linux-kernel - Re: [PATCH 2/5] rcu/nocb: Move rcu_nocb_is_setup to rcu

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20220217113012.GB742892@lothringen>
Date:   Thu, 17 Feb 2022 12:30:12 +0100
From:   Frederic Weisbecker <frederic@...nel.org>
To:     "Paul E. McKenney" <paulmck@...nel.org>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        Neeraj Upadhyay <quic_neeraju@...cinc.com>,
        Boqun Feng <boqun.feng@...il.com>,
        Uladzislau Rezki <uladzislau.rezki@...y.com>,
        Joel Fernandes <joel@...lfernandes.org>
Subject: Re: [PATCH 2/5] rcu/nocb: Move rcu_nocb_is_setup to rcu_state

On Wed, Feb 16, 2022 at 09:28:04PM -0800, Paul E. McKenney wrote:
> On Wed, Feb 16, 2022 at 05:53:32PM -0800, Paul E. McKenney wrote:
> > On Wed, Feb 16, 2022 at 01:30:35PM -0800, Paul E. McKenney wrote:
> > > On Wed, Feb 16, 2022 at 04:42:05PM +0100, Frederic Weisbecker wrote:
> > > > In order to avoid scattering the global RCU states, move the RCU nocb
> > > > initialization witness within rcu_state.
> > > > 
> > > > Reported-by: Paul E. McKenney <paulmck@...nel.org>
> > > > Cc: Neeraj Upadhyay <quic_neeraju@...cinc.com>
> > > > Cc: Uladzislau Rezki <uladzislau.rezki@...y.com>
> > > > Cc: Joel Fernandes <joel@...lfernandes.org>
> > > > Cc: Boqun Feng <boqun.feng@...il.com>
> > > > Signed-off-by: Frederic Weisbecker <frederic@...nel.org>
> > > 
> > > Build testing for CONFIG_NO_HZ_FULL=n kernels suggested the following
> > > patch be merged into this one.  Or does this variable need to be used
> > > somewhere?  Either way, please let me know!
> > 
> > TASKS03, TREE04, and TREE07 don't like this much, according to git
> > bisect.  They are the ones with nohz_full CPUs, in case that is crucial.
> > Trying again after reverting this commit locally.
> > 
> > If that works, I would be tempted to try the modification shown below.
> 
> And that does pass significant rcutorture testing, but over to you.  ;-)

Ah no we still need that variable, but I screwed up and wrongly converted
need_rcu_nocb_mask to rcu_state.nocb_is_setup somewhere.

Here is the v2 of the culprit patch (sorry):

---
From: Frederic Weisbecker <frederic@...nel.org>
Date: Mon, 14 Feb 2022 14:23:39 +0100
Subject: [PATCH] rcu/nocb: Move rcu_nocb_is_setup to rcu_state

In order to avoid scattering the global RCU states, move the RCU nocb
initialization witness within rcu_state.

Reported-by: Paul E. McKenney <paulmck@...nel.org>
Cc: Neeraj Upadhyay <quic_neeraju@...cinc.com>
Cc: Uladzislau Rezki <uladzislau.rezki@...y.com>
Cc: Joel Fernandes <joel@...lfernandes.org>
Cc: Boqun Feng <boqun.feng@...il.com>
Signed-off-by: Frederic Weisbecker <frederic@...nel.org>
---
 kernel/rcu/tree.h      |  1 +
 kernel/rcu/tree_nocb.h | 13 +++++--------
 2 files changed, 6 insertions(+), 8 deletions(-)

diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h
index 19fc9acce3ce..735fc410d76a 100644
--- a/kernel/rcu/tree.h
+++ b/kernel/rcu/tree.h
@@ -368,6 +368,7 @@ struct rcu_state {
 	arch_spinlock_t ofl_lock ____cacheline_internodealigned_in_smp;
 						/* Synchronize offline with */
 						/*  GP pre-initialization. */
+	int nocb_is_setup;			/* nocb is setup from boot */
 };
 
 /* Values for rcu_state structure's gp_flags field. */
diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
index 02e1d05a11fc..3c00240833d6 100644
--- a/kernel/rcu/tree_nocb.h
+++ b/kernel/rcu/tree_nocb.h
@@ -60,9 +60,6 @@ static inline bool rcu_current_is_nocb_kthread(struct rcu_data *rdp)
  * Parse the boot-time rcu_nocb_mask CPU list from the kernel parameters.
  * If the list is invalid, a warning is emitted and all CPUs are offloaded.
  */
-
-static bool rcu_nocb_is_setup;
-
 static int __init rcu_nocb_setup(char *str)
 {
 	alloc_bootmem_cpumask_var(&rcu_nocb_mask);
@@ -72,7 +69,7 @@ static int __init rcu_nocb_setup(char *str)
 			cpumask_setall(rcu_nocb_mask);
 		}
 	}
-	rcu_nocb_is_setup = true;
+	rcu_state.nocb_is_setup = true;
 	return 1;
 }
 __setup("rcu_nocbs", rcu_nocb_setup);
@@ -1172,10 +1169,10 @@ void __init rcu_init_nohz(void)
 				return;
 			}
 		}
-		rcu_nocb_is_setup = true;
+		rcu_state.nocb_is_setup = true;
 	}
 
-	if (!rcu_nocb_is_setup)
+	if (!rcu_state.nocb_is_setup)
 		return;
 
 #if defined(CONFIG_NO_HZ_FULL)
@@ -1233,7 +1230,7 @@ static void rcu_spawn_cpu_nocb_kthread(int cpu)
 	struct task_struct *t;
 	struct sched_param sp;
 
-	if (!rcu_scheduler_fully_active || !rcu_nocb_is_setup)
+	if (!rcu_scheduler_fully_active || !rcu_state.nocb_is_setup)
 		return;
 
 	/* If there already is an rcuo kthread, then nothing to do. */
@@ -1279,7 +1276,7 @@ static void __init rcu_spawn_nocb_kthreads(void)
 {
 	int cpu;
 
-	if (rcu_nocb_is_setup) {
+	if (rcu_state.nocb_is_setup) {
 		for_each_online_cpu(cpu)
 			rcu_spawn_cpu_nocb_kthread(cpu);
 	}
-- 
2.25.1