linux/net/sched/act_bpf.c
Davide Caratti 4e1810049c net/sched: act_bpf: validate the control action inside init()
the following script:

 # tc filter add dev crash0 egress matchall \
 > action bpf bytecode '1,6 0 0 4294967295' pass index 90
 # tc actions replace action bpf \
 > bytecode '1,6 0 0 4294967295' goto chain 42 index 90 cookie c1a0c1a0
 # tc action show action bpf

had the following output:

 Error: Failed to init TC action chain.
 We have an error talking to the kernel
 total acts 1

         action order 0: bpf bytecode '1,6 0 0 4294967295' default-action goto chain 42
         index 90 ref 2 bind 1
         cookie c1a0c1a0

Then, the first packet transmitted by crash0 made the kernel crash:

 RIP: 0010:tcf_action_exec+0xb8/0x100
 Code: 00 00 00 20 74 1d 83 f8 03 75 09 49 83 c4 08 4d 39 ec 75 bc 48 83 c4 10 5b 5d 41 5c 41 5d 41 5e 41 5f c3 49 8b 97 a8 00 00 00 <48> 8b 12 48 89 55 00 48 83 c4 10 5b 5d 41 5c 41 5d 41 5e 41 5f c3
 RSP: 0018:ffffb3a0803dfa90 EFLAGS: 00010246
 RAX: 000000002000002a RBX: ffff942b347ada00 RCX: 0000000000000000
 RDX: 0000000000000000 RSI: ffffb3a08034d038 RDI: ffff942b347ada00
 RBP: ffffb3a0803dfb30 R08: 0000000000000000 R09: 0000000000000000
 R10: 0000000000000000 R11: ffffb3a0803dfb0c R12: ffff942b3b682b00
 R13: ffff942b3b682b08 R14: 0000000000000001 R15: ffff942b3b682f00
 FS:  00007f6160a72740(0000) GS:ffff942b3da80000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000000000000000 CR3: 00000000795a4002 CR4: 00000000001606e0
 Call Trace:
  tcf_classify+0x58/0x120
  __dev_queue_xmit+0x40a/0x890
  ? ip_finish_output2+0x16f/0x430
  ip_finish_output2+0x16f/0x430
  ? ip_output+0x69/0xe0
  ip_output+0x69/0xe0
  ? ip_forward_options+0x1a0/0x1a0
  ip_send_skb+0x15/0x40
  raw_sendmsg+0x8e1/0xbd0
  ? sched_clock+0x5/0x10
  ? sched_clock_cpu+0xc/0xa0
  ? try_to_wake_up+0x54/0x480
  ? ldsem_down_read+0x3f/0x280
  ? _cond_resched+0x15/0x40
  ? down_read+0xe/0x30
  ? copy_termios+0x1e/0x70
  ? tty_mode_ioctl+0x1b6/0x4c0
  ? sock_sendmsg+0x36/0x40
  sock_sendmsg+0x36/0x40
  __sys_sendto+0x10e/0x140
  ? do_vfs_ioctl+0xa4/0x640
  ? handle_mm_fault+0xdc/0x210
  ? syscall_trace_enter+0x1df/0x2e0
  ? __audit_syscall_exit+0x216/0x260
  __x64_sys_sendto+0x24/0x30
  do_syscall_64+0x5b/0x180
  entry_SYSCALL_64_after_hwframe+0x44/0xa9
 RIP: 0033:0x7f615f7e3c03
 Code: 48 8b 0d 90 62 2c 00 f7 d8 64 89 01 48 83 c8 ff c3 66 0f 1f 44 00 00 83 3d 9d c3 2c 00 00 75 13 49 89 ca b8 2c 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 34 c3 48 83 ec 08 e8 4b cc 00 00 48 89 04 24
 RSP: 002b:00007ffee5d8cc28 EFLAGS: 00000246 ORIG_RAX: 000000000000002c
 RAX: ffffffffffffffda RBX: 000055a4f28f1700 RCX: 00007f615f7e3c03
 RDX: 0000000000000040 RSI: 000055a4f28f1700 RDI: 0000000000000003
 RBP: 00007ffee5d8e340 R08: 000055a4f28ee510 R09: 0000000000000010
 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000040
 R13: 000055a4f28f16c0 R14: 000055a4f28ef69c R15: 0000000000000080
 Modules linked in: act_bpf veth ip6table_filter ip6_tables iptable_filter binfmt_misc ext4 mbcache crct10dif_pclmul jbd2 crc32_pclmul snd_hda_codec_generic ghash_clmulni_intel snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_seq snd_seq_device snd_pcm aesni_intel crypto_simd cryptd glue_helper pcspkr joydev virtio_balloon snd_timer snd i2c_piix4 soundcore nfsd auth_rpcgss nfs_acl lockd grace sunrpc ip_tables xfs ata_generic pata_acpi qxl drm_kms_helper virtio_blk virtio_net virtio_console net_failover failover syscopyarea sysfillrect sysimgblt fb_sys_fops ttm drm crc32c_intel ata_piix serio_raw libata virtio_pci virtio_ring virtio floppy dm_mirror dm_region_hash dm_log dm_mod
 CR2: 0000000000000000

