March 2002 - Glasgow-haskell-users

RE: rts/Printer.c
by Simon Marlow 19 Mar '02

19 Mar '02

> > Is there a way to compile the RTS so that the names of data > > constructors are retained? > > My memory from a long, long time ago is that we had a debugging mode > which used libbfd to get symbol names and some GHC-demangling code to > turn those back into names like the user typed in. Yes, we still have that code, but it doesn't get much use - gdb can map addresses back to symbol names. I think I always intended to beef this up using a proper hash table for the address->symbol mapping, but never got around to it. Alternatively, if you compile for profiling, then info tables contain some extra information including the constructor name for contructors. > > It looks like Printer.c has some hooks for printing data structures, > > is it feasible to make a primitive that allows Haskell code to call > > into this? I'm thinking in particular of calling something like > > printClosure() from Haskell code. > > If it's a global symbol, you should be able to access it form the ffi. > > If you want to traverse data structures the way you can using the > HugsInternals library, you might want to tweak the code a little to > provide a similar semantics/ API. Basically, all you have to do is > take the C code and split it into handy bits. This is a nice idea - we'll be happy to incorporate the changes. Cheers, Simon

3 2

RE: Linux version, libraries
by Simon Marlow 19 Mar '02

19 Mar '02

> uname -a > Linux ...<machine>... > 2.2.12 #6 SMP mar oct 5 16:44:59 CEST 1999 i686 unknown > > ls -l /lib/libc* > > -rwxr-xr-x 1 root root 4118299 Sep 20 1999 /lib/libc-2.1.2.so* > lrwxrwxrwx 1 root root 13 Nov 5 1999 /lib/libc.so.6 -> > libc-2.1.2.so* > lrwxrwxrwx 1 ... 1999 /lib/libcom_err.so.2 -> > libcom_err.so.2.0* > -rwxr-xr-x 1 root root 7889 Oct 26 1999 /lib/libcom_err.so.2.0* > -rwxr-xr-x 1 root root 64595 Sep 20 1999 /lib/libcrypt-2.1.2.so* > lrwxrwxrwx 1 .. 1999 /lib/libcrypt.so.1 -> > libcrypt-2.1.2.so* It looks like it might be a bit earlier than RedHat 6.2, our machine here has libc-2.1.3. The RedHat 6.2 bundle should work fine. One possible way to tell what your system is: > rpm -qa | grep redhat redhat-release-6.2-1 rpmdb-redhat-6.2-0.20000308 redhat-logos-1.1.0-2 > On GMP library: > > > find /usr/lib -name '*gmp*' > > /usr/lib/libgmp.so.2.0.2 > /usr/lib/libgmp.so.2 > find: /usr/lib/netscape/de/nethelp: Permission ... > /usr/lib/perl5/5.00503/i386-linux/linux/igmp.ph > /usr/lib/perl5/5.00503/i386-linux/netinet/igmp.ph Well, you have GMP 2 installed by the looks of it. I don't know why GHC 4.04 is failing, but it's probably not worth tracking down. Cheers, Simon

1 0

Linux version, libraries
by S.D.Mechveliani 19 Mar '02

19 Mar '02

To my >> How to find out in a regular way the info about Linux system and >> libc version? Simon Marlow writes > Try 'uname -a' & 'ls -l /lib/libc*'. uname -a Linux ...<machine>... 2.2.12 #6 SMP mar oct 5 16:44:59 CEST 1999 i686 unknown ls -l /lib/libc* -rwxr-xr-x 1 root root 4118299 Sep 20 1999 /lib/libc-2.1.2.so* lrwxrwxrwx 1 root root 13 Nov 5 1999 /lib/libc.so.6 -> libc-2.1.2.so* lrwxrwxrwx 1 ... 1999 /lib/libcom_err.so.2 -> libcom_err.so.2.0* -rwxr-xr-x 1 root root 7889 Oct 26 1999 /lib/libcom_err.so.2.0* -rwxr-xr-x 1 root root 64595 Sep 20 1999 /lib/libcrypt-2.1.2.so* lrwxrwxrwx 1 .. 1999 /lib/libcrypt.so.1 -> libcrypt-2.1.2.so* But how to find whether it is RedHat (version) ? On GMP library: > find /usr/lib -name '*gmp*' /usr/lib/libgmp.so.2.0.2 /usr/lib/libgmp.so.2 find: /usr/lib/netscape/de/nethelp: Permission ... /usr/lib/perl5/5.00503/i386-linux/linux/igmp.ph /usr/lib/perl5/5.00503/i386-linux/netinet/igmp.ph ? ----------------- Serge Mechveliani mechvel(a)botik.ru

1 0

RE: rts/Printer.c
by Simon Marlow 19 Mar '02

19 Mar '02

> Say, for argument's sake, that I want to make the function > printObj() available in Haskell, where: > > void printObj (StgClosure *obj); > > (as defined in and exported from rts/Printer.c) > > In Haskell I want to reflect this as: > > printObj :: a -> IO () The FFI doesn't really provide for this sort of thing, because it assumes that any data you want to pass to the outside world will be marshaled into something that the far end can understand (a C data structure, say) before being passed. Passing the raw address of the Haskell value on the heap is a bit dodgy, because the RTS will move these objects around from time to time. It's good practice to instead pass a StablePtr to the value. So you really want something like foreign import ccall unsafe printObj :: StablePtr a -> IO () and use deRefStablePtr() inside printObj to get at the real pointer. Note that this might still break in a threaded environment unless you take care to prevent other parts of the system from modifying the heap while you inspect it, but I wouldn't worry too much about that right now. > Alistair mentioned that I could use the FFI to access some C code > in the RTS. > > My intuition would have been to go about adding a primop as > described in > /ghc/compiler/prelude/primops.txt (ie without calling through > the FFI). primops are a little harder to add than FFI calls, and are generally used for two reasons: when you want the compiler to know more about the underlying operation (eg. so it can do constant folding) or because the compiler can directly generate inline code for the operation. An FFI call is clearly the way to go here. Cheers, Simon

1 0

RE: installing 5.02.2
by Simon Marlow 19 Mar '02

19 Mar '02

> I am going to install ghc-5.02.2 > > on certain Linux machine to which I have ssh access and > permission to install ghc-5.02.2 in my user directory > (not a system user). Then you need to get the normal .tar.gz binary bundle. The RPMs will only install in the system directories. > The situation is as follows. > It has i386-unknown (Pentium Pro II ...), much of memory and disk. > I am not sure, but it looks like it is under RedHat-6.2, > glibc-2.1.2. The binary bundle was built on RedHat 6.2, so you should be fine. > It has ghc-4.04 installed in the system area. > But it failed to link Main.o of the program main = putStr "hello\n" Do you mean ghc-4.04 failed to link the program? What was the error message? > Probably, Happy, is not installed. You don't need Happy to install binaries, only if you build from source (and even then, building from a normal source distribution doesn't require Happy because we provide the Happy-generated sources. You might need to tweak the configure script to avoid it complaining about a missing Happy though.). > Also I applied certain ghc --make M > on my Debian Linux machine with ghc-5.02.2, > scp -ed the a.out.zip file to the target machine and unzipped it. > And it cannot run there, some library not found, something of this > sort. This is to be expected - shared libraries are bound to differ between the various flavours and versions of Linux distributions. > Could the GHC installation experts guide me, please? > What is easier in this situation, to install from source or > from binary? If it is a RedHat 6.2 (or close) system, then installing the .tar.gz binary bundle should work fine. If not, you may need to try to convince the system admin to install an RPM package for you. > How to install Happy for the first time? There are various binaries of Happy. If none of them work for you, then Happy has to be built from source using GHC. Happy isn't necessary for building GHC from source. > How to find out in a regular way the info about Linux system and > libc version? Try 'uname -a' & 'ls -l /lib/libc*'. Cheers, Simon

