patch-2.1.99 linux/include/net/pkt_sched.h

Next file: linux/include/net/snmp.h
Previous file: linux/include/net/pkt_cls.h
Back to the patch index
Back to the overall index

diff -u --recursive --new-file v2.1.98/linux/include/net/pkt_sched.h linux/include/net/pkt_sched.h
@@ -1,21 +1,64 @@
 #ifndef __NET_PKT_SCHED_H
 #define __NET_PKT_SCHED_H
 
+#define PSCHED_GETTIMEOFDAY	1
+#define PSCHED_JIFFIES 		2
+#define PSCHED_CPU 		3
+
+#define PSCHED_CLOCK_SOURCE	PSCHED_GETTIMEOFDAY
+
 #include <linux/pkt_sched.h>
+#include <net/pkt_cls.h>
+
+struct rtattr;
+struct Qdisc;
+
+struct qdisc_walker
+{
+	int	stop;
+	int	skip;
+	int	count;
+	int	(*fn)(struct Qdisc *, unsigned long cl, struct qdisc_walker *);
+};
+
+struct Qdisc_class_ops
+{
+	/* Child qdisc manipulation */
+	int			(*graft)(struct Qdisc *, unsigned long cl, struct Qdisc *, struct Qdisc **);
+
+	/* Class manipulation routines */
+	unsigned long		(*get)(struct Qdisc *, u32 classid);
+	void			(*put)(struct Qdisc *, unsigned long);
+	int			(*change)(struct Qdisc *, u32, u32, struct rtattr **, unsigned long *);
+	int			(*delete)(struct Qdisc *, unsigned long);
+	void			(*walk)(struct Qdisc *, struct qdisc_walker * arg);
+
+	/* Filter manipulation */
+	struct tcf_proto **	(*tcf_chain)(struct Qdisc *, unsigned long);
+	unsigned long		(*bind_tcf)(struct Qdisc *, u32 classid);
+	void			(*unbind_tcf)(struct Qdisc *, unsigned long);
+
+	/* rtnetlink specific */
+	int			(*dump)(struct Qdisc *, unsigned long, struct sk_buff *skb, struct tcmsg*);
+};
 
 struct Qdisc_ops
 {
 	struct Qdisc_ops	*next;
+	struct Qdisc_class_ops	*cl_ops;
 	char			id[IFNAMSIZ];
-	int			refcnt;
 	int			priv_size;
-	int 			(*enqueue)(struct sk_buff *skb, struct Qdisc *);
+
+	int 			(*enqueue)(struct sk_buff *, struct Qdisc *);
 	struct sk_buff *	(*dequeue)(struct Qdisc *);
+	int 			(*requeue)(struct sk_buff *, struct Qdisc *);
+	int			(*drop)(struct Qdisc *);
+
+	int			(*init)(struct Qdisc *, struct rtattr *arg);
 	void			(*reset)(struct Qdisc *);
 	void			(*destroy)(struct Qdisc *);
-	int			(*init)(struct Qdisc *, void *arg);
-	int			(*control)(struct Qdisc *, void *);
-	int 			(*requeue)(struct sk_buff *skb, struct Qdisc *);
+
+	int			(*dump)(struct Qdisc *, struct sk_buff *);
 };
 
 struct Qdisc_head
@@ -30,23 +73,35 @@
 	struct Qdisc_head	h;
 	int 			(*enqueue)(struct sk_buff *skb, struct Qdisc *dev);
 	struct sk_buff *	(*dequeue)(struct Qdisc *dev);
+	unsigned		flags;
+#define TCQ_F_DEFAULT	1
+#define TCQ_F_BUILTIN	2
 	struct Qdisc_ops	*ops;
-	int			handle;
+	struct Qdisc		*next;
+	u32			handle;
+	u32			classid;
 	struct Qdisc		*parent;
 	struct sk_buff_head	q;
 	struct device 		*dev;
-	unsigned long		dropped;
-	unsigned long		tx_last;
+
+	struct tc_stats		stats;
 	unsigned long		tx_timeo;
+	unsigned long		tx_last;
+	int			(*reshape_fail)(struct sk_buff *skb, struct Qdisc *q);
 
 	char			data[0];
 };
 
+struct qdisc_rate_table
+{
+	struct tc_ratespec rate;
+	u32		data[256];
+	struct qdisc_rate_table *next;
+	int		refcnt;
+};
 
-/* Yes, it is slow for [34]86, but we have no choice.
-   10 msec resolution is appropriate only for bandwidth < 32Kbit/sec.
 
-   RULE:
+/* 
    Timer resolution MUST BE < 10% of min_schedulable_packet_size/bandwidth
    
    Normal IP packet size ~ 512byte, hence:
@@ -57,22 +112,96 @@
    10msec resolution -> <50Kbit/sec.
    
    The result: [34]86 is not good choice for QoS router :-(
+
+   The things are not so bad, because we may use artifical
+   clock evaluated by integration of network data flow
+   in the most critical places.
+
+   Note: we do not use fastgettimeofday.
+   The reason is that, when it is not the same thing as
+   gettimeofday, it returns invalid timestamp, which is
+   not updated, when net_bh is active.
+
+   So, use PSCHED_CLOCK_SOURCE = PSCHED_CPU on alpha and pentiums
+   with rtdsc. And PSCHED_JIFFIES on all other architectures, including [34]86
+   and pentiums without rtdsc.
+   You can use PSCHED_GETTIMEOFDAY on another architectures,
+   which have fast and precise clock source, but it is too expensive.
  */
 
 
-typedef struct timeval psched_time_t;
+#if PSCHED_CLOCK_SOURCE == PSCHED_GETTIMEOFDAY
 
-/* On 64bit architecures it would be clever to define:
-typedef u64 psched_time_t;
-   and make all this boring arithmetics directly
- */
+typedef struct timeval	psched_time_t;
+typedef long		psched_tdiff_t;
 
-#ifndef SCHEDULE_ONLY_LOW_BANDWIDTH
 #define PSCHED_GET_TIME(stamp) do_gettimeofday(&(stamp))
+#define PSCHED_US2JIFFIE(usecs) (((usecs)+(1000000/HZ-1))/(1000000/HZ))
+
+#else /* PSCHED_CLOCK_SOURCE != PSCHED_GETTIMEOFDAY */
+
+typedef u64	psched_time_t;
+typedef long	psched_tdiff_t;
+
+extern psched_time_t	psched_time_base;
+
+#if PSCHED_CLOCK_SOURCE == PSCHED_JIFFIES
+
+#define PSCHED_WATCHER
+
+extern unsigned long psched_time_mark;
+
+#if HZ == 100
+#define PSCHED_JSCALE 7
+#elif HZ == 1024
+#define PSCHED_JSCALE 10
 #else
-#define PSCHED_GET_TIME(stamp) ((stamp) = xtime)
+#define PSCHED_JSCALE 0
 #endif
 
+#define PSCHED_GET_TIME(stamp) ((stamp) = psched_time_base + (((unsigned long)(jiffies-psched_time_mark))<<PSCHED_JSCALE))
+#define PSCHED_US2JIFFIE(delay) ((delay)>>PSCHED_JSCALE)
+
+#elif PSCHED_CLOCK_SOURCE == PSCHED_CPU
+
+extern psched_tdiff_t psched_clock_per_hz;
+extern int psched_clock_scale;
+
+#define PSCHED_US2JIFFIE(delay) (((delay)+psched_clock_per_hz-1)/psched_clock_per_hz)
+
+#if CPU == 586 || CPU == 686
+
+#define PSCHED_GET_TIME(stamp) \
+({ u32 hi, lo; \
+   __asm__ __volatile__ (".byte 0x0f,0x31" :"=a" (lo), "=d" (hi)); \
+   (stamp) = ((((u64)hi)<<32) + lo)>>psched_clock_scale; \
+})
+
+#elif defined (__alpha__)
+
+#define PSCHED_WATCHER
+
+extern u32 psched_time_mark;
+
+#define PSCHED_GET_TIME(stamp) \
+({ u32 __res; \
+   __asm__ __volatile__ ("rpcc %0" : "r="(__res)); \
+   if (__res <= psched_time_mark) psched_time_base += 0x100000000UL; \
+   psched_time_mark = __res; \
+   (stamp) = (psched_time_base + __res)>>psched_clock_scale; \
+})
+
+#else
+
+#error PSCHED_CLOCK_SOURCE=PSCHED_CPU is not supported on this arch.
+
+#endif /* ARCH */
+
+#endif /* PSCHED_CLOCK_SOURCE == PSCHED_JIFFIES */
+
+#endif /* PSCHED_CLOCK_SOURCE == PSCHED_GETTIMEOFDAY */
+
+#if PSCHED_CLOCK_SOURCE == PSCHED_GETTIMEOFDAY
 #define PSCHED_TDIFF(tv1, tv2) \
 ({ \
 	   int __delta_sec = (tv1).tv_sec - (tv2).tv_sec; \
@@ -106,8 +235,6 @@
 	   __delta; \
 })
 
