October 2016 - ghc-tickets

[GHC] #12562: GHC panic on rebuild (idInfo r_XsTP)
by GHC 18 Oct '16

18 Oct '16

#12562: GHC panic on rebuild (idInfo r_XsTP) -------------------------------------+------------------------------------- Reporter: cocreature | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- First of all let me apologize for the terrible testcase. I tried to find a smaller one (and am still trying) but haven’t managed to do so. The good news is that at least for me it’s reproducible === Steps to reproduce === {{{ git clone 'git@github.com:bscarlet/llvm-general.git' -b llvm-3.8 cd llvm-general cabal new-build llvm-general curl -s https://gist.githubusercontent.com/cocreature/c6807d7906756a2d58b8fe774141b… > out.patch git apply out.patch cabal new-build llvm-general }}} The patch just adds a newline somewhere to create a rebuild. I’ve seen this panic for completely different changes (all in this project) so I don’t think the change matters. I’ve also seen this panic with stack, cabal new-build and cabal build (using sandboxes). === Output === {{{ In order, the following will be built (use -v for more details): llvm-general-3.8.0.0 Preprocessing library llvm-general-3.8.0.0... [84 of 92] Compiling LLVM.General.Internal.Module ( src/LLVM/General/Internal/Module.hs, /home/moritz/tmp/llvm-general/dist- newstyle/build/llvm-general-3.8.0.0/build/LLVM/General/Internal/Module.o ) [85 of 92] Compiling LLVM.General.Module ( src/LLVM/General/Module.hs, /home/moritz/tmp/llvm-general/dist-newstyle/build/llvm- general-3.8.0.0/build/LLVM/General/Module.o ) [LLVM.General.Internal.Module changed] [89 of 92] Compiling LLVM.General.Internal.PassManager ( src/LLVM/General/Internal/PassManager.hs, /home/moritz/tmp/llvm-general /dist-newstyle/build/llvm- general-3.8.0.0/build/LLVM/General/Internal/PassManager.o ) [TH] ghc: panic! (the 'impossible' happened) (GHC version 8.0.1 for x86_64-unknown-linux): idInfo r_XsTP Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} === Comments === In this case this can be resolved by running cabal new-build again. However I’ve also had cases where this didn’t fix it and I had to blow away dist-newstyle. I haven’t yet managed to find a reproducible testcase for the latter. Matthew Pickering mentioned that he encountered this (or a similar) bug while building GHC. I tried core-lint but that didn’t help. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12562> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 17

[GHC] #12589: GHC panic with defer-typed-holes
by GHC 18 Oct '16

18 Oct '16

#12589: GHC panic with defer-typed-holes -------------------------------------+------------------------------------- Reporter: Iceland_jack | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Does this still happen in HEAD? {{{#!hs import qualified GHC.Generics as GHC import Generics.SOP import Generics.SOP.TH import Data.Proxy gminbound' :: (Generic a, Code a ~ (xs:xss), All2 Bounded (Code a)) => a gminbound' = minBound & I & hcpure (Proxy @Bounded) & apInjs_POP & head & to }}} Fails but doesn't crash with {{{ $ ghci -ignore-dot-ghci -XTypeApplications -XTypeOperators -XDataKinds -XGADTs -XFlexibleContexts /tmp/tX96.hs GHCi, version 8.0.1: http://www.haskell.org/ghc/ :? for help [1 of 1] Compiling Main ( /tmp/tX96.hs, interpreted ) /tmp/tX96.hs:8:3: error: Variable not in scope: (&) :: t4 -> (a0 -> I a0) -> t3 /tmp/tX96.hs:9:3: error: Variable not in scope: (&) :: t3 -> t5 -> t2 /tmp/tX96.hs:10:3: error: Variable not in scope: (&) :: t2 -> (POP f1 xss0 -> [SOP f1 xss0]) -> t1 /tmp/tX96.hs:11:3: error: Variable not in scope: (&) :: t1 -> ([a1] -> a1) -> t0 /tmp/tX96.hs:12:3: error: Variable not in scope: (&) :: t0 -> (Rep a2 -> a2) -> a Failed, modules loaded: none. Prelude> }}} but crashes with `-fdefer-typed-holes` {{{ $ ghci -ignore-dot-ghci -fdefer-typed-holes -XTypeApplications -XTypeOperators -XDataKinds -XGADTs -XFlexibleContexts /tmp/tX96.hs GHCi, version 8.0.1: http://www.haskell.org/ghc/ :? for help [1 of 1] Compiling Main ( /tmp/tX96.hs, interpreted ) /tmp/tX96.hs:8:3: warning: [-Wtyped-holes] Variable not in scope: (&) :: t4 -> (a0 -> I a0) -> t3 /tmp/tX96.hs:9:3: warning: [-Wtyped-holes] Variable not in scope: (&) :: t3 -> t5 -> t2 /tmp/tX96.hs:10:3: warning: [-Wtyped-holes] Variable not in scope: (&) :: t2 -> (POP f1 xss0 -> [SOP f1 xss0]) -> t1 /tmp/tX96.hs:11:3: warning: [-Wtyped-holes] Variable not in scope: (&) :: t1 -> ([a1] -> a1) -> t0 /tmp/tX96.hs:12:3: warning: [-Wtyped-holes] Variable not in scope: (&) :: t0 -> (Rep a2 -> a2) -> a ghc: panic! (the 'impossible' happened) (GHC version 8.0.1 for x86_64-unknown-linux): opt_univ fell into a hole {aabX} Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug > }}} ---- More minimal example, still requires [https://hackage.haskell.org/package /generics-sop-0.2.2.0 generics-sop]: {{{#!hs import Generics.SOP a = minBound & hcpure (Proxy @Bounded) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12589> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #12676: Panic with piResultTys1 again
by GHC 18 Oct '16

18 Oct '16

#12676: Panic with piResultTys1 again -------------------------------------+------------------------------------- Reporter: pjmagni | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Compile-time Unknown/Multiple | crash Test Case: | Blocked By: Blocking: | Related Tickets: #12634 Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The code {{{#!hs {-# LANGUAGE RankNTypes, TypeFamilies #-} module Panic where class C a where data T a panic :: (forall s . T a) -> () instance C () where data T () = T panic T = () }}} produces the output {{{ ghc: panic! (the 'impossible' happened) (GHC version 8.0.1 for x86_64-apple-darwin): piResultTys1 R:T() [Any] }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12676> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #12698: GHC panic on pattern synonym
by GHC 18 Oct '16

18 Oct '16

#12698: GHC panic on pattern synonym -------------------------------------+------------------------------------- Reporter: Iceland_jack | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.1 Keywords: | Operating System: Unknown/Multiple PatternSynonyms | Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- {{{#!hs {-# Language ViewPatterns, TypeOperators, KindSignatures, PolyKinds, TypeInType, StandaloneDeriving, GADTs, RebindableSyntax, RankNTypes, LambdaCase, PatternSynonyms, TypeApplications #-} import GHC.Types import Prelude hiding ( fromInteger ) import Data.Type.Equality import Data.Kind import qualified Prelude class Ty (a :: k) where ty :: T a instance Ty Int where ty = TI instance Ty Bool where ty = TB instance Ty a => Ty [a] where ty = TA TL ty instance Ty [] where ty = TL instance Ty (,) where ty = TP data AppResult (t :: k) where App :: T a -> T b -> AppResult (a b) data T :: forall k. k -> Type where TI :: T Int TB :: T Bool TL :: T [] TP :: T (,) TA :: T f -> T x -> T (f x) deriving instance Show (T a) splitApp :: T a -> Maybe (AppResult a) splitApp = \case TI -> Nothing TB -> Nothing TL -> Nothing TP -> Nothing TA f x -> Just (App f x) data (a :: k1) :~~: (b :: k2) where HRefl :: a :~~: a eqT :: T a -> T b -> Maybe (a :~~: b) eqT a b = case (a, b) of (TI, TI) -> Just HRefl (TB, TB) -> Just HRefl (TL, TL) -> Just HRefl (TP, TP) -> Just HRefl pattern List :: () => [] ~~ b => T b pattern List <- (eqT (ty @(Type -> Type) @[]) -> Just HRefl) where List = ty pattern Int :: () => Int ~~ b => T b pattern Int <- (eqT (ty @Type @Int) -> Just HRefl) where Int = ty pattern (:<->:) :: () => fx ~ f x => T f -> T x -> T fx pattern f :<->: x <- (splitApp -> Just (App f x)) where f :<->: x = TA f x pattern Foo <- List :<->: Int }}} Using GHCi version 8.1.20160930: {{{ G$ ./ghc-stage2 -ignore-dot-ghci --interactive /tmp/td2W.hs GHCi, version 8.1.20160930: http://www.haskell.org/ghc/ :? for help [1 of 1] Compiling Main ( /tmp/td2W.hs, interpreted ) ghc-stage2: panic! (the 'impossible' happened) (GHC version 8.1.20160930 for x86_64-unknown-linux): ASSERT failed! k1_a2mO k1_a2mO [TCvSubst In scope: InScope {k_a2mM fx_a2mN f_a2mP x_a2mQ rep_a2ny r_a2nz scrut_a2nB $cshowsPrec_a2ob $cshow_a2p1 $cshowList_a2p8 $cty_a2pi $cty_a2po $cty_a2pw $cty_a2pJ $cty_a2pP splitApp eqT $fTy(->)(,) $fTy(->)[] $fTyTYPE[] $fTyTYPEBool $fTyTYPEInt $fShowT $mList $bList $mInt $bInt $m:<->: $b:<->:} Type env: [a2mN :-> fx_a2mN, a2mP :-> f_a2mP, a2mQ :-> x_a2mQ, a2nz :-> r_a2nz] Co env: []] Call stack: CallStack (from HasCallStack): prettyCurrentCallStack, called at compiler/utils/Outputable.hs:1076:58 in ghc:Outputable callStackDoc, called at compiler/utils/Outputable.hs:1125:22 in ghc:Outputable assertPprPanic, called at compiler/types/TyCoRep.hs:2318:46 in ghc:TyCoRep Call stack: CallStack (from HasCallStack): prettyCurrentCallStack, called at compiler/utils/Outputable.hs:1076:58 in ghc:Outputable callStackDoc, called at compiler/utils/Outputable.hs:1080:37 in ghc:Outputable pprPanic, called at compiler/utils/Outputable.hs:1123:5 in ghc:Outputable assertPprPanic, called at compiler/types/TyCoRep.hs:2318:46 in ghc:TyCoRep Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug > }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12698> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #12653: reexported-modules sometimes gets mangled, causes problems for bootstrap
by GHC 18 Oct '16

18 Oct '16

#12653: reexported-modules sometimes gets mangled, causes problems for bootstrap -------------------------------------+------------------------------------- Reporter: ezyang | Owner: Type: bug | Status: new Priority: highest | Milestone: 8.0.2 Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- There is a pretty bad bug in Cabal 1.24's support for reexported modules (https://github.com/haskell/cabal/pull/3906) which means that, under some circumstances, Cabal library will output `reexported-modules` in a form that ghc-pkg will misinterpret. The particular instance I've seen is GHC's build system incorrectly concluding that GHC.Lexeme is provided by ghc- boot rather than ghc-boot-th (because the reexport wasn't parsed correctly). It does not actually seem possible to work around this given a GHC 8.0.1 bootstrap, because we need to register ghc-boot using the bootstrapping ghc-pkg, which will output the package configuration in mangled form. So my advice is: 1. Fix this bug for GHC 8.0.2, so that at least 8.0.2 bootstrap will work even if we have reexported modules in the bootstrap libraries 2. Stop using reexports in ghc-boot; they're not being used for any good reason and will help avoid the bug. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12653> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #12664: System.Environment.withArgs and friends leak memory
by GHC 18 Oct '16

18 Oct '16

#12664: System.Environment.withArgs and friends leak memory -------------------------------------+------------------------------------- Reporter: simonmar | Owner: Type: bug | Status: new Priority: normal | Milestone: 8.0.2 Component: | Version: 8.0.1 libraries/base | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- `ArgvLeak.hs`: {{{ import System.Environment main :: IO () main = withArgs ["a","b","c"] $ return () }}} repro: {{{ > ghc ArgvLeak.hs [1 of 1] Compiling Main ( ArgvLeak.hs, ArgvLeak.o ) Linking ArgvLeak ... > valgrind --leak-check=full ./ArgvLeak ==27227== Memcheck, a memory error detector ==27227== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al. ==27227== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info ==27227== Command: ./ArgvLeak ==27227== ==27227== Warning: set address range perms: large range [0x200000000, 0x600100000) (noaccess) ==27227== ==27227== HEAP SUMMARY: ==27227== in use at exit: 107 bytes in 9 blocks ==27227== total heap usage: 62 allocs, 53 frees, 60,175 bytes allocated ==27227== ==27227== 55 (40 direct, 15 indirect) bytes in 1 blocks are definitely lost in loss record 6 of 6 ==27227== at 0x4C2FB55: calloc (in /usr/lib/valgrind /vgpreload_memcheck-amd64-linux.so) ==27227== by 0x470057: stgCallocBytes (RtsUtils.c:107) ==27227== by 0x4652C1: copyArgv (RtsFlags.c:1811) ==27227== by 0x465FAE: setProgArgv (RtsFlags.c:1877) ==27227== by 0x41CC4A: base_SystemziEnvironment_zdLr7BcsetProgArgv_info (in /home/smarlow/scratch/ArgvLeak) ==27227== ==27227== LEAK SUMMARY: ==27227== definitely lost: 40 bytes in 1 blocks ==27227== indirectly lost: 15 bytes in 4 blocks ==27227== possibly lost: 0 bytes in 0 blocks ==27227== still reachable: 52 bytes in 4 blocks ==27227== suppressed: 0 bytes in 0 blocks ==27227== Reachable blocks (those to which a pointer was found) are not shown. ==27227== To see them, rerun with: --leak-check=full --show-leak-kinds=all ==27227== ==27227== For counts of detected and suppressed errors, rerun with: -v ==27227== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 0 from 0) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12664> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #12719: master branch broken with alex 3.2
by GHC 18 Oct '16

18 Oct '16

#12719: master branch broken with alex 3.2 -------------------------------------+------------------------------------- Reporter: bgamari | Owner: bgamari Type: task | Status: new Priority: highest | Milestone: 8.2.1 Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The build fails with all sorts of type errors. Need to work out what is going on when there is time. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12719> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #12616: type synonyms confuse generalized newtype deriving role checking
by GHC 18 Oct '16

