Re: hppa1.1 cross gnat question

[Joel Sherrill <joel dot sherrill at OARcorp dot com>]

John David Anglin wrote:
> 
> > When building hppa1.1-rtems cross gnat I ran into the problem
> > where the build used -mdisable-indexing with the native
> > GNU/Linux compiler.  This is an hppa specific option.  I
> 
> This was the behaviour before yesterday.  Is it possible you're
> looking at the new code and used the old code for the build?
> I don't see this when I configure for --target=hppa1.1-rtems
> on i386-linux.

Sounds possible.  I am looking at a lot of targets,
updating my source daily, and doing a lot of automated
builds. :)  I think I had done my daily update before
going to look for disable-indexing.

For sure my overnight build hppa1.1-rtems did not fail
in the same way -- it failed due to something about
the newlib/rtems sys/dirent.h.

../../xgcc -B../../ -c -DCROSS_COMPILE -DIN_GCC   `echo -O2 -g  -DIN_RTS
|sed -e
 's/-pedantic//g' -e 's/-Wtraditional//g'`  -O2 -g    -I. -I.. -I../..
-I/opt/us
r3/ftp_archive/gnu/gcc/ss/b3.1-branch/gcc-to-merge/gcc/ada
-I/opt/usr3/ftp_archi
ve/gnu/gcc/ss/b3.1-branch/gcc-to-merge/gcc/ada/..
-I/opt/usr3/ftp_archive/gnu/gc
c/ss/b3.1-branch/gcc-to-merge/gcc/ada/../config
-I/opt/usr3/ftp_archive/gnu/gcc/
ss/b3.1-branch/gcc-to-merge/gcc/ada/../../include -I./../.. adaint.c
In file included from
/opt/gcc-3.1-test/hppa1.1-rtems/include/dirent.h:6,
                 from adaint.h:35,
                 from adaint.c:129:
/opt/gcc-3.1-test/hppa1.1-rtems/include/sys/dirent.h:44: `NAME_MAX'
undeclared h
ere (not in a function)
/opt/gcc-3.1-test/hppa1.1-rtems/include/sys/dirent.h:45: confused by
earlier err
ors, bailing out          



> > tracked it down to here:
> >
> > ../../gcc/config/pa/x-ada:ADA_CFLAGS=-mdisable-indexing
> >
> > This should only be used for native hppa->hppa compiles.
> 
> No. It's not just for native compiles.  It's a host option.  What
> host was specified for the cross?  If the host is not one of the
> hppa ports that have xmake_file defines in config.gcc, do you see
> this override in both the gcc and gcc/ada Makefiles?

OK.  If this is how it is now, then it is OK.  I just made a mistake
looking at a log and source tree that were slightly out of sync.

Whoever fixed this -- thanks. :) 

> Dave
> --
> J. David Anglin                                  dave.anglin@nrc.ca
> National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel@OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985