This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] fortran/arith.c -- Fix the range of integers in overflow checks


On Sun, Aug 27, 2006 at 08:30:37PM -0400, Jack Howarth wrote:
> ps Since I don't see the failure of intrinsic_set_exponent
> at -m64 on Darwin PPC, I conclude we don't have the bug
> on that arch.

I haven't converted the octal constant to an integer
value, so failure may not occur.  The test has lines
like
   i4 = o'0001234561'    ! integer(4) constant
   i8 = o'0001234561'_8  ! integer(8) constant
The integer(4) portion of code fails due to an overflow.
The i8 may not overflow.  However, I believe the BOZ is 
converted to an integer(16) in both cases, and then  
converted upon assign to the lower integer(*) form.

-- 
Steve


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]