- Haskell-Cafe - Haskell.org

Segfault in the libc malloc using FFI (occurs only in x86_64)
by Vincent Gerard 20 Aug '11

20 Aug '11

Hi cafe, I have been struggling with this issue for the past days. I have investigated at the Haskell, C and even at assembly level... Perhaps I'm missing something big ?? I hope someone familiar with FFI could help me on this Segfault. My env is: Linux Debian 3.0.0-1, SMP, GHC 7.0.4, x86_64, eglibc 2.13-10 (This issues occurs as well on others people with /= env, but only with x86_64 arch) The bug is in the hsmagick library (FFI bindings to GraphicsMagick), for which I am the maintainer. -- So here is a reproducer (Works fine in 32 bit, segfault in 64). You should have hsmagick + GraphicsMagick dev libs installed. import Graphics.Transform.Magick.Images main = do initializeMagick c <- readImage "image.jpg" putStrLn "End" -- Executing this program, compiled with standard options, a Segfault occurs at runtime. If I compile this small program with -threaded option an assertion error is thrown at runtime: main: malloc.c:3096: sYSMALLOc: Assertion `(old_top == (((mbinptr) (((char *) &((av)->bins[((1) - 1) * 2])) - __builtin_offsetof (struct malloc_chunk, fd)))) && old_size == 0) || ((unsigned long) (old_size) >= (unsigned long)((((__builtin_offsetof (struct malloc_chunk, >fd_nextsize))+((2 * (sizeof(size_t))) - 1)) & ~((2 * (sizeof(size_t))) >- 1))) && ((old_top)->size & 0x1) && ((unsigned long)old_end & >pagemask) == 0)' failed. This assertion error is cryptic for me... but it maybe could help someone here. However, having a C experience, investigating a Segfault is in my habits, so let's install debugs symbols on everything and dig with gdb. Here is the backtrace: Program received signal SIGSEGV, Segmentation fault. 0x00007ffff510292c in malloc_consolidate (av=0x7ffff540fe60) at malloc.c:5161 (this is inside the eglibc source) 5161 unlink(p, bck, fwd); (gdb) bt #0 0x00007ffff510292c in malloc_consolidate (av=0x7ffff540fe60) at malloc.c:5161 #1 0x00007ffff5104d64 in _int_malloc (av=0x7ffff540fe60,bytes=8520) at malloc.c:4373 #2 0x00007ffff5107420 in __libc_malloc (bytes=8520) at malloc.c:3660 #3 0x00007ffff67bf311 in CloneImageInfo (image_info=0x0) at magick/image.c:1012 #4 0x0000000000439dfa in hsmagickzm0zi5_GraphicsziTransformziMagickziFFIHelpers_mkNewImageInfo1_info () Yay, the Segfault is deep in the libc after a regular malloc (8520bytes) (I dug into the GraphicsMagick source starting from magick/image.c:1012) And this libc call is called by a call to CloneImageInfo(NULL) from Haskell, which is correct as it is graphicsMagick way to allocate an empty ImageInfo data structure. (http://www.graphicsmagick.org/api/image.html#cloneimageinfo) I tried in a C program to execute (after initializing ImageMagick) image_info=CloneImageInfo((ImageInfo *) NULL); And of course .... no Segfault... Let's look at the FFI call... This FFI call is the done by the mkNewImageInfo as shown in the trace. The Haskell FFI layer here has done the right job, calling the C function with the right args (NULL) Here is all the code related to mkNewImageInfo (the last Haskell part we see in the backtrace) --- mkNewImageInfo :: IO (ForeignPtr HImageInfo) mkNewImageInfo = mkFinalizedImageInfo =<< mkNewImageInfo_ mkFinalizedImageInfo :: Ptr HImageInfo -> IO (ForeignPtr HImageInfo) mkFinalizedImageInfo = newForeignPtr imageInfoFinalizer mkNewImageInfo_ :: IO (Ptr HImageInfo) mkNewImageInfo_ = clone_image_info nullPtr -- CALL before the Segfault destroyImageInfo :: Ptr HImageInfo -> IO () destroyImageInfo = destroy_image_info foreign import ccall "static magick/api.h &DestroyImageInfo" imageInfoFinalizer :: FunPtr (Ptr HImageInfo -> IO ()) foreign import ccall "static magick/api.h CloneImageInfo" clone_image_info :: Ptr HImageInfo -> IO (Ptr HImageInfo) -- To me, this code is right (and works perfectly in 32 bit). Furthermore, as the call is made with a nullPtr arg, no data structure is used ... I tried this code without using ForeignPtr Finalizers (hsmagick <= 0.4) and the Segfault occurs as well ... I even had a look at the generated assembly, which also looks right... It seems there is a real segmentation error, or an illegal access to a memory page... But I could not find the root cause of this error. And an additionnal piece to the puzzle, when running the program inside valgrind, there is no Segfault and the program works as expected. So if anyone have an idea on how only on 64bits arch an Haskell FFI call could led to a Segfault in the libc malloc, it would make my day, perhaps my week :) Thanks ! Vincent Gerard

3 2

Simon PJ in Australia
by Simon Peyton-Jones 20 Aug '11

20 Aug '11

This is a message for Australian FPers -- sorry to spam the rest of you. I'm crossing the planet to Australia for YOW (Melbourne Dec 1-2, Brisbane Dec 5-6). http://www.yowconference.com.au/YOW2011/ I'm committed for the conference dates of course, but I could spend a few days before or afterwards, hanging out with some of you guys. Hack on GHC, give a talk, eat pizza, university departments, Haskell users groups, surfing (on water, that is).... Do drop me (not the list) a line if any of that sounds fun. Simon PS: John Hughes is coming to Yow too, so you could try luring him into your lair too.

5 4

Question about data
by Paul Reiners 19 Aug '11

19 Aug '11

I've created a simple type declaration: data MathExpression = Float | Add MathExpression MathExpression | Subtract MathExpression MathExpression | Multiply MathExpression MathExpression | Divide MathExpression MathExpression deriving (Show) Now how do I create an instance of MathExpression which is just a Float? This doesn't work: *Main> let pi = 3.14 :: MathExpression <interactive>:1:10: No instance for (Fractional MathExpression) arising from the literal `3.14' Possible fix: add an instance declaration for (Fractional MathExpression) In the expression: 3.14 :: MathExpression In an equation for `pi': pi = 3.14 :: MathExpression

5 4

why is Random in System?
by Evan Laforge 19 Aug '11

19 Aug '11

I've noticed there's a convention to put modules having to deal with randomness into System.Random. I thought System was for OS interaction? Granted getting a random seed usually means going to the OS, but isn't the rest of it, like generating random sequences, distributions, selecting based on probability, shuffling, etc. all non-OS related algorithms? I'm not sure where I would expect Random to go, perhaps Math or maybe the toplevel, but under System seems, well, random... I notice random-fu puts it under Data, which is also not where I'd look, except that you always look in Data because everything goes into Data... but algorithms dealing with random numbers aren't really data structures either, are they?

8 14

Can I have a typeclass for topological spaces?
by Grigory Sarnitskiy 19 Aug '11

19 Aug '11

Hello! I just wonder whether it is possible to have a typeclass for topological spaces?

3 3

Failed link to mixed-language shared object.
by David Banas 19 Aug '11

19 Aug '11

Hi all, Does this trigger recollection in anyone: dbanas@dbanas-eeepc:~/prj/haskell/AMIParse/trunk$ make ghc -dynamic -o ami_test -L. -lami ami_test.o ./libami.so: undefined reference to `__stginit_haskell98_MarshalArray_' ./libami.so: undefined reference to `__stginit_haskell98_MarshalError_' collect2: ld returned 1 exit status make: *** [ami_test] Error 1 dbanas@dbanas-eeepc:~/prj/haskell/AMIParse/trunk$ ? Know what I need to do? Thanks, -db

2 2

Problem with types
by Anupam Jain 19 Aug '11

19 Aug '11

Hi all, Suppose I have a compound data type - data M o = M (String,o) Now, I can define a function that works for ALL M irrespective of o. For example - f :: M o -> M o f (M (s,o)) = M (s++"!", o) I can also use this function in an expression, applying it to different types without problem - p = (m1',m2') where m1 = M ("1", ()) m2 = M ("2", True) m1' = f m1 m2' = f m2 Main*> p (M ("1!",()),M ("2!",True)) However, if I try to parameterise over the function 'f' it does not work! - p f = (m1',m2') where m1 = M ("1", ()) m2 = M ("2", True) m1' = f m1 m2' = f m2 It doesn't even typecheck, producing the error - "Couldn't match expected type 'Bool' with actual type '()'" Is there a particular reason for this? How can I define a function like 'p' within Haskell? Thanks, Anupam Jain

4 3

Compilation error in Chapter 5 of "Real World Haskell"
by Paul Reiners 18 Aug '11

18 Aug '11

I'm trying to do the following from Chapter 5 of "Real World Haskell": Our choice of naming for the source file and function is deliberate. To create an executable, *ghc* expects a module named Main that contains a function named main. The main function is the one that will be called when we run the program once we've built it. 6 comments<comments:%20show%20/%20hide> *ghc -o simple Main.hs SimpleJSON.o* ---from http://book.realworldhaskell.org/read/writing-a-library-working-with-json-d… When I do that, I get this error: C:\ch05>ghc -o simple Main.hs SimpleJSON.o [2 of 2] Compiling Main ( Main.hs, Main.o ) Linking simple.exe ... SimpleJSON.o:fake:(.data+0x0): multiple definition of `SimpleJSON_getArray_closure' .\SimpleJSON.o:fake:(.data+0x0): first defined here SimpleJSON.o:fake:(.text+0x54): multiple definition of `SimpleJSON_getArray_info' .\SimpleJSON.o:fake:(.text+0x54): first defined here SimpleJSON.o:fake:(.data+0x4): multiple definition of `SimpleJSON_getObject_closure' .\SimpleJSON.o:fake:(.data+0x4): first defined here What's going wrong here?

3 2

writing a good Setup.hs is *hard* (was: broken on install encoding 0.6.6)
by Rogan Creswick 18 Aug '11

18 Aug '11

On Wed, Aug 17, 2011 at 6:29 AM, Gang <yugang.bj(a)gmail.com> wrote: > > sorry to bother > Not in the least! I was trying to figure out what problem you were encountering, and I learned a lot. I never did recreate your problem, but in the process I ran into a whole host of other annoyances and strange situations relating to slightly broken cabal files and cabal build intricacies that confound our existing tools. I don't mean to pick on encoding or HaXml - I've seen the same kinds of issues in many, many Haskell packages (including most of the packages I've written myself) these are just the ones I ran into today. (and I am quite serious when I say that writing a good Setup.hs is very, very, difficult). My objective is just to motivate a minor change to cabal that would help to make this easier. On to the build problems: encoding-0.6.6 uses it's own source during the build process! It actually has to be partially compiled before the build tool can build encoding (!). I'm *amazed* that this actually works at all (impressed, too), and indeed, it doesn't work with cabal-dev, which is unfortunate: cabal-dev would have identified one of the dependency issues in HaXml 1.19 (pretty is an undocumented dependency, although I am unsure if that was true at the time 1.19 and encoding-0.6.6 were released. I haven't tracked the pretty dependency down to the exact source - it could be a transitive dep introduced by a recent version of something HaXml 1.19 needs. Per my subject line - this is amazingly difficult - keep in mind that we're not talking about building *packages*, this is just to build the *build tool*, and none of this is documented. How long before it is effectively impossible to recreate the circumstances that result in a successful encoding-0.6.6 build?). cabal-dev can't build encoding-0.6.6 because the encoding source that is referenced from encoding's Setup.hs depends on (at least) HaXml >= 1.19 && < 1.22.5; this is OK with cabal, since cabal-install installs a necessary HaXml into your user package db prior to building encoding's Setup.hs (modulo the missing upper version bound). Therefore HaXml "just happens" to be visible when cabal builds Setup.hs -- a build step that cabal-dev isn't capable of controlling (and probably never will be -- we could intercept system calls, but that's difficult when only supporting one OS, much less three). You can, of course, manually install HaXml to your local package db and then cabal-dev install encodings, but that's somewhat of a hack (and you'll find that the current HaXml won't work for the current encodings release, as Gang Yu did). Adding a section to the cabal file that specifies build dependencies for Setup.hs would make some very important parts of this process obvious - and it would *greatly* assist in debugging failing builds, not to mention allowing us to make use of the extensive libraries on hackage to write more expressive build systems. (We could actually use openshake in Setup.hs, once it's released, for example!) I've floated the idea past a number of Haskell programmers with positive responses, and there is a GSOC ticket that provides some more details (the project was not selected - no students proposed to work on it): http://hackage.haskell.org/trac/summer-of-code/ticket/1602 --Rogan > Regards > Gang > On 08/17/2011 04:59 PM, Gang Yu wrote: > > hi, cafe: > > I go through to the encoding package installation problem again. > > cabal install encoding > Resolving dependencies... > /tmp/encoding-0.6.68093/encoding-0.6.6/dist/setup/setup: 4: Syntax error: > ";" unexpected > > when I have a check on the this issue, I get: > > file > ~/.cabal/packages/hackage.haskell.org/encoding/0.6.6/encoding-0.6.6/dist/setup/setup > /home/gang/.cabal/packages/hackage.haskell.org/encoding/0.6.6/encoding-0.6.6/dist/setup/setup: > ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked > (uses shared libs), for GNU/Linux 2.6.27, not stripped > > My machine is: > > gang@gang-laptop:~$ uname -a > Linux gang-laptop 2.6.38-8-generic #42-Ubuntu SMP Mon Apr 11 03:31:50 UTC > 2011 i686 i686 i386 GNU/Linux > > is this related? How can I manually install the package from scratch? Any > helps or directs will be appreciated > > Thanks > Gang > > > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe > >

3 3

Gaussian elimination
by Tobias Nipkow 18 Aug '11

18 Aug '11

Hi, I came up with the following extremely compact version of Gaussian elimination for matrices represented as functions. I searched the web but found nothing resembling it, probably because of its inefficiency. Has anybody seen something like it before? Thanks Tobias gauss :: Int -> (Int -> Int -> Rational) -> Maybe (Int -> Int -> Rational) gauss 0 a = Just a gauss (n+1) a = case dropWhile (\i -> a i n == 0) [0..n] of [] -> Nothing (k:_) -> let ak' j = a k j / a k n a' i = if i == k then ak' else (\j -> a i j - a i n * ak' j) in gauss n (swap k n a') swap :: Int -> Int -> (Int -> a) -> (Int -> a) swap k n f = g where g x | x == k = f n | x == n = f k | otherwise = f x

1 0