This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: RFA: Breakpoint infrastructure cleanups [0/8]
- From: Michael Snyder <msnyder at redhat dot com>
- To: Michael Elizabeth Chastain <mec at shout dot net>
- Cc: ezannoni at redhat dot com, ac131313 at redhat dot com, drow at mvista dot com, gdb-patches at sources dot redhat dot com
- Date: Wed, 15 Oct 2003 15:00:18 -0700
- Subject: Re: RFA: Breakpoint infrastructure cleanups [0/8]
- Organization: Red Hat, Inc.
- References: <200310151948.h9FJm0W2025223@duracef.shout.net>
Michael Elizabeth Chastain wrote:
My two cents:
It's not just breakpoints. What happens when the user types:
(gdb) disassemble Foo::Foo
That's true -- or taking a step back, we currently use one semantics
(and syntax) for all commands that take a source location as an argument
(basicly break and list). Are we going to 'break' that? I think it's
OK to do so, because for list, you really do mean a source location,
whereas for break, you may mean many target locations.
Disassemble doesn't share the same syntax/semantics (decode_line_1);
it has its own (I think?) -- but you're right, it's another command
that takes a source location and "translates" it into a target
location.
Also, here is a use case to consider:
(gdb) break *0x12345678
gdb has this notion that one source address corresponds to one
object-code address, but now it's one source address corresponds
to N object-code addresses.
Now I don't follow you. "*0x12345678" isn't a source address.
Is it? Unles you've got multiple memory address spaces or
something... something more deviant than just Harvard, which
we already know how to handle.