Validating the control action within tcf_bpf_init() proved to fix the
above issue. A TDC selftest is added to verify the correct behavior.

Fixes: db50514f9a ("net: sched: add termination action to allow goto chain")
Fixes: 97763dc0f4 ("net_sched: reject unknown tcfa_action values")
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-03-21 13:26:41 -07:00

458 lines
10 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 <net/pkt_cls.h>
#include <linux/tc_act/tc_bpf.h>
#include <net/tc_act/tc_bpf.h>
#define ACT_BPF_NAME_LEN 256
struct tcf_bpf_cfg {
struct bpf_prog *filter;
struct sock_filter *bpf_ops;
const char *bpf_name;
u16 bpf_num_ops;
bool is_ebpf;
};
static unsigned int bpf_net_id;
static struct tc_action_ops act_bpf_ops;
static int tcf_bpf_act(struct sk_buff *skb, const struct tc_action *act,
struct tcf_result *res)
{
bool at_ingress = skb_at_tc_ingress(skb);
struct tcf_bpf *prog = to_bpf(act);
struct bpf_prog *filter;
int action, filter_res;
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);
bpf_compute_data_pointers(skb);
filter_res = BPF_PROG_RUN(filter, skb);
__skb_pull(skb, skb->mac_len);
} else {
bpf_compute_data_pointers(skb);
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)
{
struct nlattr *nla;
if (prog->bpf_name &&
nla_put_string(skb, TCA_ACT_BPF_NAME, prog->bpf_name))
return -EMSGSIZE;
if (nla_put_u32(skb, TCA_ACT_BPF_ID, prog->filter->aux->id))
return -EMSGSIZE;
nla = nla_reserve(skb, TCA_ACT_BPF_TAG, sizeof(prog->filter->tag));
if (nla == NULL)
return -EMSGSIZE;
memcpy(nla_data(nla), prog->filter->tag, nla_len(nla));
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 = to_bpf(act);
struct tc_act_bpf opt = {
.index = prog->tcf_index,
.refcnt = refcount_read(&prog->tcf_refcnt) - ref,
.bindcnt = atomic_read(&prog->tcf_bindcnt) - bind,
};
struct tcf_t tm;
int ret;
spin_lock_bh(&prog->tcf_lock);
opt.action = prog->tcf_action;
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;
tcf_tm_dump(&tm, &prog->tcf_tm);
if (nla_put_64bit(skb, TCA_ACT_BPF_TM, sizeof(tm), &tm,
TCA_ACT_BPF_PAD))
goto nla_put_failure;
spin_unlock_bh(&prog->tcf_lock);
return skb->len;
nla_put_failure:
spin_unlock_bh(&prog->tcf_lock);
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 = kmemdup(nla_data(tb[TCA_ACT_BPF_OPS]), bpf_size, GFP_KERNEL);
if (bpf_ops == NULL)
return -ENOMEM;
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_type(bpf_fd, BPF_PROG_TYPE_SCHED_ACT);
if (IS_ERR(fp))
return PTR_ERR(fp);
if (tb[TCA_ACT_BPF_NAME]) {
name = nla_memdup(tb[TCA_ACT_BPF_NAME], GFP_KERNEL);
if (!name) {
bpf_prog_put(fp);
return -ENOMEM;
}
}
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)
{
struct bpf_prog *filter = cfg->filter;
if (filter) {
if (cfg->is_ebpf)
bpf_prog_put(filter);
else
bpf_prog_destroy(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 tcf 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, bool rtnl_held,
struct tcf_proto *tp, struct netlink_ext_ack *extack)
{
struct tc_action_net *tn = net_generic(net, bpf_net_id);
