egcs-971127 on m68k-next-nextstep3; Status Report

egcs-971127 on m68k-next-nextstep3; Status Report

[Toon Moene]

Hi,

I managed to build all of egcs-971127 by using a trick to get the  
fixinclude'd include files in the right directory for gcc to find  
them.

I copied all of .../gcc/include/ansi/*.h and  
.../gcc/include/bsd/*.h to .../gcc/include, with the exception of  
ansi/stdio.h, because otherwise I get the following failure during  
the build of the Objective C compiler:

./xgcc -B./  -DIN_GCC    -O2 -g -O2 -I./include  -I. -I. -I./config \
-c ./objc/misc.c -o objc/misc.o
In file included from objc/runtime.h:31,
                 from ./objc/misc.c:30:
include/stdio.h:302: conflicting types for `_flsbuf'
include/stdio.h:153: previous declaration of `_flsbuf'
make[2]: *** [objc/misc.o] Error 1

which is interesting, because there only are 300 lines in stdio.h :-(

This is line 153:

   153                          extern int _flsbuf(unsigned char c,  
FILE *iop);

Anyway, because dejagnu doesn't build on my system, the only  
checking I did was to run c-torture-1.45 (same results as for  
egcs-971105 and much better than gcc-2.7.2.3).

I also ran my own batch of Fortran programs on it (around 200,000  
lines in 1,300 routines spread over half a dozen executables), which  
completed without problems.

I hope the release is progressing on schedule.

Regards,
Toon.

Re: egcs-971127 on m68k-next-nextstep3; Status Report

[Jeffrey A Law]

  In message < 9711301456.AA14740@moene.indiv.nluug.nl >you write:
  > Hi,
  > 
  > I managed to build all of egcs-971127 by using a trick to get the  
  > fixinclude'd include files in the right directory for gcc to find  
  > them.
I think this is the sequence to workaround the include problems.

cd objdir
make all-texinfo all-bison all-byacc all-binutils all-gas all-ld
cd gcc
make bootstrap
make install-headers-tar
cd ..
make bootstrap3


Seems easier than telling them to interrupt the make after building
the fixed include files, but before the C++ runtime libraries
are built.


  > which is interesting, because there only are 300 lines in
  > stdio.h :-(
Make sure you got the right stdio, since there's probably 3 or more
lying around :(

Or add -save-temps and work backwards from the .i file to find the
stdio.h that got included.

  > I hope the release is progressing on schedule.
It's moving along.  doc stuff is a pain and it's difficult to keep
working on it -- too easily distracted.

jeff

egcs-971201 on m68k-next-nextstep3; Status Report

[Toon Moene]

Jeff,

You wrote:

> I think this is the sequence to workaround the include problems.

> cd objdir
> make all-texinfo all-bison all-byacc all-binutils all-gas all-ld
> cd gcc
> make bootstrap
> make install-headers-tar
> cd ..
> make bootstrap3

I just tried this today on egcs-971201, but apparently this only  
works if the user doing the build also has write permission to the  
install directory; otherwise, `make install-headers-tar' won't  
succeed, and the subsequent `make bootstrap3' also fails.

HTH,
Toon.

Re: egcs-971201 on m68k-next-nextstep3; Status Report

[Jeffrey A Law]

  In message < 9712022154.AA18041@moene.indiv.nluug.nl >you write:
  > Jeff,
  > 
  > You wrote:
  > 
  > > I think this is the sequence to workaround the include problems.
  > 
  > > cd objdir
  > > make all-texinfo all-bison all-byacc all-binutils all-gas all-ld
  > > cd gcc
  > > make bootstrap
  > > make install-headers-tar
  > > cd ..
  > > make bootstrap3
  > 
  > I just tried this today on egcs-971201, but apparently this only  
  > works if the user doing the build also has write permission to the  
  > install directory; otherwise, `make install-headers-tar' won't  
  > succeed, and the subsequent `make bootstrap3' also fails.
Yup.  No suprise there.  I don't see a good way to solve the problem 
in-place.

jeff