On Mon, Jul 25, 2011 at 03:45:15PM -0600, Matthias Kilian wrote: > Modified files: > lang/ghc : ghc.port.mk > > Log message: > Switch back to compiling Setup.l?hs and running the resulting program > instead of using runghc (the interpreter). [..] > Note that this is not a *fix*. I'm sure there's some misbehaving > signal handler, either in runghc/ghci or in the ghc runtime library.
Note to myself (and everyone interested in ghc): to be sure it's not a problem in libpthread, build a non-threaded ghc and run some build loops with interpreted Setup.l?hs. Lack of failure in this case wouldn't mean much, however. There are a lot of #ifdef's in GHC, and what if a signal handler calls some dangerous function iff ghc is built with threading support?
