July 2014 - ghc-tickets

[GHC] #9319: nofib-analyze doesn’t provide per-benchmark compile time/alloc numbers
by GHC 08 Oct '15

08 Oct '15

#9319: nofib-analyze doesn’t provide per-benchmark compile time/alloc numbers -------------------------------------+------------------------------------- Reporter: nomeata | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: NoFib benchmark | Version: 7.8.2 suite | Differential Revisions: Keywords: | Architecture: Operating System: Unknown/Multiple | Unknown/Multiple Type of failure: None/Unknown | Difficulty: Unknown Test Case: | Blocked By: Blocking: | Related Tickets: -------------------------------------+------------------------------------- The compile time and allocation numbers calculated by nofib are only available per module. While this is useful to find the cause of problems, it would be nice if these numbers were also accumulated for each benchmark, for easy of general analysis and uniformity, especially with automated monitoring of benchmark numbers. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9319> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #9058: System.IO.openTempFile does not scale
by GHC 07 Oct '15

07 Oct '15

#9058: System.IO.openTempFile does not scale ------------------------------------+------------------------------------- Reporter: slyfox | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.2 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- In search of a bug in darcs http://bugs.darcs.net/issue2364 i've notice very bad property of openTempFile: it's pattern is very predictable and has O(n^2) of already created temp files. Predictability allows very fun bugs survive in buggy programs, like: {{{ thread1: (fn, fh) <- openTempFile "." "hello" renameFile fn "something" -- some time after when (some_rare_buggy_condition) $ -- oops, reused temp name, but too late, other thread killed it writeFileFile fn thread2: (fn, fh) <- openTempFile "." "hello" workWithFn fn -- nobody should touch it, right? }}} It's _very_ hard to debug data corruption when all temp files are named "foo${pid}" and sometimes "foo${pid+1}". And more serious bug: the more threads you have trying to create similar temps performance drops significantly: Attached program shows the following numbers: {{{ $ time ./bench-temps same 2000 real 0m2.795s user 0m1.516s sys 0m1.190s $ time ./bench-temps diff 2000 real 0m0.161s user 0m0.043s sys 0m0.115s }}} It's O(N^2) growing open() storm. https://github.com/ghc/ghc/blob/master/libraries/base/System/IO.hs#L465 {{{ FileExists -> findTempName (x + 1) }}} This is the source of the problem. I'd suggest always using random name for it. For portability reasons I suggest adding at least insecure random '''rand()''' value from C library. That way we will succeed in opening temp file at the first attempt. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9058> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #8364: equip GHC with an accurate internal model of floating point
by GHC 07 Oct '15

07 Oct '15

#8364: equip GHC with an accurate internal model of floating point ----------------------------------------------+---------------------------- Reporter: carter | Owner: Type: feature request | Status: new Priority: high | Milestone: 7.10.1 Component: Compiler | Version: 7.6.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Project (more than a week) | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: ----------------------------------------------+---------------------------- currently there really isn't a systematic / portable story for doing compile time reduction / evaluation for floating point numbers. there are some folks in the larger haskell community who are working on high precision soft float implementations / modelling. Once that work becomes available, might be worth exploring having something like that for doing controlled high precision evaluation of intermediate expressions in ghc proper. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8364> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #8335: Create more specialized entries to GC
by GHC 06 Oct '15

06 Oct '15

#8335: Create more specialized entries to GC ------------------------------------+------------------------------------- Reporter: jstolarek | Owner: Type: task | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.7 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- Consider this Cmm code (taken from #8326): {{{ {offset cFd: // stack check if ((Sp + -16) < SpLim) goto cFr; else goto cFs; cFr: // not enough place on the stack, call GC R2 = R2; R1 = A.f_closure; call (stg_gc_fun)(R2, R1) args: 8, res: 0, upd: 8; cFs: // scrutinize (x ># 0#) _sEU::I64 = R2; _sEV::I64 = %MO_S_Gt_W64(R2, 0); if (_sEV::I64 != 1) goto cFg; else goto cFo; cFg: // False branch Hp = Hp + 16; if (Hp > HpLim) goto cFy; else goto cFx; cFy: // not enough heap, call GC HpAlloc = 16; I64[Sp - 16] = cFf; R1 = _sEV::I64; I64[Sp - 8] = _sEU::I64; Sp = Sp - 16; call stg_gc_unbx_r1(R1) returns to cFf, args: 8, res: 8, upd: 8; cFf: // re-do the False branch _sEU::I64 = I64[Sp + 8]; Sp = Sp + 16; _sEV::I64 = R1; goto cFg; cFx: // RHS of False branch I64[Hp - 8] = GHC.Types.I#_con_info; I64[Hp] = -_sEU::I64; R1 = Hp - 7; call (P64[Sp])(R1) args: 8, res: 0, upd: 8; cFo: // True branch Hp = Hp + 16; if (Hp > HpLim) goto cFv; else goto cFu; cFv: // not enough heap, call GC HpAlloc = 16; I64[Sp - 16] = cFn; R1 = _sEV::I64; I64[Sp - 8] = _sEU::I64; Sp = Sp - 16; call stg_gc_unbx_r1(R1) returns to cFn, args: 8, res: 8, upd: 8; cFn: // re-do the True branch _sEU::I64 = I64[Sp + 8]; Sp = Sp + 16; _sEV::I64 = R1; goto cFo; cFu: // RHS of True branch I64[Hp - 8] = GHC.Types.I#_con_info; I64[Hp] = _sEU::I64; R1 = Hp - 7; call (P64[Sp])(R1) args: 8, res: 0, upd: 8; } }}} We perform two different calls to garbage collection here. Calling GC after making initial stack check obviously can't use any stack, so we pass parameter to GC in regsiters: {{{ R2 = R2; R1 = A.f_closure; call (stg_gc_fun)(R2, R1) args: 8, res: 0, upd: 8; }}} But in the calls that are made later on inside the function we do put some parameters on the stack, thus increasing stack usage: {{{ cFv: // not enough heap, call GC HpAlloc = 16; I64[Sp - 16] = cFn; R1 = _sEV::I64; I64[Sp - 8] = _sEU::I64; Sp = Sp - 16; call stg_gc_unbx_r1(R1) returns to cFn, args: 8, res: 8, upd: 8; }}} If we had more specialized entries to GC that would allow us to pass parameters in registers (or in some fixed memory location associated with TCO?) then this function would not require any stack at all. Note that there is a separate issue here - `_sEV` variable is in fact dead and should be eliminated (see #8327). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8335> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #8920: Alternative GADT syntax
by GHC 05 Oct '15

05 Oct '15

#8920: Alternative GADT syntax ----------------------------------------------+---------------------------- Reporter: guest | Owner: Type: feature request | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler (Parser) | Version: Keywords: gadts | 7.8.1-rc2 Architecture: Unknown/Multiple | Operating System: Difficulty: Moderate (less than a day) | Unknown/Multiple Blocked By: | Type of failure: Other Related Tickets: | Test Case: | Blocking: ----------------------------------------------+---------------------------- An alternative syntax for GADTs Instead of {{{ data Term x where K :: Term (a -> b -> a) S :: Term ((a -> b -> c) -> (a -> b) -> a -> c) Const :: a -> Term a (:@) :: Term (a -> b) -> (Term a) -> Term b }}} You can write {{{ data Term x = K :: Term (a -> b -> a) | S :: Term ((a -> b -> c) -> (a -> b) -> a -> c) | Const a :: Term a | (:@) (Term (a -> b)) (Term a) :: Term b }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8920> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #9238: Negative zero broken
by GHC 03 Oct '15

03 Oct '15

#9238: Negative zero broken --------------------------+------------------------------------------------ Reporter: | Owner: augustss | Status: new Type: bug | Milestone: Priority: normal | Version: 7.8.2 Component: | Operating System: Windows Compiler | Type of failure: Incorrect result at runtime Keywords: | Test Case: Architecture: | Blocking: Unknown/Multiple | Difficulty: | Unknown | Blocked By: | Related Tickets: | --------------------------+------------------------------------------------ Try the following program {{{ compareDouble :: Double -> Double -> Ordering compareDouble x y = case (isNaN x, isNaN y) of (True, True) -> EQ (True, False) -> LT (False, True) -> GT (False, False) -> -- Make -0 less than 0 case (x == 0, y == 0, isNegativeZero x, isNegativeZero y) of (True, True, True, False) -> LT (True, True, False, True) -> GT _ -> x `compare` y main = do let l = [-0, 0] print [ (x, y, compareDouble x y) | x <- l, y <- l ] }}} Compile and run with -O0 {{{ $ ghc -O0 D.hs [1 of 1] Compiling Main ( D.hs, D.o ) Linking D.exe ... $ ./D [(-0.0,-0.0,EQ),(-0.0,0.0,LT),(0.0,-0.0,GT),(0.0,0.0,EQ)] }}} This is the correct output. Compile and run with -O1 {{{ $ ghc -O1 D.hs [1 of 1] Compiling Main ( D.hs, D.o ) Linking D.exe ... $ ./D [(-0.0,-0.0,LT),(-0.0,0.0,LT),(0.0,-0.0,EQ),(0.0,0.0,EQ)] }}} This is wrong. Put a NOINLINE pragma on compareDouble: {{{ $ ghc -O1 D.hs [1 of 1] Compiling Main ( D.hs, D.o ) Linking D.exe ... $ ./D [(-0.0,-0.0,EQ),(-0.0,0.0,EQ),(0.0,-0.0,EQ),(0.0,0.0,EQ)] }}} This is wrong in a different way. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9238> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 17

[GHC] #9342: Branchless arithmetic operations
by GHC 28 Sep '15

28 Sep '15

#9342: Branchless arithmetic operations -------------------------------------+------------------------------------- Reporter: hvr | Owner: Type: feature request | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler | Version: 7.8.3 (CodeGen) | Differential Revisions: Keywords: | Architecture: Operating System: Unknown/Multiple | Unknown/Multiple Type of failure: None/Unknown | Difficulty: Unknown Test Case: | Blocked By: Blocking: | Related Tickets: -------------------------------------+------------------------------------- While working on #9281 I noticed sub-optimal code generation for common arithmetic operations on `Int`. Below are some examples of code generation, where the branchless versions may be desirable on modern CPUs: = `abs` {{{#!hs -- base version absI# :: Int# -> Int# absI# i# = i'# where !(I# i'#) = abs (I# i#) -- optimized version optAbsI# :: Int# -> Int# optAbsI# i# = (i# `xorI#` nsign) -# nsign where -- nsign = i# <# 0# nsign = uncheckedIShiftRA# i# (WORD_SIZE_IN_BITS# -# 1#) }}} results in {{{#!asm absI#_info: _c1To: testq %r14,%r14 jge _c1Tx _c1Ty: movq %r14,%rbx negq %rbx jmp *(%rbp) _c1Tx: movq %r14,%rbx jmp *(%rbp) optAbsI#_info: _c1SX: movq %r14,%rax sarq $63,%rax movq %r14,%rbx xorq %rax,%rbx subq %rax,%rbx jmp *(%rbp) }}} = `signum` {{{#!hs sgnI# :: Int# -> Int# sgnI# i# = i'# where !(I# i'#) = signum (I# i#) optSgnI# :: Int# -> Int# optSgnI# x# = (x# ># 0#) -# (x# <# 0#) }}} {{{#!asm sgnI#_info: _c27W: testq %r14,%r14 jl _c283 _c284: testq %r14,%r14 jne _c28a _c28b: xorl %ebx,%ebx jmp *(%rbp) _c283: movq $-1,%rbx jmp *(%rbp) _c28a: movl $1,%ebx jmp *(%rbp) optSgnI#_info: _c26P: testq %r14,%r14 setl %al movzbl %al,%eax testq %r14,%r14 setg %bl movzbl %bl,%ebx subq %rax,%rbx jmp *(%rbp) }}} = `compare` {{{#!hs cmpI# :: Int# -> Int# -> Int# cmpI# x# y# = dataToTag# (compare (I# x#) (I# y#)) -- returns 0, 1 or 2, can be fed to tagToEnum# :: Ordering optCmpI# :: Int# -> Int# -> Int# optCmpI# x# y# = 1# +# (x# ># y#) -# (x# <# y#) }}} {{{#!asm cmpI#_info: _c25s: cmpq %rsi,%r14 jl _c25z _c25A: cmpq %rsi,%r14 je _c25M _c25N: movl $2,%ebx jmp *(%rbp) _c25z: xorl %ebx,%ebx jmp *(%rbp) _c25M: movl $1,%ebx jmp *(%rbp) optCmpI#_info: _c24N: cmpq %rsi,%r14 setl %al movzbl %al,%eax movl $1,%ebx subq %rax,%rbx cmpq %rsi,%r14 setg %al movq %rbx,%rcx movzbl %al,%ebx addq %rcx,%rbx jmp *(%rbp) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9342> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #9079: Foreign.C.Types in haskell2010
by GHC 27 Sep '15

