rcu: Kill rnp->ofl_seq and use only rcu_state.ofl_lock for exclusion

If we allow architectures to bring APs online in parallel, then we end
up requiring rcu_cpu_starting() to be reentrant. But currently, the
manipulation of rnp->ofl_seq is not thread-safe.

However, rnp->ofl_seq is also fairly much pointless anyway since both
rcu_cpu_starting() and rcu_report_dead() hold rcu_state.ofl_lock for
fairly much the whole time that rnp->ofl_seq is set to an odd number
to indicate that an operation is in progress.

So drop rnp->ofl_seq completely, and use only rcu_state.ofl_lock.

This has a couple of minor complexities: lockdep will complain when we
take rcu_state.ofl_lock, and currently accepts the 'excuse' of having
an odd value in rnp->ofl_seq. So switch it to an arch_spinlock_t to
avoid that false positive complaint. Since we're killing rnp->ofl_seq
of course that 'excuse' has to be changed too, so make it check for
arch_spin_is_locked(rcu_state.ofl_lock).

There's no arch_spin_lock_irqsave() so we have to manually save and
restore local interrupts around the locking.

At Paul's request based on Neeraj's analysis, make rcu_gp_init not just
wait but *exclude* any CPU online/offline activity, which was fairly
much true already by virtue of it holding rcu_state.ofl_lock.

Signed-off-by: David Woodhouse <[e-mail]>
Signed-off-by: Paul E. McKenney <[e-mail]>

kernel/rcu/tree.c

···
 	.abbr = RCU_ABBR,
 	.exp_mutex = __MUTEX_INITIALIZER(rcu_state.exp_mutex),
 	.exp_wake_mutex = __MUTEX_INITIALIZER(rcu_state.exp_wake_mutex),
-	.ofl_lock = __RAW_SPIN_LOCK_UNLOCKED(rcu_state.ofl_lock),
+	.ofl_lock = __ARCH_SPIN_LOCK_UNLOCKED,
 };
 
 /* Dump rcu_node combining tree at boot to verify correct setup. */
···
 	preempt_disable_notrace();
 	rdp = this_cpu_ptr(&rcu_data);
 	rnp = rdp->mynode;
-	if (rdp->grpmask & rcu_rnp_online_cpus(rnp) || READ_ONCE(rnp->ofl_seq) & 0x1)
+	/*
+	 * Strictly, we care here about the case where the current CPU is
+	 * in rcu_cpu_starting() and thus has an excuse for rdp->grpmask
+	 * not being up to date. So arch_spin_is_locked() might have a
+	 * false positive if it's held by some *other* CPU, but that's
+	 * OK because that just means a false *negative* on the warning.
+	 */
+	if (rdp->grpmask & rcu_rnp_online_cpus(rnp) ||
+	    arch_spin_is_locked(&rcu_state.ofl_lock))
 		ret = true;
 	preempt_enable_notrace();
 	return ret;
···
  */
 static noinline_for_stack bool rcu_gp_init(void)
 {
-	unsigned long firstseq;
 	unsigned long flags;
 	unsigned long oldmask;
 	unsigned long mask;
···
 	 * of RCU's Requirements documentation.
 	 */
 	WRITE_ONCE(rcu_state.gp_state, RCU_GP_ONOFF);
+	/* Exclude CPU hotplug operations. */
 	rcu_for_each_leaf_node(rnp) {
-		// Wait for CPU-hotplug operations that might have
-		// started before this grace period did.
-		smp_mb(); // Pair with barriers used when updating ->ofl_seq to odd values.
-		firstseq = READ_ONCE(rnp->ofl_seq);
-		if (firstseq & 0x1)
-			while (firstseq == READ_ONCE(rnp->ofl_seq))
-				schedule_timeout_idle(1);  // Can't wake unless RCU is watching.
-		smp_mb(); // Pair with barriers used when updating ->ofl_seq to even values.
-		raw_spin_lock(&rcu_state.ofl_lock);
-		raw_spin_lock_irq_rcu_node(rnp);
+		local_irq_save(flags);
+		arch_spin_lock(&rcu_state.ofl_lock);
+		raw_spin_lock_rcu_node(rnp);
 		if (rnp->qsmaskinit == rnp->qsmaskinitnext &&
 		    !rnp->wait_blkd_tasks) {
 			/* Nothing to do on this leaf rcu_node structure. */
-			raw_spin_unlock_irq_rcu_node(rnp);
-			raw_spin_unlock(&rcu_state.ofl_lock);
+			raw_spin_unlock_rcu_node(rnp);
+			arch_spin_unlock(&rcu_state.ofl_lock);
+			local_irq_restore(flags);
 			continue;
 		}
 
···
 				rcu_cleanup_dead_rnp(rnp);
 		}
 
-		raw_spin_unlock_irq_rcu_node(rnp);
-		raw_spin_unlock(&rcu_state.ofl_lock);
+		raw_spin_unlock_rcu_node(rnp);
+		arch_spin_unlock(&rcu_state.ofl_lock);
+		local_irq_restore(flags);
 	}
 	rcu_gp_slow(gp_preinit_delay); /* Races with CPU hotplug. */
 
···
 
 	rnp = rdp->mynode;
 	mask = rdp->grpmask;
-	WRITE_ONCE(rnp->ofl_seq, rnp->ofl_seq + 1);
-	WARN_ON_ONCE(!(rnp->ofl_seq & 0x1));
+	local_irq_save(flags);
+	arch_spin_lock(&rcu_state.ofl_lock);
 	rcu_dynticks_eqs_online();
-	smp_mb(); // Pair with rcu_gp_cleanup()'s ->ofl_seq barrier().
-	raw_spin_lock_irqsave_rcu_node(rnp, flags);
+	raw_spin_lock_rcu_node(rnp);
 	WRITE_ONCE(rnp->qsmaskinitnext, rnp->qsmaskinitnext | mask);
 	newcpu = !(rnp->expmaskinitnext & mask);
 	rnp->expmaskinitnext |= mask;
···
 
 	/* An incoming CPU should never be blocking a grace period. */
 	if (WARN_ON_ONCE(rnp->qsmask & mask)) { /* RCU waiting on incoming CPU? */
+		/* rcu_report_qs_rnp() *really* wants some flags to restore */
+		unsigned long flags2;
+
+		local_irq_save(flags2);
 		rcu_disable_urgency_upon_qs(rdp);
 		/* Report QS -after- changing ->qsmaskinitnext! */
-		rcu_report_qs_rnp(mask, rnp, rnp->gp_seq, flags);
+		rcu_report_qs_rnp(mask, rnp, rnp->gp_seq, flags2);
 	} else {
-		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
+		raw_spin_unlock_rcu_node(rnp);
 	}
-	smp_mb(); // Pair with rcu_gp_cleanup()'s ->ofl_seq barrier().
-	WRITE_ONCE(rnp->ofl_seq, rnp->ofl_seq + 1);
-	WARN_ON_ONCE(rnp->ofl_seq & 0x1);
+	arch_spin_unlock(&rcu_state.ofl_lock);
+	local_irq_restore(flags);
 	smp_mb(); /* Ensure RCU read-side usage follows above initialization. */
 }
 
···
  */
 void rcu_report_dead(unsigned int cpu)
 {
-	unsigned long flags;
+	unsigned long flags, seq_flags;
 	unsigned long mask;
 	struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu);
 	struct rcu_node *rnp = rdp->mynode;  /* Outgoing CPU's rdp & rnp. */
···
 
 	/* Remove outgoing CPU from mask in the leaf rcu_node structure. */
 	mask = rdp->grpmask;
-	WRITE_ONCE(rnp->ofl_seq, rnp->ofl_seq + 1);
-	WARN_ON_ONCE(!(rnp->ofl_seq & 0x1));
-	smp_mb(); // Pair with rcu_gp_cleanup()'s ->ofl_seq barrier().
-	raw_spin_lock(&rcu_state.ofl_lock);
+	local_irq_save(seq_flags);
+	arch_spin_lock(&rcu_state.ofl_lock);
 	raw_spin_lock_irqsave_rcu_node(rnp, flags); /* Enforce GP memory-order guarantee. */
 	rdp->rcu_ofl_gp_seq = READ_ONCE(rcu_state.gp_seq);
 	rdp->rcu_ofl_gp_flags = READ_ONCE(rcu_state.gp_flags);
···
 	}
 	WRITE_ONCE(rnp->qsmaskinitnext, rnp->qsmaskinitnext & ~mask);
 	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
-	raw_spin_unlock(&rcu_state.ofl_lock);
-	smp_mb(); // Pair with rcu_gp_cleanup()'s ->ofl_seq barrier().
-	WRITE_ONCE(rnp->ofl_seq, rnp->ofl_seq + 1);
-	WARN_ON_ONCE(rnp->ofl_seq & 0x1);
+	arch_spin_unlock(&rcu_state.ofl_lock);
+	local_irq_restore(seq_flags);
 
 	rdp->cpu_started = false;
 }

kernel/rcu/tree.h

···
 				/*  Initialized from ->qsmaskinitnext at the */
 				/*  beginning of each grace period. */
 	unsigned long qsmaskinitnext;
-	unsigned long ofl_seq;	/* CPU-hotplug operation sequence count. */
-				/* Online CPUs for next grace period. */
 	unsigned long expmask;	/* CPUs or groups that need to check in */
 				/*  to allow the current expedited GP */
 				/*  to complete. */
···
 	const char *name;			/* Name of structure. */
 	char abbr;				/* Abbreviated name. */
 
-	raw_spinlock_t ofl_lock ____cacheline_internodealigned_in_smp;
+	arch_spinlock_t ofl_lock ____cacheline_internodealigned_in_smp;
 						/* Synchronize offline with */
 						/*  GP pre-initialization. */
 };