linux-kernel - [PATCH] f2fs: use aggressive GC policy during f2fs_disable

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-Id: <20220317090634.27354-1-chao@kernel.org>
Date:   Thu, 17 Mar 2022 17:06:34 +0800
From:   Chao Yu <chao@...nel.org>
To:     jaegeuk@...nel.org
Cc:     linux-f2fs-devel@...ts.sourceforge.net,
        linux-kernel@...r.kernel.org, Chao Yu <chao@...nel.org>,
        Chao Yu <chao.yu@...o.com>
Subject: [PATCH] f2fs: use aggressive GC policy during f2fs_disable_checkpoint()

Use more aggressive GC policy as below to improve performance of GC
during f2fs_disable_checkpoint():
- use GC_URGENT_HIGH mode to enable greedy algorithm and SSR allocator;
- use asynchronous GC and in-batch write to imporve possibility of IO
merge.

Signed-off-by: Chao Yu <chao.yu@...o.com>
---
 fs/f2fs/f2fs.h  |  3 +++
 fs/f2fs/super.c | 13 ++++++++++++-
 2 files changed, 15 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 3016b6354a68..e00b3993e2be 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -589,6 +589,9 @@ enum {
 /* dirty segments threshold for triggering CP */
 #define DEFAULT_DIRTY_THRESHOLD		4
 
+/* count of section which will be written in-batch during disabling checkpoint */
+#define INBATCH_WRITE_SECTION_COUNT	8
+
 /* for in-memory extent cache entry */
 #define F2FS_MIN_EXTENT_LEN	64	/* minimum extent length */
 
diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
index 9176597fdf94..08f384b0bd0d 100644
--- a/fs/f2fs/super.c
+++ b/fs/f2fs/super.c
@@ -2080,9 +2080,11 @@ static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
 {
 	unsigned int s_flags = sbi->sb->s_flags;
 	struct cp_control cpc;
+	unsigned int gc_mode;
 	int err = 0;
 	int ret;
 	block_t unusable;
+	int inbatch_cnt = 0;
 
 	if (s_flags & SB_RDONLY) {
 		f2fs_err(sbi, "checkpoint=disable on readonly fs");
@@ -2092,15 +2094,23 @@ static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
 
 	f2fs_update_time(sbi, DISABLE_TIME);
 
+	gc_mode = sbi->gc_mode;
+	sbi->gc_mode = GC_URGENT_HIGH;
+
 	while (!f2fs_time_over(sbi, DISABLE_TIME)) {
 		f2fs_down_write(&sbi->gc_lock);
-		err = f2fs_gc(sbi, true, false, false, NULL_SEGNO);
+		err = f2fs_gc(sbi, false, false, false, NULL_SEGNO);
 		if (err == -ENODATA) {
 			err = 0;
 			break;
 		}
 		if (err && err != -EAGAIN)
 			break;
+
+		if (++inbatch_cnt == INBATCH_WRITE_SECTION_COUNT) {
+			writeback_inodes_sb(sbi->sb, WB_REASON_SYNC);
+			inbatch_cnt = 0;
+		}
 	}
 
 	ret = sync_filesystem(sbi->sb);
@@ -2129,6 +2139,7 @@ static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
 out_unlock:
 	f2fs_up_write(&sbi->gc_lock);
 restore_flag:
+	sbi->gc_mode = gc_mode;
 	sbi->sb->s_flags = s_flags;	/* Restore SB_RDONLY status */
 	return err;
 }
-- 
2.32.0