#5100: Stack space overflow when using -N2 and not with -N1
-------------------------------+--------------------------------------------
    Reporter:  mitar           |       Owner:              
        Type:  bug             |      Status:  new         
    Priority:  normal          |   Component:  Compiler    
     Version:  7.1             |    Keywords:              
    Testcase:                  |   Blockedby:              
          Os:  Linux           |    Blocking:              
Architecture:  x86_64 (amd64)  |     Failure:  None/Unknown
-------------------------------+--------------------------------------------
 I get a stack space overflow error when using -N2 (or -N4), but the
 program works with -N1. The program is my test and benchmarking program
 for [http://hackage.haskell.org/package/Etage-Graph Etage-Graph] package.

 So this works normally: `./etage-graph-test -s 10 +RTS -N1`

 {{{
 Generating a random graph of size 10.
 Graph contains 10 nodes.
 Dijkstra search time for shortest paths: 0.000875s
 Etage search time for shortest paths: 0.010341s (0.5s timeout)
 Found 100.00 % shortest paths.
 etage-graph-test: DissolvingException "()"
 }}}

 But this does not: `./etage-graph-test -s 10 +RTS -N2`

 {{{
 Generating a random graph of size 10.
 Graph contains 10 nodes.
 Stack space overflow: current size 8388608 bytes.
 Use `+RTS -Ksize -RTS' to increase it.
 }}}

 The most interesting thing is that this happens only for graph sizes (-s
 parameter) between (including) 1-19. For larger sizes it works without
 problems. What means that the problem is this line:

 {{{
 map spFromSource (nodes graph) `using` parListChunk (noNodes graph `div`
 (numCapabilities * 10)) rdeepseq
 }}}

 Yes. So also `/etage-graph-test -s 5 +RTS -N1` fails. OK, so
 `parListChunk` with 0 as an argument is not well defined, or it is defined
 to diverge. Is this a wanted feature or it is a bug?

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5100>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Reply via email to