From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [PATCH 3/3] fs: move i_hash out from under inode_lock Date: Thu, 28 Oct 2010 10:24:14 -0400 Message-ID: <20101028142414.GA28390@infradead.org> References: <1288266161-28897-1-git-send-email-david@fromorbit.com> <1288266161-28897-4-git-send-email-david@fromorbit.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: viro@ZenIV.linux.org.uk, linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org To: Dave Chinner Return-path: Received: from bombadil.infradead.org ([18.85.46.34]:44133 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752648Ab0J1OYQ (ORCPT ); Thu, 28 Oct 2010 10:24:16 -0400 Content-Disposition: inline In-Reply-To: <1288266161-28897-4-git-send-email-david@fromorbit.com> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: > @@ -925,8 +934,8 @@ static struct inode *get_new_inode(struct super_block *sb, > inode->i_state = I_NEW; > spin_unlock(&inode->i_lock); > hlist_add_head(&inode->i_hash, head); > + spin_unlock(&inode_hash_lock); > inode_sb_list_add(inode); > - spin_unlock(&inode_lock); Al said he wanted to have the sb lock nest inside the hash lock for now I think. Doubt it matters much, but it keeps the behaviour that we can't look up an inode which is not added to the per-sb list yet. After that a better patch description might be: "rename inode_lock to inode_hash_lock" as the inode_lock coverage after the previous patches should be 100% identical to the new hash lock.