Program lives on with dead main thread

http://hackage.haskell.org/packages/archive/base/latest/doc/html/Control-Concurrent.html#g:13 says:

In a standalone GHC program, only the main thread is required to terminate in order for the process to terminate. Thus all other forked threads will simply terminate at the same time as the main thread (the terminology for this kind of behaviour is "daemonic threads").

This isn't always true in ghc, as the following program demonstrates:

% cat try.hs
import Control.Concurrent
import Control.Concurrent.STM
import Control.Monad
import System.Exit

main :: IO ()
main = do
    x <- newTVarIO 0
    forkIO $ forever $ do
        --atomically $ writeTVar x 42
        --yield
        return ()
    putStrLn "about to exit"
    exitSuccess
    putStrLn "dead"

% ghc -fforce-recomp try.hs
[1 of 1] Compiling Main             ( try.hs, try.o )
Linking try ...
% ./try
about to exit
^C^C

Instead of actually exiting, the program consumes 100% CPU despite the main thread being gone.

The behavior is slightly different with -threaded:

% ghc -fforce-recomp -threaded try.hs
[1 of 1] Compiling Main             ( try.hs, try.o )
Linking try ...
% ./try
^C^C

Here it doesn't even execute the main thread.

% ./try +RTS -N2
try: Using large values for -N is not allowed by default. Link with -rtsopts to allow full control.

I don't understand what's up with this. 2 isn't what I'd call "large".

% ghc -fforce-recomp -threaded -rtsopts=all try.hs
[1 of 1] Compiling Main             ( try.hs, try.o )
Linking try ...
% ./try +RTS -N2
about to exit

This version works as expected.

Anyway, I'm not sure whether this is a bug in the RTS or in the documentation, but they clearly don't agree.

Trac metadata

Trac field	Value
Version	7.6.2
Type	Bug
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	Compiler
Test case
Differential revisions
BlockedBy
Related
Blocking
CC
Operating system
Architecture