[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20230623171232.892937-3-bigeasy@linutronix.de>
Date: Fri, 23 Jun 2023 19:12:32 +0200
From: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
To: linux-mm@...ck.org, linux-kernel@...r.kernel.org
Cc: "Luis Claudio R. Goncalves" <lgoncalv@...hat.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Boqun Feng <boqun.feng@...il.com>,
Ingo Molnar <mingo@...hat.com>,
John Ogness <john.ogness@...utronix.de>,
Mel Gorman <mgorman@...hsingularity.net>,
Michal Hocko <mhocko@...e.com>,
Peter Zijlstra <peterz@...radead.org>,
Petr Mladek <pmladek@...e.com>,
Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>,
Thomas Gleixner <tglx@...utronix.de>,
Waiman Long <longman@...hat.co>, Will Deacon <will@...nel.org>,
Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Subject: [PATCH v2 2/2] mm/page_alloc: Use write_seqlock_irqsave() instead write_seqlock() + local_irq_save().
__build_all_zonelists() acquires zonelist_update_seq by first disabling
interrupts via local_irq_save() and then acquiring the seqlock with
write_seqlock(). This is troublesome and leads to problems on
PREEMPT_RT. The problem is that the inner spinlock_t becomes a sleeping
lock on PREEMPT_RT and must not be acquired with disabled interrupts.
The API provides write_seqlock_irqsave() which does the right thing in
one step.
printk_deferred_enter() has to be invoked in non-migrate-able context to
ensure that deferred printing is enabled and disabled on the same CPU.
This is the case after zonelist_update_seq has been acquired.
There was discussion on the first submission that the order should be:
local_irq_disable();
printk_deferred_enter();
write_seqlock();
to avoid pitfalls like having an unaccounted printk() coming from
write_seqlock_irqsave() before printk_deferred_enter() is invoked. The
only origin of such a printk() can be a lockdep splat because the
lockdep annotation happens after the sequence count is incremented.
This is exceptional and subject to change.
It was also pointed that PREEMPT_RT can be affected by the printk
problem since its write_seqlock_irqsave() does not really disable
interrupts. This isn't the case because PREEMPT_RT's printk
implementation differs from the mainline implementation in two important
aspects:
- Printing happens in a dedicated threads and not at during the
invocation of printk().
- In emergency cases where synchronous printing is used, a different
driver is used which does not use tty_port::lock.
Acquire zonelist_update_seq with write_seqlock_irqsave() and then defer
printk output.
Fixes: 1007843a91909 ("mm/page_alloc: fix potential deadlock on zonelist_update_seq seqlock")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
---
mm/page_alloc.c | 11 ++++-------
1 file changed, 4 insertions(+), 7 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 47421bedc12b7..99b7e7d09c5c0 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5808,11 +5808,10 @@ static void __build_all_zonelists(void *data)
unsigned long flags;
/*
- * Explicitly disable this CPU's interrupts before taking seqlock
- * to prevent any IRQ handler from calling into the page allocator
- * (e.g. GFP_ATOMIC) that could hit zonelist_iter_begin and livelock.
+ * The zonelist_update_seq must be acquired with irqsave because the
+ * reader can be invoked from IRQ with GFP_ATOMIC.
*/
- local_irq_save(flags);
+ write_seqlock_irqsave(&zonelist_update_seq, flags);
/*
* Explicitly disable this CPU's synchronous printk() before taking
* seqlock to prevent any printk() from trying to hold port->lock, for
@@ -5820,7 +5819,6 @@ static void __build_all_zonelists(void *data)
* calling kmalloc(GFP_ATOMIC | __GFP_NOWARN) with port->lock held.
*/
printk_deferred_enter();
- write_seqlock(&zonelist_update_seq);
#ifdef CONFIG_NUMA
memset(node_load, 0, sizeof(node_load));
@@ -5857,9 +5855,8 @@ static void __build_all_zonelists(void *data)
#endif
}
- write_sequnlock(&zonelist_update_seq);
printk_deferred_exit();
- local_irq_restore(flags);
+ write_sequnlock_irqrestore(&zonelist_update_seq, flags);
}
static noinline void __init
--
2.40.1
Powered by blists - more mailing lists