struct nlattr *tb[TCA_ACT_BPF_MAX + 1];
struct tcf_chain *goto_ch = NULL;
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, NULL);
if (ret < 0)
return ret;
if (!tb[TCA_ACT_BPF_PARMS])
return -EINVAL;
parm = nla_data(tb[TCA_ACT_BPF_PARMS]);
ret = tcf_idr_check_alloc(tn, &parm->index, act, bind);
if (!ret) {
ret = tcf_idr_create(tn, parm->index, est, act,
&act_bpf_ops, bind, true);
if (ret < 0) {
tcf_idr_cleanup(tn, parm->index);
return ret;
}
res = ACT_P_CREATED;
} else if (ret > 0) {
/* Don't override defaults. */
if (bind)
return 0;
if (!replace) {
tcf_idr_release(*act, bind);
return -EEXIST;
}
} else {
return ret;
}
ret = tcf_action_check_ctrlact(parm->action, tp, &goto_ch, extack);
if (ret < 0)
goto release_idr;
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 put_chain;
}
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 put_chain;
prog = to_bpf(*act);
spin_lock_bh(&prog->tcf_lock);
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;
goto_ch = tcf_action_set_ctrlact(*act, parm->action, goto_ch);
rcu_assign_pointer(prog->filter, cfg.filter);
spin_unlock_bh(&prog->tcf_lock);
if (goto_ch)
tcf_chain_put_by_act(goto_ch);
if (res == ACT_P_CREATED) {
tcf_idr_insert(tn, *act);
} else {
/* make sure the program being replaced is no longer executing */
synchronize_rcu();
tcf_bpf_cfg_cleanup(&old);
}
return res;
put_chain:
if (goto_ch)
tcf_chain_put_by_act(goto_ch);
release_idr:
tcf_idr_release(*act, bind);
return ret;
}
static void tcf_bpf_cleanup(struct tc_action *act)
{
struct tcf_bpf_cfg tmp;
tcf_bpf_prog_fill_cfg(to_bpf(act), &tmp);
tcf_bpf_cfg_cleanup(&tmp);
}
static int tcf_bpf_walker(struct net *net, struct sk_buff *skb,
struct netlink_callback *cb, int type,
const struct tc_action_ops *ops,
struct netlink_ext_ack *extack)
{
struct tc_action_net *tn = net_generic(net, bpf_net_id);
return tcf_generic_walker(tn, skb, cb, type, ops, extack);
}
static int tcf_bpf_search(struct net *net, struct tc_action **a, u32 index)
{
struct tc_action_net *tn = net_generic(net, bpf_net_id);
return tcf_idr_search(tn, a, index);
}
static struct tc_action_ops act_bpf_ops __read_mostly = {
.kind = "bpf",
.id = TCA_ID_BPF,
.owner = THIS_MODULE,
.act = tcf_bpf_act,
.dump = tcf_bpf_dump,
.cleanup = tcf_bpf_cleanup,
.init = tcf_bpf_init,
.walk = tcf_bpf_walker,
.lookup = tcf_bpf_search,
.size = sizeof(struct tcf_bpf),
};
static __net_init int bpf_init_net(struct net *net)
{
struct tc_action_net *tn = net_generic(net, bpf_net_id);
return tc_action_net_init(tn, &act_bpf_ops);
}
static void __net_exit bpf_exit_net(struct list_head *net_list)
{
tc_action_net_exit(net_list, bpf_net_id);
}
static struct pernet_operations bpf_net_ops = {
.init = bpf_init_net,
.exit_batch = bpf_exit_net,
.id = &bpf_net_id,
.size = sizeof(struct tc_action_net),
};
static int __init bpf_init_module(void)
{
return tcf_register_action(&act_bpf_ops, &bpf_net_ops);
}
static void __exit bpf_cleanup_module(void)
{
tcf_unregister_action(&act_bpf_ops, &bpf_net_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");