This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


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

Re: g++ egcs-1.1a internal compiler error



> > JHB NIJHOF <j.h.b.nijhof@aston.ac.uk> writes:
> > 
> > > Version of g++ (g++ -V): egcs-2.91.57
> > > Operating system: Debian Linux 1.3
> > 
> > I couldn't reproduce your problem with the 1.1-pre-release on neither
> > RedHat Linux 4.0/x86 nor 5.0/x86, and, AFAIK, no x86-related change
> > was done between the pre-release and the final release.  Did you run
> > the compiler with any argument other than the source file name?
> > I tried with and without optimization.
> 
> I was able to reproduce the problem on 
> 
> i486-pc-linux-gnulibc1
> 
> on a mixed Slackware / Debian Linux 1.3 system.

I checked again, and it can't be the #include <math.h> or
#include <stdlib.h> : I could minimalise the file to

// bug2.cc                             //  1
                                       //  2
#include <fstream.h>                   //  3
                                       //  4
void bug()                             //  5
{                                      //  6
  ifstream infile("kww");              //  7
  int npts = 1;                        //  8
  float floatarray[4 * npts + 1];      //  9
  char scratch[28];                    // 10
  infile.read(scratch,28);             // 11
  int arraylength = 20;                // 12
  if (!arraylength)                    // 13
    ;                                  // 14
  infile.read(floatarray,arraylength); // 15
  return;                              // 16
}                                      // 17

Still with egcs-1.1a (i686-pc-linux-gnulibc1, egcs-2.91.57),
I get the same error: g++ -c bug2.cc =>  Internal compliler error on line 15.

Comment out line 11, `infile.read(scratch, 28);', and the error disappears;
Comment out line 13, `if (!arraylength)', and the error changes to
   sorry, not implemented: testing field_decl for template parms.
Change line 9 to `float floatarray[5];' and all goes well (hardly surprising).



I have reproduced this error on egcs-1.1a compiled on a Sun,
sparc-sun-solaris2.5 , with egcs configured as follows: [=config.status]:

#!/bin/sh
# This file was generated automatically by configure.  Do not edit.
# This directory was configured as follows:
../egcs-1.1a/configure --with-gcc-version-trigger=/net/dibble/export/home/nijhojhb/egcs-compile/egcs-1.1a/gcc/version.c --host=sparc-sun-solaris2.5 --prefix=/net/dibble/export/home/nijhojhb/egcs --enable-version-specific-runtime-libs --enable-shared --norecursion
# using "mh-frag" and "mt-frag"

(i.e. compiled again using --prefix,  --enable-version-specific-runtime-libs,
 and --enable-shared).

Now the error message is:

bug2.cc:15: Internal compiler error 274.

That might narrow down the search a bit further.

Yours,

Jeroen Nijhof


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