diff options
Diffstat (limited to 'net/sched/sch_htb.c')
| -rw-r--r-- | net/sched/sch_htb.c | 608 |
1 files changed, 328 insertions, 280 deletions
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c index 01b519d6c52..9f949abcace 100644 --- a/net/sched/sch_htb.c +++ b/net/sched/sch_htb.c @@ -38,6 +38,7 @@ #include <linux/workqueue.h> #include <linux/slab.h> #include <net/netlink.h> +#include <net/sch_generic.h> #include <net/pkt_sched.h> /* HTB algorithm. @@ -64,6 +65,10 @@ static int htb_hysteresis __read_mostly = 0; /* whether to use mode hysteresis f module_param (htb_hysteresis, int, 0640); MODULE_PARM_DESC(htb_hysteresis, "Hysteresis mode, less CPU load, less accurate"); +static int htb_rate_est = 0; /* htb classes have a default rate estimator */ +module_param(htb_rate_est, int, 0640); +MODULE_PARM_DESC(htb_rate_est, "setup a default rate estimator (4sec 16sec) for htb classes"); + /* used internaly to keep status of single class */ enum htb_cmode { HTB_CANT_SEND, /* class can't send and can't borrow */ @@ -71,94 +76,105 @@ enum htb_cmode { HTB_CAN_SEND /* class can send */ }; -/* interior & leaf nodes; props specific to leaves are marked L: */ +struct htb_prio { + union { + struct rb_root row; + struct rb_root feed; + }; + struct rb_node *ptr; + /* When class changes from state 1->2 and disconnects from + * parent's feed then we lost ptr value and start from the + * first child again. Here we store classid of the + * last valid ptr (used when ptr is NULL). + */ + u32 last_ptr_id; +}; + +/* interior & leaf nodes; props specific to leaves are marked L: + * To reduce false sharing, place mostly read fields at beginning, + * and mostly written ones at the end. + */ struct htb_class { struct Qdisc_class_common common; - /* general class parameters */ - struct gnet_stats_basic_packed bstats; - struct gnet_stats_queue qstats; - struct gnet_stats_rate_est rate_est; - struct tc_htb_xstats xstats; /* our special stats */ - int refcnt; /* usage count of this class */ + struct psched_ratecfg rate; + struct psched_ratecfg ceil; + s64 buffer, cbuffer;/* token bucket depth/rate */ + s64 mbuffer; /* max wait time */ + u32 prio; /* these two are used only by leaves... */ + int quantum; /* but stored for parent-to-leaf return */ + + struct tcf_proto *filter_list; /* class attached filters */ + int filter_cnt; + int refcnt; /* usage count of this class */ - /* topology */ - int level; /* our level (see above) */ - unsigned int children; - struct htb_class *parent; /* parent class */ + int level; /* our level (see above) */ + unsigned int children; + struct htb_class *parent; /* parent class */ - int prio; /* these two are used only by leaves... */ - int quantum; /* but stored for parent-to-leaf return */ + struct gnet_stats_rate_est64 rate_est; + + /* + * Written often fields + */ + struct gnet_stats_basic_packed bstats; + struct gnet_stats_queue qstats; + struct tc_htb_xstats xstats; /* our special stats */ + + /* token bucket parameters */ + s64 tokens, ctokens;/* current number of tokens */ + s64 t_c; /* checkpoint time */ union { struct htb_class_leaf { - struct Qdisc *q; - int deficit[TC_HTB_MAXDEPTH]; struct list_head drop_list; + int deficit[TC_HTB_MAXDEPTH]; + struct Qdisc *q; } leaf; struct htb_class_inner { - struct rb_root feed[TC_HTB_NUMPRIO]; /* feed trees */ - struct rb_node *ptr[TC_HTB_NUMPRIO]; /* current class ptr */ - /* When class changes from state 1->2 and disconnects from - parent's feed then we lost ptr value and start from the - first child again. Here we store classid of the - last valid ptr (used when ptr is NULL). */ - u32 last_ptr_id[TC_HTB_NUMPRIO]; + struct htb_prio clprio[TC_HTB_NUMPRIO]; } inner; } un; - struct rb_node node[TC_HTB_NUMPRIO]; /* node for self or feed tree */ - struct rb_node pq_node; /* node for event queue */ - psched_time_t pq_key; - - int prio_activity; /* for which prios are we active */ - enum htb_cmode cmode; /* current mode of the class */ + s64 pq_key; - /* class attached filters */ - struct tcf_proto *filter_list; - int filter_cnt; + int prio_activity; /* for which prios are we active */ + enum htb_cmode cmode; /* current mode of the class */ + struct rb_node pq_node; /* node for event queue */ + struct rb_node node[TC_HTB_NUMPRIO]; /* node for self or feed tree */ +}; - /* token bucket parameters */ - struct qdisc_rate_table *rate; /* rate table of the class itself */ - struct qdisc_rate_table *ceil; /* ceiling rate (limits borrows too) */ - long buffer, cbuffer; /* token bucket depth/rate */ - psched_tdiff_t mbuffer; /* max wait time */ - long tokens, ctokens; /* current number of tokens */ - psched_time_t t_c; /* checkpoint time */ +struct htb_level { + struct rb_root wait_pq; + struct htb_prio hprio[TC_HTB_NUMPRIO]; }; struct htb_sched { struct Qdisc_class_hash clhash; - struct list_head drops[TC_HTB_NUMPRIO];/* active leaves (for drops) */ + int defcls; /* class where unclassified flows go to */ + int rate2quantum; /* quant = rate / rate2quantum */ - /* self list - roots of self generating tree */ - struct rb_root row[TC_HTB_MAXDEPTH][TC_HTB_NUMPRIO]; - int row_mask[TC_HTB_MAXDEPTH]; - struct rb_node *ptr[TC_HTB_MAXDEPTH][TC_HTB_NUMPRIO]; - u32 last_ptr_id[TC_HTB_MAXDEPTH][TC_HTB_NUMPRIO]; - - /* self wait list - roots of wait PQs per row */ - struct rb_root wait_pq[TC_HTB_MAXDEPTH]; + /* filters for qdisc itself */ + struct tcf_proto *filter_list; - /* time of nearest event per level (row) */ - psched_time_t near_ev_cache[TC_HTB_MAXDEPTH]; +#define HTB_WARN_TOOMANYEVENTS 0x1 + unsigned int warned; /* only one warning */ + int direct_qlen; + struct work_struct work; - int defcls; /* class where unclassified flows go to */ + /* non shaped skbs; let them go directly thru */ + struct sk_buff_head direct_queue; + long direct_pkts; - /* filters for qdisc itself */ - struct tcf_proto *filter_list; + struct qdisc_watchdog watchdog; - int rate2quantum; /* quant = rate / rate2quantum */ - psched_time_t now; /* cached dequeue time */ - struct qdisc_watchdog watchdog; + s64 now; /* cached dequeue time */ + struct list_head drops[TC_HTB_NUMPRIO];/* active leaves (for drops) */ - /* non shaped skbs; let them go directly thru */ - struct sk_buff_head direct_queue; - int direct_qlen; /* max qlen of above */ + /* time of nearest event per level (row) */ + s64 near_ev_cache[TC_HTB_MAXDEPTH]; - long direct_pkts; + int row_mask[TC_HTB_MAXDEPTH]; -#define HTB_WARN_TOOMANYEVENTS 0x1 - unsigned int warned; /* only one warning */ - struct work_struct work; + struct htb_level hlevel[TC_HTB_MAXDEPTH]; }; /* find class in global hash table using given handle */ @@ -182,10 +198,10 @@ static inline struct htb_class *htb_find(u32 handle, struct Qdisc *sch) * filters in qdisc and in inner nodes (if higher filter points to the inner * node). If we end up with classid MAJOR:0 we enqueue the skb into special * internal fifo (direct). These packets then go directly thru. If we still - * have no valid leaf we try to use MAJOR:default leaf. It still unsuccessfull + * have no valid leaf we try to use MAJOR:default leaf. It still unsuccessful * then finish and return direct queue. */ -#define HTB_DIRECT (struct htb_class*)-1 +#define HTB_DIRECT ((struct htb_class *)-1L) static struct htb_class *htb_classify(struct sk_buff *skb, struct Qdisc *sch, int *qerr) @@ -197,15 +213,22 @@ static struct htb_class *htb_classify(struct sk_buff *skb, struct Qdisc *sch, int result; /* allow to select class by setting skb->priority to valid classid; - note that nfmark can be used too by attaching filter fw with no - rules in it */ + * note that nfmark can be used too by attaching filter fw with no + * rules in it + */ if (skb->priority == sch->handle) return HTB_DIRECT; /* X:0 (direct flow) selected */ - if ((cl = htb_find(skb->priority, sch)) != NULL && cl->level == 0) - return cl; + cl = htb_find(skb->priority, sch); + if (cl) { + if (cl->level == 0) + return cl; + /* Start with inner filter chain if a non-leaf class is selected */ + tcf = cl->filter_list; + } else { + tcf = q->filter_list; + } *qerr = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS; - tcf = q->filter_list; while (tcf && (result = tc_classify(skb, tcf, &res)) >= 0) { #ifdef CONFIG_NET_CLS_ACT switch (result) { @@ -216,10 +239,12 @@ static struct htb_class *htb_classify(struct sk_buff *skb, struct Qdisc *sch, return NULL; } #endif - if ((cl = (void *)res.class) == NULL) { + cl = (void *)res.class; + if (!cl) { if (res.classid == sch->handle) return HTB_DIRECT; /* X:0 (direct flow) */ - if ((cl = htb_find(res.classid, sch)) == NULL) + cl = htb_find(res.classid, sch); + if (!cl) break; /* filter selected invalid classid */ } if (!cl->level) @@ -268,9 +293,9 @@ static void htb_add_to_id_tree(struct rb_root *root, * already in the queue. */ static void htb_add_to_wait_tree(struct htb_sched *q, - struct htb_class *cl, long delay) + struct htb_class *cl, s64 delay) { - struct rb_node **p = &q->wait_pq[cl->level].rb_node, *parent = NULL; + struct rb_node **p = &q->hlevel[cl->level].wait_pq.rb_node, *parent = NULL; cl->pq_key = q->now + delay; if (cl->pq_key == q->now) @@ -290,7 +315,7 @@ static void htb_add_to_wait_tree(struct htb_sched *q, p = &parent->rb_left; } rb_link_node(&cl->pq_node, parent, p); - rb_insert_color(&cl->pq_node, &q->wait_pq[cl->level]); + rb_insert_color(&cl->pq_node, &q->hlevel[cl->level].wait_pq); } /** @@ -317,7 +342,7 @@ static inline void htb_add_class_to_row(struct htb_sched *q, while (mask) { int prio = ffz(~mask); mask &= ~(1 << prio); - htb_add_to_id_tree(q->row[cl->level] + prio, cl, prio); + htb_add_to_id_tree(&q->hlevel[cl->level].hprio[prio].row, cl, prio); } } @@ -343,16 +368,18 @@ static inline void htb_remove_class_from_row(struct htb_sched *q, struct htb_class *cl, int mask) { int m = 0; + struct htb_level *hlevel = &q->hlevel[cl->level]; while (mask) { int prio = ffz(~mask); + struct htb_prio *hprio = &hlevel->hprio[prio]; mask &= ~(1 << prio); - if (q->ptr[cl->level][prio] == cl->node + prio) - htb_next_rb_node(q->ptr[cl->level] + prio); + if (hprio->ptr == cl->node + prio) + htb_next_rb_node(&hprio->ptr); - htb_safe_rb_erase(cl->node + prio, q->row[cl->level] + prio); - if (!q->row[cl->level][prio].rb_node) + htb_safe_rb_erase(cl->node + prio, &hprio->row); + if (!hprio->row.rb_node) m |= 1 << prio; } q->row_mask[cl->level] &= ~m; @@ -376,12 +403,13 @@ static void htb_activate_prios(struct htb_sched *q, struct htb_class *cl) int prio = ffz(~m); m &= ~(1 << prio); - if (p->un.inner.feed[prio].rb_node) + if (p->un.inner.clprio[prio].feed.rb_node) /* parent already has its feed in use so that - reset bit in mask as parent is already ok */ + * reset bit in mask as parent is already ok + */ mask &= ~(1 << prio); - htb_add_to_id_tree(p->un.inner.feed + prio, cl, prio); + htb_add_to_id_tree(&p->un.inner.clprio[prio].feed, cl, prio); } p->prio_activity |= mask; cl = p; @@ -411,17 +439,19 @@ static void htb_deactivate_prios(struct htb_sched *q, struct htb_class *cl) int prio = ffz(~m); m &= ~(1 << prio); - if (p->un.inner.ptr[prio] == cl->node + prio) { + if (p->un.inner.clprio[prio].ptr == cl->node + prio) { /* we are removing child which is pointed to from - parent feed - forget the pointer but remember - classid */ - p->un.inner.last_ptr_id[prio] = cl->common.classid; - p->un.inner.ptr[prio] = NULL; + * parent feed - forget the pointer but remember + * classid + */ + p->un.inner.clprio[prio].last_ptr_id = cl->common.classid; + p->un.inner.clprio[prio].ptr = NULL; } - htb_safe_rb_erase(cl->node + prio, p->un.inner.feed + prio); + htb_safe_rb_erase(cl->node + prio, + &p->un.inner.clprio[prio].feed); - if (!p->un.inner.feed[prio].rb_node) + if (!p->un.inner.clprio[prio].feed.rb_node) mask |= 1 << prio; } @@ -434,14 +464,14 @@ static void htb_deactivate_prios(struct htb_sched *q, struct htb_class *cl) htb_remove_class_from_row(q, cl, mask); } -static inline long htb_lowater(const struct htb_class *cl) +static inline s64 htb_lowater(const struct htb_class *cl) { if (htb_hysteresis) return cl->cmode != HTB_CANT_SEND ? -cl->cbuffer : 0; else return 0; } -static inline long htb_hiwater(const struct htb_class *cl) +static inline s64 htb_hiwater(const struct htb_class *cl) { if (htb_hysteresis) return cl->cmode == HTB_CAN_SEND ? -cl->buffer : 0; @@ -462,9 +492,9 @@ static inline long htb_hiwater(const struct htb_class *cl) * mode transitions per time unit. The speed gain is about 1/6. */ static inline enum htb_cmode -htb_class_mode(struct htb_class *cl, long *diff) +htb_class_mode(struct htb_class *cl, s64 *diff) { - long toks; + s64 toks; if ((toks = (cl->ctokens + *diff)) < htb_lowater(cl)) { *diff = -toks; @@ -488,7 +518,7 @@ htb_class_mode(struct htb_class *cl, long *diff) * to mode other than HTB_CAN_SEND (see htb_add_to_wait_tree). */ static void -htb_change_class_mode(struct htb_sched *q, struct htb_class *cl, long *diff) +htb_change_class_mode(struct htb_sched *q, struct htb_class *cl, s64 *diff) { enum htb_cmode new_mode = htb_class_mode(cl, diff); @@ -551,9 +581,7 @@ static int htb_enqueue(struct sk_buff *skb, struct Qdisc *sch) __skb_queue_tail(&q->direct_queue, skb); q->direct_pkts++; } else { - kfree_skb(skb); - sch->qstats.drops++; - return NET_XMIT_DROP; + return qdisc_drop(skb, sch); } #ifdef CONFIG_NET_CLS_ACT } else if (!cl) { @@ -569,38 +597,33 @@ static int htb_enqueue(struct sk_buff *skb, struct Qdisc *sch) } return ret; } else { - cl->bstats.packets += - skb_is_gso(skb)?skb_shinfo(skb)->gso_segs:1; - cl->bstats.bytes += qdisc_pkt_len(skb); htb_activate(q, cl); } sch->q.qlen++; - sch->bstats.packets += skb_is_gso(skb)?skb_shinfo(skb)->gso_segs:1; - sch->bstats.bytes += qdisc_pkt_len(skb); return NET_XMIT_SUCCESS; } -static inline void htb_accnt_tokens(struct htb_class *cl, int bytes, long diff) +static inline void htb_accnt_tokens(struct htb_class *cl, int bytes, s64 diff) { - long toks = diff + cl->tokens; + s64 toks = diff + cl->tokens; if (toks > cl->buffer) toks = cl->buffer; - toks -= (long) qdisc_l2t(cl->rate, bytes); + toks -= (s64) psched_l2t_ns(&cl->rate, bytes); if (toks <= -cl->mbuffer) toks = 1 - cl->mbuffer; cl->tokens = toks; } -static inline void htb_accnt_ctokens(struct htb_class *cl, int bytes, long diff) +static inline void htb_accnt_ctokens(struct htb_class *cl, int bytes, s64 diff) { - long toks = diff + cl->ctokens; + s64 toks = diff + cl->ctokens; if (toks > cl->cbuffer) toks = cl->cbuffer; - toks -= (long) qdisc_l2t(cl->ceil, bytes); + toks -= (s64) psched_l2t_ns(&cl->ceil, bytes); if (toks <= -cl->mbuffer) toks = 1 - cl->mbuffer; @@ -623,10 +646,10 @@ static void htb_charge_class(struct htb_sched *q, struct htb_class *cl, { int bytes = qdisc_pkt_len(skb); enum htb_cmode old_mode; - long diff; + s64 diff; while (cl) { - diff = psched_tdiff_bounded(q->now, cl->t_c, cl->mbuffer); + diff = min_t(s64, q->now - cl->t_c, cl->mbuffer); if (cl->level >= level) { if (cl->level == level) cl->xstats.lends++; @@ -643,17 +666,15 @@ static void htb_charge_class(struct htb_sched *q, struct htb_class *cl, htb_change_class_mode(q, cl, &diff); if (old_mode != cl->cmode) { if (old_mode != HTB_CAN_SEND) - htb_safe_rb_erase(&cl->pq_node, q->wait_pq + cl->level); + htb_safe_rb_erase(&cl->pq_node, &q->hlevel[cl->level].wait_pq); if (cl->cmode != HTB_CAN_SEND) htb_add_to_wait_tree(q, cl, diff); } - /* update byte stats except for leaves which are already updated */ - if (cl->level) { - cl->bstats.bytes += bytes; - cl->bstats.packets += skb_is_gso(skb)? - skb_shinfo(skb)->gso_segs:1; - } + /* update basic stats except for leaves which are already updated */ + if (cl->level) + bstats_update(&cl->bstats, skb); + cl = cl->parent; } } @@ -665,17 +686,20 @@ static void htb_charge_class(struct htb_sched *q, struct htb_class *cl, * next pending event (0 for no event in pq, q->now for too many events). * Note: Applied are events whose have cl->pq_key <= q->now. */ -static psched_time_t htb_do_events(struct htb_sched *q, int level, - unsigned long start) +static s64 htb_do_events(struct htb_sched *q, const int level, + unsigned long start) { /* don't run for longer than 2 jiffies; 2 is used instead of - 1 to simplify things when jiffy is going to be incremented - too soon */ + * 1 to simplify things when jiffy is going to be incremented + * too soon + */ unsigned long stop_at = start + 2; + struct rb_root *wait_pq = &q->hlevel[level].wait_pq; + while (time_before(jiffies, stop_at)) { struct htb_class *cl; - long diff; - struct rb_node *p = rb_first(&q->wait_pq[level]); + s64 diff; + struct rb_node *p = rb_first(wait_pq); if (!p) return 0; @@ -684,8 +708,8 @@ static psched_time_t htb_do_events(struct htb_sched *q, int level, if (cl->pq_key > q->now) return cl->pq_key; - htb_safe_rb_erase(p, q->wait_pq + level); - diff = psched_tdiff_bounded(q->now, cl->t_c, cl->mbuffer); + htb_safe_rb_erase(p, wait_pq); + diff = min_t(s64, q->now - cl->t_c, cl->mbuffer); htb_change_class_mode(q, cl, &diff); if (cl->cmode != HTB_CAN_SEND) htb_add_to_wait_tree(q, cl, diff); @@ -693,7 +717,7 @@ static psched_time_t htb_do_events(struct htb_sched *q, int level, /* too much load - let's continue after a break for scheduling */ if (!(q->warned & HTB_WARN_TOOMANYEVENTS)) { - printk(KERN_WARNING "htb: too many events!\n"); + pr_warn("htb: too many events!\n"); q->warned |= HTB_WARN_TOOMANYEVENTS; } @@ -701,7 +725,8 @@ static psched_time_t htb_do_events(struct htb_sched *q, int level, } /* Returns class->node+prio from id-tree where classe's id is >= id. NULL - is no such one exists. */ + * is no such one exists. + */ static struct rb_node *htb_id_find_next_upper(int prio, struct rb_node *n, u32 id) { @@ -727,8 +752,7 @@ static struct rb_node *htb_id_find_next_upper(int prio, struct rb_node *n, * * Find leaf where current feed pointers points to. */ -static struct htb_class *htb_lookup_leaf(struct rb_root *tree, int prio, - struct rb_node **pptr, u32 * pid) +static struct htb_class *htb_lookup_leaf(struct htb_prio *hprio, const int prio) { int i; struct { @@ -737,20 +761,22 @@ static struct htb_class *htb_lookup_leaf(struct rb_root *tree, int prio, u32 *pid; } stk[TC_HTB_MAXDEPTH], *sp = stk; - BUG_ON(!tree->rb_node); - sp->root = tree->rb_node; - sp->pptr = pptr; - sp->pid = pid; + BUG_ON(!hprio->row.rb_node); + sp->root = hprio->row.rb_node; + sp->pptr = &hprio->ptr; + sp->pid = &hprio->last_ptr_id; for (i = 0; i < 65535; i++) { if (!*sp->pptr && *sp->pid) { /* ptr was invalidated but id is valid - try to recover - the original or next ptr */ + * the original or next ptr + */ *sp->pptr = htb_id_find_next_upper(prio, sp->root, *sp->pid); } *sp->pid = 0; /* ptr is valid now so that remove this hint as it - can become out of date quickly */ + * can become out of date quickly + */ if (!*sp->pptr) { /* we are at right end; rewind & go up */ *sp->pptr = sp->root; while ((*sp->pptr)->rb_left) @@ -765,12 +791,15 @@ static struct htb_class *htb_lookup_leaf(struct rb_root *tree, int prio, } } else { struct htb_class *cl; + struct htb_prio *clp; + cl = rb_entry(*sp->pptr, struct htb_class, node[prio]); if (!cl->level) return cl; - (++sp)->root = cl->un.inner.feed[prio].rb_node; - sp->pptr = cl->un.inner.ptr + prio; - sp->pid = cl->un.inner.last_ptr_id + prio; + clp = &cl->un.inner.clprio[prio]; + (++sp)->root = clp->feed.rb_node; + sp->pptr = &clp->ptr; + sp->pid = &clp->last_ptr_id; } } WARN_ON(1); @@ -778,16 +807,18 @@ static struct htb_class *htb_lookup_leaf(struct rb_root *tree, int prio, } /* dequeues packet at given priority and level; call only if - you are sure that there is active class at prio/level */ -static struct sk_buff *htb_dequeue_tree(struct htb_sched *q, int prio, - int level) + * you are sure that there is active class at prio/level + */ +static struct sk_buff *htb_dequeue_tree(struct htb_sched *q, const int prio, + const int level) { struct sk_buff *skb = NULL; struct htb_class *cl, *start; + struct htb_level *hlevel = &q->hlevel[level]; + struct htb_prio *hprio = &hlevel->hprio[prio]; + /* look initial class up in the row */ - start = cl = htb_lookup_leaf(q->row[level] + prio, prio, - q->ptr[level] + prio, - q->last_ptr_id[level] + prio); + start = cl = htb_lookup_leaf(hprio, prio); do { next: @@ -795,9 +826,10 @@ next: return NULL; /* class can be empty - it is unlikely but can be true if leaf - qdisc drops packets in enqueue routine or if someone used - graft operation on the leaf since last dequeue; - simply deactivate and skip such class */ + * qdisc drops packets in enqueue routine or if someone used + * graft operation on the leaf since last dequeue; + * simply deactivate and skip such class + */ if (unlikely(cl->un.leaf.q->q.qlen == 0)) { struct htb_class *next; htb_deactivate(q, cl); @@ -806,9 +838,7 @@ next: if ((q->row_mask[level] & (1 << prio)) == 0) return NULL; - next = htb_lookup_leaf(q->row[level] + prio, - prio, q->ptr[level] + prio, - q->last_ptr_id[level] + prio); + next = htb_lookup_leaf(hprio, prio); if (cl == start) /* fix start if we just deleted it */ start = next; @@ -821,23 +851,23 @@ next: break; qdisc_warn_nonwc("htb", cl->un.leaf.q); - htb_next_rb_node((level ? cl->parent->un.inner.ptr : q-> - ptr[0]) + prio); - cl = htb_lookup_leaf(q->row[level] + prio, prio, - q->ptr[level] + prio, - q->last_ptr_id[level] + prio); + htb_next_rb_node(level ? &cl->parent->un.inner.clprio[prio].ptr: + &q->hlevel[0].hprio[prio].ptr); + cl = htb_lookup_leaf(hprio, prio); } while (cl != start); if (likely(skb != NULL)) { + bstats_update(&cl->bstats, skb); cl->un.leaf.deficit[level] -= qdisc_pkt_len(skb); if (cl->un.leaf.deficit[level] < 0) { cl->un.leaf.deficit[level] += cl->quantum; - htb_next_rb_node((level ? cl->parent->un.inner.ptr : q-> - ptr[0]) + prio); + htb_next_rb_node(level ? &cl->parent->un.inner.clprio[prio].ptr : + &q->hlevel[0].hprio[prio].ptr); } /* this used to be after charge_class but this constelation - gives us slightly better performance */ + * gives us slightly better performance + */ if (!cl->un.leaf.q->q.qlen) htb_deactivate(q, cl); htb_charge_class(q, cl, level, skb); @@ -847,39 +877,40 @@ next: static struct sk_buff *htb_dequeue(struct Qdisc *sch) { - struct sk_buff *skb = NULL; + struct sk_buff *skb; struct htb_sched *q = qdisc_priv(sch); int level; - psched_time_t next_event; + s64 next_event; unsigned long start_at; /* try to dequeue direct packets as high prio (!) to minimize cpu work */ skb = __skb_dequeue(&q->direct_queue); if (skb != NULL) { - sch->flags &= ~TCQ_F_THROTTLED; +ok: + qdisc_bstats_update(sch, skb); + qdisc_unthrottled(sch); sch->q.qlen--; return skb; } if (!sch->q.qlen) goto fin; - q->now = psched_get_time(); + q->now = ktime_to_ns(ktime_get()); start_at = jiffies; - next_event = q->now + 5 * PSCHED_TICKS_PER_SEC; + next_event = q->now + 5LLU * NSEC_PER_SEC; for (level = 0; level < TC_HTB_MAXDEPTH; level++) { /* common case optimization - skip event handler quickly */ int m; - psched_time_t event; + s64 event = q->near_ev_cache[level]; - if (q->now >= q->near_ev_cache[level]) { + if (q->now >= event) { event = htb_do_events(q, level, start_at); if (!event) - event = q->now + PSCHED_TICKS_PER_SEC; + event = q->now + NSEC_PER_SEC; q->near_ev_cache[level] = event; - } else - event = q->near_ev_cache[level]; + } if (next_event > event) next_event = event; @@ -887,20 +918,25 @@ static struct sk_buff *htb_dequeue(struct Qdisc *sch) m = ~q->row_mask[level]; while (m != (int)(-1)) { int prio = ffz(m); + m |= 1 << prio; skb = htb_dequeue_tree(q, prio, level); - if (likely(skb != NULL)) { - sch->q.qlen--; - sch->flags &= ~TCQ_F_THROTTLED; - goto fin; - } + if (likely(skb != NULL)) + goto ok; } } sch->qstats.overlimits++; - if (likely(next_event > q->now)) - qdisc_watchdog_schedule(&q->watchdog, next_event); - else + if (likely(next_event > q->now)) { + if (!test_bit(__QDISC_STATE_DEACTIVATED, + &qdisc_root_sleeping(q->watchdog.qdisc)->state)) { + ktime_t time = ns_to_ktime(next_event); + qdisc_throttled(q->watchdog.qdisc); + hrtimer_start(&q->watchdog.timer, time, + HRTIMER_MODE_ABS); + } + } else { schedule_work(&q->work); + } fin: return skb; } @@ -935,11 +971,10 @@ static void htb_reset(struct Qdisc *sch) { struct htb_sched *q = qdisc_priv(sch); struct htb_class *cl; - struct hlist_node *n; unsigned int i; for (i = 0; i < q->clhash.hashsize; i++) { - hlist_for_each_entry(cl, n, &q->clhash.hash[i], common.hnode) { + hlist_for_each_entry(cl, &q->clhash.hash[i], common.hnode) { if (cl->level) memset(&cl->un.inner, 0, sizeof(cl->un.inner)); else { @@ -955,10 +990,8 @@ static void htb_reset(struct Qdisc *sch) qdisc_watchdog_cancel(&q->watchdog); __skb_queue_purge(&q->direct_queue); sch->q.qlen = 0; - memset(q->row, 0, sizeof(q->row)); + memset(q->hlevel, 0, sizeof(q->hlevel)); memset(q->row_mask, 0, sizeof(q->row_mask)); - memset(q->wait_pq, 0, sizeof(q->wait_pq)); - memset(q->ptr, 0, sizeof(q->ptr)); for (i = 0; i < TC_HTB_NUMPRIO; i++) INIT_LIST_HEAD(q->drops + i); } @@ -968,6 +1001,9 @@ static const struct nla_policy htb_policy[TCA_HTB_MAX + 1] = { [TCA_HTB_INIT] = { .len = sizeof(struct tc_htb_glob) }, [TCA_HTB_CTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE }, [TCA_HTB_RTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE }, + [TCA_HTB_DIRECT_QLEN] = { .type = NLA_U32 }, + [TCA_HTB_RATE64] = { .type = NLA_U64 }, + [TCA_HTB_CEIL64] = { .type = NLA_U64 }, }; static void htb_work_func(struct work_struct *work) @@ -981,7 +1017,7 @@ static void htb_work_func(struct work_struct *work) static int htb_init(struct Qdisc *sch, struct nlattr *opt) { struct htb_sched *q = qdisc_priv(sch); - struct nlattr *tb[TCA_HTB_INIT + 1]; + struct nlattr *tb[TCA_HTB_MAX + 1]; struct tc_htb_glob *gopt; int err; int i; @@ -989,21 +1025,16 @@ static int htb_init(struct Qdisc *sch, struct nlattr *opt) if (!opt) return -EINVAL; - err = nla_parse_nested(tb, TCA_HTB_INIT, opt, htb_policy); + err = nla_parse_nested(tb, TCA_HTB_MAX, opt, htb_policy); if (err < 0) return err; - if (tb[TCA_HTB_INIT] == NULL) { - printk(KERN_ERR "HTB: hey probably you have bad tc tool ?\n"); + if (!tb[TCA_HTB_INIT]) return -EINVAL; - } + gopt = nla_data(tb[TCA_HTB_INIT]); - if (gopt->version != HTB_VER >> 16) { - printk(KERN_ERR - "HTB: need tc/htb version %d (minor is %d), you have %d\n", - HTB_VER >> 16, HTB_VER & 0xffff, gopt->version); + if (gopt->version != HTB_VER >> 16) return -EINVAL; - } err = qdisc_class_hash_init(&q->clhash); if (err < 0) @@ -1015,10 +1046,13 @@ static int htb_init(struct Qdisc *sch, struct nlattr *opt) INIT_WORK(&q->work, htb_work_func); skb_queue_head_init(&q->direct_queue); - q->direct_qlen = qdisc_dev(sch)->tx_queue_len; - if (q->direct_qlen < 2) /* some devices have zero tx_queue_len */ - q->direct_qlen = 2; - + if (tb[TCA_HTB_DIRECT_QLEN]) + q->direct_qlen = nla_get_u32(tb[TCA_HTB_DIRECT_QLEN]); + else { + q->direct_qlen = qdisc_dev(sch)->tx_queue_len; + if (q->direct_qlen < 2) /* some devices have zero tx_queue_len */ + q->direct_qlen = 2; + } if ((q->rate2quantum = gopt->rate2quantum) < 1) q->rate2quantum = 1; q->defcls = gopt->defcls; @@ -1028,12 +1062,13 @@ static int htb_init(struct Qdisc *sch, struct nlattr *opt) static int htb_dump(struct Qdisc *sch, struct sk_buff *skb) { - spinlock_t *root_lock = qdisc_root_sleeping_lock(sch); struct htb_sched *q = qdisc_priv(sch); struct nlattr *nest; struct tc_htb_glob gopt; - spin_lock_bh(root_lock); + /* Its safe to not acquire qdisc lock. As we hold RTNL, + * no change can happen on the qdisc parameters. + */ gopt.direct_pkts = q->direct_pkts; gopt.version = HTB_VER; @@ -1044,14 +1079,13 @@ static int htb_dump(struct Qdisc *sch, struct sk_buff *skb) nest = nla_nest_start(skb, TCA_OPTIONS); if (nest == NULL) goto nla_put_failure; - NLA_PUT(skb, TCA_HTB_INIT, sizeof(gopt), &gopt); - nla_nest_end(skb, nest); + if (nla_put(skb, TCA_HTB_INIT, sizeof(gopt), &gopt) || + nla_put_u32(skb, TCA_HTB_DIRECT_QLEN, q->direct_qlen)) + goto nla_put_failure; - spin_unlock_bh(root_lock); - return skb->len; + return nla_nest_end(skb, nest); nla_put_failure: - spin_unlock_bh(root_lock); nla_nest_cancel(skb, nest); return -1; } @@ -1060,11 +1094,12 @@ static int htb_dump_class(struct Qdisc *sch, unsigned long arg, struct sk_buff *skb, struct tcmsg *tcm) { struct htb_class *cl = (struct htb_class *)arg; - spinlock_t *root_lock = qdisc_root_sleeping_lock(sch); struct nlattr *nest; struct tc_htb_opt opt; - spin_lock_bh(root_lock); + /* Its safe to not acquire qdisc lock. As we hold RTNL, + * no change can happen on the class parameters. + */ tcm->tcm_parent = cl->parent ? cl->parent->common.classid : TC_H_ROOT; tcm->tcm_handle = cl->common.classid; if (!cl->level && cl->un.leaf.q) @@ -1076,21 +1111,25 @@ static int htb_dump_class(struct Qdisc *sch, unsigned long arg, memset(&opt, 0, sizeof(opt)); - opt.rate = cl->rate->rate; - opt.buffer = cl->buffer; - opt.ceil = cl->ceil->rate; - opt.cbuffer = cl->cbuffer; + psched_ratecfg_getrate(&opt.rate, &cl->rate); + opt.buffer = PSCHED_NS2TICKS(cl->buffer); + psched_ratecfg_getrate(&opt.ceil, &cl->ceil); + opt.cbuffer = PSCHED_NS2TICKS(cl->cbuffer); opt.quantum = cl->quantum; opt.prio = cl->prio; opt.level = cl->level; - NLA_PUT(skb, TCA_HTB_PARMS, sizeof(opt), &opt); + if (nla_put(skb, TCA_HTB_PARMS, sizeof(opt), &opt)) + goto nla_put_failure; + if ((cl->rate.rate_bytes_ps >= (1ULL << 32)) && + nla_put_u64(skb, TCA_HTB_RATE64, cl->rate.rate_bytes_ps)) + goto nla_put_failure; + if ((cl->ceil.rate_bytes_ps >= (1ULL << 32)) && + nla_put_u64(skb, TCA_HTB_CEIL64, cl->ceil.rate_bytes_ps)) + goto nla_put_failure; - nla_nest_end(skb, nest); - spin_unlock_bh(root_lock); - return skb->len; + return nla_nest_end(skb, nest); nla_put_failure: - spin_unlock_bh(root_lock); nla_nest_cancel(skb, nest); return -1; } @@ -1102,8 +1141,8 @@ htb_dump_class_stats(struct Qdisc *sch, unsigned long arg, struct gnet_dump *d) if (!cl->level && cl->un.leaf.q) cl->qstats.qlen = cl->un.leaf.q->q.qlen; - cl->xstats.tokens = cl->tokens; - cl->xstats.ctokens = cl->ctokens; + cl->xstats.tokens = PSCHED_NS2TICKS(cl->tokens); + cl->xstats.ctokens = PSCHED_NS2TICKS(cl->ctokens); if (gnet_stats_copy_basic(d, &cl->bstats) < 0 || gnet_stats_copy_rate_est(d, NULL, &cl->rate_est) < 0 || @@ -1177,7 +1216,8 @@ static void htb_parent_to_leaf(struct htb_sched *q, struct htb_class *cl, WARN_ON(cl->level || !cl->un.leaf.q || cl->prio_activity); if (parent->cmode != HTB_CAN_SEND) - htb_safe_rb_erase(&parent->pq_node, q->wait_pq + parent->level); + htb_safe_rb_erase(&parent->pq_node, + &q->hlevel[parent->level].wait_pq); parent->level = 0; memset(&parent->un.inner, 0, sizeof(parent->un.inner)); @@ -1185,7 +1225,7 @@ static void htb_parent_to_leaf(struct htb_sched *q, struct htb_class *cl, parent->un.leaf.q = new_q ? new_q : &noop_qdisc; parent->tokens = parent->buffer; parent->ctokens = parent->cbuffer; - parent->t_c = psched_get_time(); + parent->t_c = ktime_to_ns(ktime_get()); parent->cmode = HTB_CAN_SEND; } @@ -1196,9 +1236,6 @@ static void htb_destroy_class(struct Qdisc *sch, struct htb_class *cl) qdisc_destroy(cl->un.leaf.q); } gen_kill_estimator(&cl->bstats, &cl->rate_est); - qdisc_put_rtab(cl->rate); - qdisc_put_rtab(cl->ceil); - tcf_destroy_chain(&cl->filter_list); kfree(cl); } @@ -1206,24 +1243,25 @@ static void htb_destroy_class(struct Qdisc *sch, struct htb_class *cl) static void htb_destroy(struct Qdisc *sch) { struct htb_sched *q = qdisc_priv(sch); - struct hlist_node *n, *next; + struct hlist_node *next; struct htb_class *cl; unsigned int i; cancel_work_sync(&q->work); qdisc_watchdog_cancel(&q->watchdog); /* This line used to be after htb_destroy_class call below - and surprisingly it worked in 2.4. But it must precede it - because filter need its target class alive to be able to call - unbind_filter on it (without Oops). */ + * and surprisingly it worked in 2.4. But it must precede it + * because filter need its target class alive to be able to call + * unbind_filter on it (without Oops). + */ tcf_destroy_chain(&q->filter_list); for (i = 0; i < q->clhash.hashsize; i++) { - hlist_for_each_entry(cl, n, &q->clhash.hash[i], common.hnode) + hlist_for_each_entry(cl, &q->clhash.hash[i], common.hnode) tcf_destroy_chain(&cl->filter_list); } for (i = 0; i < q->clhash.hashsize; i++) { - hlist_for_each_entry_safe(cl, n, next, &q->clhash.hash[i], + hlist_for_each_entry_safe(cl, next, &q->clhash.hash[i], common.hnode) htb_destroy_class(sch, cl); } @@ -1239,9 +1277,10 @@ static int htb_delete(struct Qdisc *sch, unsigned long arg) struct Qdisc *new_q = NULL; int last_child = 0; - // TODO: why don't allow to delete subtree ? references ? does - // tc subsys quarantee us that in htb_destroy it holds no class - // refs so that we can remove children safely there ? + /* TODO: why don't allow to delete subtree ? references ? does + * tc subsys guarantee us that in htb_destroy it holds no class + * refs so that we can remove children safely there ? + */ if (cl->children || cl->filter_cnt) return -EBUSY; @@ -1268,7 +1307,8 @@ static int htb_delete(struct Qdisc *sch, unsigned long arg) htb_deactivate(q, cl); if (cl->cmode != HTB_CAN_SEND) - htb_safe_rb_erase(&cl->pq_node, q->wait_pq + cl->level); + htb_safe_rb_erase(&cl->pq_node, + &q->hlevel[cl->level].wait_pq); if (last_child) htb_parent_to_leaf(q, cl, new_q); @@ -1299,9 +1339,9 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, struct htb_sched *q = qdisc_priv(sch); struct htb_class *cl = (struct htb_class *)*arg, *parent; struct nlattr *opt = tca[TCA_OPTIONS]; - struct qdisc_rate_table *rtab = NULL, *ctab = NULL; - struct nlattr *tb[__TCA_HTB_MAX]; + struct nlattr *tb[TCA_HTB_MAX + 1]; struct tc_htb_opt *hopt; + u64 rate64, ceil64; /* extract all subattrs from opt attr */ if (!opt) @@ -1318,12 +1358,16 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, parent = parentid == TC_H_ROOT ? NULL : htb_find(parentid, sch); hopt = nla_data(tb[TCA_HTB_PARMS]); - - rtab = qdisc_get_rtab(&hopt->rate, tb[TCA_HTB_RTAB]); - ctab = qdisc_get_rtab(&hopt->ceil, tb[TCA_HTB_CTAB]); - if (!rtab || !ctab) + if (!hopt->rate.rate || !hopt->ceil.rate) goto failure; + /* Keeping backward compatible with rate_table based iproute2 tc */ + if (hopt->rate.linklayer == TC_LINKLAYER_UNAWARE) + qdisc_put_rtab(qdisc_get_rtab(&hopt->rate, tb[TCA_HTB_RTAB])); + + if (hopt->ceil.linklayer == TC_LINKLAYER_UNAWARE) + qdisc_put_rtab(qdisc_get_rtab(&hopt->ceil, tb[TCA_HTB_CTAB])); + if (!cl) { /* new class */ struct Qdisc *new_q; int prio; @@ -1349,19 +1393,22 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, /* check maximal depth */ if (parent && parent->parent && parent->parent->level < 2) { - printk(KERN_ERR "htb: tree is too deep\n"); + pr_err("htb: tree is too deep\n"); goto failure; } err = -ENOBUFS; - if ((cl = kzalloc(sizeof(*cl), GFP_KERNEL)) == NULL) + cl = kzalloc(sizeof(*cl), GFP_KERNEL); + if (!cl) goto failure; - err = gen_new_estimator(&cl->bstats, &cl->rate_est, - qdisc_root_sleeping_lock(sch), - tca[TCA_RATE] ? : &est.nla); - if (err) { - kfree(cl); - goto failure; + if (htb_rate_est || tca[TCA_RATE]) { + err = gen_new_estimator(&cl->bstats, &cl->rate_est, + qdisc_root_sleeping_lock(sch), + tca[TCA_RATE] ? : &est.nla); + if (err) { + kfree(cl); + goto failure; + } } cl->refcnt = 1; @@ -1373,8 +1420,9 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, RB_CLEAR_NODE(&cl->node[prio]); /* create leaf qdisc early because it uses kmalloc(GFP_KERNEL) - so that can't be used inside of sch_tree_lock - -- thanks to Karlis Peisenieks */ + * so that can't be used inside of sch_tree_lock + * -- thanks to Karlis Peisenieks + */ new_q = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops, classid); sch_tree_lock(sch); @@ -1390,7 +1438,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, /* remove from evt list because of level change */ if (parent->cmode != HTB_CAN_SEND) { - htb_safe_rb_erase(&parent->pq_node, q->wait_pq); + htb_safe_rb_erase(&parent->pq_node, &q->hlevel[0].wait_pq); parent->cmode = HTB_CAN_SEND; } parent->level = (parent->parent ? parent->parent->level @@ -1404,10 +1452,10 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, cl->parent = parent; /* set class to be in HTB_CAN_SEND state */ - cl->tokens = hopt->buffer; - cl->ctokens = hopt->cbuffer; - cl->mbuffer = 60 * PSCHED_TICKS_PER_SEC; /* 1min */ - cl->t_c = psched_get_time(); + cl->tokens = PSCHED_TICKS2NS(hopt->buffer); + cl->ctokens = PSCHED_TICKS2NS(hopt->cbuffer); + cl->mbuffer = 60ULL * NSEC_PER_SEC; /* 1min */ + cl->t_c = ktime_to_ns(ktime_get()); cl->cmode = HTB_CAN_SEND; /* attach to the hash list and parent's family */ @@ -1425,20 +1473,30 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, sch_tree_lock(sch); } + rate64 = tb[TCA_HTB_RATE64] ? nla_get_u64(tb[TCA_HTB_RATE64]) : 0; + + ceil64 = tb[TCA_HTB_CEIL64] ? nla_get_u64(tb[TCA_HTB_CEIL64]) : 0; + + psched_ratecfg_precompute(&cl->rate, &hopt->rate, rate64); + psched_ratecfg_precompute(&cl->ceil, &hopt->ceil, ceil64); + /* it used to be a nasty bug here, we have to check that node - is really leaf before changing cl->un.leaf ! */ + * is really leaf before changing cl->un.leaf ! + */ if (!cl->level) { - cl->quantum = rtab->rate.rate / q->rate2quantum; + u64 quantum = cl->rate.rate_bytes_ps; + + do_div(quantum, q->rate2quantum); + cl->quantum = min_t(u64, quantum, INT_MAX); + if (!hopt->quantum && cl->quantum < 1000) { - printk(KERN_WARNING - "HTB: quantum of class %X is small. Consider r2q change.\n", - cl->common.classid); + pr_warn("HTB: quantum of class %X is small. Consider r2q change.\n", + cl->common.classid); cl->quantum = 1000; } if (!hopt->quantum && cl->quantum > 200000) { - printk(KERN_WARNING - "HTB: quantum of class %X is big. Consider r2q change.\n", - cl->common.classid); + pr_warn("HTB: quantum of class %X is big. Consider r2q change.\n", + cl->common.classid); cl->quantum = 200000; } if (hopt->quantum) @@ -1447,14 +1505,9 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, cl->prio = TC_HTB_NUMPRIO - 1; } - cl->buffer = hopt->buffer; - cl->cbuffer = hopt->cbuffer; - if (cl->rate) - qdisc_put_rtab(cl->rate); - cl->rate = rtab; - if (cl->ceil) - qdisc_put_rtab(cl->ceil); - cl->ceil = ctab; + cl->buffer = PSCHED_TICKS2NS(hopt->buffer); + cl->cbuffer = PSCHED_TICKS2NS(hopt->cbuffer); + sch_tree_unlock(sch); qdisc_class_hash_grow(sch, &q->clhash); @@ -1463,10 +1516,6 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, return 0; failure: - if (rtab) - qdisc_put_rtab(rtab); - if (ctab) - qdisc_put_rtab(ctab); return err; } @@ -1485,13 +1534,13 @@ static unsigned long htb_bind_filter(struct Qdisc *sch, unsigned long parent, struct htb_class *cl = htb_find(classid, sch); /*if (cl && !cl->level) return 0; - The line above used to be there to prevent attaching filters to - leaves. But at least tc_index filter uses this just to get class - for other reasons so that we have to allow for it. - ---- - 19.6.2002 As Werner explained it is ok - bind filter is just - another way to "lock" the class - unlike "get" this lock can - be broken by class during destroy IIUC. + * The line above used to be there to prevent attaching filters to + * leaves. But at least tc_index filter uses this just to get class + * for other reasons so that we have to allow for it. + * ---- + * 19.6.2002 As Werner explained it is ok - bind filter is just + * another way to "lock" the class - unlike "get" this lock can + * be broken by class during destroy IIUC. */ if (cl) cl->filter_cnt++; @@ -1510,14 +1559,13 @@ static void htb_walk(struct Qdisc *sch, struct qdisc_walker *arg) { struct htb_sched *q = qdisc_priv(sch); struct htb_class *cl; - struct hlist_node *n; unsigned int i; if (arg->stop) return; for (i = 0; i < q->clhash.hashsize; i++) { - hlist_for_each_entry(cl, n, &q->clhash.hash[i], common.hnode) { + hlist_for_each_entry(cl, &q->clhash.hash[i], common.hnode) { if (arg->count < arg->skip) { arg->count++; continue; |