From 4559071063270999d016c92a0b9241692cbbb522 Mon Sep 17 00:00:00 2001 From: Gu Zheng Date: Mon, 15 Jul 2013 17:57:38 +0800 Subject: f2fs: introduce help function F2FS_NODE() Introduce help function F2FS_NODE() to simplify the conversion of node_page to f2fs_node. Signed-off-by: Gu Zheng Signed-off-by: Jaegeuk Kim --- fs/f2fs/inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/inode.c') diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c index 2b2d45d19e3..debf7430804 100644 --- a/fs/f2fs/inode.c +++ b/fs/f2fs/inode.c @@ -56,7 +56,7 @@ static int do_read_inode(struct inode *inode) if (IS_ERR(node_page)) return PTR_ERR(node_page); - rn = page_address(node_page); + rn = F2FS_NODE(node_page); ri = &(rn->i); inode->i_mode = le16_to_cpu(ri->i_mode); @@ -153,7 +153,7 @@ void update_inode(struct inode *inode, struct page *node_page) wait_on_page_writeback(node_page); - rn = page_address(node_page); + rn = F2FS_NODE(node_page); ri = &(rn->i); ri->i_mode = cpu_to_le16(inode->i_mode); -- cgit v1.2.3-70-g09d2 From a569469e967022d9ceeaa4b73619f96614087d2d Mon Sep 17 00:00:00 2001 From: Jin Xu Date: Mon, 5 Aug 2013 20:02:04 +0800 Subject: f2fs: fix a deadlock in fsync This patch fixes a deadlock bug that occurs quite often when there are concurrent write and fsync on a same file. Following is the simplified call trace when tasks get hung. fsync thread: - f2fs_sync_file ... - f2fs_write_data_pages ... - update_extent_cache ... - update_inode - wait_on_page_writeback bdi writeback thread - __writeback_single_inode - f2fs_write_data_pages - mutex_lock(sbi->writepages) The deadlock happens when the fsync thread waits on a inode page that has been added to the f2fs' cached bio sbi->bio[NODE], and unfortunately, no one else could be able to submit the cached bio to block layer for writeback. This is because the fsync thread already hold a sbi->fs_lock and the sbi->writepages lock, causing the bdi thread being blocked when attempt to write data pages for the same inode. At the same time, f2fs_gc thread does not notice the situation and could not help. Even the sync syscall gets blocked. To fix it, we could submit the cached bio first before waiting on a inode page that is being written back. Signed-off-by: Jin Xu [Jaegeuk Kim: add more cases to use f2fs_wait_on_page_writeback] Signed-off-by: Jaegeuk Kim --- fs/f2fs/data.c | 2 +- fs/f2fs/f2fs.h | 3 ++- fs/f2fs/gc.c | 8 ++------ fs/f2fs/inode.c | 2 +- fs/f2fs/segment.c | 10 ++++++++++ 5 files changed, 16 insertions(+), 9 deletions(-) (limited to 'fs/f2fs/inode.c') diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index f458883af81..a7eb5292572 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -37,7 +37,7 @@ static void __set_data_blkaddr(struct dnode_of_data *dn, block_t new_addr) struct page *node_page = dn->node_page; unsigned int ofs_in_node = dn->ofs_in_node; - wait_on_page_writeback(node_page); + f2fs_wait_on_page_writeback(node_page, NODE, false); rn = F2FS_NODE(node_page); diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 63813befdd8..13db10b70e6 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -1023,7 +1023,8 @@ int npages_for_summary_flush(struct f2fs_sb_info *); void allocate_new_segments(struct f2fs_sb_info *); struct page *get_sum_page(struct f2fs_sb_info *, unsigned int); struct bio *f2fs_bio_alloc(struct block_device *, int); -void f2fs_submit_bio(struct f2fs_sb_info *, enum page_type, bool sync); +void f2fs_submit_bio(struct f2fs_sb_info *, enum page_type, bool); +void f2fs_wait_on_page_writeback(struct page *, enum page_type, bool); void write_meta_page(struct f2fs_sb_info *, struct page *); void write_node_page(struct f2fs_sb_info *, struct page *, unsigned int, block_t, block_t *); diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c index d286d8be8e6..e6b3ffd5ff6 100644 --- a/fs/f2fs/gc.c +++ b/fs/f2fs/gc.c @@ -422,8 +422,7 @@ next_step: /* set page dirty and write it */ if (gc_type == FG_GC) { - f2fs_submit_bio(sbi, NODE, true); - wait_on_page_writeback(node_page); + f2fs_wait_on_page_writeback(node_page, NODE, true); set_page_dirty(node_page); } else { if (!PageWriteback(node_page)) @@ -523,10 +522,7 @@ static void move_data_page(struct inode *inode, struct page *page, int gc_type) } else { struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb); - if (PageWriteback(page)) { - f2fs_submit_bio(sbi, DATA, true); - wait_on_page_writeback(page); - } + f2fs_wait_on_page_writeback(page, DATA, true); if (clear_page_dirty_for_io(page) && S_ISDIR(inode->i_mode)) { diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c index debf7430804..9ab81e7472c 100644 --- a/fs/f2fs/inode.c +++ b/fs/f2fs/inode.c @@ -151,7 +151,7 @@ void update_inode(struct inode *inode, struct page *node_page) struct f2fs_node *rn; struct f2fs_inode *ri; - wait_on_page_writeback(node_page); + f2fs_wait_on_page_writeback(node_page, NODE, false); rn = F2FS_NODE(node_page); ri = &(rn->i); diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 9b74ae2137d..68e344f9e04 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -705,6 +705,16 @@ retry: trace_f2fs_submit_write_page(page, blk_addr, type); } +void f2fs_wait_on_page_writeback(struct page *page, + enum page_type type, bool sync) +{ + struct f2fs_sb_info *sbi = F2FS_SB(page->mapping->host->i_sb); + if (PageWriteback(page)) { + f2fs_submit_bio(sbi, type, sync); + wait_on_page_writeback(page); + } +} + static bool __has_curseg_space(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); -- cgit v1.2.3-70-g09d2 From 92c4342fb72a6baf9ee9fcd079b46ed0286ebe33 Mon Sep 17 00:00:00 2001 From: Jin Xu Date: Thu, 15 Aug 2013 19:17:01 +0800 Subject: f2fs: avoid writing inode redundantly when creating a file In f2fs_write_inode, updating inode after f2fs_balance_fs is not a optimized way in the case that f2fs_gc is performed ahead. The inode page will be unnecessarily written out twice, one of which is in f2fs_gc->...->sync_node_pages and the other is in update_inode_page. Let's update the inode page in prior to f2fs_balance_fs to avoid this. To reproduce it, $ touch file (before this step, should make the device need f2fs_gc) $ sync (or wait the bdi to write dirty inode) Signed-off-by: Jin Xu Signed-off-by: Jaegeuk Kim --- fs/f2fs/inode.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'fs/f2fs/inode.c') diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c index 9ab81e7472c..7f8569bd875 100644 --- a/fs/f2fs/inode.c +++ b/fs/f2fs/inode.c @@ -221,9 +221,6 @@ int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc) if (!is_inode_flag_set(F2FS_I(inode), FI_DIRTY_INODE)) return 0; - if (wbc) - f2fs_balance_fs(sbi); - /* * We need to lock here to prevent from producing dirty node pages * during the urgent cleaning time when runing out of free sections. @@ -231,6 +228,10 @@ int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc) ilock = mutex_lock_op(sbi); ret = update_inode_page(inode); mutex_unlock_op(sbi, ilock); + + if (wbc) + f2fs_balance_fs(sbi); + return ret; } -- cgit v1.2.3-70-g09d2 From 444c580f7e9ad29927a5d5269d576bd7cdccebb8 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 8 Aug 2013 15:16:22 +0900 Subject: f2fs: add flags for inline xattrs This patch adds basic inode flags for inline xattrs, F2FS_INLINE_XATTR, and add a mount option, inline_xattr, which is enabled when xattr is set. If the mount option is enabled, all the files are marked with the inline_xattrs flag. Signed-off-by: Jaegeuk Kim --- fs/f2fs/f2fs.h | 18 ++++++++++++++++++ fs/f2fs/inode.c | 2 ++ fs/f2fs/super.c | 14 ++++++++++++++ include/linux/f2fs_fs.h | 4 +++- 4 files changed, 37 insertions(+), 1 deletion(-) (limited to 'fs/f2fs/inode.c') diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 5348b63adbe..b82f1419992 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -29,6 +29,7 @@ #define F2FS_MOUNT_XATTR_USER 0x00000010 #define F2FS_MOUNT_POSIX_ACL 0x00000020 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 +#define F2FS_MOUNT_INLINE_XATTR 0x00000080 #define clear_opt(sbi, option) (sbi->mount_opt.opt &= ~F2FS_MOUNT_##option) #define set_opt(sbi, option) (sbi->mount_opt.opt |= F2FS_MOUNT_##option) @@ -892,6 +893,7 @@ enum { FI_NO_ALLOC, /* should not allocate any blocks */ FI_UPDATE_DIR, /* should update inode block for consistency */ FI_DELAY_IPUT, /* used for the recovery */ + FI_INLINE_XATTR, /* used for inline xattr */ }; static inline void set_inode_flag(struct f2fs_inode_info *fi, int flag) @@ -924,6 +926,22 @@ static inline int cond_clear_inode_flag(struct f2fs_inode_info *fi, int flag) return 0; } +static inline void get_inline_info(struct f2fs_inode_info *fi, + struct f2fs_inode *ri) +{ + if (ri->i_inline & F2FS_INLINE_XATTR) + set_inode_flag(fi, FI_INLINE_XATTR); +} + +static inline void set_raw_inline(struct f2fs_inode_info *fi, + struct f2fs_inode *ri) +{ + ri->i_inline = 0; + + if (is_inode_flag_set(fi, FI_INLINE_XATTR)) + ri->i_inline |= F2FS_INLINE_XATTR; +} + static inline int f2fs_readonly(struct super_block *sb) { return sb->s_flags & MS_RDONLY; diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c index 7f8569bd875..9339cd29204 100644 --- a/fs/f2fs/inode.c +++ b/fs/f2fs/inode.c @@ -85,6 +85,7 @@ static int do_read_inode(struct inode *inode) fi->i_advise = ri->i_advise; fi->i_pino = le32_to_cpu(ri->i_pino); get_extent_info(&fi->ext, ri->i_ext); + get_inline_info(fi, ri); f2fs_put_page(node_page, 1); return 0; } @@ -164,6 +165,7 @@ void update_inode(struct inode *inode, struct page *node_page) ri->i_size = cpu_to_le64(i_size_read(inode)); ri->i_blocks = cpu_to_le64(inode->i_blocks); set_raw_extent(&F2FS_I(inode)->ext, &ri->i_ext); + set_raw_inline(F2FS_I(inode), ri); ri->i_atime = cpu_to_le64(inode->i_atime.tv_sec); ri->i_ctime = cpu_to_le64(inode->i_ctime.tv_sec); diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c index d28c4528eff..70ecf484e7e 100644 --- a/fs/f2fs/super.c +++ b/fs/f2fs/super.c @@ -47,6 +47,7 @@ enum { Opt_noacl, Opt_active_logs, Opt_disable_ext_identify, + Opt_inline_xattr, Opt_err, }; @@ -59,6 +60,7 @@ static match_table_t f2fs_tokens = { {Opt_noacl, "noacl"}, {Opt_active_logs, "active_logs=%u"}, {Opt_disable_ext_identify, "disable_ext_identify"}, + {Opt_inline_xattr, "inline_xattr"}, {Opt_err, NULL}, }; @@ -238,11 +240,18 @@ static int parse_options(struct super_block *sb, char *options) case Opt_nouser_xattr: clear_opt(sbi, XATTR_USER); break; + case Opt_inline_xattr: + set_opt(sbi, INLINE_XATTR); + break; #else case Opt_nouser_xattr: f2fs_msg(sb, KERN_INFO, "nouser_xattr options not supported"); break; + case Opt_inline_xattr: + f2fs_msg(sb, KERN_INFO, + "inline_xattr options not supported"); + break; #endif #ifdef CONFIG_F2FS_FS_POSIX_ACL case Opt_noacl: @@ -292,6 +301,9 @@ static struct inode *f2fs_alloc_inode(struct super_block *sb) set_inode_flag(fi, FI_NEW_INODE); + if (test_opt(F2FS_SB(sb), INLINE_XATTR)) + set_inode_flag(fi, FI_INLINE_XATTR); + return &fi->vfs_inode; } @@ -444,6 +456,8 @@ static int f2fs_show_options(struct seq_file *seq, struct dentry *root) seq_puts(seq, ",user_xattr"); else seq_puts(seq, ",nouser_xattr"); + if (test_opt(sbi, INLINE_XATTR)) + seq_puts(seq, ",inline_xattr"); #endif #ifdef CONFIG_F2FS_FS_POSIX_ACL if (test_opt(sbi, POSIX_ACL)) diff --git a/include/linux/f2fs_fs.h b/include/linux/f2fs_fs.h index 383d5e39b28..10ab11f8f99 100644 --- a/include/linux/f2fs_fs.h +++ b/include/linux/f2fs_fs.h @@ -144,10 +144,12 @@ struct f2fs_extent { #define ADDRS_PER_BLOCK 1018 /* Address Pointers in a Direct Block */ #define NIDS_PER_BLOCK 1018 /* Node IDs in an Indirect Block */ +#define F2FS_INLINE_XATTR 0x01 /* file inline xattr flag */ + struct f2fs_inode { __le16 i_mode; /* file mode */ __u8 i_advise; /* file hints */ - __u8 i_reserved; /* reserved */ + __u8 i_inline; /* file inline flags */ __le32 i_uid; /* user ID */ __le32 i_gid; /* group ID */ __le32 i_links; /* links count */ -- cgit v1.2.3-70-g09d2