- Glasgow-haskell-users

OSThreads in the RTS
by Cristian Perfumo 26 Sep '07

26 Sep '07

Hi all, I am diving into the RTS, trying to identify where an OSThread is created and where it is destroyed (join). The idea is to read some hardware counters when it is just created and read them again when it is about to be destroyed (for the time being let's ignore, for the sake of understanding, the possibility of thread migration and context switches). Does anyone know where to place these reads? As shown in the code below, I tried "printfing" within "workerStart" (in the places where I thought I should read the counters), but it doesn't seem to print the same throughout several executions. I mean, running with -N2 and forking 2 Haskell threads (plus the main one), sometimes I get three "Scheduling" and two "Unscheduling" and sometimes I get only one "Unscheduling" which discourages me about reading the counters there. Any clue about the reason? Probably "workerStart" is not the place to put it, I don't know. Any light thrown on it or any documentation to read would be welcome. Cheers Cristian Perfumo void workerStart(Task *task) { Capability *cap; // See startWorkerTask(). ACQUIRE_LOCK(&task->lock); cap = task->cap; RELEASE_LOCK(&task->lock); // set the thread-local pointer to the Task: taskEnter(task); printf ("Scheduling %p\n", task); // schedule() runs without a lock. cap = schedule(cap,task); printf ("Unscheduling %p\n", task); // On exit from schedule(), we have a Capability. releaseCapability(cap); workerTaskStop(task); }

2 1

Re[2]: Hackage and GHC 6.8
by Bulat Ziganshin 24 Sep '07

24 Sep '07

Hello Thomas, Monday, September 24, 2007, 6:42:11 PM, you wrote: > This is not a pretty solution and also has the problem of incorrectly > dealing with changes to the base interface. This, IMO, is due to > keeping the name 'base' instead of opting to clarify the > incompatibilities by choosing "base2". it's not yet too late to fix it. i think that this Base2 Hell is serious problem and we need to develop and introduce solution to it before we can issue GHC 6.8.1 from my POV, the best solution is to provide base 1.0.* package that just reexports all the functionality that was available in ghc 6.6; and base 2.0 package - the current one -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

1 0

Re: [Haskell-cafe] GHC 6.7 on Windows / containers-0.1 package?
by Victor Nazarov 21 Sep '07

21 Sep '07

> Is this still up-to-date with the way GHC/GHCi work internally? Then > I'll certainly check it out. I think it isn't since STG is pretty much different with the G-machine describes in the book. The original paper about Spineless Tagless G-machine is here [1]. It is quite verbose. Recent paper "How to make a fast curry: push/enter vs eval/apply" [2] is much smaller, but it represents more recent state of GHC and can be read independently with the first paper. The book from 1987 [3] is a very good and detailed overview of the subject and it is valuable as itself, no matter how it actually implemented now. -- 1. https://research.microsoft.com/users/simonpj/papers/spineless-tagless-gmach… 2. https://research.microsoft.com/users/simonpj/Papers/eval-apply/index.htm 3. https://research.microsoft.com/users/simonpj/Papers/slpj-book-1987/index.htm

1 0

Additional thunk for function
by Victor Nazarov 21 Sep '07

21 Sep '07

What is the purpose for GHC to allocate a thunk for some functions? Why Test.map is not a function, but updatable thunk, which should become equal to the function just after the first call? Here is the details: > % ghc -c test.hs -ddump-stg > > ==================== STG syntax: ==================== > Test.map = > \u [] > let { > map1_sdR = > \r [f_sdN ds_sdI] > case ds_sdI of wild_sdU { > [] -> [] []; > : x_sdM xs_sdQ -> > let { sat_sdT = \u [] map1_sdR f_sdN xs_sdQ; } in > let { sat_sdP = \u [] f_sdN x_sdM; } in : [sat_sdP sat_sdT]; > }; > } in map1_sdR; > SRT(Test.map): [] > > > % cat test.hs > module Test where > > map f [] = [] > map f (x:xs) = f x : Test.map f xs -- Thanks in advance Victor

