Re: [GHC] #7953: Segfault on x86 with -O2

18 Jun 2013

      #7953: Segfault on x86 with -O2
-------------------------+--------------------------------------------------
    Reporter:  igloo     |       Owner:                  
        Type:  bug       |      Status:  new             
    Priority:  highest   |   Milestone:  7.8.1           
   Component:  Compiler  |     Version:  7.7             
    Keywords:            |          Os:  Unknown/Multiple
Architecture:  x86       |     Failure:  None/Unknown    
  Difficulty:  Unknown   |    Testcase:                  
   Blockedby:            |    Blocking:                  
     Related:            |  
-------------------------+--------------------------------------------------

Comment(by kazu-yamamoto):

 If we change
 {{{
 atMost pt q = case q of
     (Winner e _)
         | prio e > pt -> ([], q)
 }}}
 to
 {{{
 atMost pt q = case q of
     (Winner e _)
         | gt (prio e) pt -> ([], q)

 {-# NOINLINE gt #-}
 gt :: Prio -> Prio -> Bool
 gt = (>)
 }}}
 this segfault disappears.

 So, inlining (>) for Float/Double has a bug.

-- 
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/7953#comment:5
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler