diff --git a/gdb/gdbserver/ChangeLog b/gdb/gdbserver/ChangeLog index 6c23574494..11e1c4fca5 100644 --- a/gdb/gdbserver/ChangeLog +++ b/gdb/gdbserver/ChangeLog @@ -1,3 +1,7 @@ +2016-03-03 Yao Qi + + * linux-low.c: Update comments to start_step_over. + 2016-03-03 Yao Qi PR server/19736 diff --git a/gdb/gdbserver/linux-low.c b/gdb/gdbserver/linux-low.c index 6b07c90047..c008847a42 100644 --- a/gdb/gdbserver/linux-low.c +++ b/gdb/gdbserver/linux-low.c @@ -4625,18 +4625,10 @@ need_step_over_p (struct inferior_list_entry *entry, void *dummy) of the way. If we let other threads run while we do that, they may pass by the breakpoint location and miss hitting it. To avoid that, a step-over momentarily stops all threads while LWP is - single-stepped while the breakpoint is temporarily uninserted from - the inferior. When the single-step finishes, we reinsert the - breakpoint, and let all threads that are supposed to be running, - run again. - - On targets that don't support hardware single-step, we don't - currently support full software single-stepping. Instead, we only - support stepping over the thread event breakpoint, by asking the - low target where to place a reinsert breakpoint. Since this - routine assumes the breakpoint being stepped over is a thread event - breakpoint, it usually assumes the return address of the current - function is a good enough place to set the reinsert breakpoint. */ + single-stepped by either hardware or software while the breakpoint + is temporarily uninserted from the inferior. When the single-step + finishes, we reinsert the breakpoint, and let all threads that are + supposed to be running, run again. */ static int start_step_over (struct lwp_info *lwp)