Yes, we've seen behaviour like this here on that platform.
If you try feeding the failing compiler invocation the extra
command-line option "-dshow-passes" and it shows up as
failing pretty early on in the game (parsing, renaming),
here's something you may want to try out:
- in mk/build.mk (in the toplevel directory), define the
following
GHC_HAPPY_OPTS=-g
- in ghc/compiler/ remove the following files
parser/Parser.o parser/Parser.hs
parser/ParseIface.o parser/ParseIface.hs
main/ParsePkgConf.o main/ParsePkgConf.hs
and do a 'make all'
- with the freshly built contents of ghc/compiler in hand,
proceed to ghc/lib and continue.
This may or may not address your problem -- I've yet to
isolate the bug we were running into, but poking around
with 'gdb' suggested that the string-encoded tables that
Happy generates when using the options "-agc" was causing
the problem.
--sigbjorn
----- Original Message -----
From: "Akop Pogosian"
I am trying to build ghc-5.02 on Solaris 2.6. I already have a ghc-4.08 distribution that works. ghc build fails after nearly two hours of compiling with this error:
/opt/local/bin/ld -r -x -o PrelDynamic.p_o PrelDynamic/*.p_o; rm -f PrelIOBase.p_o; if [ ! -d PrelIOBase ]; then mkdir PrelIOBase; else /usr/b in/find PrelIOBase -name '*.p_o' | xargs rm -f __rm_food; fi;
../../../ghc/compiler/ghc-inplace -ldl -fvia-C -cpp -fglasgow-exts -O -Rghc- timing -package-name std -split-objs -hisuf p_hi -hcsuf p_hc -osuf p_o -prof -funbox-strict-fields -fno-ignore-asserts -c PrelIOBase.lhs -o PrelIOBase.p_o
Segmentation Fault gmake[4]: *** [PrelIOBase.p_o] Error 1 gmake[3]: *** [all] Error 1 gmake[2]: *** [all] Error 1 gmake[1]: *** [all] Error 1 gmake[1]: Leaving directory `/home/src/akopps/ghc-5.02/ghc' gmake: *** [all] Error 1
Has anyone seen this happen before?
-akop
_______________________________________________ Glasgow-haskell-users mailing list Glasgow-haskell-users@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users