Bugs item #1032241, was opened at 2004-09-22 01:30
Message generated for change (Comment added) made by wthaller
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=108032&aid=1032241&group_id=8032

>Category: Compiler (NCG)
Group: 6.2.1
>Status: Closed
>Resolution: Wont Fix
Priority: 5
Submitted By: Mark Tullsen (mtullsen)
>Assigned to: Wolfgang Thaller (wthaller)
Summary: power-pc code generation error

Initial Comment:
The following code (in Test.hs, attached):
  module Test where 
  data T = C1
            | C2
            ...
            | C200
results in an assembler error when compiling natively (for the 
PPC):
  $ ghc -c Test.hs 
  /tmp/ghc8899.s:6652:Fixup of 45900 too large for field width of 
16 bits
  /tmp/ghc8899.s:6650:Fixup of 45908 too large for field width of 
16 bits

I don't have a problem when I compile via C:
  $ ghc -fvia-C -c Test.hs 
  $ 

Environment:
  ghc 6.2.1
  Mac OS X
  PPC 


----------------------------------------------------------------------

>Comment By: Wolfgang Thaller (wthaller)
Date: 2004-10-01 04:52

Message:
Logged In: YES 
user_id=566359

Thanks for the report. 
Yes, I know that this can happen. Your large datatype results in more 
than 32K of code being generated, which the PowerPC's conditional 
branches can't handle. GHC currently doesn't do anything to work around 
this.
GHC 6.4 will feature a completely rewritten backend, which, while not 
really fixing this issue, will make it much less likely to happen in 
practice. Therefore, I am closing this bug as "Won't Fix".
For now, I recommend that you just add {-# OPTIONS -fvia-C #-} to the 
top of your source file; after GHC 6.4 is released, you'll be able to 
remove this.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=108032&aid=1032241&group_id=8032
_______________________________________________
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Reply via email to