From 650d3c4e56e1e92ee6e004648c9deb243e5963e0 Mon Sep 17 00:00:00 2001 From: Yunlei He Date: Thu, 22 Dec 2016 11:46:24 +0800 Subject: f2fs: fix a missing discard prefree segments If userspace issue a fstrim with a range not involve prefree segments, it will reuse these segments without discard. This patch fix it. Signed-off-by: Yunlei He Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 0d8802453758..cf98ba7bf645 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -916,9 +916,13 @@ void clear_prefree_segments(struct f2fs_sb_info *sbi, struct cp_control *cpc) dirty_i->nr_dirty[PRE] -= end - start; - if (force || !test_opt(sbi, DISCARD)) + if (!test_opt(sbi, DISCARD)) continue; + if (force && start >= cpc->trim_start && + (end - 1) <= cpc->trim_end) + continue; + if (!test_opt(sbi, LFS) || sbi->segs_per_sec == 1) { f2fs_issue_discard(sbi, START_BLOCK(sbi, start), (end - start) << sbi->log_blocks_per_seg); @@ -2263,8 +2267,12 @@ void flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc) f2fs_bug_on(sbi, sit_i->dirty_sentries); out: if (cpc->reason == CP_DISCARD) { + __u64 trim_start = cpc->trim_start; + for (; cpc->trim_start <= cpc->trim_end; cpc->trim_start++) add_discard_addrs(sbi, cpc); + + cpc->trim_start = trim_start; } mutex_unlock(&sit_i->sentry_lock); -- cgit From 9d52a504db6db9e4e254576130aa867838daff55 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 21 Dec 2016 11:51:32 -0800 Subject: f2fs: reassign new segment for mode=lfs Otherwise we can remain wrong curseg->next_blkoff, resulting in fsck failure. Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index cf98ba7bf645..b6bb6490a640 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1428,9 +1428,6 @@ void allocate_new_segments(struct f2fs_sb_info *sbi) unsigned int old_segno; int i; - if (test_opt(sbi, LFS)) - return; - for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) { curseg = CURSEG_I(sbi, i); old_segno = curseg->segno; -- cgit From 0a595ebaaa6b53a2226d3fee2a2fd616ea5ba378 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 14 Dec 2016 10:12:56 -0800 Subject: f2fs: support IO alignment for DATA and NODE writes This patch implements IO alignment by filling dummy blocks in DATA and NODE write bios. If we can guarantee, for example, 32KB or 64KB for such the IOs, we can eliminate underlying dummy page problem which FTL conducts in order to close MLC or TLC partial written pages. Note that, - it requires "-o mode=lfs". - IO size should be power of 2, not exceed BIO_MAX_PAGES, 256. - read IO is still 4KB. - do checkpoint at fsync, if dummy NODE page was written. Signed-off-by: Jaegeuk Kim --- fs/f2fs/data.c | 55 +++++++++++++++++++++++++++++++++++++++++++++++-- fs/f2fs/f2fs.h | 4 +++- fs/f2fs/segment.c | 9 ++++++-- fs/f2fs/segment.h | 3 +++ fs/f2fs/super.c | 13 +++++++++++- include/linux/f2fs_fs.h | 6 ++++++ 6 files changed, 84 insertions(+), 6 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index a06b2d187aec..12d235f2c771 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -93,6 +93,17 @@ static void f2fs_write_end_io(struct bio *bio) struct page *page = bvec->bv_page; enum count_type type = WB_DATA_TYPE(page); + if (IS_DUMMY_WRITTEN_PAGE(page)) { + set_page_private(page, (unsigned long)NULL); + ClearPagePrivate(page); + unlock_page(page); + mempool_free(page, sbi->write_io_dummy); + + if (unlikely(bio->bi_error)) + f2fs_stop_checkpoint(sbi, true); + continue; + } + fscrypt_pullback_bio_page(&page, true); if (unlikely(bio->bi_error)) { @@ -171,10 +182,42 @@ static inline void __submit_bio(struct f2fs_sb_info *sbi, struct bio *bio, enum page_type type) { if (!is_read_io(bio_op(bio))) { + unsigned int start; + if (f2fs_sb_mounted_blkzoned(sbi->sb) && current->plug && (type == DATA || type == NODE)) blk_finish_plug(current->plug); + + if (type != DATA && type != NODE) + goto submit_io; + + start = bio->bi_iter.bi_size >> F2FS_BLKSIZE_BITS; + start %= F2FS_IO_SIZE(sbi); + + if (start == 0) + goto submit_io; + + /* fill dummy pages */ + for (; start < F2FS_IO_SIZE(sbi); start++) { + struct page *page = + mempool_alloc(sbi->write_io_dummy, + GFP_NOIO | __GFP_ZERO | __GFP_NOFAIL); + f2fs_bug_on(sbi, !page); + + SetPagePrivate(page); + set_page_private(page, (unsigned long)DUMMY_WRITTEN_PAGE); + lock_page(page); + if (bio_add_page(bio, page, PAGE_SIZE, 0) < PAGE_SIZE) + f2fs_bug_on(sbi, 1); + } + /* + * In the NODE case, we lose next block address chain. So, we + * need to do checkpoint in f2fs_sync_file. + */ + if (type == NODE) + set_sbi_flag(sbi, SBI_NEED_CP); } +submit_io: if (is_read_io(bio_op(bio))) trace_f2fs_submit_read_bio(sbi->sb, type, bio); else @@ -319,13 +362,14 @@ int f2fs_submit_page_bio(struct f2fs_io_info *fio) return 0; } -void f2fs_submit_page_mbio(struct f2fs_io_info *fio) +int f2fs_submit_page_mbio(struct f2fs_io_info *fio) { struct f2fs_sb_info *sbi = fio->sbi; enum page_type btype = PAGE_TYPE_OF_BIO(fio->type); struct f2fs_bio_info *io; bool is_read = is_read_io(fio->op); struct page *bio_page; + int err = 0; io = is_read ? &sbi->read_io : &sbi->write_io[btype]; @@ -346,6 +390,12 @@ void f2fs_submit_page_mbio(struct f2fs_io_info *fio) __submit_merged_bio(io); alloc_new: if (io->bio == NULL) { + if ((fio->type == DATA || fio->type == NODE) && + fio->new_blkaddr & F2FS_IO_SIZE_MASK(sbi)) { + err = -EAGAIN; + dec_page_count(sbi, WB_DATA_TYPE(bio_page)); + goto out_fail; + } io->bio = __bio_alloc(sbi, fio->new_blkaddr, BIO_MAX_PAGES, is_read); io->fio = *fio; @@ -359,9 +409,10 @@ alloc_new: io->last_block_in_bio = fio->new_blkaddr; f2fs_trace_ios(fio, 0); - +out_fail: up_write(&io->io_rwsem); trace_f2fs_submit_page_mbio(fio->page, fio); + return err; } static void __set_data_blkaddr(struct dnode_of_data *dn) diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 2da8c3aa0ce5..0d7eaddef4a0 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -792,6 +792,8 @@ struct f2fs_sb_info { struct f2fs_bio_info read_io; /* for read bios */ struct f2fs_bio_info write_io[NR_PAGE_TYPE]; /* for write bios */ struct mutex wio_mutex[NODE + 1]; /* bio ordering for NODE/DATA */ + int write_io_size_bits; /* Write IO size bits */ + mempool_t *write_io_dummy; /* Dummy pages */ /* for checkpoint */ struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ @@ -2174,7 +2176,7 @@ void f2fs_submit_merged_bio_cond(struct f2fs_sb_info *, struct inode *, struct page *, nid_t, enum page_type, int); void f2fs_flush_merged_bios(struct f2fs_sb_info *); int f2fs_submit_page_bio(struct f2fs_io_info *); -void f2fs_submit_page_mbio(struct f2fs_io_info *); +int f2fs_submit_page_mbio(struct f2fs_io_info *); struct block_device *f2fs_target_device(struct f2fs_sb_info *, block_t, struct bio *); int f2fs_target_device_index(struct f2fs_sb_info *, block_t); diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index b6bb6490a640..2e8d12e9cae9 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1604,15 +1604,20 @@ void allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, static void do_write_page(struct f2fs_summary *sum, struct f2fs_io_info *fio) { int type = __get_segment_type(fio->page, fio->type); + int err; if (fio->type == NODE || fio->type == DATA) mutex_lock(&fio->sbi->wio_mutex[fio->type]); - +reallocate: allocate_data_block(fio->sbi, fio->page, fio->old_blkaddr, &fio->new_blkaddr, sum, type); /* writeout dirty page into bdev */ - f2fs_submit_page_mbio(fio); + err = f2fs_submit_page_mbio(fio); + if (err == -EAGAIN) { + fio->old_blkaddr = fio->new_blkaddr; + goto reallocate; + } if (fio->type == NODE || fio->type == DATA) mutex_unlock(&fio->sbi->wio_mutex[fio->type]); diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index 9d44ce83acb2..08f1455c812c 100644 --- a/fs/f2fs/segment.h +++ b/fs/f2fs/segment.h @@ -186,9 +186,12 @@ struct segment_allocation { * the page is atomically written, and it is in inmem_pages list. */ #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) +#define DUMMY_WRITTEN_PAGE ((unsigned long)-2) #define IS_ATOMIC_WRITTEN_PAGE(page) \ (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE) +#define IS_DUMMY_WRITTEN_PAGE(page) \ + (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE) struct inmem_pages { struct list_head list; diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c index 46fd30d8af77..00fc36e49368 100644 --- a/fs/f2fs/super.c +++ b/fs/f2fs/super.c @@ -1763,6 +1763,8 @@ static int f2fs_scan_devices(struct f2fs_sb_info *sbi) FDEV(i).total_segments, FDEV(i).start_blk, FDEV(i).end_blk); } + f2fs_msg(sbi->sb, KERN_INFO, + "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi)); return 0; } @@ -1880,12 +1882,19 @@ try_onemore: if (err) goto free_options; + if (F2FS_IO_SIZE(sbi) > 1) { + sbi->write_io_dummy = + mempool_create_page_pool(F2FS_IO_SIZE(sbi) - 1, 0); + if (!sbi->write_io_dummy) + goto free_options; + } + /* get an inode for meta space */ sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi)); if (IS_ERR(sbi->meta_inode)) { f2fs_msg(sb, KERN_ERR, "Failed to read F2FS meta data inode"); err = PTR_ERR(sbi->meta_inode); - goto free_options; + goto free_io_dummy; } err = get_valid_checkpoint(sbi); @@ -2103,6 +2112,8 @@ free_devices: free_meta_inode: make_bad_inode(sbi->meta_inode); iput(sbi->meta_inode); +free_io_dummy: + mempool_destroy(sbi->write_io_dummy); free_options: destroy_percpu_info(sbi); kfree(options); diff --git a/include/linux/f2fs_fs.h b/include/linux/f2fs_fs.h index cea41a124a80..f0748524ca8c 100644 --- a/include/linux/f2fs_fs.h +++ b/include/linux/f2fs_fs.h @@ -36,6 +36,12 @@ #define F2FS_NODE_INO(sbi) (sbi->node_ino_num) #define F2FS_META_INO(sbi) (sbi->meta_ino_num) +#define F2FS_IO_SIZE(sbi) (1 << (sbi)->write_io_size_bits) /* Blocks */ +#define F2FS_IO_SIZE_KB(sbi) (1 << ((sbi)->write_io_size_bits + 2)) /* KB */ +#define F2FS_IO_SIZE_BYTES(sbi) (1 << ((sbi)->write_io_size_bits + 12)) /* B */ +#define F2FS_IO_SIZE_BITS(sbi) ((sbi)->write_io_size_bits) /* power of 2 */ +#define F2FS_IO_SIZE_MASK(sbi) (F2FS_IO_SIZE(sbi) - 1) + /* This flag is used by node and meta inodes, and by recovery */ #define GFP_F2FS_ZERO (GFP_NOFS | __GFP_ZERO) #define GFP_F2FS_HIGH_ZERO (GFP_NOFS | __GFP_ZERO | __GFP_HIGHMEM) -- cgit From 26a28a0c1eb756ba18bfb1f93309c4b4406b9cd9 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 28 Dec 2016 13:55:09 -0800 Subject: f2fs: show the max number of atomic operations This patch adds to show the max number of atomic operations which are conducting concurrently. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/debug.c | 7 +++++++ fs/f2fs/f2fs.h | 17 +++++++++++++++++ fs/f2fs/file.c | 8 ++++++-- fs/f2fs/segment.c | 1 + 4 files changed, 31 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c index fbd5184140d0..29cdf0c1da1d 100644 --- a/fs/f2fs/debug.c +++ b/fs/f2fs/debug.c @@ -50,6 +50,8 @@ static void update_general_status(struct f2fs_sb_info *sbi) si->ndirty_files = sbi->ndirty_inode[FILE_INODE]; si->ndirty_all = sbi->ndirty_inode[DIRTY_META]; si->inmem_pages = get_pages(sbi, F2FS_INMEM_PAGES); + si->aw_cnt = atomic_read(&sbi->aw_cnt); + si->max_aw_cnt = atomic_read(&sbi->max_aw_cnt); si->nr_wb_cp_data = get_pages(sbi, F2FS_WB_CP_DATA); si->nr_wb_data = get_pages(sbi, F2FS_WB_DATA); si->total_count = (int)sbi->user_block_count / sbi->blocks_per_seg; @@ -256,6 +258,8 @@ static int stat_show(struct seq_file *s, void *v) si->inline_dir); seq_printf(s, " - Orphan Inode: %u\n", si->orphans); + seq_printf(s, " - Atomic write count: %4d (Max. %4d)\n", + si->aw_cnt, si->max_aw_cnt); seq_printf(s, "\nMain area: %d segs, %d secs %d zones\n", si->main_area_segs, si->main_area_sections, si->main_area_zones); @@ -414,6 +418,9 @@ int f2fs_build_stats(struct f2fs_sb_info *sbi) atomic_set(&sbi->inline_dir, 0); atomic_set(&sbi->inplace_count, 0); + atomic_set(&sbi->aw_cnt, 0); + atomic_set(&sbi->max_aw_cnt, 0); + mutex_lock(&f2fs_stat_mutex); list_add_tail(&si->stat_list, &f2fs_stat_list); mutex_unlock(&f2fs_stat_mutex); diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 0d7eaddef4a0..bdcfe2a9b532 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -884,6 +884,8 @@ struct f2fs_sb_info { atomic_t inline_xattr; /* # of inline_xattr inodes */ atomic_t inline_inode; /* # of inline_data inodes */ atomic_t inline_dir; /* # of inline_dentry inodes */ + atomic_t aw_cnt; /* # of atomic writes */ + atomic_t max_aw_cnt; /* max # of atomic writes */ int bg_gc; /* background gc calls */ unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ #endif @@ -2236,6 +2238,7 @@ struct f2fs_stat_info { int total_count, utilization; int bg_gc, nr_wb_cp_data, nr_wb_data; int inline_xattr, inline_inode, inline_dir, orphans; + int aw_cnt, max_aw_cnt; unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; unsigned int bimodal, avg_vblocks; int util_free, util_valid, util_invalid; @@ -2307,6 +2310,17 @@ static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) ((sbi)->block_count[(curseg)->alloc_type]++) #define stat_inc_inplace_blocks(sbi) \ (atomic_inc(&(sbi)->inplace_count)) +#define stat_inc_atomic_write(inode) \ + (atomic_inc(&F2FS_I_SB(inode)->aw_cnt)); +#define stat_dec_atomic_write(inode) \ + (atomic_dec(&F2FS_I_SB(inode)->aw_cnt)); +#define stat_update_max_atomic_write(inode) \ + do { \ + int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \ + int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ + if (cur > max) \ + atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ + } while (0) #define stat_inc_seg_count(sbi, type, gc_type) \ do { \ struct f2fs_stat_info *si = F2FS_STAT(sbi); \ @@ -2360,6 +2374,9 @@ void f2fs_destroy_root_stats(void); #define stat_dec_inline_inode(inode) #define stat_inc_inline_dir(inode) #define stat_dec_inline_dir(inode) +#define stat_inc_atomic_write(inode) +#define stat_dec_atomic_write(inode) +#define stat_update_max_atomic_write(inode) #define stat_inc_seg_type(sbi, curseg) #define stat_inc_block_count(sbi, curseg) #define stat_inc_inplace_blocks(sbi) diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 49f10dce817d..291d2ca53a4c 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1542,6 +1542,8 @@ static int f2fs_ioc_start_atomic_write(struct file *filp) if (ret) clear_inode_flag(inode, FI_ATOMIC_FILE); out: + stat_inc_atomic_write(inode); + stat_update_max_atomic_write(inode); inode_unlock(inode); mnt_drop_write_file(filp); return ret; @@ -1571,9 +1573,11 @@ static int f2fs_ioc_commit_atomic_write(struct file *filp) set_inode_flag(inode, FI_ATOMIC_FILE); goto err_out; } + ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true); + stat_dec_atomic_write(inode); + } else { + ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true); } - - ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true); err_out: inode_unlock(inode); mnt_drop_write_file(filp); diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 2e8d12e9cae9..9cabe935afc7 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -243,6 +243,7 @@ void drop_inmem_pages(struct inode *inode) struct f2fs_inode_info *fi = F2FS_I(inode); clear_inode_flag(inode, FI_ATOMIC_FILE); + stat_dec_atomic_write(inode); mutex_lock(&fi->inmem_lock); __revoke_inmem_pages(inode, &fi->inmem_pages, true, false); -- cgit From 4e6a8d9b224f886362ea6e8f6046b541437c944f Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 29 Dec 2016 14:07:53 -0800 Subject: f2fs: relax async discard commands more This patch relaxes async discard commands to avoid waiting its end_io during checkpoint. Instead of waiting them during checkpoint, it will be done when actually reusing them. Test on initial partition of nvme drive. # time fstrim /mnt/test Before : 6.158s After : 4.822s Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/checkpoint.c | 7 ++----- fs/f2fs/f2fs.h | 4 +++- fs/f2fs/segment.c | 24 +++++++++++++++++++----- fs/f2fs/super.c | 3 +++ 4 files changed, 27 insertions(+), 11 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c index f73ee9534d83..1a9ba69a22ba 100644 --- a/fs/f2fs/checkpoint.c +++ b/fs/f2fs/checkpoint.c @@ -1254,7 +1254,6 @@ int write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc) f2fs_bug_on(sbi, prefree_segments(sbi)); flush_sit_entries(sbi, cpc); clear_prefree_segments(sbi, cpc); - f2fs_wait_all_discard_bio(sbi); unblock_operations(sbi); goto out; } @@ -1273,12 +1272,10 @@ int write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc) /* unlock all the fs_lock[] in do_checkpoint() */ err = do_checkpoint(sbi, cpc); - if (err) { + if (err) release_discard_addrs(sbi); - } else { + else clear_prefree_segments(sbi, cpc); - f2fs_wait_all_discard_bio(sbi); - } unblock_operations(sbi); stat_inc_cp_count(sbi->stat_info); diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index bdcfe2a9b532..e0db895fd84c 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -183,6 +183,8 @@ struct discard_entry { struct bio_entry { struct list_head list; + block_t lstart; + block_t len; struct bio *bio; struct completion event; int error; @@ -2111,7 +2113,7 @@ void destroy_flush_cmd_control(struct f2fs_sb_info *, bool); void invalidate_blocks(struct f2fs_sb_info *, block_t); bool is_checkpointed_data(struct f2fs_sb_info *, block_t); void refresh_sit_entry(struct f2fs_sb_info *, block_t, block_t); -void f2fs_wait_all_discard_bio(struct f2fs_sb_info *); +void f2fs_wait_discard_bio(struct f2fs_sb_info *, block_t); void clear_prefree_segments(struct f2fs_sb_info *, struct cp_control *); void release_discard_addrs(struct f2fs_sb_info *); int npages_for_summary_flush(struct f2fs_sb_info *, bool); diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 9cabe935afc7..44d69f90be2a 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -625,20 +625,23 @@ static void locate_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno) } static struct bio_entry *__add_bio_entry(struct f2fs_sb_info *sbi, - struct bio *bio) + struct bio *bio, block_t lstart, block_t len) { struct list_head *wait_list = &(SM_I(sbi)->wait_list); struct bio_entry *be = f2fs_kmem_cache_alloc(bio_entry_slab, GFP_NOFS); INIT_LIST_HEAD(&be->list); be->bio = bio; + be->lstart = lstart; + be->len = len; init_completion(&be->event); list_add_tail(&be->list, wait_list); return be; } -void f2fs_wait_all_discard_bio(struct f2fs_sb_info *sbi) +/* This should be covered by global mutex, &sit_i->sentry_lock */ +void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) { struct list_head *wait_list = &(SM_I(sbi)->wait_list); struct bio_entry *be, *tmp; @@ -647,7 +650,15 @@ void f2fs_wait_all_discard_bio(struct f2fs_sb_info *sbi) struct bio *bio = be->bio; int err; - wait_for_completion_io(&be->event); + if (!completion_done(&be->event)) { + if ((be->lstart <= blkaddr && + blkaddr < be->lstart + be->len) || + blkaddr == NULL_ADDR) + wait_for_completion_io(&be->event); + else + continue; + } + err = be->error; if (err == -EOPNOTSUPP) err = 0; @@ -675,6 +686,7 @@ static int __f2fs_issue_discard_async(struct f2fs_sb_info *sbi, struct block_device *bdev, block_t blkstart, block_t blklen) { struct bio *bio = NULL; + block_t lblkstart = blkstart; int err; trace_f2fs_issue_discard(sbi->sb, blkstart, blklen); @@ -689,14 +701,14 @@ static int __f2fs_issue_discard_async(struct f2fs_sb_info *sbi, SECTOR_FROM_BLOCK(blklen), GFP_NOFS, 0, &bio); if (!err && bio) { - struct bio_entry *be = __add_bio_entry(sbi, bio); + struct bio_entry *be = __add_bio_entry(sbi, bio, + lblkstart, blklen); bio->bi_private = be; bio->bi_end_io = f2fs_submit_bio_wait_endio; bio->bi_opf |= REQ_SYNC; submit_bio(bio); } - return err; } @@ -1575,6 +1587,8 @@ void allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, *new_blkaddr = NEXT_FREE_BLKADDR(sbi, curseg); + f2fs_wait_discard_bio(sbi, *new_blkaddr); + /* * __add_sum_entry should be resided under the curseg_mutex * because, this function updates a summary entry in the diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c index b7efbd4f6af9..e68cec492f06 100644 --- a/fs/f2fs/super.c +++ b/fs/f2fs/super.c @@ -770,6 +770,9 @@ static void f2fs_put_super(struct super_block *sb) write_checkpoint(sbi, &cpc); } + /* be sure to wait for any on-going discard commands */ + f2fs_wait_discard_bio(sbi, NULL_ADDR); + /* write_checkpoint can update stat informaion */ f2fs_destroy_stats(sbi); -- cgit From 25290fa5591d81767713db304e0d567bf991786f Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 29 Dec 2016 22:06:15 -0800 Subject: f2fs: return fs_trim if there is no candidate If there is no candidate to submit discard command during f2fs_trim_fs, let's return without checkpoint. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/checkpoint.c | 5 +++++ fs/f2fs/f2fs.h | 1 + fs/f2fs/segment.c | 33 ++++++++++++++++++++++++++++----- 3 files changed, 34 insertions(+), 5 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c index 917b5c5053ae..ccea40763d9d 100644 --- a/fs/f2fs/checkpoint.c +++ b/fs/f2fs/checkpoint.c @@ -1249,6 +1249,11 @@ int write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc) /* this is the case of multiple fstrims without any changes */ if (cpc->reason == CP_DISCARD) { + if (!exist_trim_candidates(sbi, cpc)) { + unblock_operations(sbi); + goto out; + } + if (NM_I(sbi)->dirty_nat_cnt == 0 && SIT_I(sbi)->dirty_sentries == 0 && prefree_segments(sbi) == 0) { diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index e0db895fd84c..a4e8e6278a17 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -2119,6 +2119,7 @@ void release_discard_addrs(struct f2fs_sb_info *); int npages_for_summary_flush(struct f2fs_sb_info *, bool); void allocate_new_segments(struct f2fs_sb_info *); int f2fs_trim_fs(struct f2fs_sb_info *, struct fstrim_range *); +bool exist_trim_candidates(struct f2fs_sb_info *, struct cp_control *); struct page *get_sum_page(struct f2fs_sb_info *, unsigned int); void update_meta_page(struct f2fs_sb_info *, void *, block_t); void write_meta_page(struct f2fs_sb_info *, struct page *); diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 44d69f90be2a..8b54b1fafa70 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -834,7 +834,8 @@ done: SM_I(sbi)->nr_discards += end - start; } -static void add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc) +static bool add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc, + bool check_only) { int entries = SIT_VBLOCK_MAP_SIZE / sizeof(unsigned long); int max_blocks = sbi->blocks_per_seg; @@ -848,12 +849,12 @@ static void add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc) int i; if (se->valid_blocks == max_blocks || !f2fs_discard_en(sbi)) - return; + return false; if (!force) { if (!test_opt(sbi, DISCARD) || !se->valid_blocks || SM_I(sbi)->nr_discards >= SM_I(sbi)->max_discards) - return; + return false; } /* SIT_VBLOCK_MAP_SIZE should be multiple of sizeof(unsigned long) */ @@ -871,8 +872,12 @@ static void add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc) && (end - start) < cpc->trim_minlen) continue; + if (check_only) + return true; + __add_discard_entry(sbi, cpc, se, start, end); } + return false; } void release_discard_addrs(struct f2fs_sb_info *sbi) @@ -1453,6 +1458,24 @@ static const struct segment_allocation default_salloc_ops = { .allocate_segment = allocate_segment_by_default, }; +bool exist_trim_candidates(struct f2fs_sb_info *sbi, struct cp_control *cpc) +{ + __u64 trim_start = cpc->trim_start; + bool has_candidate = false; + + mutex_lock(&SIT_I(sbi)->sentry_lock); + for (; cpc->trim_start <= cpc->trim_end; cpc->trim_start++) { + if (add_discard_addrs(sbi, cpc, true)) { + has_candidate = true; + break; + } + } + mutex_unlock(&SIT_I(sbi)->sentry_lock); + + cpc->trim_start = trim_start; + return has_candidate; +} + int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range) { __u64 start = F2FS_BYTES_TO_BLK(range->start); @@ -2249,7 +2272,7 @@ void flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc) /* add discard candidates */ if (cpc->reason != CP_DISCARD) { cpc->trim_start = segno; - add_discard_addrs(sbi, cpc); + add_discard_addrs(sbi, cpc, false); } if (to_journal) { @@ -2287,7 +2310,7 @@ out: __u64 trim_start = cpc->trim_start; for (; cpc->trim_start <= cpc->trim_end; cpc->trim_start++) - add_discard_addrs(sbi, cpc); + add_discard_addrs(sbi, cpc, false); cpc->trim_start = trim_start; } -- cgit From 5fe457430e554a2f5188f13c1a2e36ad845640c5 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Sat, 7 Jan 2017 18:50:26 +0800 Subject: f2fs: introduce FI_ATOMIC_COMMIT This patch introduces a new flag to indicate inode status of doing atomic write committing, so that, we can keep atomic write status for inode during atomic committing, then we can skip GCing pages of atomic write inode, that avoids random GCed datas being mixed with current transaction, so isolation of transaction can be kept. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/data.c | 2 +- fs/f2fs/f2fs.h | 6 ++++++ fs/f2fs/file.c | 11 ++++++----- fs/f2fs/gc.c | 6 ++++++ fs/f2fs/segment.c | 10 +++++++--- 5 files changed, 26 insertions(+), 9 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index ab2008dea921..848d110dc1ca 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -1977,7 +1977,7 @@ static int f2fs_set_data_page_dirty(struct page *page) if (!PageUptodate(page)) SetPageUptodate(page); - if (f2fs_is_atomic_file(inode)) { + if (f2fs_is_atomic_file(inode) && !f2fs_is_commit_atomic_write(inode)) { if (!IS_ATOMIC_WRITTEN_PAGE(page)) { register_inmem_page(inode, page); return 1; diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index a4e8e6278a17..94250a69762a 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -1639,6 +1639,7 @@ enum { FI_UPDATE_WRITE, /* inode has in-place-update data */ FI_NEED_IPU, /* used for ipu per file */ FI_ATOMIC_FILE, /* indicate atomic file */ + FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ FI_VOLATILE_FILE, /* indicate volatile file */ FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ FI_DROP_CACHE, /* drop dirty page cache */ @@ -1828,6 +1829,11 @@ static inline bool f2fs_is_atomic_file(struct inode *inode) return is_inode_flag_set(inode, FI_ATOMIC_FILE); } +static inline bool f2fs_is_commit_atomic_write(struct inode *inode) +{ + return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); +} + static inline bool f2fs_is_volatile_file(struct inode *inode) { return is_inode_flag_set(inode, FI_VOLATILE_FILE); diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 6c335180b9d8..e45522115b1c 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1569,14 +1569,15 @@ static int f2fs_ioc_commit_atomic_write(struct file *filp) goto err_out; if (f2fs_is_atomic_file(inode)) { - clear_inode_flag(inode, FI_ATOMIC_FILE); ret = commit_inmem_pages(inode); - if (ret) { - set_inode_flag(inode, FI_ATOMIC_FILE); + if (ret) goto err_out; - } + ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true); - stat_dec_atomic_write(inode); + if (!ret) { + clear_inode_flag(inode, FI_ATOMIC_FILE); + stat_dec_atomic_write(inode); + } } else { ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true); } diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c index 88bfc3dff496..88e5e7b10ab6 100644 --- a/fs/f2fs/gc.c +++ b/fs/f2fs/gc.c @@ -569,6 +569,9 @@ static void move_encrypted_block(struct inode *inode, block_t bidx, if (!check_valid_map(F2FS_I_SB(inode), segno, off)) goto out; + if (f2fs_is_atomic_file(inode)) + goto out; + set_new_dnode(&dn, inode, NULL, NULL, 0); err = get_dnode_of_data(&dn, bidx, LOOKUP_NODE); if (err) @@ -661,6 +664,9 @@ static void move_data_page(struct inode *inode, block_t bidx, int gc_type, if (!check_valid_map(F2FS_I_SB(inode), segno, off)) goto out; + if (f2fs_is_atomic_file(inode)) + goto out; + if (gc_type == BG_GC) { if (PageWriteback(page)) goto out; diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 8b54b1fafa70..02a8d4ee65eb 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -242,12 +242,12 @@ void drop_inmem_pages(struct inode *inode) { struct f2fs_inode_info *fi = F2FS_I(inode); - clear_inode_flag(inode, FI_ATOMIC_FILE); - stat_dec_atomic_write(inode); - mutex_lock(&fi->inmem_lock); __revoke_inmem_pages(inode, &fi->inmem_pages, true, false); mutex_unlock(&fi->inmem_lock); + + clear_inode_flag(inode, FI_ATOMIC_FILE); + stat_dec_atomic_write(inode); } static int __commit_inmem_pages(struct inode *inode, @@ -316,6 +316,8 @@ int commit_inmem_pages(struct inode *inode) f2fs_balance_fs(sbi, true); f2fs_lock_op(sbi); + set_inode_flag(inode, FI_ATOMIC_COMMIT); + mutex_lock(&fi->inmem_lock); err = __commit_inmem_pages(inode, &revoke_list); if (err) { @@ -337,6 +339,8 @@ int commit_inmem_pages(struct inode *inode) } mutex_unlock(&fi->inmem_lock); + clear_inode_flag(inode, FI_ATOMIC_COMMIT); + f2fs_unlock_op(sbi); return err; } -- cgit From 355e78913c0d57492076d545b6f44b94fec2bf6b Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Sat, 7 Jan 2017 18:51:01 +0800 Subject: f2fs: check in-memory block bitmap This patch adds a mirror for valid block bitmap, and use it to detect in-memory bitmap corruption which may be caused by bit-transition of cache or memory overflow. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 32 ++++++++++++++++++++++++++++++-- fs/f2fs/segment.h | 6 ++++++ 2 files changed, 36 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 02a8d4ee65eb..56a097d22150 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1021,14 +1021,32 @@ static void update_sit_entry(struct f2fs_sb_info *sbi, block_t blkaddr, int del) /* Update valid block bitmap */ if (del > 0) { - if (f2fs_test_and_set_bit(offset, se->cur_valid_map)) + if (f2fs_test_and_set_bit(offset, se->cur_valid_map)) { +#ifdef CONFIG_F2FS_CHECK_FS + if (f2fs_test_and_set_bit(offset, + se->cur_valid_map_mir)) + f2fs_bug_on(sbi, 1); + else + WARN_ON(1); +#else f2fs_bug_on(sbi, 1); +#endif + } if (f2fs_discard_en(sbi) && !f2fs_test_and_set_bit(offset, se->discard_map)) sbi->discard_blks--; } else { - if (!f2fs_test_and_clear_bit(offset, se->cur_valid_map)) + if (!f2fs_test_and_clear_bit(offset, se->cur_valid_map)) { +#ifdef CONFIG_F2FS_CHECK_FS + if (!f2fs_test_and_clear_bit(offset, + se->cur_valid_map_mir)) + f2fs_bug_on(sbi, 1); + else + WARN_ON(1); +#else f2fs_bug_on(sbi, 1); +#endif + } if (f2fs_discard_en(sbi) && f2fs_test_and_clear_bit(offset, se->discard_map)) sbi->discard_blks++; @@ -2357,6 +2375,13 @@ static int build_sit_info(struct f2fs_sb_info *sbi) !sit_i->sentries[start].ckpt_valid_map) return -ENOMEM; +#ifdef CONFIG_F2FS_CHECK_FS + sit_i->sentries[start].cur_valid_map_mir + = kzalloc(SIT_VBLOCK_MAP_SIZE, GFP_KERNEL); + if (!sit_i->sentries[start].cur_valid_map_mir) + return -ENOMEM; +#endif + if (f2fs_discard_en(sbi)) { sit_i->sentries[start].discard_map = kzalloc(SIT_VBLOCK_MAP_SIZE, GFP_KERNEL); @@ -2786,6 +2811,9 @@ static void destroy_sit_info(struct f2fs_sb_info *sbi) if (sit_i->sentries) { for (start = 0; start < MAIN_SEGS(sbi); start++) { kfree(sit_i->sentries[start].cur_valid_map); +#ifdef CONFIG_F2FS_CHECK_FS + kfree(sit_i->sentries[start].cur_valid_map_mir); +#endif kfree(sit_i->sentries[start].ckpt_valid_map); kfree(sit_i->sentries[start].discard_map); } diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index 08f1455c812c..9af95194db06 100644 --- a/fs/f2fs/segment.h +++ b/fs/f2fs/segment.h @@ -164,6 +164,9 @@ struct seg_entry { unsigned int ckpt_valid_blocks:10; /* # of valid blocks last cp */ unsigned int padding:6; /* padding */ unsigned char *cur_valid_map; /* validity bitmap of blocks */ +#ifdef CONFIG_F2FS_CHECK_FS + unsigned char *cur_valid_map_mir; /* mirror of current valid bitmap */ +#endif /* * # of valid blocks and the validity bitmap stored in the the last * checkpoint pack. This information is used by the SSR mode. @@ -320,6 +323,9 @@ static inline void seg_info_from_raw_sit(struct seg_entry *se, se->ckpt_valid_blocks = GET_SIT_VBLOCKS(rs); memcpy(se->cur_valid_map, rs->valid_map, SIT_VBLOCK_MAP_SIZE); memcpy(se->ckpt_valid_map, rs->valid_map, SIT_VBLOCK_MAP_SIZE); +#ifdef CONFIG_F2FS_CHECK_FS + memcpy(se->cur_valid_map_mir, rs->valid_map, SIT_VBLOCK_MAP_SIZE); +#endif se->type = GET_SIT_TYPE(rs); se->mtime = le64_to_cpu(rs->mtime); } -- cgit From ae27d62e6befd3cac4ffa702e644cc52019642e8 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Sat, 7 Jan 2017 18:52:34 +0800 Subject: f2fs: check in-memory sit version bitmap This patch adds a mirror for sit version bitmap, and use it to detect in-memory bitmap corruption which may be caused by bit-transition of cache or memory overflow. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 16 ++++++++++++---- fs/f2fs/segment.h | 18 ++++++++++++++++++ 2 files changed, 30 insertions(+), 4 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 56a097d22150..c8c825574fed 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -2346,7 +2346,7 @@ static int build_sit_info(struct f2fs_sb_info *sbi) struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi); struct sit_info *sit_i; unsigned int sit_segs, start; - char *src_bitmap, *dst_bitmap; + char *src_bitmap; unsigned int bitmap_size; /* allocate memory for SIT information */ @@ -2408,17 +2408,22 @@ static int build_sit_info(struct f2fs_sb_info *sbi) bitmap_size = __bitmap_size(sbi, SIT_BITMAP); src_bitmap = __bitmap_ptr(sbi, SIT_BITMAP); - dst_bitmap = kmemdup(src_bitmap, bitmap_size, GFP_KERNEL); - if (!dst_bitmap) + sit_i->sit_bitmap = kmemdup(src_bitmap, bitmap_size, GFP_KERNEL); + if (!sit_i->sit_bitmap) return -ENOMEM; +#ifdef CONFIG_F2FS_CHECK_FS + sit_i->sit_bitmap_mir = kmemdup(src_bitmap, bitmap_size, GFP_KERNEL); + if (!sit_i->sit_bitmap_mir) + return -ENOMEM; +#endif + /* init SIT information */ sit_i->s_ops = &default_salloc_ops; sit_i->sit_base_addr = le32_to_cpu(raw_super->sit_blkaddr); sit_i->sit_blocks = sit_segs << sbi->log_blocks_per_seg; sit_i->written_valid_blocks = 0; - sit_i->sit_bitmap = dst_bitmap; sit_i->bitmap_size = bitmap_size; sit_i->dirty_sentries = 0; sit_i->sents_per_block = SIT_ENTRY_PER_BLOCK; @@ -2826,6 +2831,9 @@ static void destroy_sit_info(struct f2fs_sb_info *sbi) SM_I(sbi)->sit_info = NULL; kfree(sit_i->sit_bitmap); +#ifdef CONFIG_F2FS_CHECK_FS + kfree(sit_i->sit_bitmap_mir); +#endif kfree(sit_i); } diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index 9af95194db06..5cb5755c75d9 100644 --- a/fs/f2fs/segment.h +++ b/fs/f2fs/segment.h @@ -209,6 +209,9 @@ struct sit_info { block_t sit_blocks; /* # of blocks used by SIT area */ block_t written_valid_blocks; /* # of valid blocks in main area */ char *sit_bitmap; /* SIT bitmap pointer */ +#ifdef CONFIG_F2FS_CHECK_FS + char *sit_bitmap_mir; /* SIT bitmap mirror */ +#endif unsigned int bitmap_size; /* SIT bitmap size */ unsigned long *tmp_map; /* bitmap for temporal use */ @@ -423,6 +426,12 @@ static inline void get_sit_bitmap(struct f2fs_sb_info *sbi, void *dst_addr) { struct sit_info *sit_i = SIT_I(sbi); + +#ifdef CONFIG_F2FS_CHECK_FS + if (memcmp(sit_i->sit_bitmap, sit_i->sit_bitmap_mir, + sit_i->bitmap_size)) + f2fs_bug_on(sbi, 1); +#endif memcpy(dst_addr, sit_i->sit_bitmap, sit_i->bitmap_size); } @@ -643,6 +652,12 @@ static inline pgoff_t current_sit_addr(struct f2fs_sb_info *sbi, check_seg_range(sbi, start); +#ifdef CONFIG_F2FS_CHECK_FS + if (f2fs_test_bit(offset, sit_i->sit_bitmap) != + f2fs_test_bit(offset, sit_i->sit_bitmap_mir)) + f2fs_bug_on(sbi, 1); +#endif + /* calculate sit block address */ if (f2fs_test_bit(offset, sit_i->sit_bitmap)) blk_addr += sit_i->sit_blocks; @@ -668,6 +683,9 @@ static inline void set_to_next_sit(struct sit_info *sit_i, unsigned int start) unsigned int block_off = SIT_BLOCK_OFFSET(start); f2fs_change_bit(block_off, sit_i->sit_bitmap); +#ifdef CONFIG_F2FS_CHECK_FS + f2fs_change_bit(block_off, sit_i->sit_bitmap_mir); +#endif } static inline unsigned long long get_mtime(struct f2fs_sb_info *sbi) -- cgit From b01a92019cac30398ef75b560d2668b399f4e393 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 9 Jan 2017 14:13:03 -0800 Subject: f2fs: clean up flush/discard command namings This patch simply cleans up the names for flush/discard commands. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/debug.c | 2 +- fs/f2fs/f2fs.h | 20 +++++------- fs/f2fs/segment.c | 98 +++++++++++++++++++++++++++---------------------------- 3 files changed, 59 insertions(+), 61 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c index 29cdf0c1da1d..883f1ea9e0b6 100644 --- a/fs/f2fs/debug.c +++ b/fs/f2fs/debug.c @@ -194,7 +194,7 @@ get_cache: si->cache_mem += sizeof(struct f2fs_gc_kthread); /* build merge flush thread */ - if (SM_I(sbi)->cmd_control_info) + if (SM_I(sbi)->fcc_info) si->cache_mem += sizeof(struct flush_cmd_control); /* free nids */ diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 4a84b3fbbfd1..548e75d18ec1 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -181,13 +181,12 @@ struct discard_entry { int len; /* # of consecutive blocks of the discard */ }; -struct bio_entry { - struct list_head list; - block_t lstart; - block_t len; - struct bio *bio; - struct completion event; - int error; +struct discard_cmd { + struct list_head list; /* command list */ + struct completion wait; /* compleation */ + block_t lstart; /* logical start address */ + block_t len; /* length */ + struct bio *bio; /* bio */ }; /* for the list of fsync inodes, used only during recovery */ @@ -634,8 +633,8 @@ struct f2fs_sm_info { unsigned int rec_prefree_segments; /* for small discard management */ - struct list_head discard_list; /* 4KB discard list */ - struct list_head wait_list; /* linked with issued discard bio */ + struct list_head discard_entry_list; /* 4KB discard entry list */ + struct list_head discard_cmd_list; /* discard cmd list */ int nr_discards; /* # of discards in the list */ int max_discards; /* max. discards to be issued */ @@ -649,8 +648,7 @@ struct f2fs_sm_info { unsigned int min_fsync_blocks; /* threshold for fsync */ /* for flush command control */ - struct flush_cmd_control *cmd_control_info; - + struct flush_cmd_control *fcc_info; }; /* diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index c8c825574fed..fa30c117b4cc 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -26,7 +26,7 @@ #define __reverse_ffz(x) __reverse_ffs(~(x)) static struct kmem_cache *discard_entry_slab; -static struct kmem_cache *bio_entry_slab; +static struct kmem_cache *discard_cmd_slab; static struct kmem_cache *sit_entry_set_slab; static struct kmem_cache *inmem_entry_slab; @@ -439,7 +439,7 @@ static int submit_flush_wait(struct f2fs_sb_info *sbi) static int issue_flush_thread(void *data) { struct f2fs_sb_info *sbi = data; - struct flush_cmd_control *fcc = SM_I(sbi)->cmd_control_info; + struct flush_cmd_control *fcc = SM_I(sbi)->fcc_info; wait_queue_head_t *q = &fcc->flush_wait_queue; repeat: if (kthread_should_stop()) @@ -468,7 +468,7 @@ repeat: int f2fs_issue_flush(struct f2fs_sb_info *sbi) { - struct flush_cmd_control *fcc = SM_I(sbi)->cmd_control_info; + struct flush_cmd_control *fcc = SM_I(sbi)->fcc_info; struct flush_cmd cmd; trace_f2fs_issue_flush(sbi->sb, test_opt(sbi, NOBARRIER), @@ -511,8 +511,8 @@ int create_flush_cmd_control(struct f2fs_sb_info *sbi) struct flush_cmd_control *fcc; int err = 0; - if (SM_I(sbi)->cmd_control_info) { - fcc = SM_I(sbi)->cmd_control_info; + if (SM_I(sbi)->fcc_info) { + fcc = SM_I(sbi)->fcc_info; goto init_thread; } @@ -522,14 +522,14 @@ int create_flush_cmd_control(struct f2fs_sb_info *sbi) atomic_set(&fcc->submit_flush, 0); init_waitqueue_head(&fcc->flush_wait_queue); init_llist_head(&fcc->issue_list); - SM_I(sbi)->cmd_control_info = fcc; + SM_I(sbi)->fcc_info = fcc; init_thread: fcc->f2fs_issue_flush = kthread_run(issue_flush_thread, sbi, "f2fs_flush-%u:%u", MAJOR(dev), MINOR(dev)); if (IS_ERR(fcc->f2fs_issue_flush)) { err = PTR_ERR(fcc->f2fs_issue_flush); kfree(fcc); - SM_I(sbi)->cmd_control_info = NULL; + SM_I(sbi)->fcc_info = NULL; return err; } @@ -538,7 +538,7 @@ init_thread: void destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free) { - struct flush_cmd_control *fcc = SM_I(sbi)->cmd_control_info; + struct flush_cmd_control *fcc = SM_I(sbi)->fcc_info; if (fcc && fcc->f2fs_issue_flush) { struct task_struct *flush_thread = fcc->f2fs_issue_flush; @@ -548,7 +548,7 @@ void destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free) } if (free) { kfree(fcc); - SM_I(sbi)->cmd_control_info = NULL; + SM_I(sbi)->fcc_info = NULL; } } @@ -628,42 +628,43 @@ static void locate_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno) mutex_unlock(&dirty_i->seglist_lock); } -static struct bio_entry *__add_bio_entry(struct f2fs_sb_info *sbi, +static struct discard_cmd *__add_discard_cmd(struct f2fs_sb_info *sbi, struct bio *bio, block_t lstart, block_t len) { - struct list_head *wait_list = &(SM_I(sbi)->wait_list); - struct bio_entry *be = f2fs_kmem_cache_alloc(bio_entry_slab, GFP_NOFS); + struct list_head *wait_list = &(SM_I(sbi)->discard_cmd_list); + struct discard_cmd *dc; - INIT_LIST_HEAD(&be->list); - be->bio = bio; - be->lstart = lstart; - be->len = len; - init_completion(&be->event); - list_add_tail(&be->list, wait_list); + dc = f2fs_kmem_cache_alloc(discard_cmd_slab, GFP_NOFS); + INIT_LIST_HEAD(&dc->list); + dc->bio = bio; + dc->lstart = lstart; + dc->len = len; + init_completion(&dc->wait); + list_add_tail(&dc->list, wait_list); - return be; + return dc; } /* This should be covered by global mutex, &sit_i->sentry_lock */ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) { - struct list_head *wait_list = &(SM_I(sbi)->wait_list); - struct bio_entry *be, *tmp; + struct list_head *wait_list = &(SM_I(sbi)->discard_cmd_list); + struct discard_cmd *dc, *tmp; - list_for_each_entry_safe(be, tmp, wait_list, list) { - struct bio *bio = be->bio; + list_for_each_entry_safe(dc, tmp, wait_list, list) { + struct bio *bio = dc->bio; int err; - if (!completion_done(&be->event)) { - if ((be->lstart <= blkaddr && - blkaddr < be->lstart + be->len) || + if (!completion_done(&dc->wait)) { + if ((dc->lstart <= blkaddr && + blkaddr < dc->lstart + dc->len) || blkaddr == NULL_ADDR) - wait_for_completion_io(&be->event); + wait_for_completion_io(&dc->wait); else continue; } - err = be->error; + err = bio->bi_error; if (err == -EOPNOTSUPP) err = 0; @@ -672,17 +673,16 @@ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) "Issue discard failed, ret: %d", err); bio_put(bio); - list_del(&be->list); - kmem_cache_free(bio_entry_slab, be); + list_del(&dc->list); + kmem_cache_free(discard_cmd_slab, dc); } } -static void f2fs_submit_bio_wait_endio(struct bio *bio) +static void f2fs_submit_discard_endio(struct bio *bio) { - struct bio_entry *be = (struct bio_entry *)bio->bi_private; + struct discard_cmd *dc = (struct discard_cmd *)bio->bi_private; - be->error = bio->bi_error; - complete(&be->event); + complete(&dc->wait); } /* this function is copied from blkdev_issue_discard from block/blk-lib.c */ @@ -705,11 +705,11 @@ static int __f2fs_issue_discard_async(struct f2fs_sb_info *sbi, SECTOR_FROM_BLOCK(blklen), GFP_NOFS, 0, &bio); if (!err && bio) { - struct bio_entry *be = __add_bio_entry(sbi, bio, + struct discard_cmd *dc = __add_discard_cmd(sbi, bio, lblkstart, blklen); - bio->bi_private = be; - bio->bi_end_io = f2fs_submit_bio_wait_endio; + bio->bi_private = dc; + bio->bi_end_io = f2fs_submit_discard_endio; bio->bi_opf |= REQ_SYNC; submit_bio(bio); } @@ -817,7 +817,7 @@ static void __add_discard_entry(struct f2fs_sb_info *sbi, struct cp_control *cpc, struct seg_entry *se, unsigned int start, unsigned int end) { - struct list_head *head = &SM_I(sbi)->discard_list; + struct list_head *head = &SM_I(sbi)->discard_entry_list; struct discard_entry *new, *last; if (!list_empty(head)) { @@ -886,7 +886,7 @@ static bool add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc, void release_discard_addrs(struct f2fs_sb_info *sbi) { - struct list_head *head = &(SM_I(sbi)->discard_list); + struct list_head *head = &(SM_I(sbi)->discard_entry_list); struct discard_entry *entry, *this; /* drop caches */ @@ -912,7 +912,7 @@ static void set_prefree_as_free_segments(struct f2fs_sb_info *sbi) void clear_prefree_segments(struct f2fs_sb_info *sbi, struct cp_control *cpc) { - struct list_head *head = &(SM_I(sbi)->discard_list); + struct list_head *head = &(SM_I(sbi)->discard_entry_list); struct discard_entry *entry, *this; struct dirty_seglist_info *dirty_i = DIRTY_I(sbi); struct blk_plug plug; @@ -2708,8 +2708,8 @@ int build_segment_manager(struct f2fs_sb_info *sbi) sm_info->min_ipu_util = DEF_MIN_IPU_UTIL; sm_info->min_fsync_blocks = DEF_MIN_FSYNC_BLOCKS; - INIT_LIST_HEAD(&sm_info->discard_list); - INIT_LIST_HEAD(&sm_info->wait_list); + INIT_LIST_HEAD(&sm_info->discard_entry_list); + INIT_LIST_HEAD(&sm_info->discard_cmd_list); sm_info->nr_discards = 0; sm_info->max_discards = 0; @@ -2859,15 +2859,15 @@ int __init create_segment_manager_caches(void) if (!discard_entry_slab) goto fail; - bio_entry_slab = f2fs_kmem_cache_create("bio_entry", - sizeof(struct bio_entry)); - if (!bio_entry_slab) + discard_cmd_slab = f2fs_kmem_cache_create("discard_cmd", + sizeof(struct discard_cmd)); + if (!discard_cmd_slab) goto destroy_discard_entry; sit_entry_set_slab = f2fs_kmem_cache_create("sit_entry_set", sizeof(struct sit_entry_set)); if (!sit_entry_set_slab) - goto destroy_bio_entry; + goto destroy_discard_cmd; inmem_entry_slab = f2fs_kmem_cache_create("inmem_page_entry", sizeof(struct inmem_pages)); @@ -2877,8 +2877,8 @@ int __init create_segment_manager_caches(void) destroy_sit_entry_set: kmem_cache_destroy(sit_entry_set_slab); -destroy_bio_entry: - kmem_cache_destroy(bio_entry_slab); +destroy_discard_cmd: + kmem_cache_destroy(discard_cmd_slab); destroy_discard_entry: kmem_cache_destroy(discard_entry_slab); fail: @@ -2888,7 +2888,7 @@ fail: void destroy_segment_manager_caches(void) { kmem_cache_destroy(sit_entry_set_slab); - kmem_cache_destroy(bio_entry_slab); + kmem_cache_destroy(discard_cmd_slab); kmem_cache_destroy(discard_entry_slab); kmem_cache_destroy(inmem_entry_slab); } -- cgit From 0b54fb8458199dbed409abb06933c27439ea0911 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 11 Jan 2017 14:40:24 -0800 Subject: f2fs: factor out discard command info into discard_cmd_control This patch adds discard_cmd_control with the existing discarding controls. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/debug.c | 2 ++ fs/f2fs/f2fs.h | 16 ++++++++----- fs/f2fs/segment.c | 68 +++++++++++++++++++++++++++++++++++++++++++------------ fs/f2fs/super.c | 5 +++- 4 files changed, 69 insertions(+), 22 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c index cd338ca24941..f9f6b0aeba02 100644 --- a/fs/f2fs/debug.c +++ b/fs/f2fs/debug.c @@ -196,6 +196,8 @@ get_cache: /* build merge flush thread */ if (SM_I(sbi)->fcc_info) si->cache_mem += sizeof(struct flush_cmd_control); + if (SM_I(sbi)->dcc_info) + si->cache_mem += sizeof(struct discard_cmd_control); /* free nids */ si->cache_mem += (NM_I(sbi)->nid_cnt[FREE_NID_LIST] + diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 548e75d18ec1..90eb2b3645a3 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -189,6 +189,13 @@ struct discard_cmd { struct bio *bio; /* bio */ }; +struct discard_cmd_control { + struct list_head discard_entry_list; /* 4KB discard entry list */ + int nr_discards; /* # of discards in the list */ + struct list_head discard_cmd_list; /* discard cmd list */ + int max_discards; /* max. discards to be issued */ +}; + /* for the list of fsync inodes, used only during recovery */ struct fsync_inode_entry { struct list_head list; /* list head */ @@ -632,12 +639,6 @@ struct f2fs_sm_info { /* a threshold to reclaim prefree segments */ unsigned int rec_prefree_segments; - /* for small discard management */ - struct list_head discard_entry_list; /* 4KB discard entry list */ - struct list_head discard_cmd_list; /* discard cmd list */ - int nr_discards; /* # of discards in the list */ - int max_discards; /* max. discards to be issued */ - /* for batched trimming */ unsigned int trim_sections; /* # of sections to trim */ @@ -649,6 +650,9 @@ struct f2fs_sm_info { /* for flush command control */ struct flush_cmd_control *fcc_info; + + /* for discard command control */ + struct discard_cmd_control *dcc_info; }; /* diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index fa30c117b4cc..a009f8a70c3d 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -631,7 +631,8 @@ static void locate_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno) static struct discard_cmd *__add_discard_cmd(struct f2fs_sb_info *sbi, struct bio *bio, block_t lstart, block_t len) { - struct list_head *wait_list = &(SM_I(sbi)->discard_cmd_list); + struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; + struct list_head *cmd_list = &(dcc->discard_cmd_list); struct discard_cmd *dc; dc = f2fs_kmem_cache_alloc(discard_cmd_slab, GFP_NOFS); @@ -640,7 +641,7 @@ static struct discard_cmd *__add_discard_cmd(struct f2fs_sb_info *sbi, dc->lstart = lstart; dc->len = len; init_completion(&dc->wait); - list_add_tail(&dc->list, wait_list); + list_add_tail(&dc->list, cmd_list); return dc; } @@ -648,7 +649,8 @@ static struct discard_cmd *__add_discard_cmd(struct f2fs_sb_info *sbi, /* This should be covered by global mutex, &sit_i->sentry_lock */ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) { - struct list_head *wait_list = &(SM_I(sbi)->discard_cmd_list); + struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; + struct list_head *wait_list = &(dcc->discard_cmd_list); struct discard_cmd *dc, *tmp; list_for_each_entry_safe(dc, tmp, wait_list, list) { @@ -817,7 +819,7 @@ static void __add_discard_entry(struct f2fs_sb_info *sbi, struct cp_control *cpc, struct seg_entry *se, unsigned int start, unsigned int end) { - struct list_head *head = &SM_I(sbi)->discard_entry_list; + struct list_head *head = &SM_I(sbi)->dcc_info->discard_entry_list; struct discard_entry *new, *last; if (!list_empty(head)) { @@ -835,7 +837,7 @@ static void __add_discard_entry(struct f2fs_sb_info *sbi, new->len = end - start; list_add_tail(&new->list, head); done: - SM_I(sbi)->nr_discards += end - start; + SM_I(sbi)->dcc_info->nr_discards += end - start; } static bool add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc, @@ -857,7 +859,8 @@ static bool add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc, if (!force) { if (!test_opt(sbi, DISCARD) || !se->valid_blocks || - SM_I(sbi)->nr_discards >= SM_I(sbi)->max_discards) + SM_I(sbi)->dcc_info->nr_discards >= + SM_I(sbi)->dcc_info->max_discards) return false; } @@ -866,7 +869,8 @@ static bool add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc, dmap[i] = force ? ~ckpt_map[i] & ~discard_map[i] : (cur_map[i] ^ ckpt_map[i]) & ckpt_map[i]; - while (force || SM_I(sbi)->nr_discards <= SM_I(sbi)->max_discards) { + while (force || SM_I(sbi)->dcc_info->nr_discards <= + SM_I(sbi)->dcc_info->max_discards) { start = __find_rev_next_bit(dmap, max_blocks, end + 1); if (start >= max_blocks) break; @@ -886,7 +890,7 @@ static bool add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc, void release_discard_addrs(struct f2fs_sb_info *sbi) { - struct list_head *head = &(SM_I(sbi)->discard_entry_list); + struct list_head *head = &(SM_I(sbi)->dcc_info->discard_entry_list); struct discard_entry *entry, *this; /* drop caches */ @@ -912,7 +916,7 @@ static void set_prefree_as_free_segments(struct f2fs_sb_info *sbi) void clear_prefree_segments(struct f2fs_sb_info *sbi, struct cp_control *cpc) { - struct list_head *head = &(SM_I(sbi)->discard_entry_list); + struct list_head *head = &(SM_I(sbi)->dcc_info->discard_entry_list); struct discard_entry *entry, *this; struct dirty_seglist_info *dirty_i = DIRTY_I(sbi); struct blk_plug plug; @@ -972,13 +976,47 @@ next: cpc->trimmed += entry->len; skip: list_del(&entry->list); - SM_I(sbi)->nr_discards -= entry->len; + SM_I(sbi)->dcc_info->nr_discards -= entry->len; kmem_cache_free(discard_entry_slab, entry); } blk_finish_plug(&plug); } +int create_discard_cmd_control(struct f2fs_sb_info *sbi) +{ + struct discard_cmd_control *dcc; + int err = 0; + + if (SM_I(sbi)->dcc_info) { + dcc = SM_I(sbi)->dcc_info; + goto init_thread; + } + + dcc = kzalloc(sizeof(struct discard_cmd_control), GFP_KERNEL); + if (!dcc) + return -ENOMEM; + + INIT_LIST_HEAD(&dcc->discard_entry_list); + INIT_LIST_HEAD(&dcc->discard_cmd_list); + dcc->nr_discards = 0; + dcc->max_discards = 0; + + SM_I(sbi)->dcc_info = dcc; +init_thread: + return err; +} + +void destroy_discard_cmd_control(struct f2fs_sb_info *sbi, bool free) +{ + struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; + + if (free) { + kfree(dcc); + SM_I(sbi)->dcc_info = NULL; + } +} + static bool __mark_sit_entry_dirty(struct f2fs_sb_info *sbi, unsigned int segno) { struct sit_info *sit_i = SIT_I(sbi); @@ -2708,11 +2746,6 @@ int build_segment_manager(struct f2fs_sb_info *sbi) sm_info->min_ipu_util = DEF_MIN_IPU_UTIL; sm_info->min_fsync_blocks = DEF_MIN_FSYNC_BLOCKS; - INIT_LIST_HEAD(&sm_info->discard_entry_list); - INIT_LIST_HEAD(&sm_info->discard_cmd_list); - sm_info->nr_discards = 0; - sm_info->max_discards = 0; - sm_info->trim_sections = DEF_BATCHED_TRIM_SECTIONS; INIT_LIST_HEAD(&sm_info->sit_entry_set); @@ -2723,6 +2756,10 @@ int build_segment_manager(struct f2fs_sb_info *sbi) return err; } + err = create_discard_cmd_control(sbi); + if (err) + return err; + err = build_sit_info(sbi); if (err) return err; @@ -2844,6 +2881,7 @@ void destroy_segment_manager(struct f2fs_sb_info *sbi) if (!sm_info) return; destroy_flush_cmd_control(sbi, true); + destroy_discard_cmd_control(sbi, true); destroy_dirty_segmap(sbi); destroy_curseg(sbi); destroy_free_segmap(sbi); diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c index e68cec492f06..921228189acd 100644 --- a/fs/f2fs/super.c +++ b/fs/f2fs/super.c @@ -145,6 +145,7 @@ static match_table_t f2fs_tokens = { enum { GC_THREAD, /* struct f2fs_gc_thread */ SM_INFO, /* struct f2fs_sm_info */ + DCC_INFO, /* struct discard_cmd_control */ NM_INFO, /* struct f2fs_nm_info */ F2FS_SBI, /* struct f2fs_sb_info */ #ifdef CONFIG_F2FS_FAULT_INJECTION @@ -168,6 +169,8 @@ static unsigned char *__struct_ptr(struct f2fs_sb_info *sbi, int struct_type) return (unsigned char *)sbi->gc_thread; else if (struct_type == SM_INFO) return (unsigned char *)SM_I(sbi); + else if (struct_type == DCC_INFO) + return (unsigned char *)SM_I(sbi)->dcc_info; else if (struct_type == NM_INFO) return (unsigned char *)NM_I(sbi); else if (struct_type == F2FS_SBI) @@ -283,7 +286,7 @@ F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_max_sleep_time, max_sleep_time); F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_no_gc_sleep_time, no_gc_sleep_time); F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_idle, gc_idle); F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, reclaim_segments, rec_prefree_segments); -F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, max_small_discards, max_discards); +F2FS_RW_ATTR(DCC_INFO, discard_cmd_control, max_small_discards, max_discards); F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, batched_trim_sections, trim_sections); F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, ipu_policy, ipu_policy); F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_ipu_util, min_ipu_util); -- cgit From 1546996348b33dc44dff829bc86fea8a8536164d Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 9 Jan 2017 20:32:07 -0800 Subject: f2fs: add a kernel thread to issue discard commands asynchronously This patch adds a kernel thread to issue discard commands. It proposes three states, D_PREP, D_SUBMIT, and D_DONE to identify current bio status. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/f2fs.h | 12 +++++ fs/f2fs/segment.c | 128 +++++++++++++++++++++++++++++++++++++++++------------- 2 files changed, 109 insertions(+), 31 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 90eb2b3645a3..3a57d97ed9b4 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -133,6 +133,8 @@ enum { (SM_I(sbi)->trim_sections * (sbi)->segs_per_sec) #define BATCHED_TRIM_BLOCKS(sbi) \ (BATCHED_TRIM_SEGMENTS(sbi) << (sbi)->log_blocks_per_seg) + +#define DISCARD_ISSUE_RATE 8 #define DEF_CP_INTERVAL 60 /* 60 secs */ #define DEF_IDLE_INTERVAL 5 /* 5 secs */ @@ -181,18 +183,28 @@ struct discard_entry { int len; /* # of consecutive blocks of the discard */ }; +enum { + D_PREP, + D_SUBMIT, + D_DONE, +}; + struct discard_cmd { struct list_head list; /* command list */ struct completion wait; /* compleation */ block_t lstart; /* logical start address */ block_t len; /* length */ struct bio *bio; /* bio */ + int state; /* state */ }; struct discard_cmd_control { + struct task_struct *f2fs_issue_discard; /* discard thread */ struct list_head discard_entry_list; /* 4KB discard entry list */ int nr_discards; /* # of discards in the list */ struct list_head discard_cmd_list; /* discard cmd list */ + wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ + struct mutex cmd_lock; int max_discards; /* max. discards to be issued */ }; diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index a009f8a70c3d..ed38644907ba 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -628,7 +628,7 @@ static void locate_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno) mutex_unlock(&dirty_i->seglist_lock); } -static struct discard_cmd *__add_discard_cmd(struct f2fs_sb_info *sbi, +static void __add_discard_cmd(struct f2fs_sb_info *sbi, struct bio *bio, block_t lstart, block_t len) { struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; @@ -638,12 +638,30 @@ static struct discard_cmd *__add_discard_cmd(struct f2fs_sb_info *sbi, dc = f2fs_kmem_cache_alloc(discard_cmd_slab, GFP_NOFS); INIT_LIST_HEAD(&dc->list); dc->bio = bio; + bio->bi_private = dc; dc->lstart = lstart; dc->len = len; + dc->state = D_PREP; init_completion(&dc->wait); + + mutex_lock(&dcc->cmd_lock); list_add_tail(&dc->list, cmd_list); + mutex_unlock(&dcc->cmd_lock); +} + +static void __remove_discard_cmd(struct f2fs_sb_info *sbi, struct discard_cmd *dc) +{ + int err = dc->bio->bi_error; - return dc; + if (err == -EOPNOTSUPP) + err = 0; + + if (err) + f2fs_msg(sbi->sb, KERN_INFO, + "Issue discard failed, ret: %d", err); + bio_put(dc->bio); + list_del(&dc->list); + kmem_cache_free(discard_cmd_slab, dc); } /* This should be covered by global mutex, &sit_i->sentry_lock */ @@ -653,31 +671,28 @@ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) struct list_head *wait_list = &(dcc->discard_cmd_list); struct discard_cmd *dc, *tmp; + mutex_lock(&dcc->cmd_lock); list_for_each_entry_safe(dc, tmp, wait_list, list) { - struct bio *bio = dc->bio; - int err; - if (!completion_done(&dc->wait)) { - if ((dc->lstart <= blkaddr && - blkaddr < dc->lstart + dc->len) || - blkaddr == NULL_ADDR) + if (blkaddr == NULL_ADDR) { + if (dc->state == D_PREP) { + dc->state = D_SUBMIT; + submit_bio(dc->bio); + } + wait_for_completion_io(&dc->wait); + + __remove_discard_cmd(sbi, dc); + continue; + } + + if (dc->lstart <= blkaddr && blkaddr < dc->lstart + dc->len) { + if (dc->state == D_SUBMIT) wait_for_completion_io(&dc->wait); else - continue; + __remove_discard_cmd(sbi, dc); } - - err = bio->bi_error; - if (err == -EOPNOTSUPP) - err = 0; - - if (err) - f2fs_msg(sbi->sb, KERN_INFO, - "Issue discard failed, ret: %d", err); - - bio_put(bio); - list_del(&dc->list); - kmem_cache_free(discard_cmd_slab, dc); } + mutex_unlock(&dcc->cmd_lock); } static void f2fs_submit_discard_endio(struct bio *bio) @@ -685,8 +700,48 @@ static void f2fs_submit_discard_endio(struct bio *bio) struct discard_cmd *dc = (struct discard_cmd *)bio->bi_private; complete(&dc->wait); + dc->state = D_DONE; } +static int issue_discard_thread(void *data) +{ + struct f2fs_sb_info *sbi = data; + struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; + wait_queue_head_t *q = &dcc->discard_wait_queue; + struct list_head *cmd_list = &dcc->discard_cmd_list; + struct discard_cmd *dc, *tmp; + struct blk_plug plug; + int iter = 0; +repeat: + if (kthread_should_stop()) + return 0; + + blk_start_plug(&plug); + + mutex_lock(&dcc->cmd_lock); + list_for_each_entry_safe(dc, tmp, cmd_list, list) { + if (dc->state == D_PREP) { + dc->state = D_SUBMIT; + submit_bio(dc->bio); + if (iter++ > DISCARD_ISSUE_RATE) + break; + } else if (dc->state == D_DONE) { + __remove_discard_cmd(sbi, dc); + } + } + mutex_unlock(&dcc->cmd_lock); + + blk_finish_plug(&plug); + + iter = 0; + congestion_wait(BLK_RW_SYNC, HZ/50); + + wait_event_interruptible(*q, + kthread_should_stop() || !list_empty(&dcc->discard_cmd_list)); + goto repeat; +} + + /* this function is copied from blkdev_issue_discard from block/blk-lib.c */ static int __f2fs_issue_discard_async(struct f2fs_sb_info *sbi, struct block_device *bdev, block_t blkstart, block_t blklen) @@ -707,13 +762,11 @@ static int __f2fs_issue_discard_async(struct f2fs_sb_info *sbi, SECTOR_FROM_BLOCK(blklen), GFP_NOFS, 0, &bio); if (!err && bio) { - struct discard_cmd *dc = __add_discard_cmd(sbi, bio, - lblkstart, blklen); - - bio->bi_private = dc; bio->bi_end_io = f2fs_submit_discard_endio; bio->bi_opf |= REQ_SYNC; - submit_bio(bio); + + __add_discard_cmd(sbi, bio, lblkstart, blklen); + wake_up(&SM_I(sbi)->dcc_info->discard_wait_queue); } return err; } @@ -919,14 +972,11 @@ void clear_prefree_segments(struct f2fs_sb_info *sbi, struct cp_control *cpc) struct list_head *head = &(SM_I(sbi)->dcc_info->discard_entry_list); struct discard_entry *entry, *this; struct dirty_seglist_info *dirty_i = DIRTY_I(sbi); - struct blk_plug plug; unsigned long *prefree_map = dirty_i->dirty_segmap[PRE]; unsigned int start = 0, end = -1; unsigned int secno, start_segno; bool force = (cpc->reason == CP_DISCARD); - blk_start_plug(&plug); - mutex_lock(&dirty_i->seglist_lock); while (1) { @@ -979,12 +1029,11 @@ skip: SM_I(sbi)->dcc_info->nr_discards -= entry->len; kmem_cache_free(discard_entry_slab, entry); } - - blk_finish_plug(&plug); } int create_discard_cmd_control(struct f2fs_sb_info *sbi) { + dev_t dev = sbi->sb->s_bdev->bd_dev; struct discard_cmd_control *dcc; int err = 0; @@ -999,11 +1048,22 @@ int create_discard_cmd_control(struct f2fs_sb_info *sbi) INIT_LIST_HEAD(&dcc->discard_entry_list); INIT_LIST_HEAD(&dcc->discard_cmd_list); + mutex_init(&dcc->cmd_lock); dcc->nr_discards = 0; dcc->max_discards = 0; + init_waitqueue_head(&dcc->discard_wait_queue); SM_I(sbi)->dcc_info = dcc; init_thread: + dcc->f2fs_issue_discard = kthread_run(issue_discard_thread, sbi, + "f2fs_discard-%u:%u", MAJOR(dev), MINOR(dev)); + if (IS_ERR(dcc->f2fs_issue_discard)) { + err = PTR_ERR(dcc->f2fs_issue_discard); + kfree(dcc); + SM_I(sbi)->dcc_info = NULL; + return err; + } + return err; } @@ -1011,6 +1071,12 @@ void destroy_discard_cmd_control(struct f2fs_sb_info *sbi, bool free) { struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; + if (dcc && dcc->f2fs_issue_discard) { + struct task_struct *discard_thread = dcc->f2fs_issue_discard; + + dcc->f2fs_issue_discard = NULL; + kthread_stop(discard_thread); + } if (free) { kfree(dcc); SM_I(sbi)->dcc_info = NULL; -- cgit From dcc9165dbf9961cf2848af728f8be31f28a3c790 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 11 Jan 2017 10:20:04 -0800 Subject: f2fs: show # of on-going flush and discard bios This patch adds stat information for flush and discard commands. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/debug.c | 11 +++++++++-- fs/f2fs/f2fs.h | 3 ++- fs/f2fs/segment.c | 6 ++++++ 3 files changed, 17 insertions(+), 3 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c index f9f6b0aeba02..0ca977a94c13 100644 --- a/fs/f2fs/debug.c +++ b/fs/f2fs/debug.c @@ -54,6 +54,12 @@ static void update_general_status(struct f2fs_sb_info *sbi) si->max_aw_cnt = atomic_read(&sbi->max_aw_cnt); si->nr_wb_cp_data = get_pages(sbi, F2FS_WB_CP_DATA); si->nr_wb_data = get_pages(sbi, F2FS_WB_DATA); + if (SM_I(sbi) && SM_I(sbi)->fcc_info) + si->nr_flush = + atomic_read(&SM_I(sbi)->fcc_info->submit_flush); + if (SM_I(sbi) && SM_I(sbi)->dcc_info) + si->nr_discard = + atomic_read(&SM_I(sbi)->dcc_info->submit_discard); si->total_count = (int)sbi->user_block_count / sbi->blocks_per_seg; si->rsvd_segs = reserved_segments(sbi); si->overp_segs = overprovision_segments(sbi); @@ -318,8 +324,9 @@ static int stat_show(struct seq_file *s, void *v) seq_printf(s, " - Inner Struct Count: tree: %d(%d), node: %d\n", si->ext_tree, si->zombie_tree, si->ext_node); seq_puts(s, "\nBalancing F2FS Async:\n"); - seq_printf(s, " - IO (CP: %4d, Data: %4d)\n", - si->nr_wb_cp_data, si->nr_wb_data); + seq_printf(s, " - IO (CP: %4d, Data: %4d, Flush: %4d, Discard: %4d)\n", + si->nr_wb_cp_data, si->nr_wb_data, + si->nr_flush, si->nr_discard); seq_printf(s, " - inmem: %4d, atomic IO: %4d (Max. %4d)\n", si->inmem_pages, si->aw_cnt, si->max_aw_cnt); seq_printf(s, " - nodes: %4d in %4d\n", diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 3a57d97ed9b4..5d9731036306 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -206,6 +206,7 @@ struct discard_cmd_control { wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ struct mutex cmd_lock; int max_discards; /* max. discards to be issued */ + atomic_t submit_discard; /* # of issued discard */ }; /* for the list of fsync inodes, used only during recovery */ @@ -2262,7 +2263,7 @@ struct f2fs_stat_info { unsigned int ndirty_dirs, ndirty_files, ndirty_all; int nats, dirty_nats, sits, dirty_sits, free_nids, alloc_nids; int total_count, utilization; - int bg_gc, nr_wb_cp_data, nr_wb_data; + int bg_gc, nr_wb_cp_data, nr_wb_data, nr_flush, nr_discard; int inline_xattr, inline_inode, inline_dir, orphans; int aw_cnt, max_aw_cnt; unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index ed38644907ba..52b133725b7f 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -653,6 +653,9 @@ static void __remove_discard_cmd(struct f2fs_sb_info *sbi, struct discard_cmd *d { int err = dc->bio->bi_error; + if (dc->state == D_DONE) + atomic_dec(&(SM_I(sbi)->dcc_info->submit_discard)); + if (err == -EOPNOTSUPP) err = 0; @@ -678,6 +681,7 @@ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) if (dc->state == D_PREP) { dc->state = D_SUBMIT; submit_bio(dc->bio); + atomic_inc(&dcc->submit_discard); } wait_for_completion_io(&dc->wait); @@ -723,6 +727,7 @@ repeat: if (dc->state == D_PREP) { dc->state = D_SUBMIT; submit_bio(dc->bio); + atomic_inc(&dcc->submit_discard); if (iter++ > DISCARD_ISSUE_RATE) break; } else if (dc->state == D_DONE) { @@ -1049,6 +1054,7 @@ int create_discard_cmd_control(struct f2fs_sb_info *sbi) INIT_LIST_HEAD(&dcc->discard_entry_list); INIT_LIST_HEAD(&dcc->discard_cmd_list); mutex_init(&dcc->cmd_lock); + atomic_set(&dcc->submit_discard, 0); dcc->nr_discards = 0; dcc->max_discards = 0; -- cgit From 73545817c90edcb367a65720f669cddce633bc46 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Wed, 25 Jan 2017 10:52:40 +0800 Subject: f2fs: fix null pointer dereference when issuing flush in ->fsync We only allocate flush merge control structure sbi::sm_info::fcc_info when flush_merge option is on, but in f2fs_issue_flush we still try to access member of the control structure without that option, it incurs panic as show below, fix it. Call Trace: __remove_ino_entry+0xa9/0xc0 [f2fs] f2fs_do_sync_file.isra.27+0x214/0x6d0 [f2fs] f2fs_sync_file+0x18/0x20 [f2fs] vfs_fsync_range+0x3d/0xb0 __do_page_fault+0x261/0x4d0 do_fsync+0x3d/0x70 SyS_fsync+0x10/0x20 do_syscall_64+0x6e/0x180 entry_SYSCALL64_slow_path+0x25/0x25 RIP: 0033:0x7f18ce260de0 RSP: 002b:00007ffdd4589258 EFLAGS: 00000246 ORIG_RAX: 000000000000004a RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007f18ce260de0 RDX: 0000000000000006 RSI: 00000000016c0360 RDI: 0000000000000003 RBP: 00000000016c0360 R08: 000000000000ffff R09: 000000000000001f R10: 00007ffdd4589020 R11: 0000000000000246 R12: 00000000016c0100 R13: 0000000000000000 R14: 00000000016c1f00 R15: 00000000016c0100 Code: fb 81 e3 00 08 00 00 48 89 45 a0 0f 1f 44 00 00 31 c0 85 db 75 27 41 81 e7 00 04 00 00 74 0c 41 8b 45 20 85 c0 0f 85 81 00 00 00 41 ff 45 20 4c 89 e7 e8 f8 e9 ff ff f0 41 ff 4d 20 48 83 c4 RIP: f2fs_issue_flush+0x5b/0x170 [f2fs] RSP: ffffc90003b5fd78 CR2: 0000000000000020 ---[ end trace a09314c24f037648 ]--- Reported-by: Shuoran Liu Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 52b133725b7f..4e515bbbeddc 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -477,7 +477,10 @@ int f2fs_issue_flush(struct f2fs_sb_info *sbi) if (test_opt(sbi, NOBARRIER)) return 0; - if (!test_opt(sbi, FLUSH_MERGE) || !atomic_read(&fcc->submit_flush)) { + if (!test_opt(sbi, FLUSH_MERGE)) + return submit_flush_wait(sbi); + + if (!atomic_read(&fcc->submit_flush)) { int ret; atomic_inc(&fcc->submit_flush); -- cgit From 8ed5974552086363aec2ac96fe01e3f2970baeab Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Sun, 29 Jan 2017 14:27:02 +0900 Subject: f2fs: declare missing static function We missed two functions declared as static functions. Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 4e515bbbeddc..5c0b59fe9bef 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1039,7 +1039,7 @@ skip: } } -int create_discard_cmd_control(struct f2fs_sb_info *sbi) +static int create_discard_cmd_control(struct f2fs_sb_info *sbi) { dev_t dev = sbi->sb->s_bdev->bd_dev; struct discard_cmd_control *dcc; @@ -1076,7 +1076,7 @@ init_thread: return err; } -void destroy_discard_cmd_control(struct f2fs_sb_info *sbi, bool free) +static void destroy_discard_cmd_control(struct f2fs_sb_info *sbi, bool free) { struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; -- cgit From c1b221078baf3d3275dc4309a716f6115696e2eb Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 2 Feb 2017 16:40:55 -0800 Subject: f2fs: move flush tracepoint This patch moves the tracepoint location for flush command. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 5c0b59fe9bef..2f6d7370904c 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -426,6 +426,9 @@ static int submit_flush_wait(struct f2fs_sb_info *sbi) int ret = __submit_flush_wait(sbi->sb->s_bdev); int i; + trace_f2fs_issue_flush(sbi->sb, test_opt(sbi, NOBARRIER), + test_opt(sbi, FLUSH_MERGE)); + if (sbi->s_ndevs && !ret) { for (i = 1; i < sbi->s_ndevs; i++) { ret = __submit_flush_wait(FDEV(i).bdev); @@ -471,9 +474,6 @@ int f2fs_issue_flush(struct f2fs_sb_info *sbi) struct flush_cmd_control *fcc = SM_I(sbi)->fcc_info; struct flush_cmd cmd; - trace_f2fs_issue_flush(sbi->sb, test_opt(sbi, NOBARRIER), - test_opt(sbi, FLUSH_MERGE)); - if (test_opt(sbi, NOBARRIER)) return 0; -- cgit From 942fd3192f83cef54bc0d485937fd5382ac5acd0 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 1 Feb 2017 16:51:22 -0800 Subject: f2fs: check last page index in cached bio to decide submission If the cached bio has the last page's index, then we need to submit it. Otherwise, we don't need to submit it and can wait for further IO merges. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/checkpoint.c | 3 ++- fs/f2fs/data.c | 43 ++++++++++++++++++++++--------------------- fs/f2fs/f2fs.h | 4 ++-- fs/f2fs/node.c | 12 +++++++----- fs/f2fs/segment.c | 13 +++++++------ 5 files changed, 40 insertions(+), 35 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c index fb8cdfcaece6..e6e42a4b1344 100644 --- a/fs/f2fs/checkpoint.c +++ b/fs/f2fs/checkpoint.c @@ -249,7 +249,8 @@ static int f2fs_write_meta_page(struct page *page, dec_page_count(sbi, F2FS_DIRTY_META); if (wbc->for_reclaim) - f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, META, WRITE); + f2fs_submit_merged_bio_cond(sbi, page->mapping->host, + 0, page->index, META, WRITE); unlock_page(page); diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index e0b74b9ea388..d14cc8be14f1 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -243,8 +243,8 @@ static void __submit_merged_bio(struct f2fs_bio_info *io) io->bio = NULL; } -static bool __has_merged_page(struct f2fs_bio_info *io, struct inode *inode, - struct page *page, nid_t ino) +static bool __has_merged_page(struct f2fs_bio_info *io, + struct inode *inode, nid_t ino, pgoff_t idx) { struct bio_vec *bvec; struct page *target; @@ -253,7 +253,7 @@ static bool __has_merged_page(struct f2fs_bio_info *io, struct inode *inode, if (!io->bio) return false; - if (!inode && !page && !ino) + if (!inode && !ino) return true; bio_for_each_segment_all(bvec, io->bio, i) { @@ -263,10 +263,11 @@ static bool __has_merged_page(struct f2fs_bio_info *io, struct inode *inode, else target = fscrypt_control_page(bvec->bv_page); + if (idx != target->index) + continue; + if (inode && inode == target->mapping->host) return true; - if (page && page == target) - return true; if (ino && ino == ino_of_node(target)) return true; } @@ -275,22 +276,21 @@ static bool __has_merged_page(struct f2fs_bio_info *io, struct inode *inode, } static bool has_merged_page(struct f2fs_sb_info *sbi, struct inode *inode, - struct page *page, nid_t ino, - enum page_type type) + nid_t ino, pgoff_t idx, enum page_type type) { enum page_type btype = PAGE_TYPE_OF_BIO(type); struct f2fs_bio_info *io = &sbi->write_io[btype]; bool ret; down_read(&io->io_rwsem); - ret = __has_merged_page(io, inode, page, ino); + ret = __has_merged_page(io, inode, ino, idx); up_read(&io->io_rwsem); return ret; } static void __f2fs_submit_merged_bio(struct f2fs_sb_info *sbi, - struct inode *inode, struct page *page, - nid_t ino, enum page_type type, int rw) + struct inode *inode, nid_t ino, pgoff_t idx, + enum page_type type, int rw) { enum page_type btype = PAGE_TYPE_OF_BIO(type); struct f2fs_bio_info *io; @@ -299,7 +299,7 @@ static void __f2fs_submit_merged_bio(struct f2fs_sb_info *sbi, down_write(&io->io_rwsem); - if (!__has_merged_page(io, inode, page, ino)) + if (!__has_merged_page(io, inode, ino, idx)) goto out; /* change META to META_FLUSH in the checkpoint procedure */ @@ -318,15 +318,15 @@ out: void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi, enum page_type type, int rw) { - __f2fs_submit_merged_bio(sbi, NULL, NULL, 0, type, rw); + __f2fs_submit_merged_bio(sbi, NULL, 0, 0, type, rw); } void f2fs_submit_merged_bio_cond(struct f2fs_sb_info *sbi, - struct inode *inode, struct page *page, - nid_t ino, enum page_type type, int rw) + struct inode *inode, nid_t ino, pgoff_t idx, + enum page_type type, int rw) { - if (has_merged_page(sbi, inode, page, ino, type)) - __f2fs_submit_merged_bio(sbi, inode, page, ino, type, rw); + if (has_merged_page(sbi, inode, ino, idx, type)) + __f2fs_submit_merged_bio(sbi, inode, ino, idx, type, rw); } void f2fs_flush_merged_bios(struct f2fs_sb_info *sbi) @@ -1432,7 +1432,8 @@ out: ClearPageUptodate(page); if (wbc->for_reclaim) { - f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, DATA, WRITE); + f2fs_submit_merged_bio_cond(sbi, inode, 0, page->index, + DATA, WRITE); remove_dirty_inode(inode); submitted = NULL; } @@ -1480,10 +1481,10 @@ static int f2fs_write_cache_pages(struct address_space *mapping, pgoff_t index; pgoff_t end; /* Inclusive */ pgoff_t done_index; + pgoff_t last_idx = ULONG_MAX; int cycled; int range_whole = 0; int tag; - int nwritten = 0; pagevec_init(&pvec, 0); @@ -1569,7 +1570,7 @@ continue_unlock: done = 1; break; } else if (submitted) { - nwritten++; + last_idx = page->index; } if (--wbc->nr_to_write <= 0 && @@ -1591,9 +1592,9 @@ continue_unlock: if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0)) mapping->writeback_index = done_index; - if (nwritten) + if (last_idx != ULONG_MAX) f2fs_submit_merged_bio_cond(F2FS_M_SB(mapping), mapping->host, - NULL, 0, DATA, WRITE); + 0, last_idx, DATA, WRITE); return ret; } diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 77ae2a8f15ce..b8070f9e8285 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -2233,8 +2233,8 @@ void destroy_checkpoint_caches(void); void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi, enum page_type type, int rw); void f2fs_submit_merged_bio_cond(struct f2fs_sb_info *sbi, - struct inode *inode, struct page *page, - nid_t ino, enum page_type type, int rw); + struct inode *inode, nid_t ino, pgoff_t idx, + enum page_type type, int rw); void f2fs_flush_merged_bios(struct f2fs_sb_info *sbi); int f2fs_submit_page_bio(struct f2fs_io_info *fio); int f2fs_submit_page_mbio(struct f2fs_io_info *fio); diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 5bd05e552d19..8203a2f8b350 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1374,7 +1374,8 @@ static int __write_node_page(struct page *page, bool atomic, bool *submitted, up_read(&sbi->node_write); if (wbc->for_reclaim) { - f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, NODE, WRITE); + f2fs_submit_merged_bio_cond(sbi, page->mapping->host, 0, + page->index, NODE, WRITE); submitted = NULL; } @@ -1404,12 +1405,12 @@ int fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, struct writeback_control *wbc, bool atomic) { pgoff_t index, end; + pgoff_t last_idx = ULONG_MAX; struct pagevec pvec; int ret = 0; struct page *last_page = NULL; bool marked = false; nid_t ino = inode->i_ino; - int nwritten = 0; if (atomic) { last_page = last_fsync_dnode(sbi, ino); @@ -1488,7 +1489,7 @@ continue_unlock: f2fs_put_page(last_page, 0); break; } else if (submitted) { - nwritten++; + last_idx = page->index; } if (page == last_page) { @@ -1514,8 +1515,9 @@ continue_unlock: goto retry; } out: - if (nwritten) - f2fs_submit_merged_bio_cond(sbi, NULL, NULL, ino, NODE, WRITE); + if (last_idx != ULONG_MAX) + f2fs_submit_merged_bio_cond(sbi, NULL, ino, last_idx, + NODE, WRITE); return ret ? -EIO: 0; } diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 2f6d7370904c..b2e0769a09d0 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -263,7 +263,7 @@ static int __commit_inmem_pages(struct inode *inode, .op_flags = REQ_SYNC | REQ_PRIO, .encrypted_page = NULL, }; - bool submit_bio = false; + pgoff_t last_idx = ULONG_MAX; int err = 0; list_for_each_entry_safe(cur, tmp, &fi->inmem_pages, list) { @@ -289,15 +289,15 @@ static int __commit_inmem_pages(struct inode *inode, /* record old blkaddr for revoking */ cur->old_addr = fio.old_blkaddr; - - submit_bio = true; + last_idx = page->index; } unlock_page(page); list_move_tail(&cur->list, revoke_list); } - if (submit_bio) - f2fs_submit_merged_bio_cond(sbi, inode, NULL, 0, DATA, WRITE); + if (last_idx != ULONG_MAX) + f2fs_submit_merged_bio_cond(sbi, inode, 0, last_idx, + DATA, WRITE); if (!err) __revoke_inmem_pages(inode, revoke_list, false, false); @@ -1932,7 +1932,8 @@ void f2fs_wait_on_page_writeback(struct page *page, if (PageWriteback(page)) { struct f2fs_sb_info *sbi = F2FS_P_SB(page); - f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, type, WRITE); + f2fs_submit_merged_bio_cond(sbi, page->mapping->host, + 0, page->index, type, WRITE); if (ordered) wait_on_page_writeback(page); else -- cgit From 5b6c6be2d878bd7ec4dc2cb4e2a2da2779fe52ab Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Tue, 14 Feb 2017 19:32:51 -0800 Subject: f2fs: use SSR for warm node as well We have had node chains, but haven't used it so far due to stale node blocks. Now, we have crc|cp_ver in node footer and give random cp_ver at format time, we can start to use it again. Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index b2e0769a09d0..0b42b0cdd674 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1563,7 +1563,8 @@ static void allocate_segment_by_default(struct f2fs_sb_info *sbi, if (force) new_curseg(sbi, type, true); - else if (type == CURSEG_WARM_NODE) + else if (!is_set_ckpt_flags(sbi, CP_CRC_RECOVERY_FLAG) && + type == CURSEG_WARM_NODE) new_curseg(sbi, type, false); else if (curseg->alloc_type == LFS && is_next_segment_free(sbi, type)) new_curseg(sbi, type, false); -- cgit From d50aaeec905baebcef0c401768b987ed18458015 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 15 Feb 2017 11:14:06 -0800 Subject: f2fs: show actual device info in tracepoints This patch shows actual device information in the tracepoints. Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 10 ++++----- include/trace/events/f2fs.h | 49 ++++++++++++++++++++++++--------------------- 2 files changed, 31 insertions(+), 28 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 0b42b0cdd674..68db07f857f8 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -426,11 +426,11 @@ static int submit_flush_wait(struct f2fs_sb_info *sbi) int ret = __submit_flush_wait(sbi->sb->s_bdev); int i; - trace_f2fs_issue_flush(sbi->sb, test_opt(sbi, NOBARRIER), - test_opt(sbi, FLUSH_MERGE)); - if (sbi->s_ndevs && !ret) { for (i = 1; i < sbi->s_ndevs; i++) { + trace_f2fs_issue_flush(FDEV(i).bdev, + test_opt(sbi, NOBARRIER), + test_opt(sbi, FLUSH_MERGE)); ret = __submit_flush_wait(FDEV(i).bdev); if (ret) break; @@ -758,7 +758,7 @@ static int __f2fs_issue_discard_async(struct f2fs_sb_info *sbi, block_t lblkstart = blkstart; int err; - trace_f2fs_issue_discard(sbi->sb, blkstart, blklen); + trace_f2fs_issue_discard(bdev, blkstart, blklen); if (sbi->s_ndevs) { int devi = f2fs_target_device_index(sbi, blkstart); @@ -815,7 +815,7 @@ static int __f2fs_issue_discard_zone(struct f2fs_sb_info *sbi, return __f2fs_issue_discard_async(sbi, bdev, blkstart, blklen); case BLK_ZONE_TYPE_SEQWRITE_REQ: case BLK_ZONE_TYPE_SEQWRITE_PREF: - trace_f2fs_issue_reset_zone(sbi->sb, blkstart); + trace_f2fs_issue_reset_zone(bdev, blkstart); return blkdev_reset_zones(bdev, sector, nr_sects, GFP_NOFS); default: diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h index 82236792b50c..153572e256d6 100644 --- a/include/trace/events/f2fs.h +++ b/include/trace/events/f2fs.h @@ -6,8 +6,8 @@ #include -#define show_dev(entry) MAJOR(entry->dev), MINOR(entry->dev) -#define show_dev_ino(entry) show_dev(entry), (unsigned long)entry->ino +#define show_dev(dev) MAJOR(dev), MINOR(dev) +#define show_dev_ino(entry) show_dev(entry->dev), (unsigned long)entry->ino TRACE_DEFINE_ENUM(NODE); TRACE_DEFINE_ENUM(DATA); @@ -245,7 +245,7 @@ TRACE_EVENT(f2fs_sync_fs, ), TP_printk("dev = (%d,%d), superblock is %s, wait = %d", - show_dev(__entry), + show_dev(__entry->dev), __entry->dirty ? "dirty" : "not dirty", __entry->wait) ); @@ -544,7 +544,7 @@ TRACE_EVENT(f2fs_background_gc, ), TP_printk("dev = (%d,%d), wait_ms = %ld, prefree = %u, free = %u", - show_dev(__entry), + show_dev(__entry->dev), __entry->wait_ms, __entry->prefree, __entry->free) @@ -586,7 +586,7 @@ TRACE_EVENT(f2fs_get_victim, TP_printk("dev = (%d,%d), type = %s, policy = (%s, %s, %s), victim = %u " "ofs_unit = %u, pre_victim_secno = %d, prefree = %u, free = %u", - show_dev(__entry), + show_dev(__entry->dev), show_data_type(__entry->type), show_gc_type(__entry->gc_type), show_alloc_mode(__entry->alloc_mode), @@ -723,7 +723,7 @@ TRACE_EVENT(f2fs_reserve_new_blocks, ), TP_printk("dev = (%d,%d), nid = %u, ofs_in_node = %u, count = %llu", - show_dev(__entry), + show_dev(__entry->dev), (unsigned int)__entry->nid, __entry->ofs_in_node, (unsigned long long)__entry->count) @@ -793,6 +793,7 @@ DECLARE_EVENT_CLASS(f2fs__bio, TP_STRUCT__entry( __field(dev_t, dev) + __field(dev_t, target) __field(int, op) __field(int, op_flags) __field(int, type) @@ -802,6 +803,7 @@ DECLARE_EVENT_CLASS(f2fs__bio, TP_fast_assign( __entry->dev = sb->s_dev; + __entry->target = bio->bi_bdev->bd_dev; __entry->op = bio_op(bio); __entry->op_flags = bio->bi_opf; __entry->type = type; @@ -809,8 +811,9 @@ DECLARE_EVENT_CLASS(f2fs__bio, __entry->size = bio->bi_iter.bi_size; ), - TP_printk("dev = (%d,%d), rw = %s%s, %s, sector = %lld, size = %u", - show_dev(__entry), + TP_printk("dev = (%d,%d)/(%d,%d), rw = %s%s, %s, sector = %lld, size = %u", + show_dev(__entry->target), + show_dev(__entry->dev), show_bio_type(__entry->op, __entry->op_flags), show_block_type(__entry->type), (unsigned long long)__entry->sector, @@ -1107,16 +1110,16 @@ TRACE_EVENT(f2fs_write_checkpoint, ), TP_printk("dev = (%d,%d), checkpoint for %s, state = %s", - show_dev(__entry), + show_dev(__entry->dev), show_cpreason(__entry->reason), __entry->msg) ); TRACE_EVENT(f2fs_issue_discard, - TP_PROTO(struct super_block *sb, block_t blkstart, block_t blklen), + TP_PROTO(struct block_device *dev, block_t blkstart, block_t blklen), - TP_ARGS(sb, blkstart, blklen), + TP_ARGS(dev, blkstart, blklen), TP_STRUCT__entry( __field(dev_t, dev) @@ -1125,22 +1128,22 @@ TRACE_EVENT(f2fs_issue_discard, ), TP_fast_assign( - __entry->dev = sb->s_dev; + __entry->dev = dev->bd_dev; __entry->blkstart = blkstart; __entry->blklen = blklen; ), TP_printk("dev = (%d,%d), blkstart = 0x%llx, blklen = 0x%llx", - show_dev(__entry), + show_dev(__entry->dev), (unsigned long long)__entry->blkstart, (unsigned long long)__entry->blklen) ); TRACE_EVENT(f2fs_issue_reset_zone, - TP_PROTO(struct super_block *sb, block_t blkstart), + TP_PROTO(struct block_device *dev, block_t blkstart), - TP_ARGS(sb, blkstart), + TP_ARGS(dev, blkstart), TP_STRUCT__entry( __field(dev_t, dev) @@ -1148,21 +1151,21 @@ TRACE_EVENT(f2fs_issue_reset_zone, ), TP_fast_assign( - __entry->dev = sb->s_dev; + __entry->dev = dev->bd_dev; __entry->blkstart = blkstart; ), TP_printk("dev = (%d,%d), reset zone at block = 0x%llx", - show_dev(__entry), + show_dev(__entry->dev), (unsigned long long)__entry->blkstart) ); TRACE_EVENT(f2fs_issue_flush, - TP_PROTO(struct super_block *sb, unsigned int nobarrier, + TP_PROTO(struct block_device *dev, unsigned int nobarrier, unsigned int flush_merge), - TP_ARGS(sb, nobarrier, flush_merge), + TP_ARGS(dev, nobarrier, flush_merge), TP_STRUCT__entry( __field(dev_t, dev) @@ -1171,13 +1174,13 @@ TRACE_EVENT(f2fs_issue_flush, ), TP_fast_assign( - __entry->dev = sb->s_dev; + __entry->dev = dev->bd_dev; __entry->nobarrier = nobarrier; __entry->flush_merge = flush_merge; ), TP_printk("dev = (%d,%d), %s %s", - show_dev(__entry), + show_dev(__entry->dev), __entry->nobarrier ? "skip (nobarrier)" : "issue", __entry->flush_merge ? " with flush_merge" : "") ); @@ -1292,7 +1295,7 @@ TRACE_EVENT(f2fs_shrink_extent_tree, ), TP_printk("dev = (%d,%d), shrunk: node_cnt = %u, tree_cnt = %u", - show_dev(__entry), + show_dev(__entry->dev), __entry->node_cnt, __entry->tree_cnt) ); @@ -1339,7 +1342,7 @@ DECLARE_EVENT_CLASS(f2fs_sync_dirty_inodes, ), TP_printk("dev = (%d,%d), %s, dirty count = %lld", - show_dev(__entry), + show_dev(__entry->dev), show_file_type(__entry->type), __entry->count) ); -- cgit From 3436c4bdb30de421d46f58c9174669fbcfd40ce0 Mon Sep 17 00:00:00 2001 From: Yunlong Song Date: Tue, 21 Feb 2017 16:59:26 +0800 Subject: f2fs: put allocate_segment after refresh_sit_entry SIT information should be updated before segment allocation, since SSR needs latest valid block information. Current code does not update the old_blkaddr info in sit_entry, so adjust the allocate_segment to its proper location. Commit 5e443818fa0b2a2845561ee25bec181424fb2889 ("f2fs: handle dirty segments inside refresh_sit_entry") puts it into wrong location. Signed-off-by: Yunlong Song Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 68db07f857f8..454c07d470fc 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1759,14 +1759,15 @@ void allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, stat_inc_block_count(sbi, curseg); - if (!__has_curseg_space(sbi, type)) - sit_i->s_ops->allocate_segment(sbi, type, false); /* * SIT information should be updated before segment allocation, * since SSR needs latest valid block information. */ refresh_sit_entry(sbi, old_blkaddr, *new_blkaddr); + if (!__has_curseg_space(sbi, type)) + sit_i->s_ops->allocate_segment(sbi, type, false); + mutex_unlock(&sit_i->sentry_lock); if (page && IS_NODESEG(type)) -- cgit From 035e97adab26c1121cedaeb9bd04cf48a8e8cf51 Mon Sep 17 00:00:00 2001 From: Yunlong Song Date: Wed, 22 Feb 2017 20:50:49 +0800 Subject: f2fs: do SSR for data when there is enough free space In allocate_segment_by_default(), need_SSR() already detected it's time to do SSR. So, let's try to find victims for data segments more aggressively in time. Signed-off-by: Yunlong Song Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 454c07d470fc..46e29c555299 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1540,7 +1540,7 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - if (IS_NODESEG(type) || !has_not_enough_free_secs(sbi, 0, 0)) + if (IS_NODESEG(type)) return v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR); -- cgit From d0db7703ac186ff2fbed1e8b862bd43ec3fd2812 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 22 Feb 2017 16:39:11 -0800 Subject: f2fs: do SSR in higher priority Let's check SSR in prior to LFS allocation. Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 17 +---------------- 1 file changed, 1 insertion(+), 16 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 46e29c555299..8ba802e1bb06 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1324,17 +1324,6 @@ static void write_current_sum_page(struct f2fs_sb_info *sbi, f2fs_put_page(page, 1); } -static int is_next_segment_free(struct f2fs_sb_info *sbi, int type) -{ - struct curseg_info *curseg = CURSEG_I(sbi, type); - unsigned int segno = curseg->segno + 1; - struct free_segmap_info *free_i = FREE_I(sbi); - - if (segno < MAIN_SEGS(sbi) && segno % sbi->segs_per_sec) - return !test_bit(segno, free_i->free_segmap); - return 0; -} - /* * Find a new segment from the free segments bitmap to right order * This function should be returned with success, otherwise BUG @@ -1559,21 +1548,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) static void allocate_segment_by_default(struct f2fs_sb_info *sbi, int type, bool force) { - struct curseg_info *curseg = CURSEG_I(sbi, type); - if (force) new_curseg(sbi, type, true); else if (!is_set_ckpt_flags(sbi, CP_CRC_RECOVERY_FLAG) && type == CURSEG_WARM_NODE) new_curseg(sbi, type, false); - else if (curseg->alloc_type == LFS && is_next_segment_free(sbi, type)) - new_curseg(sbi, type, false); else if (need_SSR(sbi) && get_ssr_segment(sbi, type)) change_curseg(sbi, type, true); else new_curseg(sbi, type, false); - stat_inc_seg_type(sbi, curseg); + stat_inc_seg_type(sbi, CURSEG_I(sbi, type)); } void allocate_new_segments(struct f2fs_sb_info *sbi) -- cgit From c192f7a4779559dddeb8c03de2b6ef499115c938 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 22 Feb 2017 17:10:18 -0800 Subject: f2fs: find data segments across all the types Previously, if type is CURSEG_HOT_DATA, we only check CURSEG_HOT_DATA only. This patch fixes to search all the different types for SSR. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 8ba802e1bb06..630e7045c53e 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1528,16 +1528,23 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; + int i; + + /* need_SSR() already forces to do this */ + if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) + return 1; if (IS_NODESEG(type)) - return v_ops->get_victim(sbi, - &(curseg)->next_segno, BG_GC, type, SSR); + return 0; /* For data segments, let's do SSR more intensively */ - for (; type >= CURSEG_HOT_DATA; type--) + for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) { + if (i == type) + continue; if (v_ops->get_victim(sbi, &(curseg)->next_segno, - BG_GC, type, SSR)) + BG_GC, i, SSR)) return 1; + } return 0; } -- cgit From 70d625cbdbbb99e38953b8b7ac792a6471e68efb Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 22 Feb 2017 17:02:32 -0800 Subject: f2fs: do SSR for node segments more aggresively This patch gives more SSR chances for node blocks. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 630e7045c53e..d780ecccc4ac 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1528,17 +1528,22 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - int i; + int i, n; /* need_SSR() already forces to do this */ if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) return 1; - if (IS_NODESEG(type)) - return 0; + /* For node segments, let's do SSR more intensively */ + if (IS_NODESEG(type)) { + i = CURSEG_HOT_NODE; + n = CURSEG_COLD_NODE; + } else { + i = CURSEG_HOT_DATA; + n = CURSEG_COLD_DATA; + } - /* For data segments, let's do SSR more intensively */ - for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) { + for (; i <= n; i++) { if (i == type) continue; if (v_ops->get_victim(sbi, &(curseg)->next_segno, -- cgit From ad4d307fce0909a5f70635826f779321ab95b469 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 22 Feb 2017 19:10:35 -0800 Subject: f2fs: avoid very large discard command This patch adds MAX_DISCARD_BLOCKS() to avoid issuing too much large single discard command. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/f2fs.h | 3 ++- fs/f2fs/segment.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 146cc71ff91c..2ce9fba27918 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -133,7 +133,8 @@ enum { (SM_I(sbi)->trim_sections * (sbi)->segs_per_sec) #define BATCHED_TRIM_BLOCKS(sbi) \ (BATCHED_TRIM_SEGMENTS(sbi) << (sbi)->log_blocks_per_seg) - +#define MAX_DISCARD_BLOCKS(sbi) \ + ((1 << (sbi)->log_blocks_per_seg) * (sbi)->segs_per_sec) #define DISCARD_ISSUE_RATE 8 #define DEF_CP_INTERVAL 60 /* 60 secs */ #define DEF_IDLE_INTERVAL 5 /* 5 secs */ diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index d780ecccc4ac..028db55375a4 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -886,7 +886,8 @@ static void __add_discard_entry(struct f2fs_sb_info *sbi, if (!list_empty(head)) { last = list_last_entry(head, struct discard_entry, list); if (START_BLOCK(sbi, cpc->trim_start) + start == - last->blkaddr + last->len) { + last->blkaddr + last->len && + last->len < MAX_DISCARD_BLOCKS(sbi)) { last->len += end - start; goto done; } -- cgit From 40465257ac4159e9ebfba10ee502e795b917da76 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 22 Feb 2017 19:58:23 -0800 Subject: f2fs: wait for discard completion after submission We don't need to wait for each discard commands when unmounting the image. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 028db55375a4..df60e1502b82 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -676,8 +676,12 @@ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info; struct list_head *wait_list = &(dcc->discard_cmd_list); struct discard_cmd *dc, *tmp; + struct blk_plug plug; mutex_lock(&dcc->cmd_lock); + + blk_start_plug(&plug); + list_for_each_entry_safe(dc, tmp, wait_list, list) { if (blkaddr == NULL_ADDR) { @@ -686,9 +690,6 @@ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) submit_bio(dc->bio); atomic_inc(&dcc->submit_discard); } - wait_for_completion_io(&dc->wait); - - __remove_discard_cmd(sbi, dc); continue; } @@ -699,6 +700,15 @@ void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr) __remove_discard_cmd(sbi, dc); } } + blk_finish_plug(&plug); + + /* this comes from f2fs_put_super */ + if (blkaddr == NULL_ADDR) { + list_for_each_entry_safe(dc, tmp, wait_list, list) { + wait_for_completion_io(&dc->wait); + __remove_discard_cmd(sbi, dc); + } + } mutex_unlock(&dcc->cmd_lock); } -- cgit From 92592285710aebe2de9afa38315b5c3914556905 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Wed, 22 Feb 2017 20:18:35 -0800 Subject: f2fs: check discard alignment only for SEQWRITE zones For converntional zones, we don't need to align discard commands to exact zone size. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 24 ++++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index df60e1502b82..3e95db5375ed 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -793,24 +793,13 @@ static int __f2fs_issue_discard_async(struct f2fs_sb_info *sbi, static int __f2fs_issue_discard_zone(struct f2fs_sb_info *sbi, struct block_device *bdev, block_t blkstart, block_t blklen) { - sector_t nr_sects = SECTOR_FROM_BLOCK(blklen); - sector_t sector; + sector_t sector, nr_sects; int devi = 0; if (sbi->s_ndevs) { devi = f2fs_target_device_index(sbi, blkstart); blkstart -= FDEV(devi).start_blk; } - sector = SECTOR_FROM_BLOCK(blkstart); - - if (sector & (bdev_zone_sectors(bdev) - 1) || - nr_sects != bdev_zone_sectors(bdev)) { - f2fs_msg(sbi->sb, KERN_INFO, - "(%d) %s: Unaligned discard attempted (block %x + %x)", - devi, sbi->s_ndevs ? FDEV(devi).path: "", - blkstart, blklen); - return -EIO; - } /* * We need to know the type of the zone: for conventional zones, @@ -825,6 +814,17 @@ static int __f2fs_issue_discard_zone(struct f2fs_sb_info *sbi, return __f2fs_issue_discard_async(sbi, bdev, blkstart, blklen); case BLK_ZONE_TYPE_SEQWRITE_REQ: case BLK_ZONE_TYPE_SEQWRITE_PREF: + sector = SECTOR_FROM_BLOCK(blkstart); + nr_sects = SECTOR_FROM_BLOCK(blklen); + + if (sector & (bdev_zone_sectors(bdev) - 1) || + nr_sects != bdev_zone_sectors(bdev)) { + f2fs_msg(sbi->sb, KERN_INFO, + "(%d) %s: Unaligned discard attempted (block %x + %x)", + devi, sbi->s_ndevs ? FDEV(devi).path: "", + blkstart, blklen); + return -EIO; + } trace_f2fs_issue_reset_zone(bdev, blkstart); return blkdev_reset_zones(bdev, sector, nr_sects, GFP_NOFS); -- cgit From 22ad0b6ab46683975c6da032f1c2593066c7b3bd Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 9 Feb 2017 10:38:09 -0800 Subject: f2fs: add bitmaps for empty or full NAT blocks This patches adds bitmaps to represent empty or full NAT blocks containing free nid entries. If we can find valid crc|cp_ver in the last block of checkpoint pack, we'll use these bitmaps when building free nids. In order to avoid checkpointing burden, up-to-date bitmaps will be flushed only during umount time. So, normally we can get this gain, but when power-cut happens, we rely on fsck.f2fs which recovers this bitmap again. After this patch, we build free nids from nid #0 at mount time to make more full NAT blocks, but in runtime, we check empty NAT blocks to load free nids without loading any NAT pages from disk. Signed-off-by: Jaegeuk Kim --- fs/f2fs/checkpoint.c | 28 +++++++- fs/f2fs/debug.c | 1 + fs/f2fs/f2fs.h | 31 +++++++- fs/f2fs/node.c | 188 +++++++++++++++++++++++++++++++++++++++++++----- fs/f2fs/segment.c | 2 +- include/linux/f2fs_fs.h | 1 + 6 files changed, 231 insertions(+), 20 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c index 042f8d9afe44..cd7132121573 100644 --- a/fs/f2fs/checkpoint.c +++ b/fs/f2fs/checkpoint.c @@ -1024,6 +1024,10 @@ static void update_ckpt_flags(struct f2fs_sb_info *sbi, struct cp_control *cpc) spin_lock(&sbi->cp_lock); + if (cpc->reason == CP_UMOUNT && ckpt->cp_pack_total_block_count > + sbi->blocks_per_seg - NM_I(sbi)->nat_bits_blocks) + disable_nat_bits(sbi, false); + if (cpc->reason == CP_UMOUNT) __set_ckpt_flags(ckpt, CP_UMOUNT_FLAG); else @@ -1136,6 +1140,28 @@ static int do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc) start_blk = __start_cp_next_addr(sbi); + /* write nat bits */ + if (enabled_nat_bits(sbi, cpc)) { + __u64 cp_ver = cur_cp_version(ckpt); + unsigned int i; + block_t blk; + + cp_ver |= ((__u64)crc32 << 32); + *(__le64 *)nm_i->nat_bits = cpu_to_le64(cp_ver); + + blk = start_blk + sbi->blocks_per_seg - nm_i->nat_bits_blocks; + for (i = 0; i < nm_i->nat_bits_blocks; i++) + update_meta_page(sbi, nm_i->nat_bits + + (i << F2FS_BLKSIZE_BITS), blk + i); + + /* Flush all the NAT BITS pages */ + while (get_pages(sbi, F2FS_DIRTY_META)) { + sync_meta_pages(sbi, META, LONG_MAX); + if (unlikely(f2fs_cp_error(sbi))) + return -EIO; + } + } + /* need to wait for end_io results */ wait_on_all_pages_writeback(sbi); if (unlikely(f2fs_cp_error(sbi))) @@ -1272,7 +1298,7 @@ int write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc) ckpt->checkpoint_ver = cpu_to_le64(++ckpt_ver); /* write cached NAT/SIT entries to NAT/SIT area */ - flush_nat_entries(sbi); + flush_nat_entries(sbi, cpc); flush_sit_entries(sbi, cpc); /* unlock all the fs_lock[] in do_checkpoint() */ diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c index de8da9fc5c99..015ad2b73a92 100644 --- a/fs/f2fs/debug.c +++ b/fs/f2fs/debug.c @@ -193,6 +193,7 @@ static void update_mem_info(struct f2fs_sb_info *sbi) /* build nm */ si->base_mem += sizeof(struct f2fs_nm_info); si->base_mem += __bitmap_size(sbi, NAT_BITMAP); + si->base_mem += (NM_I(sbi)->nat_bits_blocks << F2FS_BLKSIZE_BITS); get_cache: si->cache_mem = 0; diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index e26cc6909a54..d1156cdd211e 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -554,6 +554,7 @@ struct f2fs_nm_info { struct list_head nat_entries; /* cached nat entry list (clean) */ unsigned int nat_cnt; /* the # of cached nat entries */ unsigned int dirty_nat_cnt; /* total num of nat entries in set */ + unsigned int nat_blocks; /* # of nat blocks */ /* free node ids management */ struct radix_tree_root free_nid_root;/* root of the free_nid cache */ @@ -564,6 +565,11 @@ struct f2fs_nm_info { /* for checkpoint */ char *nat_bitmap; /* NAT bitmap pointer */ + + unsigned int nat_bits_blocks; /* # of nat bits blocks */ + unsigned char *nat_bits; /* NAT bits blocks */ + unsigned char *full_nat_bits; /* full NAT pages */ + unsigned char *empty_nat_bits; /* empty NAT pages */ #ifdef CONFIG_F2FS_CHECK_FS char *nat_bitmap_mir; /* NAT bitmap mirror */ #endif @@ -1171,6 +1177,27 @@ static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) spin_unlock(&sbi->cp_lock); } +static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) +{ + set_sbi_flag(sbi, SBI_NEED_FSCK); + + if (lock) + spin_lock(&sbi->cp_lock); + __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); + kfree(NM_I(sbi)->nat_bits); + NM_I(sbi)->nat_bits = NULL; + if (lock) + spin_unlock(&sbi->cp_lock); +} + +static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, + struct cp_control *cpc) +{ + bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); + + return (cpc) ? (cpc->reason == CP_UMOUNT) && set : set; +} + static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) { down_read(&sbi->cp_rwsem); @@ -2131,7 +2158,7 @@ void move_node_page(struct page *node_page, int gc_type); int fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, struct writeback_control *wbc, bool atomic); int sync_node_pages(struct f2fs_sb_info *sbi, struct writeback_control *wbc); -void build_free_nids(struct f2fs_sb_info *sbi, bool sync); +void build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); bool alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); void alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); void alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); @@ -2142,7 +2169,7 @@ int recover_xattr_data(struct inode *inode, struct page *page, int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); int restore_node_summary(struct f2fs_sb_info *sbi, unsigned int segno, struct f2fs_summary_block *sum); -void flush_nat_entries(struct f2fs_sb_info *sbi); +void flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); int build_node_manager(struct f2fs_sb_info *sbi); void destroy_node_manager(struct f2fs_sb_info *sbi); int __init create_node_manager_caches(void); diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 8ebc4c78e6a4..43d35ec11851 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -338,6 +338,9 @@ static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni, set_nat_flag(e, IS_CHECKPOINTED, false); __set_nat_cache_dirty(nm_i, e); + if (enabled_nat_bits(sbi, NULL) && new_blkaddr == NEW_ADDR) + clear_bit_le(NAT_BLOCK_OFFSET(ni->nid), nm_i->empty_nat_bits); + /* update fsync_mark if its inode nat entry is still alive */ if (ni->nid != ni->ino) e = __lookup_nat_cache(nm_i, ni->ino); @@ -1841,7 +1844,60 @@ static void scan_nat_page(struct f2fs_sb_info *sbi, } } -static void __build_free_nids(struct f2fs_sb_info *sbi, bool sync) +static int scan_nat_bits(struct f2fs_sb_info *sbi) +{ + struct f2fs_nm_info *nm_i = NM_I(sbi); + struct page *page; + unsigned int i = 0; + nid_t target = FREE_NID_PAGES * NAT_ENTRY_PER_BLOCK; + nid_t nid; + + if (!enabled_nat_bits(sbi, NULL)) + return -EAGAIN; + + down_read(&nm_i->nat_tree_lock); +check_empty: + i = find_next_bit_le(nm_i->empty_nat_bits, nm_i->nat_blocks, i); + if (i >= nm_i->nat_blocks) { + i = 0; + goto check_partial; + } + + for (nid = i * NAT_ENTRY_PER_BLOCK; nid < (i + 1) * NAT_ENTRY_PER_BLOCK; + nid++) { + if (unlikely(nid >= nm_i->max_nid)) + break; + add_free_nid(sbi, nid, true); + } + + if (nm_i->nid_cnt[FREE_NID_LIST] >= target) + goto out; + i++; + goto check_empty; + +check_partial: + i = find_next_zero_bit_le(nm_i->full_nat_bits, nm_i->nat_blocks, i); + if (i >= nm_i->nat_blocks) { + disable_nat_bits(sbi, true); + up_read(&nm_i->nat_tree_lock); + return -EINVAL; + } + + nid = i * NAT_ENTRY_PER_BLOCK; + page = get_current_nat_page(sbi, nid); + scan_nat_page(sbi, page, nid); + f2fs_put_page(page, 1); + + if (nm_i->nid_cnt[FREE_NID_LIST] < target) { + i++; + goto check_partial; + } +out: + up_read(&nm_i->nat_tree_lock); + return 0; +} + +static void __build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount) { struct f2fs_nm_info *nm_i = NM_I(sbi); struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA); @@ -1856,6 +1912,21 @@ static void __build_free_nids(struct f2fs_sb_info *sbi, bool sync) if (!sync && !available_free_memory(sbi, FREE_NIDS)) return; + /* try to find free nids with nat_bits */ + if (!mount && !scan_nat_bits(sbi) && nm_i->nid_cnt[FREE_NID_LIST]) + return; + + /* find next valid candidate */ + if (enabled_nat_bits(sbi, NULL)) { + int idx = find_next_zero_bit_le(nm_i->full_nat_bits, + nm_i->nat_blocks, 0); + + if (idx >= nm_i->nat_blocks) + set_sbi_flag(sbi, SBI_NEED_FSCK); + else + nid = idx * NAT_ENTRY_PER_BLOCK; + } + /* readahead nat pages to be scanned */ ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nid), FREE_NID_PAGES, META_NAT, true); @@ -1898,10 +1969,10 @@ static void __build_free_nids(struct f2fs_sb_info *sbi, bool sync) nm_i->ra_nid_pages, META_NAT, false); } -void build_free_nids(struct f2fs_sb_info *sbi, bool sync) +void build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount) { mutex_lock(&NM_I(sbi)->build_lock); - __build_free_nids(sbi, sync); + __build_free_nids(sbi, sync, mount); mutex_unlock(&NM_I(sbi)->build_lock); } @@ -1943,7 +2014,7 @@ retry: spin_unlock(&nm_i->nid_list_lock); /* Let's scan nat pages and its caches to get free nids */ - build_free_nids(sbi, true); + build_free_nids(sbi, true, false); goto retry; } @@ -2235,8 +2306,39 @@ add_out: list_add_tail(&nes->set_list, head); } +void __update_nat_bits(struct f2fs_sb_info *sbi, nid_t start_nid, + struct page *page) +{ + struct f2fs_nm_info *nm_i = NM_I(sbi); + unsigned int nat_index = start_nid / NAT_ENTRY_PER_BLOCK; + struct f2fs_nat_block *nat_blk = page_address(page); + int valid = 0; + int i; + + if (!enabled_nat_bits(sbi, NULL)) + return; + + for (i = 0; i < NAT_ENTRY_PER_BLOCK; i++) { + if (start_nid == 0 && i == 0) + valid++; + if (nat_blk->entries[i].block_addr) + valid++; + } + if (valid == 0) { + set_bit_le(nat_index, nm_i->empty_nat_bits); + clear_bit_le(nat_index, nm_i->full_nat_bits); + return; + } + + clear_bit_le(nat_index, nm_i->empty_nat_bits); + if (valid == NAT_ENTRY_PER_BLOCK) + set_bit_le(nat_index, nm_i->full_nat_bits); + else + clear_bit_le(nat_index, nm_i->full_nat_bits); +} + static void __flush_nat_entry_set(struct f2fs_sb_info *sbi, - struct nat_entry_set *set) + struct nat_entry_set *set, struct cp_control *cpc) { struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA); struct f2fs_journal *journal = curseg->journal; @@ -2251,7 +2353,8 @@ static void __flush_nat_entry_set(struct f2fs_sb_info *sbi, * #1, flush nat entries to journal in current hot data summary block. * #2, flush nat entries to nat page. */ - if (!