[GHC] #8103: Segfault when passing unboxed Float# and Double# across modules

[GHC]

#8103: Segfault when passing unboxed Float# and Double# across modules
----------------------------------+---------------------------
        Reporter:  jstolarek      |            Owner:
            Type:  bug            |           Status:  new
        Priority:  highest        |        Milestone:  7.8.1
       Component:  Compiler       |          Version:  7.7
      Resolution:                 |         Keywords:
Operating System:  Linux          |     Architecture:  x86
 Type of failure:  Runtime crash  |       Difficulty:  Unknown
       Test Case:                 |       Blocked By:
        Blocking:  6135           |  Related Tickets:
----------------------------------+---------------------------

Comment (by jstolarek):

 I still can reproduce it on one of the older HEAD builds:

 {{{
 [t-jastol at cam-05-unx : ~/tests/bool-primops/T8103]
 /5playpen/t-jastol/sandboxes/ghc/master-i386/bin/ghc --print-build-
 platform
 i386-unknown-linux
 [t-jastol at cam-05-unx : ~/tests/bool-primops/T8103]
 /5playpen/t-jastol/sandboxes/ghc/master-i386/bin/ghc --print-target-
 platform
 i386-unknown-linux
 [t-jastol at cam-05-unx : ~/tests/bool-primops/T8103]
 /5playpen/t-jastol/sandboxes/ghc/master-i386/bin/ghc -fforce-recomp
 8103.hs -ddump-cmm | grep D1
            D1 = _B2::F64;
            call AddWraps.foo_info(D2, D1, R1) args: 4, res: 0, upd: 4;
  AddWraps.foo_entry() //  [D2, D1]
            _B2::F64 = D1;
            D1 = _cfE::F64;
            call (P32[(old + 4)])(D1) args: 4, res: 0, upd: 4;
       D1 = _B2::F64;
       call AddWraps.foo_info(D2, D1, R1) args: 4, res: 0, upd: 4;

 ... Truncated for clarity

 [t-jastol at cam-05-unx : ~/tests/bool-primops/T8103] ./8103
 Segmentation fault (core dumped)
 }}}

 I'm doing a fresh build at the moment - if this changes anything I'll
 report back.

 Perhaps an important detail here is that this is not a true 32bit machine,
 but a 32-bit sandbox. I don't know if this is relevant in any way.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8103#comment:3>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler