linux/net/sched/cls_flower.c

3241 lines
92 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-or-later
/*
* net/sched/cls_flower.c Flower classifier
*
* Copyright (c) 2015 Jiri Pirko <jiri@resnulli.us>
*/
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/rhashtable.h>
net, sched: respect rcu grace period on cls destruction Roi reported a crash in flower where tp->root was NULL in ->classify() callbacks. Reason is that in ->destroy() tp->root is set to NULL via RCU_INIT_POINTER(). It's problematic for some of the classifiers, because this doesn't respect RCU grace period for them, and as a result, still outstanding readers from tc_classify() will try to blindly dereference a NULL tp->root. The tp->root object is strictly private to the classifier implementation and holds internal data the core such as tc_ctl_tfilter() doesn't know about. Within some classifiers, such as cls_bpf, cls_basic, etc, tp->root is only checked for NULL in ->get() callback, but nowhere else. This is misleading and seemed to be copied from old classifier code that was not cleaned up properly. For example, d3fa76ee6b4a ("[NET_SCHED]: cls_basic: fix NULL pointer dereference") moved tp->root initialization into ->init() routine, where before it was part of ->change(), so ->get() had to deal with tp->root being NULL back then, so that was indeed a valid case, after d3fa76ee6b4a, not really anymore. We used to set tp->root to NULL long ago in ->destroy(), see 47a1a1d4be29 ("pkt_sched: remove unnecessary xchg() in packet classifiers"); but the NULLifying was reintroduced with the RCUification, but it's not correct for every classifier implementation. In the cases that are fixed here with one exception of cls_cgroup, tp->root object is allocated and initialized inside ->init() callback, which is always performed at a point in time after we allocate a new tp, which means tp and thus tp->root was not globally visible in the tp chain yet (see tc_ctl_tfilter()). Also, on destruction tp->root is strictly kfree_rcu()'ed in ->destroy() handler, same for the tp which is kfree_rcu()'ed right when we return from ->destroy() in tcf_destroy(). This means, the head object's lifetime for such classifiers is always tied to the tp lifetime. The RCU callback invocation for the two kfree_rcu() could be out of order, but that's fine since both are independent. Dropping the RCU_INIT_POINTER(tp->root, NULL) for these classifiers here means that 1) we don't need a useless NULL check in fast-path and, 2) that outstanding readers of that tp in tc_classify() can still execute under respect with RCU grace period as it is actually expected. Things that haven't been touched here: cls_fw and cls_route. They each handle tp->root being NULL in ->classify() path for historic reasons, so their ->destroy() implementation can stay as is. If someone actually cares, they could get cleaned up at some point to avoid the test in fast path. cls_u32 doesn't set tp->root to NULL. For cls_rsvp, I just added a !head should anyone actually be using/testing it, so it at least aligns with cls_fw and cls_route. For cls_flower we additionally need to defer rhashtable destruction (to a sleepable context) after RCU grace period as concurrent readers might still access it. (Note that in this case we need to hold module reference to keep work callback address intact, since we only wait on module unload for all call_rcu()s to finish.) This fixes one race to bring RCU grace period guarantees back. Next step as worked on by Cong however is to fix 1e052be69d04 ("net_sched: destroy proto tp when all filters are gone") to get the order of unlinking the tp in tc_ctl_tfilter() for the RTM_DELTFILTER case right by moving RCU_INIT_POINTER() before tcf_destroy() and let the notification for removal be done through the prior ->delete() callback. Both are independant issues. Once we have that right, we can then clean tp->root up for a number of classifiers by not making them RCU pointers, which requires a new callback (->uninit) that is triggered from tp's RCU callback, where we just kfree() tp->root from there. Fixes: 1f947bf151e9 ("net: sched: rcu'ify cls_bpf") Fixes: 9888faefe132 ("net: sched: cls_basic use RCU") Fixes: 70da9f0bf999 ("net: sched: cls_flow use RCU") Fixes: 77b9900ef53a ("tc: introduce Flower classifier") Fixes: bf3994d2ed31 ("net/sched: introduce Match-all classifier") Fixes: 952313bd6258 ("net: sched: cls_cgroup use RCU") Reported-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: John Fastabend <john.fastabend@gmail.com> Cc: Roi Dayan <roid@mellanox.com> Cc: Jiri Pirko <jiri@mellanox.com> Acked-by: John Fastabend <john.r.fastabend@intel.com> Acked-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-27 08:18:01 +08:00
#include <linux/workqueue.h>
#include <linux/refcount.h>
#include <linux/if_ether.h>
#include <linux/in6.h>
#include <linux/ip.h>
#include <linux/mpls.h>
#include <net/sch_generic.h>
#include <net/pkt_cls.h>
#include <net/ip.h>
#include <net/flow_dissector.h>
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
#include <net/geneve.h>
#include <net/vxlan.h>
#include <net/erspan.h>
#include <net/dst.h>
#include <net/dst_metadata.h>
#include <uapi/linux/netfilter/nf_conntrack_common.h>
#define TCA_FLOWER_KEY_CT_FLAGS_MAX \
((__TCA_FLOWER_KEY_CT_FLAGS_MAX - 1) << 1)
#define TCA_FLOWER_KEY_CT_FLAGS_MASK \
(TCA_FLOWER_KEY_CT_FLAGS_MAX - 1)
struct fl_flow_key {
struct flow_dissector_key_meta meta;
struct flow_dissector_key_control control;
struct flow_dissector_key_control enc_control;
struct flow_dissector_key_basic basic;
struct flow_dissector_key_eth_addrs eth;
struct flow_dissector_key_vlan vlan;
struct flow_dissector_key_vlan cvlan;
union {
struct flow_dissector_key_ipv4_addrs ipv4;
struct flow_dissector_key_ipv6_addrs ipv6;
};
struct flow_dissector_key_ports tp;
struct flow_dissector_key_icmp icmp;
struct flow_dissector_key_arp arp;
struct flow_dissector_key_keyid enc_key_id;
union {
struct flow_dissector_key_ipv4_addrs enc_ipv4;
struct flow_dissector_key_ipv6_addrs enc_ipv6;
};
struct flow_dissector_key_ports enc_tp;
struct flow_dissector_key_mpls mpls;
struct flow_dissector_key_tcp tcp;
struct flow_dissector_key_ip ip;
struct flow_dissector_key_ip enc_ip;
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
struct flow_dissector_key_enc_opts enc_opts;
cls_flower: Fix the behavior using port ranges with hw-offload The recent commit 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") had added filtering based on port ranges to tc flower. However the commit missed necessary changes in hw-offload code, so the feature gave rise to generating incorrect offloaded flow keys in NIC. One more detailed example is below: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 100-200 action drop With the setup above, an exact match filter with dst_port == 0 will be installed in NIC by hw-offload. IOW, the NIC will have a rule which is equivalent to the following one. $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 0 action drop The behavior was caused by the flow dissector which extracts packet data into the flow key in the tc flower. More specifically, regardless of exact match or specified port ranges, fl_init_dissector() set the FLOW_DISSECTOR_KEY_PORTS flag in struct flow_dissector to extract port numbers from skb in skb_flow_dissect() called by fl_classify(). Note that device drivers received the same struct flow_dissector object as used in skb_flow_dissect(). Thus, offloaded drivers could not identify which of these is used because the FLOW_DISSECTOR_KEY_PORTS flag was set to struct flow_dissector in either case. This patch adds the new FLOW_DISSECTOR_KEY_PORTS_RANGE flag and the new tp_range field in struct fl_flow_key to recognize which filters are applied to offloaded drivers. At this point, when filters based on port ranges passed to drivers, drivers return the EOPNOTSUPP error because they do not support the feature (the newly created FLOW_DISSECTOR_KEY_PORTS_RANGE flag). Fixes: 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-03 18:40:12 +08:00
union {
struct flow_dissector_key_ports tp;
struct {
struct flow_dissector_key_ports tp_min;
struct flow_dissector_key_ports tp_max;
};
} tp_range;
struct flow_dissector_key_ct ct;
struct flow_dissector_key_hash hash;
} __aligned(BITS_PER_LONG / 8); /* Ensure that we can do comparisons as longs. */
struct fl_flow_mask_range {
unsigned short int start;
unsigned short int end;
};
struct fl_flow_mask {
struct fl_flow_key key;
struct fl_flow_mask_range range;
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
u32 flags;
struct rhash_head ht_node;
struct rhashtable ht;
struct rhashtable_params filter_ht_params;
struct flow_dissector dissector;
struct list_head filters;
cls_flower: fix use after free in flower S/W path If flower filter is created without the skip_sw flag, fl_mask_put() can race with fl_classify() and we can destroy the mask rhashtable while a lookup operation is accessing it. BUG: unable to handle kernel paging request at 00000000000911d1 PGD 0 P4D 0 SMP PTI CPU: 3 PID: 5582 Comm: vhost-5541 Not tainted 4.18.0-rc1.vanilla+ #1950 Hardware name: Dell Inc. PowerEdge R730/072T6D, BIOS 2.1.7 06/16/2016 RIP: 0010:rht_bucket_nested+0x20/0x60 Code: 31 c8 c1 c1 18 29 c8 c3 66 90 8b 4f 04 ba 01 00 00 00 8b 07 48 8b bf 80 00 00 0 RSP: 0018:ffffafc5cfbb7a48 EFLAGS: 00010206 RAX: 0000000000001978 RBX: ffff9f12dff88a00 RCX: 00000000ffff9f12 RDX: 00000000000911d1 RSI: 0000000000000148 RDI: 0000000000000001 RBP: ffff9f12dff88a00 R08: 000000005f1cc119 R09: 00000000a715fae2 R10: ffffafc5cfbb7aa8 R11: ffff9f1cb4be804e R12: ffff9f1265e13000 R13: 0000000000000000 R14: ffffafc5cfbb7b48 R15: ffff9f12dff88b68 FS: 0000000000000000(0000) GS:ffff9f1d3f0c0000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000911d1 CR3: 0000001575a94006 CR4: 00000000001626e0 Call Trace: fl_lookup+0x134/0x140 [cls_flower] fl_classify+0xf3/0x180 [cls_flower] tcf_classify+0x78/0x150 __netif_receive_skb_core+0x69e/0xa50 netif_receive_skb_internal+0x42/0xf0 tun_get_user+0xdd5/0xfd0 [tun] tun_sendmsg+0x52/0x70 [tun] handle_tx+0x2b3/0x5f0 [vhost_net] vhost_worker+0xab/0x100 [vhost] kthread+0xf8/0x130 ret_from_fork+0x35/0x40 Modules linked in: act_mirred act_gact cls_flower vhost_net vhost tap sch_ingress CR2: 00000000000911d1 Fix the above waiting for a RCU grace period before destroying the rhashtable: we need to use tcf_queue_work(), as rhashtable_destroy() must run in process context, as pointed out by Cong Wang. v1 -> v2: use tcf_queue_work to run rhashtable_destroy(). Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-22 02:02:16 +08:00
struct rcu_work rwork;
struct list_head list;
refcount_t refcnt;
};
struct fl_flow_tmplt {
struct fl_flow_key dummy_key;
struct fl_flow_key mask;
struct flow_dissector dissector;
struct tcf_chain *chain;
};
struct cls_fl_head {
struct rhashtable ht;
spinlock_t masks_lock; /* Protect masks list */
struct list_head masks;
2019-04-24 14:53:31 +08:00
struct list_head hw_filters;
struct rcu_work rwork;
struct idr handle_idr;
};
struct cls_fl_filter {
struct fl_flow_mask *mask;
struct rhash_head ht_node;
struct fl_flow_key mkey;
struct tcf_exts exts;
struct tcf_result res;
struct fl_flow_key key;
struct list_head list;
2019-04-24 14:53:31 +08:00
struct list_head hw_list;
u32 handle;
u32 flags;
u32 in_hw_count;
struct rcu_work rwork;
struct net_device *hw_dev;
/* Flower classifier is unlocked, which means that its reference counter
* can be changed concurrently without any kind of external
* synchronization. Use atomic reference counter to be concurrency-safe.
*/
refcount_t refcnt;
bool deleted;
};
static const struct rhashtable_params mask_ht_params = {
.key_offset = offsetof(struct fl_flow_mask, key),
.key_len = sizeof(struct fl_flow_key),
.head_offset = offsetof(struct fl_flow_mask, ht_node),
.automatic_shrinking = true,
};
static unsigned short int fl_mask_range(const struct fl_flow_mask *mask)
{
return mask->range.end - mask->range.start;
}
static void fl_mask_update_range(struct fl_flow_mask *mask)
{
const u8 *bytes = (const u8 *) &mask->key;
size_t size = sizeof(mask->key);
size_t i, first = 0, last;
for (i = 0; i < size; i++) {
if (bytes[i]) {
first = i;
break;
}
}
last = first;
for (i = size - 1; i != first; i--) {
if (bytes[i]) {
last = i;
break;
}
}
mask->range.start = rounddown(first, sizeof(long));
mask->range.end = roundup(last + 1, sizeof(long));
}
static void *fl_key_get_start(struct fl_flow_key *key,
const struct fl_flow_mask *mask)
{
return (u8 *) key + mask->range.start;
}
static void fl_set_masked_key(struct fl_flow_key *mkey, struct fl_flow_key *key,
struct fl_flow_mask *mask)
{
const long *lkey = fl_key_get_start(key, mask);
const long *lmask = fl_key_get_start(&mask->key, mask);
long *lmkey = fl_key_get_start(mkey, mask);
int i;
for (i = 0; i < fl_mask_range(mask); i += sizeof(long))
*lmkey++ = *lkey++ & *lmask++;
}
static bool fl_mask_fits_tmplt(struct fl_flow_tmplt *tmplt,
struct fl_flow_mask *mask)
{
const long *lmask = fl_key_get_start(&mask->key, mask);
const long *ltmplt;
int i;
if (!tmplt)
return true;
ltmplt = fl_key_get_start(&tmplt->mask, mask);
for (i = 0; i < fl_mask_range(mask); i += sizeof(long)) {
if (~*ltmplt++ & *lmask++)
return false;
}
return true;
}
static void fl_clear_masked_range(struct fl_flow_key *key,
struct fl_flow_mask *mask)
{
memset(fl_key_get_start(key, mask), 0, fl_mask_range(mask));
}
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
static bool fl_range_port_dst_cmp(struct cls_fl_filter *filter,
struct fl_flow_key *key,
struct fl_flow_key *mkey)
{
u16 min_mask, max_mask, min_val, max_val;
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
min_mask = ntohs(filter->mask->key.tp_range.tp_min.dst);
max_mask = ntohs(filter->mask->key.tp_range.tp_max.dst);
min_val = ntohs(filter->key.tp_range.tp_min.dst);
max_val = ntohs(filter->key.tp_range.tp_max.dst);
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
if (min_mask && max_mask) {
if (ntohs(key->tp_range.tp.dst) < min_val ||
ntohs(key->tp_range.tp.dst) > max_val)
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
return false;
/* skb does not have min and max values */
cls_flower: Fix the behavior using port ranges with hw-offload The recent commit 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") had added filtering based on port ranges to tc flower. However the commit missed necessary changes in hw-offload code, so the feature gave rise to generating incorrect offloaded flow keys in NIC. One more detailed example is below: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 100-200 action drop With the setup above, an exact match filter with dst_port == 0 will be installed in NIC by hw-offload. IOW, the NIC will have a rule which is equivalent to the following one. $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 0 action drop The behavior was caused by the flow dissector which extracts packet data into the flow key in the tc flower. More specifically, regardless of exact match or specified port ranges, fl_init_dissector() set the FLOW_DISSECTOR_KEY_PORTS flag in struct flow_dissector to extract port numbers from skb in skb_flow_dissect() called by fl_classify(). Note that device drivers received the same struct flow_dissector object as used in skb_flow_dissect(). Thus, offloaded drivers could not identify which of these is used because the FLOW_DISSECTOR_KEY_PORTS flag was set to struct flow_dissector in either case. This patch adds the new FLOW_DISSECTOR_KEY_PORTS_RANGE flag and the new tp_range field in struct fl_flow_key to recognize which filters are applied to offloaded drivers. At this point, when filters based on port ranges passed to drivers, drivers return the EOPNOTSUPP error because they do not support the feature (the newly created FLOW_DISSECTOR_KEY_PORTS_RANGE flag). Fixes: 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-03 18:40:12 +08:00
mkey->tp_range.tp_min.dst = filter->mkey.tp_range.tp_min.dst;
mkey->tp_range.tp_max.dst = filter->mkey.tp_range.tp_max.dst;
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
}
return true;
}
static bool fl_range_port_src_cmp(struct cls_fl_filter *filter,
struct fl_flow_key *key,
struct fl_flow_key *mkey)
{
u16 min_mask, max_mask, min_val, max_val;
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
min_mask = ntohs(filter->mask->key.tp_range.tp_min.src);
max_mask = ntohs(filter->mask->key.tp_range.tp_max.src);
min_val = ntohs(filter->key.tp_range.tp_min.src);
max_val = ntohs(filter->key.tp_range.tp_max.src);
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
if (min_mask && max_mask) {
if (ntohs(key->tp_range.tp.src) < min_val ||
ntohs(key->tp_range.tp.src) > max_val)
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
return false;
/* skb does not have min and max values */
cls_flower: Fix the behavior using port ranges with hw-offload The recent commit 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") had added filtering based on port ranges to tc flower. However the commit missed necessary changes in hw-offload code, so the feature gave rise to generating incorrect offloaded flow keys in NIC. One more detailed example is below: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 100-200 action drop With the setup above, an exact match filter with dst_port == 0 will be installed in NIC by hw-offload. IOW, the NIC will have a rule which is equivalent to the following one. $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 0 action drop The behavior was caused by the flow dissector which extracts packet data into the flow key in the tc flower. More specifically, regardless of exact match or specified port ranges, fl_init_dissector() set the FLOW_DISSECTOR_KEY_PORTS flag in struct flow_dissector to extract port numbers from skb in skb_flow_dissect() called by fl_classify(). Note that device drivers received the same struct flow_dissector object as used in skb_flow_dissect(). Thus, offloaded drivers could not identify which of these is used because the FLOW_DISSECTOR_KEY_PORTS flag was set to struct flow_dissector in either case. This patch adds the new FLOW_DISSECTOR_KEY_PORTS_RANGE flag and the new tp_range field in struct fl_flow_key to recognize which filters are applied to offloaded drivers. At this point, when filters based on port ranges passed to drivers, drivers return the EOPNOTSUPP error because they do not support the feature (the newly created FLOW_DISSECTOR_KEY_PORTS_RANGE flag). Fixes: 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-03 18:40:12 +08:00
mkey->tp_range.tp_min.src = filter->mkey.tp_range.tp_min.src;
mkey->tp_range.tp_max.src = filter->mkey.tp_range.tp_max.src;
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
}
return true;
}
static struct cls_fl_filter *__fl_lookup(struct fl_flow_mask *mask,
struct fl_flow_key *mkey)
{
return rhashtable_lookup_fast(&mask->ht, fl_key_get_start(mkey, mask),
mask->filter_ht_params);
}
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
static struct cls_fl_filter *fl_lookup_range(struct fl_flow_mask *mask,
struct fl_flow_key *mkey,
struct fl_flow_key *key)
{
struct cls_fl_filter *filter, *f;
list_for_each_entry_rcu(filter, &mask->filters, list) {
if (!fl_range_port_dst_cmp(filter, key, mkey))
continue;
if (!fl_range_port_src_cmp(filter, key, mkey))
continue;
f = __fl_lookup(mask, mkey);
if (f)
return f;
}
return NULL;
}
static noinline_for_stack
struct cls_fl_filter *fl_mask_lookup(struct fl_flow_mask *mask, struct fl_flow_key *key)
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
{
struct fl_flow_key mkey;
fl_set_masked_key(&mkey, key, mask);
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
if ((mask->flags & TCA_FLOWER_MASK_FLAGS_RANGE))
return fl_lookup_range(mask, &mkey, key);
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
return __fl_lookup(mask, &mkey);
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
}
static u16 fl_ct_info_to_flower_map[] = {
[IP_CT_ESTABLISHED] = TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED,
[IP_CT_RELATED] = TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
TCA_FLOWER_KEY_CT_FLAGS_RELATED,
[IP_CT_ESTABLISHED_REPLY] = TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED |
TCA_FLOWER_KEY_CT_FLAGS_REPLY,
[IP_CT_RELATED_REPLY] = TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
TCA_FLOWER_KEY_CT_FLAGS_RELATED |
TCA_FLOWER_KEY_CT_FLAGS_REPLY,
[IP_CT_NEW] = TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
TCA_FLOWER_KEY_CT_FLAGS_NEW,
};
static int fl_classify(struct sk_buff *skb, const struct tcf_proto *tp,
struct tcf_result *res)
{
struct cls_fl_head *head = rcu_dereference_bh(tp->root);
bool post_ct = qdisc_skb_cb(skb)->post_ct;
struct fl_flow_key skb_key;
struct fl_flow_mask *mask;
struct cls_fl_filter *f;
list_for_each_entry_rcu(mask, &head->masks, list) {
flow_dissector_init_keys(&skb_key.control, &skb_key.basic);
fl_clear_masked_range(&skb_key, mask);
skb_flow_dissect_meta(skb, &mask->dissector, &skb_key);
/* skb_flow_dissect() does not set n_proto in case an unknown
* protocol, so do it rather here.
*/
sched: consistently handle layer3 header accesses in the presence of VLANs There are a couple of places in net/sched/ that check skb->protocol and act on the value there. However, in the presence of VLAN tags, the value stored in skb->protocol can be inconsistent based on whether VLAN acceleration is enabled. The commit quoted in the Fixes tag below fixed the users of skb->protocol to use a helper that will always see the VLAN ethertype. However, most of the callers don't actually handle the VLAN ethertype, but expect to find the IP header type in the protocol field. This means that things like changing the ECN field, or parsing diffserv values, stops working if there's a VLAN tag, or if there are multiple nested VLAN tags (QinQ). To fix this, change the helper to take an argument that indicates whether the caller wants to skip the VLAN tags or not. When skipping VLAN tags, we make sure to skip all of them, so behaviour is consistent even in QinQ mode. To make the helper usable from the ECN code, move it to if_vlan.h instead of pkt_sched.h. v3: - Remove empty lines - Move vlan variable definitions inside loop in skb_protocol() - Also use skb_protocol() helper in IP{,6}_ECN_decapsulate() and bpf_skb_ecn_set_ce() v2: - Use eth_type_vlan() helper in skb_protocol() - Also fix code that reads skb->protocol directly - Change a couple of 'if/else if' statements to switch constructs to avoid calling the helper twice Reported-by: Ilya Ponetayev <i.ponetaev@ndmsystems.com> Fixes: d8b9605d2697 ("net: sched: fix skb->protocol use in case of accelerated vlan path") Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-07-04 04:26:43 +08:00
skb_key.basic.n_proto = skb_protocol(skb, false);
skb_flow_dissect_tunnel_info(skb, &mask->dissector, &skb_key);
skb_flow_dissect_ct(skb, &mask->dissector, &skb_key,
fl_ct_info_to_flower_map,
ARRAY_SIZE(fl_ct_info_to_flower_map),
post_ct);
skb_flow_dissect_hash(skb, &mask->dissector, &skb_key);
skb_flow_dissect(skb, &mask->dissector, &skb_key, 0);
f = fl_mask_lookup(mask, &skb_key);
if (f && !tc_skip_sw(f->flags)) {
*res = f->res;
return tcf_exts_exec(skb, &f->exts, res);
}
}
return -1;
}
static int fl_init(struct tcf_proto *tp)
{
struct cls_fl_head *head;
head = kzalloc(sizeof(*head), GFP_KERNEL);
if (!head)
return -ENOBUFS;
spin_lock_init(&head->masks_lock);
INIT_LIST_HEAD_RCU(&head->masks);
2019-04-24 14:53:31 +08:00
INIT_LIST_HEAD(&head->hw_filters);
rcu_assign_pointer(tp->root, head);
idr_init(&head->handle_idr);
return rhashtable_init(&head->ht, &mask_ht_params);
}
static void fl_mask_free(struct fl_flow_mask *mask, bool mask_init_done)
cls_flower: fix use after free in flower S/W path If flower filter is created without the skip_sw flag, fl_mask_put() can race with fl_classify() and we can destroy the mask rhashtable while a lookup operation is accessing it. BUG: unable to handle kernel paging request at 00000000000911d1 PGD 0 P4D 0 SMP PTI CPU: 3 PID: 5582 Comm: vhost-5541 Not tainted 4.18.0-rc1.vanilla+ #1950 Hardware name: Dell Inc. PowerEdge R730/072T6D, BIOS 2.1.7 06/16/2016 RIP: 0010:rht_bucket_nested+0x20/0x60 Code: 31 c8 c1 c1 18 29 c8 c3 66 90 8b 4f 04 ba 01 00 00 00 8b 07 48 8b bf 80 00 00 0 RSP: 0018:ffffafc5cfbb7a48 EFLAGS: 00010206 RAX: 0000000000001978 RBX: ffff9f12dff88a00 RCX: 00000000ffff9f12 RDX: 00000000000911d1 RSI: 0000000000000148 RDI: 0000000000000001 RBP: ffff9f12dff88a00 R08: 000000005f1cc119 R09: 00000000a715fae2 R10: ffffafc5cfbb7aa8 R11: ffff9f1cb4be804e R12: ffff9f1265e13000 R13: 0000000000000000 R14: ffffafc5cfbb7b48 R15: ffff9f12dff88b68 FS: 0000000000000000(0000) GS:ffff9f1d3f0c0000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000911d1 CR3: 0000001575a94006 CR4: 00000000001626e0 Call Trace: fl_lookup+0x134/0x140 [cls_flower] fl_classify+0xf3/0x180 [cls_flower] tcf_classify+0x78/0x150 __netif_receive_skb_core+0x69e/0xa50 netif_receive_skb_internal+0x42/0xf0 tun_get_user+0xdd5/0xfd0 [tun] tun_sendmsg+0x52/0x70 [tun] handle_tx+0x2b3/0x5f0 [vhost_net] vhost_worker+0xab/0x100 [vhost] kthread+0xf8/0x130 ret_from_fork+0x35/0x40 Modules linked in: act_mirred act_gact cls_flower vhost_net vhost tap sch_ingress CR2: 00000000000911d1 Fix the above waiting for a RCU grace period before destroying the rhashtable: we need to use tcf_queue_work(), as rhashtable_destroy() must run in process context, as pointed out by Cong Wang. v1 -> v2: use tcf_queue_work to run rhashtable_destroy(). Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-22 02:02:16 +08:00
{
/* temporary masks don't have their filters list and ht initialized */
if (mask_init_done) {
WARN_ON(!list_empty(&mask->filters));
rhashtable_destroy(&mask->ht);
}
cls_flower: fix use after free in flower S/W path If flower filter is created without the skip_sw flag, fl_mask_put() can race with fl_classify() and we can destroy the mask rhashtable while a lookup operation is accessing it. BUG: unable to handle kernel paging request at 00000000000911d1 PGD 0 P4D 0 SMP PTI CPU: 3 PID: 5582 Comm: vhost-5541 Not tainted 4.18.0-rc1.vanilla+ #1950 Hardware name: Dell Inc. PowerEdge R730/072T6D, BIOS 2.1.7 06/16/2016 RIP: 0010:rht_bucket_nested+0x20/0x60 Code: 31 c8 c1 c1 18 29 c8 c3 66 90 8b 4f 04 ba 01 00 00 00 8b 07 48 8b bf 80 00 00 0 RSP: 0018:ffffafc5cfbb7a48 EFLAGS: 00010206 RAX: 0000000000001978 RBX: ffff9f12dff88a00 RCX: 00000000ffff9f12 RDX: 00000000000911d1 RSI: 0000000000000148 RDI: 0000000000000001 RBP: ffff9f12dff88a00 R08: 000000005f1cc119 R09: 00000000a715fae2 R10: ffffafc5cfbb7aa8 R11: ffff9f1cb4be804e R12: ffff9f1265e13000 R13: 0000000000000000 R14: ffffafc5cfbb7b48 R15: ffff9f12dff88b68 FS: 0000000000000000(0000) GS:ffff9f1d3f0c0000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000911d1 CR3: 0000001575a94006 CR4: 00000000001626e0 Call Trace: fl_lookup+0x134/0x140 [cls_flower] fl_classify+0xf3/0x180 [cls_flower] tcf_classify+0x78/0x150 __netif_receive_skb_core+0x69e/0xa50 netif_receive_skb_internal+0x42/0xf0 tun_get_user+0xdd5/0xfd0 [tun] tun_sendmsg+0x52/0x70 [tun] handle_tx+0x2b3/0x5f0 [vhost_net] vhost_worker+0xab/0x100 [vhost] kthread+0xf8/0x130 ret_from_fork+0x35/0x40 Modules linked in: act_mirred act_gact cls_flower vhost_net vhost tap sch_ingress CR2: 00000000000911d1 Fix the above waiting for a RCU grace period before destroying the rhashtable: we need to use tcf_queue_work(), as rhashtable_destroy() must run in process context, as pointed out by Cong Wang. v1 -> v2: use tcf_queue_work to run rhashtable_destroy(). Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-22 02:02:16 +08:00
kfree(mask);
}
static void fl_mask_free_work(struct work_struct *work)
{
struct fl_flow_mask *mask = container_of(to_rcu_work(work),
struct fl_flow_mask, rwork);
fl_mask_free(mask, true);
}
static void fl_uninit_mask_free_work(struct work_struct *work)
{
struct fl_flow_mask *mask = container_of(to_rcu_work(work),
struct fl_flow_mask, rwork);
fl_mask_free(mask, false);
cls_flower: fix use after free in flower S/W path If flower filter is created without the skip_sw flag, fl_mask_put() can race with fl_classify() and we can destroy the mask rhashtable while a lookup operation is accessing it. BUG: unable to handle kernel paging request at 00000000000911d1 PGD 0 P4D 0 SMP PTI CPU: 3 PID: 5582 Comm: vhost-5541 Not tainted 4.18.0-rc1.vanilla+ #1950 Hardware name: Dell Inc. PowerEdge R730/072T6D, BIOS 2.1.7 06/16/2016 RIP: 0010:rht_bucket_nested+0x20/0x60 Code: 31 c8 c1 c1 18 29 c8 c3 66 90 8b 4f 04 ba 01 00 00 00 8b 07 48 8b bf 80 00 00 0 RSP: 0018:ffffafc5cfbb7a48 EFLAGS: 00010206 RAX: 0000000000001978 RBX: ffff9f12dff88a00 RCX: 00000000ffff9f12 RDX: 00000000000911d1 RSI: 0000000000000148 RDI: 0000000000000001 RBP: ffff9f12dff88a00 R08: 000000005f1cc119 R09: 00000000a715fae2 R10: ffffafc5cfbb7aa8 R11: ffff9f1cb4be804e R12: ffff9f1265e13000 R13: 0000000000000000 R14: ffffafc5cfbb7b48 R15: ffff9f12dff88b68 FS: 0000000000000000(0000) GS:ffff9f1d3f0c0000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000911d1 CR3: 0000001575a94006 CR4: 00000000001626e0 Call Trace: fl_lookup+0x134/0x140 [cls_flower] fl_classify+0xf3/0x180 [cls_flower] tcf_classify+0x78/0x150 __netif_receive_skb_core+0x69e/0xa50 netif_receive_skb_internal+0x42/0xf0 tun_get_user+0xdd5/0xfd0 [tun] tun_sendmsg+0x52/0x70 [tun] handle_tx+0x2b3/0x5f0 [vhost_net] vhost_worker+0xab/0x100 [vhost] kthread+0xf8/0x130 ret_from_fork+0x35/0x40 Modules linked in: act_mirred act_gact cls_flower vhost_net vhost tap sch_ingress CR2: 00000000000911d1 Fix the above waiting for a RCU grace period before destroying the rhashtable: we need to use tcf_queue_work(), as rhashtable_destroy() must run in process context, as pointed out by Cong Wang. v1 -> v2: use tcf_queue_work to run rhashtable_destroy(). Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-22 02:02:16 +08:00
}
net: sched: flower: fix filter net reference counting Fix net reference counting in fl_change() and remove redundant call to tcf_exts_get_net() from __fl_delete(). __fl_put() already tries to get net before releasing exts and deallocating a filter, so this code caused flower classifier to obtain net twice per filter that is being deleted. Implementation of __fl_delete() called tcf_exts_get_net() to pass its result as 'async' flag to fl_mask_put(). However, 'async' flag is redundant and only complicates fl_mask_put() implementation. This functionality seems to be copied from filter cleanup code, where it was added by Cong with following explanation: This patchset tries to fix the race between call_rcu() and cleanup_net() again. Without holding the netns refcnt the tc_action_net_exit() in netns workqueue could be called before filter destroy works in tc filter workqueue. This patchset moves the netns refcnt from tc actions to tcf_exts, without breaking per-netns tc actions. This doesn't apply to flower mask, which doesn't call any tc action code during cleanup. Simplify fl_mask_put() by removing the flag parameter and always use tcf_queue_work() to free mask objects. Fixes: 061775583e35 ("net: sched: flower: introduce reference counting for filters") Fixes: 1f17f7742eeb ("net: sched: flower: insert filter to ht before offloading it to hw") Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Reported-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-12 05:54:19 +08:00
static bool fl_mask_put(struct cls_fl_head *head, struct fl_flow_mask *mask)
{
if (!refcount_dec_and_test(&mask->refcnt))
return false;
rhashtable_remove_fast(&head->ht, &mask->ht_node, mask_ht_params);
spin_lock(&head->masks_lock);
list_del_rcu(&mask->list);
spin_unlock(&head->masks_lock);
net: sched: flower: fix filter net reference counting Fix net reference counting in fl_change() and remove redundant call to tcf_exts_get_net() from __fl_delete(). __fl_put() already tries to get net before releasing exts and deallocating a filter, so this code caused flower classifier to obtain net twice per filter that is being deleted. Implementation of __fl_delete() called tcf_exts_get_net() to pass its result as 'async' flag to fl_mask_put(). However, 'async' flag is redundant and only complicates fl_mask_put() implementation. This functionality seems to be copied from filter cleanup code, where it was added by Cong with following explanation: This patchset tries to fix the race between call_rcu() and cleanup_net() again. Without holding the netns refcnt the tc_action_net_exit() in netns workqueue could be called before filter destroy works in tc filter workqueue. This patchset moves the netns refcnt from tc actions to tcf_exts, without breaking per-netns tc actions. This doesn't apply to flower mask, which doesn't call any tc action code during cleanup. Simplify fl_mask_put() by removing the flag parameter and always use tcf_queue_work() to free mask objects. Fixes: 061775583e35 ("net: sched: flower: introduce reference counting for filters") Fixes: 1f17f7742eeb ("net: sched: flower: insert filter to ht before offloading it to hw") Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Reported-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-12 05:54:19 +08:00
tcf_queue_work(&mask->rwork, fl_mask_free_work);
return true;
}
2019-04-24 14:53:31 +08:00
static struct cls_fl_head *fl_head_dereference(struct tcf_proto *tp)
{
/* Flower classifier only changes root pointer during init and destroy.
* Users must obtain reference to tcf_proto instance before calling its
* API, so tp->root pointer is protected from concurrent call to
* fl_destroy() by reference counting.
*/
return rcu_dereference_raw(tp->root);
}
static void __fl_destroy_filter(struct cls_fl_filter *f)
{
tcf_exts_destroy(&f->exts);
tcf_exts_put_net(&f->exts);
kfree(f);
}
static void fl_destroy_filter_work(struct work_struct *work)
{
struct cls_fl_filter *f = container_of(to_rcu_work(work),
struct cls_fl_filter, rwork);
__fl_destroy_filter(f);
}
static void fl_hw_destroy_filter(struct tcf_proto *tp, struct cls_fl_filter *f,
bool rtnl_held, struct netlink_ext_ack *extack)
{
struct tcf_block *block = tp->chain->block;
struct flow_cls_offload cls_flower = {};
tc_cls_common_offload_init(&cls_flower.common, tp, f->flags, extack);
cls_flower.command = FLOW_CLS_DESTROY;
cls_flower.cookie = (unsigned long) f;
net: sched: refactor block offloads counter usage Without rtnl lock protection filters can no longer safely manage block offloads counter themselves. Refactor cls API to protect block offloadcnt with tcf_block->cb_lock that is already used to protect driver callback list and nooffloaddevcnt counter. The counter can be modified by concurrent tasks by new functions that execute block callbacks (which is safe with previous patch that changed its type to atomic_t), however, block bind/unbind code that checks the counter value takes cb_lock in write mode to exclude any concurrent modifications. This approach prevents race conditions between bind/unbind and callback execution code but allows for concurrency for tc rule update path. Move block offload counter, filter in hardware counter and filter flags management from classifiers into cls hardware offloads API. Make functions tcf_block_offload_{inc|dec}() and tc_cls_offload_cnt_update() to be cls API private. Implement following new cls API to be used instead: tc_setup_cb_add() - non-destructive filter add. If filter that wasn't already in hardware is successfully offloaded, increment block offloads counter, set filter in hardware counter and flag. On failure, previously offloaded filter is considered to be intact and offloads counter is not decremented. tc_setup_cb_replace() - destructive filter replace. Release existing filter block offload counter and reset its in hardware counter and flag. Set new filter in hardware counter and flag. On failure, previously offloaded filter is considered to be destroyed and offload counter is decremented. tc_setup_cb_destroy() - filter destroy. Unconditionally decrement block offloads counter. tc_setup_cb_reoffload() - reoffload filter to single cb. Execute cb() and call tc_cls_offload_cnt_update() if cb() didn't return an error. Refactor all offload-capable classifiers to atomically offload filters to hardware, change block offload counter, and set filter in hardware counter and flag by means of the new cls API functions. Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-26 21:44:59 +08:00
tc_setup_cb_destroy(block, tp, TC_SETUP_CLSFLOWER, &cls_flower, false,
&f->flags, &f->in_hw_count, rtnl_held);
}
static int fl_hw_replace_filter(struct tcf_proto *tp,
struct cls_fl_filter *f, bool rtnl_held,
struct netlink_ext_ack *extack)
{
struct tcf_block *block = tp->chain->block;
struct flow_cls_offload cls_flower = {};
bool skip_sw = tc_skip_sw(f->flags);
int err = 0;
cls_flower.rule = flow_rule_alloc(tcf_exts_num_actions(&f->exts));
if (!cls_flower.rule)
return -ENOMEM;
tc_cls_common_offload_init(&cls_flower.common, tp, f->flags, extack);
cls_flower.command = FLOW_CLS_REPLACE;
cls_flower.cookie = (unsigned long) f;
cls_flower.rule->match.dissector = &f->mask->dissector;
cls_flower.rule->match.mask = &f->mask->key;
cls_flower.rule->match.key = &f->mkey;
cls_flower.classid = f->res.classid;
err = tc_setup_flow_action(&cls_flower.rule->action, &f->exts);
if (err) {
kfree(cls_flower.rule);
if (skip_sw) {
NL_SET_ERR_MSG_MOD(extack, "Failed to setup flow action");
return err;
}
return 0;
}
net: sched: refactor block offloads counter usage Without rtnl lock protection filters can no longer safely manage block offloads counter themselves. Refactor cls API to protect block offloadcnt with tcf_block->cb_lock that is already used to protect driver callback list and nooffloaddevcnt counter. The counter can be modified by concurrent tasks by new functions that execute block callbacks (which is safe with previous patch that changed its type to atomic_t), however, block bind/unbind code that checks the counter value takes cb_lock in write mode to exclude any concurrent modifications. This approach prevents race conditions between bind/unbind and callback execution code but allows for concurrency for tc rule update path. Move block offload counter, filter in hardware counter and filter flags management from classifiers into cls hardware offloads API. Make functions tcf_block_offload_{inc|dec}() and tc_cls_offload_cnt_update() to be cls API private. Implement following new cls API to be used instead: tc_setup_cb_add() - non-destructive filter add. If filter that wasn't already in hardware is successfully offloaded, increment block offloads counter, set filter in hardware counter and flag. On failure, previously offloaded filter is considered to be intact and offloads counter is not decremented. tc_setup_cb_replace() - destructive filter replace. Release existing filter block offload counter and reset its in hardware counter and flag. Set new filter in hardware counter and flag. On failure, previously offloaded filter is considered to be destroyed and offload counter is decremented. tc_setup_cb_destroy() - filter destroy. Unconditionally decrement block offloads counter. tc_setup_cb_reoffload() - reoffload filter to single cb. Execute cb() and call tc_cls_offload_cnt_update() if cb() didn't return an error. Refactor all offload-capable classifiers to atomically offload filters to hardware, change block offload counter, and set filter in hardware counter and flag by means of the new cls API functions. Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-26 21:44:59 +08:00
err = tc_setup_cb_add(block, tp, TC_SETUP_CLSFLOWER, &cls_flower,
skip_sw, &f->flags, &f->in_hw_count, rtnl_held);
tc_cleanup_flow_action(&cls_flower.rule->action);
kfree(cls_flower.rule);
net: sched: refactor block offloads counter usage Without rtnl lock protection filters can no longer safely manage block offloads counter themselves. Refactor cls API to protect block offloadcnt with tcf_block->cb_lock that is already used to protect driver callback list and nooffloaddevcnt counter. The counter can be modified by concurrent tasks by new functions that execute block callbacks (which is safe with previous patch that changed its type to atomic_t), however, block bind/unbind code that checks the counter value takes cb_lock in write mode to exclude any concurrent modifications. This approach prevents race conditions between bind/unbind and callback execution code but allows for concurrency for tc rule update path. Move block offload counter, filter in hardware counter and filter flags management from classifiers into cls hardware offloads API. Make functions tcf_block_offload_{inc|dec}() and tc_cls_offload_cnt_update() to be cls API private. Implement following new cls API to be used instead: tc_setup_cb_add() - non-destructive filter add. If filter that wasn't already in hardware is successfully offloaded, increment block offloads counter, set filter in hardware counter and flag. On failure, previously offloaded filter is considered to be intact and offloads counter is not decremented. tc_setup_cb_replace() - destructive filter replace. Release existing filter block offload counter and reset its in hardware counter and flag. Set new filter in hardware counter and flag. On failure, previously offloaded filter is considered to be destroyed and offload counter is decremented. tc_setup_cb_destroy() - filter destroy. Unconditionally decrement block offloads counter. tc_setup_cb_reoffload() - reoffload filter to single cb. Execute cb() and call tc_cls_offload_cnt_update() if cb() didn't return an error. Refactor all offload-capable classifiers to atomically offload filters to hardware, change block offload counter, and set filter in hardware counter and flag by means of the new cls API functions. Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-26 21:44:59 +08:00
if (err) {
fl_hw_destroy_filter(tp, f, rtnl_held, NULL);
return err;
}
if (skip_sw && !(f->flags & TCA_CLS_FLAGS_IN_HW))
return -EINVAL;
return 0;
}
static void fl_hw_update_stats(struct tcf_proto *tp, struct cls_fl_filter *f,
bool rtnl_held)
{
struct tcf_block *block = tp->chain->block;
struct flow_cls_offload cls_flower = {};
tc_cls_common_offload_init(&cls_flower.common, tp, f->flags, NULL);
cls_flower.command = FLOW_CLS_STATS;
cls_flower.cookie = (unsigned long) f;
cls_flower.classid = f->res.classid;
tc_setup_cb_call(block, TC_SETUP_CLSFLOWER, &cls_flower, false,
rtnl_held);
tcf_exts_stats_update(&f->exts, cls_flower.stats.bytes,
cls_flower.stats.pkts,
cls_flower.stats.drops,
cls_flower.stats.lastused,
cls_flower.stats.used_hw_stats,
cls_flower.stats.used_hw_stats_valid);
}
static void __fl_put(struct cls_fl_filter *f)
{
if (!refcount_dec_and_test(&f->refcnt))
return;
if (tcf_exts_get_net(&f->exts))
tcf_queue_work(&f->rwork, fl_destroy_filter_work);
else
__fl_destroy_filter(f);
}
static struct cls_fl_filter *__fl_get(struct cls_fl_head *head, u32 handle)
{
struct cls_fl_filter *f;
rcu_read_lock();
f = idr_find(&head->handle_idr, handle);
if (f && !refcount_inc_not_zero(&f->refcnt))
f = NULL;
rcu_read_unlock();
return f;
}
static int __fl_delete(struct tcf_proto *tp, struct cls_fl_filter *f,
bool *last, bool rtnl_held,
struct netlink_ext_ack *extack)
{
struct cls_fl_head *head = fl_head_dereference(tp);
*last = false;
spin_lock(&tp->lock);
if (f->deleted) {
spin_unlock(&tp->lock);
return -ENOENT;
}
f->deleted = true;
rhashtable_remove_fast(&f->mask->ht, &f->ht_node,
f->mask->filter_ht_params);
idr_remove(&head->handle_idr, f->handle);
list_del_rcu(&f->list);
spin_unlock(&tp->lock);
net: sched: flower: fix filter net reference counting Fix net reference counting in fl_change() and remove redundant call to tcf_exts_get_net() from __fl_delete(). __fl_put() already tries to get net before releasing exts and deallocating a filter, so this code caused flower classifier to obtain net twice per filter that is being deleted. Implementation of __fl_delete() called tcf_exts_get_net() to pass its result as 'async' flag to fl_mask_put(). However, 'async' flag is redundant and only complicates fl_mask_put() implementation. This functionality seems to be copied from filter cleanup code, where it was added by Cong with following explanation: This patchset tries to fix the race between call_rcu() and cleanup_net() again. Without holding the netns refcnt the tc_action_net_exit() in netns workqueue could be called before filter destroy works in tc filter workqueue. This patchset moves the netns refcnt from tc actions to tcf_exts, without breaking per-netns tc actions. This doesn't apply to flower mask, which doesn't call any tc action code during cleanup. Simplify fl_mask_put() by removing the flag parameter and always use tcf_queue_work() to free mask objects. Fixes: 061775583e35 ("net: sched: flower: introduce reference counting for filters") Fixes: 1f17f7742eeb ("net: sched: flower: insert filter to ht before offloading it to hw") Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Reported-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-12 05:54:19 +08:00
*last = fl_mask_put(head, f->mask);
if (!tc_skip_hw(f->flags))
fl_hw_destroy_filter(tp, f, rtnl_held, extack);
tcf_unbind_filter(tp, &f->res);
__fl_put(f);
return 0;
}
net, sched: respect rcu grace period on cls destruction Roi reported a crash in flower where tp->root was NULL in ->classify() callbacks. Reason is that in ->destroy() tp->root is set to NULL via RCU_INIT_POINTER(). It's problematic for some of the classifiers, because this doesn't respect RCU grace period for them, and as a result, still outstanding readers from tc_classify() will try to blindly dereference a NULL tp->root. The tp->root object is strictly private to the classifier implementation and holds internal data the core such as tc_ctl_tfilter() doesn't know about. Within some classifiers, such as cls_bpf, cls_basic, etc, tp->root is only checked for NULL in ->get() callback, but nowhere else. This is misleading and seemed to be copied from old classifier code that was not cleaned up properly. For example, d3fa76ee6b4a ("[NET_SCHED]: cls_basic: fix NULL pointer dereference") moved tp->root initialization into ->init() routine, where before it was part of ->change(), so ->get() had to deal with tp->root being NULL back then, so that was indeed a valid case, after d3fa76ee6b4a, not really anymore. We used to set tp->root to NULL long ago in ->destroy(), see 47a1a1d4be29 ("pkt_sched: remove unnecessary xchg() in packet classifiers"); but the NULLifying was reintroduced with the RCUification, but it's not correct for every classifier implementation. In the cases that are fixed here with one exception of cls_cgroup, tp->root object is allocated and initialized inside ->init() callback, which is always performed at a point in time after we allocate a new tp, which means tp and thus tp->root was not globally visible in the tp chain yet (see tc_ctl_tfilter()). Also, on destruction tp->root is strictly kfree_rcu()'ed in ->destroy() handler, same for the tp which is kfree_rcu()'ed right when we return from ->destroy() in tcf_destroy(). This means, the head object's lifetime for such classifiers is always tied to the tp lifetime. The RCU callback invocation for the two kfree_rcu() could be out of order, but that's fine since both are independent. Dropping the RCU_INIT_POINTER(tp->root, NULL) for these classifiers here means that 1) we don't need a useless NULL check in fast-path and, 2) that outstanding readers of that tp in tc_classify() can still execute under respect with RCU grace period as it is actually expected. Things that haven't been touched here: cls_fw and cls_route. They each handle tp->root being NULL in ->classify() path for historic reasons, so their ->destroy() implementation can stay as is. If someone actually cares, they could get cleaned up at some point to avoid the test in fast path. cls_u32 doesn't set tp->root to NULL. For cls_rsvp, I just added a !head should anyone actually be using/testing it, so it at least aligns with cls_fw and cls_route. For cls_flower we additionally need to defer rhashtable destruction (to a sleepable context) after RCU grace period as concurrent readers might still access it. (Note that in this case we need to hold module reference to keep work callback address intact, since we only wait on module unload for all call_rcu()s to finish.) This fixes one race to bring RCU grace period guarantees back. Next step as worked on by Cong however is to fix 1e052be69d04 ("net_sched: destroy proto tp when all filters are gone") to get the order of unlinking the tp in tc_ctl_tfilter() for the RTM_DELTFILTER case right by moving RCU_INIT_POINTER() before tcf_destroy() and let the notification for removal be done through the prior ->delete() callback. Both are independant issues. Once we have that right, we can then clean tp->root up for a number of classifiers by not making them RCU pointers, which requires a new callback (->uninit) that is triggered from tp's RCU callback, where we just kfree() tp->root from there. Fixes: 1f947bf151e9 ("net: sched: rcu'ify cls_bpf") Fixes: 9888faefe132 ("net: sched: cls_basic use RCU") Fixes: 70da9f0bf999 ("net: sched: cls_flow use RCU") Fixes: 77b9900ef53a ("tc: introduce Flower classifier") Fixes: bf3994d2ed31 ("net/sched: introduce Match-all classifier") Fixes: 952313bd6258 ("net: sched: cls_cgroup use RCU") Reported-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: John Fastabend <john.fastabend@gmail.com> Cc: Roi Dayan <roid@mellanox.com> Cc: Jiri Pirko <jiri@mellanox.com> Acked-by: John Fastabend <john.r.fastabend@intel.com> Acked-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-27 08:18:01 +08:00
static void fl_destroy_sleepable(struct work_struct *work)
{
struct cls_fl_head *head = container_of(to_rcu_work(work),
struct cls_fl_head,
rwork);
rhashtable_destroy(&head->ht);
net, sched: respect rcu grace period on cls destruction Roi reported a crash in flower where tp->root was NULL in ->classify() callbacks. Reason is that in ->destroy() tp->root is set to NULL via RCU_INIT_POINTER(). It's problematic for some of the classifiers, because this doesn't respect RCU grace period for them, and as a result, still outstanding readers from tc_classify() will try to blindly dereference a NULL tp->root. The tp->root object is strictly private to the classifier implementation and holds internal data the core such as tc_ctl_tfilter() doesn't know about. Within some classifiers, such as cls_bpf, cls_basic, etc, tp->root is only checked for NULL in ->get() callback, but nowhere else. This is misleading and seemed to be copied from old classifier code that was not cleaned up properly. For example, d3fa76ee6b4a ("[NET_SCHED]: cls_basic: fix NULL pointer dereference") moved tp->root initialization into ->init() routine, where before it was part of ->change(), so ->get() had to deal with tp->root being NULL back then, so that was indeed a valid case, after d3fa76ee6b4a, not really anymore. We used to set tp->root to NULL long ago in ->destroy(), see 47a1a1d4be29 ("pkt_sched: remove unnecessary xchg() in packet classifiers"); but the NULLifying was reintroduced with the RCUification, but it's not correct for every classifier implementation. In the cases that are fixed here with one exception of cls_cgroup, tp->root object is allocated and initialized inside ->init() callback, which is always performed at a point in time after we allocate a new tp, which means tp and thus tp->root was not globally visible in the tp chain yet (see tc_ctl_tfilter()). Also, on destruction tp->root is strictly kfree_rcu()'ed in ->destroy() handler, same for the tp which is kfree_rcu()'ed right when we return from ->destroy() in tcf_destroy(). This means, the head object's lifetime for such classifiers is always tied to the tp lifetime. The RCU callback invocation for the two kfree_rcu() could be out of order, but that's fine since both are independent. Dropping the RCU_INIT_POINTER(tp->root, NULL) for these classifiers here means that 1) we don't need a useless NULL check in fast-path and, 2) that outstanding readers of that tp in tc_classify() can still execute under respect with RCU grace period as it is actually expected. Things that haven't been touched here: cls_fw and cls_route. They each handle tp->root being NULL in ->classify() path for historic reasons, so their ->destroy() implementation can stay as is. If someone actually cares, they could get cleaned up at some point to avoid the test in fast path. cls_u32 doesn't set tp->root to NULL. For cls_rsvp, I just added a !head should anyone actually be using/testing it, so it at least aligns with cls_fw and cls_route. For cls_flower we additionally need to defer rhashtable destruction (to a sleepable context) after RCU grace period as concurrent readers might still access it. (Note that in this case we need to hold module reference to keep work callback address intact, since we only wait on module unload for all call_rcu()s to finish.) This fixes one race to bring RCU grace period guarantees back. Next step as worked on by Cong however is to fix 1e052be69d04 ("net_sched: destroy proto tp when all filters are gone") to get the order of unlinking the tp in tc_ctl_tfilter() for the RTM_DELTFILTER case right by moving RCU_INIT_POINTER() before tcf_destroy() and let the notification for removal be done through the prior ->delete() callback. Both are independant issues. Once we have that right, we can then clean tp->root up for a number of classifiers by not making them RCU pointers, which requires a new callback (->uninit) that is triggered from tp's RCU callback, where we just kfree() tp->root from there. Fixes: 1f947bf151e9 ("net: sched: rcu'ify cls_bpf") Fixes: 9888faefe132 ("net: sched: cls_basic use RCU") Fixes: 70da9f0bf999 ("net: sched: cls_flow use RCU") Fixes: 77b9900ef53a ("tc: introduce Flower classifier") Fixes: bf3994d2ed31 ("net/sched: introduce Match-all classifier") Fixes: 952313bd6258 ("net: sched: cls_cgroup use RCU") Reported-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: John Fastabend <john.fastabend@gmail.com> Cc: Roi Dayan <roid@mellanox.com> Cc: Jiri Pirko <jiri@mellanox.com> Acked-by: John Fastabend <john.r.fastabend@intel.com> Acked-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-27 08:18:01 +08:00
kfree(head);
module_put(THIS_MODULE);
}
static void fl_destroy(struct tcf_proto *tp, bool rtnl_held,
struct netlink_ext_ack *extack)
{
struct cls_fl_head *head = fl_head_dereference(tp);
struct fl_flow_mask *mask, *next_mask;
struct cls_fl_filter *f, *next;
bool last;
list_for_each_entry_safe(mask, next_mask, &head->masks, list) {
list_for_each_entry_safe(f, next, &mask->filters, list) {
__fl_delete(tp, f, &last, rtnl_held, extack);
if (last)
break;
}
}
idr_destroy(&head->handle_idr);
net, sched: respect rcu grace period on cls destruction Roi reported a crash in flower where tp->root was NULL in ->classify() callbacks. Reason is that in ->destroy() tp->root is set to NULL via RCU_INIT_POINTER(). It's problematic for some of the classifiers, because this doesn't respect RCU grace period for them, and as a result, still outstanding readers from tc_classify() will try to blindly dereference a NULL tp->root. The tp->root object is strictly private to the classifier implementation and holds internal data the core such as tc_ctl_tfilter() doesn't know about. Within some classifiers, such as cls_bpf, cls_basic, etc, tp->root is only checked for NULL in ->get() callback, but nowhere else. This is misleading and seemed to be copied from old classifier code that was not cleaned up properly. For example, d3fa76ee6b4a ("[NET_SCHED]: cls_basic: fix NULL pointer dereference") moved tp->root initialization into ->init() routine, where before it was part of ->change(), so ->get() had to deal with tp->root being NULL back then, so that was indeed a valid case, after d3fa76ee6b4a, not really anymore. We used to set tp->root to NULL long ago in ->destroy(), see 47a1a1d4be29 ("pkt_sched: remove unnecessary xchg() in packet classifiers"); but the NULLifying was reintroduced with the RCUification, but it's not correct for every classifier implementation. In the cases that are fixed here with one exception of cls_cgroup, tp->root object is allocated and initialized inside ->init() callback, which is always performed at a point in time after we allocate a new tp, which means tp and thus tp->root was not globally visible in the tp chain yet (see tc_ctl_tfilter()). Also, on destruction tp->root is strictly kfree_rcu()'ed in ->destroy() handler, same for the tp which is kfree_rcu()'ed right when we return from ->destroy() in tcf_destroy(). This means, the head object's lifetime for such classifiers is always tied to the tp lifetime. The RCU callback invocation for the two kfree_rcu() could be out of order, but that's fine since both are independent. Dropping the RCU_INIT_POINTER(tp->root, NULL) for these classifiers here means that 1) we don't need a useless NULL check in fast-path and, 2) that outstanding readers of that tp in tc_classify() can still execute under respect with RCU grace period as it is actually expected. Things that haven't been touched here: cls_fw and cls_route. They each handle tp->root being NULL in ->classify() path for historic reasons, so their ->destroy() implementation can stay as is. If someone actually cares, they could get cleaned up at some point to avoid the test in fast path. cls_u32 doesn't set tp->root to NULL. For cls_rsvp, I just added a !head should anyone actually be using/testing it, so it at least aligns with cls_fw and cls_route. For cls_flower we additionally need to defer rhashtable destruction (to a sleepable context) after RCU grace period as concurrent readers might still access it. (Note that in this case we need to hold module reference to keep work callback address intact, since we only wait on module unload for all call_rcu()s to finish.) This fixes one race to bring RCU grace period guarantees back. Next step as worked on by Cong however is to fix 1e052be69d04 ("net_sched: destroy proto tp when all filters are gone") to get the order of unlinking the tp in tc_ctl_tfilter() for the RTM_DELTFILTER case right by moving RCU_INIT_POINTER() before tcf_destroy() and let the notification for removal be done through the prior ->delete() callback. Both are independant issues. Once we have that right, we can then clean tp->root up for a number of classifiers by not making them RCU pointers, which requires a new callback (->uninit) that is triggered from tp's RCU callback, where we just kfree() tp->root from there. Fixes: 1f947bf151e9 ("net: sched: rcu'ify cls_bpf") Fixes: 9888faefe132 ("net: sched: cls_basic use RCU") Fixes: 70da9f0bf999 ("net: sched: cls_flow use RCU") Fixes: 77b9900ef53a ("tc: introduce Flower classifier") Fixes: bf3994d2ed31 ("net/sched: introduce Match-all classifier") Fixes: 952313bd6258 ("net: sched: cls_cgroup use RCU") Reported-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: John Fastabend <john.fastabend@gmail.com> Cc: Roi Dayan <roid@mellanox.com> Cc: Jiri Pirko <jiri@mellanox.com> Acked-by: John Fastabend <john.r.fastabend@intel.com> Acked-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-27 08:18:01 +08:00
__module_get(THIS_MODULE);
tcf_queue_work(&head->rwork, fl_destroy_sleepable);
}
static void fl_put(struct tcf_proto *tp, void *arg)
{
struct cls_fl_filter *f = arg;
__fl_put(f);
}
static void *fl_get(struct tcf_proto *tp, u32 handle)
{
struct cls_fl_head *head = fl_head_dereference(tp);
return __fl_get(head, handle);
}
static const struct nla_policy fl_policy[TCA_FLOWER_MAX + 1] = {
[TCA_FLOWER_UNSPEC] = { .type = NLA_UNSPEC },
[TCA_FLOWER_CLASSID] = { .type = NLA_U32 },
[TCA_FLOWER_INDEV] = { .type = NLA_STRING,
.len = IFNAMSIZ },
[TCA_FLOWER_KEY_ETH_DST] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_ETH_DST_MASK] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_ETH_SRC] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_ETH_SRC_MASK] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_ETH_TYPE] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_IP_PROTO] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_IPV4_SRC] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_IPV4_SRC_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_IPV4_DST] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_IPV4_DST_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_IPV6_SRC] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_IPV6_SRC_MASK] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_IPV6_DST] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_IPV6_DST_MASK] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_TCP_SRC] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_TCP_DST] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_UDP_SRC] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_UDP_DST] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_VLAN_ID] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_VLAN_PRIO] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_VLAN_ETH_TYPE] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_ENC_KEY_ID] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ENC_IPV4_SRC] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ENC_IPV4_SRC_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ENC_IPV4_DST] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ENC_IPV4_DST_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ENC_IPV6_SRC] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_ENC_IPV6_SRC_MASK] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_ENC_IPV6_DST] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_ENC_IPV6_DST_MASK] = { .len = sizeof(struct in6_addr) },
[TCA_FLOWER_KEY_TCP_SRC_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_TCP_DST_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_UDP_SRC_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_UDP_DST_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_SCTP_SRC_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_SCTP_DST_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_SCTP_SRC] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_SCTP_DST] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_ENC_UDP_SRC_PORT] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_ENC_UDP_SRC_PORT_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_ENC_UDP_DST_PORT] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_ENC_UDP_DST_PORT_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_FLAGS] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_FLAGS_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ICMPV4_TYPE] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ICMPV4_TYPE_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ICMPV4_CODE] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ICMPV4_CODE_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ICMPV6_TYPE] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ICMPV6_TYPE_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ICMPV6_CODE] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ICMPV6_CODE_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ARP_SIP] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ARP_SIP_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ARP_TIP] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ARP_TIP_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ARP_OP] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ARP_OP_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ARP_SHA] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_ARP_SHA_MASK] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_ARP_THA] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_ARP_THA_MASK] = { .len = ETH_ALEN },
[TCA_FLOWER_KEY_MPLS_TTL] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_MPLS_BOS] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_MPLS_TC] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_MPLS_LABEL] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_MPLS_OPTS] = { .type = NLA_NESTED },
[TCA_FLOWER_KEY_TCP_FLAGS] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_TCP_FLAGS_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_IP_TOS] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_IP_TOS_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_IP_TTL] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_IP_TTL_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_CVLAN_ID] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_CVLAN_PRIO] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_CVLAN_ETH_TYPE] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_ENC_IP_TOS] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ENC_IP_TOS_MASK] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ENC_IP_TTL] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ENC_IP_TTL_MASK] = { .type = NLA_U8 },
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
[TCA_FLOWER_KEY_ENC_OPTS] = { .type = NLA_NESTED },
[TCA_FLOWER_KEY_ENC_OPTS_MASK] = { .type = NLA_NESTED },
[TCA_FLOWER_KEY_CT_STATE] =
NLA_POLICY_MASK(NLA_U16, TCA_FLOWER_KEY_CT_FLAGS_MASK),
[TCA_FLOWER_KEY_CT_STATE_MASK] =
NLA_POLICY_MASK(NLA_U16, TCA_FLOWER_KEY_CT_FLAGS_MASK),
[TCA_FLOWER_KEY_CT_ZONE] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_CT_ZONE_MASK] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_CT_MARK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_CT_MARK_MASK] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_CT_LABELS] = { .type = NLA_BINARY,
.len = 128 / BITS_PER_BYTE },
[TCA_FLOWER_KEY_CT_LABELS_MASK] = { .type = NLA_BINARY,
.len = 128 / BITS_PER_BYTE },
[TCA_FLOWER_FLAGS] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_HASH] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_HASH_MASK] = { .type = NLA_U32 },
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
};
static const struct nla_policy
enc_opts_policy[TCA_FLOWER_KEY_ENC_OPTS_MAX + 1] = {
[TCA_FLOWER_KEY_ENC_OPTS_UNSPEC] = {
.strict_start_type = TCA_FLOWER_KEY_ENC_OPTS_VXLAN },
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
[TCA_FLOWER_KEY_ENC_OPTS_GENEVE] = { .type = NLA_NESTED },
[TCA_FLOWER_KEY_ENC_OPTS_VXLAN] = { .type = NLA_NESTED },
[TCA_FLOWER_KEY_ENC_OPTS_ERSPAN] = { .type = NLA_NESTED },
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
};
static const struct nla_policy
geneve_opt_policy[TCA_FLOWER_KEY_ENC_OPT_GENEVE_MAX + 1] = {
[TCA_FLOWER_KEY_ENC_OPT_GENEVE_CLASS] = { .type = NLA_U16 },
[TCA_FLOWER_KEY_ENC_OPT_GENEVE_TYPE] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ENC_OPT_GENEVE_DATA] = { .type = NLA_BINARY,
.len = 128 },
};
static const struct nla_policy
vxlan_opt_policy[TCA_FLOWER_KEY_ENC_OPT_VXLAN_MAX + 1] = {
[TCA_FLOWER_KEY_ENC_OPT_VXLAN_GBP] = { .type = NLA_U32 },
};
static const struct nla_policy
erspan_opt_policy[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_MAX + 1] = {
[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_VER] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_INDEX] = { .type = NLA_U32 },
[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_DIR] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_HWID] = { .type = NLA_U8 },
};
static const struct nla_policy
mpls_stack_entry_policy[TCA_FLOWER_KEY_MPLS_OPT_LSE_MAX + 1] = {
[TCA_FLOWER_KEY_MPLS_OPT_LSE_DEPTH] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_MPLS_OPT_LSE_TTL] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_MPLS_OPT_LSE_BOS] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_MPLS_OPT_LSE_TC] = { .type = NLA_U8 },
[TCA_FLOWER_KEY_MPLS_OPT_LSE_LABEL] = { .type = NLA_U32 },
};
static void fl_set_key_val(struct nlattr **tb,
void *val, int val_type,
void *mask, int mask_type, int len)
{
if (!tb[val_type])
return;
nla_memcpy(val, tb[val_type], len);
if (mask_type == TCA_FLOWER_UNSPEC || !tb[mask_type])
memset(mask, 0xff, len);
else
nla_memcpy(mask, tb[mask_type], len);
}
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
static int fl_set_key_port_range(struct nlattr **tb, struct fl_flow_key *key,
struct fl_flow_key *mask,
struct netlink_ext_ack *extack)
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
{
cls_flower: Fix the behavior using port ranges with hw-offload The recent commit 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") had added filtering based on port ranges to tc flower. However the commit missed necessary changes in hw-offload code, so the feature gave rise to generating incorrect offloaded flow keys in NIC. One more detailed example is below: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 100-200 action drop With the setup above, an exact match filter with dst_port == 0 will be installed in NIC by hw-offload. IOW, the NIC will have a rule which is equivalent to the following one. $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 0 action drop The behavior was caused by the flow dissector which extracts packet data into the flow key in the tc flower. More specifically, regardless of exact match or specified port ranges, fl_init_dissector() set the FLOW_DISSECTOR_KEY_PORTS flag in struct flow_dissector to extract port numbers from skb in skb_flow_dissect() called by fl_classify(). Note that device drivers received the same struct flow_dissector object as used in skb_flow_dissect(). Thus, offloaded drivers could not identify which of these is used because the FLOW_DISSECTOR_KEY_PORTS flag was set to struct flow_dissector in either case. This patch adds the new FLOW_DISSECTOR_KEY_PORTS_RANGE flag and the new tp_range field in struct fl_flow_key to recognize which filters are applied to offloaded drivers. At this point, when filters based on port ranges passed to drivers, drivers return the EOPNOTSUPP error because they do not support the feature (the newly created FLOW_DISSECTOR_KEY_PORTS_RANGE flag). Fixes: 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-03 18:40:12 +08:00
fl_set_key_val(tb, &key->tp_range.tp_min.dst,
TCA_FLOWER_KEY_PORT_DST_MIN, &mask->tp_range.tp_min.dst,
TCA_FLOWER_UNSPEC, sizeof(key->tp_range.tp_min.dst));
fl_set_key_val(tb, &key->tp_range.tp_max.dst,
TCA_FLOWER_KEY_PORT_DST_MAX, &mask->tp_range.tp_max.dst,
TCA_FLOWER_UNSPEC, sizeof(key->tp_range.tp_max.dst));
fl_set_key_val(tb, &key->tp_range.tp_min.src,
TCA_FLOWER_KEY_PORT_SRC_MIN, &mask->tp_range.tp_min.src,
TCA_FLOWER_UNSPEC, sizeof(key->tp_range.tp_min.src));
fl_set_key_val(tb, &key->tp_range.tp_max.src,
TCA_FLOWER_KEY_PORT_SRC_MAX, &mask->tp_range.tp_max.src,
TCA_FLOWER_UNSPEC, sizeof(key->tp_range.tp_max.src));
if (mask->tp_range.tp_min.dst && mask->tp_range.tp_max.dst &&
ntohs(key->tp_range.tp_max.dst) <=
ntohs(key->tp_range.tp_min.dst)) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_PORT_DST_MIN],
"Invalid destination port range (min must be strictly smaller than max)");
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
return -EINVAL;
}
if (mask->tp_range.tp_min.src && mask->tp_range.tp_max.src &&
ntohs(key->tp_range.tp_max.src) <=
ntohs(key->tp_range.tp_min.src)) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_PORT_SRC_MIN],
"Invalid source port range (min must be strictly smaller than max)");
return -EINVAL;
}
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
return 0;
}
static int fl_set_key_mpls_lse(const struct nlattr *nla_lse,
struct flow_dissector_key_mpls *key_val,
struct flow_dissector_key_mpls *key_mask,
struct netlink_ext_ack *extack)
{
struct nlattr *tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_MAX + 1];
struct flow_dissector_mpls_lse *lse_mask;
struct flow_dissector_mpls_lse *lse_val;
u8 lse_index;
u8 depth;
int err;
err = nla_parse_nested(tb, TCA_FLOWER_KEY_MPLS_OPT_LSE_MAX, nla_lse,
mpls_stack_entry_policy, extack);
if (err < 0)
return err;
if (!tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_DEPTH]) {
NL_SET_ERR_MSG(extack, "Missing MPLS option \"depth\"");
return -EINVAL;
}
depth = nla_get_u8(tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_DEPTH]);
/* LSE depth starts at 1, for consistency with terminology used by
* RFC 3031 (section 3.9), where depth 0 refers to unlabeled packets.
*/
if (depth < 1 || depth > FLOW_DIS_MPLS_MAX) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_DEPTH],
"Invalid MPLS depth");
return -EINVAL;
}
lse_index = depth - 1;
dissector_set_mpls_lse(key_val, lse_index);
dissector_set_mpls_lse(key_mask, lse_index);
lse_val = &key_val->ls[lse_index];
lse_mask = &key_mask->ls[lse_index];
if (tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_TTL]) {
lse_val->mpls_ttl = nla_get_u8(tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_TTL]);
lse_mask->mpls_ttl = MPLS_TTL_MASK;
}
if (tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_BOS]) {
u8 bos = nla_get_u8(tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_BOS]);
if (bos & ~MPLS_BOS_MASK) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_BOS],
"Bottom Of Stack (BOS) must be 0 or 1");
return -EINVAL;
}
lse_val->mpls_bos = bos;
lse_mask->mpls_bos = MPLS_BOS_MASK;
}
if (tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_TC]) {
u8 tc = nla_get_u8(tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_TC]);
if (tc & ~MPLS_TC_MASK) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_TC],
"Traffic Class (TC) must be between 0 and 7");
return -EINVAL;
}
lse_val->mpls_tc = tc;
lse_mask->mpls_tc = MPLS_TC_MASK;
}
if (tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_LABEL]) {
u32 label = nla_get_u32(tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_LABEL]);
if (label & ~MPLS_LABEL_MASK) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_OPT_LSE_LABEL],
"Label must be between 0 and 1048575");
return -EINVAL;
}
lse_val->mpls_label = label;
lse_mask->mpls_label = MPLS_LABEL_MASK;
}
return 0;
}
static int fl_set_key_mpls_opts(const struct nlattr *nla_mpls_opts,
struct flow_dissector_key_mpls *key_val,
struct flow_dissector_key_mpls *key_mask,
struct netlink_ext_ack *extack)
{
struct nlattr *nla_lse;
int rem;
int err;
if (!(nla_mpls_opts->nla_type & NLA_F_NESTED)) {
NL_SET_ERR_MSG_ATTR(extack, nla_mpls_opts,
"NLA_F_NESTED is missing");
return -EINVAL;
}
nla_for_each_nested(nla_lse, nla_mpls_opts, rem) {
if (nla_type(nla_lse) != TCA_FLOWER_KEY_MPLS_OPTS_LSE) {
NL_SET_ERR_MSG_ATTR(extack, nla_lse,
"Invalid MPLS option type");
return -EINVAL;
}
err = fl_set_key_mpls_lse(nla_lse, key_val, key_mask, extack);
if (err < 0)
return err;
}
if (rem) {
NL_SET_ERR_MSG(extack,
"Bytes leftover after parsing MPLS options");
return -EINVAL;
}
return 0;
}
static int fl_set_key_mpls(struct nlattr **tb,
struct flow_dissector_key_mpls *key_val,
struct flow_dissector_key_mpls *key_mask,
struct netlink_ext_ack *extack)
{
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
struct flow_dissector_mpls_lse *lse_mask;
struct flow_dissector_mpls_lse *lse_val;
if (tb[TCA_FLOWER_KEY_MPLS_OPTS]) {
if (tb[TCA_FLOWER_KEY_MPLS_TTL] ||
tb[TCA_FLOWER_KEY_MPLS_BOS] ||
tb[TCA_FLOWER_KEY_MPLS_TC] ||
tb[TCA_FLOWER_KEY_MPLS_LABEL]) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_OPTS],
"MPLS label, Traffic Class, Bottom Of Stack and Time To Live must be encapsulated in the MPLS options attribute");
return -EBADMSG;
}
return fl_set_key_mpls_opts(tb[TCA_FLOWER_KEY_MPLS_OPTS],
key_val, key_mask, extack);
}
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_val = &key_val->ls[0];
lse_mask = &key_mask->ls[0];
if (tb[TCA_FLOWER_KEY_MPLS_TTL]) {
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_val->mpls_ttl = nla_get_u8(tb[TCA_FLOWER_KEY_MPLS_TTL]);
lse_mask->mpls_ttl = MPLS_TTL_MASK;
dissector_set_mpls_lse(key_val, 0);
dissector_set_mpls_lse(key_mask, 0);
}
if (tb[TCA_FLOWER_KEY_MPLS_BOS]) {
u8 bos = nla_get_u8(tb[TCA_FLOWER_KEY_MPLS_BOS]);
if (bos & ~MPLS_BOS_MASK) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_BOS],
"Bottom Of Stack (BOS) must be 0 or 1");
return -EINVAL;
}
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_val->mpls_bos = bos;
lse_mask->mpls_bos = MPLS_BOS_MASK;
dissector_set_mpls_lse(key_val, 0);
dissector_set_mpls_lse(key_mask, 0);
}
if (tb[TCA_FLOWER_KEY_MPLS_TC]) {
u8 tc = nla_get_u8(tb[TCA_FLOWER_KEY_MPLS_TC]);
if (tc & ~MPLS_TC_MASK) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_TC],
"Traffic Class (TC) must be between 0 and 7");
return -EINVAL;
}
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_val->mpls_tc = tc;
lse_mask->mpls_tc = MPLS_TC_MASK;
dissector_set_mpls_lse(key_val, 0);
dissector_set_mpls_lse(key_mask, 0);
}
if (tb[TCA_FLOWER_KEY_MPLS_LABEL]) {
u32 label = nla_get_u32(tb[TCA_FLOWER_KEY_MPLS_LABEL]);
if (label & ~MPLS_LABEL_MASK) {
NL_SET_ERR_MSG_ATTR(extack,
tb[TCA_FLOWER_KEY_MPLS_LABEL],
"Label must be between 0 and 1048575");
return -EINVAL;
}
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_val->mpls_label = label;
lse_mask->mpls_label = MPLS_LABEL_MASK;
dissector_set_mpls_lse(key_val, 0);
dissector_set_mpls_lse(key_mask, 0);
}
return 0;
}
static void fl_set_key_vlan(struct nlattr **tb,
__be16 ethertype,
int vlan_id_key, int vlan_prio_key,
struct flow_dissector_key_vlan *key_val,
struct flow_dissector_key_vlan *key_mask)
{
#define VLAN_PRIORITY_MASK 0x7
if (tb[vlan_id_key]) {
key_val->vlan_id =
nla_get_u16(tb[vlan_id_key]) & VLAN_VID_MASK;
key_mask->vlan_id = VLAN_VID_MASK;
}
if (tb[vlan_prio_key]) {
key_val->vlan_priority =
nla_get_u8(tb[vlan_prio_key]) &
VLAN_PRIORITY_MASK;
key_mask->vlan_priority = VLAN_PRIORITY_MASK;
}
key_val->vlan_tpid = ethertype;
key_mask->vlan_tpid = cpu_to_be16(~0);
}
static void fl_set_key_flag(u32 flower_key, u32 flower_mask,
u32 *dissector_key, u32 *dissector_mask,
u32 flower_flag_bit, u32 dissector_flag_bit)
{
if (flower_mask & flower_flag_bit) {
*dissector_mask |= dissector_flag_bit;
if (flower_key & flower_flag_bit)
*dissector_key |= dissector_flag_bit;
}
}
static int fl_set_key_flags(struct nlattr **tb, u32 *flags_key,
u32 *flags_mask, struct netlink_ext_ack *extack)
{
u32 key, mask;
/* mask is mandatory for flags */
if (!tb[TCA_FLOWER_KEY_FLAGS_MASK]) {
NL_SET_ERR_MSG(extack, "Missing flags mask");
return -EINVAL;
}
key = be32_to_cpu(nla_get_be32(tb[TCA_FLOWER_KEY_FLAGS]));
mask = be32_to_cpu(nla_get_be32(tb[TCA_FLOWER_KEY_FLAGS_MASK]));
*flags_key = 0;
*flags_mask = 0;
fl_set_key_flag(key, mask, flags_key, flags_mask,
TCA_FLOWER_KEY_FLAGS_IS_FRAGMENT, FLOW_DIS_IS_FRAGMENT);
fl_set_key_flag(key, mask, flags_key, flags_mask,
TCA_FLOWER_KEY_FLAGS_FRAG_IS_FIRST,
FLOW_DIS_FIRST_FRAG);
return 0;
}
static void fl_set_key_ip(struct nlattr **tb, bool encap,
struct flow_dissector_key_ip *key,
struct flow_dissector_key_ip *mask)
{
int tos_key = encap ? TCA_FLOWER_KEY_ENC_IP_TOS : TCA_FLOWER_KEY_IP_TOS;
int ttl_key = encap ? TCA_FLOWER_KEY_ENC_IP_TTL : TCA_FLOWER_KEY_IP_TTL;
int tos_mask = encap ? TCA_FLOWER_KEY_ENC_IP_TOS_MASK : TCA_FLOWER_KEY_IP_TOS_MASK;
int ttl_mask = encap ? TCA_FLOWER_KEY_ENC_IP_TTL_MASK : TCA_FLOWER_KEY_IP_TTL_MASK;
fl_set_key_val(tb, &key->tos, tos_key, &mask->tos, tos_mask, sizeof(key->tos));
fl_set_key_val(tb, &key->ttl, ttl_key, &mask->ttl, ttl_mask, sizeof(key->ttl));
}
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
static int fl_set_geneve_opt(const struct nlattr *nla, struct fl_flow_key *key,
int depth, int option_len,
struct netlink_ext_ack *extack)
{
struct nlattr *tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_MAX + 1];
struct nlattr *class = NULL, *type = NULL, *data = NULL;
struct geneve_opt *opt;
int err, data_len = 0;
if (option_len > sizeof(struct geneve_opt))
data_len = option_len - sizeof(struct geneve_opt);
opt = (struct geneve_opt *)&key->enc_opts.data[key->enc_opts.len];
memset(opt, 0xff, option_len);
opt->length = data_len / 4;
opt->r1 = 0;
opt->r2 = 0;
opt->r3 = 0;
/* If no mask has been prodived we assume an exact match. */
if (!depth)
return sizeof(struct geneve_opt) + data_len;
if (nla_type(nla) != TCA_FLOWER_KEY_ENC_OPTS_GENEVE) {
NL_SET_ERR_MSG(extack, "Non-geneve option type for mask");
return -EINVAL;
}
netlink: make validation more configurable for future strictness We currently have two levels of strict validation: 1) liberal (default) - undefined (type >= max) & NLA_UNSPEC attributes accepted - attribute length >= expected accepted - garbage at end of message accepted 2) strict (opt-in) - NLA_UNSPEC attributes accepted - attribute length >= expected accepted Split out parsing strictness into four different options: * TRAILING - check that there's no trailing data after parsing attributes (in message or nested) * MAXTYPE - reject attrs > max known type * UNSPEC - reject attributes with NLA_UNSPEC policy entries * STRICT_ATTRS - strictly validate attribute size The default for future things should be *everything*. The current *_strict() is a combination of TRAILING and MAXTYPE, and is renamed to _deprecated_strict(). The current regular parsing has none of this, and is renamed to *_parse_deprecated(). Additionally it allows us to selectively set one of the new flags even on old policies. Notably, the UNSPEC flag could be useful in this case, since it can be arranged (by filling in the policy) to not be an incompatible userspace ABI change, but would then going forward prevent forgetting attribute entries. Similar can apply to the POLICY flag. We end up with the following renames: * nla_parse -> nla_parse_deprecated * nla_parse_strict -> nla_parse_deprecated_strict * nlmsg_parse -> nlmsg_parse_deprecated * nlmsg_parse_strict -> nlmsg_parse_deprecated_strict * nla_parse_nested -> nla_parse_nested_deprecated * nla_validate_nested -> nla_validate_nested_deprecated Using spatch, of course: @@ expression TB, MAX, HEAD, LEN, POL, EXT; @@ -nla_parse(TB, MAX, HEAD, LEN, POL, EXT) +nla_parse_deprecated(TB, MAX, HEAD, LEN, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse_strict(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated_strict(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression TB, MAX, NLA, POL, EXT; @@ -nla_parse_nested(TB, MAX, NLA, POL, EXT) +nla_parse_nested_deprecated(TB, MAX, NLA, POL, EXT) @@ expression START, MAX, POL, EXT; @@ -nla_validate_nested(START, MAX, POL, EXT) +nla_validate_nested_deprecated(START, MAX, POL, EXT) @@ expression NLH, HDRLEN, MAX, POL, EXT; @@ -nlmsg_validate(NLH, HDRLEN, MAX, POL, EXT) +nlmsg_validate_deprecated(NLH, HDRLEN, MAX, POL, EXT) For this patch, don't actually add the strict, non-renamed versions yet so that it breaks compile if I get it wrong. Also, while at it, make nla_validate and nla_parse go down to a common __nla_validate_parse() function to avoid code duplication. Ultimately, this allows us to have very strict validation for every new caller of nla_parse()/nlmsg_parse() etc as re-introduced in the next patch, while existing things will continue to work as is. In effect then, this adds fully strict validation for any new command. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-26 20:07:28 +08:00
err = nla_parse_nested_deprecated(tb,
TCA_FLOWER_KEY_ENC_OPT_GENEVE_MAX,
nla, geneve_opt_policy, extack);
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
if (err < 0)
return err;
/* We are not allowed to omit any of CLASS, TYPE or DATA
* fields from the key.
*/
if (!option_len &&
(!tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_CLASS] ||
!tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_TYPE] ||
!tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_DATA])) {
NL_SET_ERR_MSG(extack, "Missing tunnel key geneve option class, type or data");
return -EINVAL;
}
/* Omitting any of CLASS, TYPE or DATA fields is allowed
* for the mask.
*/
if (tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_DATA]) {
int new_len = key->enc_opts.len;
data = tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_DATA];
data_len = nla_len(data);
if (data_len < 4) {
NL_SET_ERR_MSG(extack, "Tunnel key geneve option data is less than 4 bytes long");
return -ERANGE;
}
if (data_len % 4) {
NL_SET_ERR_MSG(extack, "Tunnel key geneve option data is not a multiple of 4 bytes long");
return -ERANGE;
}
new_len += sizeof(struct geneve_opt) + data_len;
BUILD_BUG_ON(FLOW_DIS_TUN_OPTS_MAX != IP_TUNNEL_OPTS_MAX);
if (new_len > FLOW_DIS_TUN_OPTS_MAX) {
NL_SET_ERR_MSG(extack, "Tunnel options exceeds max size");
return -ERANGE;
}
opt->length = data_len / 4;
memcpy(opt->opt_data, nla_data(data), data_len);
}
if (tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_CLASS]) {
class = tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_CLASS];
opt->opt_class = nla_get_be16(class);
}
if (tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_TYPE]) {
type = tb[TCA_FLOWER_KEY_ENC_OPT_GENEVE_TYPE];
opt->type = nla_get_u8(type);
}
return sizeof(struct geneve_opt) + data_len;
}
static int fl_set_vxlan_opt(const struct nlattr *nla, struct fl_flow_key *key,
int depth, int option_len,
struct netlink_ext_ack *extack)
{
struct nlattr *tb[TCA_FLOWER_KEY_ENC_OPT_VXLAN_MAX + 1];
struct vxlan_metadata *md;
int err;
md = (struct vxlan_metadata *)&key->enc_opts.data[key->enc_opts.len];
memset(md, 0xff, sizeof(*md));
if (!depth)
return sizeof(*md);
if (nla_type(nla) != TCA_FLOWER_KEY_ENC_OPTS_VXLAN) {
NL_SET_ERR_MSG(extack, "Non-vxlan option type for mask");
return -EINVAL;
}
err = nla_parse_nested(tb, TCA_FLOWER_KEY_ENC_OPT_VXLAN_MAX, nla,
vxlan_opt_policy, extack);
if (err < 0)
return err;
if (!option_len && !tb[TCA_FLOWER_KEY_ENC_OPT_VXLAN_GBP]) {
NL_SET_ERR_MSG(extack, "Missing tunnel key vxlan option gbp");
return -EINVAL;
}
if (tb[TCA_FLOWER_KEY_ENC_OPT_VXLAN_GBP]) {
md->gbp = nla_get_u32(tb[TCA_FLOWER_KEY_ENC_OPT_VXLAN_GBP]);
md->gbp &= VXLAN_GBP_MASK;
}
return sizeof(*md);
}
static int fl_set_erspan_opt(const struct nlattr *nla, struct fl_flow_key *key,
int depth, int option_len,
struct netlink_ext_ack *extack)
{
struct nlattr *tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_MAX + 1];
struct erspan_metadata *md;
int err;
md = (struct erspan_metadata *)&key->enc_opts.data[key->enc_opts.len];
memset(md, 0xff, sizeof(*md));
md->version = 1;
if (!depth)
return sizeof(*md);
if (nla_type(nla) != TCA_FLOWER_KEY_ENC_OPTS_ERSPAN) {
NL_SET_ERR_MSG(extack, "Non-erspan option type for mask");
return -EINVAL;
}
err = nla_parse_nested(tb, TCA_FLOWER_KEY_ENC_OPT_ERSPAN_MAX, nla,
erspan_opt_policy, extack);
if (err < 0)
return err;
if (!option_len && !tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_VER]) {
NL_SET_ERR_MSG(extack, "Missing tunnel key erspan option ver");
return -EINVAL;
}
if (tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_VER])
md->version = nla_get_u8(tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_VER]);
if (md->version == 1) {
if (!option_len && !tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_INDEX]) {
NL_SET_ERR_MSG(extack, "Missing tunnel key erspan option index");
return -EINVAL;
}
if (tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_INDEX]) {
nla = tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_INDEX];
memset(&md->u, 0x00, sizeof(md->u));
md->u.index = nla_get_be32(nla);
}
} else if (md->version == 2) {
if (!option_len && (!tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_DIR] ||
!tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_HWID])) {
NL_SET_ERR_MSG(extack, "Missing tunnel key erspan option dir or hwid");
return -EINVAL;
}
if (tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_DIR]) {
nla = tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_DIR];
md->u.md2.dir = nla_get_u8(nla);
}
if (tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_HWID]) {
nla = tb[TCA_FLOWER_KEY_ENC_OPT_ERSPAN_HWID];
set_hwid(&md->u.md2, nla_get_u8(nla));
}
} else {
NL_SET_ERR_MSG(extack, "Tunnel key erspan option ver is incorrect");
return -EINVAL;
}
return sizeof(*md);
}
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key,
struct fl_flow_key *mask,
struct netlink_ext_ack *extack)
{
const struct nlattr *nla_enc_key, *nla_opt_key, *nla_opt_msk = NULL;
int err, option_len, key_depth, msk_depth = 0;
netlink: make validation more configurable for future strictness We currently have two levels of strict validation: 1) liberal (default) - undefined (type >= max) & NLA_UNSPEC attributes accepted - attribute length >= expected accepted - garbage at end of message accepted 2) strict (opt-in) - NLA_UNSPEC attributes accepted - attribute length >= expected accepted Split out parsing strictness into four different options: * TRAILING - check that there's no trailing data after parsing attributes (in message or nested) * MAXTYPE - reject attrs > max known type * UNSPEC - reject attributes with NLA_UNSPEC policy entries * STRICT_ATTRS - strictly validate attribute size The default for future things should be *everything*. The current *_strict() is a combination of TRAILING and MAXTYPE, and is renamed to _deprecated_strict(). The current regular parsing has none of this, and is renamed to *_parse_deprecated(). Additionally it allows us to selectively set one of the new flags even on old policies. Notably, the UNSPEC flag could be useful in this case, since it can be arranged (by filling in the policy) to not be an incompatible userspace ABI change, but would then going forward prevent forgetting attribute entries. Similar can apply to the POLICY flag. We end up with the following renames: * nla_parse -> nla_parse_deprecated * nla_parse_strict -> nla_parse_deprecated_strict * nlmsg_parse -> nlmsg_parse_deprecated * nlmsg_parse_strict -> nlmsg_parse_deprecated_strict * nla_parse_nested -> nla_parse_nested_deprecated * nla_validate_nested -> nla_validate_nested_deprecated Using spatch, of course: @@ expression TB, MAX, HEAD, LEN, POL, EXT; @@ -nla_parse(TB, MAX, HEAD, LEN, POL, EXT) +nla_parse_deprecated(TB, MAX, HEAD, LEN, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse_strict(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated_strict(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression TB, MAX, NLA, POL, EXT; @@ -nla_parse_nested(TB, MAX, NLA, POL, EXT) +nla_parse_nested_deprecated(TB, MAX, NLA, POL, EXT) @@ expression START, MAX, POL, EXT; @@ -nla_validate_nested(START, MAX, POL, EXT) +nla_validate_nested_deprecated(START, MAX, POL, EXT) @@ expression NLH, HDRLEN, MAX, POL, EXT; @@ -nlmsg_validate(NLH, HDRLEN, MAX, POL, EXT) +nlmsg_validate_deprecated(NLH, HDRLEN, MAX, POL, EXT) For this patch, don't actually add the strict, non-renamed versions yet so that it breaks compile if I get it wrong. Also, while at it, make nla_validate and nla_parse go down to a common __nla_validate_parse() function to avoid code duplication. Ultimately, this allows us to have very strict validation for every new caller of nla_parse()/nlmsg_parse() etc as re-introduced in the next patch, while existing things will continue to work as is. In effect then, this adds fully strict validation for any new command. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-26 20:07:28 +08:00
err = nla_validate_nested_deprecated(tb[TCA_FLOWER_KEY_ENC_OPTS],
TCA_FLOWER_KEY_ENC_OPTS_MAX,
enc_opts_policy, extack);
if (err)
return err;
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
nla_enc_key = nla_data(tb[TCA_FLOWER_KEY_ENC_OPTS]);
if (tb[TCA_FLOWER_KEY_ENC_OPTS_MASK]) {
netlink: make validation more configurable for future strictness We currently have two levels of strict validation: 1) liberal (default) - undefined (type >= max) & NLA_UNSPEC attributes accepted - attribute length >= expected accepted - garbage at end of message accepted 2) strict (opt-in) - NLA_UNSPEC attributes accepted - attribute length >= expected accepted Split out parsing strictness into four different options: * TRAILING - check that there's no trailing data after parsing attributes (in message or nested) * MAXTYPE - reject attrs > max known type * UNSPEC - reject attributes with NLA_UNSPEC policy entries * STRICT_ATTRS - strictly validate attribute size The default for future things should be *everything*. The current *_strict() is a combination of TRAILING and MAXTYPE, and is renamed to _deprecated_strict(). The current regular parsing has none of this, and is renamed to *_parse_deprecated(). Additionally it allows us to selectively set one of the new flags even on old policies. Notably, the UNSPEC flag could be useful in this case, since it can be arranged (by filling in the policy) to not be an incompatible userspace ABI change, but would then going forward prevent forgetting attribute entries. Similar can apply to the POLICY flag. We end up with the following renames: * nla_parse -> nla_parse_deprecated * nla_parse_strict -> nla_parse_deprecated_strict * nlmsg_parse -> nlmsg_parse_deprecated * nlmsg_parse_strict -> nlmsg_parse_deprecated_strict * nla_parse_nested -> nla_parse_nested_deprecated * nla_validate_nested -> nla_validate_nested_deprecated Using spatch, of course: @@ expression TB, MAX, HEAD, LEN, POL, EXT; @@ -nla_parse(TB, MAX, HEAD, LEN, POL, EXT) +nla_parse_deprecated(TB, MAX, HEAD, LEN, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse_strict(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated_strict(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression TB, MAX, NLA, POL, EXT; @@ -nla_parse_nested(TB, MAX, NLA, POL, EXT) +nla_parse_nested_deprecated(TB, MAX, NLA, POL, EXT) @@ expression START, MAX, POL, EXT; @@ -nla_validate_nested(START, MAX, POL, EXT) +nla_validate_nested_deprecated(START, MAX, POL, EXT) @@ expression NLH, HDRLEN, MAX, POL, EXT; @@ -nlmsg_validate(NLH, HDRLEN, MAX, POL, EXT) +nlmsg_validate_deprecated(NLH, HDRLEN, MAX, POL, EXT) For this patch, don't actually add the strict, non-renamed versions yet so that it breaks compile if I get it wrong. Also, while at it, make nla_validate and nla_parse go down to a common __nla_validate_parse() function to avoid code duplication. Ultimately, this allows us to have very strict validation for every new caller of nla_parse()/nlmsg_parse() etc as re-introduced in the next patch, while existing things will continue to work as is. In effect then, this adds fully strict validation for any new command. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-26 20:07:28 +08:00
err = nla_validate_nested_deprecated(tb[TCA_FLOWER_KEY_ENC_OPTS_MASK],
TCA_FLOWER_KEY_ENC_OPTS_MAX,
enc_opts_policy, extack);
if (err)
return err;
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
nla_opt_msk = nla_data(tb[TCA_FLOWER_KEY_ENC_OPTS_MASK]);
msk_depth = nla_len(tb[TCA_FLOWER_KEY_ENC_OPTS_MASK]);
if (!nla_ok(nla_opt_msk, msk_depth)) {
NL_SET_ERR_MSG(extack, "Invalid nested attribute for masks");
return -EINVAL;
}
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
}
nla_for_each_attr(nla_opt_key, nla_enc_key,
nla_len(tb[TCA_FLOWER_KEY_ENC_OPTS]), key_depth) {
switch (nla_type(nla_opt_key)) {
case TCA_FLOWER_KEY_ENC_OPTS_GENEVE:
if (key->enc_opts.dst_opt_type &&
key->enc_opts.dst_opt_type != TUNNEL_GENEVE_OPT) {
NL_SET_ERR_MSG(extack, "Duplicate type for geneve options");
return -EINVAL;
}
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
option_len = 0;
key->enc_opts.dst_opt_type = TUNNEL_GENEVE_OPT;
option_len = fl_set_geneve_opt(nla_opt_key, key,
key_depth, option_len,
extack);
if (option_len < 0)
return option_len;
key->enc_opts.len += option_len;
/* At the same time we need to parse through the mask
* in order to verify exact and mask attribute lengths.
*/
mask->enc_opts.dst_opt_type = TUNNEL_GENEVE_OPT;
option_len = fl_set_geneve_opt(nla_opt_msk, mask,
msk_depth, option_len,
extack);
if (option_len < 0)
return option_len;
mask->enc_opts.len += option_len;
if (key->enc_opts.len != mask->enc_opts.len) {
NL_SET_ERR_MSG(extack, "Key and mask miss aligned");
return -EINVAL;
}
break;
case TCA_FLOWER_KEY_ENC_OPTS_VXLAN:
if (key->enc_opts.dst_opt_type) {
NL_SET_ERR_MSG(extack, "Duplicate type for vxlan options");
return -EINVAL;
}
option_len = 0;
key->enc_opts.dst_opt_type = TUNNEL_VXLAN_OPT;
option_len = fl_set_vxlan_opt(nla_opt_key, key,
key_depth, option_len,
extack);
if (option_len < 0)
return option_len;
key->enc_opts.len += option_len;
/* At the same time we need to parse through the mask
* in order to verify exact and mask attribute lengths.
*/
mask->enc_opts.dst_opt_type = TUNNEL_VXLAN_OPT;
option_len = fl_set_vxlan_opt(nla_opt_msk, mask,
msk_depth, option_len,
extack);
if (option_len < 0)
return option_len;
mask->enc_opts.len += option_len;
if (key->enc_opts.len != mask->enc_opts.len) {
NL_SET_ERR_MSG(extack, "Key and mask miss aligned");
return -EINVAL;
}
break;
case TCA_FLOWER_KEY_ENC_OPTS_ERSPAN:
if (key->enc_opts.dst_opt_type) {
NL_SET_ERR_MSG(extack, "Duplicate type for erspan options");
return -EINVAL;
}
option_len = 0;
key->enc_opts.dst_opt_type = TUNNEL_ERSPAN_OPT;
option_len = fl_set_erspan_opt(nla_opt_key, key,
key_depth, option_len,
extack);
if (option_len < 0)
return option_len;
key->enc_opts.len += option_len;
/* At the same time we need to parse through the mask
* in order to verify exact and mask attribute lengths.
*/
mask->enc_opts.dst_opt_type = TUNNEL_ERSPAN_OPT;
option_len = fl_set_erspan_opt(nla_opt_msk, mask,
msk_depth, option_len,
extack);
if (option_len < 0)
return option_len;
mask->enc_opts.len += option_len;
if (key->enc_opts.len != mask->enc_opts.len) {
NL_SET_ERR_MSG(extack, "Key and mask miss aligned");
return -EINVAL;
}
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
break;
default:
NL_SET_ERR_MSG(extack, "Unknown tunnel option type");
return -EINVAL;
}
if (!msk_depth)
continue;
if (!nla_ok(nla_opt_msk, msk_depth)) {
NL_SET_ERR_MSG(extack, "A mask attribute is invalid");
return -EINVAL;
}
nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
}
return 0;
}
static int fl_validate_ct_state(u16 state, struct nlattr *tb,
struct netlink_ext_ack *extack)
{
if (state && !(state & TCA_FLOWER_KEY_CT_FLAGS_TRACKED)) {
NL_SET_ERR_MSG_ATTR(extack, tb,
"no trk, so no other flag can be set");
return -EINVAL;
}
if (state & TCA_FLOWER_KEY_CT_FLAGS_NEW &&
state & TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED) {
NL_SET_ERR_MSG_ATTR(extack, tb,
"new and est are mutually exclusive");
return -EINVAL;
}
if (state & TCA_FLOWER_KEY_CT_FLAGS_INVALID &&
state & ~(TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
TCA_FLOWER_KEY_CT_FLAGS_INVALID)) {
NL_SET_ERR_MSG_ATTR(extack, tb,
"when inv is set, only trk may be set");
return -EINVAL;
}
if (state & TCA_FLOWER_KEY_CT_FLAGS_NEW &&
state & TCA_FLOWER_KEY_CT_FLAGS_REPLY) {
NL_SET_ERR_MSG_ATTR(extack, tb,
"new and rpl are mutually exclusive");
return -EINVAL;
}
return 0;
}
static int fl_set_key_ct(struct nlattr **tb,
struct flow_dissector_key_ct *key,
struct flow_dissector_key_ct *mask,
struct netlink_ext_ack *extack)
{
if (tb[TCA_FLOWER_KEY_CT_STATE]) {
int err;
if (!IS_ENABLED(CONFIG_NF_CONNTRACK)) {
NL_SET_ERR_MSG(extack, "Conntrack isn't enabled");
return -EOPNOTSUPP;
}
fl_set_key_val(tb, &key->ct_state, TCA_FLOWER_KEY_CT_STATE,
&mask->ct_state, TCA_FLOWER_KEY_CT_STATE_MASK,
sizeof(key->ct_state));
err = fl_validate_ct_state(key->ct_state & mask->ct_state,
tb[TCA_FLOWER_KEY_CT_STATE_MASK],
extack);
if (err)
return err;
}
if (tb[TCA_FLOWER_KEY_CT_ZONE]) {
if (!IS_ENABLED(CONFIG_NF_CONNTRACK_ZONES)) {
NL_SET_ERR_MSG(extack, "Conntrack zones isn't enabled");
return -EOPNOTSUPP;
}
fl_set_key_val(tb, &key->ct_zone, TCA_FLOWER_KEY_CT_ZONE,
&mask->ct_zone, TCA_FLOWER_KEY_CT_ZONE_MASK,
sizeof(key->ct_zone));
}
if (tb[TCA_FLOWER_KEY_CT_MARK]) {
if (!IS_ENABLED(CONFIG_NF_CONNTRACK_MARK)) {
NL_SET_ERR_MSG(extack, "Conntrack mark isn't enabled");
return -EOPNOTSUPP;
}
fl_set_key_val(tb, &key->ct_mark, TCA_FLOWER_KEY_CT_MARK,
&mask->ct_mark, TCA_FLOWER_KEY_CT_MARK_MASK,
sizeof(key->ct_mark));
}
if (tb[TCA_FLOWER_KEY_CT_LABELS]) {
if (!IS_ENABLED(CONFIG_NF_CONNTRACK_LABELS)) {
NL_SET_ERR_MSG(extack, "Conntrack labels aren't enabled");
return -EOPNOTSUPP;
}
fl_set_key_val(tb, key->ct_labels, TCA_FLOWER_KEY_CT_LABELS,
mask->ct_labels, TCA_FLOWER_KEY_CT_LABELS_MASK,
sizeof(key->ct_labels));
}
return 0;
}
static int fl_set_key(struct net *net, struct nlattr **tb,
struct fl_flow_key *key, struct fl_flow_key *mask,
struct netlink_ext_ack *extack)
{
__be16 ethertype;
int ret = 0;
if (tb[TCA_FLOWER_INDEV]) {
int err = tcf_change_indev(net, tb[TCA_FLOWER_INDEV], extack);
if (err < 0)
return err;
key->meta.ingress_ifindex = err;
mask->meta.ingress_ifindex = 0xffffffff;
}
fl_set_key_val(tb, key->eth.dst, TCA_FLOWER_KEY_ETH_DST,
mask->eth.dst, TCA_FLOWER_KEY_ETH_DST_MASK,
sizeof(key->eth.dst));
fl_set_key_val(tb, key->eth.src, TCA_FLOWER_KEY_ETH_SRC,
mask->eth.src, TCA_FLOWER_KEY_ETH_SRC_MASK,
sizeof(key->eth.src));
if (tb[TCA_FLOWER_KEY_ETH_TYPE]) {
ethertype = nla_get_be16(tb[TCA_FLOWER_KEY_ETH_TYPE]);
if (eth_type_vlan(ethertype)) {
fl_set_key_vlan(tb, ethertype, TCA_FLOWER_KEY_VLAN_ID,
TCA_FLOWER_KEY_VLAN_PRIO, &key->vlan,
&mask->vlan);
if (tb[TCA_FLOWER_KEY_VLAN_ETH_TYPE]) {
ethertype = nla_get_be16(tb[TCA_FLOWER_KEY_VLAN_ETH_TYPE]);
if (eth_type_vlan(ethertype)) {
fl_set_key_vlan(tb, ethertype,
TCA_FLOWER_KEY_CVLAN_ID,
TCA_FLOWER_KEY_CVLAN_PRIO,
&key->cvlan, &mask->cvlan);
fl_set_key_val(tb, &key->basic.n_proto,
TCA_FLOWER_KEY_CVLAN_ETH_TYPE,
&mask->basic.n_proto,
TCA_FLOWER_UNSPEC,
sizeof(key->basic.n_proto));
} else {
key->basic.n_proto = ethertype;
mask->basic.n_proto = cpu_to_be16(~0);
}
}
} else {
key->basic.n_proto = ethertype;
mask->basic.n_proto = cpu_to_be16(~0);
}
}
if (key->basic.n_proto == htons(ETH_P_IP) ||
key->basic.n_proto == htons(ETH_P_IPV6)) {
fl_set_key_val(tb, &key->basic.ip_proto, TCA_FLOWER_KEY_IP_PROTO,
&mask->basic.ip_proto, TCA_FLOWER_UNSPEC,
sizeof(key->basic.ip_proto));
fl_set_key_ip(tb, false, &key->ip, &mask->ip);
}
if (tb[TCA_FLOWER_KEY_IPV4_SRC] || tb[TCA_FLOWER_KEY_IPV4_DST]) {
key->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
mask->control.addr_type = ~0;
fl_set_key_val(tb, &key->ipv4.src, TCA_FLOWER_KEY_IPV4_SRC,
&mask->ipv4.src, TCA_FLOWER_KEY_IPV4_SRC_MASK,
sizeof(key->ipv4.src));
fl_set_key_val(tb, &key->ipv4.dst, TCA_FLOWER_KEY_IPV4_DST,
&mask->ipv4.dst, TCA_FLOWER_KEY_IPV4_DST_MASK,
sizeof(key->ipv4.dst));
} else if (tb[TCA_FLOWER_KEY_IPV6_SRC] || tb[TCA_FLOWER_KEY_IPV6_DST]) {
key->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
mask->control.addr_type = ~0;
fl_set_key_val(tb, &key->ipv6.src, TCA_FLOWER_KEY_IPV6_SRC,
&mask->ipv6.src, TCA_FLOWER_KEY_IPV6_SRC_MASK,
sizeof(key->ipv6.src));
fl_set_key_val(tb, &key->ipv6.dst, TCA_FLOWER_KEY_IPV6_DST,
&mask->ipv6.dst, TCA_FLOWER_KEY_IPV6_DST_MASK,
sizeof(key->ipv6.dst));
}
if (key->basic.ip_proto == IPPROTO_TCP) {
fl_set_key_val(tb, &key->tp.src, TCA_FLOWER_KEY_TCP_SRC,
&mask->tp.src, TCA_FLOWER_KEY_TCP_SRC_MASK,
sizeof(key->tp.src));
fl_set_key_val(tb, &key->tp.dst, TCA_FLOWER_KEY_TCP_DST,
&mask->tp.dst, TCA_FLOWER_KEY_TCP_DST_MASK,
sizeof(key->tp.dst));
fl_set_key_val(tb, &key->tcp.flags, TCA_FLOWER_KEY_TCP_FLAGS,
&mask->tcp.flags, TCA_FLOWER_KEY_TCP_FLAGS_MASK,
sizeof(key->tcp.flags));
} else if (key->basic.ip_proto == IPPROTO_UDP) {
fl_set_key_val(tb, &key->tp.src, TCA_FLOWER_KEY_UDP_SRC,
&mask->tp.src, TCA_FLOWER_KEY_UDP_SRC_MASK,
sizeof(key->tp.src));
fl_set_key_val(tb, &key->tp.dst, TCA_FLOWER_KEY_UDP_DST,
&mask->tp.dst, TCA_FLOWER_KEY_UDP_DST_MASK,
sizeof(key->tp.dst));
} else if (key->basic.ip_proto == IPPROTO_SCTP) {
fl_set_key_val(tb, &key->tp.src, TCA_FLOWER_KEY_SCTP_SRC,
&mask->tp.src, TCA_FLOWER_KEY_SCTP_SRC_MASK,
sizeof(key->tp.src));
fl_set_key_val(tb, &key->tp.dst, TCA_FLOWER_KEY_SCTP_DST,
&mask->tp.dst, TCA_FLOWER_KEY_SCTP_DST_MASK,
sizeof(key->tp.dst));
} else if (key->basic.n_proto == htons(ETH_P_IP) &&
key->basic.ip_proto == IPPROTO_ICMP) {
fl_set_key_val(tb, &key->icmp.type, TCA_FLOWER_KEY_ICMPV4_TYPE,
&mask->icmp.type,
TCA_FLOWER_KEY_ICMPV4_TYPE_MASK,
sizeof(key->icmp.type));
fl_set_key_val(tb, &key->icmp.code, TCA_FLOWER_KEY_ICMPV4_CODE,
&mask->icmp.code,
TCA_FLOWER_KEY_ICMPV4_CODE_MASK,
sizeof(key->icmp.code));
} else if (key->basic.n_proto == htons(ETH_P_IPV6) &&
key->basic.ip_proto == IPPROTO_ICMPV6) {
fl_set_key_val(tb, &key->icmp.type, TCA_FLOWER_KEY_ICMPV6_TYPE,
&mask->icmp.type,
TCA_FLOWER_KEY_ICMPV6_TYPE_MASK,
sizeof(key->icmp.type));
fl_set_key_val(tb, &key->icmp.code, TCA_FLOWER_KEY_ICMPV6_CODE,
&mask->icmp.code,
TCA_FLOWER_KEY_ICMPV6_CODE_MASK,
sizeof(key->icmp.code));
} else if (key->basic.n_proto == htons(ETH_P_MPLS_UC) ||
key->basic.n_proto == htons(ETH_P_MPLS_MC)) {
ret = fl_set_key_mpls(tb, &key->mpls, &mask->mpls, extack);
if (ret)
return ret;
} else if (key->basic.n_proto == htons(ETH_P_ARP) ||
key->basic.n_proto == htons(ETH_P_RARP)) {
fl_set_key_val(tb, &key->arp.sip, TCA_FLOWER_KEY_ARP_SIP,
&mask->arp.sip, TCA_FLOWER_KEY_ARP_SIP_MASK,
sizeof(key->arp.sip));
fl_set_key_val(tb, &key->arp.tip, TCA_FLOWER_KEY_ARP_TIP,
&mask->arp.tip, TCA_FLOWER_KEY_ARP_TIP_MASK,
sizeof(key->arp.tip));
fl_set_key_val(tb, &key->arp.op, TCA_FLOWER_KEY_ARP_OP,
&mask->arp.op, TCA_FLOWER_KEY_ARP_OP_MASK,
sizeof(key->arp.op));
fl_set_key_val(tb, key->arp.sha, TCA_FLOWER_KEY_ARP_SHA,
mask->arp.sha, TCA_FLOWER_KEY_ARP_SHA_MASK,
sizeof(key->arp.sha));
fl_set_key_val(tb, key->arp.tha, TCA_FLOWER_KEY_ARP_THA,
mask->arp.tha, TCA_FLOWER_KEY_ARP_THA_MASK,
sizeof(key->arp.tha));
}
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
if (key->basic.ip_proto == IPPROTO_TCP ||
key->basic.ip_proto == IPPROTO_UDP ||
key->basic.ip_proto == IPPROTO_SCTP) {
ret = fl_set_key_port_range(tb, key, mask, extack);
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
if (ret)
return ret;
}
if (tb[TCA_FLOWER_KEY_ENC_IPV4_SRC] ||
tb[TCA_FLOWER_KEY_ENC_IPV4_DST]) {
key->enc_control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
mask->enc_control.addr_type = ~0;
fl_set_key_val(tb, &key->enc_ipv4.src,
TCA_FLOWER_KEY_ENC_IPV4_SRC,
&mask->enc_ipv4.src,
TCA_FLOWER_KEY_ENC_IPV4_SRC_MASK,
sizeof(key->enc_ipv4.src));
fl_set_key_val(tb, &key->enc_ipv4.dst,
TCA_FLOWER_KEY_ENC_IPV4_DST,
&mask->enc_ipv4.dst,
TCA_FLOWER_KEY_ENC_IPV4_DST_MASK,
sizeof(key->enc_ipv4.dst));
}
if (tb[TCA_FLOWER_KEY_ENC_IPV6_SRC] ||
tb[TCA_FLOWER_KEY_ENC_IPV6_DST]) {
key->enc_control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
mask->enc_control.addr_type = ~0;
fl_set_key_val(tb, &key->enc_ipv6.src,
TCA_FLOWER_KEY_ENC_IPV6_SRC,
&mask->enc_ipv6.src,
TCA_FLOWER_KEY_ENC_IPV6_SRC_MASK,
sizeof(key->enc_ipv6.src));
fl_set_key_val(tb, &key->enc_ipv6.dst,
TCA_FLOWER_KEY_ENC_IPV6_DST,
&mask->enc_ipv6.dst,
TCA_FLOWER_KEY_ENC_IPV6_DST_MASK,
sizeof(key->enc_ipv6.dst));
}
fl_set_key_val(tb, &key->enc_key_id.keyid, TCA_FLOWER_KEY_ENC_KEY_ID,
&mask->enc_key_id.keyid, TCA_FLOWER_UNSPEC,
sizeof(key->enc_key_id.keyid));
fl_set_key_val(tb, &key->enc_tp.src, TCA_FLOWER_KEY_ENC_UDP_SRC_PORT,
&mask->enc_tp.src, TCA_FLOWER_KEY_ENC_UDP_SRC_PORT_MASK,
sizeof(key->enc_tp.src));
fl_set_key_val(tb, &key->enc_tp.dst, TCA_FLOWER_KEY_ENC_UDP_DST_PORT,
&mask->enc_tp.dst, TCA_FLOWER_KEY_ENC_UDP_DST_PORT_MASK,
sizeof(key->enc_tp.dst));
fl_set_key_ip(tb, true, &key->enc_ip, &mask->enc_ip);
fl_set_key_val(tb, &key->hash.hash, TCA_FLOWER_KEY_HASH,
&mask->hash.hash, TCA_FLOWER_KEY_HASH_MASK,
sizeof(key->hash.hash));
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
if (tb[TCA_FLOWER_KEY_ENC_OPTS]) {
ret = fl_set_enc_opt(tb, key, mask, extack);
if (ret)
return ret;
}
ret = fl_set_key_ct(tb, &key->ct, &mask->ct, extack);
if (ret)
return ret;
if (tb[TCA_FLOWER_KEY_FLAGS])
ret = fl_set_key_flags(tb, &key->control.flags,
&mask->control.flags, extack);
return ret;
}
static void fl_mask_copy(struct fl_flow_mask *dst,
struct fl_flow_mask *src)
{
const void *psrc = fl_key_get_start(&src->key, src);
void *pdst = fl_key_get_start(&dst->key, src);
memcpy(pdst, psrc, fl_mask_range(src));
dst->range = src->range;
}
static const struct rhashtable_params fl_ht_params = {
.key_offset = offsetof(struct cls_fl_filter, mkey), /* base offset */
.head_offset = offsetof(struct cls_fl_filter, ht_node),
.automatic_shrinking = true,
};
static int fl_init_mask_hashtable(struct fl_flow_mask *mask)
{
mask->filter_ht_params = fl_ht_params;
mask->filter_ht_params.key_len = fl_mask_range(mask);
mask->filter_ht_params.key_offset += mask->range.start;
return rhashtable_init(&mask->ht, &mask->filter_ht_params);
}
#define FL_KEY_MEMBER_OFFSET(member) offsetof(struct fl_flow_key, member)
#define FL_KEY_MEMBER_SIZE(member) sizeof_field(struct fl_flow_key, member)
#define FL_KEY_IS_MASKED(mask, member) \
memchr_inv(((char *)mask) + FL_KEY_MEMBER_OFFSET(member), \
0, FL_KEY_MEMBER_SIZE(member)) \
#define FL_KEY_SET(keys, cnt, id, member) \
do { \
keys[cnt].key_id = id; \
keys[cnt].offset = FL_KEY_MEMBER_OFFSET(member); \
cnt++; \
} while(0);
#define FL_KEY_SET_IF_MASKED(mask, keys, cnt, id, member) \
do { \
if (FL_KEY_IS_MASKED(mask, member)) \
FL_KEY_SET(keys, cnt, id, member); \
} while(0);
static void fl_init_dissector(struct flow_dissector *dissector,
struct fl_flow_key *mask)
{
struct flow_dissector_key keys[FLOW_DISSECTOR_KEY_MAX];
size_t cnt = 0;
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_META, meta);
FL_KEY_SET(keys, cnt, FLOW_DISSECTOR_KEY_CONTROL, control);
FL_KEY_SET(keys, cnt, FLOW_DISSECTOR_KEY_BASIC, basic);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ETH_ADDRS, eth);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_IPV4_ADDRS, ipv4);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_IPV6_ADDRS, ipv6);
cls_flower: Fix the behavior using port ranges with hw-offload The recent commit 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") had added filtering based on port ranges to tc flower. However the commit missed necessary changes in hw-offload code, so the feature gave rise to generating incorrect offloaded flow keys in NIC. One more detailed example is below: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 100-200 action drop With the setup above, an exact match filter with dst_port == 0 will be installed in NIC by hw-offload. IOW, the NIC will have a rule which is equivalent to the following one. $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 0 action drop The behavior was caused by the flow dissector which extracts packet data into the flow key in the tc flower. More specifically, regardless of exact match or specified port ranges, fl_init_dissector() set the FLOW_DISSECTOR_KEY_PORTS flag in struct flow_dissector to extract port numbers from skb in skb_flow_dissect() called by fl_classify(). Note that device drivers received the same struct flow_dissector object as used in skb_flow_dissect(). Thus, offloaded drivers could not identify which of these is used because the FLOW_DISSECTOR_KEY_PORTS flag was set to struct flow_dissector in either case. This patch adds the new FLOW_DISSECTOR_KEY_PORTS_RANGE flag and the new tp_range field in struct fl_flow_key to recognize which filters are applied to offloaded drivers. At this point, when filters based on port ranges passed to drivers, drivers return the EOPNOTSUPP error because they do not support the feature (the newly created FLOW_DISSECTOR_KEY_PORTS_RANGE flag). Fixes: 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-03 18:40:12 +08:00
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_PORTS, tp);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_PORTS_RANGE, tp_range);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_IP, ip);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_TCP, tcp);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ICMP, icmp);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ARP, arp);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_MPLS, mpls);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_VLAN, vlan);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_CVLAN, cvlan);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ENC_KEYID, enc_key_id);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS, enc_ipv4);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS, enc_ipv6);
if (FL_KEY_IS_MASKED(mask, enc_ipv4) ||
FL_KEY_IS_MASKED(mask, enc_ipv6))
FL_KEY_SET(keys, cnt, FLOW_DISSECTOR_KEY_ENC_CONTROL,
enc_control);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ENC_PORTS, enc_tp);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ENC_IP, enc_ip);
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_ENC_OPTS, enc_opts);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_CT, ct);
FL_KEY_SET_IF_MASKED(mask, keys, cnt,
FLOW_DISSECTOR_KEY_HASH, hash);
skb_flow_dissector_init(dissector, keys, cnt);
}
static struct fl_flow_mask *fl_create_new_mask(struct cls_fl_head *head,
struct fl_flow_mask *mask)
{
struct fl_flow_mask *newmask;
int err;
newmask = kzalloc(sizeof(*newmask), GFP_KERNEL);
if (!newmask)
return ERR_PTR(-ENOMEM);
fl_mask_copy(newmask, mask);
cls_flower: Fix the behavior using port ranges with hw-offload The recent commit 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") had added filtering based on port ranges to tc flower. However the commit missed necessary changes in hw-offload code, so the feature gave rise to generating incorrect offloaded flow keys in NIC. One more detailed example is below: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 100-200 action drop With the setup above, an exact match filter with dst_port == 0 will be installed in NIC by hw-offload. IOW, the NIC will have a rule which is equivalent to the following one. $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 0 action drop The behavior was caused by the flow dissector which extracts packet data into the flow key in the tc flower. More specifically, regardless of exact match or specified port ranges, fl_init_dissector() set the FLOW_DISSECTOR_KEY_PORTS flag in struct flow_dissector to extract port numbers from skb in skb_flow_dissect() called by fl_classify(). Note that device drivers received the same struct flow_dissector object as used in skb_flow_dissect(). Thus, offloaded drivers could not identify which of these is used because the FLOW_DISSECTOR_KEY_PORTS flag was set to struct flow_dissector in either case. This patch adds the new FLOW_DISSECTOR_KEY_PORTS_RANGE flag and the new tp_range field in struct fl_flow_key to recognize which filters are applied to offloaded drivers. At this point, when filters based on port ranges passed to drivers, drivers return the EOPNOTSUPP error because they do not support the feature (the newly created FLOW_DISSECTOR_KEY_PORTS_RANGE flag). Fixes: 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-03 18:40:12 +08:00
if ((newmask->key.tp_range.tp_min.dst &&
newmask->key.tp_range.tp_max.dst) ||
(newmask->key.tp_range.tp_min.src &&
newmask->key.tp_range.tp_max.src))
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
newmask->flags |= TCA_FLOWER_MASK_FLAGS_RANGE;
