and the fact that it doesn't make sense for an absent or bottoming demand to carry a `SubDemand` that describes an evaluation context that is never realised. 2. So I refactored the `Demand` representation so that it has two new data constructors for `AbsDmd` and `BotDmd`. The old `(:*)` data constructor becomes a pattern synonym which expands absent demands as needed, so that it still forms a complete match and a versatile builder. The new `Demand` data constructor now carries a `CardNonAbs` and only occurs in a very limited number of internal call sites. 3. Wherever a full-blown `Card` might end up in a `CardNonAbs` field (like that of `D` or `Call`), I assert the consistency. When the smart builder of `(:*)` is called with an absent `Card`, I assert that the `SubDemand` is the same that we would expand to in the matcher. 4. `Poly` now takes a `CardNonOnce` and encodes the previously noticed invariant that we never produce `Poly C_11` or `Poly C_01`. I made sure that we never construct a `Poly` with `C_11` or `C_01`. Fixes #19050. We lose a tiny bit of anal perf overall, probably because the new `Demand` definition can't be unboxed. The biggest loser is WWRec, where allocations go from 16MB to 26MB in DmdAnal, making up for a total increase of (merely) 1.6%. It's all within acceptance thresholds. There are even two ghc/alloc metric decreases. T11545 decreases by *67%*! Metric Decrease: T11545 T18304
-
- Jun 28, 2021
-
-
This is a follow-up to #19992, which fixes the type and strictness signature for `fork#`. The `forkOn#` primop also needs analogous changes, which this patch accomplishes.
-
This commit tries to untangle the zoo of diagnostic-related functions in `Tc.Utils.Monad` so that we can have the interfaces mentions only `TcRnMessage`s while we push the creation of these messages upstream. It also ports TcRnMessage diagnostics to use the new API, in particular this commit switch to use TcRnMessage in the external interfaces of the diagnostic functions, and port the old SDoc to be wrapped into TcRnUnknownMessage.
-
-
Update submodule libffi-tarballs to upstream commit 4f9e20a. Remove C compiler flags that suppress warnings in the RTS. Those warnings have been fixed by libffi upstream. Fixes #19885
-
- Jun 27, 2021
-
-
This reverts commit d1f59540. This commit breaks the build of unordered-containers ``` [3 of 9] Compiling Data.HashMap.Internal.Array ( Data/HashMap/Internal/Array.hs, dist/build/Data/HashMap/Internal/Array.o, dist/build/Data/HashMap/Internal/Array.dyn_o ) *** Parser [Data.HashMap.Internal.Array]: Parser [Data.HashMap.Internal.Array]: alloc=21043544 time=13.621 *** Renamer/typechecker [Data.HashMap.Internal.Array]: Renamer/typechecker [Data.HashMap.Internal.Array]: alloc=151218672 time=187.083 *** Desugar [Data.HashMap.Internal.Array]: ghc: panic! (the 'impossible' happened) GHC version 9.3.20210625: expectJust splitFunTy CallStack (from HasCallStack): error, called at compiler/GHC/Data/Maybe.hs:68:27 in ghc:GHC.Data.Maybe expectJust, called at compiler/GHC/Core/Type.hs:1247:14 in ghc:GHC.Core.Type ``` Revert containers submodule update
-
In a sequel of #19414, I wrote a script that measures min and max allocation bounds of T12545 based on randomly modifying -dunique-increment. I got a spread of as much as 4.8%. But instead of widening the acceptance window further (to 5%), I committed the script as part of this commit, so that false positive increases can easily be diagnosed by comparing min and max bounds to HEAD. Indeed, for !5814 we have seen T12545 go from -0.3% to 3.3% after a rebase. I made sure that the min and max bounds actually stayed the same. In the future, this kind of check can very easily be done in a matter of a minute. Maybe we should increase the acceptance threshold if we need to check often (leave a comment on #19414 if you had to check), but I've not been bitten by it for half a year, which seems OK. Metric Increase: T12545 -
Part of fixing #19766 required the emission of `LitRubbish` as absent filler in places where we used `absentError` before. In WWRec we have the situation that such bindings occur in the argument to functions. With `LitRubbish` we inlined those functions, because 1. The absent binding was regarded as ConLike. So I fixed `exprIsHNFLike` to respond `False` to `LitRubbish`. 2. The other source of inlining was that after inlining such an absent binding, `LitRubbish` itself was regarded `ValueArg` by `interestingArg`, leading to more inlining. It now responds `TrivArg` to `LitRubbish`. Fixes #20035. There's one slight 1.6% ghc/alloc regression left in T15164 that is due to an additional specialisation `$s$cget`. I've no idea why that happens; the Core output before is identical and has the call site that we specialise for. Metric Decrease: WWRec
-
The only item left in #17819. Fixes #17819.
