From 43563069e1c1df417d2eed6eca8a22fc6b04691d Mon Sep 17 00:00:00 2001 From: Yongpeng Yang Date: Mon, 21 Oct 2024 12:48:01 +0800 Subject: [PATCH 01/16] f2fs: check curseg->inited before write_sum_page in change_curseg In the __f2fs_init_atgc_curseg->get_atssr_segment calling, curseg->segno is NULL_SEGNO, indicating that there is no summary block that needs to be written. Fixes: 093749e296e2 ("f2fs: support age threshold based garbage collection") Signed-off-by: Yongpeng Yang Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 96e7abaf42c8..bda7156a2d9d 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -2926,7 +2926,8 @@ static int change_curseg(struct f2fs_sb_info *sbi, int type) struct f2fs_summary_block *sum_node; struct page *sum_page; - write_sum_page(sbi, curseg->sum_blk, GET_SUM_BLOCK(sbi, curseg->segno)); + if (curseg->inited) + write_sum_page(sbi, curseg->sum_blk, GET_SUM_BLOCK(sbi, curseg->segno)); __set_test_and_inuse(sbi, new_segno); -- 2.51.0 From 84b5bb8bf0f6a78c232a20c2eecdbb8112ac2703 Mon Sep 17 00:00:00 2001 From: Qi Han Date: Fri, 25 Oct 2024 03:18:23 -0600 Subject: [PATCH 02/16] f2fs: modify f2fs_is_checkpoint_ready logic to allow more data to be written with the CP disable When the free segment is used up during CP disable, many write or ioctl operations will get ENOSPC error codes, even if there are still many blocks available. We can reproduce it in the following steps: dd if=/dev/zero of=f2fs.img bs=1M count=65 mkfs.f2fs -f f2fs.img mount f2fs.img f2fs_dir -o checkpoint=disable:10% cd f2fs_dir i=1 ; while [[ $i -lt 50 ]] ; do (file_name=./2M_file$i ; dd \ if=/dev/random of=$file_name bs=1M count=2); i=$((i+1)); done sync i=1 ; while [[ $i -lt 50 ]] ; do (file_name=./2M_file$i ; truncate \ -s 1K $file_name); i=$((i+1)); done sync dd if=/dev/zero of=./file bs=1M count=20 In f2fs_need_SSR() function, it is allowed to use SSR to allocate blocks when CP is disabled, so in f2fs_is_checkpoint_ready function, can we judge the number of invalid blocks when free segment is not enough, and return ENOSPC only if the number of invalid blocks is also not enough. Signed-off-by: Qi Han Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.h | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index 8bf09fb67841..0f706cfc754a 100644 --- a/fs/f2fs/segment.h +++ b/fs/f2fs/segment.h @@ -654,12 +654,30 @@ static inline bool has_enough_free_secs(struct f2fs_sb_info *sbi, return !has_not_enough_free_secs(sbi, freed, needed); } +static inline bool has_enough_free_blks(struct f2fs_sb_info *sbi) +{ + unsigned int total_free_blocks = 0; + unsigned int avail_user_block_count; + + spin_lock(&sbi->stat_lock); + + avail_user_block_count = get_available_block_count(sbi, NULL, true); + total_free_blocks = avail_user_block_count - (unsigned int)valid_user_blocks(sbi); + + spin_unlock(&sbi->stat_lock); + + return total_free_blocks > 0; +} + static inline bool f2fs_is_checkpoint_ready(struct f2fs_sb_info *sbi) { if (likely(!is_sbi_flag_set(sbi, SBI_CP_DISABLED))) return true; if (likely(has_enough_free_secs(sbi, 0, 0))) return true; + if (!f2fs_lfs_mode(sbi) && + likely(has_enough_free_blks(sbi))) + return true; return false; } -- 2.51.0 From e63ce120b41ac5a904758a0231d43c6b328cd8fa Mon Sep 17 00:00:00 2001 From: Andrew Kreimer Date: Sun, 27 Oct 2024 16:06:08 +0200 Subject: [PATCH 03/16] f2fs: fix typos Fix typos: datas -> data. Via codespell. Signed-off-by: Andrew Kreimer Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/debug.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c index 278c8855ac0a..468828288a4a 100644 --- a/fs/f2fs/debug.c +++ b/fs/f2fs/debug.c @@ -694,9 +694,9 @@ static int stat_show(struct seq_file *s, void *v) si->ndirty_node, si->node_pages); seq_printf(s, " - dents: %4d in dirs:%4d (%4d)\n", si->ndirty_dent, si->ndirty_dirs, si->ndirty_all); - seq_printf(s, " - datas: %4d in files:%4d\n", + seq_printf(s, " - data: %4d in files:%4d\n", si->ndirty_data, si->ndirty_files); - seq_printf(s, " - quota datas: %4d in quota files:%4d\n", + seq_printf(s, " - quota data: %4d in quota files:%4d\n", si->ndirty_qdata, si->nquota_files); seq_printf(s, " - meta: %4d in %4d\n", si->ndirty_meta, si->meta_pages); -- 2.51.0 From 0c3a38a4b442893f8baca72e44a2a27d52d6cc75 Mon Sep 17 00:00:00 2001 From: Zeng Heng Date: Wed, 23 Oct 2024 17:48:50 +0800 Subject: [PATCH 04/16] f2fs: Fix not used variable 'index' MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit Fix the following compilation warning: fs/f2fs/data.c:2391:10: warning: variable ‘index’ set but not used [-Wunused-but-set-variable] 2391 | pgoff_t index; Only define and set the variable index when the CONFIG_F2FS_FS_COMPRESSION is enabled. Fixes: db92e6c729d8 ("f2fs: convert f2fs_mpage_readpages() to use folio") Signed-off-by: Zeng Heng Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/data.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index 90fa8ab85194..306b86b0595d 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -2385,10 +2385,10 @@ static int f2fs_mpage_readpages(struct inode *inode, .nr_cpages = 0, }; pgoff_t nc_cluster_idx = NULL_CLUSTER; + pgoff_t index; #endif unsigned nr_pages = rac ? readahead_count(rac) : 1; unsigned max_nr_pages = nr_pages; - pgoff_t index; int ret = 0; map.m_pblk = 0; @@ -2406,9 +2406,9 @@ static int f2fs_mpage_readpages(struct inode *inode, prefetchw(&folio->flags); } +#ifdef CONFIG_F2FS_FS_COMPRESSION index = folio_index(folio); -#ifdef CONFIG_F2FS_FS_COMPRESSION if (!f2fs_compressed_file(inode)) goto read_single_page; -- 2.51.0 From f10a890308a7cd8794e21f646f09827c6cb4bf5d Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 22 Oct 2024 16:36:23 +0800 Subject: [PATCH 05/16] f2fs: fix to avoid potential deadlock in f2fs_record_stop_reason() syzbot reports deadlock issue of f2fs as below: ====================================================== WARNING: possible circular locking dependency detected 6.12.0-rc3-syzkaller-00087-gc964ced77262 #0 Not tainted ------------------------------------------------------ kswapd0/79 is trying to acquire lock: ffff888011824088 (&sbi->sb_lock){++++}-{3:3}, at: f2fs_down_write fs/f2fs/f2fs.h:2199 [inline] ffff888011824088 (&sbi->sb_lock){++++}-{3:3}, at: f2fs_record_stop_reason+0x52/0x1d0 fs/f2fs/super.c:4068 but task is already holding lock: ffff88804bd92610 (sb_internal#2){.+.+}-{0:0}, at: f2fs_evict_inode+0x662/0x15c0 fs/f2fs/inode.c:842 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (sb_internal#2){.+.+}-{0:0}: lock_acquire+0x1ed/0x550 kernel/locking/lockdep.c:5825 percpu_down_read include/linux/percpu-rwsem.h:51 [inline] __sb_start_write include/linux/fs.h:1716 [inline] sb_start_intwrite+0x4d/0x1c0 include/linux/fs.h:1899 f2fs_evict_inode+0x662/0x15c0 fs/f2fs/inode.c:842 evict+0x4e8/0x9b0 fs/inode.c:725 f2fs_evict_inode+0x1a4/0x15c0 fs/f2fs/inode.c:807 evict+0x4e8/0x9b0 fs/inode.c:725 dispose_list fs/inode.c:774 [inline] prune_icache_sb+0x239/0x2f0 fs/inode.c:963 super_cache_scan+0x38c/0x4b0 fs/super.c:223 do_shrink_slab+0x701/0x1160 mm/shrinker.c:435 shrink_slab+0x1093/0x14d0 mm/shrinker.c:662 shrink_one+0x43b/0x850 mm/vmscan.c:4818 shrink_many mm/vmscan.c:4879 [inline] lru_gen_shrink_node mm/vmscan.c:4957 [inline] shrink_node+0x3799/0x3de0 mm/vmscan.c:5937 kswapd_shrink_node mm/vmscan.c:6765 [inline] balance_pgdat mm/vmscan.c:6957 [inline] kswapd+0x1ca3/0x3700 mm/vmscan.c:7226 kthread+0x2f0/0x390 kernel/kthread.c:389 ret_from_fork+0x4b/0x80 arch/x86/kernel/process.c:147 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:244 -> #1 (fs_reclaim){+.+.}-{0:0}: lock_acquire+0x1ed/0x550 kernel/locking/lockdep.c:5825 __fs_reclaim_acquire mm/page_alloc.c:3834 [inline] fs_reclaim_acquire+0x88/0x130 mm/page_alloc.c:3848 might_alloc include/linux/sched/mm.h:318 [inline] prepare_alloc_pages+0x147/0x5b0 mm/page_alloc.c:4493 __alloc_pages_noprof+0x16f/0x710 mm/page_alloc.c:4722 alloc_pages_mpol_noprof+0x3e8/0x680 mm/mempolicy.c:2265 alloc_pages_noprof mm/mempolicy.c:2345 [inline] folio_alloc_noprof+0x128/0x180 mm/mempolicy.c:2352 filemap_alloc_folio_noprof+0xdf/0x500 mm/filemap.c:1010 do_read_cache_folio+0x2eb/0x850 mm/filemap.c:3787 read_mapping_folio include/linux/pagemap.h:1011 [inline] f2fs_commit_super+0x3c0/0x7d0 fs/f2fs/super.c:4032 f2fs_record_stop_reason+0x13b/0x1d0 fs/f2fs/super.c:4079 f2fs_handle_critical_error+0x2ac/0x5c0 fs/f2fs/super.c:4174 f2fs_write_inode+0x35f/0x4d0 fs/f2fs/inode.c:785 write_inode fs/fs-writeback.c:1503 [inline] __writeback_single_inode+0x711/0x10d0 fs/fs-writeback.c:1723 writeback_single_inode+0x1f3/0x660 fs/fs-writeback.c:1779 sync_inode_metadata+0xc4/0x120 fs/fs-writeback.c:2849 f2fs_release_file+0xa8/0x100 fs/f2fs/file.c:1941 __fput+0x23f/0x880 fs/file_table.c:431 task_work_run+0x24f/0x310 kernel/task_work.c:228 resume_user_mode_work include/linux/resume_user_mode.h:50 [inline] exit_to_user_mode_loop kernel/entry/common.c:114 [inline] exit_to_user_mode_prepare include/linux/entry-common.h:328 [inline] __syscall_exit_to_user_mode_work kernel/entry/common.c:207 [inline] syscall_exit_to_user_mode+0x168/0x370 kernel/entry/common.c:218 do_syscall_64+0x100/0x230 arch/x86/entry/common.c:89 entry_SYSCALL_64_after_hwframe+0x77/0x7f -> #0 (&sbi->sb_lock){++++}-{3:3}: check_prev_add kernel/locking/lockdep.c:3161 [inline] check_prevs_add kernel/locking/lockdep.c:3280 [inline] validate_chain+0x18ef/0x5920 kernel/locking/lockdep.c:3904 __lock_acquire+0x1384/0x2050 kernel/locking/lockdep.c:5202 lock_acquire+0x1ed/0x550 kernel/locking/lockdep.c:5825 down_write+0x99/0x220 kernel/locking/rwsem.c:1577 f2fs_down_write fs/f2fs/f2fs.h:2199 [inline] f2fs_record_stop_reason+0x52/0x1d0 fs/f2fs/super.c:4068 f2fs_handle_critical_error+0x2ac/0x5c0 fs/f2fs/super.c:4174 f2fs_evict_inode+0xa61/0x15c0 fs/f2fs/inode.c:883 evict+0x4e8/0x9b0 fs/inode.c:725 f2fs_evict_inode+0x1a4/0x15c0 fs/f2fs/inode.c:807 evict+0x4e8/0x9b0 fs/inode.c:725 dispose_list fs/inode.c:774 [inline] prune_icache_sb+0x239/0x2f0 fs/inode.c:963 super_cache_scan+0x38c/0x4b0 fs/super.c:223 do_shrink_slab+0x701/0x1160 mm/shrinker.c:435 shrink_slab+0x1093/0x14d0 mm/shrinker.c:662 shrink_one+0x43b/0x850 mm/vmscan.c:4818 shrink_many mm/vmscan.c:4879 [inline] lru_gen_shrink_node mm/vmscan.c:4957 [inline] shrink_node+0x3799/0x3de0 mm/vmscan.c:5937 kswapd_shrink_node mm/vmscan.c:6765 [inline] balance_pgdat mm/vmscan.c:6957 [inline] kswapd+0x1ca3/0x3700 mm/vmscan.c:7226 kthread+0x2f0/0x390 kernel/kthread.c:389 ret_from_fork+0x4b/0x80 arch/x86/kernel/process.c:147 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:244 other info that might help us debug this: Chain exists of: &sbi->sb_lock --> fs_reclaim --> sb_internal#2 Possible unsafe locking scenario: CPU0 CPU1 ---- ---- rlock(sb_internal#2); lock(fs_reclaim); lock(sb_internal#2); lock(&sbi->sb_lock); Root cause is there will be potential deadlock in between below tasks: Thread A Kswapd - f2fs_ioc_commit_atomic_write - mnt_want_write_file -- down_read lock A - balance_pgdat - __fs_reclaim_acquire -- lock B - shrink_node - prune_icache_sb - dispose_list - f2fs_evict_inode - sb_start_intwrite -- down_read lock A - f2fs_do_sync_file - f2fs_write_inode - f2fs_handle_critical_error - f2fs_record_stop_reason - f2fs_commit_super - read_mapping_folio - filemap_alloc_folio_noprof - fs_reclaim_acquire -- lock B Both threads try to acquire read lock of lock A, then its upcoming write lock grabber will trigger deadlock. Let's always create an asynchronous task in f2fs_handle_critical_error() rather than calling f2fs_record_stop_reason() synchronously to avoid this potential deadlock issue. Fixes: b62e71be2110 ("f2fs: support errors=remount-ro|continue|panic mountoption") Reported-by: syzbot+be4a9983e95a5e25c8d3@syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/6704d667.050a0220.1e4d62.0081.GAE@google.com Signed-off-by: Chao Yu Reviewed-by: Daejun Park Signed-off-by: Jaegeuk Kim --- fs/f2fs/checkpoint.c | 2 +- fs/f2fs/f2fs.h | 3 +-- fs/f2fs/super.c | 13 +++++++------ 3 files changed, 9 insertions(+), 9 deletions(-) diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c index 7f76460b721f..efda9a022981 100644 --- a/fs/f2fs/checkpoint.c +++ b/fs/f2fs/checkpoint.c @@ -32,7 +32,7 @@ void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, f2fs_build_fault_attr(sbi, 0, 0); if (!end_io) f2fs_flush_merged_writes(sbi); - f2fs_handle_critical_error(sbi, reason, end_io); + f2fs_handle_critical_error(sbi, reason); } /* diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 3c6f3cce5779..0ac553627771 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -3637,8 +3637,7 @@ int f2fs_quota_sync(struct super_block *sb, int type); loff_t max_file_blocks(struct inode *inode); void f2fs_quota_off_umount(struct super_block *sb); void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); -void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, - bool irq_context); +void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason); void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c index 80a53dbf1c38..49519439b770 100644 --- a/fs/f2fs/super.c +++ b/fs/f2fs/super.c @@ -4159,8 +4159,7 @@ static bool system_going_down(void) || system_state == SYSTEM_RESTART; } -void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, - bool irq_context) +void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason) { struct super_block *sb = sbi->sb; bool shutdown = reason == STOP_CP_REASON_SHUTDOWN; @@ -4172,10 +4171,12 @@ void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, if (!f2fs_hw_is_readonly(sbi)) { save_stop_reason(sbi, reason); - if (irq_context && !shutdown) - schedule_work(&sbi->s_error_work); - else - f2fs_record_stop_reason(sbi); + /* + * always create an asynchronous task to record stop_reason + * in order to avoid potential deadlock when running into + * f2fs_record_stop_reason() synchronously. + */ + schedule_work(&sbi->s_error_work); } /* -- 2.51.0 From 9395fb09e897dcd96f601715e63018fc5c69cf03 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 22 Oct 2024 10:01:49 +0800 Subject: [PATCH 06/16] f2fs: fix to parse temperature correctly in f2fs_get_segment_temp() In __get_segment_type(), __get_segment_type_6() may return CURSEG_COLD_DATA_PINNED or CURSEG_ALL_DATA_ATGC log type, but following f2fs_get_segment_temp() can only handle persistent log type, fix it. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/f2fs.h | 5 +++-- fs/f2fs/file.c | 4 ++-- fs/f2fs/segment.c | 33 +++++++++++++++++++++++++-------- fs/f2fs/segment.h | 4 ---- 4 files changed, 30 insertions(+), 16 deletions(-) diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 0ac553627771..c13812dc7d3b 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -1019,7 +1019,7 @@ static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) -enum { +enum log_type { CURSEG_HOT_DATA = 0, /* directory entry blocks */ CURSEG_WARM_DATA, /* data blocks */ CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ @@ -3758,7 +3758,8 @@ void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, block_t old_addr, block_t new_addr, unsigned char version, bool recover_curseg, bool recover_newaddr); -int f2fs_get_segment_temp(int seg_type); +enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, + enum log_type seg_type); int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, block_t old_blkaddr, block_t *new_blkaddr, struct f2fs_summary *sum, int type, diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 75a8b22da664..0e7a0195eca8 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -4858,8 +4858,8 @@ static void f2fs_dio_write_submit_io(const struct iomap_iter *iter, { struct inode *inode = iter->inode; struct f2fs_sb_info *sbi = F2FS_I_SB(inode); - int seg_type = f2fs_rw_hint_to_seg_type(sbi, inode->i_write_hint); - enum temp_type temp = f2fs_get_segment_temp(seg_type); + enum log_type type = f2fs_rw_hint_to_seg_type(sbi, inode->i_write_hint); + enum temp_type temp = f2fs_get_segment_temp(sbi, type); bio->bi_write_hint = f2fs_io_type_to_rw_hint(sbi, DATA, temp); submit_bio(bio); diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index bda7156a2d9d..d8a0c50b8dfc 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -3583,18 +3583,35 @@ static int __get_segment_type_6(struct f2fs_io_info *fio) } } -int f2fs_get_segment_temp(int seg_type) +enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, + enum log_type type) { - if (IS_HOT(seg_type)) - return HOT; - else if (IS_WARM(seg_type)) - return WARM; - return COLD; + struct curseg_info *curseg = CURSEG_I(sbi, type); + enum temp_type temp = COLD; + + switch (curseg->seg_type) { + case CURSEG_HOT_NODE: + case CURSEG_HOT_DATA: + temp = HOT; + break; + case CURSEG_WARM_NODE: + case CURSEG_WARM_DATA: + temp = WARM; + break; + case CURSEG_COLD_NODE: + case CURSEG_COLD_DATA: + temp = COLD; + break; + default: + f2fs_bug_on(sbi, 1); + } + + return temp; } static int __get_segment_type(struct f2fs_io_info *fio) { - int type = 0; + enum log_type type = CURSEG_HOT_DATA; switch (F2FS_OPTION(fio->sbi).active_logs) { case 2: @@ -3610,7 +3627,7 @@ static int __get_segment_type(struct f2fs_io_info *fio) f2fs_bug_on(fio->sbi, true); } - fio->temp = f2fs_get_segment_temp(type); + fio->temp = f2fs_get_segment_temp(fio->sbi, type); return type; } diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index 0f706cfc754a..d440fdb83ca4 100644 --- a/fs/f2fs/segment.h +++ b/fs/f2fs/segment.h @@ -34,10 +34,6 @@ static inline void sanity_check_seg_type(struct f2fs_sb_info *sbi, f2fs_bug_on(sbi, seg_type >= NR_PERSISTENT_LOG); } -#define IS_HOT(t) ((t) == CURSEG_HOT_NODE || (t) == CURSEG_HOT_DATA) -#define IS_WARM(t) ((t) == CURSEG_WARM_NODE || (t) == CURSEG_WARM_DATA) -#define IS_COLD(t) ((t) == CURSEG_COLD_NODE || (t) == CURSEG_COLD_DATA) - #define IS_CURSEG(sbi, seg) \ (((seg) == CURSEG_I(sbi, CURSEG_HOT_DATA)->segno) || \ ((seg) == CURSEG_I(sbi, CURSEG_WARM_DATA)->segno) || \ -- 2.51.0 From 296b8cb34e65fa93382cf919be5a056f719c9a26 Mon Sep 17 00:00:00 2001 From: Zhiguo Niu Date: Tue, 29 Oct 2024 11:12:49 +0800 Subject: [PATCH 07/16] f2fs: fix to avoid use GC_AT when setting gc_mode as GC_URGENT_LOW or GC_URGENT_MID If gc_mode is set to GC_URGENT_LOW or GC_URGENT_MID, cost benefit GC approach should be used, but if ATGC is enabled at the same time, Age-threshold approach will be selected, which can only do amount of GC and it is much less than the numbers of CB approach. some traces: f2fs_gc-254:48-396 [007] ..... 2311600.684028: f2fs_gc_begin: dev = (254,48), gc_type = Background GC, no_background_GC = 0, nr_free_secs = 0, nodes = 1053, dents = 2, imeta = 18, free_sec:44898, free_seg:44898, rsv_seg:239, prefree_seg:0 f2fs_gc-254:48-396 [007] ..... 2311600.684527: f2fs_get_victim: dev = (254,48), type = No TYPE, policy = (Background GC, LFS-mode, Age-threshold), victim = 10, cost = 4294364975, ofs_unit = 1, pre_victim_secno = -1, prefree = 0, free = 44898 f2fs_gc-254:48-396 [007] ..... 2311600.714835: f2fs_gc_end: dev = (254,48), ret = 0, seg_freed = 0, sec_freed = 0, nodes = 1562, dents = 2, imeta = 18, free_sec:44898, free_seg:44898, rsv_seg:239, prefree_seg:0 f2fs_gc-254:48-396 [007] ..... 2311600.714843: f2fs_background_gc: dev = (254,48), wait_ms = 50, prefree = 0, free = 44898 f2fs_gc-254:48-396 [007] ..... 2311600.771785: f2fs_gc_begin: dev = (254,48), gc_type = Background GC, no_background_GC = 0, nr_free_secs = 0, nodes = 1562, dents = 2, imeta = 18, free_sec:44898, free_seg:44898, rsv_seg:239, prefree_seg: f2fs_gc-254:48-396 [007] ..... 2311600.772275: f2fs_gc_end: dev = (254,48), ret = -61, seg_freed = 0, sec_freed = 0, nodes = 1562, dents = 2, imeta = 18, free_sec:44898, free_seg:44898, rsv_seg:239, prefree_seg:0 Fixes: 0e5e81114de1 ("f2fs: add GC_URGENT_LOW mode in gc_urgent") Fixes: d98af5f45520 ("f2fs: introduce gc_urgent_mid mode") Signed-off-by: Zhiguo Niu Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- Documentation/ABI/testing/sysfs-fs-f2fs | 7 +++++-- fs/f2fs/gc.c | 2 ++ 2 files changed, 7 insertions(+), 2 deletions(-) diff --git a/Documentation/ABI/testing/sysfs-fs-f2fs b/Documentation/ABI/testing/sysfs-fs-f2fs index fdedf1ea944b..513296bb6f29 100644 --- a/Documentation/ABI/testing/sysfs-fs-f2fs +++ b/Documentation/ABI/testing/sysfs-fs-f2fs @@ -311,10 +311,13 @@ Description: Do background GC aggressively when set. Set to 0 by default. GC approach and turns SSR mode on. gc urgent low(2): lowers the bar of checking I/O idling in order to process outstanding discard commands and GC a - little bit aggressively. uses cost benefit GC approach. + little bit aggressively. always uses cost benefit GC approach, + and will override age-threshold GC approach if ATGC is enabled + at the same time. gc urgent mid(3): does GC forcibly in a period of given gc_urgent_sleep_time and executes a mid level of I/O idling check. - uses cost benefit GC approach. + always uses cost benefit GC approach, and will override + age-threshold GC approach if ATGC is enabled at the same time. What: /sys/fs/f2fs//gc_urgent_sleep_time Date: August 2017 diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c index e40bdd12e36d..3e1b6d2ff3a7 100644 --- a/fs/f2fs/gc.c +++ b/fs/f2fs/gc.c @@ -257,6 +257,8 @@ static int select_gc_type(struct f2fs_sb_info *sbi, int gc_type) switch (sbi->gc_mode) { case GC_IDLE_CB: + case GC_URGENT_LOW: + case GC_URGENT_MID: gc_mode = GC_CB; break; case GC_IDLE_GREEDY: -- 2.51.0 From cffaa0976fcc941a618951b56745a817befa8f91 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 29 Oct 2024 08:57:03 +0800 Subject: [PATCH 08/16] f2fs: clean up opened code w/ {get,set}_nid() Just cleanup, no logic change. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index af36c6d6542b..7d904f2bd5b6 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1066,7 +1066,7 @@ static int truncate_partial_nodes(struct dnode_of_data *dn, int i; int idx = depth - 2; - nid[0] = le32_to_cpu(ri->i_nid[offset[0] - NODE_DIR1_BLOCK]); + nid[0] = get_nid(dn->inode_page, offset[0], true); if (!nid[0]) return 0; @@ -1177,7 +1177,7 @@ int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from) skip_partial: while (cont) { - dn.nid = le32_to_cpu(ri->i_nid[offset[0] - NODE_DIR1_BLOCK]); + dn.nid = get_nid(page, offset[0], true); switch (offset[0]) { case NODE_DIR1_BLOCK: case NODE_DIR2_BLOCK: @@ -1209,13 +1209,10 @@ skip_partial: } if (err < 0) goto fail; - if (offset[1] == 0 && - ri->i_nid[offset[0] - NODE_DIR1_BLOCK]) { + if (offset[1] == 0 && get_nid(page, offset[0], true)) { lock_page(page); BUG_ON(page->mapping != NODE_MAPPING(sbi)); - f2fs_wait_on_page_writeback(page, NODE, true, true); - ri->i_nid[offset[0] - NODE_DIR1_BLOCK] = 0; - set_page_dirty(page); + set_nid(page, offset[0], 0, true); unlock_page(page); } offset[1] = 0; -- 2.51.0 From 1df2bc3c8252261860787a02f61797a914b99163 Mon Sep 17 00:00:00 2001 From: LongPing Wei Date: Tue, 29 Oct 2024 18:49:07 +0800 Subject: [PATCH 09/16] f2fs: clean up the unused variable additional_reserved_segments additional_reserved_segments was introduced by commit 300a842937fb ("f2fs: fix to reserve space for IO align feature"), and its initialization was deleted by commit 87161a2b0aed ("f2fs: deprecate io_bits"). Signed-off-by: LongPing Wei Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/f2fs.h | 1 - fs/f2fs/segment.h | 3 +-- fs/f2fs/sysfs.c | 4 +--- 3 files changed, 2 insertions(+), 6 deletions(-) diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index c13812dc7d3b..119706dbaefa 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -1064,7 +1064,6 @@ struct f2fs_sm_info { unsigned int segment_count; /* total # of segments */ unsigned int main_segments; /* # of segments in main area */ unsigned int reserved_segments; /* # of reserved segments */ - unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ unsigned int ovp_segments; /* # of overprovision segments */ /* a threshold to reclaim prefree segments */ diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index d440fdb83ca4..943be4f1d6d2 100644 --- a/fs/f2fs/segment.h +++ b/fs/f2fs/segment.h @@ -522,8 +522,7 @@ static inline unsigned int free_segments(struct f2fs_sb_info *sbi) static inline unsigned int reserved_segments(struct f2fs_sb_info *sbi) { - return SM_I(sbi)->reserved_segments + - SM_I(sbi)->additional_reserved_segments; + return SM_I(sbi)->reserved_segments; } static inline unsigned int free_sections(struct f2fs_sb_info *sbi) diff --git a/fs/f2fs/sysfs.c b/fs/f2fs/sysfs.c index e51304bc65ea..bdbf24db667b 100644 --- a/fs/f2fs/sysfs.c +++ b/fs/f2fs/sysfs.c @@ -501,9 +501,7 @@ out: if (a->struct_type == RESERVED_BLOCKS) { spin_lock(&sbi->stat_lock); if (t > (unsigned long)(sbi->user_block_count - - F2FS_OPTION(sbi).root_reserved_blocks - - SEGS_TO_BLKS(sbi, - SM_I(sbi)->additional_reserved_segments))) { + F2FS_OPTION(sbi).root_reserved_blocks)) { spin_unlock(&sbi->stat_lock); return -EINVAL; } -- 2.51.0 From 51d3d952c5084393d89cce0b951bb5f18eb97bb1 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Wed, 30 Oct 2024 16:17:01 +0800 Subject: [PATCH 10/16] f2fs: fix to convert log type to segment data type correctly This patch introduces a new helper log_type_to_seg_type() to convert log type to segment data type, and uses it to clean up opened codes in build_curseg(), and also it fixes to convert log type before use in do_write_page(). Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 36 ++++++++++++++++++++++++++++-------- 1 file changed, 28 insertions(+), 8 deletions(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index d8a0c50b8dfc..edf2a74207b3 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -3812,10 +3812,35 @@ void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, } } +static int log_type_to_seg_type(enum log_type type) +{ + int seg_type = CURSEG_COLD_DATA; + + switch (type) { + case CURSEG_HOT_DATA: + case CURSEG_WARM_DATA: + case CURSEG_COLD_DATA: + case CURSEG_HOT_NODE: + case CURSEG_WARM_NODE: + case CURSEG_COLD_NODE: + seg_type = (int)type; + break; + case CURSEG_COLD_DATA_PINNED: + case CURSEG_ALL_DATA_ATGC: + seg_type = CURSEG_COLD_DATA; + break; + default: + break; + } + return seg_type; +} + static void do_write_page(struct f2fs_summary *sum, struct f2fs_io_info *fio) { - int type = __get_segment_type(fio); - bool keep_order = (f2fs_lfs_mode(fio->sbi) && type == CURSEG_COLD_DATA); + enum log_type type = __get_segment_type(fio); + int seg_type = log_type_to_seg_type(type); + bool keep_order = (f2fs_lfs_mode(fio->sbi) && + seg_type == CURSEG_COLD_DATA); if (keep_order) f2fs_down_read(&fio->sbi->io_order_lock); @@ -4797,12 +4822,7 @@ static int build_curseg(struct f2fs_sb_info *sbi) sizeof(struct f2fs_journal), GFP_KERNEL); if (!array[i].journal) return -ENOMEM; - if (i < NR_PERSISTENT_LOG) - array[i].seg_type = CURSEG_HOT_DATA + i; - else if (i == CURSEG_COLD_DATA_PINNED) - array[i].seg_type = CURSEG_COLD_DATA; - else if (i == CURSEG_ALL_DATA_ATGC) - array[i].seg_type = CURSEG_COLD_DATA; + array[i].seg_type = log_type_to_seg_type(i); reset_curseg_fields(&array[i]); } return restore_curseg_summaries(sbi); -- 2.51.0 From 744e66cb8779fcbdbe20b0ec3235ef1ee9815261 Mon Sep 17 00:00:00 2001 From: Zhiguo Niu Date: Thu, 31 Oct 2024 13:59:52 +0800 Subject: [PATCH 11/16] f2fs: remove redundant atomic file check in defragment f2fs_is_atomic_file(inode) is checked in f2fs_defragment_range, so remove the redundant checking in f2fs_ioc_defragment. Signed-off-by: Zhiguo Niu Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 0e7a0195eca8..b619f7e55640 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -2883,7 +2883,7 @@ static int f2fs_ioc_defragment(struct file *filp, unsigned long arg) if (!capable(CAP_SYS_ADMIN)) return -EPERM; - if (!S_ISREG(inode->i_mode) || f2fs_is_atomic_file(inode)) + if (!S_ISREG(inode->i_mode)) return -EINVAL; if (f2fs_readonly(sbi->sb)) -- 2.51.0 From a7a7c1d423a6351a6541e95c797da5358e5ad1ea Mon Sep 17 00:00:00 2001 From: Xiuhong Wang Date: Tue, 29 Oct 2024 14:15:35 +0800 Subject: [PATCH 12/16] f2fs: fix fiemap failure issue when page size is 16KB After enable 16K page size, an infinite loop may occur in fiemap (fm_length=UINT64_MAX) on a file, such as the 16KB scratch.img during the remount operation in Android. The condition for whether fiemap continues to map is to check whether the number of bytes corresponding to the next map.m_lblk exceeds blks_to_bytes(inode,max_inode_blocks(inode)) if there are HOLE. The latter does not take into account the maximum size of a file with 16KB page size, so the loop cannot be jumped out. The following is the fail trace: When f2fs_map_blocks reaches map.m_lblk=3936, it needs to go to the first direct node block, so the map is 3936 + 4090 = 8026, The next map is the second direct node block, that is, 8026 + 4090 = 12116, The next map is the first indirect node block, that is, 12116 + 4090 * 4090 = 16740216, The next map is the second indirect node block, that is, 16740216 + 4090 * 4090 = 33468316, The next map is the first double indirect node block, that is, 33468316 + 4090 * 4090 * 4090 = 68451397316 Since map.m_lblk represents the address of a block, which is 32 bits, truncation will occur, that is, 68451397316 becomes 4026887876, and the number of bytes corresponding to the block number does not exceed blks_to_bytes(inode,max_inode_blocks(inode)), so the loop will not be jumped out. The next time, it will be considered that it should still be a double indirect node block, that is, 4026887876 + 4090 * 4090 * 4090 = 72444816876, which will be truncated to 3725340140, and the loop will not be jumped out. 156.374871: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 0, start blkaddr = 0x8e00, len = 0x200, flags = 2,seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.374916: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 512, start blkaddr = 0x0, len = 0x0, flags = 0 , seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.374920: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 513, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 ...... 