Merge master.kernel.org:/pub/scm/linux/kernel/git/acme/net-2.6

5 files changed, 403 insertions, 40 deletions

diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index 60ffcb9c579..e87b233615b 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -93,6 +93,7 @@ struct tc_fifo_qopt
 /* PRIO section */
 
 #define TCQ_PRIO_BANDS	16
+#define TCQ_MIN_PRIO_BANDS 2
 
 struct tc_prio_qopt
 {
@@ -169,6 +170,7 @@ struct tc_red_qopt
 	unsigned char   Scell_log;	/* cell size for idle damping */
 	unsigned char	flags;
 #define TC_RED_ECN	1
+#define TC_RED_HARDDROP	2
 };
 
 struct tc_red_xstats
@@ -194,38 +196,34 @@ enum
 
 #define TCA_GRED_MAX (__TCA_GRED_MAX - 1)
 
-#define TCA_SET_OFF TCA_GRED_PARMS
 struct tc_gred_qopt
 {
-       __u32           limit;          /* HARD maximal queue length (bytes)    
-*/
-       __u32           qth_min;        /* Min average length threshold (bytes) 
-*/
-       __u32           qth_max;        /* Max average length threshold (bytes) 
-*/
-       __u32           DP;             /* upto 2^32 DPs */
-       __u32           backlog;        
-       __u32           qave;   
-       __u32           forced; 
-       __u32           early;  
-       __u32           other;  
-       __u32           pdrop;  
-
-       unsigned char   Wlog;           /* log(W)               */
-       unsigned char   Plog;           /* log(P_max/(qth_max-qth_min)) */
-       unsigned char   Scell_log;      /* cell size for idle damping */
-       __u8            prio;		/* prio of this VQ */
-       __u32	packets;
-       __u32	bytesin;
+	__u32		limit;        /* HARD maximal queue length (bytes)    */
+	__u32		qth_min;      /* Min average length threshold (bytes) */
+	__u32		qth_max;      /* Max average length threshold (bytes) */
+	__u32		DP;           /* upto 2^32 DPs */
+	__u32		backlog;
+	__u32		qave;
+	__u32		forced;
+	__u32		early;
+	__u32		other;
+	__u32		pdrop;
+	__u8		Wlog;         /* log(W)               */
+	__u8		Plog;         /* log(P_max/(qth_max-qth_min)) */
+	__u8		Scell_log;    /* cell size for idle damping */
+	__u8		prio;         /* prio of this VQ */
+	__u32		packets;
+	__u32		bytesin;
 };
+
 /* gred setup */
 struct tc_gred_sopt
 {
-       __u32		DPs;
-       __u32		def_DP;
-       __u8		grio;
-       __u8		pad1;
-       __u16		pad2;
+	__u32		DPs;
+	__u32		def_DP;
+	__u8		grio;
+	__u8		flags;
+	__u16		pad1;
 };
 
 /* HTB section */
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 4286d832166..fdfb8fe8c38 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -603,23 +603,23 @@ static inline void skb_queue_head_init(struct sk_buff_head *list)
  */
 
 /**
- *	__skb_queue_head - queue a buffer at the list head
+ *	__skb_queue_after - queue a buffer at the list head
  *	@list: list to use
+ *	@prev: place after this buffer
  *	@newsk: buffer to queue
  *
- *	Queue a buffer at the start of a list. This function takes no locks
+ *	Queue a buffer int the middle of a list. This function takes no locks
  *	and you must therefore hold required locks before calling it.
  *
  *	A buffer cannot be placed on two lists at the same time.
  */
-extern void skb_queue_head(struct sk_buff_head *list, struct sk_buff *newsk);
-static inline void __skb_queue_head(struct sk_buff_head *list,
-				    struct sk_buff *newsk)
+static inline void __skb_queue_after(struct sk_buff_head *list,
+				     struct sk_buff *prev,
+				     struct sk_buff *newsk)
 {
-	struct sk_buff *prev, *next;
-
+	struct sk_buff *next;
 	list->qlen++;
-	prev = (struct sk_buff *)list;
+
 	next = prev->next;
 	newsk->next = next;
 	newsk->prev = prev;
@@ -627,6 +627,23 @@ static inline void __skb_queue_head(struct sk_buff_head *list,
 }
 
 /**
+ *	__skb_queue_head - queue a buffer at the list head
+ *	@list: list to use
+ *	@newsk: buffer to queue
+ *
+ *	Queue a buffer at the start of a list. This function takes no locks
+ *	and you must therefore hold required locks before calling it.
+ *
+ *	A buffer cannot be placed on two lists at the same time.
+ */
+extern void skb_queue_head(struct sk_buff_head *list, struct sk_buff *newsk);
+static inline void __skb_queue_head(struct sk_buff_head *list,
+				    struct sk_buff *newsk)
+{
+	__skb_queue_after(list, (struct sk_buff *)list, newsk);
+}
+
+/**
  *	__skb_queue_tail - queue a buffer at the list tail
  *	@list: list to use
  *	@newsk: buffer to queue
@@ -1203,6 +1220,11 @@ static inline void kunmap_skb_frag(void *vaddr)
 		     prefetch(skb->next), (skb != (struct sk_buff *)(queue));	\
 		     skb = skb->next)
 
+#define skb_queue_reverse_walk(queue, skb) \
+		for (skb = (queue)->prev;					\
+		     prefetch(skb->prev), (skb != (struct sk_buff *)(queue));	\
+		     skb = skb->prev)
+
 
 extern struct sk_buff *skb_recv_datagram(struct sock *sk, unsigned flags,
 					 int noblock, int *err);
diff --git a/include/net/inet_ecn.h b/include/net/inet_ecn.h
index f87845e2e96..b0c47e2eccf 100644
--- a/include/net/inet_ecn.h
+++ b/include/net/inet_ecn.h
@@ -2,6 +2,7 @@
 #define _INET_ECN_H_
 
 #include <linux/ip.h>
+#include <linux/skbuff.h>
 #include <net/dsfield.h>
 
 enum {
@@ -48,7 +49,7 @@ static inline __u8 INET_ECN_encapsulate(__u8 outer, __u8 inner)
 		(label) |= __constant_htons(INET_ECN_ECT_0 << 4);	\
     } while (0)
 
-static inline void IP_ECN_set_ce(struct iphdr *iph)
+static inline int IP_ECN_set_ce(struct iphdr *iph)
 {
 	u32 check = iph->check;
 	u32 ecn = (iph->tos + 1) & INET_ECN_MASK;
@@ -61,7 +62,7 @@ static inline void IP_ECN_set_ce(struct iphdr *iph)
 	 * INET_ECN_CE      => 00
 	 */
 	if (!(ecn & 2))
-		return;
+		return !ecn;
 
 	/*
 	 * The following gives us:
@@ -72,6 +73,7 @@ static inline void IP_ECN_set_ce(struct iphdr *iph)
 
 	iph->check = check + (check>=0xFFFF);
 	iph->tos |= INET_ECN_CE;
+	return 1;
 }
 
 static inline void IP_ECN_clear(struct iphdr *iph)
@@ -87,11 +89,12 @@ static inline void ipv4_copy_dscp(struct iphdr *outer, struct iphdr *inner)
 
 struct ipv6hdr;
 
-static inline void IP6_ECN_set_ce(struct ipv6hdr *iph)
+static inline int IP6_ECN_set_ce(struct ipv6hdr *iph)
 {
 	if (INET_ECN_is_not_ect(ipv6_get_dsfield(iph)))
-		return;
+		return 0;
 	*(u32*)iph |= htonl(INET_ECN_CE << 20);
+	return 1;
 }
 
 static inline void IP6_ECN_clear(struct ipv6hdr *iph)
@@ -105,4 +108,21 @@ static inline void ipv6_copy_dscp(struct ipv6hdr *outer, struct ipv6hdr *inner)
 	ipv6_change_dsfield(inner, INET_ECN_MASK, dscp);
 }
 
+static inline int INET_ECN_set_ce(struct sk_buff *skb)
+{
+	switch (skb->protocol) {
+	case __constant_htons(ETH_P_IP):
+		if (skb->nh.raw + sizeof(struct iphdr) <= skb->tail)
+			return IP_ECN_set_ce(skb->nh.iph);
+		break;
+
+	case __constant_htons(ETH_P_IPV6):
+		if (skb->nh.raw + sizeof(struct ipv6hdr) <= skb->tail)
+			return IP6_ECN_set_ce(skb->nh.ipv6h);
+		break;
+	}
+
+	return 0;
+}
+
 #endif
diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index f50f9596834..07840baa934 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -125,9 +125,7 @@ struct inet_hashinfo {
 	rwlock_t			lhash_lock ____cacheline_aligned;
 	atomic_t			lhash_users;
 	wait_queue_head_t		lhash_wait;
-	spinlock_t			portalloc_lock;
 	kmem_cache_t			*bind_bucket_cachep;
-	int				port_rover;
 };
 
 static inline unsigned int inet_ehashfn(const __u32 laddr, const __u16 lport,
diff --git a/include/net/red.h b/include/net/red.h
new file mode 100644
index 00000000000..2ed4358e329
--- /dev/null
+++ b/include/net/red.h
@@ -0,0 +1,325 @@
+#ifndef __NET_SCHED_RED_H
+#define __NET_SCHED_RED_H
+
+#include <linux/config.h>
+#include <linux/types.h>
+#include <net/pkt_sched.h>
+#include <net/inet_ecn.h>
+#include <net/dsfield.h>
+
+/*	Random Early Detection (RED) algorithm.
+	=======================================
+
+	Source: Sally Floyd and Van Jacobson, "Random Early Detection Gateways
+	for Congestion Avoidance", 1993, IEEE/ACM Transactions on Networking.
+
+	This file codes a "divisionless" version of RED algorithm
+	as written down in Fig.17 of the paper.
+
+	Short description.
+	------------------
+
+	When a new packet arrives we calculate the average queue length:
+
+	avg = (1-W)*avg + W*current_queue_len,
+
+	W is the filter time constant (chosen as 2^(-Wlog)), it controls
+	the inertia of the algorithm. To allow larger bursts, W should be
+	decreased.
+
+	if (avg > th_max) -> packet marked (dropped).
+	if (avg < th_min) -> packet passes.
+	if (th_min < avg < th_max) we calculate probability:
+
+	Pb = max_P * (avg - th_min)/(th_max-th_min)
+
+	and mark (drop) packet with this probability.
+	Pb changes from 0 (at avg==th_min) to max_P (avg==th_max).
+	max_P should be small (not 1), usually 0.01..0.02 is good value.
+
+	max_P is chosen as a number, so that max_P/(th_max-th_min)
+	is a negative power of two in order arithmetics to contain
+	only shifts.
+
+
+	Parameters, settable by user:
+	-----------------------------
+
+	qth_min		- bytes (should be < qth_max/2)
+	qth_max		- bytes (should be at least 2*qth_min and less limit)
+	Wlog	       	- bits (<32) log(1/W).
+	Plog	       	- bits (<32)
+
+	Plog is related to max_P by formula:
+
+	max_P = (qth_max-qth_min)/2^Plog;
+
+	F.e. if qth_max=128K and qth_min=32K, then Plog=22
+	corresponds to max_P=0.02
+
+	Scell_log
+	Stab
+
+	Lookup table for log((1-W)^(t/t_ave).
+
+
+	NOTES:
+
+	Upper bound on W.
+	-----------------
+
+	If you want to allow bursts of L packets of size S,
+	you should choose W:
+
+	L + 1 - th_min/S < (1-(1-W)^L)/W
+
+	th_min/S = 32         th_min/S = 4
+
+	log(W)	L
+	-1	33
+	-2	35
+	-3	39
+	-4	46
+	-5	57
+	-6	75
+	-7	101
+	-8	135
+	-9	190
+	etc.
+ */
+
+#define RED_STAB_SIZE	256
+#define RED_STAB_MASK	(RED_STAB_SIZE - 1)
+
+struct red_stats
+{
+	u32		prob_drop;	/* Early probability drops */
+	u32		prob_mark;	/* Early probability marks */
+	u32		forced_drop;	/* Forced drops, qavg > max_thresh */
+	u32		forced_mark;	/* Forced marks, qavg > max_thresh */
+	u32		pdrop;          /* Drops due to queue limits */
+	u32		other;          /* Drops due to drop() calls */
+	u32		backlog;
+};
+
+struct red_parms
+{
+	/* Parameters */
+	u32		qth_min;	/* Min avg length threshold: A scaled */
+	u32		qth_max;	/* Max avg length threshold: A scaled */
+	u32		Scell_max;
+	u32		Rmask;		/* Cached random mask, see red_rmask */
+	u8		Scell_log;
+	u8		Wlog;		/* log(W)		*/
+	u8		Plog;		/* random number bits	*/
+	u8		Stab[RED_STAB_SIZE];
+
+	/* Variables */
+	int		qcount;		/* Number of packets since last random
+					   number generation */
+	u32		qR;		/* Cached random number */
+
+	unsigned long	qavg;		/* Average queue length: A scaled */
+	psched_time_t	qidlestart;	/* Start of current idle period */
+};
+
+static inline u32 red_rmask(u8 Plog)
+{
+	return Plog < 32 ? ((1 << Plog) - 1) : ~0UL;
+}
+
+static inline void red_set_parms(struct red_parms *p,
+				 u32 qth_min, u32 qth_max, u8 Wlog, u8 Plog,
+				 u8 Scell_log, u8 *stab)
+{
+	/* Reset average queue length, the value is strictly bound
+	 * to the parameters below, reseting hurts a bit but leaving
+	 * it might result in an unreasonable qavg for a while. --TGR
+	 */
+	p->qavg		= 0;
+
+	p->qcount	= -1;
+	p->qth_min	= qth_min << Wlog;
+	p->qth_max	= qth_max << Wlog;
+	p->Wlog		= Wlog;
+	p->Plog		= Plog;
+	p->Rmask	= red_rmask(Plog);
+	p->Scell_log	= Scell_log;
+	p->Scell_max	= (255 << Scell_log);
+
+	memcpy(p->Stab, stab, sizeof(p->Stab));
+}
+
+static inline int red_is_idling(struct red_parms *p)
+{
+	return !PSCHED_IS_PASTPERFECT(p->qidlestart);
+}
+
+static inline void red_start_of_idle_period(struct red_parms *p)
+{
+	PSCHED_GET_TIME(p->qidlestart);
+}
+
+static inline void red_end_of_idle_period(struct red_parms *p)
+{
+	PSCHED_SET_PASTPERFECT(p->qidlestart);
+}
+
+static inline void red_restart(struct red_parms *p)
+{
+	red_end_of_idle_period(p);
+	p->qavg = 0;
+	p->qcount = -1;
+}
+
+static inline unsigned long red_calc_qavg_from_idle_time(struct red_parms *p)
+{
+	psched_time_t now;
+	long us_idle;
+	int  shift;
+
+	PSCHED_GET_TIME(now);
+	us_idle = PSCHED_TDIFF_SAFE(now, p->qidlestart, p->Scell_max);
+
+	/*
+	 * The problem: ideally, average length queue recalcultion should
+	 * be done over constant clock intervals. This is too expensive, so
+	 * that the calculation is driven by outgoing packets.
+	 * When the queue is idle we have to model this clock by hand.
+	 *
+	 * SF+VJ proposed to "generate":
+	 *
+	 *	m = idletime / (average_pkt_size / bandwidth)
+	 *
+	 * dummy packets as a burst after idle time, i.e.
+	 *
+	 * 	p->qavg *= (1-W)^m
+	 *
