Re: [PATCH] rcu: Let rcu_dump_task() be used without preemption disabled

[Paul E. McKenney]

On Fri, Jun 28, 2024 at 01:18:26PM +0900, takakura@xxxxxxxxxxxxx wrote:
> From: Ryo Takakura <takakura@xxxxxxxxxxxxx>
> 
> The commit 2d7f00b2f0130 ("rcu: Suppress smp_processor_id() complaint
> in synchronize_rcu_expedited_wait()") disabled preemption around
> dump_cpu_task() to suppress warning on its usage within preemtible context.
> 
> Calling dump_cpu_task() doesn't required to be in non-preemptible context
> except for suppressing the smp_processor_id() warning.
> As the smp_processor_id() is evaluated along with in_hardirq()
> to check if it's in interrupt context, this patch removes the need
> for its preemtion disablement by reordering the condition so that
> smp_processor_id() only gets evaluated when it's in interrupt context.
> 
> Signed-off-by: Ryo Takakura <takakura@xxxxxxxxxxxxx>

Hearing no objections, I pulled this in for further review and testing.

I had to hand-apply this due to a recent conflicting change in the
-rcu tree, so could you please check the version below in case I messed
something up?

							Thanx, Paul

------------------------------------------------------------------------

commit ad6647a70f239aa9f2741b2f5a828a4483122a26
Author: Ryo Takakura <takakura@xxxxxxxxxxxxx>
Date:   Fri Jun 28 13:18:26 2024 +0900

    rcu: Let rcu_dump_task() be used without preemption disabled
    
    The commit 2d7f00b2f0130 ("rcu: Suppress smp_processor_id() complaint
    in synchronize_rcu_expedited_wait()") disabled preemption around
    dump_cpu_task() to suppress warning on its usage within preemtible context.
    
    Calling dump_cpu_task() doesn't required to be in non-preemptible context
    except for suppressing the smp_processor_id() warning.
    As the smp_processor_id() is evaluated along with in_hardirq()
    to check if it's in interrupt context, this patch removes the need
    for its preemtion disablement by reordering the condition so that
    smp_processor_id() only gets evaluated when it's in interrupt context.
    
    Signed-off-by: Ryo Takakura <takakura@xxxxxxxxxxxxx>
    Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxx>

diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h
index d4be644afb50..c5d9a7eb0803 100644
--- a/kernel/rcu/tree_exp.h
+++ b/kernel/rcu/tree_exp.h
@@ -597,9 +597,7 @@ static void synchronize_rcu_expedited_stall(unsigned long jiffies_start, unsigne
 			mask = leaf_node_cpu_bit(rnp, cpu);
 			if (!(READ_ONCE(rnp->expmask) & mask))
 				continue;
-			preempt_disable(); // For smp_processor_id() in dump_cpu_task().
 			dump_cpu_task(cpu);
-			preempt_enable();
 		}
 		rcu_exp_print_detail_task_stall_rnp(rnp);
 	}
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 05afa2932b5e..bdb0e0328f6a 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -11485,7 +11485,7 @@ struct cgroup_subsys cpu_cgrp_subsys = {
 
 void dump_cpu_task(int cpu)
 {
-	if (cpu == smp_processor_id() && in_hardirq()) {
+	if (in_hardirq() && cpu == smp_processor_id()) {
 		struct pt_regs *regs;
 
 		regs = get_irq_regs();