Re: Assume solib.h

[Andrew Cagney <cagney@gnu.org>]

Randolph Chung wrote:
>>There's just one non-technical nit.
>>
>>It means breaking non solib.[hc] shared library systems.  Kevin 
>>indicated that there were two - AIX and HP/UX remaining.  I think we can 
>>live with that - we've patiently waited for what, more than two years 
>>for nothing to happen, so it is now time to give things that gentle pus
> 
> 
> I'm not prepared to sign up for hpux support, but can you explain some
> more what will break and what is needed to fix it? Despite the lack of
> maintainence, there are still a good number of people out there using
> gdb on hpux (judging by private mail I've received since I started 
> working on hppa-linux support)....

Have a look at solist.h which contains:

> struct target_so_ops
>   {
>     /* Adjust the section binding addresses by the base address at
>        which the object was actually mapped.  */
>     void (*relocate_section_addresses) (struct so_list *so,
>                                         struct section_table *);
...

it just needs to implement that object (see solib-svr4.c).  Without it, 
shared libraries wouldn't work but everything else should.

We need to find a way of flushing some of these people still using GDB 
on HP/UX (or are they using HP's WDB fork?) and, unfortunate as it is, 
push-come-to-shove is one of the most effective ways of doing this.

Anyway, HP/UX has a more immediate problem - it's still using 
deprecated_registers[] and that's now past its end-of-life :-/

Andrew