From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mark Fasheh Subject: [PATCH 3/7] btrfs: Don't BUG_ON alloc_path errors in btrfs_truncate_inode_items Date: Thu, 14 Jul 2011 15:14:58 -0700 Message-ID: <1310681702-13922-4-git-send-email-mfasheh@suse.com> References: <1310681702-13922-1-git-send-email-mfasheh@suse.com> <1310681702-13922-2-git-send-email-mfasheh@suse.com> <1310681702-13922-3-git-send-email-mfasheh@suse.com> Cc: chris.mason@oracle.com, Mark Fasheh To: linux-btrfs@vger.kernel.org Return-path: In-Reply-To: <1310681702-13922-3-git-send-email-mfasheh@suse.com> List-ID: I moved the path allocation up a few lines to the top of the function so that we couldn't get into the state where we've dropped delayed items and the extent cache but fail due to -ENOMEM. Signed-off-by: Mark Fasheh --- fs/btrfs/inode.c | 9 +++++---- 1 files changed, 5 insertions(+), 4 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 8be7d7a..a0faf7d 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -3172,6 +3172,11 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY); + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + path->reada = -1; + if (root->ref_cows || root == root->fs_info->tree_root) btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0); @@ -3184,10 +3189,6 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, if (min_type == 0 && root == BTRFS_I(inode)->root) btrfs_kill_delayed_inode_items(inode); - path = btrfs_alloc_path(); - BUG_ON(!path); - path->reada = -1; - key.objectid = ino; key.offset = (u64)-1; key.type = (u8)-1; -- 1.7.5.3