linux-kernel - [PATCH 5/5] timers: Tag (hr)timer softirq as hotplug safe

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: <20230912104406.312185-6-frederic@kernel.org>
Date:   Tue, 12 Sep 2023 12:44:06 +0200
From:   Frederic Weisbecker <frederic@...nel.org>
To:     LKML <linux-kernel@...r.kernel.org>
Cc:     Frederic Weisbecker <frederic@...nel.org>,
        Joel Fernandes <joel@...lfernandes.org>,
        Thomas Gleixner <tglx@...utronix.de>, vineethrp@...il.com
Subject: [PATCH 5/5] timers: Tag (hr)timer softirq as hotplug safe

Specific stress involving frequent CPU-hotplug operations, such as
running rcutorture for example, may trigger the following message:

	"NOHZ tick-stop error: local softirq work is pending, handler #02!!!"

This happens in the CPU-down hotplug process, after
CPUHP_AP_SMPBOOT_THREADS whose teardown callback parks ksoftirqd, and
before the target CPU shuts down through CPUHP_AP_IDLE_DEAD. In this
fragile intermediate state, softirqs waiting for threaded handling may
be forever ignored and eventually reported by the idle task as in the
above example.

However some vectors are known to be safe as long as the corresponding
subsystems have teardown callbacks handling the migration of their
events. The above error message reports pending timers softirq although
this vector can be considered as hotplug safe because the
CPUHP_TIMERS_PREPARE teardown callback performs the necessary migration
of timers after the death of the CPU. Hrtimers also have a similar
hotplug handling.

Therefore this error message, as far as (hr-)timers are concerned, can
be considered spurious and the relevant softirq vectors can be marked as
hotplug safe.

Signed-off-by: Frederic Weisbecker <frederic@...nel.org>
---
 include/linux/interrupt.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index a92bce40b04b..4a1dc88ddbff 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -569,8 +569,12 @@ enum
  * 	2) rcu_report_dead() reports the final quiescent states.
  *
  * _ IRQ_POLL: irq_poll_cpu_dead() migrates the queue
+ *
+ * _ (HR)TIMER_SOFTIRQ: (hr)timers_dead_cpu() migrates the queue
  */
-#define SOFTIRQ_HOTPLUG_SAFE_MASK (BIT(RCU_SOFTIRQ) | BIT(IRQ_POLL_SOFTIRQ))
+#define SOFTIRQ_HOTPLUG_SAFE_MASK (BIT(TIMER_SOFTIRQ) | BIT(IRQ_POLL_SOFTIRQ) |\
+				   BIT(HRTIMER_SOFTIRQ) | BIT(RCU_SOFTIRQ))
+

 /* map softirq index to softirq name. update 'softirq_to_name' in
  * kernel/softirq.c when adding a new softirq.
-- 
2.34.1