[Bug fortran/17502] New: ICE: segmentation fault on (believed) correct code

[Bug fortran/17502] New: ICE: segmentation fault on (believed) correct code

[pdellar at na-net dot ornl dot gov]

g77 -c -O0 -ffixed-line-length-none -fno-second-underscore -fschedule-insns2
on the following (believed correct, or at least working with 3.4.1) code causes
an ICE. Other optimisation levels work correctly, as does -O0 without
-fschedule-insns2.

g77 3.4.2 built with 

../gcc-3.4.2/configure --prefix=/opt/gcc-3.4.2
make -j 2 bootstrap

on a dual Itanium running Linux 2.4.20-mckinley-smp.


Problematic fortran file begins here:

# 1 "surface.F"
# 1 "<built-in>"
# 1 "<command line>"
# 1 "surface.F"

      implicit none
      integer mx,mz,modes
      parameter (mx=256,mz=128)
      common /modeblk/ modes
      integer i,j
      real*8 x,z,eta,pi,delta,t,dt,dx
      parameter (pi=3.14159265358979d0,delta=1d0)

      real*8 p(mx+2,mz),surf(mx+2),realsurf(mx+2),surfdot(mx+2)
      common /p_blk/ p

      real*8 preal(mx+2,mz)

      real*8 surfx(mx+2),surfxx(mx+2)

      integer info(30),itol,istate,itask,mf,iopt,ifail,lid
      integer lrw,liw,nvars
      parameter (nvars=mx+2)
      parameter (liw=1000+3*nvars)
      parameter (lrw=1000+20*nvars)
      integer iwork(liw)
      real*8 rwork(lrw)

      real*8 tlast,tprint,tend

      external calcdot,psol,jac

      real*8 fzint              ! function for calculating volume

      real*8 rtol,atol

      real*8 maxsurf,maxcurv,fluid

      character*150 dirname,comname,logname,filename
      character*150 timestr

      common /dirblk/ dirname

      real*8 px(mx+2),peta(mx+2)
      common /pdiff_blk/ px,peta

      real*8 tension
      parameter (tension=(1d-4))

      integer slice
      parameter (slice=5)

      modes=mx                  ! no adaptivity

      dx=2d0*pi/dble(slice*mx)

C     initialise free surface
      do i=1,mx
         x=dble(i-1)*dx
         surf(i)=(1d0+0d0*cos(4d0*x))
      enddo

      do j=1,mz
         do i=1,mx
            p(i,j)=0d0          ! initial guess for p
         enddo
      enddo

 99   format(20G20.10)

      do i=1,30
         info(i)=0
      enddo


      itol=1
      itask=1
      istate=1
      iopt=0


      atol=1d-8
      rtol=1d-8


      t=0d0
      tlast=0.001d0
      tprint=0d0

      open (50,file=comname,status='unknown')


 123  continue

      tprint=tprint+tlast

      if (t.lt.100d0) goto 123

      end

-- 
           Summary: ICE: segmentation fault on (believed) correct code
           Product: gcc
           Version: 3.4.2
            Status: UNCONFIRMED
          Severity: minor
          Priority: P2
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pdellar at na-net dot ornl dot gov
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: ia64-unknown-linux-gnu
  GCC host triplet: ia64-unknown-linux-gnu
GCC target triplet: ia64-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17502

[Bug target/17502] ICE: segmentation fault on (believed) correct code

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|fortran                     |target
           Keywords|                            |wrong-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17502

[Bug target/17502] [g77] ICE: segmentation fault on (believed) correct code

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-09-29 12:48 -------
What is the ICE?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|ICE: segmentation fault on  |[g77] ICE: segmentation
                   |(believed) correct code     |fault on (believed) correct
                   |                            |code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17502

[Bug target/17502] [g77] ICE: segmentation fault on (believed) correct code

[dellar at maths dot ox dot ac dot uk]

------- Additional Comments From dellar at maths dot ox dot ac dot uk  2004-09-29 15:12 -------
Subject: Re:  [g77] ICE: segmentation fault on (believed)
 correct code

This is how I invoked the compiler on the misbehaving file (further copy
attached) and the resulting output:

amulet{pjd12}12: /opt/gcc-3.4.2/bin/g77 -c -ffixed-line-length-none -fschedule-insns2 surface.f
surface.F: In program `MAIN__':
In file included from surface.F:0:
surface.F:95: internal compiler error: Segmentation fault      <--- NB
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.
------- Additional Comments From dellar at maths dot ox dot ac dot uk  2004-09-29 15:12 -------
Created an attachment (id=7237)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7237&action=view)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17502

[Bug target/17502] [g77] ICE: segmentation fault on (believed) correct code

[sje at cup dot hp dot com]

------- Additional Comments From sje at cup dot hp dot com  2005-06-02 21:18 -------
I built a 3.4.2 GCC on an IA64 Linux box and tried to reproduce the bug but
could not.  The included test case compiled with no problem.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sje at cup dot hp dot com
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |WORKSFORME


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17502