[GHC] #14779: Compiling with -g fails -lint-core checks

[GHC]

#14779: Compiling with -g fails -lint-core checks
-------------------------------------+-------------------------------------
        Reporter:  niteria           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:
  (Debugging)                        |
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  crash or panic                     |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #14122, #14123,   |  Differential Rev(s):
  #8472, #14406                      |
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by niteria):

 Agreed, this is too broad. I need `isTopLevel` as well.

 I didn't get here in a very principled way.

 I just saw that FloatOut was doing what I expected (not putting ticks),
 then after a Simplifier pass it was broken. I noticed that the floated out
 binding was touched by pre-inline-uncond and it prompted me to look at
 `preInlineUnconditionally`. I then vaguely recalled that we intend for
 these top level unboxed strings to stay floated, so it seemed like a good
 idea not to inline them. I forgot the top level part in my impromptu
 patch.

 Why do we end up with ticks after that? I don't know yet. But arguably,
 inlining a top level unboxed string literal is the first place where we
 went wrong.

 NB: With the fixed patch that does `isTopLevel` I'm able to compile GHC
 HEAD with `-g`.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14779#comment:7>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler