[GHC] #10844: CallStack should not be inlined

[GHC]

#10844: CallStack should not be inlined
-------------------------------------+-------------------------------------
        Reporter:  nomeata           |                Owner:  gridaphobe
            Type:  task              |               Status:  patch
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  7.10.2
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):  Phab:D1259
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by simonpj):

 Well the apppend-string rule is in `PrelRules.lhs`:
 {{{
 match_append_lit :: [Expr CoreBndr] -> Maybe (Expr CoreBndr)
 match_append_lit [Type ty1,
                     Lit (MachStr s1),
                     c1,
                     Var unpk `App` Type ty2
                              `App` Lit (MachStr s2)
                              `App` c2
                              `App` n
                    ]
  = ...
 }}}
 Notice that it maches only on actual literal strings.  If they are floated
 to top level they'll be replaced by a `Var`. Probably an `Id` whose
 unfolding is the literal string, but still it won't match the above.

 Instead you probably need to use `exprIsLiteral_maybe` as in
 {{{
 match_Word64ToInteger _ id_unf id [xl]
   | Just (MachWord64 x) <- exprIsLiteral_maybe id_unf xl
   = case splitFunTy_maybe (idType id) of
 }}}
 later in the same module.

 Try that?

 Indeed, ''every'' match on a `Lit` pattern in these rules should go via
 `exprIsLiteralMaybe`.   See if this works and then maybe at least open a
 ticket for the others; preferably do them.  Could be a big win!

 Simon

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10844#comment:21>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler