From patchwork Mon Jan 19 11:36:40 2009 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Aneesh Kumar K.V" X-Patchwork-Id: 19442 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.176.167]) by ozlabs.org (Postfix) with ESMTP id D198DDDF13 for ; Tue, 20 Jan 2009 18:55:42 +1100 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753392AbZATHz3 (ORCPT ); Tue, 20 Jan 2009 02:55:29 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754372AbZATHz3 (ORCPT ); Tue, 20 Jan 2009 02:55:29 -0500 Received: from ausmtp04.au.ibm.com ([202.81.18.152]:39081 "EHLO ausmtp04.au.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754407AbZATHz0 (ORCPT ); Tue, 20 Jan 2009 02:55:26 -0500 Received: from sd0109e.au.ibm.com (d23rh905.au.ibm.com [202.81.18.225]) by ausmtp04.au.ibm.com (8.13.8/8.13.8) with ESMTP id n0JBtkkP216632 for ; Mon, 19 Jan 2009 22:57:26 +1100 Received: from d23av04.au.ibm.com (d23av04.au.ibm.com [9.190.235.139]) by sd0109e.au.ibm.com (8.13.8/8.13.8/NCO v9.1) with ESMTP id n0JBbBN8025012 for ; Mon, 19 Jan 2009 22:37:11 +1100 Received: from d23av04.au.ibm.com (loopback [127.0.0.1]) by d23av04.au.ibm.com (8.12.11.20060308/8.13.3) with ESMTP id n0JBbA7b016416 for ; Mon, 19 Jan 2009 22:37:10 +1100 Received: from skywalker ([9.77.195.141]) by d23av04.au.ibm.com (8.12.11.20060308/8.12.11) with ESMTP id n0JBalw1016210 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES128-SHA bits=128 verify=NO); Mon, 19 Jan 2009 22:37:01 +1100 Date: Mon, 19 Jan 2009 17:06:40 +0530 From: "Aneesh Kumar K.V" To: "Theodore Ts'o" Cc: linux-ext4@vger.kernel.org Subject: [PATCH] ext4: don't use blocks freed but not yet committed in buddy cache init Message-ID: <20090119113640.GG9482@skywalker> References: MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-ext4-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-ext4@vger.kernel.org From: Aneesh Kumar K.V Subject: [PATCH] ext4: don't use blocks freed but not yet committed in buddy cache init 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" --- 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 93e4c09..ea13c5e 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 ext4_mb_free_committed_blocks(struct super_block *); @@ -859,7 +861,9 @@ 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 */ @@ -873,6 +877,7 @@ 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 @@ -3583,6 +3588,32 @@ 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) @@ -4720,27 +4751,22 @@ 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; - 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, @@ -4799,7 +4825,6 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, spin_lock(&sbi->s_md_lock); list_add(&new_entry->list, &sbi->s_active_transaction); spin_unlock(&sbi->s_md_lock); - ext4_unlock_group(sb, group); return 0; } @@ -4906,15 +4931,6 @@ 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; @@ -4926,11 +4942,29 @@ 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); @@ -4953,6 +4987,10 @@ 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);