From a014e037be26b5c9ee6fb4e49e7804141cf3bb89 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Fri, 20 Jun 2014 21:44:02 -0700 Subject: f2fs: clean up an unused parameter and assignment This patch cleans up simple unnecessary codes. Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 4b697ccc9b0c..de709f0a445e 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -845,7 +845,7 @@ void remove_inode_page(struct inode *inode) truncate_node(&dn); } -struct page *new_inode_page(struct inode *inode, const struct qstr *name) +struct page *new_inode_page(struct inode *inode) { struct dnode_of_data dn; -- cgit v1.2.3 From aec71382c68135261ef6efc3d8a96b7149939446 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 24 Jun 2014 09:18:20 +0800 Subject: f2fs: refactor flush_nat_entries codes for reducing NAT writes Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time. In this patch we merge dirty entries located in same NAT block to nat entry set, and linked all set to list, sorted ascending order by entries' count of set. Later we flush entries in sparse set into journal as many as we can, and then flush merged entries to disk. In this way we can not only gain in performance, but also save lifetime of flash device. In my testing environment, it shows this patch can help to reduce NAT block writes obviously. In hard disk test case: cost time of fsstress is stablely reduced by about 5%. 1. virtual machine + hard disk: fsstress -p 20 -n 200 -l 5 node num cp count nodes/cp based 4599.6 1803.0 2.551 patched 2714.6 1829.6 1.483 2. virtual machine + 32g micro SD card: fsstress -p 20 -n 200 -l 1 -w -f chown=0 -f creat=4 -f dwrite=0 -f fdatasync=4 -f fsync=4 -f link=0 -f mkdir=4 -f mknod=4 -f rename=5 -f rmdir=5 -f symlink=0 -f truncate=4 -f unlink=5 -f write=0 -S node num cp count nodes/cp based 84.5 43.7 1.933 patched 49.2 40.0 1.23 Our latency of merging op shows not bad when handling extreme case like: merging a great number of dirty nats: latency(ns) dirty nat count 3089219 24922 5129423 27422 4000250 24523 change log from v1: o fix wrong logic in add_nat_entry when grab a new nat entry set. o swith to create slab cache in create_node_manager_caches. o use GFP_ATOMIC instead of GFP_NOFS to avoid potential long latency. change log from v2: o make comment position more appropriate suggested by Jaegeuk Kim. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 263 +++++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 179 insertions(+), 84 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index de709f0a445e..a90f51d32482 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -25,6 +25,7 @@ static struct kmem_cache *nat_entry_slab; static struct kmem_cache *free_nid_slab; +static struct kmem_cache *nat_entry_set_slab; bool available_free_memory(struct f2fs_sb_info *sbi, int type) { @@ -90,12 +91,8 @@ static struct page *get_next_nat_page(struct f2fs_sb_info *sbi, nid_t nid) /* get current nat block page with lock */ src_page = get_meta_page(sbi, src_off); - - /* Dirty src_page means that it is already the new target NAT page. */ - if (PageDirty(src_page)) - return src_page; - dst_page = grab_meta_page(sbi, dst_off); + f2fs_bug_on(PageDirty(src_page)); src_addr = page_address(src_page); dst_addr = page_address(dst_page); @@ -1744,7 +1741,90 @@ skip: return err; } -static bool flush_nats_in_journal(struct f2fs_sb_info *sbi) +static struct nat_entry_set *grab_nat_entry_set(void) +{ + struct nat_entry_set *nes = + f2fs_kmem_cache_alloc(nat_entry_set_slab, GFP_ATOMIC); + + nes->entry_cnt = 0; + INIT_LIST_HEAD(&nes->set_list); + INIT_LIST_HEAD(&nes->entry_list); + return nes; +} + +static void release_nat_entry_set(struct nat_entry_set *nes, + struct