156.385747: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 3935, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385752: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 3936, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385755: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 8026, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385758: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 12116, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385761: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 16740216, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385764: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 33468316, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385767: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 4026887876, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385770: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 3725340140, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385772: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 4026887876, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 156.385775: f2fs_map_blocks: dev = (254,57), ino = 7449, file offset = 3725340140, start blkaddr = 0x0, len = 0x0, flags = 0, seg_type = 8, may_create = 0, multidevice = 0, flag = 1, err = 0 Commit a6a010f5def5 ("f2fs: Restrict max filesize for 16K f2fs") has set the maximum allowed file size to (U32_MAX + 1) * F2FS_BLKSIZE, so max_file_blocks should be used here to limit it, that is, maxbytes defined above. And the max_inode_blocks function is not called by other functions except here, so cleanup it. Signed-off-by: Xiuhong Wang Signed-off-by: Zhiguo Niu Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/data.c | 22 +--------------------- 1 file changed, 1 insertion(+), 21 deletions(-) diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index 306b86b0595d..402e00d54c0b 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -1901,25 +1901,6 @@ static int f2fs_xattr_fiemap(struct inode *inode, return (err < 0 ? err : 0); } -static loff_t max_inode_blocks(struct inode *inode) -{ - loff_t result = ADDRS_PER_INODE(inode); - loff_t leaf_count = ADDRS_PER_BLOCK(inode); - - /* two direct node blocks */ - result += (leaf_count * 2); - - /* two indirect node blocks */ - leaf_count *= NIDS_PER_BLOCK; - result += (leaf_count * 2); - - /* one double indirect node block */ - leaf_count *= NIDS_PER_BLOCK; - result += leaf_count; - - return result; -} - int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, u64 start, u64 len) { @@ -1992,8 +1973,7 @@ next: if (!compr_cluster && !(map.m_flags & F2FS_MAP_FLAGS)) { start_blk = next_pgofs; - if (blks_to_bytes(inode, start_blk) < blks_to_bytes(inode, - max_inode_blocks(inode))) + if (blks_to_bytes(inode, start_blk) < maxbytes) goto prep_next; flags |= FIEMAP_EXTENT_LAST; -- 2.51.0 From 7b0033dbc48340a1c1c3f12448ba17d6587ca092 Mon Sep 17 00:00:00 2001 From: Long Li Date: Mon, 4 Nov 2024 10:05:42 +0800 Subject: [PATCH 13/16] f2fs: fix race in concurrent f2fs_stop_gc_thread In my test case, concurrent calls to f2fs shutdown report the following stack trace: Oops: general protection fault, probably for non-canonical address 0xc6cfff63bb5513fc: 0000 [#1] PREEMPT SMP PTI CPU: 0 UID: 0 PID: 678 Comm: f2fs_rep_shutdo Not tainted 6.12.0-rc5-next-20241029-g6fb2fa9805c5-dirty #85 Call Trace: ? show_regs+0x8b/0xa0 ? __die_body+0x26/0xa0 ? die_addr+0x54/0x90 ? exc_general_protection+0x24b/0x5c0 ? asm_exc_general_protection+0x26/0x30 ? kthread_stop+0x46/0x390 f2fs_stop_gc_thread+0x6c/0x110 f2fs_do_shutdown+0x309/0x3a0 f2fs_ioc_shutdown+0x150/0x1c0 __f2fs_ioctl+0xffd/0x2ac0 f2fs_ioctl+0x76/0xe0 vfs_ioctl+0x23/0x60 __x64_sys_ioctl+0xce/0xf0 x64_sys_call+0x2b1b/0x4540 do_syscall_64+0xa7/0x240 entry_SYSCALL_64_after_hwframe+0x76/0x7e The root cause is a race condition in f2fs_stop_gc_thread() called from different f2fs shutdown paths: [CPU0] [CPU1] ---------------------- ----------------------- f2fs_stop_gc_thread f2fs_stop_gc_thread gc_th = sbi->gc_thread gc_th = sbi->gc_thread kfree(gc_th) sbi->gc_thread = NULL < gc_th != NULL > kthread_stop(gc_th->f2fs_gc_task) //UAF The commit c7f114d864ac ("f2fs: fix to avoid use-after-free in f2fs_stop_gc_thread()") attempted to fix this issue by using a read semaphore to prevent races between shutdown and remount threads, but it fails to prevent all race conditions. Fix it by converting to write lock of s_umount in f2fs_do_shutdown(). Fixes: 7950e9ac638e ("f2fs: stop gc/discard thread after