3 3

Additional thunk for function
by Victor Nazarov 21 Sep '07

21 Sep '07

I have reproduced it on 2 Windows machines with ghc 6.6 and 6.6.1 Here is the version information for ghc used to generate output in original message. % ghc -v Glasgow Haskell Compiler, Version 6.6, for Haskell 98, compiled by GHC version 6.6 Using package config file: C:\Program Files\Visual Haskell\package.conf wired-in package base mapped to base-2.0 wired-in package rts mapped to rts-1.0 wired-in package haskell98 mapped to haskell98-1.0 wired-in package template-haskell mapped to template-haskell-2.0 Hsc static flags: -static -- vir http://vir.comtv.ru/

1 0

FFI program seg faulting
by Ben Gaster 21 Sep '07

21 Sep '07

Hi, I am trying to build a very simple Haskell function (foo) which I can then call from C, using an example I got from the GHC FFI Wiki. The problem is that the resulting program fails to run with a seg fault. Here is the code: ---Haskell Module start (Foo.hs) module Foo where foreign export ccall foo :: Int -> IO Int foo :: Int -> IO Int foo n = return (length (f n)) f :: Int -> [Int] f 0 = [] f n = n:(f (n-1)) ---Haskell Module end /*- C module start foo.c*/ #include <stdio.h> #include "Foo_stub.h" int main(void) { int x = foo(100); //printf("x = %d\n", x); return 0; } /*- C module end */ The I build the a.out with the commands: ghc -fglasgow-exts -fffi -c Foo.hs ghc -fglasgow-exts -fffi Foo.o Foo_stub.o foo.c Now if I run the resulting a.out then I get: Segmentation fault Quickly putting this through GDB I get: gdb ./a.out GNU gdb Red Hat Linux (6.5-16.el5rh) Copyright (C) 2006 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. There is absolutely no warranty for GDB. Type "show warranty" for details. This GDB was configured as "i386-redhat-linux-gnu"... (no debugging symbols found) Using host libthread_db library "/lib/libthread_db.so.1". run Starting program: /home/brg/dev/haskell/ffi/a.out (no debugging symbols found) (no debugging symbols found) (no debugging symbols found) [Thread debugging using libthread_db enabled] [New Thread 7870144 (LWP 22448)] Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 7870144 (LWP 22448)] 0x0805c313 in allocateLocal () (gdb) where #0 0x0805c313 in allocateLocal () #1 0x0000005c in ?? () #2 0x08075c97 in base_DataziTypeable_zdLr42klvl4_srt () #3 0x4ee1dca0 in _dl_argv_internal () from /lib/ld-linux.so.2 #4 0x0807c920 in last_free_capability () #5 0x4ee1dca0 in _dl_argv_internal () from /lib/ld-linux.so.2 #6 0x00000000 in ?? () I am using GHC 6.6.1 built from sources on RHE5. Many regards, Ben Benedict R. Gaster S/W Architecture Team Lead ClearSpeed Technology Plc 3110 Great Western Court Hunts Ground Road Stoke Gifford Bristol BS34 8HP Phone: <blocked::mailto:karen@clearspeed.com> + 44 117 3172016 Fax: + 44 117 3172002 brg(a)clearspeed.com <mailto:brg@clearspeed.com> www.clearspeed.com <blocked::blocked::blocked::http://www.clearspeed.com> -- The contents of this email and any attachments are confidential and may be legally privileged. If you have received this email in error please notify the sender immediately and refrain from copying or disclosing the contents of the email to any third party. ClearSpeed accepts no liability for any viruses which may be transmitted by this email or its attachments. ClearSpeed Technology PLC is a company registered in England under company number 05159262 whose registered office is at 3110 Great Western Court, Hunts Ground Road, Bristol BS34 8HP, UK. ClearSpeed Technology INC is a wholly owned subsidiary of ClearSpeed Technology PLC and is incorporated in the United States of America, with its principle place of business at 3031 Tisch Way, Suite 200, San Jose, CA 95128, US. Additional Company information can be found at the following: http://www.clearspeed.com/aboutus/company/index.html

