From 952fc18ef9ec707ebdc16c0786ec360295e5ff15 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Mon, 9 Jul 2012 16:27:05 -0400 Subject: ext4: fix overhead calculation used by ext4_statfs() Commit f975d6bcc7a introduced bug which caused ext4_statfs() to miscalculate the number of file system overhead blocks. This causes the f_blocks field in the statfs structure to be larger than it should be. This would in turn cause the "df" output to show the number of data blocks in the file system and the number of data blocks used to be larger than they should be. Signed-off-by: "Theodore Ts'o" Cc: stable@kernel.org --- fs/ext4/super.c | 174 ++++++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 124 insertions(+), 50 deletions(-) (limited to 'fs/ext4/super.c') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index eb7aa3e4ef05..78b7ede2efa0 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -3085,6 +3085,114 @@ static int set_journal_csum_feature_set(struct super_block *sb) return ret; } +/* + * Note: calculating the overhead so we can be compatible with + * historical BSD practice is quite difficult in the face of + * clusters/bigalloc. This is because multiple metadata blocks from + * different block group can end up in the same allocation cluster. + * Calculating the exact overhead in the face of clustered allocation + * requires either O(all block bitmaps) in memory or O(number of block + * groups**2) in time. We will still calculate the superblock for + * older file systems --- and if we come across with a bigalloc file + * system with zero in s_overhead_clusters the estimate will be close to + * correct especially for very large cluster sizes --- but for newer + * file systems, it's better to calculate this figure once at mkfs + * time, and store it in the superblock. If the superblock value is + * present (even for non-bigalloc file systems), we will use it. + */ +static int count_overhead(struct super_block *sb, ext4_group_t grp, + char *buf) +{ + struct ext4_sb_info *sbi = EXT4_SB(sb); + struct ext4_group_desc *gdp; + ext4_fsblk_t first_block, last_block, b; + ext4_group_t i, ngroups = ext4_get_groups_count(sb); + int s, j, count = 0; + + first_block = le32_to_cpu(sbi->s_es->s_first_data_block) + + (grp * EXT4_BLOCKS_PER_GROUP(sb)); + last_block = first_block + EXT4_BLOCKS_PER_GROUP(sb) - 1; + for (i = 0; i < ngroups; i++) { + gdp = ext4_get_group_desc(sb, i, NULL); + b = ext4_block_bitmap(sb, gdp); + if (b >= first_block && b <= last_block) { + ext4_set_bit(EXT4_B2C(sbi, b - first_block), buf); + count++; + } + b = ext4_inode_bitmap(sb, gdp); + if (b >= first_block && b <= last_block) { + ext4_set_bit(EXT4_B2C(sbi, b - first_block), buf); + count++; + } + b = ext4_inode_table(sb, gdp); + if (b >= first_block && b + sbi->s_itb_per_group <= last_block) + for (j = 0; j < sbi->s_itb_per_group; j++, b++) { + int c = EXT4_B2C(sbi, b - first_block); + ext4_set_bit(c, buf); + count++; + } + if (i != grp) + continue; + s = 0; + if (ext4_bg_has_super(sb, grp)) { + ext4_set_bit(s++, buf); + count++; + } + for (j = ext4_bg_num_gdb(sb, grp); j > 0; j--) { + ext4_set_bit(EXT4_B2C(sbi, s++), buf); + count++; + } + } + if (!count) + return 0; + return EXT4_CLUSTERS_PER_GROUP(sb) - + ext4_count_free(buf, EXT4_CLUSTERS_PER_GROUP(sb) / 8); +} + +/* + * Compute the overhead and stash it in sbi->s_overhead + */ +int ext4_calculate_overhead(struct super_block *sb) +{ + struct ext4_sb_info *sbi = EXT4_SB(sb); + struct ext4_super_block *es = sbi->s_es; + ext4_group_t i, ngroups = ext4_get_groups_count(sb); + ext4_fsblk_t overhead = 0; + char *buf = (char *) get_zeroed_page(GFP_KERNEL); + + memset(buf, 0, PAGE_SIZE); + if (!buf) + return -ENOMEM; + + /* + * Compute the overhead (FS structures). This is constant + * for a given filesystem unless the number of block groups + * changes so we cache the previous value until it does. + */ + + /* + * All of the blocks before first_data_block are overhead + */ + overhead = EXT4_B2C(sbi, le32_to_cpu(es->s_first_data_block)); + + /* + * Add the overhead found