linux/net/sched/act_bpf.c
Alexei Starovoitov 27b29f6305 bpf: add bpf_redirect() helper
Existing bpf_clone_redirect() helper clones skb before redirecting
it to RX or TX of destination netdev.
Introduce bpf_redirect() helper that does that without cloning.

Benchmarked with two hosts using 10G ixgbe NICs.
One host is doing line rate pktgen.
Another host is configured as:
$ tc qdisc add dev $dev ingress
$ tc filter add dev $dev root pref 10 u32 match u32 0 0 flowid 1:2 \
   action bpf run object-file tcbpf1_kern.o section clone_redirect_xmit drop
so it receives the packet on $dev and immediately xmits it on $dev + 1
The section 'clone_redirect_xmit' in tcbpf1_kern.o file has the program
that does bpf_clone_redirect() and performance is 2.0 Mpps

$ tc filter add dev $dev root pref 10 u32 match u32 0 0 flowid 1:2 \
   action bpf run object-file tcbpf1_kern.o section redirect_xmit drop
which is using bpf_redirect() - 2.4 Mpps

and using cls_bpf with integrated actions as:
$ tc filter add dev $dev root pref 10 \
  bpf run object-file tcbpf1_kern.o section redirect_xmit integ_act classid 1
performance is 2.5 Mpps

To summarize:
u32+act_bpf using clone_redirect - 2.0 Mpps
u32+act_bpf using redirect - 2.4 Mpps
cls_bpf using redirect - 2.5 Mpps

For comparison linux bridge in this setup is doing 2.1 Mpps
and ixgbe rx + drop in ip_rcv - 7.8 Mpps

Signed-off-by: Alexei Starovoitov <ast@plumgrid.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-17 21:09:07 -07:00

396 lines
8.8 KiB
C

/*
* Copyright (c) 2015 Jiri Pirko <jiri@resnulli.us>
*
* 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.
*/
#include <linux/module.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/skbuff.h>
#include <linux/rtnetlink.h>
#include <linux/filter.h>
#include <linux/bpf.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <linux/tc_act/tc_bpf.h>
#include <net/tc_act/tc_bpf.h>
#define BPF_TAB_MASK 15
#define ACT_BPF_NAME_LEN 256
struct tcf_bpf_cfg {
struct bpf_prog *filter;
struct sock_filter *bpf_ops;
const char *bpf_name;
u32 bpf_fd;
u16 bpf_num_ops;
bool is_ebpf;
};
static int tcf_bpf(struct sk_buff *skb, const struct tc_action *act,
struct tcf_result *res)
{
struct tcf_bpf *prog = act->priv;
struct bpf_prog *filter;
int action, filter_res;
bool at_ingress = G_TC_AT(skb->tc_verd) & AT_INGRESS;
if (unlikely(!skb_mac_header_was_set(skb)))
return TC_ACT_UNSPEC;
tcf_lastuse_update(&prog->tcf_tm);
bstats_cpu_update(this_cpu_ptr(prog->common.cpu_bstats), skb);
rcu_read_lock();
filter = rcu_dereference(prog->filter);
if (at_ingress) {
__skb_push(skb, skb->mac_len);
filter_res = BPF_PROG_RUN(filter, skb);
__skb_pull(skb, skb->mac_len);
} else {
filter_res = BPF_PROG_RUN(filter, skb);
}
rcu_read_unlock();
/* A BPF program may overwrite the default action opcode.
* Similarly as in cls_bpf, if filter_res == -1 we use the
* default action specified from tc.
*
* In case a different well-known TC_ACT opcode has been
* returned, it will overwrite the default one.
*
* For everything else that is unkown, TC_ACT_UNSPEC is
* returned.
*/
switch (filter_res) {
case TC_ACT_PIPE:
case TC_ACT_RECLASSIFY:
case TC_ACT_OK:
case TC_ACT_REDIRECT:
action = filter_res;
break;
case TC_ACT_SHOT:
action = filter_res;
qstats_drop_inc(this_cpu_ptr(prog->common.cpu_qstats));
break;
case TC_ACT_UNSPEC:
action = prog->tcf_action;
break;
default:
action = TC_ACT_UNSPEC;
break;
}
return action;
}
static bool tcf_bpf_is_ebpf(const struct tcf_bpf *prog)
{
return !prog->bpf_ops;
}
static int tcf_bpf_dump_bpf_info(const struct tcf_bpf *prog,
struct sk_buff *skb)
{
struct nlattr *nla;
if (nla_put_u16(skb, TCA_ACT_BPF_OPS_LEN, prog->bpf_num_ops))
return -EMSGSIZE;
nla = nla_reserve(skb, TCA_ACT_BPF_OPS, prog->bpf_num_ops *
sizeof(struct sock_filter));
if (nla == NULL)
return -EMSGSIZE;
memcpy(nla_data(nla), prog->bpf_ops, nla_len(nla));
return 0;
}
static int tcf_bpf_dump_ebpf_info(const struct tcf_bpf *prog,
struct sk_buff *skb)
{
if (nla_put_u32(skb, TCA_ACT_BPF_FD, prog->bpf_fd))
return -EMSGSIZE;
if (prog->bpf_name &&
nla_put_string(skb, TCA_ACT_BPF_NAME, prog->bpf_name))
return -EMSGSIZE;
return 0;
}
static int tcf_bpf_dump(struct sk_buff *skb, struct tc_action *act,
int bind, int ref)
{
unsigned char *tp = skb_tail_pointer(skb);
struct tcf_bpf *prog = act->priv;
struct tc_act_bpf opt = {
.index = prog->tcf_index,
.refcnt = prog->tcf_refcnt - ref,
.bindcnt = prog->tcf_bindcnt - bind,
.action = prog->tcf_action,
};
struct tcf_t tm;
int ret;
if (nla_put(skb, TCA_ACT_BPF_PARMS, sizeof(opt), &opt))
goto nla_put_failure;
if (tcf_bpf_is_ebpf(prog))
ret = tcf_bpf_dump_ebpf_info(prog, skb);
else
ret = tcf_bpf_dump_bpf_info(prog, skb);
if (ret)
goto nla_put_failure;
tm.install = jiffies_to_clock_t(jiffies - prog->tcf_tm.install);
tm.lastuse = jiffies_to_clock_t(jiffies - prog->tcf_tm.lastuse);
tm.expires = jiffies_to_clock_t(prog->tcf_tm.expires);
if (nla_put(skb, TCA_ACT_BPF_TM, sizeof(tm), &tm))
goto nla_put_failure;
return skb->len;
nla_put_failure:
nlmsg_trim(skb, tp);
return -1;
}
static const struct nla_policy act_bpf_policy[TCA_ACT_BPF_MAX + 1] = {
[TCA_ACT_BPF_PARMS] = { .len = sizeof(struct tc_act_bpf) },
[TCA_ACT_BPF_FD] = { .type = NLA_U32 },
[TCA_ACT_BPF_NAME] = { .type = NLA_NUL_STRING, .len = ACT_BPF_NAME_LEN },
[TCA_ACT_BPF_OPS_LEN] = { .type = NLA_U16 },
[TCA_ACT_BPF_OPS] = { .type = NLA_BINARY,
.len = sizeof(struct sock_filter) * BPF_MAXINSNS },
};
static int tcf_bpf_init_from_ops(struct nlattr **tb, struct tcf_bpf_cfg *cfg)
{
struct sock_filter *bpf_ops;
struct sock_fprog_kern fprog_tmp;
struct bpf_prog *fp;
u16 bpf_size, bpf_num_ops;
int ret;