__has_cursum_space(journal, set->entry_cnt, NAT_JOURNAL)) + if (enabled_nat_bits(sbi, cpc) || + !__has_cursum_space(journal, set->entry_cnt, NAT_JOURNAL)) to_journal = false; if (to_journal) { @@ -2291,10 +2394,12 @@ static void __flush_nat_entry_set(struct f2fs_sb_info *sbi, } } - if (to_journal) + if (to_journal) { up_write(&curseg->journal_rwsem); - else + } else { + __update_nat_bits(sbi, start_nid, page); f2fs_put_page(page, 1); + } f2fs_bug_on(sbi, set->entry_cnt); @@ -2305,7 +2410,7 @@ static void __flush_nat_entry_set(struct f2fs_sb_info *sbi, /* * This function is called during the checkpointing process. */ -void flush_nat_entries(struct f2fs_sb_info *sbi) +void flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc) { struct f2fs_nm_info *nm_i = NM_I(sbi); struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA); @@ -2326,7 +2431,8 @@ void flush_nat_entries(struct f2fs_sb_info *sbi) * entries, remove all entries from journal and merge them * into nat entry set. */ - if (!__has_cursum_space(journal, nm_i->dirty_nat_cnt, NAT_JOURNAL)) + if (cpc->reason == CP_UMOUNT || + !__has_cursum_space(journal, nm_i->dirty_nat_cnt, NAT_JOURNAL)) remove_nats_in_journal(sbi); while ((found = __gang_lookup_nat_set(nm_i, @@ -2340,27 +2446,72 @@ void flush_nat_entries(struct f2fs_sb_info *sbi) /* flush dirty nats in nat entry set */ list_for_each_entry_safe(set, tmp, &sets, set_list) - __flush_nat_entry_set(sbi, set); + __flush_nat_entry_set(sbi, set, cpc); up_write(&nm_i->nat_tree_lock); f2fs_bug_on(sbi, nm_i->dirty_nat_cnt); } +static int __get_nat_bitmaps(struct f2fs_sb_info *sbi) +{ + struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); + struct f2fs_nm_info *nm_i = NM_I(sbi); + unsigned int nat_bits_bytes = nm_i->nat_blocks / BITS_PER_BYTE; + unsigned int i; + __u64 cp_ver = cur_cp_version(ckpt); + size_t crc_offset = le32_to_cpu(ckpt->checksum_offset); + __u64 crc = le32_to_cpu(*((__le32 *) + ((unsigned char *)ckpt + crc_offset))); + block_t nat_bits_addr; + + if (!enabled_nat_bits(sbi, NULL)) + return 0; + + nm_i->nat_bits_blocks = F2FS_BYTES_TO_BLK((nat_bits_bytes << 1) + 8 + + F2FS_BLKSIZE - 1); + nm_i->nat_bits = kzalloc(nm_i->nat_bits_blocks << F2FS_BLKSIZE_BITS, + GFP_KERNEL); + if (!nm_i->nat_bits) + return -ENOMEM; + + nat_bits_addr = __start_cp_addr(sbi) + sbi->blocks_per_seg - + nm_i->nat_bits_blocks; + for (i = 0; i < nm_i->nat_bits_blocks; i++) { + struct page *page = get_meta_page(sbi, nat_bits_addr++); + + memcpy(nm_i->nat_bits + (i << F2FS_BLKSIZE_BITS), + page_address(page), F2FS_BLKSIZE); + f2fs_put_page(page, 1); + } + + cp_ver |= (crc << 32); + if (cpu_to_le64(cp_ver) != *(__le64 *)nm_i->nat_bits) { + disable_nat_bits(sbi, true); + return 0; + } + + nm_i->full_nat_bits = nm_i->nat_bits + 8; + nm_i->empty_nat_bits = nm_i->full_nat_bits + nat_bits_bytes; + + f2fs_msg(sbi->sb, KERN_NOTICE, "Found nat_bits in checkpoint"); + return 0; +} + static int init_node_manager(struct f2fs_sb_info *sbi) { struct f2fs_super_block *sb_raw = F2FS_RAW_SUPER(sbi); struct f2fs_nm_info *nm_i = NM_I(sbi); unsigned char *version_bitmap; - unsigned int nat_segs, nat_blocks; + unsigned int nat_segs; + int err; nm_i->nat_blkaddr = le32_to_cpu(sb_raw->nat_blkaddr); /* segment_count_nat includes pair segment so divide to 2. */ nat_segs = le32_to_cpu(sb_raw->segment_count_nat) >> 1; - nat_blocks = nat_segs << le32_to_cpu(sb_raw->log_blocks_per_seg); - - nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks; + nm_i->nat_blocks = nat_segs << le32_to_cpu(sb_raw->log_blocks_per_seg); + nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nm_i->nat_blocks; /* not used nids: 0, node, meta, (and root counted as valid node) */ nm_i->available_nids = nm_i->max_nid - sbi->total_valid_node_count - @@ -2394,6 +2545,10 @@ static int init_node_manager(struct f2fs_sb_info *sbi) if (!nm_i->nat_bitmap) return -ENOMEM; + err = __get_nat_bitmaps(sbi); + if (err) + return err; + #ifdef CONFIG_F2FS_CHECK_FS nm_i->nat_bitmap_mir = kmemdup(version_bitmap, nm_i->bitmap_size, GFP_KERNEL); @@ -2416,7 +2571,7 @@ int build_node_manager(struct f2fs_sb_info *sbi) if (err) return err; - build_free_nids(sbi, true); + build_free_nids(sbi, true, true); return 0; } @@ -2475,6 +2630,7 @@ void destroy_node_manager(struct f2fs_sb_info *sbi) up_write(&nm_i->nat_tree_lock); kfree(nm_i->nat_bitmap); + kfree(nm_i->nat_bits); #ifdef CONFIG_F2FS_CHECK_FS kfree(nm_i->nat_bitmap_mir); #endif diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 3e95db5375ed..9eb6d89bf9e2 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -386,7 +386,7 @@ void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi) if (!available_free_memory(sbi, FREE_NIDS)) try_to_free_nids(sbi, MAX_FREE_NIDS); else - build_free_nids(sbi, false); + build_free_nids(sbi, false, false); if (!is_idle(sbi)) return; diff --git a/include/linux/f2fs_fs.h b/include/linux/f2fs_fs.h index f0748524ca8c..1c92ace2e8f8 100644 --- a/include/linux/f2fs_fs.h +++ b/include/linux/f2fs_fs.h @@ -114,6 +114,7 @@ struct f2fs_super_block { /* * For checkpoint */ +#define CP_NAT_BITS_FLAG 0x00000080 #define CP_CRC_RECOVERY_FLAG 0x00000040 #define CP_FASTBOOT_FLAG 0x00000020 #define CP_FSCK_FLAG 0x00000010 -- cgit From 55523519bc7227e651fd4febeb3aafdd22b8af1c Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Sat, 25 Feb 2017 11:08:28 +0800 Subject: f2fs: show simple call stack in fault injection message Previously kernel message can show that in which function we do the injection, but unfortunately, most of the caller are the same, for tracking more information of injection path, it needs to show upper caller's name. This patch supports that ability. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/checkpoint.c | 1 + fs/f2fs/data.c | 4 +++- fs/f2fs/dir.c | 4 +++- fs/f2fs/f2fs.h | 20 +++++++++++++------- fs/f2fs/gc.c | 4 +++- fs/f2fs/inode.c | 4 +++- fs/f2fs/node.c | 4 +++- fs/f2fs/segment.c | 4 +++- 8 files changed, 32 insertions(+), 13 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c index cd7132121573..04d7c244c0f0 100644 --- a/fs/f2fs/checkpoint.c +++ b/fs/f2fs/checkpoint.c @@ -494,6 +494,7 @@ int acquire_orphan_inode(struct f2fs_sb_info *sbi) #ifdef CONFIG_F2FS_FAULT_INJECTION if (time_to_inject(sbi, FAULT_ORPHAN)) { spin_unlock(&im->ino_lock); + f2fs_show_injection_info(FAULT_ORPHAN); return -ENOSPC; } #endif diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index 9e51c5e40ce1..b0a2e3faabb2 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -55,8 +55,10 @@ static void f2fs_read_end_io(struct bio *bio) int i; #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(F2FS_P_SB(bio->bi_io_vec->bv_page), FAULT_IO)) + if (time_to_inject(F2FS_P_SB(bio->bi_io_vec->bv_page), FAULT_IO)) { + f2fs_show_injection_info(FAULT_IO); bio->bi_error = -EIO; + } #endif if (f2fs_bio_encrypted(bio)) { diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c index 54aa30ee028f..295a223ae11e 100644 --- a/fs/f2fs/dir.c +++ b/fs/f2fs/dir.c @@ -549,8 +549,10 @@ int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name, start: #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(F2FS_I_SB(dir), FAULT_DIR_DEPTH)) + if (time_to_inject(F2FS_I_SB(dir), FAULT_DIR_DEPTH)) { + f2fs_show_injection_info(FAULT_DIR_DEPTH); return -ENOSPC; + } #endif if (unlikely(current_depth == MAX_DIR_HASH_DEPTH)) return -ENOSPC; diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index d1156cdd211e..8c0916ae0bea 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -952,6 +952,10 @@ struct f2fs_sb_info { }; #ifdef CONFIG_F2FS_FAULT_INJECTION +#define f2fs_show_injection_info(type) \ + printk("%sF2FS-fs : inject %s in %s of %pF\n", \ + KERN_INFO, fault_name[type], \ + __func__, __builtin_return_address(0)) static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) { struct f2fs_fault_info *ffi = &sbi->fault_info; @@ -965,10 +969,6 @@ static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) atomic_inc(&ffi->inject_ops); if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { atomic_set(&ffi->inject_ops, 0); - printk("%sF2FS-fs : inject %s in %pF\n", - KERN_INFO, - fault_name[type], - __builtin_return_address(0)); return true; } return false; @@ -1277,8 +1277,10 @@ static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi, blkcnt_t diff; #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(sbi, FAULT_BLOCK)) + if (time_to_inject(sbi, FAULT_BLOCK)) { + f2fs_show_injection_info(FAULT_BLOCK); return false; + } #endif /* * let's increase this in prior to actual block count change in order @@ -1518,8 +1520,10 @@ static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, if (page) return page; - if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) + if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { + f2fs_show_injection_info(FAULT_PAGE_ALLOC); return NULL; + } #endif if (!for_write) return grab_cache_page(mapping, index); @@ -1995,8 +1999,10 @@ static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, size_t size, gfp_t flags) { #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(sbi, FAULT_KMALLOC)) + if (time_to_inject(sbi, FAULT_KMALLOC)) { + f2fs_show_injection_info(FAULT_KMALLOC); return NULL; + } #endif return kmalloc(size, flags); } diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c index 6c996e39b59a..8be5144da8e6 100644 --- a/fs/f2fs/gc.c +++ b/fs/f2fs/gc.c @@ -48,8 +48,10 @@ static int gc_thread_func(void *data) } #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(sbi, FAULT_CHECKPOINT)) + if (time_to_inject(sbi, FAULT_CHECKPOINT)) { + f2fs_show_injection_info(FAULT_CHECKPOINT); f2fs_stop_checkpoint(sbi, false); + } #endif /* diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c index af06bda51a54..24bb8213d974 100644 --- a/fs/f2fs/inode.c +++ b/fs/f2fs/inode.c @@ -373,8 +373,10 @@ void f2fs_evict_inode(struct inode *inode) goto no_delete; #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(sbi, FAULT_EVICT_INODE)) + if (time_to_inject(sbi, FAULT_EVICT_INODE)) { + f2fs_show_injection_info(FAULT_EVICT_INODE); goto no_delete; + } #endif remove_ino_entry(sbi, inode->i_ino, APPEND_INO); diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 43d35ec11851..24ea49f98891 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1987,8 +1987,10 @@ bool alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid) struct free_nid *i = NULL; retry: #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(sbi, FAULT_ALLOC_NID)) + if (time_to_inject(sbi, FAULT_ALLOC_NID)) { + f2fs_show_injection_info(FAULT_ALLOC_NID); return false; + } #endif spin_lock(&nm_i->nid_list_lock); diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 9eb6d89bf9e2..1bab09097590 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -352,8 +352,10 @@ int commit_inmem_pages(struct inode *inode) void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need) { #ifdef CONFIG_F2FS_FAULT_INJECTION - if (time_to_inject(sbi, FAULT_CHECKPOINT)) + if (time_to_inject(sbi, FAULT_CHECKPOINT)) { + f2fs_show_injection_info(FAULT_CHECKPOINT); f2fs_stop_checkpoint(sbi, false); + } #endif if (!need) -- cgit From d27c3d89db9986b6f48576169031247e4c893729 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Fri, 24 Feb 2017 18:46:00 +0800 Subject: f2fs: select target segment with closer temperature in SSR mode In SSR mode, we can allocate target segment which has different temperature type from the type of current block, in order to avoid mixing coldest and hottest data/node as much as possible, change SSR allocation policy to select closer temperature for current block prior. Signed-off-by: Yunlong Song Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 23 +++++++++++++++++------ 1 file changed, 17 insertions(+), 6 deletions(-) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 1bab09097590..9006d8ed6f52 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1541,7 +1541,8 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - int i, n; + int i, cnt; + bool reversed = false; /* need_SSR() already forces to do this */ if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) @@ -1549,14 +1550,24 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) /* For node segments, let's do SSR more intensively */ if (IS_NODESEG(type)) { - i = CURSEG_HOT_NODE; - n = CURSEG_COLD_NODE; + if (type >= CURSEG_WARM_NODE) { + reversed = true; + i = CURSEG_COLD_NODE; + } else { + i = CURSEG_HOT_NODE; + } + cnt = NR_CURSEG_NODE_TYPE; } else { - i = CURSEG_HOT_DATA; - n = CURSEG_COLD_DATA; + if (type >= CURSEG_WARM_DATA) { + reversed = true; + i = CURSEG_COLD_DATA; + } else { + i = CURSEG_HOT_DATA; + } + cnt = NR_CURSEG_DATA_TYPE; } - for (; i <= n; i++) { + for (; cnt-- > 0; reversed ? i-- : i++) { if (i == type) continue; if (v_ops->get_victim(sbi, &(curseg)->next_segno, -- cgit From 8b107f5b97772c7c0c218302e9a4d15b4edf50b4 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 27 Feb 2017 11:57:11 -0800 Subject: f2fs: avoid to issue redundant discard commands If segs_per_sec is over 1 like under SMR, previously f2fs issues discard commands redundantly on the same section, since we didn't move end position for the previous discard command. E.g., start end | | prefree_bitmap = [01111100111100] And, after issue discard for this section, end start | | prefree_bitmap = [01111100111100] Select this section again by searching from (end + 1), start end | | prefree_bitmap = [01111100111100] Fixes: 36abef4e796d38 ("f2fs: introduce mode=lfs mount option") Cc: Cc: Damien Le Moal Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 9006d8ed6f52..4bd7a8b19332 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1036,6 +1036,8 @@ next: start = start_segno + sbi->segs_per_sec; if (start < end) goto next; + else + end = start - 1; } mutex_unlock(&dirty_i->seglist_lock); -- cgit From 720037f939fa50fc3531035ae61b4cf4b0ff35e5 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 6 Mar 2017 11:59:56 -0800 Subject: f2fs: don't overwrite node block by SSR This patch fixes that SSR can overwrite previous warm node block consisting of a node chain since the last checkpoint. Fixes: 5b6c6be2d878 ("f2fs: use SSR for warm node as well") Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/segment.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs/f2fs/segment.c') diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 4bd7a8b19332..29ef7088c558 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1163,6 +1163,12 @@ static void update_sit_entry(struct f2fs_sb_info *sbi, block_t blkaddr, int del) if (f2fs_discard_en(sbi) && !f2fs_test_and_set_bit(offset, se->discard_map)) sbi->discard_blks--; + + /* don't overwrite by SSR to keep node chain */ + if (se->type == CURSEG_WARM_NODE) { + if (!f2fs_test_and_set_bit(offset, se->ckpt_valid_map)) + se->ckpt_valid_blocks++; + } } else { if (!f2fs_test_and_clear_bit(offset, se->cur_valid_map)) { #ifdef CONFIG_F2FS_CHECK_FS -- cgit