Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Thu, 14 Oct 2021 13:06:03 +0100
From:   Christoph Hellwig <hch@infradead.org>
To:     Zqiang <qiang.zhang1211@gmail.com>
Cc:     akpm@linux-foundation.org, sunhao.th@gmail.com,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH] mm: backing-dev: use kfree_rcu() instead of
 synchronize_rcu_expedited()
Message-ID: <YWgdK9OhXa/9N82Y@infradead.org>
References: <20211014082433.30733-1-qiang.zhang1211@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20211014082433.30733-1-qiang.zhang1211@gmail.com>
Precedence: bulk

On Thu, Oct 14, 2021 at 04:24:33PM +0800, Zqiang wrote:
> <IRQ>
>  __init_work+0x2d/0x50
>  synchronize_rcu_expedited+0x3af/0x650
>  bdi_remove_from_list [inline]
>  bdi_unregister+0x17f/0x5c0
>  release_bdi+0xa1/0xc0
>  kref_put [inline]
>  bdi_put+0x72/0xa0
>  bdev_free_inode+0x11e/0x220
>  i_callback+0x3f/0x70
>  rcu_do_batch [inline]
>  rcu_core+0x76d/0x16c0
>  __do_softirq+0x1d7/0x93b
>  invoke_softirq [inline]
>  __irq_exit_rcu [inline]
>  irq_exit_rcu+0xf2/0x130
>  sysvec_apic_timer_interrupt+0x93/0xc0
> 
> The bdi_remove_from_list() is called in RCU softirq, however the
> synchronize_rcu_expedited() will produce sleep action, use kfree_rcu()
> instead of it.

What workload is this running?  If we hit the RCU unregister path
from inode freeing we have a lifetime problem somewhere that we need
to fix first.