[GHC] #7211: Huge space leak on a program that shouldn't leak

[GHC]

#7211: Huge space leak on a program that shouldn't leak
-------------------------------------+--------------------------------------
 Reporter:  bartavelle               |          Owner:                
     Type:  bug                      |         Status:  new           
 Priority:  normal                   |      Component:  Compiler      
  Version:  7.4.2                    |       Keywords:                
       Os:  Linux                    |   Architecture:  x86_64 (amd64)
  Failure:  Runtime performance bug  |       Testcase:                
Blockedby:                           |       Blocking:                
  Related:                           |  
-------------------------------------+--------------------------------------
 I have a program that works in a small amount of memory on a computer
 (less than 50mb) and leaks until it takes all available memory on another
 (with 16GB RAM).

 I am unsure if this is a compiler bug, or a bug in one of the packages I
 use. I am also unsure on how to diagnose it.

 You will find attached the ghc-pkg list for both computers, the offending
 program, and the output of -hp and -hy for the offending case. To
 reproduce the test it should be used like that :

 genstats-simplepatterns dictionary output

 Where dictionary is a large text file with a word per line (such as the
 rockyou list). The program itself is a fold that compiles statistics for
 each input line.

 The "good" program is built with GHC-7.4.1. The bad one used to be built
 with it. As I suspected a package problem, I cleaned all my installation
 and started with a fresh GHC-7.4.2 + latest Haskell platform, but the
 result was identical.

 The space leak seems to occur in HashMap, but converting all instances
 into Data.Map results in another leak.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7211>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler