This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
Re: RFC: Return number of bytes written in M packet
- To: Andrew Cagney <ac131313 at cygnus dot com>
- Subject: Re: RFC: Return number of bytes written in M packet
- From: Quality Quorum <qqi at world dot std dot com>
- Date: Mon, 14 May 2001 18:01:27 -0400
- cc: GDB Discussion <gdb at sources dot redhat dot com>
On Mon, 14 May 2001, Andrew Cagney wrote:
> Hello,
>
> This is intended to close what I think is a hole in the current memory
> write packet - there isn't a way that a target can report back a partial
> (i.e. fragmanted) write.
>
> The change is simple. As well as:
>
> <- "OK"
> <- ""
> <- "Enn"
>
> allow:
>
> <- <BE-hex-nr-bytes-written>
>
> as a reply.
>
> Memory reads already allow partial transfers.
I am kind of scared by the ad hoc nature of this development, there are
number of holes in gdb protocol and I would like to address them all
in systematic way, otherwise we will simply run of characters.
Just a few examples - I suppose that the current method of collecting
thread information is way too ineffective, 'z' commands do deserve a
better support from initial handshake (stub has to have a way to recognize
initial hadnshake and clear all the break points) and some fortification,
thread model has to be settled down etc, etc, etc.
I tried to answer some of these questions:
http://world.std.com/~qqi/download/protocol.txt
http://world.std.com/~qqi/download/i386_stub.tar.gz
There are plenty to do still, especially if we going into the area of
changes which are not going to be backward compatible.
> Andrew
Thanks,
Aleksey