Received-SPF: pass (google.com: domain of linux-kernel+bounces-194857-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Subject: Re: [RFC PATCH] sbitmap: fix io hung due to race on
 sbitmap_word::cleared
To: Yang Yang <yang.yang@vivo.com>, Jens Axboe <axboe@kernel.dk>,
 Andrew Morton <akpm@linux-foundation.org>,
 Pavel Begunkov <asml.silence@gmail.com>, linux-kernel@vger.kernel.org,
 linux-block@vger.kernel.org, "yukuai (C)" <yukuai3@huawei.com>
References: <20240527042654.2404-1-yang.yang@vivo.com>
From: Yu Kuai <yukuai1@huaweicloud.com>
Message-ID: <bd176002-90e8-030d-f8d8-9388d3b40241@huaweicloud.com>
Date: Thu, 30 May 2024 09:49:12 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.8.0
Precedence: bulk
MIME-Version: 1.0
In-Reply-To: <20240527042654.2404-1-yang.yang@vivo.com>
Content-Type: text/plain; charset=gbk; format=flowed
Content-Transfer-Encoding: 8bit

Hi,

?? 2024/05/27 12:26, Yang Yang д??:
> Configuration for sbq:
>    depth=64, wake_batch=6, shift=6, map_nr=1
> 
> 1. There are 64 requests in progress:
>    map->word = 0xFFFFFFFFFFFFFFFF
> 2. After all the 64 requests complete, and no more requests come:
>    map->word = 0xFFFFFFFFFFFFFFFF, map->cleared = 0xFFFFFFFFFFFFFFFF
> 3. Now two tasks try to allocate requests:
>    T1:                                       T2:
>    __blk_mq_get_tag                          .
>    __sbitmap_queue_get                       .
>    sbitmap_get                               .
>    sbitmap_find_bit                          .
>    sbitmap_find_bit_in_word                  .
>    __sbitmap_get_word  -> nr=-1              __blk_mq_get_tag
>    sbitmap_deferred_clear                    __sbitmap_queue_get
>    /* map->cleared=0xFFFFFFFFFFFFFFFF */     sbitmap_find_bit
>      if (!READ_ONCE(map->cleared))           sbitmap_find_bit_in_word
>        return false;                         __sbitmap_get_word -> nr=-1
>      mask = xchg(&map->cleared, 0)           sbitmap_deferred_clear
>      atomic_long_andnot()                    /* map->cleared=0 */
>                                                if (!(map->cleared))
>                                                  return false;
>                                       /*
>                                        * map->cleared is cleared by T1
>                                        * T2 fail to acquire the tag
>                                        */

This looks correct.

> 
> 4. T2 is the sole tag waiter. When T1 puts the tag, T2 cannot be woken
> up due to the wake_batch being set at 6. If no more requests come, T1
> will wait here indefinitely.
> 
> Fix this issue by adding a new flag swap_inprogress to indicate whether
> the swap is ongoing.
> 
> Fixes: 661d4f55a794 ("sbitmap: remove swap_lock")
> Signed-off-by: Yang Yang <yang.yang@vivo.com>
> ---
>   include/linux/sbitmap.h |  5 +++++
>   lib/sbitmap.c           | 22 ++++++++++++++++++++--
>   2 files changed, 25 insertions(+), 2 deletions(-)
> 
> diff --git a/include/linux/sbitmap.h b/include/linux/sbitmap.h
> index d662cf136021..b88a9e4997ab 100644
> --- a/include/linux/sbitmap.h
> +++ b/include/linux/sbitmap.h
> @@ -36,6 +36,11 @@ struct sbitmap_word {
>   	 * @cleared: word holding cleared bits
>   	 */
>   	unsigned long cleared ____cacheline_aligned_in_smp;
> +
> +	/**
> +	 * @swap_inprogress: set to 1 when swapping word <-> cleared
> +	 */
> +	atomic_t swap_inprogress;
>   } ____cacheline_aligned_in_smp;
>   
>   /**
> diff --git a/lib/sbitmap.c b/lib/sbitmap.c
> index 1e453f825c05..d4bb258fe8b0 100644
> --- a/lib/sbitmap.c
> +++ b/lib/sbitmap.c
> @@ -62,10 +62,19 @@ static inline void update_alloc_hint_after_get(struct sbitmap *sb,
>    */
>   static inline bool sbitmap_deferred_clear(struct sbitmap_word *map)
>   {
> -	unsigned long mask;
> +	unsigned long mask, flags;
> +	int zero = 0;
>   
> -	if (!READ_ONCE(map->cleared))
> +	if (!READ_ONCE(map->cleared)) {
> +		if (atomic_read(&map->swap_inprogress))
> +			goto out_wait;
>   		return false;
> +	}
> +
> +	if (!atomic_try_cmpxchg(&map->swap_inprogress, &zero, 1))
> +		goto out_wait;
> +
> +	local_irq_save(flags);
>   
>   	/*
>   	 * First get a stable cleared mask, setting the old mask to 0.
> @@ -77,6 +86,15 @@ static inline bool sbitmap_deferred_clear(struct sbitmap_word *map)
>   	 */
>   	atomic_long_andnot(mask, (atomic_long_t *)&map->word);
>   	BUILD_BUG_ON(sizeof(atomic_long_t) != sizeof(map->word));
> +
> +	atomic_set(&map->swap_inprogress, 0);
> +	smp_mb__after_atomic();
> +	local_irq_restore(flags);
> +	return true;
> +
> +out_wait:
> +	while (atomic_read(&map->swap_inprogress))
> +		;

However, I really don't like this change.

How about following patch? I think it can fix the prblem as well.

diff --git a/lib/sbitmap.c b/lib/sbitmap.c
index 1e453f825c05..d1b9bd61d296 100644
--- a/lib/sbitmap.c
+++ b/lib/sbitmap.c
@@ -175,12 +175,14 @@ static int sbitmap_find_bit_in_word(struct 
sbitmap_word *map,
         int nr;

         do {
+               bool clear = READ_ONCE(map->cleared) != 0;
+
                 nr = __sbitmap_get_word(&map->word, depth,
                                         alloc_hint, wrap);
-               if (nr != -1)
-                       break;
-               if (!sbitmap_deferred_clear(map))
+               if (nr != -1 || !clear)
                         break;
+
+               sbitmap_deferred_clear(map);
         } while (1);

         return nr;

Thanks,
Kuai
>   	return true;
>   }
>   
>