- Nhc-bugs - Haskell.org

Trouble getting nhc98 to work
by Karl Crary 21 Jul '06

21 Jul '06

Dear NHC98 maintainers, I was experimenting with nhc98 because I was led to believe that using it, I might be able to use the Haskell FFI to link with libraries built using Cygwin. Unfortunately, I never got far enough to try that; it's failing right out of the box. First I tried the binary distribution. The configure and install process seemed to go smoothly, but in my first attempt to use the compiler I get: /usr/local/bin/nhc98: line 131: C:/cygwin/usr/local/lib/nhc98/ix86-CYGWIN_NT-5.1/hmake-PRAGMA: No such file or directory /usr/local/bin/nhc98: line 131: C:/cygwin/usr/local/lib/nhc98/ix86-CYGWIN_NT-5.1/hmake-PRAGMA: No such file or directory /usr/local/bin/nhc98: line 131: C:/cygwin/usr/local/lib/nhc98/ix86-CYGWIN_NT-5.1/hmake-PRAGMA: No such file or directory /usr/local/bin/nhc98: line 547: C:/cygwin/usr/local/lib/nhc98/ix86-CYGWIN_NT-5.1/nhc98comp: No such file or directory I then tried to build from source, but with no better luck. Configure seems to go smoothly, but the build breaks down quickly: bash-3.1$ make ghc -cpp -o lib/ix86-CYGWIN_NT-5.1/hmake-PRAGMA script/hmake-PRAGMA.hs cd src/runtime; make all nhc98heap.exe make[1]: Entering directory `/c/nhc98-1.18/src/runtime' Makefile:27: *** target pattern contains no `%'. Stop. make[1]: Leaving directory `/c/nhc98-1.18/src/runtime' make: *** [targets/ix86-CYGWIN_NT-5.1/runtime] Error 2 bash-3.1$ gmake bash: gmake: command not found bash-3.1$ which make /usr/bin/make bash-3.1$ make cd src/runtime; make all nhc98heap.exe make[1]: Entering directory `/c/nhc98-1.18/src/runtime' Makefile:27: *** target pattern contains no `%'. Stop. make[1]: Leaving directory `/c/nhc98-1.18/src/runtime' make: *** [targets/ix86-CYGWIN_NT-5.1/runtime] Error 2 Any idea what I might be doing wrong? -- Karl Crary

2 1

Bus Error and Segmentation Fault when Heap Profiling is On
by Shin-Cheng Mu 02 Jun '06

02 Jun '06

Hi, This is another potential bug. Unfortunately I currently cannot reproduce it on smaller programs. I tried to do heap profiling on the following code: http://www.psdlab.org/~scm/hx.tar.gz The main program is hxpc.hs. I tried to compile it with hmake -nhc98 -p hxpc and run it on a small sample input scm$ cat > temp.xml <a><b><item>123</item></b><c></c></a> scm$ ./hxpc temp.xml Bus error scm$ ./hxpc temp.xml +RTS -p Segmentation fault Increasing the heap size does not seem to help. If I compile the program without the -p option, it runs fine. Since Malcolm Wallace has already profiled the program on his machine, it seems to be a platform specific problem I installed NHC98 1.18_4 using Darwin Ports, on a PowerPC G4 based Mac, running OS X 10.4 and GCC 4.0. sincerely, Shin-Cheng Mu

2 1

ld: Undefined symbols on OS X when compiling with -t
by Shin-Cheng Mu 22 May '06

22 May '06

Hi, I just installed NHC98 using Darwin Ports on a PowerPC G4 based Mac, running OS X 10.4 with GCC 4.0. NHC works fine without profiling options. When I turned on time profiling, however, NHC98 appears to have problems linking: scm$ cat text.hs main = print (foldr (+) 0 [1..10000]) scm$ hmake -nhc98 -t text nhc98 -t -c -o text.z.o text.hs nhc98 -t -o text text.z.o /tmp/tprofusr3456.c: In function 'tprofTMInitTreeUsr': /tmp/tprofusr3456.c:21: warning: incompatible implicit declaration of built-in function 'strdup' /usr/bin/ld: Undefined symbols: _FN_NHC_46Internal_46_95apply1 _FN_NHC_46Internal_46_95apply2 _TMSUB_NHC_46DErrNo _TM_NHC_46DErrNo _TMSUB_NHC_46PackedString _TM_NHC_46PackedString collect2: ld returned 1 exit status Is it a bug, or is something missing in Darwin Ports? sincerely, Shin-Cheng Mu

2 1

What's the correct procedure for upgrading?
by Hethcoat-III, Charles L 18 Apr '06

18 Apr '06

I hope nhc-bugs is the correct venue for this question. I installed NHC 1.16 on Cygwin (Windows XP), had some problems with it, and then realized I had grabbed an old package. I'm going to upgrade to 1.18 and try again. However, I'm a little bit concerned about the right way to do it. I haven't found any instructions for upgrading properly. Do I just install 1.18 and forget about 1.16? Or should I delete all of the 1.16 files first? I didn't see a Makefile target "uninstall", although there are some "*clean*" options there. Perhaps you could add this topic to your website? Thanks. Charles Hethcoat

2 1

Linux install troubles
by Doug McIlroy 12 Apr '06

12 Apr '06

On Fedora Linux, I fetched nhc98-1.18.tar.gz (Linux executable). Got some relatively innocuous diagnostics from configure and make install, but one stood out: ./configure: line 594: src/runtime/nhc98heap.c: No such file or directory Bulling through anyway, I did an installation. and compiled a hello world program. The result appears to be related to the configure trouble: OS allocated a heap in high memory (>0x80000000) which breaks this program's run-time system. hpStart=0xb7d96008, hpEnd=0xb7e0b308

2 1

EnumFromThen
by hanatani＠mbg.nifty.com 10 Apr '06

10 Apr '06

Hello, I found a bug in nhc98-1.18. $ diff src/prelude/Prelude/Enum_Integer.hs src/prelude/Prelude/Enum_Integer.hs ~ 14c14 < enumFromThen x y = x : enumFromThen y (2*y-x) --- > enumFromThen x y = x : enumFromThen y (2*y+x) -- Yoichi Hanatani

2 1

Fixing "OS allocated a heap in high memory"
by dons＠cse.unsw.edu.au 23 Oct '05

23 Oct '05

Hey guys, Recent changes to malloc on OpenBSD mean that nhc98 builds now die with the "OS allocated a heap in high memory" error, as malloc is putting things above the 0x80000000 line. Switching the heap allocation to use mmap with a hint to allocate somewhere lower fixes it at least on x86/openbsd. The ghc testsuite runs, for example. The mmap code is stolen from MBlock.c in ghc. Any options on this? Is it reasonable? (I'd like to use it as a custom patch in the OpenBSD port of nhc98, but want to check I'm not doing anything hideously wrong :) -- Don Here's what I'm doing: --- src/runtime/Kernel/collector.c.orig Tue Mar 8 03:53:38 2005 +++ src/runtime/Kernel/collector.c Sat Oct 22 19:05:22 2005 @@ -7,6 +7,11 @@ /*#include "runtime.h" -- already included in node.h */ #include "mark.h" +#include <errno.h> +#include <sys/types.h> +#include <sys/mman.h> + + /*#define HEAPSIZE 100000 -- defined in top-level Makefile at config-time */ #define GCSTACKSIZE 20000 @@ -45,11 +50,22 @@ void initGc(Int hpSize,NodePtr *ihp,Int { Int totalSize = hpSize+spSize; Int tableSize = (totalSize+WORDSIZE)/(WORDSIZE+1)+1; /* Last one for end of marked */ + int size = (int)totalSize * sizeof(Node); - if(NULL == (hpStart = malloc ((int)totalSize * sizeof(Node)))) { - fprintf(stderr,"Not enough memory for heap and stack.\n"); - exit(-1); - } + /* Ask for memory well below the 0x80000000 line. The nhc runtime + * reserves the high bit on addresses for its own purposes */ + hpStart = mmap((void *)0x40000000, size, + PROT_EXEC|PROT_READ|PROT_WRITE, + MAP_PRIVATE|MAP_ANON, -1 , 0); + + if (hpStart == MAP_FAILED) { + if (errno == ENOMEM) { + fprintf(stderr,"initGc: out of memory (requested %lu bytes)", size); + exit(-1); + } else { + fprintf(stderr, "initGc: mmap failed: %s", strerror(errno)); + } + } hpEnd = hpStart + totalSize;

3 5

Debugging a segfault
by dons＠cse.unsw.edu.au 23 Oct '05

23 Oct '05

Hey, I need some help chasing down a segfault in nhc98 1.1{6,8} on OpenBSD/powerpc. The x86 version runs nicely (after the mmap patch), however both 1.16 and 1.18 die at the following point in the build process: cd src/prelude/powerpc-OpenBSD/NHC; gmake clean all # Patch machine-specific parts. gmake[1]: Entering directory `/home/hack/dons/build/nhc98-1.18/src/prelude/powerpc-OpenBSD/NHC' rm -f *.hi rm -f /home/hack/dons/build/nhc98-1.18/targets/powerpc-OpenBSD/obj/prelude/DErrNo/*.o *.o rm -f rm -f /home/hack/dons/build/nhc98-1.18/script/nhc98 -c +CTS -lib -redefine -CTS +RTS -H32M -RTS -o /home/hack/dons/build/nhc98-1.18/targets/powerpc-OpenBSD/obj/prelude/DErrNo/DErrNo.o DErrNo.hs Segmentation fault (core dumped) Segmentation fault (core dumped) Segmentation fault (core dumped) Segmentation fault (core dumped) gmake[1]: *** [/home/hack/dons/build/nhc98-1.18/targets/powerpc-OpenBSD/obj/prelude/DErrNo/DErrNo.o] Error 1 And we see: $ cd /home/hack/dons/build/nhc98-1.18/src/prelude/powerpc-OpenBSD/NHC $ ls DErrNo.hc DErrNo.p.c Makefile hmake-PRAGMA.core DErrNo.hs DErrNo.z.c Makefile.inc nhc98comp.core nhc98comp.core says: (gdb) where #0 0x01802224 in ?? () #1 0x01803018 in ?? () Previous frame identical to this frame (corrupt stack?) Whereas hmake-PRAGMA.core gives us: (gdb) where #0 0x01801f08 in run () #1 0x01801d68 in main () How do I go about debugging this? gdb wasn't particularly revealing. Is there a way to get debugging symbols compiled in? -- Don

2 1

programs compile, but fail to run on Linux 2.6.11
by Aaron Denney 22 Sep '05

22 Sep '05

% ./Main OS allocated a heap in high memory (>0x80000000) which breaks this program's run-time system. hpStart=0xb7e03008, hpEnd=0xb7e78308 % Fresh compile of nhc98-1.18, with ghc 6.4. I assume this could also be a libc issue, if it's defining "heap" as addresses malloc() returns, which could be done through anonymous mmap()s. I'm running glibc-2.3.5 Are there any other details I should provide? -- Aaron Denney -><-

1 0

Re: [nhc-bugs] Installation Problem
by Malcolm Wallace 31 Jul '05

31 Jul '05

Dieter Büttner <judisun(a)t-online.de> writes: > thank you very much for your answer. Yes I forgot make. > I did it all once again including make. But it doesn't make things easier. > It doesn't work. I downloaded the sources unzipped etc. > > Then I started ./configure --> see the file nhc98_configure where I > redirected the output (standard and standard error). > Then I started make --> see the file nhc98_make where I redirected the output > (standard and standard error). More than 100000 lines!! I see that your machine is an x86_64. At present unfortunately, nhc98 only works on 32-bit machines, not 64-bit. Regards, Malcolm

1 0