Segmentation fault/access violation in generated code

[NeilMitchell]

F:\Temp\supero\supero-msg\tak>ghc --make Main_.hs -o foo -O2
[1 of 1] Compiling Main             ( Main_.hs, Main_.o )
Linking foo.exe ...

F:\Temp\supero\supero-msg\tak>foo 24 16 8
Segmentation fault/access violation in generated code

F:\Temp\supero\supero-msg\tak>ghc --version
The Glorious Glasgow Haskell Compilation System, version 6.8.1

Windows XP. The test case is rather big, and needs reducing - I'll come back and do that in a few days time unless anyone beats me to it (I have a paper deadline in 23 hours...)

[NeilMitchell]

I should point out that I'm fairly certain that this code doesn't do what it was intended to (the tak benchmark), but it still shouldn't segfault.

[simonmar]

Thanks Neil, I think we have enough to go on. (hint: the binding f34 is an infinite loop, avoiding that somehow will probably work around the problem).

Simon: let's discuss this one next week, I have some suspicious-looking entrails but I'm not entirely sure where to place the blame.

[simonpj]

Fixed, thank you.

Tue Dec  4 14:58:03 GMT 2007  simonpj@microsoft.com
  * Make eta reduction check more carefully for bottoms (fix Trac #1947)

Ian: please merge. And could you make this program into a test? (Compile with -O2 and run as above.) Thanks

Simon

[igloo]

Merged. Still need to make the test.

[igloo]

I couldn't reproduce the problem with 6.8.2, but I have now added the program as test simplrun010 nevertheless.