November 2018 - ghc-tickets

[GHC] #15091: Better occurrence analysis for join points
by GHC 10 Feb '19

10 Feb '19

#15091: Better occurrence analysis for join points -------------------------------------+------------------------------------- Reporter: simonpj | Owner: (none) Type: task | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.2.2 Keywords: JoinPoints | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Consider this {{{ let x = ... in join j y = x+y in case v of A -> x B -> j 1 C -> j 2 D -> 3 }}} What does `OccAnal` say about `x`'s usage? It says `ManyOccs`! But it's plain as a pikestaff that `x` is used at most once: * Either in the `A` branch, * or in the `B` or `C` branches via `j`. If instead we had inlined `j` we'd have {{{ let x = ... in join j y = x+y in case v of A -> x B -> x + 1 C -> x + 2 D -> 3 }}} and now it's all more obvious: `x`'s occurrence info should be `OneOcc { occ_one_br = False }`, not `ManyOccs`. Does this matter? Not a great deal, but there is a reason for having `OneOcc` with `occ_one_br = False`, and it seems a shame not to take advantage. One case in point is the definition of `SimplUtils.isExitJoinId {{{ isExitJoinId :: Var -> Bool isExitJoinId id = isJoinId id && isOneOcc (idOccInfo id) && occ_in_lam (idOccInfo id) }}} Something does not cease to be an exit-join-point if it is mentioned in multiple places as above. Another use for this info is `postInlineUnconditionally`. Could we improve this situation? I think it'd be quite easy. For non- recursive join points `j = rhs` * Gather occurrence info from the RHS * Bind `j` to its occurrence info * Unleash that occurrence info at each jump-site for `j`, just as if it had been inlined. See Trac #14152, comment:40. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15091> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #15449: Nondeterministic Failure on aarch64 with -jn, n > 1
by GHC 10 Feb '19

10 Feb '19

#15449: Nondeterministic Failure on aarch64 with -jn, n > 1 -------------------------------------+------------------------------------- Reporter: tmobile | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.4.3 Keywords: | Operating System: Linux Architecture: aarch64 | Type of failure: Compile-time | crash or panic Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- GHC releases 8.2.1 through 8.4.3 exhibit various crashes when invoked with '-jn' where n > 1. GHCHQ's binary releases have this behavior, as well as GHCs I've cross-built on my own. In order to reproduce this issue there must be some parallelism in the module dependency graph; I've attached a test package for easily reproducing this. Use of deriving in the test modules isn't necessary to trigger this; it merely gives the compiler some work to do. The 'hscolour' package also triggers this issue reliably. To trigger the bad behavior, simply run: ghc --make -jn Main.hs -o test with n > 1 in the test-package. Running this repeatedly (removing .hi and .o files in between runs, of course), I've observed these outcomes with varying frequencies: - Segmentation fault. - Bus fault. - Compiler process sleeps indefinitely. - {{{ <no location info>: error: ghc: panic! (the 'impossible' happened) (GHC version 8.4.3 for aarch64-unknown-linux): Binary.UserData: no put_binding_name }}} - {{{ ghc: internal error: MUT_VAR_CLEAN object entered! (GHC version 8.4.3 for aarch64_unknown_linux) Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug Aborted (core dumped) }}} - And most strangely: {{{ A.hs:3:1: error: • Kind signature on data type declaration has non-* return kind * • In the data declaration for ‘A’ | 3 | data A = A | ^^^^^^^^^^... }}} I have not noticed issues with any other concurrent Haskell programs on aarch64. I'm using the NVIDIA Jetson TX2 for these tests. I have not yet tried GHC 8.0.x, 8.6.x, or HEAD. I'm having trouble reproducing this in GDB; I'll report back when I've got that working. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15449> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 27

[GHC] #12658: GHC 7.10.3 sometimes reports unimplemented/strange closure type 63270
by GHC 09 Feb '19

09 Feb '19

#12658: GHC 7.10.3 sometimes reports unimplemented/strange closure type 63270 --------------------------------+--------------------------------------- Reporter: clint | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.3 Keywords: | Operating System: Linux Architecture: powerpc | Type of failure: Compile-time crash Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: --------------------------------+--------------------------------------- This happened when building http-link-header. {{{ ghc: internal error: scavenge: unimplemented/strange closure type 63270 @ 0xf36de0a8ghc: internal error: scavenge: unimplemented/strange closure type 63270 @ 0xf36de0a8 (GHC version 7.10.3 for powerpc_unknown_linux) Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} Sometimes it segfaults instead. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12658> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #15336: ./System/IO.hs accidentally overridden when running ghci
by GHC 08 Feb '19