f2fs_nm_info *nm_i) +{ + f2fs_bug_on(!list_empty(&nes->entry_list)); + + nm_i->dirty_nat_cnt -= nes->entry_cnt; + list_del(&nes->set_list); + kmem_cache_free(nat_entry_set_slab, nes); +} + +static void adjust_nat_entry_set(struct nat_entry_set *nes, + struct list_head *head) +{ + struct nat_entry_set *next = nes; + + if (list_is_last(&nes->set_list, head)) + return; + + list_for_each_entry_continue(next, head, set_list) + if (nes->entry_cnt <= next->entry_cnt) + break; + + list_move_tail(&nes->set_list, &next->set_list); +} + +static void add_nat_entry(struct nat_entry *ne, struct list_head *head) +{ + struct nat_entry_set *nes; + nid_t start_nid = START_NID(ne->ni.nid); + + list_for_each_entry(nes, head, set_list) { + if (nes->start_nid == start_nid) { + list_move_tail(&ne->list, &nes->entry_list); + nes->entry_cnt++; + adjust_nat_entry_set(nes, head); + return; + } + } + + nes = grab_nat_entry_set(); + + nes->start_nid = start_nid; + list_move_tail(&ne->list, &nes->entry_list); + nes->entry_cnt++; + list_add(&nes->set_list, head); +} + +static void merge_nats_in_set(struct f2fs_sb_info *sbi) +{ + struct f2fs_nm_info *nm_i = NM_I(sbi); + struct list_head *dirty_list = &nm_i->dirty_nat_entries; + struct list_head *set_list = &nm_i->nat_entry_set; + struct nat_entry *ne, *tmp; + + write_lock(&nm_i->nat_tree_lock); + list_for_each_entry_safe(ne, tmp, dirty_list, list) { + if (nat_get_blkaddr(ne) == NEW_ADDR) + continue; + add_nat_entry(ne, set_list); + nm_i->dirty_nat_cnt++; + } + write_unlock(&nm_i->nat_tree_lock); +} + +static bool __has_cursum_space(struct f2fs_summary_block *sum, int size) +{ + if (nats_in_cursum(sum) + size <= NAT_JOURNAL_ENTRIES) + return true; + else + return false; +} + +static void remove_nats_in_journal(struct f2fs_sb_info *sbi) { struct f2fs_nm_info *nm_i = NM_I(sbi); struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA); @@ -1752,12 +1832,6 @@ static bool flush_nats_in_journal(struct f2fs_sb_info *sbi) int i; mutex_lock(&curseg->curseg_mutex); - - if (nats_in_cursum(sum) < NAT_JOURNAL_ENTRIES) { - mutex_unlock(&curseg->curseg_mutex); - return false; - } - for (i = 0; i < nats_in_cursum(sum); i++) { struct nat_entry *ne; struct f2fs_nat_entry raw_ne; @@ -1767,23 +1841,21 @@ static bool flush_nats_in_journal(struct f2fs_sb_info *sbi) retry: write_lock(&nm_i->nat_tree_lock); ne = __lookup_nat_cache(nm_i, nid); - if (ne) { - __set_nat_cache_dirty(nm_i, ne); - write_unlock(&nm_i->nat_tree_lock); - continue; - } + if (ne) + goto found; + ne = grab_nat_entry(nm_i, nid); if (!ne) { write_unlock(&nm_i->nat_tree_lock); goto retry; } node_info_from_raw_nat(&ne->ni, &raw_ne); +found: __set_nat_cache_dirty(nm_i, ne); write_unlock(&nm_i->nat_tree_lock); } update_nats_in_cursum(sum, -i); mutex_unlock(&curseg->curseg_mutex); - return true; } /* @@ -1794,80 +1866,91 @@ void flush_nat_entries(struct f2fs_sb_info *sbi) struct f2fs_nm_info *nm_i = NM_I(sbi); struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA); struct f2fs_summary_block *sum = curseg->sum_blk; - struct nat_entry *ne, *cur; - struct page *page = NULL; - struct f2fs_nat_block *nat_blk = NULL; - nid_t start_nid = 0, end_nid = 0; - bool flushed; - - flushed = flush_nats_in_journal(sbi); + struct nat_entry_set *nes, *tmp; + struct list_head *head = &nm_i->nat_entry_set; + bool to_journal = true; - if (!flushed) - mutex_lock(&curseg->curseg_mutex); + /* merge nat entries of dirty list to nat entry set temporarily */ + merge_nats_in_set(sbi); - /* 1) flush dirty nat caches */ - list_for_each_entry_safe(ne, cur, &nm_i->dirty_nat_entries, list) { - nid_t nid; - struct f2fs_nat_entry raw_ne; - int offset = -1; - - if (nat_get_blkaddr(ne) == NEW_ADDR) - continue; - - nid = nat_get_nid(ne); + /* + * if there are no enough space in journal to store dirty nat + * entries, remove all entries from journal and merge them + * into nat entry set. + */ + if (!