From: Theodore Ts'o Subject: [PATCH FOR-STABLE-2.6.28 11/24] ext4: don't use blocks freed but not yet committed in buddy cache init Date: Tue, 17 Feb 2009 10:32:29 -0500 Message-ID: <1234884762-13580-12-git-send-email-tytso@mit.edu> References: <1234884762-13580-1-git-send-email-tytso@mit.edu> <1234884762-13580-2-git-send-email-tytso@mit.edu> <1234884762-13580-3-git-send-email-tytso@mit.edu> <1234884762-13580-4-git-send-email-tytso@mit.edu> <1234884762-13580-5-git-send-email-tytso@mit.edu> <1234884762-13580-6-git-send-email-tytso@mit.edu> <1234884762-13580-7-git-send-email-tytso@mit.edu> <1234884762-13580-8-git-send-email-tytso@mit.edu> <1234884762-13580-9-git-send-email-tytso@mit.edu> <1234884762-13580-10-git-send-email-tytso@mit.edu> <1234884762-13580-11-git-send-email-tytso@mit.edu> Cc: linux-ext4@vger.kernel.org, "Aneesh Kumar K.V" , "Theodore Ts'o" To: stable@kernel.org Return-path: Received: from thunk.org ([69.25.196.29]:33251 "EHLO thunker.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753406AbZBQQKR (ORCPT ); Tue, 17 Feb 2009 11:10:17 -0500 In-Reply-To: <1234884762-13580-11-git-send-email-tytso@mit.edu> Sender: linux-ext4-owner@vger.kernel.org List-ID: From: Aneesh Kumar K.V When we generate buddy cache (especially during resize) we need to make sure we don't use the blocks freed but not yet comitted. This makes sure we have the right value of free blocks count in the group info and also in the bitmap. This also ensures the ordered mode consistency Signed-off-by: Aneesh Kumar K.V Signed-off-by: "Theodore Ts'o" Cc: stable@kernel.org (cherry picked from commit 7a2fcbf7f85737735fd44eb34b62315bccf6d6e4) --- fs/ext4/mballoc.c | 82 ++++++++++++++++++++++++++++++++++++++-------------- 1 files changed, 60 insertions(+), 22 deletions(-) diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index 2a02cf2..d4283b4 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -335,6 +335,8 @@ static struct kmem_cache *ext4_ac_cachep; static struct kmem_cache *ext4_free_ext_cachep; static void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap, ext4_group_t group); +static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap, + ext4_group_t group); static int ext4_mb_init_per_dev_proc(struct super_block *sb); static int ext4_mb_destroy_per_dev_proc(struct super_block *sb); static void release_blocks_on_commit(journal_t *journal, transaction_t *txn); @@ -858,7 +860,9 @@ static int ext4_mb_init_cache(struct page *page, char *incore) /* * incore got set to the group block bitmap below */ + ext4_lock_group(sb, group); ext4_mb_generate_buddy(sb, data, incore, group); + ext4_unlock_group(sb, group); incore = NULL; } else { /* this is block of bitmap */ @@ -872,6 +876,7 @@ static int ext4_mb_init_cache(struct page *page, char *incore) /* mark all preallocated blks used in in-core bitmap */ ext4_mb_generate_from_pa(sb, data, group); + ext4_mb_generate_from_freelist(sb, data, group); ext4_unlock_group(sb, group); /* set incore so that the buddy information can be @@ -3469,6 +3474,32 @@ ext4_mb_use_preallocated(struct ext4_allocation_context *ac) } /* + * the function goes through all block freed in the group + * but not yet committed and marks them used in in-core bitmap. + * buddy must be generated from this bitmap + * Need to be called with ext4 group lock (ext4_lock_group) + */ +static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap, + ext4_group_t group) +{ + struct rb_node *n; + struct ext4_group_info *grp; + struct ext4_free_data *entry; + + grp = ext4_get_group_info(sb, group); + n = rb_first(&(grp->bb_free_root)); + + while (n) { + entry = rb_entry(n, struct ext4_free_data, node); + mb_set_bits(sb_bgl_lock(EXT4_SB(sb), group), + bitmap, entry->start_blk, + entry->count); + n = rb_next(n); + } + return; +} + +/* * the function goes through all preallocation in this group and marks them * used in in-core bitmap. buddy must be generated from this bitmap * Need to be called with ext4 group lock (ext4_lock_group) @@ -4565,12 +4596,13 @@ static int can_merge(struct ext4_free_data *entry1, static noinline_for_stack int ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, - ext4_group_t group, ext4_grpblk_t block, int count) + struct ext4_free_data *new_entry) { + ext4_grpblk_t block; + struct ext4_free_data *entry; struct ext4_group_info *db = e4b->bd_info; struct super_block *sb = e4b->bd_sb; struct ext4_sb_info *sbi = EXT4_SB(sb); - struct ext4_free_data *entry, *new_entry; struct rb_node **n = &db->bb_free_root.rb_node, *node; struct rb_node *parent = NULL, *new_node; @@ -4578,14 +4610,9 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, BUG_ON(e4b->bd_bitmap_page == NULL); BUG_ON(e4b->bd_buddy_page == NULL); - new_entry = kmem_cache_alloc(ext4_free_ext_cachep, GFP_NOFS); - new_entry->start_blk = block; - new_entry->group = group; - new_entry->count = count; - new_entry->t_tid = handle->h_transaction->t_tid; new_node = &new_entry->node; + block = new_entry->start_blk; - ext4_lock_group(sb, group); if (!*n) { /* first free block exent. We need to protect buddy cache from being freed, @@ -4603,7 +4630,6 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, else if (block >= (entry->start_blk + entry->count)) n = &(*n)->rb_right; else { - ext4_unlock_group(sb, group); ext4_error(sb, __func__, "Double free of blocks %d (%d %d)\n", block, entry->start_blk, entry->count); @@ -4645,7 +4671,6 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, spin_lock(&sbi->s_md_lock); list_add(&new_entry->list, &handle->h_transaction->t_private_list); spin_unlock(&sbi->s_md_lock); - ext4_unlock_group(sb, group); return 0; } @@ -4750,15 +4775,6 @@ do_more: BUG_ON(!mb_test_bit(bit + i, bitmap_bh->b_data)); } #endif - mb_clear_bits(sb_bgl_lock(sbi, block_group), bitmap_bh->b_data, - bit, count); - - /* We dirtied the bitmap block */ - BUFFER_TRACE(bitmap_bh, "dirtied bitmap block"); - err = ext4_journal_dirty_metadata(handle, bitmap_bh); - if (err) - goto error_return; - if (ac) { ac->ac_b_ex.fe_group = block_group; ac->ac_b_ex.fe_start = bit; @@ -4770,11 +4786,29 @@ do_more: if (err) goto error_return; if (metadata) { - /* blocks being freed are metadata. these blocks shouldn't - * be used until this transaction is committed */ - ext4_mb_free_metadata(handle, &e4b, block_group, bit, count); + struct ext4_free_data *new_entry; + /* + * blocks being freed are metadata. these blocks shouldn't + * be used until this transaction is committed + */ + new_entry = kmem_cache_alloc(ext4_free_ext_cachep, GFP_NOFS); + new_entry->start_blk = bit; + new_entry->group = block_group; + new_entry->count = count; + new_entry->t_tid = handle->h_transaction->t_tid; + ext4_lock_group(sb, block_group); + mb_clear_bits(sb_bgl_lock(sbi, block_group), bitmap_bh->b_data, + bit, count); + ext4_mb_free_metadata(handle, &e4b, new_entry); + ext4_unlock_group(sb, block_group); } else { ext4_lock_group(sb, block_group); + /* need to update group_info->bb_free and bitmap + * with group lock held. generate_buddy look at + * them with group lock_held + */ + mb_clear_bits(sb_bgl_lock(sbi, block_group), bitmap_bh->b_data, + bit, count); mb_free_blocks(inode, &e4b, bit, count); ext4_mb_return_to_preallocation(inode, &e4b, block, count); ext4_unlock_group(sb, block_group); @@ -4797,6 +4831,10 @@ do_more: *freed += count; + /* We dirtied the bitmap block */ + BUFFER_TRACE(bitmap_bh, "dirtied bitmap block"); + err = ext4_journal_dirty_metadata(handle, bitmap_bh); + /* And the group descriptor block */ BUFFER_TRACE(gd_bh, "dirtied group descriptor block"); ret = ext4_journal_dirty_metadata(handle, gd_bh); -- 1.5.6.3