From 860d21e2c585f7ee8a4ecc06f474fdc33c9474f4 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Sat, 12 Jan 2013 16:19:36 -0500 Subject: ext4: return ENOMEM if sb_getblk() fails The only reason for sb_getblk() failing is if it can't allocate the buffer_head. So ENOMEM is more appropriate than EIO. In addition, make sure that the file system is marked as being inconsistent if sb_getblk() fails. Signed-off-by: "Theodore Ts'o" Cc: stable@vger.kernel.org --- fs/ext4/resize.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'fs/ext4/resize.c') diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c index d99387b89edd..02824dc2ff3b 100644 --- a/fs/ext4/resize.c +++ b/fs/ext4/resize.c @@ -334,7 +334,7 @@ static struct buffer_head *bclean(handle_t *handle, struct super_block *sb, bh = sb_getblk(sb, blk); if (!bh) - return ERR_PTR(-EIO); + return ERR_PTR(-ENOMEM); if ((err = ext4_journal_get_write_access(handle, bh))) { brelse(bh); bh = ERR_PTR(err); @@ -411,7 +411,7 @@ static int set_flexbg_block_bitmap(struct super_block *sb, handle_t *handle, bh = sb_getblk(sb, flex_gd->groups[group].block_bitmap); if (!bh) - return -EIO; + return -ENOMEM; err = ext4_journal_get_write_access(handle, bh); if (err) @@ -501,7 +501,7 @@ static int setup_new_flex_group_blocks(struct super_block *sb, gdb = sb_getblk(sb, block); if (!gdb) { - err = -EIO; + err = -ENOMEM; goto out; } @@ -1065,7 +1065,7 @@ static void update_backups(struct super_block *sb, int blk_off, char *data, bh = sb_getblk(sb, backup_block); if (!bh) { - err = -EIO; + err = -ENOMEM; break; } ext4_debug("update metadata backup %llu(+%llu)\n", -- cgit v1.2.3 From aebf02430d25b6bd2b8542126fdcdb90e75a24b8 Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Sat, 12 Jan 2013 16:28:47 -0500 Subject: ext4: use unlikely to improve the efficiency of the kernel Because the function 'sb_getblk' seldomly fails to return NULL value,it will be better to use 'unlikely' to optimize it. Signed-off-by: Wang Shilong Signed-off-by: "Theodore Ts'o" --- fs/ext4/resize.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs/ext4/resize.c') diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c index 02824dc2ff3b..05f8d4502d42 100644 --- a/fs/ext4/resize.c +++ b/fs/ext4/resize.c @@ -333,7 +333,7 @@ static struct buffer_head *bclean(handle_t *handle, struct super_block *sb, int err; bh = sb_getblk(sb, blk); - if (!bh) + if (unlikely(!bh)) return ERR_PTR(-ENOMEM); if ((err = ext4_journal_get_write_access(handle, bh))) { brelse(bh); @@ -410,7 +410,7 @@ static int set_flexbg_block_bitmap(struct super_block *sb, handle_t *handle, return err; bh = sb_getblk(sb, flex_gd->groups[group].block_bitmap); - if (!bh) + if (unlikely(!bh)) return -ENOMEM; err = ext4_journal_get_write_access(handle, bh); @@ -500,7 +500,7 @@ static int setup_new_flex_group_blocks(struct super_block *sb, goto out; gdb = sb_getblk(sb, block); - if (!gdb) { + if (unlikely(!gdb)) { err = -ENOMEM; goto out; } @@ -1064,7 +1064,7 @@ static void update_backups(struct super_block *sb, int blk_off, char *data, ext4_bg_has_super(sb, group)); bh = sb_getblk(sb, backup_block); - if (!bh) { + if (unlikely(!bh)) { err = -ENOMEM; break; } @@ -1168,7 +1168,7 @@ static int ext4_add_new_descs(handle_t *handle, struct super_block *sb, static struct buffer_head *ext4_get_bitmap(struct super_block *sb, __u64 block) { struct buffer_head *bh = sb_getblk(sb, block); - if (!bh) + if (unlikely(!bh)) return NULL; if (!bh_uptodate_or_lock(bh)) { if (bh_submit_read(bh) < 0) { -- cgit v1.2.3 From 7f5118629f74b82bd4ba5e47415d1b4dcb940241 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Sun, 13 Jan 2013 08:41:45 -0500 Subject: ext4: trigger the lazy inode table initialization after resize After we have finished extending the file system, we need to trigger a the lazy inode table thread to zero out the inode tables. Signed-off-by: "Theodore Ts'o" --- fs/ext4/resize.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'fs/ext4/resize.c') diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c index 05f8d4502d42..8eefb636beb8 100644 --- a/fs/ext4/resize.c +++ b/fs/ext4/resize.c @@ -1506,10 +1506,12 @@ static int ext4_setup_next_flex_gd(struct super_block *sb, group_data[i].blocks_count = blocks_per_group; overhead = ext4_group_overhead_blocks(sb, group + i); group_data[i].free_blocks_count = blocks_per_group - overhead; - if (ext4_has_group_desc_csum(sb)) + if (ext4_has_group_desc_csum(sb)) { flex_gd->bg_flags[i] = EXT4_BG_BLOCK_UNINIT | EXT4_BG_INODE_UNINIT; - else + if (!test_opt(sb, INIT_INODE_TABLE)) + flex_gd->bg_flags[i] |= EXT4_BG_INODE_ZEROED; + } else flex_gd->bg_flags[i] = EXT4_BG_INODE_ZEROED; } @@ -1594,7 +1596,7 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input) err = ext4_alloc_flex_bg_array(sb, input->group + 1); if (err) - return err; + goto out; err = ext4_mb_alloc_groupinfo(sb, input->group + 1); if (err) -- cgit v1.2.3 From 9924a92a8c217576bd2a2b1bbbb854462f1a00ae Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Fri, 8 Feb 2013 21:59:22 -0500 Subject: ext4: pass context information to jbd2__journal_start() So we can better understand what bits of ext4 are responsible for long-running jbd2 handles, use jbd2__journal_start() so we can pass context information for logging purposes. The recommended way for finding the longer-running handles is: T=/sys/kernel/debug/tracing EVENT=$T/events/jbd2/jbd2_handle_stats echo "interval > 5" > $EVENT/filter echo 1 > $EVENT/enable ./run-my-fs-benchmark cat $T/trace > /tmp/problem-handles This will list handles that were active for longer than 20ms. Having longer-running handles is bad, because a commit started at the wrong time could stall for those 20+ milliseconds, which could delay an fsync() or an O_SYNC operation. Here is an example line from the trace file describing a handle which lived on for 311 jiffies, or over 1.2 seconds: postmark-2917 [000] .... 196.435786: jbd2_handle_stats: dev 254,32 tid 570 type 2 line_no 2541 interval 311 sync 0 requested_blocks 1 dirtied_blocks 0 Signed-off-by: "Theodore Ts'o" --- fs/ext4/resize.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs/ext4/resize.c') diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c index 8eefb636beb8..c7f4d7584669 100644 --- a/fs/ext4/resize.c +++ b/fs/ext4/resize.c @@ -466,7 +466,7 @@ static int setup_new_flex_group_blocks(struct super_block *sb, meta_bg = EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_META_BG); /* This transaction may be extended/restarted along the way */ - handle = ext4_journal_start_sb(sb, EXT4_MAX_TRANS_DATA); + handle = ext4_journal_start_sb(sb, EXT4_HT_RESIZE, EXT4_MAX_TRANS_DATA); if (IS_ERR(handle)) return PTR_ERR(handle); @@ -1031,7 +1031,7 @@ static void update_backups(struct super_block *sb, int blk_off, char *data, handle_t *handle; int err = 0, err2; - handle = ext4_journal_start_sb(sb, EXT4_MAX_TRANS_DATA); + handle = ext4_journal_start_sb(sb, EXT4_HT_RESIZE, EXT4_MAX_TRANS_DATA); if (IS_ERR(handle)) { group = 1; err = PTR_ERR(handle); @@ -1412,7 +1412,7 @@ static int ext4_flex_group_add(struct super_block *sb, * modify each of the reserved GDT dindirect blocks. */ credit = flex_gd->count * 4 + reserved_gdb; - handle = ext4_journal_start_sb(sb, credit); + handle = ext4_journal_start_sb(sb, EXT4_HT_RESIZE, credit); if (IS_ERR(handle)) { err = PTR_ERR(handle); goto exit; @@ -1624,7 +1624,7 @@ static int ext4_group_extend_no_check(struct super_block *sb, /* We will update the superblock, one block bitmap, and * one group descriptor via ext4_group_add_blocks(). */ - handle = ext4_journal_start_sb(sb, 3); + handle = ext4_journal_start_sb(sb, EXT4_HT_RESIZE, 3); if (IS_ERR(handle)) { err = PTR_ERR(handle); ext4_warning(sb, "error %d on journal start", err); @@ -1788,7 +1788,7 @@ static int ext4_convert_meta_bg(struct super_block *sb, struct inode *inode) credits += 3; /* block bitmap, bg descriptor, resize inode */ } - handle = ext4_journal_start_sb(sb, credits); + handle = ext4_journal_start_sb(sb, EXT4_HT_RESIZE, credits); if (IS_ERR(handle)) return PTR_ERR(handle); -- cgit v1.2.3 From 810da240f221d64bf90020f25941b05b378186fe Mon Sep 17 00:00:00 2001 From: Lukas Czerner Date: Sat, 2 Mar 2013 17:18:58 -0500 Subject: ext4: convert number of blocks to clusters properly We're using macro EXT4_B2C() to convert number of blocks to number of clusters for bigalloc file systems. However, we should be using EXT4_NUM_B2C(). Signed-off-by: Lukas Czerner Signed-off-by: "Theodore Ts'o" Cc: stable@vger.kernel.org --- fs/ext4/resize.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/ext4/resize.c') diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c index c7f4d7584669..b2c8ee56eb98 100644 --- a/fs/ext4/resize.c +++ b/fs/ext4/resize.c @@ -1247,7 +1247,7 @@ static int ext4_setup_new_descs(handle_t *handle, struct super_block *sb, ext4_inode_table_set(sb, gdp, group_data->inode_table); ext4_free_group_clusters_set(sb, gdp, - EXT4_B2C(sbi, group_data->free_blocks_count)); + EXT4_NUM_B2C(sbi, group_data->free_blocks_count)); ext4_free_inodes_set(sb, gdp, EXT4_INODES_PER_GROUP(sb)); if (ext4_has_group_desc_csum(sb)) ext4_itable_unused_set(sb, gdp, @@ -1349,7 +1349,7 @@ static void ext4_update_super(struct super_block *sb, /* Update the free space counts */ percpu_counter_add(&sbi->s_freeclusters_counter, - EXT4_B2C(sbi, free_blocks)); + EXT4_NUM_B2C(sbi, free_blocks)); percpu_counter_add(&sbi->s_freeinodes_counter, EXT4_INODES_PER_GROUP(sb) * flex_gd->count); @@ -1360,7 +1360,7 @@ static void ext4_update_super(struct super_block *sb, sbi->s_log_groups_per_flex) { ext4_group_t flex_group; flex_group = ext4_flex_group(sbi, group_data[0].group); - atomic_add(EXT4_B2C(sbi, free_blocks), + atomic_add(EXT4_NUM_B2C(sbi, free_blocks), &sbi->s_flex_groups[flex_group].free_clusters); atomic_add(EXT4_INODES_PER_GROUP(sb) * flex_gd->count, &sbi->s_flex_groups[flex_group].free_inodes); -- cgit v1.2.3