-#define PSCHED_US2JIFFIE(usecs) (((usecs)+(1000000/HZ-1))/(1000000/HZ))
-
 #define PSCHED_TLESS(tv1, tv2) (((tv1).tv_usec < (tv2).tv_usec && \
 				(tv1).tv_sec <= (tv2).tv_sec) || \
 				 (tv1).tv_sec < (tv2).tv_sec)
@@ -127,24 +254,86 @@
 		 (tv).tv_usec -= 1000000; } \
 })
 
-/* Set/check that undertime is in the "past perfect";
+/* Set/check that time is in the "past perfect";
    it depends on concrete representation of system time
  */
 
 #define PSCHED_SET_PASTPERFECT(t)	((t).tv_sec = 0)
 #define PSCHED_IS_PASTPERFECT(t)	((t).tv_sec == 0)
 
+#define	PSCHED_AUDIT_TDIFF(t) ({ if ((t) > 2000000) (t) = 2000000; })
+
+#else
+
+#define PSCHED_TDIFF(tv1, tv2) (long)((tv1) - (tv2))
+#define PSCHED_TDIFF_SAFE(tv1, tv2, bound, guard) \
+({ \
+	   long __delta = (tv1) - (tv2); \
+	   if ( __delta > (bound)) {  __delta = (bound); guard; } \
+	   __delta; \
+})
+
+
+#define PSCHED_TLESS(tv1, tv2) ((tv1) < (tv2))
+#define PSCHED_TADD2(tv, delta, tv_res) ((tv_res) = (tv) + (delta))
+#define PSCHED_TADD(tv, delta) ((tv) += (delta))
+#define PSCHED_SET_PASTPERFECT(t)	((t) = 0)
+#define PSCHED_IS_PASTPERFECT(t)	((t) == 0)
+#define	PSCHED_AUDIT_TDIFF(t)
+
+#endif
+
+struct tcf_police
+{
+	struct tcf_police *next;
+	int		refcnt;
+	u32		index;
+
+	int		action;
+	u32		burst;
+	u32		mtu;
+
+	u32		toks;
+	u32		ptoks;
+	psched_time_t	t_c;
+	struct qdisc_rate_table *R_tab;
+	struct qdisc_rate_table *P_tab;
+};
+
+extern void tcf_police_destroy(struct tcf_police *p);
+extern struct tcf_police * tcf_police_locate(struct rtattr *rta);
+extern int tcf_police_dump(struct sk_buff *skb, struct tcf_police *p);
+extern int tcf_police(struct sk_buff *skb, struct tcf_police *p);
+
+extern __inline__ void tcf_police_release(struct tcf_police *p)
+{
+	if (p && --p->refcnt == 0)
+		tcf_police_destroy(p);
+}
 
 extern struct Qdisc noop_qdisc;
+extern struct Qdisc_ops noop_qdisc_ops;
+extern struct Qdisc_ops pfifo_qdisc_ops;
+extern struct Qdisc_ops bfifo_qdisc_ops;
 
 int register_qdisc(struct Qdisc_ops *qops);
 int unregister_qdisc(struct Qdisc_ops *qops);
+struct Qdisc *qdisc_lookup(struct device *dev, u32 handle);
+struct Qdisc *qdisc_lookup_class(struct device *dev, u32 handle);
 void dev_init_scheduler(struct device *dev);
 void dev_shutdown(struct device *dev);
 void dev_activate(struct device *dev);
 void dev_deactivate(struct device *dev);
 void qdisc_reset(struct Qdisc *qdisc);
 void qdisc_destroy(struct Qdisc *qdisc);
+struct Qdisc * qdisc_create_dflt(struct device *dev, struct Qdisc_ops *ops);
+struct Qdisc * dev_set_scheduler(struct device *dev, struct Qdisc *qdisc);
+int qdisc_new_estimator(struct tc_stats *stats, struct rtattr *opt);
+void qdisc_kill_estimator(struct tc_stats *stats);
+struct qdisc_rate_table *qdisc_get_rtab(struct tc_ratespec *r, struct rtattr *tab);
+void qdisc_put_rtab(struct qdisc_rate_table *tab);
+int teql_init(void);
+int tc_filter_init(void);
 int pktsched_init(void);
 
 void qdisc_run_queues(void);
@@ -159,6 +348,12 @@
 			qdisc_head.forw = &q->h;
 		}
 	}
+}
+
+extern __inline__ unsigned psched_mtu(struct device *dev)
+{
+	unsigned mtu = dev->mtu;
+	return dev->hard_header ? mtu + dev->hard_header_len : mtu;
 }
 
 #endif

FUNET's LINUX-ADM group, linux-adm@nic.funet.fi
TCL-scripts by Sam Shen, slshen@lbl.gov