Re: [commit/AIX] Fix error when loading core file

["Ulrich Weigand" <uweigand@de.ibm.com>]

Pedro Alves wrote:
> Pedro Alves wrote:
> > Ulrich Weigand wrote:
> >>
> >> I'd thought of defining a TARGET_OBJECT_LDINFO xfer_partial object
> >> that reflects the contents of the ldinfo, and use this in the
> >> implementation of a (platform-independent) solib-aix.c file.
> >> The rs6000 native target and the core file support would need to
> >> be extended to supply that object.
> >>
> > 
> > If AIX needs to use native calls to get at the shared libraries list,
> > perhaps you could also look if the solib-target.c Daniel is pushing
> > for remote dll support, and see if it would be a good fit for AIX too.
> > As a quick look it seems so.  You would move aix_current_sos
> > to the target_ops, and perhaps needs a few other tweaks 
> > (aix_in_dynsym_resolve_code, and aix_solib_create_inferior_hook).
> > Although he presented it as remote support, it also makes sense
> > for native win32 debugging, so it may also make sense as native
> > AIX debugging.
> > 
> 
> Hummm, looking at it for another 30 seconds, made me realize
> it isn't fit.  Although AIX tells us something loaded with
> 'wait (&status), (status == 0x57c)', it doesn't tell us
> specifically *what* was loaded - we have to build a list
> of current_sos every time.  Being so, case solib-target.c
> gets more in the way than helps.  Your approach seems best.

Another reason why I like the TARGET_OBJECT_LDINFO approach
is that the ldinfo is also present in core files.  Using a
xfer_partial object allows to cleanly support both native
and core file debugging (and it can also be extended to
support remote debugging, if we ever want to do that on AIX).

In any case, thanks for looking at this!

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com