__has_cursum_space(sum, nm_i->dirty_nat_cnt)) { + remove_nats_in_journal(sbi); - if (flushed) - goto to_nat_page; + /* + * merge nat entries of dirty list to nat entry set temporarily + */ + merge_nats_in_set(sbi); + } - /* if there is room for nat enries in curseg->sumpage */ - offset = lookup_journal_in_cursum(sum, NAT_JOURNAL, nid, 1); - if (offset >= 0) { - raw_ne = nat_in_journal(sum, offset); - goto flush_now; - } -to_nat_page: - if (!page || (start_nid > nid || nid > end_nid)) { - if (page) { - f2fs_put_page(page, 1); - page = NULL; - } - start_nid = START_NID(nid); - end_nid = start_nid + NAT_ENTRY_PER_BLOCK - 1; + if (!nm_i->dirty_nat_cnt) + return; - /* - * get nat block with dirty flag, increased reference - * count, mapped and lock - */ + /* + * there are two steps to flush nat entries: + * #1, flush nat entries to journal in current hot data summary block. + * #2, flush nat entries to nat page. + */ + list_for_each_entry_safe(nes, tmp, head, set_list) { + struct f2fs_nat_block *nat_blk; + struct nat_entry *ne, *cur; + struct page *page; + nid_t start_nid = nes->start_nid; + + if (to_journal && !__has_cursum_space(sum, nes->entry_cnt)) + to_journal = false; + + if (to_journal) { + mutex_lock(&curseg->curseg_mutex); + } else { page = get_next_nat_page(sbi, start_nid); nat_blk = page_address(page); + f2fs_bug_on(!nat_blk); } - f2fs_bug_on(!nat_blk); - raw_ne = nat_blk->entries[nid - start_nid]; -flush_now: - raw_nat_from_node_info(&raw_ne, &ne->ni); - - if (offset < 0) { - nat_blk->entries[nid - start_nid] = raw_ne; - } else { - nat_in_journal(sum, offset) = raw_ne; - nid_in_journal(sum, offset) = cpu_to_le32(nid); - } + /* flush dirty nats in nat entry set */ + list_for_each_entry_safe(ne, cur, &nes->entry_list, list) { + struct f2fs_nat_entry *raw_ne; + nid_t nid = nat_get_nid(ne); + int offset; + + if (to_journal) { + offset = lookup_journal_in_cursum(sum, + NAT_JOURNAL, nid, 1); + f2fs_bug_on(offset < 0); + raw_ne = &nat_in_journal(sum, offset); + nid_in_journal(sum, offset) = cpu_to_le32(nid); + } else { + raw_ne = &nat_blk->entries[nid - start_nid]; + } + raw_nat_from_node_info(raw_ne, &ne->ni); - if (nat_get_blkaddr(ne) == NULL_ADDR && + if (nat_get_blkaddr(ne) == NULL_ADDR && add_free_nid(sbi, nid, false) <= 0) { - write_lock(&nm_i->nat_tree_lock); - __del_from_nat_cache(nm_i, ne); - write_unlock(&nm_i->nat_tree_lock); - } else { - write_lock(&nm_i->nat_tree_lock); - __clear_nat_cache_dirty(nm_i, ne); - write_unlock(&nm_i->nat_tree_lock); + write_lock(&nm_i->nat_tree_lock); + __del_from_nat_cache(nm_i, ne); + write_unlock(&nm_i->nat_tree_lock); + } else { + write_lock(&nm_i->nat_tree_lock); + __clear_nat_cache_dirty(nm_i, ne); + write_unlock(&nm_i->nat_tree_lock); + } } + + if (to_journal) + mutex_unlock(&curseg->curseg_mutex); + else + f2fs_put_page(page, 1); + + release_nat_entry_set(nes, nm_i); } - if (!flushed) - mutex_unlock(&curseg->curseg_mutex); - f2fs_put_page(page, 1); + + f2fs_bug_on(!list_empty(head)); + f2fs_bug_on(nm_i->dirty_nat_cnt); } static int init_node_manager(struct f2fs_sb_info *sbi) @@ -1896,6 +1979,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi) INIT_RADIX_TREE(&nm_i->nat_root, GFP_ATOMIC); INIT_LIST_HEAD(&nm_i->nat_entries); INIT_LIST_HEAD(&nm_i->dirty_nat_entries); + INIT_LIST_HEAD(&nm_i->nat_entry_set); mutex_init(&nm_i->build_lock); spin_lock_init(&nm_i->free_nid_list_lock); @@ -1976,19 +2060,30 @@ int __init create_node_manager_caches(void) nat_entry_slab = f2fs_kmem_cache_create("nat_entry", sizeof(struct