This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: jit_breakpoint_re_set_internal desc_symbol issue
- From: Tom Tromey <tromey at redhat dot com>
- To: Doug Evans <dje at google dot com>
- Cc: gdb-patches <gdb-patches at sourceware dot org>
- Date: Tue, 16 Apr 2013 14:52:55 -0600
- Subject: Re: jit_breakpoint_re_set_internal desc_symbol issue
- References: <yjt2mwt3s7u5 dot fsf at ruffy2 dot mtv dot corp dot google dot com> <CADPb22SSjw-+_9a7hSeQUzmfGYPc=6FKhZuDhgQ94RP3rtfhfw at mail dot gmail dot com>
>>>>> "Doug" == Doug Evans <dje@google.com> writes:
Doug> Is there a well defined set of places where we need to start watching
Doug> for changes requiring breakpoint-resetting,
Doug> and a well defined set of places where we need the resetting done by?
Doug> Seems like there should be.
Doug> I was thinking IWBN if all the various places that need
Doug> breakpoint-resetting done could just (effectively) set a flag, and
Doug> have this flag checked in the few well-defined places that need it.
Doug> Plus, how much benefit would there be for more granularity? A lot of
Doug> times we just do a global resetting, when the code that needs the
Doug> resetting done only needs it for, say, a single objfile.
"Fine-grained breakpoint re-setting" has been on our to-do list for a while.
I think Keith was going to look at that after the linespec stuff he's
working on.
The basic idea is that many re-sets can be handled more efficiently.
For example if a .so is removed, we don't need to fully reset
all breakpoints, just remove some locations.
I think the challenge is enumerating the kinds of resets and figuring
out a clean way to distribute the information to the various spots that
need it.
Tom