fs shutdown") Signed-off-by: Long Li Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index b619f7e55640..c0818a4f87c1 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -2365,9 +2365,12 @@ int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, if (readonly) goto out; - /* grab sb->s_umount to avoid racing w/ remount() */ + /* + * grab sb->s_umount to avoid racing w/ remount() and other shutdown + * paths. + */ if (need_lock) - down_read(&sbi->sb->s_umount); + down_write(&sbi->sb->s_umount); f2fs_stop_gc_thread(sbi); f2fs_stop_discard_thread(sbi); @@ -2376,7 +2379,7 @@ int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, clear_opt(sbi, DISCARD); if (need_lock) - up_read(&sbi->sb->s_umount); + up_write(&sbi->sb->s_umount); f2fs_update_time(sbi, REQ_TIME); out: -- 2.51.0 From 5dd00ebda337b9295e7027691fa70540da369ff2 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Mon, 4 Nov 2024 09:35:51 +0800 Subject: [PATCH 14/16] f2fs: fix to map blocks correctly for direct write f2fs_map_blocks() supports to map continuous holes or preallocated address, we should avoid setting F2FS_MAP_MAPPED for these cases only, otherwise, it may fail f2fs_iomap_begin(), and make direct write fallbacking to use buffered IO and flush, result in performance regression. Fixes: 9f0f6bf42714 ("f2fs: support to map continuous holes or preallocated address") Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-lkp/202409122103.e45aa13b-oliver.sang@intel.com Cc: Cyril Hrubis Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/data.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index 402e00d54c0b..b33aca24b9ef 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -1676,7 +1676,8 @@ next_block: /* reserved delalloc block should be mapped for fiemap. */ if (blkaddr == NEW_ADDR) map->m_flags |= F2FS_MAP_DELALLOC; - if (flag != F2FS_GET_BLOCK_DIO || !is_hole) + /* DIO READ and hole case, should not map the blocks. */ + if (!(flag == F2FS_GET_BLOCK_DIO && is_hole && !map->m_may_create)) map->m_flags |= F2FS_MAP_MAPPED; map->m_pblk = blkaddr; -- 2.51.0 From 26e6f59d0bbaac76fa3413462d780bd2b5f9f653 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Mon, 4 Nov 2024 09:50:16 +0800 Subject: [PATCH 15/16] f2fs: fix to avoid forcing direct write to use buffered IO on inline_data inode Jinsu Lee reported a performance regression issue, after commit 5c8764f8679e ("f2fs: fix to force buffered IO on inline_data inode"), we forced direct write to use buffered IO on inline_data inode, it will cause performace regression due to memory copy and data flush. It's fine to not force direct write to use buffered IO, as it can convert inline inode before committing direct write IO. Fixes: 5c8764f8679e ("f2fs: fix to force buffered IO on inline_data inode") Reported-by: Jinsu Lee Closes: https://lore.kernel.org/linux-f2fs-devel/af03dd2c-e361-4f80-b2fd-39440766cf6e@kernel.org Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index c0818a4f87c1..cb2172a698be 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -883,7 +883,11 @@ static bool f2fs_force_buffered_io(struct inode *inode, int rw) return true; if (f2fs_compressed_file(inode)) return true; - if (f2fs_has_inline_data(inode)) + /* + * only force direct read to use buffered IO, for direct write, + * it expects inline data conversion before committing IO. + */ + if (f2fs_has_inline_data(inode) && rw == READ) return true; /* disallow direct IO if any of devices has unaligned blksize */ -- 2.51.0 From acff9409dd40beaca2bd982678d222e2740ad84b Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Tue, 12 Nov 2024 01:04:58 +0000 Subject: [PATCH 16/16] Revert "f2fs: remove unreachable lazytime mount option parsing" This reverts commit 54f43a10fa257ad4af02a1d157fefef6ebcfa7dc. The above commit broke the lazytime mount, given mount("/dev/vdb", "/mnt/test", "f2fs", 0, "lazytime"); CC: stable@vger.kernel.org # 6.11+ Signed-off-by: Daniel Rosenberg Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/super.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c index 49519439b770..35c4394e4fc6 100644 --- a/fs/f2fs/super.c +++ b/fs/f2fs/super.c @@ -150,6 +150,8 @@ enum { Opt_mode, Opt_fault_injection, Opt_fault_type, + Opt_lazytime, + Opt_nolazytime, Opt_quota, Opt_noquota, Opt_usrquota, @@ -226,6 +228,8 @@ static match_table_t f2fs_tokens = { {Opt_mode, "mode=%s"}, {Opt_fault_injection, "fault_injection=%u"}, {Opt_fault_type, "fault_type=%u"}, + {Opt_lazytime, "lazytime"}, + {Opt_nolazytime, "nolazytime"}, {Opt_quota, "quota"}, {Opt_noquota, "noquota"}, {Opt_usrquota, "usrquota"}, @@ -922,6 +926,12 @@ static int parse_options(struct super_block *sb, char *options, bool is_remount) f2fs_info(sbi, "fault_type options not supported"); break; #endif + case Opt_lazytime: + sb->s_flags |= SB_LAZYTIME; + break; + case Opt_nolazytime: + sb->s_flags &= ~SB_LAZYTIME; + break; #ifdef CONFIG_QUOTA case Opt_quota: case Opt_usrquota: -- 2.51.0