So it looks like {{{binary-dist}}} step is wrong. I'll soon upload a full log. No, it is fine. A binary dist is prepared so the files in it can be installed, and for that it doesn't need most of the `inplace` scripts.
ldd (on powerpc64): {{{ /usr/local/lib/ghc-8.6.3/bin/ghc-pkg: libm.so.5 => /lib/libm.so.5 (0x811318000) libncursesw.so.8 => /lib/libncursesw.so.8 (0x81135e000) libutil.so.9 => /lib/libutil.so.9 (0x8113e4000) librt.so.1 => /usr/lib/librt.so.1 (0x81140f000) libdl.so.1 => /usr/lib/libdl.so.1 (0x811427000) libthr.so.3 => /lib/libthr.so.3 (0x811439000) libc.so.7 => /lib/libc.so.7 (0x811483000) libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x8117ea000) }}} Here we are at something. `libgcc_s.so.1` is taken from `/lib` while it should be coming from `gcc8`. Did you set `CFLAGS` and`LDFLAGS` to also point gcc's
#15916: GHC doesn't build on powerpc64 architecture on systems other than GNU / Linux -------------------------------------+------------------------------------- Reporter: pkubaj | Owner: trommler Type: feature request | Status: patch Priority: normal | Milestone: 8.10.1 Component: Compiler | Version: 8.6.2 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: powerpc64 Type of failure: Building GHC | Test Case: failed | Blocked By: 15411 | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | https://gitlab.haskell.org/ghc/ghc/merge_requests/81 -------------------------------------+------------------------------------- Comment (by trommler): Replying to [comment:37 pkubaj]: directory on powerpc64 (with `-Wl,rpath=...`) when you were building on amd64? Could you also post the output of `readelf -d /usr/local/lib/ghc-8.6.3/bin /ghc-pkg` on powerpc64? I am only interested in lines that contain `RUNPATH` or `RPATH`. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15916#comment:39 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler