This is the mail archive of the
mailing list for the GDB project.
Re: [RFC] i386/amd64 biarch gdbserver
- From: Pedro Alves <pedro at codesourcery dot com>
- To: gdb-patches at sourceware dot org
- Cc: Doug Evans <dje at google dot com>
- Date: Mon, 11 May 2009 18:50:46 +0100
- Subject: Re: [RFC] i386/amd64 biarch gdbserver
- References: <20090509191225.404D984890@localhost>
On Saturday 09 May 2009 20:12:25, Doug Evans wrote:
> This patch implements biarch i386/amd64 gdbserver.
> I haven't done anything with xml target descriptions, but there
> are no regressions so it seems like a good stopping place.
Yeah, this came up on IRC, when I mentioned about gdbserver
multi-process, and debugging 32-bit and 64-bit inferiors
simultaneously. The x86 and amd64 register sets aren't compatible,
and, we've got a single target_gdbarch to represent the connection, so
this is a problem. But it's one we can solve on top of this.
> This patch also fixes a bug where current amd64 gdbserver
> doesn't handle xmm[8-15]. Testcase to follow later.
> Tested on
> - i386-linux
> - x86_64-linux with gcc -m64
> - x86_64-linux with gcc -m32 (and hacking site.exp to specify i386-linux)
> using --target_board=native-gdbserver
> [ref http://sourceware.org/gdb/wiki/Native_gdbserver_testing]
This looks good to me, but, I don't know how far can I stretch in
Small nit, you've added an `assert' call, but I think you should use `fatal'
instead -- there are no other references to `assert' in gdbserver whole.
> +/* Return non-zero if the target is 64-bit. */
> +static void
> +x86_arch_setup (void)