Gregory Nutt <spudan...@gmail.com> 于2023年4月23日周日 04:24写道：
> I am not sure what this mean. But if you mean calling > enter_critical_secion() in sched_lock(), then it is a very bad idea You > should not do that. enter_critical_section() disables interrupts on all > CPUs and this must never happen when disabling preemption. > > In the SMP case, that involves and coordiation with all other CPUs with > inter-CPU messages in order to assure that interrupts are disabled on > all CPUs. That is terrible thing and must not happen. You happen to understand the opposite. Well, I mean calling sched_lock() in enter_critical_secion() This way, many places that simultaneously call sched_lock() and enter_critical_secion() can be replaced with enter_critical_secion(). This simplifies and reduces the use of sched_lock(), laying the foundation for subsequent optimization. And as you mentioned earlier, the cost of sched_lock() is minimal. BR！
