From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753879Ab1G1El0 (ORCPT ); Thu, 28 Jul 2011 00:41:26 -0400 Received: from mail-ww0-f44.google.com ([74.125.82.44]:49700 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751467Ab1G1ElS (ORCPT ); Thu, 28 Jul 2011 00:41:18 -0400 Subject: Re: [PATCH] vfs: avoid taking locks if inode not in lists From: Eric Dumazet To: Christoph Hellwig Cc: Tim Chen , Al Viro , David Miller , Andi Kleen , Matthew Wilcox , Anton Blanchard , npiggin@kernel.dk, linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, netdev In-Reply-To: <20110727204415.GA13308@infradead.org> References: <1311294452.2576.18.camel@schen9-DESK> <20110723132411.GA22183@infradead.org> <1311633550.2576.33.camel@schen9-DESK> <20110725225154.GD22133@ZenIV.linux.org.uk> <1311636178.2576.34.camel@schen9-DESK> <1311660013.2996.6.camel@edumazet-laptop> <1311668466.2355.12.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> <20110726090357.GA13013@infradead.org> <1311672994.2355.17.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> <1311780065.2356.18.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> <20110727204415.GA13308@infradead.org> Content-Type: text/plain; charset="UTF-8" Date: Thu, 28 Jul 2011 06:41:09 +0200 Message-ID: <1311828069.2697.27.camel@edumazet-laptop> Mime-Version: 1.0 X-Mailer: Evolution 2.32.2 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Le mercredi 27 juillet 2011 à 16:44 -0400, Christoph Hellwig a écrit : > On Wed, Jul 27, 2011 at 05:21:05PM +0200, Eric Dumazet wrote: > > If I am not mistaken, we can add unlocked checks on the three hot spots. > > > > After following patch, a close(socket(PF_INET, SOCK_DGRAM, 0)) pair on > > my dev machine takes ~3us instead of ~9us. > > > > Maybe its better to split it in three patches, just let me know. > > I think three patches would be a lot cleaner. > > As for safety of the unlocked checks: > > - inode are either hashed when created or never, so that one looks > fine. > - same for the sb list. > - the writeback list is a bit more dynamic as we move things around > quite a bit. But in additon to the inode_wb_list_del call from > evict() it only ever gets remove in writeback_single_inode, which > for a freeing inode can only be called from the callers of evict(). > > Btw, I wonder if you should micro-optimize things a bit further by > moving the unhashed checks from the deletion functions into the callers > and thus save a function call for each of them. > What about following patch, addressing the micro-optimization and Andi Kleen concern about evict() readability ? Thanks ! [PATCH] vfs: avoid taking inode_hash_lock on pipes and sockets Some inodes (pipes, sockets, ...) are not hashed, no need to take contended inode_hash_lock at dismantle time. nice speedup on SMP machines on socket intensive workloads. Signed-off-by: Eric Dumazet --- fs/inode.c | 6 +++--- include/linux/fs.h | 9 ++++++++- 2 files changed, 11 insertions(+), 4 deletions(-) diff --git a/fs/inode.c b/fs/inode.c index d0c72ff..73b5598 100644 --- a/fs/inode.c +++ b/fs/inode.c @@ -399,12 +399,12 @@ void __insert_inode_hash(struct inode *inode, unsigned long hashval) EXPORT_SYMBOL(__insert_inode_hash); /** - * remove_inode_hash - remove an inode from the hash + * __remove_inode_hash - remove an inode from the hash * @inode: inode to unhash * * Remove an inode from the superblock. */ -void remove_inode_hash(struct inode *inode) +void __remove_inode_hash(struct inode *inode) { spin_lock(&inode_hash_lock); spin_lock(&inode->i_lock); @@ -412,7 +412,7 @@ void remove_inode_hash(struct inode *inode) spin_unlock(&inode->i_lock); spin_unlock(&inode_hash_lock); } -EXPORT_SYMBOL(remove_inode_hash); +EXPORT_SYMBOL(__remove_inode_hash); void end_writeback(struct inode *inode) { diff --git a/include/linux/fs.h b/include/linux/fs.h index f23bcb7..786b3b1 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -2317,11 +2317,18 @@ extern int should_remove_suid(struct dentry *); extern int file_remove_suid(struct file *); extern void __insert_inode_hash(struct inode *, unsigned long hashval); -extern void remove_inode_hash(struct inode *); static inline void insert_inode_hash(struct inode *inode) { __insert_inode_hash(inode, inode->i_ino); } + +extern void __remove_inode_hash(struct inode *); +static inline void remove_inode_hash(struct inode *inode) +{ + if (!inode_unhashed(inode)) + __remove_inode_hash(inode); +} + extern void inode_sb_list_add(struct inode *inode); #ifdef CONFIG_BLOCK