+	 * This is an apparently overcomplicated solution (f.e. we have to
+	 * precompute a table to make this calculation in reasonable time)
+	 * I believe that a simpler model may be used here,
+	 * but it is field for experiments.
+	 */
+
+	shift = p->Stab[(us_idle >> p->Scell_log) & RED_STAB_MASK];
+
+	if (shift)
+		return p->qavg >> shift;
+	else {
+		/* Approximate initial part of exponent with linear function:
+		 *
+		 * 	(1-W)^m ~= 1-mW + ...
+		 *
+		 * Seems, it is the best solution to
+		 * problem of too coarse exponent tabulation.
+		 */
+		us_idle = (p->qavg * us_idle) >> p->Scell_log;
+
+		if (us_idle < (p->qavg >> 1))
+			return p->qavg - us_idle;
+		else
+			return p->qavg >> 1;
+	}
+}
+
+static inline unsigned long red_calc_qavg_no_idle_time(struct red_parms *p,
+						       unsigned int backlog)
+{
+	/*
+	 * NOTE: p->qavg is fixed point number with point at Wlog.
+	 * The formula below is equvalent to floating point
+	 * version:
+	 *
+	 * 	qavg = qavg*(1-W) + backlog*W;
+	 *
+	 * --ANK (980924)
+	 */
+	return p->qavg + (backlog - (p->qavg >> p->Wlog));
+}
+
+static inline unsigned long red_calc_qavg(struct red_parms *p,
+					  unsigned int backlog)
+{
+	if (!red_is_idling(p))
+		return red_calc_qavg_no_idle_time(p, backlog);
+	else
+		return red_calc_qavg_from_idle_time(p);
+}
+
+static inline u32 red_random(struct red_parms *p)
+{
+	return net_random() & p->Rmask;
+}
+
+static inline int red_mark_probability(struct red_parms *p, unsigned long qavg)
+{
+	/* The formula used below causes questions.
+
+	   OK. qR is random number in the interval 0..Rmask
+	   i.e. 0..(2^Plog). If we used floating point
+	   arithmetics, it would be: (2^Plog)*rnd_num,
+	   where rnd_num is less 1.
+
+	   Taking into account, that qavg have fixed
+	   point at Wlog, and Plog is related to max_P by
+	   max_P = (qth_max-qth_min)/2^Plog; two lines
+	   below have the following floating point equivalent:
+
+	   max_P*(qavg - qth_min)/(qth_max-qth_min) < rnd/qcount
+
+	   Any questions? --ANK (980924)
+	 */
+	return !(((qavg - p->qth_min) >> p->Wlog) * p->qcount < p->qR);
+}
+
+enum {
+	RED_BELOW_MIN_THRESH,
+	RED_BETWEEN_TRESH,
+	RED_ABOVE_MAX_TRESH,
+};
+
+static inline int red_cmp_thresh(struct red_parms *p, unsigned long qavg)
+{
+	if (qavg < p->qth_min)
+		return RED_BELOW_MIN_THRESH;
+	else if (qavg >= p->qth_max)
+		return RED_ABOVE_MAX_TRESH;
+	else
+		return RED_BETWEEN_TRESH;
+}
+
+enum {
+	RED_DONT_MARK,
+	RED_PROB_MARK,
+	RED_HARD_MARK,
+};
+
+static inline int red_action(struct red_parms *p, unsigned long qavg)
+{
+	switch (red_cmp_thresh(p, qavg)) {
+		case RED_BELOW_MIN_THRESH:
+			p->qcount = -1;
+			return RED_DONT_MARK;
+
+		case RED_BETWEEN_TRESH:
+			if (++p->qcount) {
+				if (red_mark_probability(p, qavg)) {
+					p->qcount = 0;
+					p->qR = red_random(p);
+					return RED_PROB_MARK;
+				}
+			} else
+				p->qR = red_random(p);
+
+			return RED_DONT_MARK;
+
+		case RED_ABOVE_MAX_TRESH:
+			p->qcount = -1;
+			return RED_HARD_MARK;
+	}
+
+	BUG();
+	return RED_DONT_MARK;
+}
+
+#endif