This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
Re: GDB's remote protocol: My proposed extention
- To: Andrew Cagney <ac131313 at cygnus dot com>
- Subject: Re: GDB's remote protocol: My proposed extention
- From: Stephen Smith <ischis2 at home dot com>
- Date: Wed, 21 Mar 2001 13:15:31 -0700
- CC: GDB <gdb at sources dot redhat dot com>, Kevin Buettner <kevinb at cygnus dot com>, Michael Elizabeth Chastain <chastain at cygnus dot com>
- References: <3AB6CB9A.C0883CD9@home.com> <3AB7867B.3D80F74A@cygnus.com>
> Stephen Smith wrote:
> >
> > When reading chapter 13 of the gdb manual, the rcmd can be taken to mean one
> > of two things:
> >
> > 1) a command sent from the host machine to the target machine for the GDB server/stub to
> > execute
>
> This one. All (well ok almost all) packets are GDB->target->GDB.
>
> Andrew
Ok, I have been poking around in the GDB code like you suggested and here is my proposal for changes in the remote.c
file:
1) Until this is working use an environment variable to turn on this feature * especially since I don't know how to do
it right - yet*
2) Add a "qLibraries" general query. This query would expect a response of the form "sharedLib1, address1;
sharedLib2, address2; sharedLib3, address3"
3) For each library/address pair in the return, call add_symbol_file_command() from symfile.c.
Advantage is that this is a high level function and should be processor/coef/elf independent.
4) Add a "qNewLibraries" general query which would return a `1` or a `0`.
What do you think?
BTW, I keep trying to get a copy of the GNU foundation copyright assignment document so that I can submit the patch when
I am done, haven't
heard a thing from them. Any suggestions on that also?
sps