08 Feb '19

#15336: ./System/IO.hs accidentally overridden when running ghci -------------------------------------+------------------------------------- Reporter: luqui | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.1 Component: GHCi | Version: 8.2.2 Keywords: | Operating System: MacOS X Architecture: | Type of failure: GHC rejects Unknown/Multiple | valid program Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Running ghci fails if there is a System/IO.hs module present in the current directory. Doesn't seem like desirable behavior. {{{ % mkdir System % touch System/IO.hs % ghci GHCi, version 8.2.2: http://www.haskell.org/ghc/ :? for help <interactive>:1:19: error: Not in scope: ‘System.IO.hSetBuffering’ No module named ‘System.IO’ is imported. <interactive>:1:43: error: Not in scope: ‘System.IO.stdin’ No module named ‘System.IO’ is imported. <interactive>:1:60: error: Not in scope: data constructor ‘System.IO.NoBuffering’ No module named ‘System.IO’ is imported. <interactive>:1:99: error: Not in scope: ‘System.IO.hSetBuffering’ No module named ‘System.IO’ is imported. <interactive>:1:123: error: Not in scope: ‘System.IO.stdout’ No module named ‘System.IO’ is imported. <interactive>:1:140: error: Not in scope: data constructor ‘System.IO.NoBuffering’ No module named ‘System.IO’ is imported. <interactive>:1:179: error: Not in scope: ‘System.IO.hSetBuffering’ No module named ‘System.IO’ is imported. <interactive>:1:203: error: Not in scope: ‘System.IO.stderr’ No module named ‘System.IO’ is imported. <interactive>:1:220: error: Not in scope: data constructor ‘System.IO.NoBuffering’ No module named ‘System.IO’ is imported. }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15336> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #9370: large blowup in memory usage and time when doing parallel build of xmlhtml package
by GHC 08 Feb '19

08 Feb '19

#9370: large blowup in memory usage and time when doing parallel build of xmlhtml package -------------------------------------+------------------------------------- Reporter: carter | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- I've observed a blowup in ghc memory usage when invoked with parallel build flags. to reproduce {{{ cabal get xmlhtml-0.2.3.2 cabal install xmlhtml-0.2.3.2 --only-dependencies cd xmlhtml-0.2.3.2 }}} then {{{ cabal clean ; cabal configure --ghc-options="-j4" ; time cabal build -j4 -v3 }}} will take quite a while and use > 1gb of ram whereas {{{ cabal clean ; cabal configure --ghc-options="-j1 ; time cabal build -j1 -v3 }}} will use < 150mb of ram. Based upon the output of cabal build -j4 -v3, it looks like the parallel build is spending a lottt more time in the simplifier passes. I'm not sure what changes between the parallel and sequential builds to change this. I'll try to dig into this more in a few days, but recording this problem now before i forget. (though of course any insights would be appreciated) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9370> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 48

[GHC] #8635: GHC optimisation flag ignored when importing a local module with derived type classes
by GHC 08 Feb '19

08 Feb '19

#8635: GHC optimisation flag ignored when importing a local module with derived type classes ------------------------------------+------------------------------------- Reporter: NeilMitchell | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.6.3 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- Given Foo.hs and Bar.hs: {{{ module Foo where data Food = Food -- deriving Eq {-# OPTIONS_GHC -O2 -ddump-simpl #-} module Bar where import Foo bar :: Int -> Bool bar x = x == 72 }}} If I run: {{{ ghc --make Bar -fforce-recomp }}} I get (snipped): {{{ Bar.bar = \ (x_afk :: GHC.Types.Int) -> case x_afk of _ { GHC.Types.I# x1_alo -> case x1_alo of _ { __DEFAULT -> GHC.Types.False; 72 -> GHC.Types.True } } }}} `bar` now looks pretty well optimised. However, if I uncomment the `deriving Eq` I get: {{{ Bar.bar1 = GHC.Types.I# 72 Bar.bar2 = GHC.Classes.== @ GHC.Types.Int GHC.Classes.$fEqInt Bar.bar = \ (x_amD :: GHC.Types.Int) -> Bar.bar2 x_amD Bar.bar1 }}} Now `bar` looks like terrible code, using dictionaries, boxing etc. It seems adding `deriving` in the imported and unused module makes it ignore the optimisation level. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8635> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #11822: Pattern match checker exceeded (2000000) iterations
by GHC 07 Feb '19

07 Feb '19

#11822: Pattern match checker exceeded (2000000) iterations -------------------------------------+------------------------------------- Reporter: j.waldmann | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1-rc3 Keywords: | Operating System: Unknown/Multiple Architecture: x86_64 | Type of failure: Compile-time (amd64) | performance bug Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- graphviz seems a nice benchmark (not too large, few dependencies) that highlights compiler performance problems on "trivial" source code (cf. https://hackage.haskell.org/package/graphviz-2999.18.0.2/docs/src /Data-GraphViz-Attributes-Complete.html#sameAttribute) Compilation of Compiling Data.GraphViz.Attributes.Colors.X11 and Data.GraphViz.Attributes.Values takes >= 1 second each, and the following happens: {{{ cabal install graphviz-2999.18.0.2 --allow-newer ... [17 of 31] Compiling Data.GraphViz.Attributes.Complete ( Data/GraphViz/Attributes/Complete.hs, dist/dist-sandbox- e6f2c124/build/Data/GraphViz/Attributes/Complete.o ) Data/GraphViz/Attributes/Complete.hs:1013:1: warning: Pattern match checker exceeded (2000000) iterations in an equation for ‘sameAttribute’. (Use -fmax-pmcheck-iterations=n to set the maximun number of iterations to n) }}} I do welcome that ghc emits this warning - I expect you want its occurence to be reported as a bug? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11822> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 22

[GHC] #15286: "Can't use Natural in base" when compiling GHC.Natural with -O0
by GHC 07 Feb '19

07 Feb '19

#15286: "Can't use Natural in base" when compiling GHC.Natural with -O0 -------------------------------------+------------------------------------- Reporter: alpmestan | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: | Version: 8.5 libraries/base | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Building GHC Unknown/Multiple | failed Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- {{{ _build/stage0/bin/ghc -Wall -hisuf hi -osuf o -hcsuf hc -static -hide-all- packages -no-user-package-db '-package-db _build/stage1/lib/package.conf.d' '-this-unit-id base-4.12.0.0' '-package- id ghc-prim-0.5.3' '-package-id integer-simple-0.1.1.1' '-package-id rts-1.0' -i -i_build/stage1/libraries/base/build -i_build/stage1/libraries/base/build/autogen -ilibraries/base/. -Iincludes -I_build/generated -I_build/stage1/libraries/base/build -I_build/stage1/libraries/base/build/include -Ilibraries/base/include -I/home/travis/build/snowleopard/hadrian/ghc/_build/stage1/lib/x86_64 -linux-ghc-8.5.20180617/rts-1.0/include -I_build/generated -optc- I_build/generated -optP-include -optP_build/stage1/libraries/base/build/autogen/cabal_macros.h -optc- std=gnu99 -optc-fno-stack-protector -optP-std=gnu99 -odir _build/stage1/libraries/base/build -hidir _build/stage1/libraries/base/build -stubdir _build/stage1/libraries/base/build -Wnoncanonical-monad-instances -optc- Werror=unused-but-set-variable -optc-Wno-error=inline -c libraries/base/GHC/Num.hs -o _build/stage1/libraries/base/build/GHC/Num.o -O0 -H64m -this-unit-id base -Wcompat -Wnoncanonical-monad-instances -XHaskell2010 -ghcversion- file=/home/travis/build/snowleopard/hadrian/ghc/_build/generated/ghcversion.h -Wno-deprecated-flags -Wno-trustworthy-safe Exit code: 1 Stderr: ghc: panic! (the 'impossible' happened) (GHC version 8.5.20180617 for x86_64-unknown-linux): Can't use Natural in base }}} We noticed this in hadrian land (in [https://github.com/snowleopard/hadrian/issues/499 this PR] and [https://github.com/snowleopard/hadrian/issues/629] and noticed that this happens: - with 8.4.2 as the boot compiler but not 8.2.2 - with integer-simple as the integer library - only with the quickest flavour (or more generally when we build GHC.Natural with -O0) Some optimisation seems critical to making any trace of `Natural` disappear, but this seems rather fragile. The code that throws the error got introduced in fe770c211631e7b4c9b0b1e88ef9b6046c6585ef. The full build log is available [https://travis- ci.org/snowleopard/hadrian/jobs/393259151#L4209 here], anchored to where the error appears. An example of hadrian command to reproduce the problem, assuming a configured build tree: `hadrian/build.sh --flavour=quickest --integer- simple -j4`. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15286> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #14730: Missing predicate for "ResourceVanished" IOException/IOErrorType
by GHC 06 Feb '19

06 Feb '19

#14730: Missing predicate for "ResourceVanished" IOException/IOErrorType -------------------------------------+------------------------------------- Reporter: bit | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: | Version: 8.2.2 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: -------------------------------------+------------------------------------- `IOErrorType` has a "`ResourceVanished`" constructor: https://hackage.haskell.org/package/base-4.10.1.0/docs/src/GHC.IO.Exception… But there is no way to detect this for an `IOException` or `IOErrorType`. Here is some real-world code I found, where the author wants to catch "`ResourceVanished`", but resorts to catching all `IOException`: https://hackage.haskell.org/package/http-client-tls-0.3.5.1/docs/src /Network-HTTP-Client-TLS.html#convertConnection Docs for `hPutBuf` also mention the public existence of "`ResourceVanished`": https://hackage.haskell.org/package/base-4.10.1.0/docs/GHC-IO- Handle.html#v:hPutBuf Suggestion: We need to add functions: {{{#!haskell isResourceVanished :: IOError -> Bool resourceVanishedErrorType :: IOErrorType isResourceVanishedErrorType :: IOErrorType -> Bool }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14730> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #15897: Negative MUT time in +RTS -s -RTS when heap profiling is enabled
by GHC 06 Feb '19

06 Feb '19

#15897: Negative MUT time in +RTS -s -RTS when heap profiling is enabled -------------------------------------+------------------------------------- Reporter: maoe | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.3 Component: Profiling | Version: 8.6.2 Keywords: | Operating System: Linux Architecture: x86_64 | Type of failure: Debugging (amd64) | information is incorrect Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Mutator run time in +RTS -s -RTS can go negative in GHC 8.6.x when retainer profiling is enabled. Here is an example taken from https://ghc.haskell.org/trac/ghc/ticket/881#comment:9 {{{ {-# OPTIONS_GHC -fprof-auto #-} import Control.Monad import Data.IORef import System.Environment main = do [n] <- getArgs replicateM (read n) (newIORef [1,2,3]) }}} {{{ % ghc-8.6.2 -prof -rtsopts -fforce-recomp repro.hs [1 of 1] Compiling Main ( repro.hs, repro.o ) Linking repro ... % ./repro 10000000 +RTS -s -h 3,053,080,472 bytes allocated in the heap 5,538,404,848 bytes copied during GC 1,013,615,328 bytes maximum residency (16 sample(s)) 36,422,944 bytes maximum slop 966 MB total memory in use (0 MB lost due to fragmentation) Tot time (elapsed) Avg pause Max pause Gen 0 2600 colls, 0 par 1.375s 1.376s 0.0005s 0.0019s Gen 1 16 colls, 0 par 4.355s 4.355s 0.2722s 1.0136s INIT time 0.000s ( 0.000s elapsed) MUT time 0.000s ( -0.833s elapsed) GC time 4.016s ( 4.016s elapsed) RP time 0.000s ( 0.000s elapsed) PROF time 1.715s ( 1.715s elapsed) EXIT time 1.715s ( 1.715s elapsed) Total time 6.612s ( 6.612s elapsed) %GC time 0.0% (0.0% elapsed) Alloc rate 0 bytes per MUT second Productivity -12.6% of total user, -12.6% of total elapsed }}} Here is what I found so far: * GHC 8.6.1, 8.6.2, and yesterday's HEAD have the same issue. GHC 8.4.4 doesn't have this issue. * `-hr` has the same issue. `-hT` is fine. * Without heap profiling the numbers this doesn't happen -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15897> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 11