[GHC] #9142: LLVM HEAD rejects aliases used by LLVM codegen

[GHC]

#9142: LLVM HEAD rejects aliases  used by LLVM codegen
-------------------------------------+------------------------------------
        Reporter:  bgamari           |            Owner:
            Type:  bug               |           Status:  new
        Priority:  high              |        Milestone:
       Component:  Compiler          |          Version:  7.8.2
      Resolution:                    |         Keywords:
Operating System:  Unknown/Multiple  |     Architecture:  Unknown/Multiple
 Type of failure:  None/Unknown      |       Difficulty:  Unknown
       Test Case:                    |       Blocked By:
        Blocking:                    |  Related Tickets:
-------------------------------------+------------------------------------

Comment (by bgamari):

 Replying to [comment:4 scpmw]:
 > Not quite sure I understand how your patch works - isn't this basically
 allocating a bunch of static null-initialized `i8` fields to point to?
 That's hardly what we want, even if it compiles.
 >
 Frankly I'm not too clear on this either. I modelled the change after
 similar [https://github.com/llvm-
 mirror/llvm/commit/38048cdb1c016e1429004ddf4adfa40a8d853cbf#diff-
 6888511bd2c39203c5f7a7e314bdd2ebL40 changes] made in the commit in
 question to testcases. However, you are right, it would appear that the
 local definitions will be used in lieu of the external symbols the
 references are supposed to use. All I know at the moment is that the code
 compiles. It likely does the wrong thing.

 > It would be really interesting to get a statement from LLVM people on
 this.
 >
 Indeed it would.  At this point my next goal is to chase down the relevant
 LLVM people.

 > If they think that we are mis-using aliases, that's reason enough to do
 something else. One possible solution could again be type aliases - but
 instead of writing the definitions at the end of the session we'd close
 the file and /prepend/ them. That could be a cheap and easy way to get the
 LLVM parser to like us again.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9142#comment:5>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler