From patchwork Wed Nov 8 02:17:26 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Yunlong Song X-Patchwork-Id: 10047523 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id E98DB60247 for ; Wed, 8 Nov 2017 02:18:52 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id D84D32A3AE for ; Wed, 8 Nov 2017 02:18:52 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id CCF542A3B3; Wed, 8 Nov 2017 02:18:52 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.9 required=2.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=unavailable version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 3E4EF2A3AE for ; Wed, 8 Nov 2017 02:18:52 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752584AbdKHCSh (ORCPT ); Tue, 7 Nov 2017 21:18:37 -0500 Received: from szxga05-in.huawei.com ([45.249.212.191]:10026 "EHLO szxga05-in.huawei.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752251AbdKHCSg (ORCPT ); Tue, 7 Nov 2017 21:18:36 -0500 Received: from 172.30.72.60 (EHLO DGGEMS401-HUB.china.huawei.com) ([172.30.72.60]) by dggrg05-dlp.huawei.com (MOS 4.4.6-GA FastPath queued) with ESMTP id DKN68571; Wed, 08 Nov 2017 10:18:32 +0800 (CST) Received: from huawei.com (10.107.193.250) by DGGEMS401-HUB.china.huawei.com (10.3.19.201) with Microsoft SMTP Server id 14.3.361.1; Wed, 8 Nov 2017 10:17:30 +0800 From: Yunlong Song To: , , , , CC: , , , , Subject: [PATCH v4] f2fs: fix out-of-free problem caused by atomic write Date: Wed, 8 Nov 2017 10:17:26 +0800 Message-ID: <1510107446-52423-1-git-send-email-yunlong.song@huawei.com> X-Mailer: git-send-email 1.8.5.2 In-Reply-To: <1509934071-116656-1-git-send-email-yunlong.song@huawei.com> References: <1509934071-116656-1-git-send-email-yunlong.song@huawei.com> MIME-Version: 1.0 X-Originating-IP: [10.107.193.250] X-CFilter-Loop: Reflected X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A020204.5A026979.0020, ss=1, re=0.000, recu=0.000, reip=0.000, cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2014-11-16 11:51:01, dmn=2013-03-21 17:37:32 X-Mirapoint-Loop-Id: 6120644efeb75569e8058a8265449f36 Sender: linux-fsdevel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-fsdevel@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP f2fs_balance_fs only actives once in the commit_inmem_pages, but there are more than one page to commit, so all the other pages will miss the check. This will lead to out-of-free problem when commit a very large file. However, we cannot do f2fs_balance_fs for each inmem page, since this will break atomicity. As a result, we should do f2fs_balance_fs for all the inmem pages together. Signed-off-by: Yunlong Song --- fs/f2fs/debug.c | 5 +++-- fs/f2fs/f2fs.h | 26 ++++++++++++++++++++++++-- fs/f2fs/segment.c | 30 ++++++++++++++++++++++++------ fs/f2fs/segment.h | 4 +++- 4 files changed, 54 insertions(+), 11 deletions(-) diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c index f7eec50..41c47c4 100644 --- a/fs/f2fs/debug.c +++ b/fs/f2fs/debug.c @@ -50,6 +50,7 @@ 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->inmem_commit_pages = get_pages(sbi, F2FS_INMEM_COMMIT_PAGES); si->aw_cnt = atomic_read(&sbi->aw_cnt); si->vw_cnt = atomic_read(&sbi->vw_cnt); si->max_aw_cnt = atomic_read(&sbi->max_aw_cnt); @@ -360,9 +361,9 @@ static int stat_show(struct seq_file *s, void *v) si->nr_discarding, si->nr_discarded, si->nr_discard_cmd, si->undiscard_blks); seq_printf(s, " - inmem: %4d, atomic IO: %4d (Max. %4d), " - "volatile IO: %4d (Max. %4d)\n", + "volatile IO: %4d (Max. %4d), commit: %4d\n", si->inmem_pages, si->aw_cnt, si->max_aw_cnt, - si->vw_cnt, si->max_vw_cnt); + si->vw_cnt, si->max_vw_cnt, si->inmem_commit_pages); seq_printf(s, " - nodes: %4d in %4d\n", si->ndirty_node, si->node_pages); seq_printf(s, " - dents: %4d in dirs:%4d (%4d)\n", diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 13a96b8..749bdb6 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -610,6 +610,7 @@ struct f2fs_inode_info { struct list_head inmem_pages; /* inmemory pages managed by f2fs */ struct task_struct *inmem_task; /* store inmemory task */ struct mutex inmem_lock; /* lock for inmemory pages */ + unsigned long inmem_blocks; /* inmemory blocks */ struct extent_tree *extent_tree; /* cached extent_tree entry */ struct rw_semaphore dio_rwsem[2];/* avoid racing between dio and gc */ struct rw_semaphore i_mmap_sem; @@ -863,6 +864,7 @@ enum count_type { F2FS_DIRTY_NODES, F2FS_DIRTY_META, F2FS_INMEM_PAGES, + F2FS_INMEM_COMMIT_PAGES, F2FS_DIRTY_IMETA, F2FS_WB_CP_DATA, F2FS_WB_DATA, @@ -1600,7 +1602,21 @@ static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) atomic_inc(&sbi->nr_pages[count_type]); if (count_type == F2FS_DIRTY_DATA || count_type == F2FS_INMEM_PAGES || - count_type == F2FS_WB_CP_DATA || count_type == F2FS_WB_DATA) + count_type == F2FS_WB_CP_DATA || count_type == F2FS_WB_DATA || + count_type == F2FS_INMEM_COMMIT_PAGES) + return; + + set_sbi_flag(sbi, SBI_IS_DIRTY); +} + +static inline void inc_pages_count(struct f2fs_sb_info *sbi, int count_type, + int pages) +{ + atomic_add(pages, &sbi->nr_pages[count_type]); + + if (count_type == F2FS_DIRTY_DATA || count_type == F2FS_INMEM_PAGES || + count_type == F2FS_WB_CP_DATA || count_type == F2FS_WB_DATA || + count_type == F2FS_INMEM_COMMIT_PAGES) return; set_sbi_flag(sbi, SBI_IS_DIRTY); @@ -1618,6 +1634,12 @@ static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) atomic_dec(&sbi->nr_pages[count_type]); } +static inline void dec_pages_count(struct f2fs_sb_info *sbi, int count_type, + int pages) +{ + atomic_sub(pages, &sbi->nr_pages[count_type]); +} + static inline void inode_dec_dirty_pages(struct inode *inode) { if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && @@ -2716,7 +2738,7 @@ struct f2fs_stat_info { unsigned long long hit_total, total_ext; int ext_tree, zombie_tree, ext_node; int ndirty_node, ndirty_dent, ndirty_meta, ndirty_data, ndirty_imeta; - int inmem_pages; + int inmem_pages, inmem_commit_pages; unsigned int ndirty_dirs, ndirty_files, ndirty_all; int nats, dirty_nats, sits, dirty_sits; int free_nids, avail_nids, alloc_nids; diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 46dfbca..2ff1bba4 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -174,6 +174,8 @@ bool need_SSR(struct f2fs_sb_info *sbi) int node_secs = get_blocktype_secs(sbi, F2FS_DIRTY_NODES); int dent_secs = get_blocktype_secs(sbi, F2FS_DIRTY_DENTS); int imeta_secs = get_blocktype_secs(sbi, F2FS_DIRTY_IMETA); + int inmem_commit_secs = get_blocktype_secs(sbi, + F2FS_INMEM_COMMIT_PAGES); if (test_opt(sbi, LFS)) return false; @@ -181,7 +183,7 @@ bool need_SSR(struct f2fs_sb_info *sbi) return true; return free_sections(sbi) <= (node_secs + 2 * dent_secs + imeta_secs + - 2 * reserved_sections(sbi)); + inmem_commit_secs + 2 * reserved_sections(sbi)); } void register_inmem_page(struct inode *inode, struct page *page) @@ -210,6 +212,7 @@ void register_inmem_page(struct inode *inode, struct page *page) list_add_tail(&fi->inmem_ilist, &sbi->inode_list[ATOMIC_FILE]); spin_unlock(&sbi->inode_lock[ATOMIC_FILE]); inc_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES); + fi->inmem_blocks++; mutex_unlock(&fi->inmem_lock); trace_f2fs_register_inmem_page(page, INMEM); @@ -221,6 +224,7 @@ static int __revoke_inmem_pages(struct inode *inode, struct f2fs_sb_info *sbi = F2FS_I_SB(inode); struct inmem_pages *cur, *tmp; int err = 0; + struct f2fs_inode_info *fi = F2FS_I(inode); list_for_each_entry_safe(cur, tmp, head, list) { struct page *page = cur->page; @@ -263,6 +267,7 @@ static int __revoke_inmem_pages(struct inode *inode, list_del(&cur->list); kmem_cache_free(inmem_entry_slab, cur); dec_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES); + fi->inmem_blocks--; } return err; } @@ -302,6 +307,10 @@ void drop_inmem_pages(struct inode *inode) if (!list_empty(&fi->inmem_ilist)) list_del_init(&fi->inmem_ilist); spin_unlock(&sbi->inode_lock[ATOMIC_FILE]); + if (fi->inmem_blocks) { + f2fs_bug_on(sbi, 1); + fi->inmem_blocks = 0; + } mutex_unlock(&fi->inmem_lock); clear_inode_flag(inode, FI_ATOMIC_FILE); @@ -326,6 +335,7 @@ void drop_inmem_page(struct inode *inode, struct page *page) f2fs_bug_on(sbi, !cur || cur->page != page); list_del(&cur->list); + fi->inmem_blocks--; mutex_unlock(&fi->inmem_lock); dec_page_count(sbi, F2FS_INMEM_PAGES); @@ -354,7 +364,7 @@ static int __commit_inmem_pages(struct inode *inode, .io_type = FS_DATA_IO, }; pgoff_t last_idx = ULONG_MAX; - int err = 0; + int err = 0, inmem_blocks = fi->inmem_blocks; list_for_each_entry_safe(cur, tmp, &fi->inmem_pages, list) { struct page *page = cur->page; @@ -390,6 +400,8 @@ static int __commit_inmem_pages(struct inode *inode, } unlock_page(page); list_move_tail(&cur->list, revoke_list); + dec_page_count(sbi, F2FS_INMEM_COMMIT_PAGES); + inmem_blocks--; } if (last_idx != ULONG_MAX) @@ -397,6 +409,8 @@ static int __commit_inmem_pages(struct inode *inode, if (!err) __revoke_inmem_pages(inode, revoke_list, false, false); + else + dec_pages_count(sbi, F2FS_INMEM_COMMIT_PAGES, inmem_blocks); return err; } @@ -409,12 +423,12 @@ int commit_inmem_pages(struct inode *inode) int err; INIT_LIST_HEAD(&revoke_list); + set_inode_flag(inode, FI_ATOMIC_COMMIT); + mutex_lock(&fi->inmem_lock); + inc_pages_count(sbi, F2FS_INMEM_COMMIT_PAGES, fi->inmem_blocks); 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) { int ret; @@ -437,11 +451,15 @@ int commit_inmem_pages(struct inode *inode) if (!list_empty(&fi->inmem_ilist)) list_del_init(&fi->inmem_ilist); spin_unlock(&sbi->inode_lock[ATOMIC_FILE]); + if (fi->inmem_blocks) { + f2fs_bug_on(sbi, 1); + fi->inmem_blocks = 0; + } + f2fs_unlock_op(sbi); mutex_unlock(&fi->inmem_lock); clear_inode_flag(inode, FI_ATOMIC_COMMIT); - f2fs_unlock_op(sbi); return err; } diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index 8d93652..f3885de 100644 --- a/fs/f2fs/segment.h +++ b/fs/f2fs/segment.h @@ -503,12 +503,14 @@ static inline bool has_not_enough_free_secs(struct f2fs_sb_info *sbi, int node_secs = get_blocktype_secs(sbi, F2FS_DIRTY_NODES); int dent_secs = get_blocktype_secs(sbi, F2FS_DIRTY_DENTS); int imeta_secs = get_blocktype_secs(sbi, F2FS_DIRTY_IMETA); + int inmem_commit_secs = get_blocktype_secs(sbi, + F2FS_INMEM_COMMIT_PAGES); if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING))) return false; return (free_sections(sbi) + freed) <= - (node_secs + 2 * dent_secs + imeta_secs + + (node_secs + 2 * dent_secs + imeta_secs + inmem_commit_secs + reserved_sections(sbi) + needed); }