#4914: FPU initialization required again
---------------------------------+------------------------------------------
Reporter: aruiz | Owner: simonmar
Type: bug | Status: new
Priority: high | Milestone: 7.2.1
Component: Compiler (NCG) | Version: 7.0.1
Keywords: | Testcase:
Blockedby: | Difficulty:
Os: Unknown/Multiple | Blocking:
Architecture: x86 | Failure: Incorrect result at runtime
---------------------------------+------------------------------------------
Comment(by aruiz):
I have run the hmatrix tests checking the fpu state at the beginning of
the foreign calls and the presence of NaN's in the results. In a large
number of cases the stack pointer is not 0 and there is no problem.
But in a few cases in which the foreign call receives a nonzero stack
pointer the foreign computation produces a wrong NaN. I have discovered
that I must only call finit to prevent the error if the fpu is in one of
the following three states:
{{{
14624 : 0011 1001 0010 0000
14629 : = = = 0101
14631 : = = = 0111
}}}
In this particular experiment any other state is harmless. I know very
little about this, but from the info in
[http://webster.cs.ucr.edu/AoA/Windows/HTML/RealArithmetic.html#1000117
this page] it looks like the only common pattern is a top of stack = 7 and
precision = 1, which is not problematic in many other cases.
I will try to find a small testcase producing a NaN.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4914#comment:10>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