27 Sep '15

#9079: Foreign.C.Types in haskell2010 -----------------------------------------+--------------------------------- Reporter: tensor5 | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: libraries/haskell2010 | Version: 7.8.2 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: None/Unknown Blocked By: | Test Case: Related Tickets: | Blocking: -----------------------------------------+--------------------------------- According to Haskell 2010, a marshallable foreign type has to be a `newtype` of a basic type with exposed constructor. Why constructors are not exposed in the module `Foreign.C.Types` from `haskell2010`? When I replace `base` with `haskell2010` in {{{ module Test where import Foreign.C foreign import ccall unsafe "sys/mount.h umount" umount :: CString -> IO CInt }}} I get {{{ Unacceptable result type in foreign declaration: IO CInt When checking declaration: foreign import ccall unsafe "static sys/mount.h umount" umount :: CString -> IO CInt }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9079> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #8832: Constant-folding regression wrt `clearBit (bit 0) 0 `
by GHC 24 Sep '15

24 Sep '15

#8832: Constant-folding regression wrt `clearBit (bit 0) 0 ` ------------------------------+-------------------------------------------- Reporter: hvr | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.8.1 Component: Compiler | Version: 7.8.1-rc2 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Runtime performance bug Unknown/Multiple | Test Case: Difficulty: Unknown | Blocking: Blocked By: | Related Tickets: | ------------------------------+-------------------------------------------- While implementing `zeroBits` (see [83bd2f5fc7e/base]) I noticed that constant folding of the expression `clearBit (bit 0) 0` regressed (and improved at the same time) from GHC 7.6.3 to GHC 7.8.1, specifically, the following module {{{#!haskell {-# LANGUAGE CPP #-} module M where import Data.Bits import Data.Int import Data.Word #define T(s,T) \ s :: T ; \ s = clearBit (bit 0) 0 ; \ T(i,Int) T(i8,Int8) T(i16,Int16) T(i32,Int32) T(i64,Int64) T(w,Word) T(w8,Word8) T(w16,Word16) T(w32,Word32) T(w64,Word64) T(z,Integer) }}} compiled with GHC 7.8.1RC2 results in the following Core output: {{{#!haskell -- GHC 7.8.1RC2 i = I# (andI# 1 (notI# 1)) i8 = I8# 0 i16 = I16# 0 i32 = I32# 0 i64 = I64# 0 w = W# (__word 0) w8 = W8# (__word 0) w16 = W16# (__word 0) w32 = W32# (__word 0) w64 = W64# (__word 0) z2 = $w$cbit 0 z1 = complementInteger z2 z = andInteger z2 z1 }}} Thus, `i` and `z` are not properly constant-folded in GHC 7.8.1RC2. With GHC 7.6.3, however, `i` and `z` were properly folded to `0`: {{{#!haskell -- GHC 7.6.3 i = I# 0 i8 = case $fBitsInt8_$cbit i of _ { I8# x#_aDf -> case $fBitsInt8_$cbit i of _ { I8# x#1_aDr -> I8# (word2Int# (and# (int2Word# x#_aDf) (xor# (int2Word# x#1_aDr) (__word 18446744073709551615)))) } } i16,i32,i64 -- equivalent to i8 w = W# (__word 0) w8 = case $fBitsWord8_$cbit i of _ { W8# x#_aEV -> case $fBitsWord8_$cbit i of _ { W8# x#1_aF5 -> W8# (and# x#_aEV (xor# x#1_aF5 (__word 255))) } } w16,w32,w64 -- equivalent to w8 z = __integer 0 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8832> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 32

[GHC] #8213: Bad error message when using lazy IO to read from closed handle
by GHC 22 Sep '15

22 Sep '15

#8213: Bad error message when using lazy IO to read from closed handle ------------------------------------+-------------------------------------- Reporter: nh2 | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Runtime System | Version: 7.6.3 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: Documentation bug Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+-------------------------------------- Today I accidentally wrote this code: {{{ filesEqual :: FilePath -> FilePath -> IO Bool filesEqual f1 f2 = do equal <- withBinaryFile f1 ReadMode $ \f1 -> withBinaryFile f2 ReadMode $ \f2 -> c1 <- BSL.hGetContents h1 c2 <- BSL.hGetContents h2 return (c1 == c2) return $! equal }}} The problem is that I should have done the forcing before the files are closed: {{{ return $! (c1 == c2) }}} I got the incredibly unhelpful error message {{{ myprogram: myfile.txt: illegal operation }}} It took me 2 hours and strace to find out what was going on, especially because around every file open() there are multiple things that map top IllegalOperation (e.g. the ENOTTY from the ioctl that checks whether the device is a TTY). Can we give a better error message here, at least mentioning that the problem has to do with closed (and GC'd) handles? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8213> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3