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: Regressions caused by copy_string patch.


On Fri, Jul 14, 2006 at 10:19:18AM -0700, Steve Kargl wrote:
> FX,
> 
> Your patch fixes the regression for interface_derived_type_1.f90
> and present_1.f90.  The regression with actual_array_substr_2.f90
> is still occurring.   It may be that the copy_string patch has
> uncovered a latent bug in the middle/back-end because this failure
> is an ICE in explow.c.  I'll investigate.  FX, if you don't get
> around to committing this patch, I'll do it sometime this weekend.
> 

I spoke with pinskia on IRC.  The exchange was

<kargl>   troutmask:sgk[259] gfc4x -o z -O0 -static actual_array_substr_2.f90 
<kargl>   actual_array_substr_2.f90: In function 'MAIN__':
<kargl>   actual_array_substr_2.f90:10: internal compiler error: ini
          copy_to_mode_reg, at explow.c:635
<kargl>   Please submit a full bug report,
<pinskia> isn't that the thread about the fortran regressions?
<kargl>   yes.  FX sent a patch that fixes 2 of the 3.  This is the onei
          that isn't fixed.
<kargl>   Jack has reproduced it on PPC with -m64 and I see it on amd64,
<pinskia> well the assert is  gcc_assert (GET_MODE (x) == mode ||
          GET_MODE (x) == VOIDmode);
<pinskia> which most likely means there is still a type mismatch

-- 
Steve


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