Re: RFA: libunwind basic support

["J. Johnston" <jjohnstn at redhat dot com>]

Patch checked in.  I forgot acconfig.h in the original patch which just added 
the #undef HAVE_LIBUNWIND reference.  The following is the amended ChangeLog entry.

2003-11-14  Jeff Johnston  <jjohnstn@redhat.com>

        * libunwind-frame.c: New file.
        * libunwind-frame.h: New file.
        * acconfig.h (HAVE_LIBUNWIND): #undef reference added.
        * configure.in: Add --with-libunwind option support.  Define
        HAVE_LIBUNWIND if libunwind support selected / defaulted.
        Also check for libunwind.h and libunwind-ia64.h headers.
        * configure: Regenerated.
        * Makefile.in: Add support for libunwind-frame.o.
        * config.in: Regenerated.

Thanks,

-- Jeff J.

Andrew Cagney wrote:
> > It is possible for the user to enable libunwind but the header files 
> > are not there.  I am guessing you would prefer compilation to fail in 
> > this case so I have removed the check.
>
>
> Yes.  If the user forces the use of libunwind, but it isn't there, the 
> user should loose.
>
> > I have added comments for the search unwind table call.
>
>
> Thanks.
>
>
> > Index: configure.in
> > ===================================================================
> > RCS file: /cvs/src/src/gdb/configure.in,v
> > retrieving revision 1.132
> > diff -u -r1.132 configure.in
> > --- configure.in    3 Sep 2003 15:02:48 -0000    1.132
> > +++ configure.in    22 Oct 2003 23:36:21 -0000
> > @@ -192,6 +192,33 @@
> >      enable_gdbtk=no ;;
> >  esac
> >
> >
> > Based on 12.1 Working With External Software
> > http://www.gnu.org/software/autoconf/manual/autoconf-2.57/html_chapter/autoconf_12.html#SEC130 
> >
> >
> > I suspect that this should be --with (external libunwind software), 
> > and not --enable (gdb feature), although, to be honest, its hard to 
> > tell which it should be.
> >
> > Anyone?
> >
> >
> > From discussions with Jeff Law when I was doing configuration for 
> > newlib, he told me that --enable should be used for on/off decisions 
> > in preference to --with.  I don't personally care so I have switched 
> > to --with.
>
>
> Thanks.
>
> You're right that the lines are blured.  enable/disable is for 
> controlling the presence of an internal feature while with/without is 
> for controlling a dependency/use of an external program, or library. 
> Unfortunatly, here the decision to build GDB with/without the external 
> libunwind library has the consequence of determining if GDB should 
> enable/disable the internal (but not user visible) libunwind unwinder. I 
> think, based on the presence of X as an example in the manual, the 
> with/without takes precidence.
>
> So, yes all ok.
> Andrew
>
>
>
> > Index: libunwind-frame.c