Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Sat, 29 Feb 2020 22:06:13 +0000
From:   Wei Yang <richard.weiyang@gmail.com>
To:     Wei Yang <richard.weiyang@gmail.com>
Cc:     akpm@linux-foundation.org, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, Hugh Dickins <hughd@google.com>
Subject: Re: [PATCH] mm/swapfile.c: simplify the scan loop in
 scan_swap_map_slots()
Message-ID: <20200229220613.oyryopfshe6juvro@master>
Reply-To: Wei Yang <richard.weiyang@gmail.com>
References: <20200229131537.3475-1-richard.weiyang@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200229131537.3475-1-richard.weiyang@gmail.com>
User-Agent: NeoMutt/20170113 (1.7.2)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sat, Feb 29, 2020 at 01:15:37PM +0000, Wei Yang wrote:
>After commit c60aa176c6de8 ("swapfile: swap allocation cycle if
>nonrot"), swap allocation is cyclic. Current approach is done with two
>separate loop on the upper and lower half. This looks a little
>redundant.
>
>>From another point of view, the loop iterates [lowest_bit, highest_bit]
>range starting with (offset + 1) but except scan_base. So we can
>simplify the loop with condition (next_offset() != scan_base) by
>introducing next_offset() which makes sure offset fit in that range
>with correct order.
>
>Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
>CC: Hugh Dickins <hughd@google.com>
>---
> mm/swapfile.c | 26 +++++++++-----------------
> 1 file changed, 9 insertions(+), 17 deletions(-)
>
>diff --git a/mm/swapfile.c b/mm/swapfile.c
>index 95024f9b691a..42c5c2010bfc 100644
>--- a/mm/swapfile.c
>+++ b/mm/swapfile.c
>@@ -729,6 +729,14 @@ static void swap_range_free(struct swap_info_struct *si, unsigned long offset,
> 	}
> }
> 
>+static unsigned long next_offset(struct swap_info_struct *si,
>+				 unsigned long *offset)
>+{
>+	if (++(*offset) > si->highest_bit)
>+		*offset = si->lowest_bit;

Hmm... I found one potential problem here. If someone has eaten the lower
part, (si->lowest_bit > scan_base), we would fall into infinite loop.

Will wait for some comment before sending v2.

>+	return *offset;
>+}
>+
> static int scan_swap_map_slots(struct swap_info_struct *si,
> 			       unsigned char usage, int nr,
> 			       swp_entry_t slots[])
>@@ -883,7 +891,7 @@ static int scan_swap_map_slots(struct swap_info_struct *si,
> 
> scan:
> 	spin_unlock(&si->lock);
>-	while (++offset <= si->highest_bit) {
>+	while (next_offset(si, &offset) != scan_base) {
> 		if (!si->swap_map[offset]) {
> 			spin_lock(&si->lock);
> 			goto checks;
>@@ -897,22 +905,6 @@ static int scan_swap_map_slots(struct swap_info_struct *si,
> 			latency_ration = LATENCY_LIMIT;
> 		}
> 	}
>-	offset = si->lowest_bit;
>-	while (offset < scan_base) {
>-		if (!si->swap_map[offset]) {
>-			spin_lock(&si->lock);
>-			goto checks;
>-		}
>-		if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
>-			spin_lock(&si->lock);
>-			goto checks;
>-		}
>-		if (unlikely(--latency_ration < 0)) {
>-			cond_resched();
>-			latency_ration = LATENCY_LIMIT;
>-		}
>-		offset++;
>-	}
> 	spin_lock(&si->lock);
> 
> no_page:
>-- 
>2.23.0

-- 
Wei Yang
Help you, Help me