in each block group + */ + for (i = 0; i < ngroups; i++) { + int blks; + + blks = count_overhead(sb, i, buf); + overhead += blks; + if (blks) + memset(buf, 0, PAGE_SIZE); + cond_resched(); + } + sbi->s_overhead = overhead; + smp_wmb(); + free_page((unsigned long) buf); + return 0; +} + static int ext4_fill_super(struct super_block *sb, void *data, int silent) { char *orig_data = kstrdup(data, GFP_KERNEL); @@ -3734,6 +3842,18 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) percpu_counter_set(&sbi->s_dirtyclusters_counter, 0); no_journal: + /* + * Get the # of file system overhead blocks from the + * superblock if present. + */ + if (es->s_overhead_clusters) + sbi->s_overhead = le32_to_cpu(es->s_overhead_clusters); + else { + ret = ext4_calculate_overhead(sb); + if (ret) + goto failed_mount_wq; + } + /* * The maximum number of concurrent works can be high and * concurrency isn't really necessary. Limit it to 1. @@ -4600,67 +4720,21 @@ restore_opts: return err; } -/* - * Note: calculating the overhead so we can be compatible with - * historical BSD practice is quite difficult in the face of - * clusters/bigalloc. This is because multiple metadata blocks from - * different block group can end up in the same allocation cluster. - * Calculating the exact overhead in the face of clustered allocation - * requires either O(all block bitmaps) in memory or O(number of block - * groups**2) in time. We will still calculate the superblock for - * older file systems --- and if we come across with a bigalloc file - * system with zero in s_overhead_clusters the estimate will be close to - * correct especially for very large cluster sizes --- but for newer - * file systems, it's better to calculate this figure once at mkfs - * time, and store it in the superblock. If the superblock value is - * present (even for non-bigalloc file systems), we will use it. - */ static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf) { struct super_block *sb = dentry->d_sb; struct ext4_sb_info *sbi = EXT4_SB(sb); struct ext4_super_block *es = sbi->s_es; - struct ext4_group_desc *gdp; + ext4_fsblk_t overhead = 0; u64 fsid; s64 bfree; - if (test_opt(sb, MINIX_DF)) { - sbi->s_overhead_last = 0; - } else if (es->s_overhead_clusters) { - sbi->s_overhead_last = le32_to_cpu(es->s_overhead_clusters); - } else if (sbi->s_blocks_last != ext4_blocks_count(es)) { - ext4_group_t i, ngroups = ext4_get_groups_count(sb); - ext4_fsblk_t overhead = 0; - - /* - * Compute the overhead (FS structures). This is constant - * for a given filesystem unless the number of block groups - * changes so we cache the previous value until it does. - */ - - /* - * All of the blocks before first_data_block are - * overhead - */ - overhead = EXT4_B2C(sbi, le32_to_cpu(es->s_first_data_block)); - - /* - * Add the overhead found in each block group - */ - for (i = 0; i < ngroups; i++) { - gdp = ext4_get_group_desc(sb, i, NULL); - overhead += ext4_num_overhead_clusters(sb, i, gdp); - cond_resched(); - } - sbi->s_overhead_last = overhead; - smp_wmb(); - sbi->s_blocks_last = ext4_blocks_count(es); - } + if (!test_opt(sb, MINIX_DF)) + overhead = sbi->s_overhead; buf->f_type = EXT4_SUPER_MAGIC; buf->f_bsize = sb->s_blocksize; - buf->f_blocks = (ext4_blocks_count(es) - - EXT4_C2B(sbi, sbi->s_overhead_last)); + buf->f_blocks = ext4_blocks_count(es) - EXT4_C2B(sbi, sbi->s_overhead); bfree = percpu_counter_sum_positive(&sbi->s_freeclusters_counter) - percpu_counter_sum_positive(&sbi->s_dirtyclusters_counter); /* prevent underflow in case that few free space is available */ -- cgit v1.2.3 From 7c319d328505b7781b65238ae9f53293b5ee0ca8 Mon Sep 17 00:00:00 2001 From: Aditya Kali Date: Sun, 22 Jul 2012 20:21:31 -0400 Subject: ext4: make quota as first class supported feature This patch adds support for quotas as a first class feature in ext4; which is to say, the quota files are stored in hidden inodes as file system metadata, instead of as separate files visible in the file system directory hierarchy. It is based on the proposal at: https://ext4.wiki.kernel.org/index.php/Design_For_1st_Class_Quota_in_Ext4 This patch introduces a new feature - EXT4_FEATURE_RO_COMPAT_QUOTA which, when turned on, enables quota accounting at mount time iteself. Also, the quota inodes are stored in two additional superblock fields. Some changes introduced by this patch that should be pointed out are: 1) Two new ext4-superblock fields - s_usr_quota_inum and s_grp_quota_inum for storing the quota inodes in use. 2) Default quota inodes are: inode#3 for tracking userquota and inode#4 for tracking group quota. The superblock fields can be set to use other inodes as well. 3) If the QUOTA feature and corresponding quota inodes are set in superblock, the quota usage tracking is turned on at mount time. On 'quotaon' ioctl, the quota limits enforcement is turned on. 'quotaoff' ioctl turns off only the limits enforcement in this case. 4) When QUOTA feature is in use, the quota mount options 'quota', 'usrquota', 'grpquota' are ignored by the kernel. 5) mke2fs or tune2fs can be used to set the QUOTA feature and initialize quota inodes. The default reserved inodes will not be visible to user as regular files. 6) The quota-tools will need to be modified to support hidden quota files on ext4. E2fsprogs will also include support for creating and fixing quota files. 7) Support is only for the new V2 quota file format. Tested-by: Jan Kara Reviewed-by: Jan Kara Reviewed-by: Johann Lombardi Signed-off-by: Aditya Kali Signed-off-by: "Theodore Ts'o" --- fs/ext4/super.c | 137 ++++++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 134 insertions(+), 3 deletions(-) (limited to 'fs/ext4/super.c') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 78b7ede2efa0..bebf8e5bf087 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -1137,12 +1137,18 @@ static int ext4_mark_dquot_dirty(struct dquot *dquot); static int ext4_write_info(struct super_block *sb, int type); static int ext4_quota_on(struct super_block *sb, int type, int format_id, struct path *path); +static int ext4_quota_on_sysfile(struct super_block *sb, int type, + int format_id); static int ext4_quota_off(struct super_block *sb, int type); +static int ext4_quota_off_sysfile(struct super_block *sb, int type); static int ext4_quota_on_mount(struct super_block *sb, int type); static ssize_t ext4_quota_read(struct super_block *sb, int type, char *data, size_t len, loff_t off); static ssize_t ext4_quota_write(struct super_block *sb, int type, const char *data, size_t len, loff_t off); +static int ext4_quota_enable(struct super_block *sb, int type, int format_id, + unsigned int flags); +static int ext4_enable_quotas(struct super_block *sb); static const struct dquot_operations ext4_quota_operations = { .get_reserved_space = ext4_get_reserved_space, @@ -1164,6 +1170,16 @@ static const struct quotactl_ops ext4_qctl_operations = { .get_dqblk = dquot_get_dqblk, .set_dqblk = dquot_set_dqblk }; + +static const struct quotactl_ops ext4_qctl_sysfile_operations = { + .quota_on_meta = ext4_quota_on_sysfile, + .quota_off = ext4_quota_off_sysfile, + .quota_sync = dquot_quota_sync, + .get_info = dquot_get_dqinfo, + .set_info = dquot_set_dqinfo, + .get_dqblk = dquot_get_dqblk, + .set_dqblk = dquot_set_dqblk +}; #endif static const struct super_operations ext4_sops = { @@ -2661,6 +2677,16 @@ static int ext4_feature_set_ok(struct super_block *sb, int readonly) "extents feature\n"); return 0; } + +#ifndef CONFIG_QUOTA + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_QUOTA) && + !readonly) { + ext4_msg(sb, KERN_ERR, + "Filesystem with quota feature cannot be mounted RDWR " + "without CONFIG_QUOTA"); + return 0; + } +#endif /* CONFIG_QUOTA */ return 1; } @@ -3748,6 +3774,11 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) #ifdef CONFIG_QUOTA sb->s_qcop = &ext4_qctl_operations; sb->dq_op = &ext4_quota_operations; + + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_QUOTA)) { + /* Use qctl operations for hidden quota files. */ + sb->s_qcop = &ext4_qctl_sysfile_operations; + } #endif memcpy(sb->s_uuid, es->s_uuid, sizeof(es->s_uuid)); @@ -3960,6 +3991,16 @@ no_journal: } else descr = "out journal"; +#ifdef CONFIG_QUOTA + /* Enable quota usage during mount. */ + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_QUOTA) && + !(sb->s_flags & MS_RDONLY)) { + ret = ext4_enable_quotas(sb); + if (ret) + goto failed_mount7; + } +#endif /* CONFIG_QUOTA */ + ext4_msg(sb, KERN_INFO, "mounted filesystem with%s. " "Opts: %s%s%s", descr, sbi->s_es->s_mount_opts, *sbi->s_es->s_mount_opts ? "; " : "", orig_data); @@ -4682,16 +4723,26 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data) if (sbi->s_journal == NULL) ext4_commit_super(sb, 1); + unlock_super(sb); #ifdef CONFIG_QUOTA /* Release old quota file names */ for (i = 0; i < MAXQUOTAS; i++) if (old_opts.s_qf_names[i] && old_opts.s_qf_names[i] != sbi->s_qf_names[i]) kfree(old_opts.s_qf_names[i]); + if (enable_quota) { + if (sb_any_quota_suspended(sb)) + dquot_resume(sb, -1); + else if (EXT4_HAS_RO_COMPAT_FEATURE(sb, + EXT4_FEATURE_RO_COMPAT_QUOTA)) { + err = ext4_enable_quotas(sb); + if (err) { + lock_super(sb); + goto restore_opts; + } + } + } #endif - unlock_super(sb); - if (enable_quota) - dquot_resume(sb, -1); ext4_msg(sb, KERN_INFO, "re-mounted. Opts: %s", orig_data); kfree(orig_data); @@ -4904,6 +4955,74 @@ static int ext4_quota_on(struct super_block *sb, int type, int format_id, return dquot_quota_on(sb, type, format_id, path); } +static int ext4_quota_enable(struct super_block *sb, int type, int format_id, + unsigned int flags) +{ + int err; + struct inode *qf_inode; + unsigned long qf_inums[MAXQUOTAS] = { + le32_to_cpu(EXT4_SB(sb)->s_es->s_usr_quota_inum), + le32_to_cpu(EXT4_SB(sb)->s_es->s_grp_quota_inum) + }; + + BUG_ON(!EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_QUOTA)); + + if (!qf_inums[type]) + return -EPERM; + + qf_inode = ext4_iget(sb, qf_inums[type]); + if (IS_ERR(qf_inode)) { + ext4_error(sb, "Bad quota inode # %lu", qf_inums[type]); + return PTR_ERR(qf_inode); + } + + err = dquot_enable(qf_inode, type, format_id, flags); + iput(qf_inode); + + return err; +} + +/* Enable usage tracking for all quota types. */ +static int ext4_enable_quotas(struct super_block *sb) +{ + int type, err = 0; + unsigned long qf_inums[MAXQUOTAS] = { + le32_to_cpu(EXT4_SB(sb)->s_es->s_usr_quota_inum), + le32_to_cpu(EXT4_SB(sb)->s_es->s_grp_quota_inum) + }; + + sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE; + for (type = 0; type < MAXQUOTAS; type++) { + if (qf_inums[type]) { + err = ext4_quota_enable(sb, type, QFMT_VFS_V1, + DQUOT_USAGE_ENABLED); + if (err) { + ext4_warning(sb, + "Failed to enable quota (type=%d) " + "tracking. Please run e2fsck to fix.", + type); + return err; + } + } + } + return 0; +} + +/* + * quota_on function that is used when QUOTA feature is set. + */ +static int ext4_quota_on_sysfile(struct super_block *sb, int type, + int format_id) +{ + if (!EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_QUOTA)) + return -EINVAL; + + /* + * USAGE was enabled at mount time. Only need to enable LIMITS now. + */ + return ext4_quota_enable(sb, type, format_id, DQUOT_LIMITS_ENABLED); +} + static int ext4_quota_off(struct super_block *sb, int type) { struct inode *inode = sb_dqopt(sb)->files[type]; @@ -4930,6 +5049,18 @@ out: return dquot_quota_off(sb, type); } +/* + * quota_off function that is used when QUOTA feature is set. + */ +static int ext4_quota_off_sysfile(struct super_block *sb, int type) +{ + if (!EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_QUOTA)) + return -EINVAL; + + /* Disable only the limits. */ + return dquot_disable(sb, type, DQUOT_LIMITS_ENABLED); +} + /* Read data from quotafile - avoid pagecache and such because we cannot afford * acquiring the locks... As quota files are never truncated and quota code * itself serializes the operations (and no one else should touch the files) -- cgit v1.2.3 From 58c5873a769987e36265d1523d2aa5bdc18f32bd Mon Sep 17 00:00:00 2001 From: Artem Bityutskiy Date: Sun, 22 Jul 2012 20:33:31 -0400 Subject: ext4: remove unnecessary superblock dirtying This patch changes the 'ext4_handle_dirty_super()' function which submits the superblock for I/O in the following cases: 1. When creating the first large file on a file system without EXT4_FEATURE_RO_COMPAT_LARGE_FILE feature. 2. When re-sizing the file-system. 3. When creating an xattr on a file-system without the EXT4_FEATURE_COMPAT_EXT_ATTR feature. If the file-system has journal enabled, the superblock is written via the journal. We do not modify this path. If the file-system has no journal, this function, falls back to just marking the superblock as dirty using the 's_dirt' superblock flag. This means that it delays the actual superblock I/O submission by 5 seconds (default setting). Namely, the 'sync_supers()' kernel thread will call 'ext4_write_super()' later and will actually submit the superblock for I/O. And this is the behavior this patch modifies: we stop using 's_dirt' and just mark the superblock buffer as dirty right away. Indeed, all 3 cases above are extremely rare and it does not add any value to delay the I/O submission for them. Note: 'ext4_handle_dirty_super()' executes '__ext4_handle_dirty_super()' with 'now = 0'. This patch basically makes the 'now' argument unneeded and it will be deleted in one of the next patches. This patch also removes 's_dirt' condition on the unmount path because we never set it anymore, so we should not test it. Tested using xfstests for both journalled and non-journalled ext4. Signed-off-by: Artem Bityutskiy Signed-off-by: "Theodore Ts'o" Reviewed-by: Jan Kara --- fs/ext4/super.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/ext4/super.c') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index bebf8e5bf087..662e93e8c25b 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -896,7 +896,7 @@ static void ext4_put_super(struct super_block *sb) EXT4_CLEAR_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER); es->s_state = cpu_to_le16(sbi->s_mount_state); } - if (sb->s_dirt || !(sb->s_flags & MS_RDONLY)) + if (!(sb->s_flags & MS_RDONLY)) ext4_commit_super(sb, 1); if (sbi->s_proc) { -- cgit v1.2.3 From 4d47603d9703e6fff8ff2618bc108d6280e2439d Mon Sep 17 00:00:00 2001 From: Artem Bityutskiy Date: Sun, 22 Jul 2012 20:35:31 -0400 Subject: ext4: weed out ext4_write_super We do not depend on VFS's '->write_super()' anymore and do not need the 's_dirt' flag anymore, so weed out 'ext4_write_super()' and 's_dirt'. Signed-off-by: Artem Bityutskiy Signed-off-by: "Theodore Ts'o" Reviewed-by: Jan Kara --- fs/ext4/super.c | 10 ---------- 1 file changed, 10 deletions(-) (limited to 'fs/ext4/super.c') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 662e93e8c25b..a2a59796cde0 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -74,7 +74,6 @@ static const char *ext4_decode_error(struct super_block *sb, int errno, static int ext4_remount(struct super_block *sb, int *flags, char *data); static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf); static int ext4_unfreeze(struct super_block *sb); -static void ext4_write_super(struct super_block *sb); static int ext4_freeze(struct super_block *sb); static struct dentry *ext4_mount(struct file_system_type *fs_type, int flags, const char *dev_name, void *data); @@ -1210,7 +1209,6 @@ static const struct super_operations ext4_nojournal_sops = { .dirty_inode = ext4_dirty_inode, .drop_inode = ext4_drop_inode, .evict_inode = ext4_evict_inode, - .write_super = ext4_write_super, .put_super = ext4_put_super, .statfs = ext4_statfs, .remount_fs = ext4_remount, @@ -4364,7 +4362,6 @@ static int ext4_commit_super(struct super_block *sb, int sync) es->s_free_inodes_count = cpu_to_le32(percpu_counter_sum_positive( &EXT4_SB(sb)->s_freeinodes_counter)); - sb->s_dirt = 0; BUFFER_TRACE(sbh, "marking dirty"); ext4_superblock_csum_set(sb, es); mark_buffer_dirty(sbh); @@ -4471,13 +4468,6 @@ int ext4_force_commit(struct super_block *sb) return ret; } -static void ext4_write_super(struct super_block *sb) -{ - lock_super(sb); - ext4_commit_super(sb, 1); - unlock_super(sb); -} - static int ext4_sync_fs(struct super_block *sb, int wait) { int ret = 0; -- cgit v1.2.3