1 1

Re: STG to JavaScript translation
by Victor Nazarov 21 Sep '07

21 Sep '07

I still have some questions regarding the GHC internals. There is a description of STG language in the "Making a Fast Curry: Push/Enter vs. Eval/Apply for Higher-oder Languages" (2004) by Simon Marlow and Simon Peyton Jones paper. In this description the constructor application (CONS closure) can only appear on the right hand side of the bindings. This is totally reasonable if "let" is the only construct that allocates objects. But in the GHC's StgSyn.hs any expression can be constructor application. How does constructor applications are compiled? Are they implicitly transformed to let? For example: f = let g = (THUNK h x) in (CONS g y) Is this exactly the same as (right variant following the paper) f = let g = (THUNK h x) in let freshvar = (CONS g y) in freshvar ? And the second question is how does constructor tag is passed to case when non-vector return is used? In register? In constructor closure? Are there any cases when closure is not build for constructor application? What the case binder is bound to if there is no closure for constructor application? -- vir http://vir.comtv.ru/

3 6

GHC 6.8.1 RC debugger only breaking on first evaluation of a function
by Olivier Boudry 21 Sep '07

21 Sep '07

Hi all, I just tried the new GHCi debugger. A great new feature of GHCi 6.8.1. When debugging a function, as for example the qsort function given as an example in the "3.5 The GHCi Debugger" documentation page, the debugger will only break on first function evaluation. As haskell is pure and lazy it's probably a normal behavior (reuse last result instead of recompute) but it's not very practical in a debugger. I tried to reload (:r) but it doesn't seems to help. Is there a way to force the function to be re-evalutated without quitting, starting, loading prog and setting breakpoints again, other than making the function part of the IO monad. ;-) Thanks, Olivier.

5 8

combination not supported: Threaded/Profiling when building NDP library
by Ben Gaster 20 Sep '07

20 Sep '07

Hello Having read the papers on Nested Data Parallelism in the Haskell I wanted to play around with the de-sugared implementation in the GHC library NDP. I have built GHC from source, on RHE5, and then installed the NDP library from the tar ball. Following the build instructions in the README I got to building the library which resulted in the following error being reported: ghc-6-6.1: combination not supported: Threaded/Profiling This seems to be down to the fact that -prof and -threaded have been specified on a particular build line, although I emit not to be completely sure about this. Looking through the GHC page on its SMP support it seems to imply that the option -threaded is not needed when building code, rather when linking a parallel program, but it is unclear if this is the case when building a library such as NDP. Having scanned through the GHC source tree it seems that -prof is referenced in numerous places such that it would be a bit of a pain to have to remove these over removing the single reference to -threaded in the ndp.mk file. Thanks for any help that you can provide on this. Many Regards, Ben Benedict R. Gaster S/W Architecture Team Lead ClearSpeed Technology Plc 3110 Great Western Court Hunts Ground Road Stoke Gifford Bristol BS34 8HP -- The contents of this email and any attachments are confidential and may be legally privileged. If you have received this email in error please notify the sender immediately and refrain from copying or disclosing the contents of the email to any third party. ClearSpeed accepts no liability for any viruses which may be transmitted by this email or its attachments. ClearSpeed Technology PLC is a company registered in England under company number 05159262 whose registered office is at 3110 Great Western Court, Hunts Ground Road, Bristol BS34 8HP, UK. ClearSpeed Technology INC is a wholly owned subsidiary of ClearSpeed Technology PLC and is incorporated in the United States of America, with its principle place of business at 3031 Tisch Way, Suite 200, San Jose, CA 95128, US. Additional Company information can be found at the following: http://www.clearspeed.com/aboutus/company/index.html

2 2

Unknown option -XPatternSig used in warning
by Bas van Dijk 20 Sep '07

20 Sep '07

Hello, $ ghc --version The Glorious Glasgow Haskell Compilation System, version 6.7.20070915 When loading the following incorrect program in ghci: foo (x :: a) = x ghci rightly complains: Illegal signature in pattern: a Use -XPatternSigs to permit it OK so I follow the advice and load: {-# OPTIONS_GHC -XPatternSigs #-} foo (x :: a) = x Now ghci complains: unknown flags in {-# OPTIONS #-} pragma: -XPatternSigs ??? The following works though: {-# OPTIONS_GHC -fglasgow-exts #-} foo (x :: a) = x results in: A pattern type signature cannot bind scoped type variables `a' unless the pattern has a rigid type context In the pattern: x :: a In the definition of `foo': foo (x :: a) = x Which I expect. Should I file a bug report, or is there an easy fix? regards, Bas van Dijk

6 9