This is the mail archive of the
gdb-prs@sources.redhat.com
mailing list for the GDB project.
gdb/469: replace flush_cached_frames() and registers_changed() with target_changed()
- From: ac131313 at redhat dot com
- To: gdb-gnats at sources dot redhat dot com
- Date: 6 Apr 2002 15:04:54 -0000
- Subject: gdb/469: replace flush_cached_frames() and registers_changed() with target_changed()
- Reply-to: ac131313 at redhat dot com
>Number: 469
>Category: gdb
>Synopsis: replace flush_cached_frames() and registers_changed() with target_changed()
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: unassigned
>State: open
>Class: change-request
>Submitter-Id: net
>Arrival-Date: Sat Apr 06 07:08:00 PST 2002
>Closed-Date:
>Last-Modified:
>Originator: ac131313@redhat.com
>Release: 2002-04-05
>Organization:
>Environment:
>Description:
At present GDB tries to be smart over what needs to be rebuilt when - if the registers are changed, it is assumed that memory didn't change for instance. In general that isn't true. A single byte change has the potential of invaliding everything. There are edge conditions that are not correctly handled: modifying a register can corrupt the entire stack frame; writing to memory can modify a register / stack frame; some h/w has dual ported memory - also a register; ...
Instead of trying to be smart (and failing), be stupid. If the target is modified (memory write, register write, continue), invalidate everything.
Then concentrate on the most important case - recovering from a stop / continue / inferior function call.
>How-To-Repeat:
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted: