This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [rfa:sparc/rfc] multi-arch SOFTWARE_SINGLE_STEP et.al.
- To: Andrew Cagney <ac131313 at cygnus dot com>, gdb-patches at sources dot redhat dot com
- Subject: Re: [rfa:sparc/rfc] multi-arch SOFTWARE_SINGLE_STEP et.al.
- From: Michael Snyder <msnyder at cygnus dot com>
- Date: Thu, 01 Feb 2001 13:49:50 -0800
- Organization: Red Hat
- References: <3A6A1DF4.C6760CEF@cygnus.com> <3A774357.D5DB0361@cygnus.com> <3A774B50.4B9C93B9@cygnus.com> <3A790917.D58BDD87@cygnus.com>
Andrew Cagney wrote:
>
> Michael Snyder wrote:
> >
> > Andrew Cagney wrote:
> > >
> > > Andrew Cagney wrote:
> > >
> > > > For the sparc64 this multi-arches software_single_step. Maintainer?
> > > > Test on a solaris N+1 machine?
> > >
> > > So, I've managed to get a solaris 7 machine. How should I configure
> > > things so that I know I'm testing the sparc64 support?
> >
> > Just make sure that the default gcc on your path (or $CC) defaults to 64-bit.
> > Everything else should follow automatically.
>
> Ok, found one. Must be working because:
>
> $ file corelow.o
> corelow.o: ELF 64-bit MSB relocatable SPARCV9 Version 1
>
> > During development, when I had a prototype gcc that would do 64-bit
> > but not as the default, I simply defined CC = "gcc -m64".
>
> Only ....
>
> gcc -c -g -O -I. -I/skil/cagney/SINGLE-STEP/src/gdb
> -I/skil/cagney/SINGLE-STEP/src/gdb/config -DHAVE_CONFIG_H
> -I/skil/cagney/SINGLE-STEP/src/gdb/../include/opcode
> -I/skil/cagney/SINGLE-STEP/src/gdb/../readline/.. -I../bfd
> -I/skil/cagney/SINGLE-STEP/src/gdb/../bfd
> -I/skil/cagney/SINGLE-STEP/src/gdb/../include -I../intl
> -I/skil/cagney/SINGLE-STEP/src/gdb/../intl -DUI_OUT=1 -Werror
> /skil/cagney/SINGLE-STEP/src/gdb/core-sol2.c
> In file included from /skil/cagney/SINGLE-STEP/src/gdb/core-sol2.c:35:
> /usr/include/v9/sys/privregs.h:20:20: sys/asi.h: No such file or
> directory
> make: *** [core-sol2.o] Error 1
> cagney@skil$ file corelow.o
>
> I appended an extra -I/usr/include/v9 to CFLAGS. Should that be needed?
No it shouldn't. This seems to be a bug in Sun's system include files.
As you can see, /usr/include/v9/sys/privregs.h attempts to
#include <sys/asi.h>, which does not exist. Your workaround will
get you past it, but I wish Sun would fix it. We brought it to
their attention some time ago.