Hi Daniil
I have a binary dist here:
http://www.dfki.de/sks/hets/solaris/ghcs/ghc-6.8.3-sparc-sun-solaris2.tar.bz2
that runs and was built with gcc-4.0.3 on Solaris 8 sparc. It runs under
Solaris 10 with gcc-4.2.2, too. (It also runs with gcc-3.4, but
building fails as in #951). I've not tried gcc-2.95.
my ldd output of ghc-6.8.3 is:
librt.so.1 => /usr/lib/librt.so.1
libreadline.so.5 => /usr/local/lib/libreadline.so.5
libncurses.so.5 => /usr/local/lib/libncurses.so.5
libdl.so.1 => /usr/lib/libdl.so.1
libm.so.2 => /usr/local/lib/libm.so.2
libgmp.so.3 => /usr/local/lib/libgmp.so.3
libpthread.so.1 => /usr/lib/libpthread.so.1
libc.so.1 => /usr/lib/libc.so.1
libaio.so.1 => /usr/lib/libaio.so.1
libthread.so.1 => /usr/lib/libthread.so.1
/usr/platform/SUNW,Ultra-4/lib/libc_psr.so.1
My "ld" is a Solaris Link Editors: 5.8-1.302 (resp. 5.10-1.489), but a
gnu-linker should work, too. In fact gcc uses a gnu linker, i.e. my gcc
was configured:
--with-gnu-as --with-as=/usr/local/bin/gnu-as
--with-gnu-ld --with-ld=/usr/local/bin/gnu-ld
My build.mk file for building ghc contains:
SRC_HC_OPTS += -optc-mcpu=ultrasparc -opta-mcpu=ultrasparc
I've no idea what goes wrong for you, but maybe my above information
helps you to figure this out yourself.
Cheers Christian
Daniil Elovkov wrote:
Hello folks, Christian
I'm trying to get ghc 6.8.2 running on Solaris 10 and having problems.
To be precise, I'm trying to compile a 'hello world' program by ghc
6.8.2 which I got in binary form haskell.org.
gcc is 2.95, it uses sun linker. I remember there were problems with
that in the past. Is ghc supposed to work only with gnu ld or sun ld
as well?
So, how it went
first I got compiler errors in many places of Reg.hs:
global register variable follows a function definition
Googling showed that Don Stewart used to fix it by swapping 2 includes
in Stg.h - putting MachRegs.h after Regs.h instead of before. It
helped
Then there was assembler error:
cannot use v8plus instructions in a non-v8plus target binary
It was caused by -mcpu=v9, which ghc passes to gcc. I blindly added
-optc -mcpu=v8 and it helped :)
Then the linker complained that it could not resolve aio_fork and
__aio_suspend64, referenced from librt.so.
-lrt is passed by ghc to the linker. On this machine there is
/lib/libaio.so. Linking with it didn't help. It doesn't really contain
exactly those functions, only with slightly different names, like
_aio_forkinit, _libaio_fork.
Also, librt which was linked in was the one lying close to where gcc
is installed. Apart from that, there is also librt.so in /lib. I
thought, maybe the wrong librt was used and said -optl -L/lib to link
against the one in /lib.
No complaints, but the resulting binary segfaults.
Does anybody have any ideas?
_______________________________________________
Glasgow-haskell-users mailing list
Glasgow-haskell-users@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-users
