Re: RFA: AIX 64-bit mega-patch

[Andrew Cagney <ac131313 at cygnus dot com>]

Nick Duffek wrote:
> 
> The appended patch implements 64-bit AIX support using the multi-arch
> framework.
> 
> In 32-bit mode, there are no test suite regressions and many progressions.
> There are a number of 64-bit failures that I'm in the process of fixing.
> 
> The patch tries to set the current architecture from the current core file
> if there is one and from the current target machine otherwise.  There's an
> obscure problem with this: if "core-file <file>" is entered while an
> inferior process exists, the architecture should remain the same until the
> process exits, at which point the architecture should switch to that of
> the core file.  Instead, the architecture is immediately switched to that
> of the core file.
> 
> One way to fix that would be to postpone the set_gdbarch_from_file() in
> core_open() if !ontop.  I haven't attempted such a fix.
> 
> The patch relies on some BFD changes that I'll be posting to the binutils
> mailing list in a few minutes.
> 
> Okay to apply?  If there are no major problems, I'd prefer to commit this
> as-is and then address any issues with follow-on patches.

Just some suggestions:

Drop a tag before the commit.  For the MIPS I went as far as creating
branch since I know I'm going to adding debug code to it so I can figure
out why obscure target xyz broke.  I suspect for here a branch is over
kill.

Since there is a BFD dependency you'll need to wait for that (sigh :-). 
It may be possible to carve up this patch and get some of it in
immediatly.  I guess, wait and see.

As for the "core-file <file>" problem.  It will need a little thought. 
The theory is that, eventully, an architecture will be bound to each
``thread''.  An interum solution could have the architecture bound to
the target.

	Andrew