FreeChainXenon/binutils-gdb
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

* linux-low.c (linux_wait_for_event): Update messages. Do not

Browse source 
reinsert auto-delete breakpoints.
	* mem-break.c (struct breakpoint): Change return type of handler to
	int.
	(set_breakpoint_at): Update handler type.
	(reinsert_breakpoint_handler): Return 1 instead of calling
	delete_breakpoint.
	(reinsert_breakpoint_by_bp): Check for the original breakpoint before
	setting a new one.
	(check_breakpoints): Delete auto-delete breakpoints and return 2.
	* mem-break.h (set_breakpoint_at): Update handler type.
	* thread-db.c (thread_db_create_event, thread_db_create_event): Update.
	* win32-low.c (auto_delete_breakpoint): New.
	(get_child_debug_event): Use it.
This commit is contained in:
Daniel Jacobowitz 2007-12-18 21:58:01 +00:00
parent 0205c1752c
commit b65d95c576
6 changed files with 65 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

19
gdb/gdbserver/linux-low.c
View file

@ -623,6 +623,7 @@ linux_wait_for_event (struct thread_info *child)
CORE_ADDR stop_pc;
struct process_info *event_child;
int wstat;
int bp_status;
/* Check for a process with a pending status. */
/* It is possible that the user changed the pending task's registers since
@ -786,18 +787,20 @@ linux_wait_for_event (struct thread_info *child)
continue;
}
if (debug_threads)
fprintf (stderr, "Hit a (non-reinsert) breakpoint.\n");
bp_status = check_breakpoints (stop_pc);
if (check_breakpoints (stop_pc) != 0)
if (bp_status != 0)
{
if (debug_threads)
fprintf (stderr, "Hit a gdbserver breakpoint.\n");
/* We hit one of our own breakpoints. We mark it as a pending
breakpoint, so that check_removed_breakpoint () will do the PC
adjustment for us at the appropriate time. */
event_child->pending_is_breakpoint = 1;
event_child->pending_stop_pc = stop_pc;
/* Now we need to put the breakpoint back. We continue in the event
/* We may need to put the breakpoint back. We continue in the event
loop instead of simply replacing the breakpoint right away,
in order to not lose signals sent to the thread that hit the
breakpoint. Unfortunately this increases the window where another
@ -815,7 +818,10 @@ linux_wait_for_event (struct thread_info *child)
Otherwise, call the target function to figure out where we need
our temporary breakpoint, create it, and continue executing this
process. */
if (the_low_target.breakpoint_reinsert_addr == NULL)
if (bp_status == 2)
/* No need to reinsert. */
linux_resume_one_process (&event_child->head, 0, 0, NULL);
else if (the_low_target.breakpoint_reinsert_addr == NULL)
{
event_child->bp_reinsert = stop_pc;
uninsert_breakpoint (stop_pc);
@ -831,6 +837,9 @@ linux_wait_for_event (struct thread_info *child)
continue;
}
if (debug_threads)
fprintf (stderr, "Hit a non-gdbserver breakpoint.\n");
/* If we were single-stepping, we definitely want to report the
SIGTRAP. The single-step operation has completed, so also
clear the stepping flag; in general this does not matter,