f2fs: reduce expensive checkpoint trigger frequency
Subsystems (2)
Chao Yu
maintainerPATCH v4 1/1 f2fs: reduce expensive checkpoint trigger frequency
We may trigger high frequent checkpoint for below case:
1. mkdir /mnt/dir1; set dir1 encrypted
2. touch /mnt/file1; fsync /mnt/file1
3. mkdir /mnt/dir2; set dir2 encrypted
4. touch /mnt/file2; fsync /mnt/file2
...
Although, newly created dir and file are not related, due to commit bbf156f7afa7 ("f2fs: fix lost xattrs of directories"), we will trigger checkpoint whenever fsync() comes after a new encrypted dir created.
In order to avoid such performance regression issue, let's record an entry including directory's ino in global cache whenever we update directory's xattr data, and then triggerring checkpoint() only if xattr metadata of target file's parent was updated.
This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2) set_xattr(dir) w/ new xnid
3) create(file)
4) fsync(file)
--- fs/f2fs/f2fs.h | 2 ++ fs/f2fs/file.c | 3 +++ fs/f2fs/xattr.c | 14 ++++++++++++-- include/trace/events/f2fs.h | 3 ++- 4 files changed, 19 insertions(+), 3 deletions(-)
fs/f2fs/f2fs.h +2 -0
@@ -284,6 +284,7 @@ enum {APPEND_INO ,/* for append ino list */ UPDATE_INO ,/* for update ino list */ TRANS_DIR_INO ,/* for transactions dir ino list */ +ENC_DIR_INO ,/* for encrypted dir ino list */ FLUSH_INO ,/* for multiple device flushing */ MAX_INO_ENTRY, /* max. list */ };
@@ -1150,6 +1151,7 @@ enum cp_reason_type {CP_FASTBOOT_MODE ,CP_SPEC_LOG_NUM ,CP_RECOVER_DIR , + CP_ENC_DIR, }; enum iostat_type {
fs/f2fs/file.c +3 -0
@@ -217,6 +217,9 @@ static inline enum cp_reason_type need_do_checkpoint(struct inode *inode) f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, TRANS_DIR_INO)) cp_reason =CP_RECOVER_DIR ; + else if (f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, +ENC_DIR_INO )) + cp_reason =CP_ENC_DIR ;return cp_reason; }
fs/f2fs/xattr.c +12 -2
@@ -629,6 +629,7 @@ static int __f2fs_setxattr(struct inode *inode, int index, const char *name, const void *value, size_t size, struct page *ipage, int flags) { + struct f2fs_sb_info *sbi = F2FS_I_SB(inode);struct f2fs_xattr_entry *here, *last;void *base_addr, *last_base_addr;int found, newsize;
@@ -772,9 +773,18 @@ static int __f2fs_setxattr(struct inode *inode, int index,if (index ==F2FS_XATTR_INDEX_ENCRYPTION && !strcmp (name,F2FS_XATTR_NAME_ENCRYPTION_CONTEXT ))f2fs_set_encrypted_inode (inode); -if (S_ISDIR (inode->i_mode)) -set_sbi_flag (F2FS_I_SB (inode),SBI_NEED_CP ); +if (!S_ISDIR (inode->i_mode)) + goto same; +/* +* In restrict mode, fsync() always try to trigger checkpoint for all +* metadata consistency, in other mode, it triggers checkpoint when +* parent's xattr metadata was updated. +*/ +if (F2FS_OPTION (sbi).fsync_mode ==FSYNC_MODE_STRICT ) +set_sbi_flag (sbi,SBI_NEED_CP ); +else +f2fs_add_ino_entry (sbi, inode->i_ino,ENC_DIR_INO ); same:if (is_inode_flag_set (inode,FI_ACL_MODE )) { inode->i_mode =F2FS_I (inode)->i_acl_mode;
include/trace/events/f2fs.h +2 -1
@@ -139,7 +139,8 @@ TRACE_DEFINE_ENUM(EX_BLOCK_AGE); { CP_NODE_NEED_CP, "node needs cp" }, \ { CP_FASTBOOT_MODE, "fastboot mode" }, \ { CP_SPEC_LOG_NUM, "log type is 2" }, \ - { CP_RECOVER_DIR, "dir needs recovery" }) + { CP_RECOVER_DIR, "dir needs recovery" }, \ + { CP_ENC_DIR, "persist encryption policy" }) #define show_shutdown_mode(type) \ __print_symbolic(type, \
Signature · 1 line
2.40.1
-----邮件原件-----
发件人: Chao Yu <[email protected]>
发送时间: 2024年6月25日 14:55
收件人: [email protected]
抄送: [email protected]; [email protected]; Chao Yu <[email protected]>; wangzijie <[email protected]>; 牛志国 (Zhiguo Niu) <[email protected]>; Yunlei He <[email protected]> 主题: [PATCH v4] f2fs: reduce expensive checkpoint trigger frequency
注意: 这封邮件来自于外部。除非你确定邮件内容安全,否则不要点击任何链接和附件。
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
We may trigger high frequent checkpoint for below case:
1. mkdir /mnt/dir1; set dir1 encrypted
2. touch /mnt/file1; fsync /mnt/file1
3. mkdir /mnt/dir2; set dir2 encrypted
4. touch /mnt/file2; fsync /mnt/file2
...
Although, newly created dir and file are not related, due to commit bbf156f7afa7 ("f2fs: fix lost xattrs of directories"), we will trigger checkpoint whenever fsync() comes after a new encrypted dir created.
In order to avoid such performance regression issue, let's record an entry including directory's ino in global cache whenever we update directory's xattr data, and then triggerring checkpoint() only if xattr metadata of target file's parent was updated.
This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2) set_xattr(dir) w/ new xnid
3) create(file)
4) fsync(file)
--- fs/f2fs/f2fs.h | 2 ++ fs/f2fs/file.c | 3 +++ fs/f2fs/xattr.c | 14 ++++++++++++-- include/trace/events/f2fs.h | 3 ++- 4 files changed, 19 insertions(+), 3 deletions(-)
fs/f2fs/f2fs.h +2 -0
@@ -284,6 +284,7 @@ enum {APPEND_INO ,/* for append ino list */ UPDATE_INO ,/* for update ino list */ TRANS_DIR_INO ,/* for transactions dir ino list */ +ENC_DIR_INO ,/* for encrypted dir ino list */ FLUSH_INO ,/* for multiple device flushing */ MAX_INO_ENTRY, /* max. list */ };
@@ -1150,6 +1151,7 @@ enum cp_reason_type {CP_FASTBOOT_MODE ,CP_SPEC_LOG_NUM ,CP_RECOVER_DIR , + CP_ENC_DIR, }; enum iostat_type {
fs/f2fs/file.c +3 -0
@@ -217,6 +217,9 @@ static inline enum cp_reason_type need_do_checkpoint(struct inode *inode) f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, TRANS_DIR_INO)) cp_reason =CP_RECOVER_DIR ; + else if (f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, +ENC_DIR_INO )) + cp_reason =CP_ENC_DIR ;return cp_reason; }
fs/f2fs/xattr.c +10 -2
@@ -629,6 +629,7 @@ static int __f2fs_setxattr(struct inode *inode, int index, const char *name, const void *value, size_t size, struct page *ipage, int flags) { + struct f2fs_sb_info *sbi = F2FS_I_SB(inode);struct f2fs_xattr_entry *here, *last;void *base_addr, *last_base_addr;int found, newsize;
@@ -772,9 +773,18 @@ static int __f2fs_setxattr(struct inode *inode, int index,if (index ==F2FS_XATTR_INDEX_ENCRYPTION && !strcmp (name,F2FS_XATTR_NAME_ENCRYPTION_CONTEXT ))f2fs_set_encrypted_inode (inode); -if (S_ISDIR (inode->i_mode)) -set_sbi_flag (F2FS_I_SB (inode),SBI_NEED_CP ); +if (!S_ISDIR (inode->i_mode)) + goto same; +/* +* In restrict mode, fsync() always try to trigger checkpoint for all +* metadata consistency, in other mode, it triggers checkpoint when +* parent's xattr metadata was updated. +*/ +if (F2FS_OPTION (sbi).fsync_mode ==FSYNC_MODE_STRICT ) +set_sbi_flag (sbi,SBI_NEED_CP );
Hi Chao,
For this case, will it also cause the same issue with original issue when fsync_mode == FSYNC_MODE_STRICT ? if ckpt thread is blocked by some reasons and SBI_NEED_CP is not cleared in time, Subsequent fsync will trigger cp?
+ else
+ f2fs_add_ino_entry(sbi, inode->i_ino, ENC_DIR_INO);
This patch version regardless of whether dir is encrypted or not, so this name(ENC_DIR_INO) can be rename other for more accurate? Thanks!
same:
if (is_inode_flag_set(inode, FI_ACL_MODE)) {
inode->i_mode = F2FS_I(inode)->i_acl_mode; diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h index ed794b5fefbe..e4a94995e9a8 100644include/trace/events/f2fs.h +2 -1
@@ -139,7 +139,8 @@ TRACE_DEFINE_ENUM(EX_BLOCK_AGE); { CP_NODE_NEED_CP, "node needs cp" }, \ { CP_FASTBOOT_MODE, "fastboot mode" }, \ { CP_SPEC_LOG_NUM, "log type is 2" }, \ - { CP_RECOVER_DIR, "dir needs recovery" }) + { CP_RECOVER_DIR, "dir needs recovery" }, \ + { CP_ENC_DIR, "persist encryption policy" }) #define show_shutdown_mode(type) \ __print_symbolic(type, \
--
2.40.1
Chao Yu
maintainerQuoted 104 lines · 牛志国 (Zhiguo Niu)
-----邮件原件-----
发件人: Chao Yu <[email protected]>
发送时间: 2024年6月25日 14:55
收件人: [email protected]
抄送: [email protected]; [email protected]; Chao Yu <[email protected]>; wangzijie <[email protected]>; 牛志国 (Zhiguo Niu) <[email protected]>; Yunlei He <[email protected]> 主题: [PATCH v4] f2fs: reduce expensive checkpoint trigger frequency注意: 这封邮件来自于外部。除非你确定邮件内容安全,否则不要点击任何链接和附件。
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.We may trigger high frequent checkpoint for below case:
1. mkdir/mnt/dir1; set dir1 encrypted
2. touch/mnt/file1; fsync/mnt/file1
3. mkdir/mnt/dir2; set dir2 encrypted
4. touch/mnt/file2; fsync/mnt/file2
...Although, newly created dir and file are not related, due to commit
bbf156f7afa7("f2fs: fix lost xattrs of directories"), we will trigger checkpoint wheneverfsync()comes after a new encrypted dir created.In order to avoid such performance regression issue, let's record an entry including directory's ino in global cache whenever we update directory's xattr data, and then triggerring
checkpoint()only if xattr metadata of target file's parent was updated.This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2)set_xattr(dir)w/ new xnid
3) create(file)
4) fsync(file)Fixes bbf156f7afa7 ("f2fs: fix lost xattrs of directories")Reported-by wangzijie <[email protected]>Reported-by Zhiguo Niu <[email protected]>Tested-by Zhiguo Niu <[email protected]>Reported-by Yunlei He <[email protected]>Signed-off-by Chao Yu <[email protected]>--- fs/f2fs/f2fs.h | 2 ++ fs/f2fs/file.c | 3 +++ fs/f2fs/xattr.c | 14 ++++++++++++-- include/trace/events/f2fs.h | 3 ++- 4 files changed, 19 insertions(+), 3 deletions(-)fs/f2fs/f2fs.h +2
-0@@ -284,6 +284,7 @@ enum {APPEND_INO ,/* for append ino list */ UPDATE_INO ,/* for update ino list */ TRANS_DIR_INO ,/* for transactions dir ino list */ +ENC_DIR_INO ,/* for encrypted dir ino list */ FLUSH_INO ,/* for multiple device flushing */ MAX_INO_ENTRY, /* max. list */ };@@ -1150,6 +1151,7 @@ enum cp_reason_type {CP_FASTBOOT_MODE ,CP_SPEC_LOG_NUM ,CP_RECOVER_DIR , + CP_ENC_DIR, }; enum iostat_type {fs/f2fs/file.c +3
-0@@ -217,6 +217,9 @@ static inline enum cp_reason_type need_do_checkpoint(struct inode *inode) f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, TRANS_DIR_INO)) cp_reason =CP_RECOVER_DIR ; + else if (f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, +ENC_DIR_INO )) + cp_reason =CP_ENC_DIR ;return cp_reason; }fs/f2fs/xattr.c +10
-2@@ -629,6 +629,7 @@ static int __f2fs_setxattr(struct inode *inode, int index, const char *name, const void *value, size_t size, struct page *ipage, int flags) { + struct f2fs_sb_info *sbi = F2FS_I_SB(inode);struct f2fs_xattr_entry *here, *last;void *base_addr, *last_base_addr;int found, newsize;@@ -772,9 +773,18 @@ static int __f2fs_setxattr(struct inode *inode, int index,if (index ==F2FS_XATTR_INDEX_ENCRYPTION && !strcmp (name,F2FS_XATTR_NAME_ENCRYPTION_CONTEXT ))f2fs_set_encrypted_inode (inode); -if (S_ISDIR (inode->i_mode)) -set_sbi_flag (F2FS_I_SB (inode),SBI_NEED_CP ); +if (!S_ISDIR (inode->i_mode)) + goto same; +/* +* In restrict mode, fsync() always try to trigger checkpoint for all +* metadata consistency, in other mode, it triggers checkpoint when +* parent's xattr metadata was updated. +*/ +if (F2FS_OPTION (sbi).fsync_mode ==FSYNC_MODE_STRICT ) +set_sbi_flag (sbi,SBI_NEED_CP ); Hi Chao,
Quoted 2 lines
For this case, will it also cause the same issue with original issue when
fsync_mode==FSYNC_MODE_STRICT? if ckpt thread is blocked by some reasons andSBI_NEED_CPis not cleared in time, Subsequent fsync will trigger cp?
Hi Zhiguo,
If there is no dirty data after previous CP, later CP triggered by subsequent fsync will return directly?
if (!is_sbi_flag_set(sbi, SBI_IS_DIRTY) && ((cpc->reason & CP_FASTBOOT) || (cpc->reason & CP_SYNC) || ((cpc->reason & CP_DISCARD) && !sbi->discard_blks))) goto out;
Quoted 3 lines
fs/f2fs/xattr.c +2
-0+else +f2fs_add_ino_entry (sbi, inode->i_ino,ENC_DIR_INO );This patch version regardless of whether dir is encrypted or not, so this
name(ENC_DIR_INO)can be rename other for more accurate?
Thanks for pointing out this, let me fix it in v5.
Thanks,
Quoted 19 lines
Thanks!
same:if (is_inode_flag_set(inode, FI_ACL_MODE)) { inode->i_mode = F2FS_I(inode)->i_acl_mode; diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h index ed794b5fefbe..e4a94995e9a8 100644include/trace/events/f2fs.h +2
-2@@ -139,7 +139,8 @@ TRACE_DEFINE_ENUM(EX_BLOCK_AGE); { CP_NODE_NEED_CP, "node needs cp" }, \ { CP_FASTBOOT_MODE, "fastboot mode" }, \ { CP_SPEC_LOG_NUM, "log type is 2" }, \ - { CP_RECOVER_DIR, "dir needs recovery" }) + { CP_RECOVER_DIR, "dir needs recovery" }, \ + { CP_ENC_DIR, "persist encryption policy" }) #define show_shutdown_mode(type) \ __print_symbolic(type, \ --2.40.1
Quoted 111 lines · Chao Yu <[email protected]> 于2024年6月25日周二 18:38写道:
Quoted 104 lines · 牛志国 (Zhiguo Niu)
-----邮件原件-----
发件人: Chao Yu <[email protected]>
发送时间: 2024年6月25日 14:55
收件人: [email protected]
抄送: [email protected]; [email protected]; Chao Yu <[email protected]>; wangzijie <[email protected]>; 牛志国 (Zhiguo Niu) <[email protected]>; Yunlei He <[email protected]> 主题: [PATCH v4] f2fs: reduce expensive checkpoint trigger frequency注意: 这封邮件来自于外部。除非你确定邮件内容安全,否则不要点击任何链接和附件。
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.We may trigger high frequent checkpoint for below case:
1. mkdir/mnt/dir1; set dir1 encrypted
2. touch/mnt/file1; fsync/mnt/file1
3. mkdir/mnt/dir2; set dir2 encrypted
4. touch/mnt/file2; fsync/mnt/file2
...Although, newly created dir and file are not related, due to commit
bbf156f7afa7("f2fs: fix lost xattrs of directories"), we will trigger checkpoint wheneverfsync()comes after a new encrypted dir created.In order to avoid such performance regression issue, let's record an entry including directory's ino in global cache whenever we update directory's xattr data, and then triggerring
checkpoint()only if xattr metadata of target file's parent was updated.This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2)set_xattr(dir)w/ new xnid
3) create(file)
4) fsync(file)Fixes bbf156f7afa7 ("f2fs: fix lost xattrs of directories")Reported-by wangzijie <[email protected]>Reported-by Zhiguo Niu <[email protected]>Tested-by Zhiguo Niu <[email protected]>Reported-by Yunlei He <[email protected]>Signed-off-by Chao Yu <[email protected]>--- fs/f2fs/f2fs.h | 2 ++ fs/f2fs/file.c | 3 +++ fs/f2fs/xattr.c | 14 ++++++++++++-- include/trace/events/f2fs.h | 3 ++- 4 files changed, 19 insertions(+), 3 deletions(-)fs/f2fs/f2fs.h +2
-0@@ -284,6 +284,7 @@ enum {APPEND_INO ,/* for append ino list */ UPDATE_INO ,/* for update ino list */ TRANS_DIR_INO ,/* for transactions dir ino list */ +ENC_DIR_INO ,/* for encrypted dir ino list */ FLUSH_INO ,/* for multiple device flushing */ MAX_INO_ENTRY, /* max. list */ };@@ -1150,6 +1151,7 @@ enum cp_reason_type {CP_FASTBOOT_MODE ,CP_SPEC_LOG_NUM ,CP_RECOVER_DIR , + CP_ENC_DIR, }; enum iostat_type {fs/f2fs/file.c +3
-0@@ -217,6 +217,9 @@ static inline enum cp_reason_type need_do_checkpoint(struct inode *inode) f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, TRANS_DIR_INO)) cp_reason =CP_RECOVER_DIR ; + else if (f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, +ENC_DIR_INO )) + cp_reason =CP_ENC_DIR ;return cp_reason; }fs/f2fs/xattr.c +10
-2@@ -629,6 +629,7 @@ static int __f2fs_setxattr(struct inode *inode, int index, const char *name, const void *value, size_t size, struct page *ipage, int flags) { + struct f2fs_sb_info *sbi = F2FS_I_SB(inode);struct f2fs_xattr_entry *here, *last;void *base_addr, *last_base_addr;int found, newsize;@@ -772,9 +773,18 @@ static int __f2fs_setxattr(struct inode *inode, int index,if (index ==F2FS_XATTR_INDEX_ENCRYPTION && !strcmp (name,F2FS_XATTR_NAME_ENCRYPTION_CONTEXT ))f2fs_set_encrypted_inode (inode); -if (S_ISDIR (inode->i_mode)) -set_sbi_flag (F2FS_I_SB (inode),SBI_NEED_CP ); +if (!S_ISDIR (inode->i_mode)) + goto same; +/* +* In restrict mode, fsync() always try to trigger checkpoint for all +* metadata consistency, in other mode, it triggers checkpoint when +* parent's xattr metadata was updated. +*/ +if (F2FS_OPTION (sbi).fsync_mode ==FSYNC_MODE_STRICT ) +set_sbi_flag (sbi,SBI_NEED_CP ); Hi Chao,Quoted 2 lines
For this case, will it also cause the same issue with original issue when
fsync_mode==FSYNC_MODE_STRICT? if ckpt thread is blocked by some reasons andSBI_NEED_CPis not cleared in time, Subsequent fsync will trigger cp?Hi Zhiguo,
Quoted 7 lines
If there is no dirty data after previous CP, later CP triggered by subsequent fsync will return directly?
if (!is_sbi_flag_set(sbi, SBI_IS_DIRTY) && ((cpc->reason & CP_FASTBOOT) || (cpc->reason & CP_SYNC) || ((cpc->reason & CP_DISCARD) && !sbi->discard_blks))) goto out;
Hi Chao,
I mean in the case of F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT, this will still cause the same issue with original question.
Let me repost the problems I encountered .
1.set_sbi_flag(F2FS_I_SB(inode), SBI_NEED_CP);
2. cp threa is blocked by io resource
#0 [ffffffc015fcb330] __switch_to at ffffffc010196350
#1 [ffffffc015fcb390] __schedule at ffffffc01168e53c
#2 [ffffffc015fcb3f0] schedule at ffffffc01168ed7c
#3 [ffffffc015fcb450] io_schedule at ffffffc01168f7a0
#4 [ffffffc015fcb4c0] blk_mq_get_tag at ffffffc0101008a4
#5 [ffffffc015fcb530] blk_mq_get_request at ffffffc0109241b0 #6 [ffffffc015fcb5f0] blk_mq_make_request at ffffffc0109233bc #7 [ffffffc015fcb680] generic_make_request at ffffffc0100fc6ec #8 [ffffffc015fcb700] submit_bio at ffffffc0100fc3b8
#9 [ffffffc015fcb750] __submit_bio at ffffffc01081a2e0
#10 [ffffffc015fcb7d0] __submit_merged_bio at ffffffc01081b07c #11 [ffffffc015fcb8a0] f2fs_submit_page_write at ffffffc0100ecd3c #12 [ffffffc015fcb990] f2fs_do_write_meta_page at ffffffc010845738 #13 [ffffffc015fcb9d0] __f2fs_write_meta_page at ffffffc01080a8f4 #14 [ffffffc015fcbb60] f2fs_sync_meta_pages at ffffffc01080a684 #15 [ffffffc015fcbca0] do_checkpoint at ffffffc01080f0a8
#16 [ffffffc015fcbd10] f2fs_write_checkpoint at ffffffc01080e50c #17 [ffffffc015fcbdb0] __checkpoint_and_complete_reqs at ffffffc010810f54 #18 [ffffffc015fcbe40] issue_checkpoint_thread at ffffffc0108113ec #19 [ffffffc015fcbe80] kthread at ffffffc0102665b0
3. subsequent fsync will still issue ckpt becasue SBI_NEED_CP is set, and if this is a very important thread, it will cause system issue, such as init.
but this is should be avoided.
PID: 287 TASK: ffffff80f9eb0ec0 CPU: 2 COMMAND: "init"
#0 [ffffffc01389bab0] __switch_to at ffffffc010196350 #1 [ffffffc01389bb10] __schedule at ffffffc01168e53c #2 [ffffffc01389bb70] schedule at ffffffc01168ed7c
#3 [ffffffc01389bbc0] wait_for_completion at ffffffc011692368 #4 [ffffffc01389bca0] f2fs_issue_checkpoint at ffffffc010810cb0 #5 [ffffffc01389bd00] f2fs_sync_fs at ffffffc0107f4e1c #6 [ffffffc01389bdc0] f2fs_do_sync_file at ffffffc0107d4d44 #7 [ffffffc01389be20] f2fs_sync_file at ffffffc0107d492c #8 [ffffffc01389be30] __arm64_sys_fsync at ffffffc0105d31d8 #9 [ffffffc01389be70] el0_svc_common at ffffffc0101aa550 #10 [ffffffc01389beb0] el0_svc_handler at ffffffc0100886fc
thanks!
Quoted 34 lines
Quoted 4 lines
fs/f2fs/xattr.c +2
-0+else +f2fs_add_ino_entry (sbi, inode->i_ino,ENC_DIR_INO );This patch version regardless of whether dir is encrypted or not, so this
name(ENC_DIR_INO)can be rename other for more accurate?Thanks for pointing out this, let me fix it in v5.
Thanks,
Quoted 17 lines
Thanks!
same:if (is_inode_flag_set(inode, FI_ACL_MODE)) { inode->i_mode = F2FS_I(inode)->i_acl_mode; diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h index ed794b5fefbe..e4a94995e9a8 100644include/trace/events/f2fs.h +2
-2@@ -139,7 +139,8 @@ TRACE_DEFINE_ENUM(EX_BLOCK_AGE); { CP_NODE_NEED_CP, "node needs cp" }, \ { CP_FASTBOOT_MODE, "fastboot mode" }, \ { CP_SPEC_LOG_NUM, "log type is 2" }, \ - { CP_RECOVER_DIR, "dir needs recovery" }) + { CP_RECOVER_DIR, "dir needs recovery" }, \ + { CP_ENC_DIR, "persist encryption policy" }) #define show_shutdown_mode(type) \ __print_symbolic(type, \ --Quoted 2 lines
2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
Chao Yu
maintainerQuoted 126 lines · Zhiguo Niu
Quoted 111 lines · Chao Yu <[email protected]> 于2024年6月25日周二 18:38写道:
Quoted 104 lines · 牛志国 (Zhiguo Niu)
-----邮件原件-----
发件人: Chao Yu <[email protected]>
发送时间: 2024年6月25日 14:55
收件人: [email protected]
抄送: [email protected]; [email protected]; Chao Yu <[email protected]>; wangzijie <[email protected]>; 牛志国 (Zhiguo Niu) <[email protected]>; Yunlei He <[email protected]> 主题: [PATCH v4] f2fs: reduce expensive checkpoint trigger frequency注意: 这封邮件来自于外部。除非你确定邮件内容安全,否则不要点击任何链接和附件。
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.We may trigger high frequent checkpoint for below case:
1. mkdir/mnt/dir1; set dir1 encrypted
2. touch/mnt/file1; fsync/mnt/file1
3. mkdir/mnt/dir2; set dir2 encrypted
4. touch/mnt/file2; fsync/mnt/file2
...Although, newly created dir and file are not related, due to commit
bbf156f7afa7("f2fs: fix lost xattrs of directories"), we will trigger checkpoint wheneverfsync()comes after a new encrypted dir created.In order to avoid such performance regression issue, let's record an entry including directory's ino in global cache whenever we update directory's xattr data, and then triggerring
checkpoint()only if xattr metadata of target file's parent was updated.This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2)set_xattr(dir)w/ new xnid
3) create(file)
4) fsync(file)Fixes bbf156f7afa7 ("f2fs: fix lost xattrs of directories")Reported-by wangzijie <[email protected]>Reported-by Zhiguo Niu <[email protected]>Tested-by Zhiguo Niu <[email protected]>Reported-by Yunlei He <[email protected]>Signed-off-by Chao Yu <[email protected]>--- fs/f2fs/f2fs.h | 2 ++ fs/f2fs/file.c | 3 +++ fs/f2fs/xattr.c | 14 ++++++++++++-- include/trace/events/f2fs.h | 3 ++- 4 files changed, 19 insertions(+), 3 deletions(-)fs/f2fs/f2fs.h +2
-0@@ -284,6 +284,7 @@ enum {APPEND_INO ,/* for append ino list */ UPDATE_INO ,/* for update ino list */ TRANS_DIR_INO ,/* for transactions dir ino list */ +ENC_DIR_INO ,/* for encrypted dir ino list */ FLUSH_INO ,/* for multiple device flushing */ MAX_INO_ENTRY, /* max. list */ };@@ -1150,6 +1151,7 @@ enum cp_reason_type {CP_FASTBOOT_MODE ,CP_SPEC_LOG_NUM ,CP_RECOVER_DIR , + CP_ENC_DIR, }; enum iostat_type {fs/f2fs/file.c +3
-0@@ -217,6 +217,9 @@ static inline enum cp_reason_type need_do_checkpoint(struct inode *inode) f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, TRANS_DIR_INO)) cp_reason =CP_RECOVER_DIR ; + else if (f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, +ENC_DIR_INO )) + cp_reason =CP_ENC_DIR ;return cp_reason; }fs/f2fs/xattr.c +10
-2@@ -629,6 +629,7 @@ static int __f2fs_setxattr(struct inode *inode, int index, const char *name, const void *value, size_t size, struct page *ipage, int flags) { + struct f2fs_sb_info *sbi = F2FS_I_SB(inode);struct f2fs_xattr_entry *here, *last;void *base_addr, *last_base_addr;int found, newsize;@@ -772,9 +773,18 @@ static int __f2fs_setxattr(struct inode *inode, int index,if (index ==F2FS_XATTR_INDEX_ENCRYPTION && !strcmp (name,F2FS_XATTR_NAME_ENCRYPTION_CONTEXT ))f2fs_set_encrypted_inode (inode); -if (S_ISDIR (inode->i_mode)) -set_sbi_flag (F2FS_I_SB (inode),SBI_NEED_CP ); +if (!S_ISDIR (inode->i_mode)) + goto same; +/* +* In restrict mode, fsync() always try to trigger checkpoint for all +* metadata consistency, in other mode, it triggers checkpoint when +* parent's xattr metadata was updated. +*/ +if (F2FS_OPTION (sbi).fsync_mode ==FSYNC_MODE_STRICT ) +set_sbi_flag (sbi,SBI_NEED_CP ); Hi Chao,Quoted 2 lines
For this case, will it also cause the same issue with original issue when
fsync_mode==FSYNC_MODE_STRICT? if ckpt thread is blocked by some reasons andSBI_NEED_CPis not cleared in time, Subsequent fsync will trigger cp?Hi Zhiguo,
Quoted 7 lines
If there is no dirty data after previous CP, later CP triggered by subsequent fsync will return directly?
if (!is_sbi_flag_set(sbi, SBI_IS_DIRTY) && ((cpc->reason & CP_FASTBOOT) || (cpc->reason & CP_SYNC) || ((cpc->reason & CP_DISCARD) && !sbi->discard_blks))) goto out;Hi Chao,
I mean in the case of F2FS_OPTION(sbi).fsync_mode ==FSYNC_MODE_STRICT, this will still cause the same issue with original question.
Let me repost the problems I encountered .
Quoted 8 lines
1.set_sbi_flag(F2FS_I_SB(inode),
SBI_NEED_CP);2. cp threa is blocked by io resource
#0 [ffffffc015fcb330]__switch_toatffffffc010196350#1 [ffffffc015fcb390]__scheduleatffffffc01168e53c#2 [ffffffc015fcb3f0] schedule atffffffc01168ed7c#3 [ffffffc015fcb450]io_scheduleatffffffc01168f7a0#4 [ffffffc015fcb4c0]blk_mq_get_tagatffffffc0101008a4
It seems there is no free tag at that time, can you check its context to see why IO queue can be congested once you can reproduce this issue?
If the bottleneck is performance of storage, how about using other fsync_mode if its data consistency performace matches the requirement.
Thanks,
Quoted 69 lines
#5 [
ffffffc015fcb530]blk_mq_get_requestatffffffc0109241b0#6 [ffffffc015fcb5f0]blk_mq_make_requestatffffffc0109233bc#7 [ffffffc015fcb680]generic_make_requestatffffffc0100fc6ec#8 [ffffffc015fcb700]submit_bioatffffffc0100fc3b8
#9 [ffffffc015fcb750]__submit_bioatffffffc01081a2e0
#10 [ffffffc015fcb7d0]__submit_merged_bioatffffffc01081b07c#11 [ffffffc015fcb8a0]f2fs_submit_page_writeatffffffc0100ecd3c#12 [ffffffc015fcb990]f2fs_do_write_meta_pageatffffffc010845738#13 [ffffffc015fcb9d0]__f2fs_write_meta_pageatffffffc01080a8f4#14 [ffffffc015fcbb60]f2fs_sync_meta_pagesatffffffc01080a684#15 [ffffffc015fcbca0]do_checkpointatffffffc01080f0a8
#16 [ffffffc015fcbd10]f2fs_write_checkpointatffffffc01080e50c#17 [ffffffc015fcbdb0]__checkpoint_and_complete_reqsatffffffc010810f54#18 [ffffffc015fcbe40]issue_checkpoint_threadatffffffc0108113ec#19 [ffffffc015fcbe80] kthread atffffffc0102665b03. subsequent fsync will still issue ckpt becasue
SBI_NEED_CPis set, and if this is a very important thread, it will cause system issue, such as init.
but this is should be avoided.
PID: 287 TASK:ffffff80f9eb0ec0CPU: 2 COMMAND: "init"#0 [
ffffffc01389bab0]__switch_toatffffffc010196350#1 [ffffffc01389bb10]__scheduleatffffffc01168e53c#2 [ffffffc01389bb70] schedule atffffffc01168ed7c
#3 [ffffffc01389bbc0]wait_for_completionatffffffc011692368#4 [ffffffc01389bca0]f2fs_issue_checkpointatffffffc010810cb0#5 [ffffffc01389bd00]f2fs_sync_fsatffffffc0107f4e1c#6 [ffffffc01389bdc0]f2fs_do_sync_fileatffffffc0107d4d44#7 [ffffffc01389be20]f2fs_sync_fileatffffffc0107d492c#8 [ffffffc01389be30]__arm64_sys_fsyncatffffffc0105d31d8#9 [ffffffc01389be70]el0_svc_commonatffffffc0101aa550#10 [ffffffc01389beb0]el0_svc_handleratffffffc0100886fcthanks!
Quoted 34 lines
Quoted 4 lines
fs/f2fs/xattr.c +2
-0+else +f2fs_add_ino_entry (sbi, inode->i_ino,ENC_DIR_INO );This patch version regardless of whether dir is encrypted or not, so this
name(ENC_DIR_INO)can be rename other for more accurate?Thanks for pointing out this, let me fix it in v5.
Thanks,
Quoted 17 lines
Thanks!
same:if (is_inode_flag_set(inode, FI_ACL_MODE)) { inode->i_mode = F2FS_I(inode)->i_acl_mode; diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h index ed794b5fefbe..e4a94995e9a8 100644include/trace/events/f2fs.h +2
-2@@ -139,7 +139,8 @@ TRACE_DEFINE_ENUM(EX_BLOCK_AGE); { CP_NODE_NEED_CP, "node needs cp" }, \ { CP_FASTBOOT_MODE, "fastboot mode" }, \ { CP_SPEC_LOG_NUM, "log type is 2" }, \ - { CP_RECOVER_DIR, "dir needs recovery" }) + { CP_RECOVER_DIR, "dir needs recovery" }, \ + { CP_ENC_DIR, "persist encryption policy" }) #define show_shutdown_mode(type) \ __print_symbolic(type, \ --Quoted 2 lines
2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
Quoted 128 lines · Chao Yu <[email protected]> 于2024年6月26日周三 09:36写道:
Quoted 126 lines · Zhiguo Niu
Quoted 111 lines · Chao Yu <[email protected]> 于2024年6月25日周二 18:38写道:
Quoted 104 lines · 牛志国 (Zhiguo Niu)
-----邮件原件-----
发件人: Chao Yu <[email protected]>
发送时间: 2024年6月25日 14:55
收件人: [email protected]
抄送: [email protected]; [email protected]; Chao Yu <[email protected]>; wangzijie <[email protected]>; 牛志国 (Zhiguo Niu) <[email protected]>; Yunlei He <[email protected]> 主题: [PATCH v4] f2fs: reduce expensive checkpoint trigger frequency注意: 这封邮件来自于外部。除非你确定邮件内容安全,否则不要点击任何链接和附件。
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.We may trigger high frequent checkpoint for below case:
1. mkdir/mnt/dir1; set dir1 encrypted
2. touch/mnt/file1; fsync/mnt/file1
3. mkdir/mnt/dir2; set dir2 encrypted
4. touch/mnt/file2; fsync/mnt/file2
...Although, newly created dir and file are not related, due to commit
bbf156f7afa7("f2fs: fix lost xattrs of directories"), we will trigger checkpoint wheneverfsync()comes after a new encrypted dir created.In order to avoid such performance regression issue, let's record an entry including directory's ino in global cache whenever we update directory's xattr data, and then triggerring
checkpoint()only if xattr metadata of target file's parent was updated.This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2)set_xattr(dir)w/ new xnid
3) create(file)
4) fsync(file)Fixes bbf156f7afa7 ("f2fs: fix lost xattrs of directories")Reported-by wangzijie <[email protected]>Reported-by Zhiguo Niu <[email protected]>Tested-by Zhiguo Niu <[email protected]>Reported-by Yunlei He <[email protected]>Signed-off-by Chao Yu <[email protected]>--- fs/f2fs/f2fs.h | 2 ++ fs/f2fs/file.c | 3 +++ fs/f2fs/xattr.c | 14 ++++++++++++-- include/trace/events/f2fs.h | 3 ++- 4 files changed, 19 insertions(+), 3 deletions(-)fs/f2fs/f2fs.h +2
-0@@ -284,6 +284,7 @@ enum {APPEND_INO ,/* for append ino list */ UPDATE_INO ,/* for update ino list */ TRANS_DIR_INO ,/* for transactions dir ino list */ +ENC_DIR_INO ,/* for encrypted dir ino list */ FLUSH_INO ,/* for multiple device flushing */ MAX_INO_ENTRY, /* max. list */ };@@ -1150,6 +1151,7 @@ enum cp_reason_type {CP_FASTBOOT_MODE ,CP_SPEC_LOG_NUM ,CP_RECOVER_DIR , + CP_ENC_DIR, }; enum iostat_type {fs/f2fs/file.c +3
-0@@ -217,6 +217,9 @@ static inline enum cp_reason_type need_do_checkpoint(struct inode *inode) f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, TRANS_DIR_INO)) cp_reason =CP_RECOVER_DIR ; + else if (f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino, +ENC_DIR_INO )) + cp_reason =CP_ENC_DIR ;return cp_reason; }fs/f2fs/xattr.c +10
-2@@ -629,6 +629,7 @@ static int __f2fs_setxattr(struct inode *inode, int index, const char *name, const void *value, size_t size, struct page *ipage, int flags) { + struct f2fs_sb_info *sbi = F2FS_I_SB(inode);struct f2fs_xattr_entry *here, *last;void *base_addr, *last_base_addr;int found, newsize;@@ -772,9 +773,18 @@ static int __f2fs_setxattr(struct inode *inode, int index,if (index ==F2FS_XATTR_INDEX_ENCRYPTION && !strcmp (name,F2FS_XATTR_NAME_ENCRYPTION_CONTEXT ))f2fs_set_encrypted_inode (inode); -if (S_ISDIR (inode->i_mode)) -set_sbi_flag (F2FS_I_SB (inode),SBI_NEED_CP ); +if (!S_ISDIR (inode->i_mode)) + goto same; +/* +* In restrict mode, fsync() always try to trigger checkpoint for all +* metadata consistency, in other mode, it triggers checkpoint when +* parent's xattr metadata was updated. +*/ +if (F2FS_OPTION (sbi).fsync_mode ==FSYNC_MODE_STRICT ) +set_sbi_flag (sbi,SBI_NEED_CP ); Hi Chao,Quoted 2 lines
For this case, will it also cause the same issue with original issue when
fsync_mode==FSYNC_MODE_STRICT? if ckpt thread is blocked by some reasons andSBI_NEED_CPis not cleared in time, Subsequent fsync will trigger cp?Hi Zhiguo,
Quoted 7 lines
If there is no dirty data after previous CP, later CP triggered by subsequent fsync will return directly?
if (!is_sbi_flag_set(sbi, SBI_IS_DIRTY) && ((cpc->reason & CP_FASTBOOT) || (cpc->reason & CP_SYNC) || ((cpc->reason & CP_DISCARD) && !sbi->discard_blks))) goto out;Hi Chao,
I mean in the case of F2FS_OPTION(sbi).fsync_mode ==FSYNC_MODE_STRICT, this will still cause the same issue with original question.
Let me repost the problems I encountered .
Quoted 9 lines
Quoted 8 lines
1.set_sbi_flag(F2FS_I_SB(inode),
SBI_NEED_CP);2. cp threa is blocked by io resource
#0 [ffffffc015fcb330]__switch_toatffffffc010196350#1 [ffffffc015fcb390]__scheduleatffffffc01168e53c#2 [ffffffc015fcb3f0] schedule atffffffc01168ed7c#3 [ffffffc015fcb450]io_scheduleatffffffc01168f7a0#4 [ffffffc015fcb4c0]blk_mq_get_tagatffffffc0101008a4
Hi Chao
Quoted 2 lines
It seems there is no free tag at that time, can you check its context to see why IO queue can be congested once you can reproduce this issue?
Yes, its monkek test report this issue, and IO load is heavy
Quoted 3 lines
If the bottleneck is performance of storage, how about using other
fsync_modeif its data consistency performace matches the requirement.
agree this, thanks a lot.
Quoted 72 lines
Thanks,
Quoted 69 lines
#5 [
ffffffc015fcb530]blk_mq_get_requestatffffffc0109241b0#6 [ffffffc015fcb5f0]blk_mq_make_requestatffffffc0109233bc#7 [ffffffc015fcb680]generic_make_requestatffffffc0100fc6ec#8 [ffffffc015fcb700]submit_bioatffffffc0100fc3b8
#9 [ffffffc015fcb750]__submit_bioatffffffc01081a2e0
#10 [ffffffc015fcb7d0]__submit_merged_bioatffffffc01081b07c#11 [ffffffc015fcb8a0]f2fs_submit_page_writeatffffffc0100ecd3c#12 [ffffffc015fcb990]f2fs_do_write_meta_pageatffffffc010845738#13 [ffffffc015fcb9d0]__f2fs_write_meta_pageatffffffc01080a8f4#14 [ffffffc015fcbb60]f2fs_sync_meta_pagesatffffffc01080a684#15 [ffffffc015fcbca0]do_checkpointatffffffc01080f0a8
#16 [ffffffc015fcbd10]f2fs_write_checkpointatffffffc01080e50c#17 [ffffffc015fcbdb0]__checkpoint_and_complete_reqsatffffffc010810f54#18 [ffffffc015fcbe40]issue_checkpoint_threadatffffffc0108113ec#19 [ffffffc015fcbe80] kthread atffffffc0102665b03. subsequent fsync will still issue ckpt becasue
SBI_NEED_CPis set, and if this is a very important thread, it will cause system issue, such as init.
but this is should be avoided.
PID: 287 TASK:ffffff80f9eb0ec0CPU: 2 COMMAND: "init"#0 [
ffffffc01389bab0]__switch_toatffffffc010196350#1 [ffffffc01389bb10]__scheduleatffffffc01168e53c#2 [ffffffc01389bb70] schedule atffffffc01168ed7c
#3 [ffffffc01389bbc0]wait_for_completionatffffffc011692368#4 [ffffffc01389bca0]f2fs_issue_checkpointatffffffc010810cb0#5 [ffffffc01389bd00]f2fs_sync_fsatffffffc0107f4e1c#6 [ffffffc01389bdc0]f2fs_do_sync_fileatffffffc0107d4d44#7 [ffffffc01389be20]f2fs_sync_fileatffffffc0107d492c#8 [ffffffc01389be30]__arm64_sys_fsyncatffffffc0105d31d8#9 [ffffffc01389be70]el0_svc_commonatffffffc0101aa550#10 [ffffffc01389beb0]el0_svc_handleratffffffc0100886fcthanks!
Quoted 34 lines
Quoted 4 lines
fs/f2fs/xattr.c +2
-0+else +f2fs_add_ino_entry (sbi, inode->i_ino,ENC_DIR_INO );This patch version regardless of whether dir is encrypted or not, so this
name(ENC_DIR_INO)can be rename other for more accurate?Thanks for pointing out this, let me fix it in v5.
Thanks,
Quoted 17 lines
Thanks!
same:if (is_inode_flag_set(inode, FI_ACL_MODE)) { inode->i_mode = F2FS_I(inode)->i_acl_mode; diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h index ed794b5fefbe..e4a94995e9a8 100644include/trace/events/f2fs.h +2
-2@@ -139,7 +139,8 @@ TRACE_DEFINE_ENUM(EX_BLOCK_AGE); { CP_NODE_NEED_CP, "node needs cp" }, \ { CP_FASTBOOT_MODE, "fastboot mode" }, \ { CP_SPEC_LOG_NUM, "log type is 2" }, \ - { CP_RECOVER_DIR, "dir needs recovery" }) + { CP_RECOVER_DIR, "dir needs recovery" }, \ + { CP_ENC_DIR, "persist encryption policy" }) #define show_shutdown_mode(type) \ __print_symbolic(type, \ --Quoted 2 lines
2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel