200-sched_esfq.patch

Linux Home Server 是专门为家庭和SOHO/SMB 设计的高性价比的ISCSI 存储服务器, 具有如下的特色： 强大的iscsi 存储服务器软件； 混合iscsi 和NAS 服务；

Index: linux-2.6.21.7/include/linux/pkt_sched.h
===================================================================
--- linux-2.6.21.7.orig/include/linux/pkt_sched.h
+++ linux-2.6.21.7/include/linux/pkt_sched.h
@@ -146,8 +146,40 @@ struct tc_sfq_qopt
  *
  *	The only reason for this is efficiency, it is possible
  *	to change these parameters in compile time.
+ *	
+ *	If you need to play with these values use esfq instead.
  */
 
+/* ESFQ section */
+
+enum
+{
+        /* traditional */
+	TCA_SFQ_HASH_CLASSIC,
+	TCA_SFQ_HASH_DST,
+	TCA_SFQ_HASH_SRC,
+	TCA_SFQ_HASH_FWMARK,
+        /* direct */
+	TCA_SFQ_HASH_DSTDIR,
+	TCA_SFQ_HASH_SRCDIR,
+	TCA_SFQ_HASH_FWMARKDIR,
+	/* conntrack */
+	TCA_SFQ_HASH_CTORIGDST,
+	TCA_SFQ_HASH_CTORIGSRC,
+	TCA_SFQ_HASH_CTREPLDST,
+	TCA_SFQ_HASH_CTREPLSRC,
+};
+
+struct tc_esfq_qopt
+{
+	unsigned	quantum;	/* Bytes per round allocated to flow */
+	int		perturb_period;	/* Period of hash perturbation */
+	__u32		limit;		/* Maximal packets in queue */
+	unsigned	divisor;	/* Hash divisor  */
+	unsigned	flows;		/* Maximal number of flows  */
+	unsigned	hash_kind;	/* Hash function to use for flow identification */
+};
+
 /* RED section */
 
 enum
Index: linux-2.6.21.7/net/sched/Kconfig
===================================================================
--- linux-2.6.21.7.orig/net/sched/Kconfig
+++ linux-2.6.21.7/net/sched/Kconfig
@@ -189,6 +189,26 @@ config NET_SCH_SFQ
 	  To compile this code as a module, choose M here: the
 	  module will be called sch_sfq.
 
+config NET_SCH_ESFQ
+	tristate "Enhanced Stochastic Fairness Queueing (ESFQ)"
+	---help---
+	  Say Y here if you want to use the Enhanced Stochastic Fairness
+	  Queueing (ESFQ) packet scheduling algorithm for some of your network
+	  devices or as a leaf discipline for a classful qdisc such as HTB or
+	  CBQ (see the top of <file:net/sched/sch_esfq.c> for details and
+	  references to the SFQ algorithm).
+
+	  This is an enchanced SFQ version which allows you to control some
+	  hardcoded values in the SFQ scheduler.
+
+	  ESFQ also adds control of the hash function used to identify packet
+	  flows. The original SFQ discipline hashes by connection; ESFQ add
+	  several other hashing methods, such as by src IP or by dst IP, which
+	  can be more fair to users in some networking situations.
+	  
+	  To compile this code as a module, choose M here: the
+	  module will be called sch_esfq.
+
 config NET_SCH_TEQL
 	tristate "True Link Equalizer (TEQL)"
 	---help---
Index: linux-2.6.21.7/net/sched/Makefile
===================================================================
--- linux-2.6.21.7.orig/net/sched/Makefile
+++ linux-2.6.21.7/net/sched/Makefile
@@ -23,6 +23,7 @@ obj-$(CONFIG_NET_SCH_GRED)	+= sch_gred.o
 obj-$(CONFIG_NET_SCH_INGRESS)	+= sch_ingress.o 
 obj-$(CONFIG_NET_SCH_DSMARK)	+= sch_dsmark.o
 obj-$(CONFIG_NET_SCH_SFQ)	+= sch_sfq.o
+obj-$(CONFIG_NET_SCH_ESFQ)	+= sch_esfq.o
 obj-$(CONFIG_NET_SCH_TBF)	+= sch_tbf.o
 obj-$(CONFIG_NET_SCH_TEQL)	+= sch_teql.o
 obj-$(CONFIG_NET_SCH_PRIO)	+= sch_prio.o
Index: linux-2.6.21.7/net/sched/sch_esfq.c
===================================================================
--- /dev/null
+++ linux-2.6.21.7/net/sched/sch_esfq.c
@@ -0,0 +1,704 @@
+/*
+ * net/sched/sch_esfq.c	Extended Stochastic Fairness Queueing discipline.
+ *
+ *		This program is free software; you can redistribute it and/or
+ *		modify it under the terms of the GNU General Public License
+ *		as published by the Free Software Foundation; either version
+ *		2 of the License, or (at your option) any later version.
+ *
+ * Authors:	Alexey Kuznetsov, <[e-mail]>
+ *
+ * Changes:	Alexander Atanasov, <[e-mail]>
+ *		Added dynamic depth,limit,divisor,hash_kind options.
+ *		Added dst and src hashes.
+ *
+ * 		Alexander Clouter, <[e-mail]>
+ *		Ported ESFQ to Linux 2.6.
+ *
+ * 		Corey Hickey, <[e-mail]>
+ *		Maintenance of the Linux 2.6 port.
+ *		Added fwmark hash (thanks to Robert Kurjata).
+ *		Added direct hashing for src, dst, and fwmark.
+ *		Added usage of jhash.
+ *		
+ */
+
+#include <linux/module.h>
+#include <asm/uaccess.h>
+#include <asm/system.h>
+#include <linux/bitops.h>
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/jiffies.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/socket.h>
+#include <linux/sockios.h>
+#include <linux/in.h>
+#include <linux/errno.h>
+#include <linux/interrupt.h>
+#include <linux/if_ether.h>
+#include <linux/inet.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/notifier.h>
+#include <linux/init.h>
+#include <net/ip.h>
+#include <linux/ipv6.h>
+#include <net/route.h>
+#include <linux/skbuff.h>
+#include <net/sock.h>
+#include <net/pkt_sched.h>
+#include <linux/jhash.h>
+
+#ifdef CONFIG_NF_CONNTRACK_ENABLED
+#include <net/netfilter/nf_conntrack.h>
+#endif
+
+/*	Stochastic Fairness Queuing algorithm.
+	For more comments look at sch_sfq.c.
+	The difference is that you can change limit, depth,
+	hash table size and choose alternate hash types.
+	
+	classic:	same as in sch_sfq.c
+	dst:		destination IP address
+	src:		source IP address
+	fwmark:         netfilter mark value
+	dst_direct:
+	src_direct:
+	fwmark_direct:  direct hashing of the above sources
+	ctorigdst:	original destination IP address
+	ctorigsrc:	original source IP address
+	ctrepldst:	reply destination IP address
+	ctreplsrc:	reply source IP 
+	
+*/
+
+
+/* This type should contain at least SFQ_DEPTH*2 values */
+typedef unsigned int esfq_index;
+
+struct esfq_head
+{
+	esfq_index	next;
+	esfq_index	prev;
+};
+
+struct esfq_sched_data
+{
+/* Parameters */
+	int		perturb_period;
+	unsigned	quantum;	/* Allotment per round: MUST BE >= MTU */
+	int		limit;
+	unsigned	depth;
+	unsigned	hash_divisor;
+	unsigned	hash_kind;
+/* Variables */
+	struct timer_list perturb_timer;
+	int		perturbation;
+	esfq_index	tail;		/* Index of current slot in round */
+	esfq_index	max_depth;	/* Maximal depth */
+
+	esfq_index	*ht;			/* Hash table */
+	esfq_index	*next;			/* Active slots link */
+	short		*allot;			/* Current allotment per slot */
+	unsigned short	*hash;			/* Hash value indexed by slots */
+	struct sk_buff_head	*qs;		/* Slot queue */
+	struct esfq_head	*dep;		/* Linked list of slots, indexed by depth */
+	unsigned	dyn_min;	/* For dynamic divisor adjustment; minimum value seen */
+	unsigned	dyn_max;	/*                                 maximum value seen */
+	unsigned	dyn_range;	/*	        		   saved range */
+};
+
+/* This contains the info we will hash. */
+struct esfq_packet_info
+{
+	u32	proto;		/* protocol or port */
+	u32	src;		/* source from packet header */
+	u32	dst;		/* destination from packet header */
+	u32	ctorigsrc;	/* original source from conntrack */
+	u32	ctorigdst;	/* original destination from conntrack */
+	u32	ctreplsrc;	/* reply source from conntrack */
+	u32	ctrepldst;	/* reply destination from conntrack */
+	u32	mark;		/* netfilter mark (fwmark) */
+};
+
+/* Hash input values directly into the "nearest" slot, taking into account the
+ * range of input values seen. This is most useful when the hash table is at
+ * least as large as the range of possible values.
+ * Note: this functionality was added before the change to using jhash, and may
+ * no longer be useful. */
+static __inline__ unsigned esfq_hash_direct(struct esfq_sched_data *q, u32 h)
+{
+	/* adjust minimum and maximum */
+	if (h < q->dyn_min || h > q->dyn_max) {
+		q->dyn_min = h < q->dyn_min ? h : q->dyn_min;
+		q->dyn_max = h > q->dyn_max ? h : q->dyn_max;
+	
+		/* find new range */
+		if ((q->dyn_range = q->dyn_max - q->dyn_min) >= q->hash_divisor)
+			printk(KERN_WARNING "ESFQ: (direct hash) Input range %u is larger than hash "
+					"table. See ESFQ README for details.\n", q->dyn_range);
+	}
+	
+	/* hash input values into slot numbers */
+	if (q->dyn_min == q->dyn_max)
+		return 0; /* only one value seen; avoid division by 0 */
+	else
+		return (h - q->dyn_min) * (q->hash_divisor - 1) / q->dyn_range;
+}
+
+static __inline__ unsigned esfq_jhash_1word(struct esfq_sched_data *q,u32 a)
+{
+	return jhash_1word(a, q->perturbation) & (q->hash_divisor-1);
+}
+
+static __inline__ unsigned esfq_jhash_2words(struct esfq_sched_data *q, u32 a, u32 b)
+{
+	return jhash_2words(a, b, q->perturbation) & (q->hash_divisor-1);
+}
+
+static __inline__ unsigned esfq_jhash_3words(struct esfq_sched_data *q, u32 a, u32 b, u32 c)
+{
+	return jhash_3words(a, b, c, q->perturbation) & (q->hash_divisor-1);
+}
+
+
+static unsigned esfq_hash(struct esfq_sched_data *q, struct sk_buff *skb)
+{
+	struct esfq_packet_info info;
+#ifdef CONFIG_NF_CONNTRACK_ENABLED
+	enum ip_conntrack_info ctinfo;
+	struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
+#endif
+	
+	switch (skb->protocol) {
+	case __constant_htons(ETH_P_IP):
+	{
+		struct iphdr *iph = skb->nh.iph;
+		info.dst = iph->daddr;
+		info.src = iph->saddr;
+		if (!(iph->frag_off&htons(IP_MF|IP_OFFSET)) &&
+		    (iph->protocol == IPPROTO_TCP ||
+		     iph->protocol == IPPROTO_UDP ||
+		     iph->protocol == IPPROTO_SCTP ||
+		     iph->protocol == IPPROTO_DCCP ||
+		     iph->protocol == IPPROTO_ESP))
+			info.proto = *(((u32*)iph) + iph->ihl);
+		else
+			info.proto = iph->protocol;
+		break;
+	}
+	case __constant_htons(ETH_P_IPV6):
+	{
+		struct ipv6hdr *iph = skb->nh.ipv6h;
+		/* Hash ipv6 addresses into a u32. This isn't ideal,
+		 * but the code is simple. */
+		info.dst = jhash2(iph->daddr.s6_addr32, 4, q->perturbation);
+		info.src = jhash2(iph->saddr.s6_addr32, 4, q->perturbation);
+		if (iph->nexthdr == IPPROTO_TCP ||
+		    iph->nexthdr == IPPROTO_UDP ||
+		    iph->nexthdr == IPPROTO_SCTP ||
+		    iph->nexthdr == IPPROTO_DCCP ||
+		    iph->nexthdr == IPPROTO_ESP)
+			info.proto = *(u32*)&iph[1];
+		else
+			info.proto = iph->nexthdr;
+		break;
+	}
+	default:
+		info.dst   = (u32)(unsigned long)skb->dst;
+		info.src   = (u32)(unsigned long)skb->sk;
+		info.proto = skb->protocol;
+	}
+
+	info.mark = skb->mark;
+
+#ifdef CONFIG_NF_CONNTRACK_ENABLED
+	/* defaults if there is no conntrack info */
+	info.ctorigsrc = info.src;
+	info.ctorigdst = info.dst;
+	info.ctreplsrc = info.dst;
+	info.ctrepldst = info.src;
+	/* collect conntrack info */
+	if (ct && ct != &nf_conntrack_untracked) {
+		if (skb->protocol == __constant_htons(ETH_P_IP)) {
+			info.ctorigsrc = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.src.u3.ip;
+			info.ctorigdst = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.dst.u3.ip;
+			info.ctreplsrc = ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.u3.ip;
+			info.ctrepldst = ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3.ip;
+		}
+		else if (skb->protocol == __constant_htons(ETH_P_IPV6)) {
+			/* Again, hash ipv6 addresses into a single u32. */
+			info.ctorigsrc = jhash2(ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.src.u3.ip6, 4, q->perturbation);
+			info.ctorigdst = jhash2(ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.dst.u3.ip6, 4, q->perturbation);
+			info.ctreplsrc = jhash2(ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.u3.ip6, 4, q->perturbation);
+			info.ctrepldst = jhash2(ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3.ip6, 4, q->perturbation);
+		}
+
+	}
+#endif
+
+	switch(q->hash_kind)
+	{
+	case TCA_SFQ_HASH_CLASSIC:
+		return esfq_jhash_3words(q, info.dst, info.src, info.proto);
+	case TCA_SFQ_HASH_DST:
+		return esfq_jhash_1word(q, info.dst);
+	case TCA_SFQ_HASH_DSTDIR:
+		return esfq_hash_direct(q, ntohl(info.dst));
+	case TCA_SFQ_HASH_SRC:
+		return esfq_jhash_1word(q, info.src);
+	case TCA_SFQ_HASH_SRCDIR:
+		return esfq_hash_direct(q, ntohl(info.src));
+	case TCA_SFQ_HASH_FWMARK:
+		return esfq_jhash_1word(q, info.mark);
+	case TCA_SFQ_HASH_FWMARKDIR:
+		return esfq_hash_direct(q, info.mark);
+#ifdef CONFIG_NF_CONNTRACK_ENABLED
+	case TCA_SFQ_HASH_CTORIGDST:
+		return esfq_jhash_1word(q, info.ctorigdst);
+	case TCA_SFQ_HASH_CTORIGSRC:
+		return esfq_jhash_1word(q, info.ctorigsrc);
+	case TCA_SFQ_HASH_CTREPLDST:
+		return esfq_jhash_1word(q, info.ctrepldst);
+	case TCA_SFQ_HASH_CTREPLSRC:
+		return esfq_jhash_1word(q, info.ctreplsrc);
+#endif
+	default:
+		if (net_ratelimit())
+			printk(KERN_WARNING "ESFQ: Unknown hash method. Falling back to classic.\n");
+	}
+	return esfq_jhash_3words(q, info.dst, info.src, info.proto);
+}
+
+static inline void esfq_link(struct esfq_sched_data *q, esfq_index x)
+{
+	esfq_index p, n;
+	int d = q->qs[x].qlen + q->depth;
+
+	p = d;
+	n = q->dep[d].next;
+	q->dep[x].next = n;
+	q->dep[x].prev = p;
+	q->dep[p].next = q->dep[n].prev = x;
+}
+
+static inline void esfq_dec(struct esfq_sched_data *q, esfq_index x)
+{
+	esfq_index p, n;
+
+	n = q->dep[x].next;
+	p = q->dep[x].prev;
+	q->dep[p].next = n;
+	q->dep[n].prev = p;
+
+	if (n == p && q->max_depth == q->qs[x].qlen + 1)
+		q->max_depth--;
+
+	esfq_link(q, x);
+}
+
+static inline void esfq_inc(struct esfq_sched_data *q, esfq_index x)
+{
+	esfq_index p, n;
+	int d;
+