18 Oct '16

#12616: type synonyms confuse generalized newtype deriving role checking -------------------------------------+------------------------------------- Reporter: daviddarais | Owner: Type: bug | Status: new Priority: low | Milestone: Component: Compiler | Version: 8.0.1 Keywords: generalized | Operating System: Unknown/Multiple newtype deriving roles rankntypes | Architecture: | Type of failure: GHC rejects Unknown/Multiple | valid program Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Below is a small Haskell file showing the problem. It defines a simple `MonadTrans` typeclass, but using a type operator `~>`. When using the type operator, GHC complains that generalized newtype deriving gets stuck on a nominal role, but when the type operator is not used then GND works just fine. Perhaps this is just expected behavior with mixing RankNTypes and GND (and roles), but it's rather unfortunate. {{{#!hs {-# LANGUAGE GeneralizedNewtypeDeriving #-} {-# LANGUAGE RankNTypes #-} {-# LANGUAGE TypeOperators #-} module GND where type m ~> n = forall a. m a -> n a class MonadTrans t where -- > this line works: -- lift :: (Monad m) => m a -> t m a -- > this line doesn't: lift :: (Monad m) => m ~> t m data StateT s m a = StateT { runStateT :: s -> m (a, s) } instance MonadTrans (StateT s) where lift xM = StateT $ \ s -> do { x <- xM ; return (x,s) } newtype OtherStateT s m a = OtherStateT { runOtherStateT :: StateT s m a } deriving (MonadTrans) }}} The error message is: {{{ GND.hs:23:13: error: • Couldn't match representation of type ‘m1 ~> StateT s m1’ with that of ‘m1 a1 -> OtherStateT s m1 a1’ arising from a use of ‘GHC.Prim.coerce’ • In the expression: GHC.Prim.coerce (lift :: (~>) m (StateT s m)) :: forall (m :: TYPE GHC.Types.PtrRepLifted -> TYPE GHC.Types.PtrRepLifted). Monad m => (~>) m (OtherStateT s m) In an equation for ‘lift’: lift = GHC.Prim.coerce (lift :: (~>) m (StateT s m)) :: forall (m :: TYPE GHC.Types.PtrRepLifted -> TYPE GHC.Types.PtrRepLifted). Monad m => (~>) m (OtherStateT s m) When typechecking the code for ‘lift’ in a derived instance for ‘MonadTrans (OtherStateT s)’: To see the code I am typechecking, use -ddump-deriv In the instance declaration for ‘MonadTrans (OtherStateT s)’ • Relevant bindings include lift :: m ~> OtherStateT s m (bound at GND.hs:23:13) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12616> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

Re: [GHC] #1600: Optimisation: CPR the results of IO
by GHC 18 Oct '16

18 Oct '16

#1600: Optimisation: CPR the results of IO -------------------------------------+------------------------------------- Reporter: simonmar | Owner: Type: task | Status: new Priority: lowest | Milestone: Component: Compiler | Version: 6.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Runtime | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: #8598 | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Changes (by simonpj): * cc: osa1 (added) Comment: I wonder why there's a consistent 2% increase in binary size; I was not expecting that. I have not looked at the code... when you are ready, seeing it all as one Phab diff (not lots) would be helpful. Omer may be thinking about nested CPR, and has an implementation on a branch somewhere. This is a particular instance of nested CPR, so it'd be good to sync up with him (cc'd). Simon -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/1600#comment:55> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #12723: Family instance modules are not fingerprinted in ABI
by GHC 18 Oct '16

18 Oct '16

#12723: Family instance modules are not fingerprinted in ABI -------------------------------------+------------------------------------- Reporter: ezyang | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Driver | Version: 8.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- This leads to the following delightful, five module bug: {{{ -- A.hs {-# LANGUAGE TypeFamilies #-} module A where type family F a type instance F Int = Bool -- B.hs module B where import A -- C.hs module C where import B -- D.hs module D where import C -- E.hs module E where import D import B }}} Build these modules. Then rename A to A2 (fixing B's import), and build them again. You'll get: {{{ ezyang@sabre:~/Dev/labs/T3871$ ghc --make E.hs [1 of 5] Compiling A2 ( A2.hs, A2.o ) [2 of 5] Compiling B ( B.hs, B.o ) [3 of 5] Compiling C ( C.hs, C.o ) [B changed] [5 of 5] Compiling E ( E.hs, E.o ) [B changed] attempting to use module ‘A’ (./A.hs) which is not loaded }}} The problem is clear: D was not recompiled, but it needs to be, because when the module gets renamed, we need to update its list of family instance modules to rename A to A2. When we don't do this, the subsequent family instance check chokes because it tries to load A. And why did D decide not to get recompiled? Because the ABI hash of B did not change. And that's WRONG. (Also, family instances are really awful, you really do have to rebuild everything when you change them. UGH.) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12723> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5