linux-nfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Greg Banks <gnb@sgi.com>
To: "J. Bruce Fields" <bfields@fieldses.org>
Cc: Linux NFS ML <linux-nfs@vger.kernel.org>
Subject: [patch 19/29] knfsd: faster probing in the reply cache
Date: Wed, 01 Apr 2009 07:28:19 +1100	[thread overview]
Message-ID: <20090331202944.736093000@sgi.com> (raw)
In-Reply-To: 20090331202800.739621000@sgi.com

Slightly reorganise the CPU-intensive hash chain probing to check
for more commonly varying entry keys first, and to reduce arithmetic
operations by hoisting out an invariant.  This slightly reduces the
CPU time spent in nfsd_cache_lookup() under heavy nonidempotent loads.

This code predates the IPv6 support in the NFS server but should be
harmless with IPv6 clients.

Signed-off-by: Greg Banks <gnb@sgi.com>
---

 fs/nfsd/nfscache.c |   25 ++++++++++++++++++++++---
 1 file changed, 22 insertions(+), 3 deletions(-)

Index: bfields/fs/nfsd/nfscache.c
===================================================================
--- bfields.orig/fs/nfsd/nfscache.c
+++ bfields/fs/nfsd/nfscache.c
@@ -108,6 +108,23 @@ static inline u32 request_hash(u32 xid, 
 	return h;
 }
 
+
+/*
+ * Fast compare of two sockaddr_in's.  Ignore the zero padding at
+ * the end of a sockaddr_in; compares sin_family and sin_port in
+ * one comparison.  Returns 1 if identical, 0 otherwise.
+ *
+ * Alignment issues prevent us from comparing all the relevant
+ * fields in a single 64b comparison on 64b platforms: the most
+ * we can assume is that sin_addr is 32b-aligned.
+ */
+static inline int
+compare_sockaddr_in(const struct sockaddr_in *sina, const struct sockaddr_in *sinb)
+{
+	return ((*(u32 *)&sina->sin_family == *(u32 *)&sinb->sin_family) &&
+		(sina->sin_addr.s_addr == sinb->sin_addr.s_addr));
+}
+
 static int	nfsd_cache_append(struct svc_rqst *rqstp, struct kvec *vec);
 
 /*
@@ -263,12 +280,14 @@ nfsd_cache_lookup(struct svc_rqst *rqstp
 	rtn = RC_DOIT;
 
 	rh = &b->hash[h];
+	age = jiffies - 120*HZ;
 	hlist_for_each_entry(rp, hn, rh, c_hash) {
 		if (rp->c_state != RC_UNUSED &&
-		    xid == rp->c_xid && proc == rp->c_proc &&
+		    xid == rp->c_xid &&
+		    compare_sockaddr_in(svc_addr_in(rqstp), &rp->c_addr) &&
+		    proc == rp->c_proc &&
 		    proto == rp->c_prot && vers == rp->c_vers &&
-		    time_before(jiffies, rp->c_timestamp + 120*HZ) &&
-		    memcmp((char*)&rqstp->rq_addr, (char*)&rp->c_addr, sizeof(rp->c_addr))==0) {
+		    time_after(rp->c_timestamp, age)) {
 			nfsdstats.rchits++;
 			goto found_entry;
 		}

--
Greg

  parent reply	other threads:[~2009-03-31 21:02 UTC|newest]

Thread overview: 63+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-03-31 20:28 [patch 00/29] SGI enhancedNFS patches Greg Banks
2009-03-31 20:28 ` [patch 01/29] knfsd: Add infrastructure for measuring RPC service times Greg Banks
2009-04-25  2:13   ` J. Bruce Fields
2009-04-25  2:14     ` J. Bruce Fields
2009-04-25  2:52     ` Greg Banks
2009-03-31 20:28 ` [patch 02/29] knfsd: Add stats table infrastructure Greg Banks
2009-04-25  3:56   ` J. Bruce Fields
2009-04-26  4:12     ` Greg Banks
2009-03-31 20:28 ` [patch 03/29] knfsd: add userspace controls for stats tables Greg Banks
2009-04-25 21:57   ` J. Bruce Fields
2009-04-25 22:03     ` J. Bruce Fields
2009-04-27 16:06       ` Chuck Lever
2009-04-27 23:22         ` J. Bruce Fields
2009-04-28 15:37           ` Chuck Lever
2009-04-28 15:57             ` J. Bruce Fields
2009-04-28 16:03               ` Chuck Lever
2009-04-28 16:26                 ` J. Bruce Fields
2009-04-29  1:45               ` Greg Banks
     [not found]         ` <ac442c870904271827w6041a67ew82fe36a843beeac3@mail.gmail.com>
     [not found]           ` <ac442c870904271827w6041a67ew82fe36a843beeac3-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2009-04-28  1:31             ` Greg Banks
2009-04-26  4:14     ` Greg Banks
2009-03-31 20:28 ` [patch 04/29] knfsd: Add stats updating API Greg Banks
2009-03-31 20:28 ` [patch 05/29] knfsd: Infrastructure for providing stats to userspace Greg Banks
2009-04-01  0:28   ` J. Bruce Fields
2009-04-01  3:43     ` Greg Banks
2009-03-31 20:28 ` [patch 06/29] knfsd: Gather per-export stats Greg Banks
2009-03-31 20:28 ` [patch 07/29] knfsd: Prefetch the per-export stats entry Greg Banks
2009-03-31 20:28 ` [patch 08/29] knfsd: Gather per-client stats Greg Banks
2009-03-31 20:28 ` [patch 09/29] knfsd: Cache per-client stats entry on TCP transports Greg Banks
2009-03-31 20:28 ` [patch 10/29] knfsd: Update per-client & per-export stats from NFSv3 Greg Banks
2009-03-31 20:28 ` [patch 11/29] knfsd: Update per-client & per-export stats from NFSv2 Greg Banks
2009-03-31 20:28 ` [patch 12/29] knfsd: Update per-client & per-export stats from NFSv4 Greg Banks
2009-03-31 20:28 ` [patch 13/29] knfsd: reply cache cleanups Greg Banks
2009-05-12 19:54   ` J. Bruce Fields
2009-03-31 20:28 ` [patch 14/29] knfsd: better hashing in the reply cache Greg Banks
2009-05-08 22:01   ` J. Bruce Fields
2009-03-31 20:28 ` [patch 15/29] knfsd: fix reply cache memory corruption Greg Banks
2009-05-12 19:55   ` J. Bruce Fields
2009-03-31 20:28 ` [patch 16/29] knfsd: use client IPv4 address in reply cache hash Greg Banks
2009-05-11 21:48   ` J. Bruce Fields
2009-03-31 20:28 ` [patch 17/29] knfsd: make the reply cache SMP-friendly Greg Banks
2009-03-31 20:28 ` [patch 18/29] knfsd: dynamically expand the reply cache Greg Banks
2009-05-26 18:57   ` J. Bruce Fields
2009-05-26 19:04     ` J. Bruce Fields
2009-05-26 21:24     ` Rob Gardner
2009-05-26 21:52       ` J. Bruce Fields
2009-05-27  0:28       ` Greg Banks
2009-03-31 20:28 ` Greg Banks [this message]
2009-03-31 20:28 ` [patch 20/29] knfsd: add extended reply cache stats Greg Banks
2009-03-31 20:28 ` [patch 21/29] knfsd: remove unreported filehandle stats counters Greg Banks
2009-05-12 20:00   ` J. Bruce Fields
2009-03-31 20:28 ` [patch 22/29] knfsd: make svc_authenticate() scale Greg Banks
2009-05-12 21:24   ` J. Bruce Fields
2009-03-31 20:28 ` [patch 23/29] knfsd: introduce SVC_INC_STAT Greg Banks
2009-03-31 20:28 ` [patch 24/29] knfsd: remove the program field from struct svc_stat Greg Banks
2009-03-31 20:28 ` [patch 25/29] knfsd: allocate svc_serv.sv_stats dynamically Greg Banks
2009-03-31 20:28 ` [patch 26/29] knfsd: make svc_serv.sv_stats per-CPU Greg Banks
2009-03-31 20:28 ` [patch 27/29] knfsd: move hot procedure count field out of svc_procedure Greg Banks
2009-03-31 20:28 ` [patch 28/29] knfsd: introduce NFSD_INC_STAT() Greg Banks
2009-03-31 20:28 ` [patch 29/29] knfsd: make nfsdstats per-CPU Greg Banks
2009-04-01  0:23 ` [patch 00/29] SGI enhancedNFS patches J. Bruce Fields
2009-04-01  3:32   ` Greg Banks
     [not found]     ` <ac442c870903312032t34630c6dvdbb644cb510f8079-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2009-04-01  6:34       ` Jeff Garzik
2009-04-01  6:41         ` Greg Banks

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20090331202944.736093000@sgi.com \
    --to=gnb@sgi.com \
    --cc=bfields@fieldses.org \
    --cc=linux-nfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).