-
In !5814 (comment 355144), Simon noted that `mkWWstr` and `mkWWcpr` could generate fewer let bindings and be implemented less indirectly by returning the rebuilt expressions directly, e.g. instead of ``` f :: (Int, Int) -> Int f (x, y) = x+y ==> f :: (Int, Int) -> Int f p = case p of (x, y) -> case x of I# x' -> case y of I# y' -> case $wf x' y' of r' -> let r = I# r' -- immediately returned in r f :: Int# -> Int# -> Int# $wf x' y' = let x = I# x' in -- only used in p let y = I# y' in -- only used in p let p = (x, y) in -- only used in the App below case (\(x,y) -> x+y) p of I# r' -> r' ``` we know generate ``` f :: (Int, Int) -> Int f p = case p of (x, y) -> case x of I# x' -> case y of I# y' -> case $wf x' y' of r' -> I# r' -- 1 fewer let f :: Int# -> Int# -> Int# $wf x' y' = case (\(x,y) -> x+y) (I# x, I# y) of I# r' -> -- 3 fewer lets r' ``` Which is much nicer and makes it easier to comprehend the output of worker-wrapper pre-Simplification as well as puts less strain on the Simplifier. I had to drop support for #18983, but we found that it's broken anyway. Simon is working on a patch that provides a bit more justification.
-
`mkWWargs`'s job was pushing casts inwards and doing eta expansion to match the arity with the number of argument demands we w/w for. Nowadays, we use the Simplifier to eta expand to arity. In fact, in recent years we have even seen the eta expansion done by w/w as harmful, see Note [Don't eta expand in w/w]. If a function hasn't enough manifest lambdas, don't w/w it! What purpose does `mkWWargs` serve in this world? Not a great one, it turns out! I could remove it by pulling some important bits, notably Note [Freshen WW arguments] and Note [Join points and beta-redexes]. Result: We reuse the freshened binder names of the wrapper in the worker where possible (see testuite changes), much nicer! In order to avoid scoping errors due to lambda-bound unfoldings in worker arguments, we zap those unfoldings now. In doing so, we fix #19766. Fixes #19874.
-
Strong loop-breakers never inline, INLINE pragma or not. Hence they should be treated as if there was no INLINE pragma on them. Also not doing Cast W/W for INLINE strong loop-breakers will trip up Strictness W/W, because it treats them as if there was no INLINE pragma. Subsequently, that will lead to a panic once Strictness W/W will no longer do eta-expansion, as we discovered while implementing !5814. I also renamed to `unfoldingInfo` to `realUnfoldingInfo` and redefined `unfoldingInfo` to zap the unfolding it returns in case of a strong loop-breaker. Now the naming and semantics is symmetrical to `idUnfolding`/`realIdUnfolding`. Now there was no more reason for `hasInlineUnfolding` to operate on `Id`, because the zapping of strong loop-breaker unfoldings moved from `idUnfolding` to `unfoldingInfo`, so I refactored it to take `IdInfo` and call it both from the Simplifier and WorkWrap, making it utterly clear that both checks are equivalent.
-
Previously, when using Typeable in a quantified constraint, GHC would complain that user-specified instances of Typeable aren't allowed. This was because checking for SigmaCtxt was missing from a check for whether an instance head is a hand-written binding. Fixes #20033
-
-
-
Previously we branched unnecessarily on IF_NONMOVING_WRITE_BARRIER_ENABLED on every trip through the array barrier push loop.
-
Previously the code generator's logic for invoking the nonmoving write barrier was inconsistent with the write barrier itself. Namely, the code generator treated the header size argument as being in words whereas the barrier expected bytes. This was the cause of #19715. Fixes #19715.
-
- Jun 25, 2021
-
-
When TemplateHaskellQuotes is enabled, we also generate programs which mention symbols from the template-haskell module. So that package is added conditionally if the extension is turned on. We should really do the same for other wired-in packages: * base * ghc-bignum * ghc-prim * rts When we link an executable, we must also link against these libraries. In accordance with every other package, these dependencies should be added into the direct dependencies for a module automatically and end up in the interface file to record the fact the object file was created by linking against these packages. Unfortunately it is not so easy to work out when symbols from each of these libraries ends up in the generated program. You might think that `base` would always be used but the `ghc-prim` package doesn't depend on `base`, so you have to be a bit careful and this futher enhancement is left to a future patch.
-
This fixes an error message regression.
-
-
-
Fixed in 25977ab5 Fixes #17481
-
This test was fixed by 25977ab5 Fixes #18330
-
-
