This is the mail archive of the
gdb-patches@sourceware.cygnus.com
mailing list for the GDB project.
Re: change to remote protocol qOffsets response handling
- To: jtc at redback dot com
- Subject: Re: change to remote protocol qOffsets response handling
- From: Michael Snyder <msnyder at cygnus dot com>
- Date: Tue, 05 Oct 1999 16:14:58 -0700
- CC: gdb-patches at sourceware dot cygnus dot com
- Organization: Cygnus Solutions
- References: <5m7ll176w8.fsf@jtc.redbacknetworks.com>
J.T. Conklin wrote:
>
> The qOffsets query, which was originally added for NetWare systems, is
> used to handle cases where the image being debugged has been relocated
> by the target. It returns the new addresses of the text, data, and
> bss sections. But GDB used the value returned for the data section
> for the bss section as well, with the comment:
>
> /* This is a temporary kludge to force data and bss to use the
> same offsets because that's what nlmconv does now. The real
> solution requires changes to the stub and remote.c that I don't
> have time to do right now. */
>
> But this makes it impossible for other systems that do relocate data
> and bss sections separately to communicate that fact to GDB. IMO, it
> seems better for the NetWare stub to return the same value to GDB for
> data and bss offsets, rather than GDB to make that assumption for all
> targets.
Agreed, but is it polite to just change it out from under them,
without notice, consent or discussion?