sch_atm.c

linux 内核源代码

/* net/sched/sch_atm.c - ATM VC selection "queueing discipline" */

/* Written 1998-2000 by Werner Almesberger, EPFL ICA */

#include <linux/module.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/atmdev.h>
#include <linux/atmclip.h>
#include <linux/rtnetlink.h>
#include <linux/file.h>		/* for fput */
#include <net/netlink.h>
#include <net/pkt_sched.h>

extern struct socket *sockfd_lookup(int fd, int *err);	/* @@@ fix this */

#if 0 /* control */
#define DPRINTK(format,args...) printk(KERN_DEBUG format,##args)
#else
#define DPRINTK(format,args...)
#endif

#if 0 /* data */
#define D2PRINTK(format,args...) printk(KERN_DEBUG format,##args)
#else
#define D2PRINTK(format,args...)
#endif

/*
 * The ATM queuing discipline provides a framework for invoking classifiers
 * (aka "filters"), which in turn select classes of this queuing discipline.
 * Each class maps the flow(s) it is handling to a given VC. Multiple classes
 * may share the same VC.
 *
 * When creating a class, VCs are specified by passing the number of the open
 * socket descriptor by which the calling process references the VC. The kernel
 * keeps the VC open at least until all classes using it are removed.
 *
 * In this file, most functions are named atm_tc_* to avoid confusion with all
 * the atm_* in net/atm. This naming convention differs from what's used in the
 * rest of net/sched.
 *
 * Known bugs:
 *  - sometimes messes up the IP stack
 *  - any manipulations besides the few operations described in the README, are
 *    untested and likely to crash the system
 *  - should lock the flow while there is data in the queue (?)
 */

#define PRIV(sch) qdisc_priv(sch)
#define VCC2FLOW(vcc) ((struct atm_flow_data *) ((vcc)->user_back))

struct atm_flow_data {
	struct Qdisc		*q;	/* FIFO, TBF, etc. */
	struct tcf_proto	*filter_list;
	struct atm_vcc		*vcc;	/* VCC; NULL if VCC is closed */
	void			(*old_pop)(struct atm_vcc *vcc,
					   struct sk_buff * skb); /* chaining */
	struct atm_qdisc_data	*parent;	/* parent qdisc */
	struct socket		*sock;		/* for closing */
	u32			classid;	/* x:y type ID */
	int			ref;		/* reference count */
	struct gnet_stats_basic	bstats;
	struct gnet_stats_queue	qstats;
	struct atm_flow_data	*next;
	struct atm_flow_data	*excess;	/* flow for excess traffic;
						   NULL to set CLP instead */
	int			hdr_len;
	unsigned char		hdr[0];		/* header data; MUST BE LAST */
};

struct atm_qdisc_data {
	struct atm_flow_data	link;		/* unclassified skbs go here */
	struct atm_flow_data	*flows;		/* NB: "link" is also on this
						   list */
	struct tasklet_struct	task;		/* requeue tasklet */
};

/* ------------------------- Class/flow operations ------------------------- */

static int find_flow(struct atm_qdisc_data *qdisc, struct atm_flow_data *flow)
{
	struct atm_flow_data *walk;

	DPRINTK("find_flow(qdisc %p,flow %p)\n", qdisc, flow);
	for (walk = qdisc->flows; walk; walk = walk->next)
		if (walk == flow)
			return 1;
	DPRINTK("find_flow: not found\n");
	return 0;
}

static inline struct atm_flow_data *lookup_flow(struct Qdisc *sch, u32 classid)
{
	struct atm_qdisc_data *p = PRIV(sch);
	struct atm_flow_data *flow;

	for (flow = p->flows; flow; flow = flow->next)
		if (flow->classid == classid)
			break;
	return flow;
}

static int atm_tc_graft(struct Qdisc *sch, unsigned long arg,
			struct Qdisc *new, struct Qdisc **old)
{
	struct atm_qdisc_data *p = PRIV(sch);
	struct atm_flow_data *flow = (struct atm_flow_data *)arg;

	DPRINTK("atm_tc_graft(sch %p,[qdisc %p],flow %p,new %p,old %p)\n",
		sch, p, flow, new, old);
	if (!find_flow(p, flow))
		return -EINVAL;
	if (!new)
		new = &noop_qdisc;
	*old = xchg(&flow->q, new);
	if (*old)
		qdisc_reset(*old);
	return 0;
}

static struct Qdisc *atm_tc_leaf(struct Qdisc *sch, unsigned long cl)
{
	struct atm_flow_data *flow = (struct atm_flow_data *)cl;

	DPRINTK("atm_tc_leaf(sch %p,flow %p)\n", sch, flow);
	return flow ? flow->q : NULL;
}

static unsigned long atm_tc_get(struct Qdisc *sch, u32 classid)
{
	struct atm_qdisc_data *p __maybe_unused = PRIV(sch);
	struct atm_flow_data *flow;

	DPRINTK("atm_tc_get(sch %p,[qdisc %p],classid %x)\n", sch, p, classid);
	flow = lookup_flow(sch, classid);
	if (flow)
		flow->ref++;
	DPRINTK("atm_tc_get: flow %p\n", flow);
	return (unsigned long)flow;
}

static unsigned long atm_tc_bind_filter(struct Qdisc *sch,
					unsigned long parent, u32 classid)
{
	return atm_tc_get(sch, classid);
}

/*
 * atm_tc_put handles all destructions, including the ones that are explicitly
 * requested (atm_tc_destroy, etc.). The assumption here is that we never drop
 * anything that still seems to be in use.
 */
static void atm_tc_put(struct Qdisc *sch, unsigned long cl)
{
	struct atm_qdisc_data *p = PRIV(sch);
	struct atm_flow_data *flow = (struct atm_flow_data *)cl;
	struct atm_flow_data **prev;

	DPRINTK("atm_tc_put(sch %p,[qdisc %p],flow %p)\n", sch, p, flow);
	if (--flow->ref)
		return;
	DPRINTK("atm_tc_put: destroying\n");
	for (prev = &p->flows; *prev; prev = &(*prev)->next)
		if (*prev == flow)
			break;
	if (!*prev) {
		printk(KERN_CRIT "atm_tc_put: class %p not found\n", flow);
		return;
	}
	*prev = flow->next;
	DPRINTK("atm_tc_put: qdisc %p\n", flow->q);
	qdisc_destroy(flow->q);
	tcf_destroy_chain(flow->filter_list);
	if (flow->sock) {
		DPRINTK("atm_tc_put: f_count %d\n",
			file_count(flow->sock->file));
		flow->vcc->pop = flow->old_pop;
		sockfd_put(flow->sock);
	}
	if (flow->excess)
		atm_tc_put(sch, (unsigned long)flow->excess);
	if (flow != &p->link)
		kfree(flow);
	/*
	 * If flow == &p->link, the qdisc no longer works at this point and
	 * needs to be removed. (By the caller of atm_tc_put.)
	 */
}

static void sch_atm_pop(struct atm_vcc *vcc, struct sk_buff *skb)
{
	struct atm_qdisc_data *p = VCC2FLOW(vcc)->parent;

	D2PRINTK("sch_atm_pop(vcc %p,skb %p,[qdisc %p])\n", vcc, skb, p);
	VCC2FLOW(vcc)->old_pop(vcc, skb);
	tasklet_schedule(&p->task);
}

static const u8 llc_oui_ip[] = {
	0xaa,			/* DSAP: non-ISO */
	0xaa,			/* SSAP: non-ISO */
	0x03,			/* Ctrl: Unnumbered Information Command PDU */
	0x00,			/* OUI: EtherType */
	0x00, 0x00,
	0x08, 0x00
};				/* Ethertype IP (0800) */

static int atm_tc_change(struct Qdisc *sch, u32 classid, u32 parent,
			 struct rtattr **tca, unsigned long *arg)
{
	struct atm_qdisc_data *p = PRIV(sch);
	struct atm_flow_data *flow = (struct atm_flow_data *)*arg;
	struct atm_flow_data *excess = NULL;
	struct rtattr *opt = tca[TCA_OPTIONS - 1];
	struct rtattr *tb[TCA_ATM_MAX];
	struct socket *sock;
	int fd, error, hdr_len;
	void *hdr;

	DPRINTK("atm_tc_change(sch %p,[qdisc %p],classid %x,parent %x,"
		"flow %p,opt %p)\n", sch, p, classid, parent, flow, opt);
	/*
	 * The concept of parents doesn't apply for this qdisc.
	 */
	if (parent && parent != TC_H_ROOT && parent != sch->handle)
		return -EINVAL;
	/*
	 * ATM classes cannot be changed. In order to change properties of the
	 * ATM connection, that socket needs to be modified directly (via the
	 * native ATM API. In order to send a flow to a different VC, the old
	 * class needs to be removed and a new one added. (This may be changed
	 * later.)
	 */
	if (flow)
		return -EBUSY;
	if (opt == NULL || rtattr_parse_nested(tb, TCA_ATM_MAX, opt))
		return -EINVAL;
	if (!tb[TCA_ATM_FD - 1] || RTA_PAYLOAD(tb[TCA_ATM_FD - 1]) < sizeof(fd))
		return -EINVAL;
	fd = *(int *)RTA_DATA(tb[TCA_ATM_FD - 1]);
	DPRINTK("atm_tc_change: fd %d\n", fd);
	if (tb[TCA_ATM_HDR - 1]) {
		hdr_len = RTA_PAYLOAD(tb[TCA_ATM_HDR - 1]);
		hdr = RTA_DATA(tb[TCA_ATM_HDR - 1]);
	} else {
		hdr_len = RFC1483LLC_LEN;
		hdr = NULL;	/* default LLC/SNAP for IP */
	}
	if (!tb[TCA_ATM_EXCESS - 1])
		excess = NULL;
	else {
		if (RTA_PAYLOAD(tb[TCA_ATM_EXCESS - 1]) != sizeof(u32))
			return -EINVAL;
		excess = (struct atm_flow_data *)
			atm_tc_get(sch, *(u32 *)RTA_DATA(tb[TCA_ATM_EXCESS - 1]));
		if (!excess)
			return -ENOENT;
	}
	DPRINTK("atm_tc_change: type %d, payload %d, hdr_len %d\n",
		opt->rta_type, RTA_PAYLOAD(opt), hdr_len);
	if (!(sock = sockfd_lookup(fd, &error)))
		return error;	/* f_count++ */
	DPRINTK("atm_tc_change: f_count %d\n", file_count(sock->file));
	if (sock->ops->family != PF_ATMSVC && sock->ops->family != PF_ATMPVC) {
		error = -EPROTOTYPE;
		goto err_out;
	}
	/* @@@ should check if the socket is really operational or we'll crash
	   on vcc->send */
	if (classid) {
		if (TC_H_MAJ(classid ^ sch->handle)) {
			DPRINTK("atm_tc_change: classid mismatch\n");
			error = -EINVAL;
			goto err_out;
		}
		if (find_flow(p, flow)) {
			error = -EEXIST;
			goto err_out;
		}
	} else {
		int i;
		unsigned long cl;

		for (i = 1; i < 0x8000; i++) {
			classid = TC_H_MAKE(sch->handle, 0x8000 | i);
			if (!(cl = atm_tc_get(sch, classid)))
				break;
			atm_tc_put(sch, cl);
		}
	}
	DPRINTK("atm_tc_change: new id %x\n", classid);
	flow = kzalloc(sizeof(struct atm_flow_data) + hdr_len, GFP_KERNEL);
	DPRINTK("atm_tc_change: flow %p\n", flow);
	if (!flow) {
		error = -ENOBUFS;
		goto err_out;
	}
	flow->filter_list = NULL;
	if (!(flow->q = qdisc_create_dflt(sch->dev, &pfifo_qdisc_ops, classid)))
		flow->q = &noop_qdisc;
	DPRINTK("atm_tc_change: qdisc %p\n", flow->q);
	flow->sock = sock;
	flow->vcc = ATM_SD(sock);	/* speedup */
	flow->vcc->user_back = flow;
	DPRINTK("atm_tc_change: vcc %p\n", flow->vcc);
	flow->old_pop = flow->vcc->pop;
	flow->parent = p;
	flow->vcc->pop = sch_atm_pop;
	flow->classid = classid;
	flow->ref = 1;
	flow->excess = excess;
	flow->next = p->link.next;
	p->link.next = flow;
	flow->hdr_len = hdr_len;
	if (hdr)
		memcpy(flow->hdr, hdr, hdr_len);
	else
		memcpy(flow->hdr, llc_oui_ip, sizeof(llc_oui_ip));
	*arg = (unsigned long)flow;
	return 0;
err_out:
	if (excess)
		atm_tc_put(sch, (unsigned long)excess);
	sockfd_put(sock);
	return error;
}

static int atm_tc_delete(struct Qdisc *sch, unsigned long arg)
{
	struct atm_qdisc_data *p = PRIV(sch);
	struct atm_flow_data *flow = (struct atm_flow_data *)arg;

	DPRINTK("atm_tc_delete(sch %p,[qdisc %p],flow %p)\n", sch, p, flow);
	if (!find_flow(PRIV(sch), flow))
		return -EINVAL;
	if (flow->filter_list || flow == &p->link)
		return -EBUSY;
	/*
	 * Reference count must be 2: one for "keepalive" (set at class
	 * creation), and one for the reference held when calling delete.
	 */
	if (flow->ref < 2) {
		printk(KERN_ERR "atm_tc_delete: flow->ref == %d\n", flow->ref);
		return -EINVAL;
	}
	if (flow->ref > 2)
		return -EBUSY;	/* catch references via excess, etc. */
	atm_tc_put(sch, arg);
	return 0;
}

static void atm_tc_walk(struct Qdisc *sch, struct qdisc_walker *walker)
{
	struct atm_qdisc_data *p = PRIV(sch);