bpf_num_ops = nla_get_u16(tb[TCA_ACT_BPF_OPS_LEN]);
if (bpf_num_ops > BPF_MAXINSNS || bpf_num_ops == 0)
return -EINVAL;
bpf_size = bpf_num_ops * sizeof(*bpf_ops);
if (bpf_size != nla_len(tb[TCA_ACT_BPF_OPS]))
return -EINVAL;
bpf_ops = kzalloc(bpf_size, GFP_KERNEL);
if (bpf_ops == NULL)
return -ENOMEM;
memcpy(bpf_ops, nla_data(tb[TCA_ACT_BPF_OPS]), bpf_size);
fprog_tmp.len = bpf_num_ops;
fprog_tmp.filter = bpf_ops;
ret = bpf_prog_create(&fp, &fprog_tmp);
if (ret < 0) {
kfree(bpf_ops);
return ret;
}
cfg->bpf_ops = bpf_ops;
cfg->bpf_num_ops = bpf_num_ops;
cfg->filter = fp;
cfg->is_ebpf = false;
return 0;
}
static int tcf_bpf_init_from_efd(struct nlattr **tb, struct tcf_bpf_cfg *cfg)
{
struct bpf_prog *fp;
char *name = NULL;
u32 bpf_fd;
bpf_fd = nla_get_u32(tb[TCA_ACT_BPF_FD]);
fp = bpf_prog_get(bpf_fd);
if (IS_ERR(fp))
return PTR_ERR(fp);
if (fp->type != BPF_PROG_TYPE_SCHED_ACT) {
bpf_prog_put(fp);
return -EINVAL;
}
if (tb[TCA_ACT_BPF_NAME]) {
name = kmemdup(nla_data(tb[TCA_ACT_BPF_NAME]),
nla_len(tb[TCA_ACT_BPF_NAME]),
GFP_KERNEL);
if (!name) {
bpf_prog_put(fp);
return -ENOMEM;
}
}
cfg->bpf_fd = bpf_fd;
cfg->bpf_name = name;
cfg->filter = fp;
cfg->is_ebpf = true;
return 0;
}
static void tcf_bpf_cfg_cleanup(const struct tcf_bpf_cfg *cfg)
{
if (cfg->is_ebpf)
bpf_prog_put(cfg->filter);
else
bpf_prog_destroy(cfg->filter);
kfree(cfg->bpf_ops);
kfree(cfg->bpf_name);
}
static void tcf_bpf_prog_fill_cfg(const struct tcf_bpf *prog,
struct tcf_bpf_cfg *cfg)
{
cfg->is_ebpf = tcf_bpf_is_ebpf(prog);
/* updates to prog->filter are prevented, since it's called either
* with rtnl lock or during final cleanup in rcu callback
*/
cfg->filter = rcu_dereference_protected(prog->filter, 1);
cfg->bpf_ops = prog->bpf_ops;
cfg->bpf_name = prog->bpf_name;
}
static int tcf_bpf_init(struct net *net, struct nlattr *nla,
struct nlattr *est, struct tc_action *act,
int replace, int bind)
{
struct nlattr *tb[TCA_ACT_BPF_MAX + 1];
struct tcf_bpf_cfg cfg, old;
struct tc_act_bpf *parm;
struct tcf_bpf *prog;
bool is_bpf, is_ebpf;
int ret, res = 0;
if (!nla)
return -EINVAL;
ret = nla_parse_nested(tb, TCA_ACT_BPF_MAX, nla, act_bpf_policy);
if (ret < 0)
return ret;
if (!tb[TCA_ACT_BPF_PARMS])
return -EINVAL;
parm = nla_data(tb[TCA_ACT_BPF_PARMS]);
if (!tcf_hash_check(parm->index, act, bind)) {
ret = tcf_hash_create(parm->index, est, act,
sizeof(*prog), bind, true);
if (ret < 0)
return ret;
res = ACT_P_CREATED;
} else {
/* Don't override defaults. */
if (bind)
return 0;
tcf_hash_release(act, bind);
if (!replace)
return -EEXIST;
}
is_bpf = tb[TCA_ACT_BPF_OPS_LEN] && tb[TCA_ACT_BPF_OPS];
is_ebpf = tb[TCA_ACT_BPF_FD];
if ((!is_bpf && !is_ebpf) || (is_bpf && is_ebpf)) {
ret = -EINVAL;
goto out;
}
memset(&cfg, 0, sizeof(cfg));
ret = is_bpf ? tcf_bpf_init_from_ops(tb, &cfg) :
tcf_bpf_init_from_efd(tb, &cfg);
if (ret < 0)
goto out;
prog = to_bpf(act);
ASSERT_RTNL();
if (res != ACT_P_CREATED)
tcf_bpf_prog_fill_cfg(prog, &old);
prog->bpf_ops = cfg.bpf_ops;
prog->bpf_name = cfg.bpf_name;
if (cfg.bpf_num_ops)
prog->bpf_num_ops = cfg.bpf_num_ops;
if (cfg.bpf_fd)
prog->bpf_fd = cfg.bpf_fd;
prog->tcf_action = parm->action;
rcu_assign_pointer(prog->filter, cfg.filter);
if (res == ACT_P_CREATED) {
tcf_hash_insert(act);
} else {
/* make sure the program being replaced is no longer executing */
synchronize_rcu();
tcf_bpf_cfg_cleanup(&old);
}
return res;
out:
if (res == ACT_P_CREATED)
tcf_hash_cleanup(act, est);
return ret;
}
static void tcf_bpf_cleanup(struct tc_action *act, int bind)
{
struct tcf_bpf_cfg tmp;
tcf_bpf_prog_fill_cfg(act->priv, &tmp);
tcf_bpf_cfg_cleanup(&tmp);
}
static struct tc_action_ops act_bpf_ops __read_mostly = {
.kind = "bpf",
.type = TCA_ACT_BPF,
.owner = THIS_MODULE,
.act = tcf_bpf,
.dump = tcf_bpf_dump,
.cleanup = tcf_bpf_cleanup,
.init = tcf_bpf_init,
};
static int __init bpf_init_module(void)
{
return tcf_register_action(&act_bpf_ops, BPF_TAB_MASK);
}
static void __exit bpf_cleanup_module(void)
{
tcf_unregister_action(&act_bpf_ops);
}
module_init(bpf_init_module);
module_exit(bpf_cleanup_module);
MODULE_AUTHOR("Jiri Pirko <jiri@resnulli.us>");
MODULE_DESCRIPTION("TC BPF based action");
MODULE_LICENSE("GPL v2");