Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Peng Zhang <zhangpeng.00@bytedance.com>
To:     Liam.Howlett@oracle.com
Cc:     akpm@linux-foundation.org, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, maple-tree@lists.infradead.org,
        Peng Zhang <zhangpeng.00@bytedance.com>
Subject: [PATCH 07/10] maple_tree: Add mas_wr_new_end() to calculate new_end accurately
Date:   Mon, 15 May 2023 21:17:54 +0800
Message-Id: <20230515131757.60035-8-zhangpeng.00@bytedance.com>
In-Reply-To: <20230515131757.60035-1-zhangpeng.00@bytedance.com>
References: <20230515131757.60035-1-zhangpeng.00@bytedance.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

The previous new_end calculation is inaccurate, because it assumes that
two new pivots must be added (this is inaccurate), and sometimes it will
miss the fast path and enter the slow path. Add mas_wr_new_end() to
accurately calculate new_end to make the conditions for entering the
fast path more accurate.

Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
---
 lib/maple_tree.c | 33 ++++++++++++++++++++++-----------
 1 file changed, 22 insertions(+), 11 deletions(-)

diff --git a/lib/maple_tree.c b/lib/maple_tree.c
index ce695adc670ec..20082ef8c396c 100644
--- a/lib/maple_tree.c
+++ b/lib/maple_tree.c
@@ -4301,6 +4301,20 @@ static inline bool mas_wr_replace(struct ma_wr_state *wr_mas)
 	return false;
 }
 
+static inline unsigned char mas_wr_new_end(struct ma_wr_state *wr_mas)
+{
+	struct ma_state *mas = wr_mas->mas;
+	unsigned char new_end = wr_mas->node_end + 2;
+
+	new_end -= wr_mas->offset_end - mas->offset;
+	if (wr_mas->r_min == mas->index)
+		new_end--;
+	if (wr_mas->end_piv == mas->last)
+		new_end--;
+
+	return new_end;
+}
+
 static inline bool mas_wr_append(struct ma_wr_state *wr_mas)
 {
 	unsigned char end = wr_mas->node_end;
@@ -4356,25 +4370,22 @@ static void mas_wr_bnode(struct ma_wr_state *wr_mas)
 
 static inline void mas_wr_modify(struct ma_wr_state *wr_mas)
 {
-	unsigned char node_slots;
-	unsigned char node_size;
 	struct ma_state *mas = wr_mas->mas;
+	unsigned char new_end;
 
 	/* Attempt to direct replace */
 	if (mas_wr_replace(wr_mas))
 		return;
 
-	/* Attempt to append */
-	node_slots = mt_slots[wr_mas->type];
-	node_size = wr_mas->node_end - wr_mas->offset_end + mas->offset + 2;
-	if (mas->max == ULONG_MAX)
-		node_size++;
-
-	/* slot and node store will not fit, go to the slow path */
-	if (unlikely(node_size >= node_slots))
+	/*
+	 * new_end exceeds the size of the maple node and cannot enter the fast
+	 * path.
+	 */
+	new_end = mas_wr_new_end(wr_mas);
+	if (new_end >= mt_slots[wr_mas->type])
 		goto slow_path;
 
-	if (wr_mas->entry && (wr_mas->node_end < node_slots - 1) &&
+	if (wr_mas->entry && (wr_mas->node_end < mt_slots[wr_mas->type] - 1) &&
 	    (mas->offset == wr_mas->node_end) && mas_wr_append(wr_mas)) {
 		if (!wr_mas->content || !wr_mas->entry)
 			mas_update_gap(mas);
-- 
2.20.1