nat_entry)); if (!nat_entry_slab) - return -ENOMEM; + goto fail; free_nid_slab = f2fs_kmem_cache_create("free_nid", sizeof(struct free_nid)); - if (!free_nid_slab) { - kmem_cache_destroy(nat_entry_slab); - return -ENOMEM; - } + if (!free_nid_slab) + goto destory_nat_entry; + + nat_entry_set_slab = f2fs_kmem_cache_create("nat_entry_set", + sizeof(struct nat_entry_set)); + if (!nat_entry_set_slab) + goto destory_free_nid; return 0; + +destory_free_nid: + kmem_cache_destroy(free_nid_slab); +destory_nat_entry: + kmem_cache_destroy(nat_entry_slab); +fail: + return -ENOMEM; } void destroy_node_manager_caches(void) { + kmem_cache_destroy(nat_entry_set_slab); kmem_cache_destroy(free_nid_slab); kmem_cache_destroy(nat_entry_slab); } -- cgit v1.2.3 From b3582c68920105e29d219714d8a6fbde25a43379 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Thu, 3 Jul 2014 18:58:39 +0800 Subject: f2fs: reduce competition among node page writes We do not need to block on ->node_write among different node page writers e.g. fsync/flush, unless we have a node page writer from write_checkpoint. So it's better use rw_semaphore instead of mutex type for ->node_write to promote performance. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index a90f51d32482..7b5b5def65fe 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1231,12 +1231,12 @@ static int f2fs_write_node_page(struct page *page, if (wbc->for_reclaim) goto redirty_out; - mutex_lock(&sbi->node_write); + down_read(&sbi->node_write); set_page_writeback(page); write_node_page(sbi, page, &fio, nid, ni.blk_addr, &new_addr); set_node_addr(sbi, &ni, new_addr, is_fsync_dnode(page)); dec_page_count(sbi, F2FS_DIRTY_NODES); - mutex_unlock(&sbi->node_write); + up_read(&sbi->node_write); unlock_page(page); return 0; -- cgit v1.2.3 From 70cfed88efa760fd165fc413cfd1801b5cc8acd2 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Sat, 2 Aug 2014 15:26:04 +0800 Subject: f2fs: avoid skipping recover_inline_xattr after recover_inline_data When we recover data of inode in roll-forward procedure, and the inode has both inline data and inline xattr. We may skip recovering inline xattr if we recover inline data form node page first. This patch will fix the problem that we lost inline xattr data in above scenario. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 7b5b5def65fe..d3d90d284631 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1549,7 +1549,7 @@ void recover_node_page(struct f2fs_sb_info *sbi, struct page *page, clear_node_page_dirty(page); } -static void recover_inline_xattr(struct inode *inode, struct page *page) +void recover_inline_xattr(struct inode *inode, struct page *page) { struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb); void *src_addr, *dst_addr; @@ -1588,8 +1588,6 @@ bool recover_xattr_data(struct inode *inode, struct page *page, block_t blkaddr) nid_t new_xnid = nid_of_node(page); struct node_info ni; - recover_inline_xattr(inode, page); - if (!f2fs_has_xattr_block(ofs_of_node(page))) return false; -- cgit v1.2.3 From e1c42045203071c4634b89e696037357810d3083 Mon Sep 17 00:00:00 2001 From: arter97 Date: Wed, 6 Aug 2014 23:22:50 +0900 Subject: f2fs: fix typo Fix typo and some grammatical errors. The words "filesystem" and "readahead" are being used without the space treewide. Signed-off-by: Park Ju Hyung Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index d3d90d284631..1f3329907483 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -237,7 +237,7 @@ retry: nat_get_blkaddr(e) != NULL_ADDR && new_blkaddr == NEW_ADDR); - /* increament version no as node is removed */ + /* increment version no as node is removed */ if (nat_get_blkaddr(e) != NEW_ADDR && new_blkaddr == NULL_ADDR) { unsigned char version = nat_get_version(e); nat_set_version(e, inc_node_version(version)); @@ -274,7 +274,7 @@ int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink) } /* - * This function returns always success + * This function always returns success */ void get_node_info(struct f2fs_sb_info *sbi, nid_t nid, struct node_info *ni) { @@ -650,7 +650,7 @@ static int truncate_partial_nodes(struct dnode_of_data *dn, /* get indirect nodes in the path */ for (i = 0; i < idx + 1; i++) { - /* refernece count'll be increased */ + /* reference count'll be increased */ pages[i] = get_node_page(sbi, nid[i]); if (IS_ERR(pages[i])) { err = PTR_ERR(pages[i]); @@ -836,7 +836,7 @@ void remove_inode_page(struct inode *inode) f2fs_put_page(page, 1); return; } - /* 0 is possible, after f2fs_new_inode() is failed */ + /* 0 is possible, after f2fs_new_inode() has failed */ f2fs_bug_on(inode->i_blocks != 0 && inode->i_blocks != 1); set_new_dnode(&dn, inode, page, page, ino); truncate_node(&dn); @@ -1637,7 +1637,7 @@ int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page) if (!ipage) return -ENOMEM; - /* Should not use this inode from free nid list */ + /* Should not use this inode from free nid list */ remove_free_nid(NM_I(sbi), ino); SetPageUptodate(ipage); @@ -1665,7 +1665,7 @@ int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page) /* * ra_sum_pages() merge contiguous pages into one bio and submit. - * these pre-readed pages are alloced in bd_inode's mapping tree. + * these pre-read pages are allocated in bd_inode's mapping tree. */ static int ra_sum_pages(struct f2fs_sb_info *sbi, struct page **pages, int start, int nrpages) @@ -1709,7 +1709,7 @@ int restore_node_summary(struct f2fs_sb_info *sbi, for (i = 0; !err && i < last_offset; i += nrpages, addr += nrpages) { nrpages = min(last_offset - i, bio_blocks); - /* read ahead node pages */ + /* readahead node pages */ nrpages = ra_sum_pages(sbi, pages, addr, nrpages); if (!nrpages) return -ENOMEM; -- cgit v1.2.3 From 617deb8c053aeec06c7aa16ac7225f046fab95e8 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 7 Aug 2014 17:04:24 -0700 Subject: f2fs: fix the initial inode page for recovery If a new inode page is needed for recover_dentry, we should assing i_inline as zero. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 1f3329907483..093d7991b7ed 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1651,6 +1651,7 @@ int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page) dst->i_blocks = cpu_to_le64(1); dst->i_links = cpu_to_le32(1); dst->i_xattr_nid = 0; + dst->i_inline = src->i_inline & F2FS_INLINE_XATTR; new_ni = old_ni; new_ni.ino = ino; @@ -1659,6 +1660,7 @@ int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page) WARN_ON(1); set_node_addr(sbi, &new_ni, NEW_ADDR, false); inc_valid_inode_count(sbi); + set_page_dirty(ipage); f2fs_put_page(ipage, 1); return 0; } -- cgit v1.2.3 From e3b4d43f7c233c6fce21fe4b4cb55b6d59afddae Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 7 Aug 2014 23:45:42 -0700 Subject: f2fs: should clear the inline_xattr flag During the recovery, we should clear the inline_xattr flag if its xattr node block is recovered. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 093d7991b7ed..151045f3e7b7 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1557,26 +1557,25 @@ void recover_inline_xattr(struct inode *inode, struct page *page) struct page *ipage; struct f2fs_inode *ri; - if (!f2fs_has_inline_xattr(inode)) - return; - if (!IS_INODE(page)) return; - ri = F2FS_INODE(page); - if (!(ri->i_inline & F2FS_INLINE_XATTR)) - return; - ipage = get_node_page(sbi, inode->i_ino); f2fs_bug_on(IS_ERR(ipage)); + ri = F2FS_INODE(page); + if (!(ri->i_inline & F2FS_INLINE_XATTR)) { + clear_inode_flag(F2FS_I(inode), FI_INLINE_XATTR); + goto update_inode; + } + dst_addr = inline_xattr_addr(ipage); src_addr = inline_xattr_addr(page); inline_size = inline_xattr_size(inode); f2fs_wait_on_page_writeback(ipage, NODE); memcpy(dst_addr, src_addr, inline_size); - +update_inode: update_inode(inode, ipage); f2fs_put_page(ipage, 1); } -- cgit v1.2.3 From 1c35a90e8ab57cd34b8e806b9c75ba05b3b5c7a3 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 7 Aug 2014 23:49:17 -0700 Subject: f2fs: fix to recover inline_xattr/data and blocks This patch fixes not to skip xattr recovery and inline xattr/data recovery order. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 151045f3e7b7..c80e3d59314d 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1557,9 +1557,6 @@ void recover_inline_xattr(struct inode *inode, struct page *page) struct page *ipage; struct f2fs_inode *ri; - if (!IS_INODE(page)) - return; - ipage = get_node_page(sbi, inode->i_ino); f2fs_bug_on(IS_ERR(ipage)); @@ -1580,16 +1577,13 @@ update_inode: f2fs_put_page(ipage, 1); } -bool recover_xattr_data(struct inode *inode, struct page *page, block_t blkaddr) +void recover_xattr_data(struct inode *inode, struct page *page, block_t blkaddr) { struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb); nid_t prev_xnid = F2FS_I(inode)->i_xattr_nid; nid_t new_xnid = nid_of_node(page); struct node_info ni; - if (!f2fs_has_xattr_block(ofs_of_node(page))) - return false; - /* 1: invalidate the previous xattr nid */ if (!prev_xnid) goto recover_xnid; @@ -1617,7 +1611,6 @@ recover_xnid: set_node_addr(sbi, &ni, blkaddr, false); update_inode_page(inode); - return true; } int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page) -- cgit v1.2.3 From 5274651927a76c947469a589e3d2a9adbd075da6 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 11 Aug 2014 18:18:36 -0700 Subject: f2fs: unlock_page when node page is redirtied out This patch fixes missing unlock_page when a node page is redirtied out. Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index c80e3d59314d..9f126f80813d 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1129,8 +1129,11 @@ continue_unlock: set_fsync_mark(page, 0); set_dentry_mark(page, 0); } - NODE_MAPPING(sbi)->a_ops->writepage(page, wbc); - wrote++; + + if (NODE_MAPPING(sbi)->a_ops->writepage(page, wbc)) + unlock_page(page); + else + wrote++; if (--wbc->nr_to_write == 0) break; -- cgit v1.2.3 From cf779cab14d50a84b61399f758da269654b863db Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 11 Aug 2014 18:37:46 -0700 Subject: f2fs: handle EIO not to break fs consistency There are two rules when EIO is occurred. 1. don't write any checkpoint data to preserve the previous checkpoint 2. don't lose the cached dentry/node/meta pages So, at first, this patch adds set_page_dirty in f2fs_write_end_io's failure. Then, writing checkpoint/dentry/node blocks is not allowed. Note that, for the data pages, we can't just throw away by redirtying them. Otherwise, kworker can fall into infinite loop to flush them. (Ref. xfstests/019) Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 9f126f80813d..d2f784283425 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1215,6 +1215,8 @@ static int f2fs_write_node_page(struct page *page, if (unlikely(sbi->por_doing)) goto redirty_out; + if (unlikely(f2fs_cp_error(sbi))) + goto redirty_out; f2fs_wait_on_page_writeback(page, NODE); -- cgit v1.2.3 From 202095a7a0ec075b924cb15dde330bf76e485f61 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Fri, 15 Aug 2014 09:56:46 -0700 Subject: f2fs: remove rewrite_node_page I think we need to let the dirty node pages remain in the page cache instead of rewriting them in their places. So, after done with successful recovery, write_checkpoint will flush all of them through the normal write path. Through this, we can avoid potential error cases in terms of block allocation. Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 9 --------- 1 file changed, 9 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index d2f784283425..b4d964029fc7 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1545,15 +1545,6 @@ void alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid) kmem_cache_free(free_nid_slab, i); } -void recover_node_page(struct f2fs_sb_info *sbi, struct page *page, - struct f2fs_summary *sum, struct node_info *ni, - block_t new_blkaddr) -{ - rewrite_node_page(sbi, page, sum, ni->blk_addr, new_blkaddr); - set_node_addr(sbi, ni, new_blkaddr, false); - clear_node_page_dirty(page); -} - void recover_inline_xattr(struct inode *inode, struct page *page) { struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb); -- cgit v1.2.3 From c200b1aa6cb460ce8c3ecf6fdc690d3949c3cc5d Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Wed, 20 Aug 2014 18:36:46 +0800 Subject: f2fs: fix incorrect calculation with total/free inode num Theoretically, our total inodes number is the same as total node number, but there are three node ids are reserved in f2fs, they are 0, 1 (node nid), and 2 (meta nid), and they should never be used by user, so our total/free inode number calculated in ->statfs is wrong. This patch indroduces F2FS_RESERVED_NODE_NUM and then fixes this issue by recalculating total/free inode number with the macro. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index b4d964029fc7..044395c20ee9 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1957,7 +1957,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi) nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks; /* not used nids: 0, node, meta, (and root counted as valid node) */ - nm_i->available_nids = nm_i->max_nid - 3; + nm_i->available_nids = nm_i->max_nid - F2FS_RESERVED_NODE_NUM; nm_i->fcnt = 0; nm_i->nat_cnt = 0; nm_i->ram_thresh = DEF_RAM_THRESHOLD; -- cgit v1.2.3 From c2e69583a4787b252f6be9a9daea4662eebc26f8 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 25 Aug 2014 14:45:59 -0700 Subject: f2fs: truncate stale block for inline_data This verifies to truncate any allocated blocks, offset[0], by inline_data. Not figured out, but for making sure. Signed-off-by: Jaegeuk Kim --- fs/f2fs/node.c | 20 ++++++++++++-------- 1 file changed, 12 insertions(+), 8 deletions(-) (limited to 'fs/f2fs/node.c') diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 044395c20ee9..45378196e19a 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -823,22 +823,26 @@ int truncate_xattr_node(struct inode *inode, struct page *page) */ void remove_inode_page(struct inode *inode) { - struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb); - struct page *page; - nid_t ino = inode->i_ino; struct dnode_of_data dn; - page = get_node_page(sbi, ino); - if (IS_ERR(page)) + set_new_dnode(&dn, inode, NULL, NULL, inode->i_ino); + if (get_dnode_of_data(&dn, 0, LOOKUP_NODE)) return; - if (truncate_xattr_node(inode, page)) { - f2fs_put_page(page, 1); + if (truncate_xattr_node(inode, dn.inode_page)) { + f2fs_put_dnode(&dn); return; } + + /* remove potential inline_data blocks */ + if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || + S_ISLNK(inode->i_mode)) + truncate_data_blocks_range(&dn, 1); + /* 0 is possible, after f2fs_new_inode() has failed */ f2fs_bug_on(inode->i_blocks != 0 && inode->i_blocks != 1); - set_new_dnode(&dn, inode, page, page, ino); + + /* will put inode & node pages */ truncate_node(&dn); } -- cgit v1.2.3