Fixup last entry.
authorPedro Alves <palves@redhat.com>
Fri, 9 Sep 2011 19:29:51 +0000 (19:29 +0000)
committerPedro Alves <palves@redhat.com>
Fri, 9 Sep 2011 19:29:51 +0000 (19:29 +0000)
commitddf17726b28e169d5ec0252a777a9c528c6a772f
treed79dd345d7ed240eed05c5127075bacb3549543d
parent25289eb274bff1f2052ecea034171ff2b6970c94
Fixup last entry.

2011-09-09  Pedro Alves  <pedro@codesourcery.com>

* linux-nat.h (enum resume_kind): New.
(struct lwp_info) <last_resume_kind>: New field.
* linux-nat.c (linux_child_follow_fork): Set last_resume_kind to
resume_stop on the new lwp.
(add_lwp): Set last_resume_kind as resume_continue by default.
(lin_lwp_attach_lwp): Set last_resume_kind as resume_stop.
(resume_lwp): New, factored out from resume_callback.  Also check
for pending status in lp->waitstatus.
(resume_callback): Reimplement.
(resume_clear_callback): Set last_resume_kind as resume_stop.
(resume_set_callback): Set last_resume_kind as resume_continue.
(linux_nat_resume, linux_handle_extended_wait): Set
last_resume_kind.
(running_callback): Also check lp->waitstatus for pending events.
(select_singlestep_lwp_callback): Check that lp->last_resume_kind
is resume_step.
(stop_and_resume_callback): Don't re-resume if the core wanted the
lwp stopped.  Use resume_lwp instead of resume_callback.  Avoid
using an invalidated pointer.
(linux_nat_filter_event): Don't discard SIGSTOPs as delayed
SIGSTOPs if the core wanted the LWP to stop.
(linux_nat_wait_1) Don't consume a pending SIGSTOP if the core
wanted the lwp to stop.  If the core wanted the lwp to stop, and
the lwp stopped with a SIGSTOP, report a TARGET_SIGNAL_0 instead
of TARGET_SIGNAL_STOP.
(linux_nat_stop_lwp): Don't synchronously wait for the lwp to stop
here.  Instead, signal the lwp, and set the last_resume_kind to
resume_stop.
gdb/ChangeLog