February 2015 - ghc-tickets

Re: [GHC] #7495: generalizing overloaded list syntax to Sized Lists, HLists, HRecords, etc
by GHC 23 Feb '15

23 Feb '15

#7495: generalizing overloaded list syntax to Sized Lists, HLists, HRecords, etc -------------------------------------+------------------------------------- Reporter: nwf | Owner: carter Type: feature request | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: None/Unknown | Unknown/Multiple Blocked By: | Test Case: Related Tickets: #9883 | Blocking: 9883 | Differential Revisions: -------------------------------------+------------------------------------- Changes (by gershomb): * cc: gershomb@… (added) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7495#comment:10> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7634: MD5 collision could lead to SafeHaskell violation
by GHC 23 Feb '15

23 Feb '15

#7634: MD5 collision could lead to SafeHaskell violation -------------------------------------+------------------------------------- Reporter: shachaf | Owner: ekmett Type: bug | Status: new Priority: normal | Milestone: ⊥ Component: Core Libraries | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Other | Unknown/Multiple Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by ekmett): The move out from 7.12 was more an acknowledgement that the CLC wasn't planning on tackling this issue in 7.12. If everything is a priority then nothing is a priority, and we have lots of actionable issues with far more immediate impact. Switching fingerprints to something like SHA-256 could be done. In fact it is probably a pretty good idea, despite doubling the size of fingerprints and having a measurable performance impact -- if a patch was put forth then it would by all means be considered. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7634#comment:7> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #9103: hackage's type-level-0.2.4 fails to compile
by GHC 23 Feb '15

23 Feb '15

#9103: hackage's type-level-0.2.4 fails to compile -------------------------------------+------------------------------------- Reporter: slyfox | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 7.8.2 checker) | Operating System: Unknown/Multiple Keywords: | Type of failure: GHC rejects Architecture: Unknown/Multiple | valid program Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | -------------------------------------+------------------------------------- Attached trimmed-down dependencyless archive. Steps to reproduce: {{{ $ tar -xzf type-level-0.2.4.tar.gz $ ghci -hide-all-packages -package=base -package=template-haskell Data/TypeLevel.hs }}} The error on 7.8.2 and in -HEAD is: {{{ [6 of 8] Compiling Data.TypeLevel.Num.Ops ( Data/TypeLevel/Num/Ops.hs, interpreted ) Data/TypeLevel/Num/Ops.hs:90:10: Illegal instance declaration for ‘Succ' (x, x) (x, x) D0 D0 True’ The liberal coverage condition fails in class ‘Succ'’ for functional dependency: ‘yh yl yz -> xh xl’ Reason: lhs types ‘D0’, ‘D0’, ‘True’ do not jointly determine rhs types ‘(x, x)’, ‘(x, x)’ In the instance declaration for ‘Succ' (x, x) (x, x) D0 D0 True’ ... }}} Typechecked fine on ghc-7.6.3 {{{ [1 of 8] Compiling Data.TypeLevel.Num.Reps ( Data/TypeLevel/Num/Reps.hs, interpreted ) [2 of 8] Compiling Data.TypeLevel.Num.Sets ( Data/TypeLevel/Num/Sets.hs, interpreted ) [3 of 8] Compiling Data.TypeLevel.Num.Aliases.TH ( Data/TypeLevel/Num/Aliases/TH.hs, interpreted ) [4 of 8] Compiling Data.TypeLevel.Num.Aliases ( Data/TypeLevel/Num/Aliases.hs, interpreted ) Generating and compiling a zillion numerical type aliases, this might take a while [5 of 8] Compiling Data.TypeLevel.Bool ( Data/TypeLevel/Bool.hs, interpreted ) [6 of 8] Compiling Data.TypeLevel.Num.Ops ( Data/TypeLevel/Num/Ops.hs, interpreted ) [7 of 8] Compiling Data.TypeLevel.Num ( Data/TypeLevel/Num.hs, interpreted ) [8 of 8] Compiling Data.TypeLevel ( Data/TypeLevel.hs, interpreted ) Ok, modules loaded: Data.TypeLevel, Data.TypeLevel.Num, Data.TypeLevel.Bool, Data.TypeLevel.Num.Reps, Data.TypeLevel.Num.Aliases, Data.TypeLevel.Num.Sets, Data.TypeLevel.Num.Ops, Data.TypeLevel.Num.Aliases.TH. *Data.TypeLevel> }}} I have failed to decipher an error message, but looks like it tries to instantiate a thing it should not. My apologies if it's not-a-bug. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9103> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 18

Re: [GHC] #3055: Int / Word / IntN / WordN are unequally optimized
by GHC 23 Feb '15

23 Feb '15

#3055: Int / Word / IntN / WordN are unequally optimized -------------------------------------+------------------------------------- Reporter: claus | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 6.11 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Runtime | Unknown/Multiple performance bug | Test Case: Blocked By: | simplCore/should_compile/T3055 Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Changes (by thomie): * testcase: => simplCore/should_compile/T3055 Comment: The test has been passing for a while. Needs investigation to see if this ticket can be closed. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/3055#comment:16> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #17: Separate warnings for unused local and top-level bindings
by GHC 23 Feb '15

23 Feb '15

#17: Separate warnings for unused local and top-level bindings -------------------------------------+------------------------------------- Reporter: magunter | Owner: Type: feature request | Status: new Priority: lowest | Milestone: 7.12.1 Component: Compiler | Version: None Resolution: None | Keywords: -fwarn- Operating System: Unknown/Multiple | unused-binds newcomer Type of failure: None/Unknown | Architecture: Blocked By: | Unknown/Multiple Related Tickets: #3283 | Test Case: | Blocking: | Differential Revisions: Phab:D591 -------------------------------------+------------------------------------- Comment (by Austin Seipp <austin@…>): In [changeset:"aead01902e1c41e85b758dbafd15e60d08956374/ghc"]: {{{ #!CommitTicketReference repository="ghc" revision="aead01902e1c41e85b758dbafd15e60d08956374" driver: split -fwarn-unused-binds into 3 flags (fixes #17) Summary: New flags: -fwarn-unused-top-binds -fwarn-unused-local-binds -fwarn-unused-pattern-binds Test Plan: `tests/rename/should_compile/T17` tests Correct other tests Reviewers: austin, rwbarton Reviewed By: austin, rwbarton Subscribers: rwbarton, carter, thomie Differential Revision: https://phabricator.haskell.org/D591 GHC Trac Issues: #17 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/17#comment:17> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7336: Defined but not used is not detected for data types with instances
by GHC 23 Feb '15

23 Feb '15

#7336: Defined but not used is not detected for data types with instances -------------------------------------+------------------------------------- Reporter: maeder | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: None/Unknown | Unknown/Multiple Blocked By: | Test Case: Related Tickets: | rename/should_compile/T7336 | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Changes (by thomie): * cc: cactus (added) * status: closed => new * resolution: fixed => Comment: Commit [changeset:"22ddcfff16476a6d19190f10743a78bde0b2c28f/ghc"] mentions: "T7336 is expected to be broken because of a regression introduced by pattern synonym implementation" -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7336#comment:2> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #9778: Namespace collision detection for promoted types
by GHC 23 Feb '15

23 Feb '15

#9778: Namespace collision detection for promoted types -------------------------------------+------------------------------------- Reporter: crockeea | Owner: Type: feature request | 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: -------------------------------------+------------------------------------- In the following example {{{#!hs {-# LANGUAGE DataKinds #-} module Foo where data Nat = Z | S Nat data S foo :: S n -> S n foo = id }}} GHC displays the error {{{ Foo.hs:12:8: ‘S’ is applied to too many type arguments In the type signature for ‘foo’: foo :: S n -> S n Failed, modules loaded: none. }}} Although it's clear in my example where the problem is, this example was distilled from a case where I imported module1 with type `S` defined and module2 with the promoted type `S`, which is what I was trying to use. Can we get namespace collision detection instead? Something like: {{{ Foo.hs:9:15 Ambiguous occurence ‘S’ It could refer to either Foo.S defined at Foo.hs:7:1 or to the promoted constructor S defined at Foo.hs:5:16 Use 'S to refer to the promoted constructor. }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9778> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 16

[GHC] #9986: lhs -> hs fallout
by GHC 23 Feb '15

23 Feb '15

#9986: lhs -> hs fallout -------------------------------------+------------------------------------- Reporter: jstolarek | Owner: Type: task | Status: new Priority: normal | Milestone: Component: | Version: 7.9 Documentation | Operating System: Unknown/Multiple Keywords: | Type of failure: None/Unknown Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- It looks that during .lhs -> .hs switch the comments were not updated. Grepping for ".lhs" reveals 117 lines where .lhs files are referenced: {{{ basicTypes/OccName.hs:154:-- see also: Note [Data Constructor Naming] in DataCon.lhs basicTypes/BasicTypes.hs:447:-- instance. See Note [Safe Haskell isSafeOverlap] (in `InstEnv.lhs`) for a basicTypes/BasicTypes.hs:874: Note [Expanding variables] in Rules.lhs. basicTypes/Id.hs:579:-- See Note [Specialisations and RULES in IdInfo] in IdInfo.lhs basicTypes/Id.hs:689: -- Another good example is in fill_in in PrelPack.lhs. We should be able to basicTypes/DataCon.hs:407: -- See also Note [Data-con worker strictness] in MkId.lhs basicTypes/MkId.hs:1194: See DsUtils.lhs Note [Desugaring seq (1)] and (2) and (3) basicTypes/Var.hs:340:-- Ids, because Id.lhs uses 'mkGlobalId' etc with different types basicTypes/IdInfo.hs:423:-- CafInfo is used to build Static Reference Tables (see simplStg/SRT.lhs). cmm/CmmBuildInfoTables.hs:201:-- Adapted from simpleStg/SRT.lhs, which expects Id's. cmm/cmm-notes:98:SMRep.lhs cmm/cmm-notes:123:CodeGen.lhs cmm/cmm-notes:124:CgBindery.lhs cmm/cmm-notes:126:CgCase.lhs cmm/cmm-notes:127:CgClosure.lhs cmm/cmm-notes:128:CgCon.lhs cmm/cmm-notes:129:CgExpr.lhs cmm/cmm-notes:130:CgLetNoEscape.lhs cmm/cmm-notes:132:CgHeapery.lhs cmm/cmm-notes:135:CgMonad.lhs cmm/cmm-notes:139:CgStackery.lhs cmm/cmm-notes:140:CgTailCall.lhs cmm/cmm-notes:143:ClosureInfo.lhs codeGen/StgCmmBind.hs:744: -- CgConTbls.lhs with a description generated from the data constructor codeGen/StgCmmExpr.hs:704:-- stgSyn/CoreToStg.lhs). Consider this: coreSyn/CorePrep.hs:111:9. Replace (lazy e) by e. See Note [lazyId magic] in MkId.lhs coreSyn/CoreUtils.hs:803:~~~~~~~~~~~~~~~~~~ in CoreUnfold.lhs coreSyn/CoreSyn.hs:245:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs coreSyn/CoreSyn.hs:268:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs coreSyn/CoreSyn.hs:274:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs coreSyn/CoreSyn.hs:289:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs coreSyn/CoreSyn.hs:438:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs coreSyn/CoreSyn.hs:764: -- See Note [Extra args in rule matching] in Rules.lhs coreSyn/CoreSyn.hs:921: -- (see MkId.lhs, calls to mkCompulsoryUnfolding). coreSyn/CoreSyn.hs:1100:-- See Note [Expanding variables] in Rules.lhs coreSyn/CoreSyn.hs:1256:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs coreSyn/CoreSyn.hs:1445:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs coreSyn/CoreArity.hs:612:See Note [Dictionary-like types] in TcType.lhs for why we use coreSyn/CoreFVs.hs:294: -- See Note [Finding rule RHS free vars] in OccAnal.lhs coreSyn/CoreFVs.hs:393:-- This actually happened in the defn of errorIO in IOBase.lhs: deSugar/DsExpr.hs:479: T (recConErr t1 "M.lhs/230/op1") deSugar/DsExpr.hs:481: (recConErr t1 "M.lhs/230/op3") deSugar/DsExpr.hs:486: M.lhs, line 230: missing field op1 was evaluated deSugar/DsExpr.hs:531: other -> recUpdError "M.lhs/230" deSugar/DsForeign.hs:673: -- codeGen/CodeGen.lhs; however, stginit functions have a negative impact deSugar/Check.hs:56: @Match.lhs@ to generate code) deSugar/Check.hs:631:tidy_eqn does more or less the same thing as @tidy@ in @Match.lhs@; ghci/ByteCodeGen.hs:1157: -- the DataCon. See DataCon.lhs for details. ghci/Linker.hs:1109: -- compiler/ghci/Linker.lhs + rts/Linker.c link the ghc.mk:243:# Generate supporting stuff for prelude/PrimOp.lhs hsSyn/HsDecls.hs:371: (DefMeth is defined in Class.lhs) hsSyn/Convert.hs:884: -- Convert.lhs, hence panic iface/IfaceType.hs:467:-- See equivalent function in TypeRep.lhs iface/LoadIface.hs:197: -- compiling Float.lhs, which mentions Float of course main/HscMain.hs:356: -- - filter out the .hs/.lhs source filename if we have one main/TidyPgm.hs:1266: -- split (see Note [Worker-wrapper for bottoming functions] in WorkWrap.lhs main/GHC.hs:1182:{- ToDo: Move the primary logic here to compiler/main/Packages.lhs main/DriverPhases.hs:52: literate pre-processor | .lhs | - | - main/Finder.hs:307:-- not a reexport (this invariant is upheld by @Packages.lhs@) and (2) main/DriverPipeline.hs:827: -- libraries/hpc/tests/function/subdir/tough2.lhs main/DriverPipeline.hs:1687: -- compiler/nativeGen/AsmCodeGen.lhs for another instance main/HscTypes.hs:1066: -- See Note [Overall plumbing for rules] in Rules.lhs main/HscTypes.hs:2704:This stuff is in here, rather than (say) in Linker.lhs, because the Linker.lhs nativeGen/PPC/CodeGen.hs:893: nonvolatile registers differs (see MachRegs.lhs). nativeGen/PIC.hs:748:-- (See PprMach.lhs) nativeGen/RegAlloc/Graph/TrivColorable.hs:73:32% of total compile time and 42% of total alloc when compiling SHA1.lhs from darcs. nativeGen/RegAlloc/Graph/TrivColorable.hs:81:the most efficient variant tried. Benchmark compiling 10-times SHA1.lhs follows. parser/cutils.h:9:// Out-of-line string functions, see PrimPacked.lhs prelude/TysPrim.hs:333:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs prelude/TysPrim.hs:531:defined in \tr{TysWiredIn.lhs}, not here. prelude/PrimOp.hs:188:is done with plain ccalls now (see PrelNumExtra.lhs). prelude/PrimOp.hs:486:@primOpIsCheap@, as used in \tr{SimplUtils.lhs}. For now (HACK prelude/PrelRules.hs:1006: -- See Note [Integer division constant folding] in libraries/base/GHC/Real.lhs prelude/PrelRules.hs:1137:-- See Note [magicDictId magic] in `basicTypes/MkId.lhs` rename/RnPat.hs:238: -- before going on to the RHSes (see RnSource.lhs). rename/RnEnv.hs:1374: -- See Note [Binders in Template Haskell] in Convert.lhs rename/RnExpr.hs:316:wired-in. See the Note about the NameSorts in Name.lhs. simplCore/SimplUtils.hs:1491: poly_id = transferPolyIdInfo var tvs_here $ -- Note [transferPolyIdInfo] in Id.lhs simplCore/SimplUtils.hs:1624:was like this (base/Foreign/C/Err/Error.lhs): simplCore/OccurAnal.hs:286:recursive functions] in Simplify.lhs simplCore/FloatIn.hs:86:SetLevels.lhs module, which tags things with their level numbers. simplCore/SetLevels.hs:830: -- See Note [Computing one-shot info] in Demand.lhs simplCore/SetLevels.hs:1053: mk_poly_bndr bndr uniq = transferPolyIdInfo bndr abs_vars $ -- Note [transferPolyIdInfo] in Id.lhs simplCore/SimplCore.hs:299: -- Reason: see Note [Shadowing] in SpecConstr.lhs simplCore/SimplCore.hs:634: -- See Note [Overall plumbing for rules] in Rules.lhs simplCore/Simplify.hs:55:the simplifier is in SimplCore.lhs. simplCore/Simplify.hs:2161: -- See Note [Data-con worker strictness] in MkId.lhs simplCore/Simplify.hs:2630:The Ord instance of Maybe in PrelMaybe.lhs, for example, took several extra simplStg/UnariseStg.hs:66: where -- See Note [Nullary unboxed tuple] in Type.lhs specialise/SpecConstr.hs:1839: -- See Note [Matching lets] in Rule.lhs specialise/SpecConstr.hs:1845:{- Disabled; see Note [Matching cases] in Rule.lhs stgSyn/CoreToStg.hs:92:-- constructing SRTs (see SRT.lhs). stgSyn/CoreToStg.hs:444: -- See Note [Nullary unboxed tuple] in Type.lhs stranal/DmdAnal.hs:1089:-- See Note [Trimming a demand to a type] in Demand.lhs typecheck/TcForeign.hs:137: -- See Note [Expanding newtypes] in TyCon.lhs typecheck/FamInst.hs:327:-- in FamInstEnv.lhs typecheck/TcExpr.hs:1321:Note [Converting strings] in Convert.lhs typecheck/TcType.hs:259:It's knot-tied back to Var.lhs. There is no reason in principle typecheck/TcType.hs:260:why Var.lhs shouldn't actually have the definition, but it "belongs" here. typecheck/TcEvidence.hs:138: [TcCoercion] -- See [CoAxiom Index] in Coercion.lhs typecheck/TcPat.hs:679:As explained by [Wrappers for data instance tycons] in MkIds.lhs, the typecheck/TcInteract.hs:1524:To achieve this required some refactoring of FunDeps.lhs (nicer typecheck/TcTyClsDecls.hs:1490: -- See Note [GADT record selectors] in MkId.lhs types/Kind.hs:218:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/Kind.hs:234:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/CoAxiom.hs:231:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/TyCon.hs:162:See also Note [Wrappers for data instance tycons] in MkId.lhs types/TyCon.hs:349:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/TypeRep.hs:98:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/Coercion.hs:126:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/Coercion.hs:203:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/Coercion.hs:945:-- of Coercion and the Note [Refl invariant] in types/TypeRep.lhs. types/Type.hs:606:interfaces. Notably this plays a role in tcTySigs in TcBinds.lhs. types/Type.hs:819:-- See Note [GHC Formalism] in coreSyn/CoreLint.lhs types/Type.hs:1324:-- See Note [Kind Constraint and kind *] in Kinds.lhs types/Unify.hs:235:-- thing to do. C.f. Note [Matching variable types] in Rules.lhs utils/Binary.hs:476:-- yes, we need Binary Integer and Binary Rational in basicTypes/Literal.lhs }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9986> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8539: Data.Complex shouldn't use default implementation of (**)
by GHC 23 Feb '15

23 Feb '15

#8539: Data.Complex shouldn't use default implementation of (**) -------------------------------------+------------------------------------- Reporter: jjaredsimpson | Owner: Type: bug | Status: new Priority: low | Milestone: Component: Prelude | Version: 7.6.3 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: Incorrect result Difficulty: Easy (less than 1 | at runtime hour) | Test Case: Blocked By: | Blocking: Related Tickets: | -------------------------------------+------------------------------------- {{{ Prelude Data.Complex> 0**2 0.0 Prelude Data.Complex> 0**2 :: Complex Double NaN :+ NaN Prelude Data.Complex> exp $ 2 * log 0 :: Complex Double NaN :+ NaN Prelude Data.Complex> log 0 :: Complex Double (-Infinity) :+ 0.0 Prelude Data.Complex> 2 * it :: Complex Double (-Infinity) :+ NaN Prelude Data.Complex> exp it :: Complex Double NaN :+ NaN }}} So Complex uses the default implementation of (!**). Then when 2*(-inf :+ 0) is evaluated. We do (2 * -inf - 0*0) :+ (2*0 + -inf*0). Which because of -inf*0 sets the imaginary part to NaN. Then exp (-inf :+ NaN) = exp x cos y :+ exp x sin y which becomes 0 * cos NaN :+ 0 * sin NaN. So we end up with NaN :+ NaN. Ross Paterson suggested: I would say the default implementation of (!**) is wrong: to match the Float/Double instances it should be x !** y = if x == 0 then 0 else exp (log x * y) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8539> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 44

[GHC] #8309: traceEvent truncates to 512 bytes
by GHC 23 Feb '15

23 Feb '15

#8309: traceEvent truncates to 512 bytes ------------------------------------+------------------------------------- Reporter: duncan | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Runtime System | Version: 7.6.3 Keywords: eventlog tracing | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- The `Debug.Trace.traceEvent` (& `traceEventIO`) use a code path that unnecessarily uses printf format strings and uses a fixed size 512-byte buffer, hence truncating all user trace messages at that size. Here is the call path: * `Debug.Trace.traceEvent` (& `traceEventIO`) call `traceEvent#` primop * cmm impl of the primop `stg_traceEventzh` calls C RTS function `traceUserMsg` * `traceUserMsg` calls `traceFormatUserMsg(cap, "%s", msg);`, using the printf format string * `traceFormatUserMsg` uses `postUserMsg` which eventually calls `postLogMsg` * `postLogMsg` does the printf stuff Here's what `postLogMsg` does: {{{ #define BUF 512 void postLogMsg(EventsBuf *eb, EventTypeNum type, char *msg, va_list ap) { char buf[BUF]; nat size; size = vsnprintf(buf,BUF,msg,ap); if (size > BUF) { buf[BUF-1] = '\0'; size = BUF; } .... }}} This is obviously designed for RTS-internal users, not the user path. So the problem starts with this bit: {{{ void traceUserMsg(Capability *cap, char *msg) { traceFormatUserMsg(cap, "%s", msg); } }}} It just should not use that code path. It should call something that directly posts the message, without any silly printf strings. In fact, the code path from the primop down should really be changed to use an explicit given length, rather than using a null-terminated string and having to call strlen. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8309> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5