err = fl_init_mask_hashtable(newmask);
if (err)
goto errout_free;
fl_init_dissector(&newmask->dissector, &newmask->key);
INIT_LIST_HEAD_RCU(&newmask->filters);
refcount_set(&newmask->refcnt, 1);
err = rhashtable_replace_fast(&head->ht, &mask->ht_node,
&newmask->ht_node, mask_ht_params);
if (err)
goto errout_destroy;
spin_lock(&head->masks_lock);
list_add_tail_rcu(&newmask->list, &head->masks);
spin_unlock(&head->masks_lock);
return newmask;
errout_destroy:
rhashtable_destroy(&newmask->ht);
errout_free:
kfree(newmask);
return ERR_PTR(err);
}
static int fl_check_assign_mask(struct cls_fl_head *head,
struct cls_fl_filter *fnew,
struct cls_fl_filter *fold,
struct fl_flow_mask *mask)
{
struct fl_flow_mask *newmask;
int ret = 0;
rcu_read_lock();
/* Insert mask as temporary node to prevent concurrent creation of mask
* with same key. Any concurrent lookups with same key will return
* -EAGAIN because mask's refcnt is zero.
*/
fnew->mask = rhashtable_lookup_get_insert_fast(&head->ht,
&mask->ht_node,
mask_ht_params);
if (!fnew->mask) {
rcu_read_unlock();
if (fold) {
ret = -EINVAL;
goto errout_cleanup;
}
newmask = fl_create_new_mask(head, mask);
if (IS_ERR(newmask)) {
ret = PTR_ERR(newmask);
goto errout_cleanup;
}
fnew->mask = newmask;
return 0;
} else if (IS_ERR(fnew->mask)) {
ret = PTR_ERR(fnew->mask);
} else if (fold && fold->mask != fnew->mask) {
ret = -EINVAL;
} else if (!refcount_inc_not_zero(&fnew->mask->refcnt)) {
/* Mask was deleted concurrently, try again */
ret = -EAGAIN;
}
rcu_read_unlock();
return ret;
errout_cleanup:
rhashtable_remove_fast(&head->ht, &mask->ht_node,
mask_ht_params);
return ret;
}
static int fl_set_parms(struct net *net, struct tcf_proto *tp,
struct cls_fl_filter *f, struct fl_flow_mask *mask,
unsigned long base, struct nlattr **tb,
struct nlattr *est,
struct fl_flow_tmplt *tmplt, u32 flags,
struct netlink_ext_ack *extack)
{
int err;
err = tcf_exts_validate(net, tp, tb, est, &f->exts, flags, extack);
if (err < 0)
return err;
if (tb[TCA_FLOWER_CLASSID]) {
f->res.classid = nla_get_u32(tb[TCA_FLOWER_CLASSID]);
if (flags & TCA_ACT_FLAGS_NO_RTNL)
rtnl_lock();
tcf_bind_filter(tp, &f->res, base);
if (flags & TCA_ACT_FLAGS_NO_RTNL)
rtnl_unlock();
}
err = fl_set_key(net, tb, &f->key, &mask->key, extack);
if (err)
return err;
fl_mask_update_range(mask);
fl_set_masked_key(&f->mkey, &f->key, mask);
if (!fl_mask_fits_tmplt(tmplt, mask)) {
NL_SET_ERR_MSG_MOD(extack, "Mask does not fit the template");
return -EINVAL;
}
return 0;
}
net: sched: flower: insert filter to ht before offloading it to hw John reports: Recent refactoring of fl_change aims to use the classifier spinlock to avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer() function was moved to before the lock is taken. This can create problems for drivers if duplicate filters are created (commmon in ovs tc offload due to filters being triggered by user-space matches). Drivers registered for such filters will now receive multiple copies of the same rule, each with a different cookie value. This means that the drivers would need to do a full match field lookup to determine duplicates, repeating work that will happen in flower __fl_lookup(). Currently, drivers do not expect to receive duplicate filters. To fix this, verify that filter with same key is not present in flower classifier hash table and insert the new filter to the flower hash table before offloading it to hardware. Implement helper function fl_ht_insert_unique() to atomically verify/insert a filter. This change makes filter visible to fast path at the beginning of fl_change() function, which means it can no longer be freed directly in case of error. Refactor fl_change() error handling code to deallocate the filter with rcu timeout. Fixes: 620da4860827 ("net: sched: flower: refactor fl_change") Reported-by: John Hurley <john.hurley@netronome.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-06 01:56:26 +08:00
static int fl_ht_insert_unique(struct cls_fl_filter *fnew,
struct cls_fl_filter *fold,
bool *in_ht)
{
struct fl_flow_mask *mask = fnew->mask;
int err;
err = rhashtable_lookup_insert_fast(&mask->ht,
&fnew->ht_node,
mask->filter_ht_params);
net: sched: flower: insert filter to ht before offloading it to hw John reports: Recent refactoring of fl_change aims to use the classifier spinlock to avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer() function was moved to before the lock is taken. This can create problems for drivers if duplicate filters are created (commmon in ovs tc offload due to filters being triggered by user-space matches). Drivers registered for such filters will now receive multiple copies of the same rule, each with a different cookie value. This means that the drivers would need to do a full match field lookup to determine duplicates, repeating work that will happen in flower __fl_lookup(). Currently, drivers do not expect to receive duplicate filters. To fix this, verify that filter with same key is not present in flower classifier hash table and insert the new filter to the flower hash table before offloading it to hardware. Implement helper function fl_ht_insert_unique() to atomically verify/insert a filter. This change makes filter visible to fast path at the beginning of fl_change() function, which means it can no longer be freed directly in case of error. Refactor fl_change() error handling code to deallocate the filter with rcu timeout. Fixes: 620da4860827 ("net: sched: flower: refactor fl_change") Reported-by: John Hurley <john.hurley@netronome.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-06 01:56:26 +08:00
if (err) {
*in_ht = false;
/* It is okay if filter with same key exists when
* overwriting.
*/
return fold && err == -EEXIST ? 0 : err;
}
*in_ht = true;
return 0;
}
static int fl_change(struct net *net, struct sk_buff *in_skb,
struct tcf_proto *tp, unsigned long base,
u32 handle, struct nlattr **tca,
void **arg, u32 flags,
struct netlink_ext_ack *extack)
{
struct cls_fl_head *head = fl_head_dereference(tp);
bool rtnl_held = !(flags & TCA_ACT_FLAGS_NO_RTNL);
struct cls_fl_filter *fold = *arg;
struct cls_fl_filter *fnew;
struct fl_flow_mask *mask;
struct nlattr **tb;
net: sched: flower: insert filter to ht before offloading it to hw John reports: Recent refactoring of fl_change aims to use the classifier spinlock to avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer() function was moved to before the lock is taken. This can create problems for drivers if duplicate filters are created (commmon in ovs tc offload due to filters being triggered by user-space matches). Drivers registered for such filters will now receive multiple copies of the same rule, each with a different cookie value. This means that the drivers would need to do a full match field lookup to determine duplicates, repeating work that will happen in flower __fl_lookup(). Currently, drivers do not expect to receive duplicate filters. To fix this, verify that filter with same key is not present in flower classifier hash table and insert the new filter to the flower hash table before offloading it to hardware. Implement helper function fl_ht_insert_unique() to atomically verify/insert a filter. This change makes filter visible to fast path at the beginning of fl_change() function, which means it can no longer be freed directly in case of error. Refactor fl_change() error handling code to deallocate the filter with rcu timeout. Fixes: 620da4860827 ("net: sched: flower: refactor fl_change") Reported-by: John Hurley <john.hurley@netronome.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-06 01:56:26 +08:00
bool in_ht;
int err;
if (!tca[TCA_OPTIONS]) {
err = -EINVAL;
goto errout_fold;
}
mask = kzalloc(sizeof(struct fl_flow_mask), GFP_KERNEL);
if (!mask) {
err = -ENOBUFS;
goto errout_fold;
}
tb = kcalloc(TCA_FLOWER_MAX + 1, sizeof(struct nlattr *), GFP_KERNEL);
if (!tb) {
err = -ENOBUFS;
goto errout_mask_alloc;
}
netlink: make validation more configurable for future strictness We currently have two levels of strict validation: 1) liberal (default) - undefined (type >= max) & NLA_UNSPEC attributes accepted - attribute length >= expected accepted - garbage at end of message accepted 2) strict (opt-in) - NLA_UNSPEC attributes accepted - attribute length >= expected accepted Split out parsing strictness into four different options: * TRAILING - check that there's no trailing data after parsing attributes (in message or nested) * MAXTYPE - reject attrs > max known type * UNSPEC - reject attributes with NLA_UNSPEC policy entries * STRICT_ATTRS - strictly validate attribute size The default for future things should be *everything*. The current *_strict() is a combination of TRAILING and MAXTYPE, and is renamed to _deprecated_strict(). The current regular parsing has none of this, and is renamed to *_parse_deprecated(). Additionally it allows us to selectively set one of the new flags even on old policies. Notably, the UNSPEC flag could be useful in this case, since it can be arranged (by filling in the policy) to not be an incompatible userspace ABI change, but would then going forward prevent forgetting attribute entries. Similar can apply to the POLICY flag. We end up with the following renames: * nla_parse -> nla_parse_deprecated * nla_parse_strict -> nla_parse_deprecated_strict * nlmsg_parse -> nlmsg_parse_deprecated * nlmsg_parse_strict -> nlmsg_parse_deprecated_strict * nla_parse_nested -> nla_parse_nested_deprecated * nla_validate_nested -> nla_validate_nested_deprecated Using spatch, of course: @@ expression TB, MAX, HEAD, LEN, POL, EXT; @@ -nla_parse(TB, MAX, HEAD, LEN, POL, EXT) +nla_parse_deprecated(TB, MAX, HEAD, LEN, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse_strict(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated_strict(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression TB, MAX, NLA, POL, EXT; @@ -nla_parse_nested(TB, MAX, NLA, POL, EXT) +nla_parse_nested_deprecated(TB, MAX, NLA, POL, EXT) @@ expression START, MAX, POL, EXT; @@ -nla_validate_nested(START, MAX, POL, EXT) +nla_validate_nested_deprecated(START, MAX, POL, EXT) @@ expression NLH, HDRLEN, MAX, POL, EXT; @@ -nlmsg_validate(NLH, HDRLEN, MAX, POL, EXT) +nlmsg_validate_deprecated(NLH, HDRLEN, MAX, POL, EXT) For this patch, don't actually add the strict, non-renamed versions yet so that it breaks compile if I get it wrong. Also, while at it, make nla_validate and nla_parse go down to a common __nla_validate_parse() function to avoid code duplication. Ultimately, this allows us to have very strict validation for every new caller of nla_parse()/nlmsg_parse() etc as re-introduced in the next patch, while existing things will continue to work as is. In effect then, this adds fully strict validation for any new command. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-26 20:07:28 +08:00
err = nla_parse_nested_deprecated(tb, TCA_FLOWER_MAX,
tca[TCA_OPTIONS], fl_policy, NULL);
if (err < 0)
goto errout_tb;
if (fold && handle && fold->handle != handle) {
err = -EINVAL;
goto errout_tb;
}
fnew = kzalloc(sizeof(*fnew), GFP_KERNEL);
if (!fnew) {
err = -ENOBUFS;
goto errout_tb;
}
2019-04-24 14:53:31 +08:00
INIT_LIST_HEAD(&fnew->hw_list);
refcount_set(&fnew->refcnt, 1);
err = tcf_exts_init(&fnew->exts, net, TCA_FLOWER_ACT, 0);
if (err < 0)
goto errout;
if (tb[TCA_FLOWER_FLAGS]) {
fnew->flags = nla_get_u32(tb[TCA_FLOWER_FLAGS]);
if (!tc_flags_valid(fnew->flags)) {
err = -EINVAL;
net: sched: flower: insert new filter to idr after setting its mask When adding new filter to flower classifier, fl_change() inserts it to handle_idr before initializing filter extensions and assigning it a mask. Normally this ordering doesn't matter because all flower classifier ops callbacks assume rtnl lock protection. However, when filter has an action that doesn't have its kernel module loaded, rtnl lock is released before call to request_module(). During this time the filter can be accessed bu concurrent task before its initialization is completed, which can lead to a crash. Example case of NULL pointer dereference in concurrent dump: Task 1 Task 2 tc_new_tfilter() fl_change() idr_alloc_u32(fnew) fl_set_parms() tcf_exts_validate() tcf_action_init() tcf_action_init_1() rtnl_unlock() request_module() ... rtnl_lock() tc_dump_tfilter() tcf_chain_dump() fl_walk() idr_get_next_ul() tcf_node_dump() tcf_fill_node() fl_dump() mask = &f->mask->key; <- NULL ptr rtnl_lock() Extension initialization and mask assignment don't depend on fnew->handle that is allocated by idr_alloc_u32(). Move idr allocation code after action creation and mask assignment in fl_change() to prevent concurrent access to not fully initialized filter when rtnl lock is released to load action module. Fixes: 01683a146999 ("net: sched: refactor flower walk to iterate over idr") Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-03-06 22:22:12 +08:00
goto errout;
}
}
err = fl_set_parms(net, tp, fnew, mask, base, tb, tca[TCA_RATE],
tp->chain->tmplt_priv, flags, extack);
if (err)
net: sched: flower: insert new filter to idr after setting its mask When adding new filter to flower classifier, fl_change() inserts it to handle_idr before initializing filter extensions and assigning it a mask. Normally this ordering doesn't matter because all flower classifier ops callbacks assume rtnl lock protection. However, when filter has an action that doesn't have its kernel module loaded, rtnl lock is released before call to request_module(). During this time the filter can be accessed bu concurrent task before its initialization is completed, which can lead to a crash. Example case of NULL pointer dereference in concurrent dump: Task 1 Task 2 tc_new_tfilter() fl_change() idr_alloc_u32(fnew) fl_set_parms() tcf_exts_validate() tcf_action_init() tcf_action_init_1() rtnl_unlock() request_module() ... rtnl_lock() tc_dump_tfilter() tcf_chain_dump() fl_walk() idr_get_next_ul() tcf_node_dump() tcf_fill_node() fl_dump() mask = &f->mask->key; <- NULL ptr rtnl_lock() Extension initialization and mask assignment don't depend on fnew->handle that is allocated by idr_alloc_u32(). Move idr allocation code after action creation and mask assignment in fl_change() to prevent concurrent access to not fully initialized filter when rtnl lock is released to load action module. Fixes: 01683a146999 ("net: sched: refactor flower walk to iterate over idr") Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-03-06 22:22:12 +08:00
goto errout;
err = fl_check_assign_mask(head, fnew, fold, mask);
if (err)
net: sched: flower: insert new filter to idr after setting its mask When adding new filter to flower classifier, fl_change() inserts it to handle_idr before initializing filter extensions and assigning it a mask. Normally this ordering doesn't matter because all flower classifier ops callbacks assume rtnl lock protection. However, when filter has an action that doesn't have its kernel module loaded, rtnl lock is released before call to request_module(). During this time the filter can be accessed bu concurrent task before its initialization is completed, which can lead to a crash. Example case of NULL pointer dereference in concurrent dump: Task 1 Task 2 tc_new_tfilter() fl_change() idr_alloc_u32(fnew) fl_set_parms() tcf_exts_validate() tcf_action_init() tcf_action_init_1() rtnl_unlock() request_module() ... rtnl_lock() tc_dump_tfilter() tcf_chain_dump() fl_walk() idr_get_next_ul() tcf_node_dump() tcf_fill_node() fl_dump() mask = &f->mask->key; <- NULL ptr rtnl_lock() Extension initialization and mask assignment don't depend on fnew->handle that is allocated by idr_alloc_u32(). Move idr allocation code after action creation and mask assignment in fl_change() to prevent concurrent access to not fully initialized filter when rtnl lock is released to load action module. Fixes: 01683a146999 ("net: sched: refactor flower walk to iterate over idr") Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-03-06 22:22:12 +08:00
goto errout;
net: sched: flower: insert filter to ht before offloading it to hw John reports: Recent refactoring of fl_change aims to use the classifier spinlock to avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer() function was moved to before the lock is taken. This can create problems for drivers if duplicate filters are created (commmon in ovs tc offload due to filters being triggered by user-space matches). Drivers registered for such filters will now receive multiple copies of the same rule, each with a different cookie value. This means that the drivers would need to do a full match field lookup to determine duplicates, repeating work that will happen in flower __fl_lookup(). Currently, drivers do not expect to receive duplicate filters. To fix this, verify that filter with same key is not present in flower classifier hash table and insert the new filter to the flower hash table before offloading it to hardware. Implement helper function fl_ht_insert_unique() to atomically verify/insert a filter. This change makes filter visible to fast path at the beginning of fl_change() function, which means it can no longer be freed directly in case of error. Refactor fl_change() error handling code to deallocate the filter with rcu timeout. Fixes: 620da4860827 ("net: sched: flower: refactor fl_change") Reported-by: John Hurley <john.hurley@netronome.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-06 01:56:26 +08:00
err = fl_ht_insert_unique(fnew, fold, &in_ht);
if (err)
goto errout_mask;
if (!tc_skip_hw(fnew->flags)) {
err = fl_hw_replace_filter(tp, fnew, rtnl_held, extack);
if (err)
net: sched: flower: insert filter to ht before offloading it to hw John reports: Recent refactoring of fl_change aims to use the classifier spinlock to avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer() function was moved to before the lock is taken. This can create problems for drivers if duplicate filters are created (commmon in ovs tc offload due to filters being triggered by user-space matches). Drivers registered for such filters will now receive multiple copies of the same rule, each with a different cookie value. This means that the drivers would need to do a full match field lookup to determine duplicates, repeating work that will happen in flower __fl_lookup(). Currently, drivers do not expect to receive duplicate filters. To fix this, verify that filter with same key is not present in flower classifier hash table and insert the new filter to the flower hash table before offloading it to hardware. Implement helper function fl_ht_insert_unique() to atomically verify/insert a filter. This change makes filter visible to fast path at the beginning of fl_change() function, which means it can no longer be freed directly in case of error. Refactor fl_change() error handling code to deallocate the filter with rcu timeout. Fixes: 620da4860827 ("net: sched: flower: refactor fl_change") Reported-by: John Hurley <john.hurley@netronome.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-06 01:56:26 +08:00
goto errout_ht;
}
if (!tc_in_hw(fnew->flags))
fnew->flags |= TCA_CLS_FLAGS_NOT_IN_HW;
spin_lock(&tp->lock);
/* tp was deleted concurrently. -EAGAIN will cause caller to lookup
* proto again or create new one, if necessary.
*/
if (tp->deleting) {
err = -EAGAIN;
goto errout_hw;
}
if (fold) {
/* Fold filter was deleted concurrently. Retry lookup. */
if (fold->deleted) {
err = -EAGAIN;
goto errout_hw;
}
fnew->handle = handle;
net: sched: flower: insert filter to ht before offloading it to hw John reports: Recent refactoring of fl_change aims to use the classifier spinlock to avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer() function was moved to before the lock is taken. This can create problems for drivers if duplicate filters are created (commmon in ovs tc offload due to filters being triggered by user-space matches). Drivers registered for such filters will now receive multiple copies of the same rule, each with a different cookie value. This means that the drivers would need to do a full match field lookup to determine duplicates, repeating work that will happen in flower __fl_lookup(). Currently, drivers do not expect to receive duplicate filters. To fix this, verify that filter with same key is not present in flower classifier hash table and insert the new filter to the flower hash table before offloading it to hardware. Implement helper function fl_ht_insert_unique() to atomically verify/insert a filter. This change makes filter visible to fast path at the beginning of fl_change() function, which means it can no longer be freed directly in case of error. Refactor fl_change() error handling code to deallocate the filter with rcu timeout. Fixes: 620da4860827 ("net: sched: flower: refactor fl_change") Reported-by: John Hurley <john.hurley@netronome.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-06 01:56:26 +08:00
if (!in_ht) {
struct rhashtable_params params =
fnew->mask->filter_ht_params;
err = rhashtable_insert_fast(&fnew->mask->ht,
&fnew->ht_node,
params);
if (err)
goto errout_hw;
in_ht = true;
}
2019-04-24 14:53:31 +08:00
refcount_inc(&fnew->refcnt);
rhashtable_remove_fast(&fold->mask->ht,
&fold->ht_node,
fold->mask->filter_ht_params);
idr_replace(&head->handle_idr, fnew, fnew->handle);
list_replace_rcu(&fold->list, &fnew->list);
fold->deleted = true;
spin_unlock(&tp->lock);
net: sched: flower: fix filter net reference counting Fix net reference counting in fl_change() and remove redundant call to tcf_exts_get_net() from __fl_delete(). __fl_put() already tries to get net before releasing exts and deallocating a filter, so this code caused flower classifier to obtain net twice per filter that is being deleted. Implementation of __fl_delete() called tcf_exts_get_net() to pass its result as 'async' flag to fl_mask_put(). However, 'async' flag is redundant and only complicates fl_mask_put() implementation. This functionality seems to be copied from filter cleanup code, where it was added by Cong with following explanation: This patchset tries to fix the race between call_rcu() and cleanup_net() again. Without holding the netns refcnt the tc_action_net_exit() in netns workqueue could be called before filter destroy works in tc filter workqueue. This patchset moves the netns refcnt from tc actions to tcf_exts, without breaking per-netns tc actions. This doesn't apply to flower mask, which doesn't call any tc action code during cleanup. Simplify fl_mask_put() by removing the flag parameter and always use tcf_queue_work() to free mask objects. Fixes: 061775583e35 ("net: sched: flower: introduce reference counting for filters") Fixes: 1f17f7742eeb ("net: sched: flower: insert filter to ht before offloading it to hw") Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Reported-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-12 05:54:19 +08:00
fl_mask_put(head, fold->mask);
if (!tc_skip_hw(fold->flags))
fl_hw_destroy_filter(tp, fold, rtnl_held, NULL);
tcf_unbind_filter(tp, &fold->res);
/* Caller holds reference to fold, so refcnt is always > 0
* after this.
*/
refcount_dec(&fold->refcnt);
__fl_put(fold);
} else {
if (handle) {
/* user specifies a handle and it doesn't exist */
err = idr_alloc_u32(&head->handle_idr, fnew, &handle,
handle, GFP_ATOMIC);
/* Filter with specified handle was concurrently
* inserted after initial check in cls_api. This is not
* necessarily an error if NLM_F_EXCL is not set in
* message flags. Returning EAGAIN will cause cls_api to
* try to update concurrently inserted rule.
*/
if (err == -ENOSPC)
err = -EAGAIN;
} else {
handle = 1;
err = idr_alloc_u32(&head->handle_idr, fnew, &handle,
INT_MAX, GFP_ATOMIC);
}
if (err)
goto errout_hw;
2019-04-24 14:53:31 +08:00
refcount_inc(&fnew->refcnt);
fnew->handle = handle;
list_add_tail_rcu(&fnew->list, &fnew->mask->filters);
spin_unlock(&tp->lock);
}
*arg = fnew;
kfree(tb);
tcf_queue_work(&mask->rwork, fl_uninit_mask_free_work);
return 0;
2019-04-24 14:53:31 +08:00
errout_ht:
spin_lock(&tp->lock);
errout_hw:
2019-04-24 14:53:31 +08:00
fnew->deleted = true;
spin_unlock(&tp->lock);
if (!tc_skip_hw(fnew->flags))
fl_hw_destroy_filter(tp, fnew, rtnl_held, NULL);
net: sched: flower: insert filter to ht before offloading it to hw John reports: Recent refactoring of fl_change aims to use the classifier spinlock to avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer() function was moved to before the lock is taken. This can create problems for drivers if duplicate filters are created (commmon in ovs tc offload due to filters being triggered by user-space matches). Drivers registered for such filters will now receive multiple copies of the same rule, each with a different cookie value. This means that the drivers would need to do a full match field lookup to determine duplicates, repeating work that will happen in flower __fl_lookup(). Currently, drivers do not expect to receive duplicate filters. To fix this, verify that filter with same key is not present in flower classifier hash table and insert the new filter to the flower hash table before offloading it to hardware. Implement helper function fl_ht_insert_unique() to atomically verify/insert a filter. This change makes filter visible to fast path at the beginning of fl_change() function, which means it can no longer be freed directly in case of error. Refactor fl_change() error handling code to deallocate the filter with rcu timeout. Fixes: 620da4860827 ("net: sched: flower: refactor fl_change") Reported-by: John Hurley <john.hurley@netronome.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-06 01:56:26 +08:00
if (in_ht)
rhashtable_remove_fast(&fnew->mask->ht, &fnew->ht_node,
fnew->mask->filter_ht_params);
net: sched: flower: insert new filter to idr after setting its mask When adding new filter to flower classifier, fl_change() inserts it to handle_idr before initializing filter extensions and assigning it a mask. Normally this ordering doesn't matter because all flower classifier ops callbacks assume rtnl lock protection. However, when filter has an action that doesn't have its kernel module loaded, rtnl lock is released before call to request_module(). During this time the filter can be accessed bu concurrent task before its initialization is completed, which can lead to a crash. Example case of NULL pointer dereference in concurrent dump: Task 1 Task 2 tc_new_tfilter() fl_change() idr_alloc_u32(fnew) fl_set_parms() tcf_exts_validate() tcf_action_init() tcf_action_init_1() rtnl_unlock() request_module() ... rtnl_lock() tc_dump_tfilter() tcf_chain_dump() fl_walk() idr_get_next_ul() tcf_node_dump() tcf_fill_node() fl_dump() mask = &f->mask->key; <- NULL ptr rtnl_lock() Extension initialization and mask assignment don't depend on fnew->handle that is allocated by idr_alloc_u32(). Move idr allocation code after action creation and mask assignment in fl_change() to prevent concurrent access to not fully initialized filter when rtnl lock is released to load action module. Fixes: 01683a146999 ("net: sched: refactor flower walk to iterate over idr") Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-03-06 22:22:12 +08:00
errout_mask:
net: sched: flower: fix filter net reference counting Fix net reference counting in fl_change() and remove redundant call to tcf_exts_get_net() from __fl_delete(). __fl_put() already tries to get net before releasing exts and deallocating a filter, so this code caused flower classifier to obtain net twice per filter that is being deleted. Implementation of __fl_delete() called tcf_exts_get_net() to pass its result as 'async' flag to fl_mask_put(). However, 'async' flag is redundant and only complicates fl_mask_put() implementation. This functionality seems to be copied from filter cleanup code, where it was added by Cong with following explanation: This patchset tries to fix the race between call_rcu() and cleanup_net() again. Without holding the netns refcnt the tc_action_net_exit() in netns workqueue could be called before filter destroy works in tc filter workqueue. This patchset moves the netns refcnt from tc actions to tcf_exts, without breaking per-netns tc actions. This doesn't apply to flower mask, which doesn't call any tc action code during cleanup. Simplify fl_mask_put() by removing the flag parameter and always use tcf_queue_work() to free mask objects. Fixes: 061775583e35 ("net: sched: flower: introduce reference counting for filters") Fixes: 1f17f7742eeb ("net: sched: flower: insert filter to ht before offloading it to hw") Fixes: 05cd271fd61a ("cls_flower: Support multiple masks per priority") Reported-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-12 05:54:19 +08:00
fl_mask_put(head, fnew->mask);
errout:
2019-04-24 14:53:31 +08:00
__fl_put(fnew);
errout_tb:
kfree(tb);
errout_mask_alloc:
tcf_queue_work(&mask->rwork, fl_uninit_mask_free_work);
errout_fold:
if (fold)
__fl_put(fold);
return err;
}
static int fl_delete(struct tcf_proto *tp, void *arg, bool *last,
bool rtnl_held, struct netlink_ext_ack *extack)
{
struct cls_fl_head *head = fl_head_dereference(tp);
struct cls_fl_filter *f = arg;
bool last_on_mask;
int err = 0;
err = __fl_delete(tp, f, &last_on_mask, rtnl_held, extack);
*last = list_empty(&head->masks);
__fl_put(f);
return err;
}
static void fl_walk(struct tcf_proto *tp, struct tcf_walker *arg,
bool rtnl_held)
{
struct cls_fl_head *head = fl_head_dereference(tp);
unsigned long id = arg->cookie, tmp;
struct cls_fl_filter *f;
arg->count = arg->skip;
net: sched: flower: protect fl_walk() with rcu Patch that refactored fl_walk() to use idr_for_each_entry_continue_ul() also removed rcu protection of individual filters which causes following use-after-free when filter is deleted concurrently. Fix fl_walk() to obtain rcu read lock while iterating and taking the filter reference and temporary release the lock while calling arg->fn() callback that can sleep. KASAN trace: [ 352.773640] ================================================================== [ 352.775041] BUG: KASAN: use-after-free in fl_walk+0x159/0x240 [cls_flower] [ 352.776304] Read of size 4 at addr ffff8881c8251480 by task tc/2987 [ 352.777862] CPU: 3 PID: 2987 Comm: tc Not tainted 5.15.0-rc2+ #2 [ 352.778980] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 [ 352.781022] Call Trace: [ 352.781573] dump_stack_lvl+0x46/0x5a [ 352.782332] print_address_description.constprop.0+0x1f/0x140 [ 352.783400] ? fl_walk+0x159/0x240 [cls_flower] [ 352.784292] ? fl_walk+0x159/0x240 [cls_flower] [ 352.785138] kasan_report.cold+0x83/0xdf [ 352.785851] ? fl_walk+0x159/0x240 [cls_flower] [ 352.786587] kasan_check_range+0x145/0x1a0 [ 352.787337] fl_walk+0x159/0x240 [cls_flower] [ 352.788163] ? fl_put+0x10/0x10 [cls_flower] [ 352.789007] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.790102] tcf_chain_dump+0x231/0x450 [ 352.790878] ? tcf_chain_tp_delete_empty+0x170/0x170 [ 352.791833] ? __might_sleep+0x2e/0xc0 [ 352.792594] ? tfilter_notify+0x170/0x170 [ 352.793400] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.794477] tc_dump_tfilter+0x385/0x4b0 [ 352.795262] ? tc_new_tfilter+0x1180/0x1180 [ 352.796103] ? __mod_node_page_state+0x1f/0xc0 [ 352.796974] ? __build_skb_around+0x10e/0x130 [ 352.797826] netlink_dump+0x2c0/0x560 [ 352.798563] ? netlink_getsockopt+0x430/0x430 [ 352.799433] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.800542] __netlink_dump_start+0x356/0x440 [ 352.801397] rtnetlink_rcv_msg+0x3ff/0x550 [ 352.802190] ? tc_new_tfilter+0x1180/0x1180 [ 352.802872] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.803668] ? tc_new_tfilter+0x1180/0x1180 [ 352.804344] ? _copy_from_iter_nocache+0x800/0x800 [ 352.805202] ? kasan_set_track+0x1c/0x30 [ 352.805900] netlink_rcv_skb+0xc6/0x1f0 [ 352.806587] ? rht_deferred_worker+0x6b0/0x6b0 [ 352.807455] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.808324] ? netlink_ack+0x4d0/0x4d0 [ 352.809086] ? netlink_deliver_tap+0x62/0x3d0 [ 352.809951] netlink_unicast+0x353/0x480 [ 352.810744] ? netlink_attachskb+0x430/0x430 [ 352.811586] ? __alloc_skb+0xd7/0x200 [ 352.812349] netlink_sendmsg+0x396/0x680 [ 352.813132] ? netlink_unicast+0x480/0x480 [ 352.813952] ? __import_iovec+0x192/0x210 [ 352.814759] ? netlink_unicast+0x480/0x480 [ 352.815580] sock_sendmsg+0x6c/0x80 [ 352.816299] ____sys_sendmsg+0x3a5/0x3c0 [ 352.817096] ? kernel_sendmsg+0x30/0x30 [ 352.817873] ? __ia32_sys_recvmmsg+0x150/0x150 [ 352.818753] ___sys_sendmsg+0xd8/0x140 [ 352.819518] ? sendmsg_copy_msghdr+0x110/0x110 [ 352.820402] ? ___sys_recvmsg+0xf4/0x1a0 [ 352.821110] ? __copy_msghdr_from_user+0x260/0x260 [ 352.821934] ? _raw_spin_lock+0x81/0xd0 [ 352.822680] ? __handle_mm_fault+0xef3/0x1b20 [ 352.823549] ? rb_insert_color+0x2a/0x270 [ 352.824373] ? copy_page_range+0x16b0/0x16b0 [ 352.825209] ? perf_event_update_userpage+0x2d0/0x2d0 [ 352.826190] ? __fget_light+0xd9/0xf0 [ 352.826941] __sys_sendmsg+0xb3/0x130 [ 352.827613] ? __sys_sendmsg_sock+0x20/0x20 [ 352.828377] ? do_user_addr_fault+0x2c5/0x8a0 [ 352.829184] ? fpregs_assert_state_consistent+0x52/0x60 [ 352.830001] ? exit_to_user_mode_prepare+0x32/0x160 [ 352.830845] do_syscall_64+0x35/0x80 [ 352.831445] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.832331] RIP: 0033:0x7f7bee973c17 [ 352.833078] Code: 0c 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10 [ 352.836202] RSP: 002b:00007ffcbb368e28 EFLAGS: 00000246 ORIG_RAX: 000000000000002e [ 352.837524] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7bee973c17 [ 352.838715] RDX: 0000000000000000 RSI: 00007ffcbb368e50 RDI: 0000000000000003 [ 352.839838] RBP: 00007ffcbb36d090 R08: 00000000cea96d79 R09: 00007f7beea34a40 [ 352.841021] R10: 00000000004059bb R11: 0000000000000246 R12: 000000000046563f [ 352.842208] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffcbb36d088 [ 352.843784] Allocated by task 2960: [ 352.844451] kasan_save_stack+0x1b/0x40 [ 352.845173] __kasan_kmalloc+0x7c/0x90 [ 352.845873] fl_change+0x282/0x22db [cls_flower] [ 352.846696] tc_new_tfilter+0x6cf/0x1180 [ 352.847493] rtnetlink_rcv_msg+0x471/0x550 [ 352.848323] netlink_rcv_skb+0xc6/0x1f0 [ 352.849097] netlink_unicast+0x353/0x480 [ 352.849886] netlink_sendmsg+0x396/0x680 [ 352.850678] sock_sendmsg+0x6c/0x80 [ 352.851398] ____sys_sendmsg+0x3a5/0x3c0 [ 352.852202] ___sys_sendmsg+0xd8/0x140 [ 352.852967] __sys_sendmsg+0xb3/0x130 [ 352.853718] do_syscall_64+0x35/0x80 [ 352.854457] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.855830] Freed by task 7: [ 352.856421] kasan_save_stack+0x1b/0x40 [ 352.857139] kasan_set_track+0x1c/0x30 [ 352.857854] kasan_set_free_info+0x20/0x30 [ 352.858609] __kasan_slab_free+0xed/0x130 [ 352.859348] kfree+0xa7/0x3c0 [ 352.859951] process_one_work+0x44d/0x780 [ 352.860685] worker_thread+0x2e2/0x7e0 [ 352.861390] kthread+0x1f4/0x220 [ 352.862022] ret_from_fork+0x1f/0x30 [ 352.862955] Last potentially related work creation: [ 352.863758] kasan_save_stack+0x1b/0x40 [ 352.864378] kasan_record_aux_stack+0xab/0xc0 [ 352.865028] insert_work+0x30/0x160 [ 352.865617] __queue_work+0x351/0x670 [ 352.866261] rcu_work_rcufn+0x30/0x40 [ 352.866917] rcu_core+0x3b2/0xdb0 [ 352.867561] __do_softirq+0xf6/0x386 [ 352.868708] Second to last potentially related work creation: [ 352.869779] kasan_save_stack+0x1b/0x40 [ 352.870560] kasan_record_aux_stack+0xab/0xc0 [ 352.871426] call_rcu+0x5f/0x5c0 [ 352.872108] queue_rcu_work+0x44/0x50 [ 352.872855] __fl_put+0x17c/0x240 [cls_flower] [ 352.873733] fl_delete+0xc7/0x100 [cls_flower] [ 352.874607] tc_del_tfilter+0x510/0xb30 [ 352.886085] rtnetlink_rcv_msg+0x471/0x550 [ 352.886875] netlink_rcv_skb+0xc6/0x1f0 [ 352.887636] netlink_unicast+0x353/0x480 [ 352.888285] netlink_sendmsg+0x396/0x680 [ 352.888942] sock_sendmsg+0x6c/0x80 [ 352.889583] ____sys_sendmsg+0x3a5/0x3c0 [ 352.890311] ___sys_sendmsg+0xd8/0x140 [ 352.891019] __sys_sendmsg+0xb3/0x130 [ 352.891716] do_syscall_64+0x35/0x80 [ 352.892395] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.893666] The buggy address belongs to the object at ffff8881c8251000 which belongs to the cache kmalloc-2k of size 2048 [ 352.895696] The buggy address is located 1152 bytes inside of 2048-byte region [ffff8881c8251000, ffff8881c8251800) [ 352.897640] The buggy address belongs to the page: [ 352.898492] page:00000000213bac35 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1c8250 [ 352.900110] head:00000000213bac35 order:3 compound_mapcount:0 compound_pincount:0 [ 352.901541] flags: 0x2ffff800010200(slab|head|node=0|zone=2|lastcpupid=0x1ffff) [ 352.902908] raw: 002ffff800010200 0000000000000000 dead000000000122 ffff888100042f00 [ 352.904391] raw: 0000000000000000 0000000000080008 00000001ffffffff 0000000000000000 [ 352.905861] page dumped because: kasan: bad access detected [ 352.907323] Memory state around the buggy address: [ 352.908218] ffff8881c8251380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.909471] ffff8881c8251400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.910735] >ffff8881c8251480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.912012] ^ [ 352.912642] ffff8881c8251500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.913919] ffff8881c8251580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.915185] ================================================================== Fixes: d39d714969cd ("idr: introduce idr_for_each_entry_continue_ul()") Signed-off-by: Vlad Buslov <vladbu@nvidia.com> Acked-by: Cong Wang <cong.wang@bytedance.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-29 23:08:49 +08:00
rcu_read_lock();
idr_for_each_entry_continue_ul(&head->handle_idr, f, tmp, id) {
/* don't return filters that are being deleted */
if (!refcount_inc_not_zero(&f->refcnt))
continue;
