Received-SPF: pass (google.com: domain of linux-ext4+bounces-1944-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
From: Zhang Yi <yi.zhang@huaweicloud.com>
To: linux-ext4@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org,
	linux-mm@kvack.org,
	linux-kernel@vger.kernel.org,
	tytso@mit.edu,
	adilger.kernel@dilger.ca,
	jack@suse.cz,
	ritesh.list@gmail.com,
	hch@infradead.org,
	djwong@kernel.org,
	willy@infradead.org,
	zokeefe@google.com,
	yi.zhang@huawei.com,
	yi.zhang@huaweicloud.com,
	chengzhihao1@huawei.com,
	yukuai3@huawei.com,
	wangkefeng.wang@huawei.com
Subject: [RFC PATCH v4 11/34] ext4: optimize the EXT4_GET_BLOCKS_DELALLOC_RESERVE flag set
Date: Wed, 10 Apr 2024 21:27:55 +0800
Message-Id: <20240410132818.2812377-12-yi.zhang@huaweicloud.com>
In-Reply-To: <20240410132818.2812377-1-yi.zhang@huaweicloud.com>
References: <20240410132818.2812377-1-yi.zhang@huaweicloud.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

From: Zhang Yi <yi.zhang@huawei.com>

Magic EXT4_GET_BLOCKS_DELALLOC_RESERVE means caller is from the delayed
allocation writeout path, and the blocks and quotas has already been
checked when the data was copied into the page cache, there is no need
to check them again when we do the real allocation. But now we only set
this magic when allocating delayed allocated clusters, it makes things
complicated because we have to deal with the case of allocating
non-delayed allocated clusters, e.g. from fallocate, because in this
case, free space has already been claimed by ext4_mb_new_blocks(), we
shouldn't claim it again. Move setting EXT4_GET_BLOCKS_DELALLOC_RESERVE
to where we actually do block allocation could simplify the handling
process a lot, it means that we always set this magic once the
allocation range covers delalloc blocks, don't distinguish the
allocation path.

Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
---
 fs/ext4/inode.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 10a256cfcaa1..fd5a27db62c0 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -489,6 +489,14 @@ static int ext4_map_create_blocks(handle_t *handle, struct inode *inode,
 	unsigned int status;
 	int err, retval = 0;
 
+	/*
+	 * We pass in the magic EXT4_GET_BLOCKS_DELALLOC_RESERVE
+	 * indicates that the blocks and quotas has already been
+	 * checked when the data was copied into the page cache.
+	 */
+	if (map->m_flags & EXT4_MAP_DELAYED)
+		flags |= EXT4_GET_BLOCKS_DELALLOC_RESERVE;
+
 	/*
 	 * Here we clear m_flags because after allocating an new extent,
 	 * it will be set again.
@@ -2216,11 +2224,6 @@ static int mpage_map_one_extent(handle_t *handle, struct mpage_da_data *mpd)
 	 * writeback and there is nothing we can do about it so it might result
 	 * in data loss.  So use reserved blocks to allocate metadata if
 	 * possible.
-	 *
-	 * We pass in the magic EXT4_GET_BLOCKS_DELALLOC_RESERVE if
-	 * the blocks in question are delalloc blocks.  This indicates
-	 * that the blocks and quotas has already been checked when
-	 * the data was copied into the page cache.
 	 */
 	get_blocks_flags = EXT4_GET_BLOCKS_CREATE |
 			   EXT4_GET_BLOCKS_METADATA_NOFAIL |
@@ -2228,8 +2231,6 @@ static int mpage_map_one_extent(handle_t *handle, struct mpage_da_data *mpd)
 	dioread_nolock = ext4_should_dioread_nolock(inode);
 	if (dioread_nolock)
 		get_blocks_flags |= EXT4_GET_BLOCKS_IO_CREATE_EXT;
-	if (map->m_flags & BIT(BH_Delay))
-		get_blocks_flags |= EXT4_GET_BLOCKS_DELALLOC_RESERVE;
 
 	err = ext4_map_blocks(handle, inode, map, get_blocks_flags);
 	if (err < 0)
-- 
2.39.2