1 0

installing 5.02.2
by S.D.Mechveliani 19 Mar '02

19 Mar '02

Dear GHC, I am going to install ghc-5.02.2 on certain Linux machine to which I have ssh access and permission to install ghc-5.02.2 in my user directory (not a system user). The situation is as follows. It has i386-unknown (Pentium Pro II ...), much of memory and disk. I am not sure, but it looks like it is under RedHat-6.2, glibc-2.1.2. It has ghc-4.04 installed in the system area. But it failed to link Main.o of the program main = putStr "hello\n" Probably, Happy, is not installed. Also I applied certain ghc --make M on my Debian Linux machine with ghc-5.02.2, scp -ed the a.out.zip file to the target machine and unzipped it. And it cannot run there, some library not found, something of this sort. Could the GHC installation experts guide me, please? What is easier in this situation, to install from source or from binary? How to install Happy for the first time? How to find out in a regular way the info about Linux system and libc version? Thank you for the advice, ----------------- Serge Mechveliani mechvel(a)botik.ru

1 0

RE: GHC 4.08.2 on OpenBSD 2.9, almost
by Simon Marlow 18 Mar '02

18 Mar '02

> I'm trying to port any recent version of ghc to OpenBSD 2.9. > > I have, with a little bit of work, been able to compile 4.08.2 > from .hc sources. > > I now get the following kind of undefined symbol errors: > > Main.o: Undefined symbol `__init_Prelude' referenced from text > segment > Main.o: Undefined symbol `PrelBase_unpackCStringzh_closure' > referenced from text segment > Main.o: Undefined symbol `PrelIO_putStr_closure' > referenced from > text segment > > [ ... ] > > PrelMain__1.o: Undefined symbol `___init_Main' > referenced from text > segment > .o::PrelMain(void): Undefined symbol > `_Main_main_closure' referenced > from text segment > .o::PrelMain(void): Undefined symbol > `_Main_main_closure' referenced > from text segment > > My guess is that this kind of thing has happened before > ... is there > an area in the source I should be looking at? This looks suspiciously like a mismatch in the handling of leading underscores in symbol names somewhere. It appears from the messages above that Main.o doesn't have extra leading underscores on its symbols, but the libraries do. Perhaps Main.o was compiled by the native code generator? Because the leading-underscore setting is compiled into the compiler via an external setting, you'll need to bootstrap the compiler before the setting will be correct for your system. Try compiling your test program with -fvia-C. Cheers, Simon

1 0

rts/Printer.c
by Bernard James POPE 17 Mar '02

17 Mar '02

Hi all, I'm looking at extending my debugger for Haskell to GHC (currently only works for Hugs, which is a pity). I have a couple of questions about printing data structures from the GHC heap. Is there a way to compile the RTS so that the names of data constructors are retained? It looks like Printer.c has some hooks for printing data structures, is it feasible to make a primitive that allows Haskell code to call into this? I'm thinking in particular of calling something like printClosure() from Haskell code. Unfortunately my understanding of the rts and the layout of data structures is fading. Cheers, Bernie.

2 1

GHC 4.08.2 on OpenBSD 2.9, almost
by dons＠cse.unsw.edu.au 16 Mar '02

16 Mar '02

Hi, I'm trying to port any recent version of ghc to OpenBSD 2.9. I have, with a little bit of work, been able to compile 4.08.2 from .hc sources. I now get the following kind of undefined symbol errors: Main.o: Undefined symbol `__init_Prelude' referenced from text segment Main.o: Undefined symbol `PrelBase_unpackCStringzh_closure' referenced from text segment Main.o: Undefined symbol `PrelIO_putStr_closure' referenced from text segment [ ... ] PrelMain__1.o: Undefined symbol `___init_Main' referenced from text segment .o::PrelMain(void): Undefined symbol `_Main_main_closure' referenced from text segment .o::PrelMain(void): Undefined symbol `_Main_main_closure' referenced from text segment My guess is that this kind of thing has happened before ... is there an area in the source I should be looking at? Thanks, Don Stewart dons(a)cse.unsw.edu.au

1 0

RE: Strictness of addition functions.
by Simon Peyton-Jones 15 Mar '02

15 Mar '02

| data Pair = Pair !Int# !Int# | | f (Pair x y) = x +# y | | I still discovered to my dismay that the type of the first | argument to f is U(LL). | | What can I do to force the arguments to be strict? Int# is *always* strict. You don't need the !'s. The L is misleading, I suppose. But since Int# means *unboxed* 32-bit integers, there is simply no way you can pass a thunk. Simon

1 0