Re: [patch V2 08/12] rseq: Implement time slice extension enforcement timer

[Sebastian Andrzej Siewior]

On 2025-10-28 14:52:09 [+0530], K Prateek Nayak wrote:
> On 10/28/2025 2:30 PM, Sebastian Andrzej Siewior wrote:
> >>> Without an interrupt on the target CPU, there is nothing stopping the
> >>> task from overstepping its fair share.
> >>
> >> When the task moves CPU, the rseq_exit_user_update() would clear all
> >> of the slice extension state before running the task again. The task
> >> will start off again with "rseq->slice_ctrl.request" and
> >> "rseq->slice_ctrl.granted" both at 0 signifying the task was
> >> rescheduled.
> > 
> > I wasn't aware this is done once the task is in userland and then
> > relocated to another CPU.
> 
> The exact path based on my understanding is:
> 
>   /* Task migrates to another CPU; Has to resume from kernel. */
>   __schedule()
>     context_switch()
>       rseq_sched_switch_event()
>         t->rseq.event.sched_switch = true;
>         set_tsk_thread_flag(t, TIF_RSEQ);
> 
>     ...
>     exit_to_user_mode_loop()
>       rseq_exit_to_user_mode_restart()
>         __rseq_exit_to_user_mode_restart()
>           /* Sees t->rseq.event.sched_switch to be true. */
>           rseq_exit_user_update()
>             if (rseq_slice_extension_enabled())
>               unsafe_put_user(0U, &rseq->slice_ctrl.all, efault); /* Unconditionally clears all of "rseq_ctrl" */

You are right. The migration thread preempts it on the old CPU and then
it gets scheduled in on the new CPU.

Sebastian