Re: MI and remote debugging

[Daniel Jacobowitz <drow@mvista.com>]

On Wed, Feb 13, 2002 at 01:26:04AM -0500, Andrew Cagney wrote:
> >Any remote config (like gdbserver, or even monitor, I think...) that
> >overrides gdb_target_cmd/gdb_file_cmd loses because of the static mi 
> >version
> >in mi-support.exp.  This leads to almost a day's worth of timeouts.  I've
> >tried to cobble together a gdbserver version of mi_gdb_file_cmd and an
> >mi-aware mi_gdb_target_cmd; I can't do it.
> >
> >Andrew, any thoughts?
> 
> For targets other than ``target remote'' I don't think we need to be too 
> concerned.  However, ``target remote'' variants should work.  I know it 
> works for SID (a remote simulator).
> 
> Can you expand a little.

For target remote it does not necessarily work.  Look at
config/gdbserver.exp to see what you need to do to spawn a gdbserver on
load; mi overrides that function entirely.  The comment
"# ``gdb_file_cmd''" in mi_gdb_file_cmd appears to be as close as we
come to thinking about this.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer