[4.9,13/41] printk: fix deadlock when kernel panic
diff mbox series

Message ID 20210305120851.935800288@linuxfoundation.org
State New, archived
Headers show
  • Untitled series #487863
Related show

Commit Message

Greg Kroah-Hartman March 5, 2021, 12:22 p.m. UTC
From: Muchun Song <songmuchun@bytedance.com>

commit 8a8109f303e25a27f92c1d8edd67d7cbbc60a4eb upstream.

printk_safe_flush_on_panic() caused the following deadlock on our

CPU0:                                         CPU1:
panic                                         rcu_dump_cpu_stacks
  kdump_nmi_shootdown_cpus                      nmi_trigger_cpumask_backtrace
    register_nmi_handler(crash_nmi_callback)      printk_safe_flush
    // send NMI to other processors
                                                        // NMI interrupt, dead loop
        // deadlock

DEADLOCK: read_lock is taken on CPU1 and will never get released.

It happens when panic() stops a CPU by NMI while it has been in
the middle of printk_safe_flush().

Handle the lock the same way as logbuf_lock. The printk_safe buffers
are flushed only when both locks can be safely taken. It can avoid
the deadlock _in this particular case_ at expense of losing contents
of printk_safe buffers.

Note: It would actually be safe to re-init the locks when all CPUs were
      stopped by NMI. But it would require passing this information
      from arch-specific code. It is not worth the complexity.
      Especially because logbuf_lock and printk_safe buffers have been
      obsoleted by the lockless ring buffer.

Fixes: cf9b1106c81c ("printk/nmi: flush NMI messages on the system panic")
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Cc: <stable@vger.kernel.org>
Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Signed-off-by: Petr Mladek <pmladek@suse.com>
Link: https://lore.kernel.org/r/20210210034823.64867-1-songmuchun@bytedance.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

 kernel/printk/nmi.c |   16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff mbox series

--- a/kernel/printk/nmi.c
+++ b/kernel/printk/nmi.c
@@ -52,6 +52,8 @@  struct nmi_seq_buf {
 static DEFINE_PER_CPU(struct nmi_seq_buf, nmi_print_seq);
+static DEFINE_RAW_SPINLOCK(nmi_read_lock);
  * Safe printk() for NMI context. It uses a per-CPU buffer to
  * store the message. NMIs are not nested, so there is always only
@@ -134,8 +136,6 @@  static void printk_nmi_flush_seq_line(st
 static void __printk_nmi_flush(struct irq_work *work)
-	static raw_spinlock_t read_lock =
 	struct nmi_seq_buf *s = container_of(work, struct nmi_seq_buf, work);
 	unsigned long flags;
 	size_t len, size;
@@ -148,7 +148,7 @@  static void __printk_nmi_flush(struct ir
 	 * different CPUs. This is especially important when printing
 	 * a backtrace.
-	raw_spin_lock_irqsave(&read_lock, flags);
+	raw_spin_lock_irqsave(&nmi_read_lock, flags);
 	i = 0;
@@ -197,7 +197,7 @@  more:
 		goto more;
-	raw_spin_unlock_irqrestore(&read_lock, flags);
+	raw_spin_unlock_irqrestore(&nmi_read_lock, flags);
@@ -239,6 +239,14 @@  void printk_nmi_flush_on_panic(void)
+	if (in_nmi() && raw_spin_is_locked(&nmi_read_lock)) {
+		if (num_online_cpus() > 1)
+			return;
+		debug_locks_off();
+		raw_spin_lock_init(&nmi_read_lock);
+	}