net: sched: flower: protect fl_walk() with rcu Patch that refactored fl_walk() to use idr_for_each_entry_continue_ul() also removed rcu protection of individual filters which causes following use-after-free when filter is deleted concurrently. Fix fl_walk() to obtain rcu read lock while iterating and taking the filter reference and temporary release the lock while calling arg->fn() callback that can sleep. KASAN trace: [ 352.773640] ================================================================== [ 352.775041] BUG: KASAN: use-after-free in fl_walk+0x159/0x240 [cls_flower] [ 352.776304] Read of size 4 at addr ffff8881c8251480 by task tc/2987 [ 352.777862] CPU: 3 PID: 2987 Comm: tc Not tainted 5.15.0-rc2+ #2 [ 352.778980] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 [ 352.781022] Call Trace: [ 352.781573] dump_stack_lvl+0x46/0x5a [ 352.782332] print_address_description.constprop.0+0x1f/0x140 [ 352.783400] ? fl_walk+0x159/0x240 [cls_flower] [ 352.784292] ? fl_walk+0x159/0x240 [cls_flower] [ 352.785138] kasan_report.cold+0x83/0xdf [ 352.785851] ? fl_walk+0x159/0x240 [cls_flower] [ 352.786587] kasan_check_range+0x145/0x1a0 [ 352.787337] fl_walk+0x159/0x240 [cls_flower] [ 352.788163] ? fl_put+0x10/0x10 [cls_flower] [ 352.789007] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.790102] tcf_chain_dump+0x231/0x450 [ 352.790878] ? tcf_chain_tp_delete_empty+0x170/0x170 [ 352.791833] ? __might_sleep+0x2e/0xc0 [ 352.792594] ? tfilter_notify+0x170/0x170 [ 352.793400] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.794477] tc_dump_tfilter+0x385/0x4b0 [ 352.795262] ? tc_new_tfilter+0x1180/0x1180 [ 352.796103] ? __mod_node_page_state+0x1f/0xc0 [ 352.796974] ? __build_skb_around+0x10e/0x130 [ 352.797826] netlink_dump+0x2c0/0x560 [ 352.798563] ? netlink_getsockopt+0x430/0x430 [ 352.799433] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.800542] __netlink_dump_start+0x356/0x440 [ 352.801397] rtnetlink_rcv_msg+0x3ff/0x550 [ 352.802190] ? tc_new_tfilter+0x1180/0x1180 [ 352.802872] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.803668] ? tc_new_tfilter+0x1180/0x1180 [ 352.804344] ? _copy_from_iter_nocache+0x800/0x800 [ 352.805202] ? kasan_set_track+0x1c/0x30 [ 352.805900] netlink_rcv_skb+0xc6/0x1f0 [ 352.806587] ? rht_deferred_worker+0x6b0/0x6b0 [ 352.807455] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.808324] ? netlink_ack+0x4d0/0x4d0 [ 352.809086] ? netlink_deliver_tap+0x62/0x3d0 [ 352.809951] netlink_unicast+0x353/0x480 [ 352.810744] ? netlink_attachskb+0x430/0x430 [ 352.811586] ? __alloc_skb+0xd7/0x200 [ 352.812349] netlink_sendmsg+0x396/0x680 [ 352.813132] ? netlink_unicast+0x480/0x480 [ 352.813952] ? __import_iovec+0x192/0x210 [ 352.814759] ? netlink_unicast+0x480/0x480 [ 352.815580] sock_sendmsg+0x6c/0x80 [ 352.816299] ____sys_sendmsg+0x3a5/0x3c0 [ 352.817096] ? kernel_sendmsg+0x30/0x30 [ 352.817873] ? __ia32_sys_recvmmsg+0x150/0x150 [ 352.818753] ___sys_sendmsg+0xd8/0x140 [ 352.819518] ? sendmsg_copy_msghdr+0x110/0x110 [ 352.820402] ? ___sys_recvmsg+0xf4/0x1a0 [ 352.821110] ? __copy_msghdr_from_user+0x260/0x260 [ 352.821934] ? _raw_spin_lock+0x81/0xd0 [ 352.822680] ? __handle_mm_fault+0xef3/0x1b20 [ 352.823549] ? rb_insert_color+0x2a/0x270 [ 352.824373] ? copy_page_range+0x16b0/0x16b0 [ 352.825209] ? perf_event_update_userpage+0x2d0/0x2d0 [ 352.826190] ? __fget_light+0xd9/0xf0 [ 352.826941] __sys_sendmsg+0xb3/0x130 [ 352.827613] ? __sys_sendmsg_sock+0x20/0x20 [ 352.828377] ? do_user_addr_fault+0x2c5/0x8a0 [ 352.829184] ? fpregs_assert_state_consistent+0x52/0x60 [ 352.830001] ? exit_to_user_mode_prepare+0x32/0x160 [ 352.830845] do_syscall_64+0x35/0x80 [ 352.831445] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.832331] RIP: 0033:0x7f7bee973c17 [ 352.833078] Code: 0c 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10 [ 352.836202] RSP: 002b:00007ffcbb368e28 EFLAGS: 00000246 ORIG_RAX: 000000000000002e [ 352.837524] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7bee973c17 [ 352.838715] RDX: 0000000000000000 RSI: 00007ffcbb368e50 RDI: 0000000000000003 [ 352.839838] RBP: 00007ffcbb36d090 R08: 00000000cea96d79 R09: 00007f7beea34a40 [ 352.841021] R10: 00000000004059bb R11: 0000000000000246 R12: 000000000046563f [ 352.842208] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffcbb36d088 [ 352.843784] Allocated by task 2960: [ 352.844451] kasan_save_stack+0x1b/0x40 [ 352.845173] __kasan_kmalloc+0x7c/0x90 [ 352.845873] fl_change+0x282/0x22db [cls_flower] [ 352.846696] tc_new_tfilter+0x6cf/0x1180 [ 352.847493] rtnetlink_rcv_msg+0x471/0x550 [ 352.848323] netlink_rcv_skb+0xc6/0x1f0 [ 352.849097] netlink_unicast+0x353/0x480 [ 352.849886] netlink_sendmsg+0x396/0x680 [ 352.850678] sock_sendmsg+0x6c/0x80 [ 352.851398] ____sys_sendmsg+0x3a5/0x3c0 [ 352.852202] ___sys_sendmsg+0xd8/0x140 [ 352.852967] __sys_sendmsg+0xb3/0x130 [ 352.853718] do_syscall_64+0x35/0x80 [ 352.854457] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.855830] Freed by task 7: [ 352.856421] kasan_save_stack+0x1b/0x40 [ 352.857139] kasan_set_track+0x1c/0x30 [ 352.857854] kasan_set_free_info+0x20/0x30 [ 352.858609] __kasan_slab_free+0xed/0x130 [ 352.859348] kfree+0xa7/0x3c0 [ 352.859951] process_one_work+0x44d/0x780 [ 352.860685] worker_thread+0x2e2/0x7e0 [ 352.861390] kthread+0x1f4/0x220 [ 352.862022] ret_from_fork+0x1f/0x30 [ 352.862955] Last potentially related work creation: [ 352.863758] kasan_save_stack+0x1b/0x40 [ 352.864378] kasan_record_aux_stack+0xab/0xc0 [ 352.865028] insert_work+0x30/0x160 [ 352.865617] __queue_work+0x351/0x670 [ 352.866261] rcu_work_rcufn+0x30/0x40 [ 352.866917] rcu_core+0x3b2/0xdb0 [ 352.867561] __do_softirq+0xf6/0x386 [ 352.868708] Second to last potentially related work creation: [ 352.869779] kasan_save_stack+0x1b/0x40 [ 352.870560] kasan_record_aux_stack+0xab/0xc0 [ 352.871426] call_rcu+0x5f/0x5c0 [ 352.872108] queue_rcu_work+0x44/0x50 [ 352.872855] __fl_put+0x17c/0x240 [cls_flower] [ 352.873733] fl_delete+0xc7/0x100 [cls_flower] [ 352.874607] tc_del_tfilter+0x510/0xb30 [ 352.886085] rtnetlink_rcv_msg+0x471/0x550 [ 352.886875] netlink_rcv_skb+0xc6/0x1f0 [ 352.887636] netlink_unicast+0x353/0x480 [ 352.888285] netlink_sendmsg+0x396/0x680 [ 352.888942] sock_sendmsg+0x6c/0x80 [ 352.889583] ____sys_sendmsg+0x3a5/0x3c0 [ 352.890311] ___sys_sendmsg+0xd8/0x140 [ 352.891019] __sys_sendmsg+0xb3/0x130 [ 352.891716] do_syscall_64+0x35/0x80 [ 352.892395] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.893666] The buggy address belongs to the object at ffff8881c8251000 which belongs to the cache kmalloc-2k of size 2048 [ 352.895696] The buggy address is located 1152 bytes inside of 2048-byte region [ffff8881c8251000, ffff8881c8251800) [ 352.897640] The buggy address belongs to the page: [ 352.898492] page:00000000213bac35 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1c8250 [ 352.900110] head:00000000213bac35 order:3 compound_mapcount:0 compound_pincount:0 [ 352.901541] flags: 0x2ffff800010200(slab|head|node=0|zone=2|lastcpupid=0x1ffff) [ 352.902908] raw: 002ffff800010200 0000000000000000 dead000000000122 ffff888100042f00 [ 352.904391] raw: 0000000000000000 0000000000080008 00000001ffffffff 0000000000000000 [ 352.905861] page dumped because: kasan: bad access detected [ 352.907323] Memory state around the buggy address: [ 352.908218] ffff8881c8251380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.909471] ffff8881c8251400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.910735] >ffff8881c8251480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.912012] ^ [ 352.912642] ffff8881c8251500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.913919] ffff8881c8251580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.915185] ================================================================== Fixes: d39d714969cd ("idr: introduce idr_for_each_entry_continue_ul()") Signed-off-by: Vlad Buslov <vladbu@nvidia.com> Acked-by: Cong Wang <cong.wang@bytedance.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-29 23:08:49 +08:00
rcu_read_unlock();
if (arg->fn(tp, f, arg) < 0) {
__fl_put(f);
arg->stop = 1;
net: sched: flower: protect fl_walk() with rcu Patch that refactored fl_walk() to use idr_for_each_entry_continue_ul() also removed rcu protection of individual filters which causes following use-after-free when filter is deleted concurrently. Fix fl_walk() to obtain rcu read lock while iterating and taking the filter reference and temporary release the lock while calling arg->fn() callback that can sleep. KASAN trace: [ 352.773640] ================================================================== [ 352.775041] BUG: KASAN: use-after-free in fl_walk+0x159/0x240 [cls_flower] [ 352.776304] Read of size 4 at addr ffff8881c8251480 by task tc/2987 [ 352.777862] CPU: 3 PID: 2987 Comm: tc Not tainted 5.15.0-rc2+ #2 [ 352.778980] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 [ 352.781022] Call Trace: [ 352.781573] dump_stack_lvl+0x46/0x5a [ 352.782332] print_address_description.constprop.0+0x1f/0x140 [ 352.783400] ? fl_walk+0x159/0x240 [cls_flower] [ 352.784292] ? fl_walk+0x159/0x240 [cls_flower] [ 352.785138] kasan_report.cold+0x83/0xdf [ 352.785851] ? fl_walk+0x159/0x240 [cls_flower] [ 352.786587] kasan_check_range+0x145/0x1a0 [ 352.787337] fl_walk+0x159/0x240 [cls_flower] [ 352.788163] ? fl_put+0x10/0x10 [cls_flower] [ 352.789007] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.790102] tcf_chain_dump+0x231/0x450 [ 352.790878] ? tcf_chain_tp_delete_empty+0x170/0x170 [ 352.791833] ? __might_sleep+0x2e/0xc0 [ 352.792594] ? tfilter_notify+0x170/0x170 [ 352.793400] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.794477] tc_dump_tfilter+0x385/0x4b0 [ 352.795262] ? tc_new_tfilter+0x1180/0x1180 [ 352.796103] ? __mod_node_page_state+0x1f/0xc0 [ 352.796974] ? __build_skb_around+0x10e/0x130 [ 352.797826] netlink_dump+0x2c0/0x560 [ 352.798563] ? netlink_getsockopt+0x430/0x430 [ 352.799433] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.800542] __netlink_dump_start+0x356/0x440 [ 352.801397] rtnetlink_rcv_msg+0x3ff/0x550 [ 352.802190] ? tc_new_tfilter+0x1180/0x1180 [ 352.802872] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.803668] ? tc_new_tfilter+0x1180/0x1180 [ 352.804344] ? _copy_from_iter_nocache+0x800/0x800 [ 352.805202] ? kasan_set_track+0x1c/0x30 [ 352.805900] netlink_rcv_skb+0xc6/0x1f0 [ 352.806587] ? rht_deferred_worker+0x6b0/0x6b0 [ 352.807455] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.808324] ? netlink_ack+0x4d0/0x4d0 [ 352.809086] ? netlink_deliver_tap+0x62/0x3d0 [ 352.809951] netlink_unicast+0x353/0x480 [ 352.810744] ? netlink_attachskb+0x430/0x430 [ 352.811586] ? __alloc_skb+0xd7/0x200 [ 352.812349] netlink_sendmsg+0x396/0x680 [ 352.813132] ? netlink_unicast+0x480/0x480 [ 352.813952] ? __import_iovec+0x192/0x210 [ 352.814759] ? netlink_unicast+0x480/0x480 [ 352.815580] sock_sendmsg+0x6c/0x80 [ 352.816299] ____sys_sendmsg+0x3a5/0x3c0 [ 352.817096] ? kernel_sendmsg+0x30/0x30 [ 352.817873] ? __ia32_sys_recvmmsg+0x150/0x150 [ 352.818753] ___sys_sendmsg+0xd8/0x140 [ 352.819518] ? sendmsg_copy_msghdr+0x110/0x110 [ 352.820402] ? ___sys_recvmsg+0xf4/0x1a0 [ 352.821110] ? __copy_msghdr_from_user+0x260/0x260 [ 352.821934] ? _raw_spin_lock+0x81/0xd0 [ 352.822680] ? __handle_mm_fault+0xef3/0x1b20 [ 352.823549] ? rb_insert_color+0x2a/0x270 [ 352.824373] ? copy_page_range+0x16b0/0x16b0 [ 352.825209] ? perf_event_update_userpage+0x2d0/0x2d0 [ 352.826190] ? __fget_light+0xd9/0xf0 [ 352.826941] __sys_sendmsg+0xb3/0x130 [ 352.827613] ? __sys_sendmsg_sock+0x20/0x20 [ 352.828377] ? do_user_addr_fault+0x2c5/0x8a0 [ 352.829184] ? fpregs_assert_state_consistent+0x52/0x60 [ 352.830001] ? exit_to_user_mode_prepare+0x32/0x160 [ 352.830845] do_syscall_64+0x35/0x80 [ 352.831445] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.832331] RIP: 0033:0x7f7bee973c17 [ 352.833078] Code: 0c 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10 [ 352.836202] RSP: 002b:00007ffcbb368e28 EFLAGS: 00000246 ORIG_RAX: 000000000000002e [ 352.837524] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7bee973c17 [ 352.838715] RDX: 0000000000000000 RSI: 00007ffcbb368e50 RDI: 0000000000000003 [ 352.839838] RBP: 00007ffcbb36d090 R08: 00000000cea96d79 R09: 00007f7beea34a40 [ 352.841021] R10: 00000000004059bb R11: 0000000000000246 R12: 000000000046563f [ 352.842208] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffcbb36d088 [ 352.843784] Allocated by task 2960: [ 352.844451] kasan_save_stack+0x1b/0x40 [ 352.845173] __kasan_kmalloc+0x7c/0x90 [ 352.845873] fl_change+0x282/0x22db [cls_flower] [ 352.846696] tc_new_tfilter+0x6cf/0x1180 [ 352.847493] rtnetlink_rcv_msg+0x471/0x550 [ 352.848323] netlink_rcv_skb+0xc6/0x1f0 [ 352.849097] netlink_unicast+0x353/0x480 [ 352.849886] netlink_sendmsg+0x396/0x680 [ 352.850678] sock_sendmsg+0x6c/0x80 [ 352.851398] ____sys_sendmsg+0x3a5/0x3c0 [ 352.852202] ___sys_sendmsg+0xd8/0x140 [ 352.852967] __sys_sendmsg+0xb3/0x130 [ 352.853718] do_syscall_64+0x35/0x80 [ 352.854457] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.855830] Freed by task 7: [ 352.856421] kasan_save_stack+0x1b/0x40 [ 352.857139] kasan_set_track+0x1c/0x30 [ 352.857854] kasan_set_free_info+0x20/0x30 [ 352.858609] __kasan_slab_free+0xed/0x130 [ 352.859348] kfree+0xa7/0x3c0 [ 352.859951] process_one_work+0x44d/0x780 [ 352.860685] worker_thread+0x2e2/0x7e0 [ 352.861390] kthread+0x1f4/0x220 [ 352.862022] ret_from_fork+0x1f/0x30 [ 352.862955] Last potentially related work creation: [ 352.863758] kasan_save_stack+0x1b/0x40 [ 352.864378] kasan_record_aux_stack+0xab/0xc0 [ 352.865028] insert_work+0x30/0x160 [ 352.865617] __queue_work+0x351/0x670 [ 352.866261] rcu_work_rcufn+0x30/0x40 [ 352.866917] rcu_core+0x3b2/0xdb0 [ 352.867561] __do_softirq+0xf6/0x386 [ 352.868708] Second to last potentially related work creation: [ 352.869779] kasan_save_stack+0x1b/0x40 [ 352.870560] kasan_record_aux_stack+0xab/0xc0 [ 352.871426] call_rcu+0x5f/0x5c0 [ 352.872108] queue_rcu_work+0x44/0x50 [ 352.872855] __fl_put+0x17c/0x240 [cls_flower] [ 352.873733] fl_delete+0xc7/0x100 [cls_flower] [ 352.874607] tc_del_tfilter+0x510/0xb30 [ 352.886085] rtnetlink_rcv_msg+0x471/0x550 [ 352.886875] netlink_rcv_skb+0xc6/0x1f0 [ 352.887636] netlink_unicast+0x353/0x480 [ 352.888285] netlink_sendmsg+0x396/0x680 [ 352.888942] sock_sendmsg+0x6c/0x80 [ 352.889583] ____sys_sendmsg+0x3a5/0x3c0 [ 352.890311] ___sys_sendmsg+0xd8/0x140 [ 352.891019] __sys_sendmsg+0xb3/0x130 [ 352.891716] do_syscall_64+0x35/0x80 [ 352.892395] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.893666] The buggy address belongs to the object at ffff8881c8251000 which belongs to the cache kmalloc-2k of size 2048 [ 352.895696] The buggy address is located 1152 bytes inside of 2048-byte region [ffff8881c8251000, ffff8881c8251800) [ 352.897640] The buggy address belongs to the page: [ 352.898492] page:00000000213bac35 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1c8250 [ 352.900110] head:00000000213bac35 order:3 compound_mapcount:0 compound_pincount:0 [ 352.901541] flags: 0x2ffff800010200(slab|head|node=0|zone=2|lastcpupid=0x1ffff) [ 352.902908] raw: 002ffff800010200 0000000000000000 dead000000000122 ffff888100042f00 [ 352.904391] raw: 0000000000000000 0000000000080008 00000001ffffffff 0000000000000000 [ 352.905861] page dumped because: kasan: bad access detected [ 352.907323] Memory state around the buggy address: [ 352.908218] ffff8881c8251380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.909471] ffff8881c8251400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.910735] >ffff8881c8251480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.912012] ^ [ 352.912642] ffff8881c8251500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.913919] ffff8881c8251580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.915185] ================================================================== Fixes: d39d714969cd ("idr: introduce idr_for_each_entry_continue_ul()") Signed-off-by: Vlad Buslov <vladbu@nvidia.com> Acked-by: Cong Wang <cong.wang@bytedance.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-29 23:08:49 +08:00
rcu_read_lock();
break;
}
__fl_put(f);
arg->count++;
net: sched: flower: protect fl_walk() with rcu Patch that refactored fl_walk() to use idr_for_each_entry_continue_ul() also removed rcu protection of individual filters which causes following use-after-free when filter is deleted concurrently. Fix fl_walk() to obtain rcu read lock while iterating and taking the filter reference and temporary release the lock while calling arg->fn() callback that can sleep. KASAN trace: [ 352.773640] ================================================================== [ 352.775041] BUG: KASAN: use-after-free in fl_walk+0x159/0x240 [cls_flower] [ 352.776304] Read of size 4 at addr ffff8881c8251480 by task tc/2987 [ 352.777862] CPU: 3 PID: 2987 Comm: tc Not tainted 5.15.0-rc2+ #2 [ 352.778980] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 [ 352.781022] Call Trace: [ 352.781573] dump_stack_lvl+0x46/0x5a [ 352.782332] print_address_description.constprop.0+0x1f/0x140 [ 352.783400] ? fl_walk+0x159/0x240 [cls_flower] [ 352.784292] ? fl_walk+0x159/0x240 [cls_flower] [ 352.785138] kasan_report.cold+0x83/0xdf [ 352.785851] ? fl_walk+0x159/0x240 [cls_flower] [ 352.786587] kasan_check_range+0x145/0x1a0 [ 352.787337] fl_walk+0x159/0x240 [cls_flower] [ 352.788163] ? fl_put+0x10/0x10 [cls_flower] [ 352.789007] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.790102] tcf_chain_dump+0x231/0x450 [ 352.790878] ? tcf_chain_tp_delete_empty+0x170/0x170 [ 352.791833] ? __might_sleep+0x2e/0xc0 [ 352.792594] ? tfilter_notify+0x170/0x170 [ 352.793400] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.794477] tc_dump_tfilter+0x385/0x4b0 [ 352.795262] ? tc_new_tfilter+0x1180/0x1180 [ 352.796103] ? __mod_node_page_state+0x1f/0xc0 [ 352.796974] ? __build_skb_around+0x10e/0x130 [ 352.797826] netlink_dump+0x2c0/0x560 [ 352.798563] ? netlink_getsockopt+0x430/0x430 [ 352.799433] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.800542] __netlink_dump_start+0x356/0x440 [ 352.801397] rtnetlink_rcv_msg+0x3ff/0x550 [ 352.802190] ? tc_new_tfilter+0x1180/0x1180 [ 352.802872] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.803668] ? tc_new_tfilter+0x1180/0x1180 [ 352.804344] ? _copy_from_iter_nocache+0x800/0x800 [ 352.805202] ? kasan_set_track+0x1c/0x30 [ 352.805900] netlink_rcv_skb+0xc6/0x1f0 [ 352.806587] ? rht_deferred_worker+0x6b0/0x6b0 [ 352.807455] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.808324] ? netlink_ack+0x4d0/0x4d0 [ 352.809086] ? netlink_deliver_tap+0x62/0x3d0 [ 352.809951] netlink_unicast+0x353/0x480 [ 352.810744] ? netlink_attachskb+0x430/0x430 [ 352.811586] ? __alloc_skb+0xd7/0x200 [ 352.812349] netlink_sendmsg+0x396/0x680 [ 352.813132] ? netlink_unicast+0x480/0x480 [ 352.813952] ? __import_iovec+0x192/0x210 [ 352.814759] ? netlink_unicast+0x480/0x480 [ 352.815580] sock_sendmsg+0x6c/0x80 [ 352.816299] ____sys_sendmsg+0x3a5/0x3c0 [ 352.817096] ? kernel_sendmsg+0x30/0x30 [ 352.817873] ? __ia32_sys_recvmmsg+0x150/0x150 [ 352.818753] ___sys_sendmsg+0xd8/0x140 [ 352.819518] ? sendmsg_copy_msghdr+0x110/0x110 [ 352.820402] ? ___sys_recvmsg+0xf4/0x1a0 [ 352.821110] ? __copy_msghdr_from_user+0x260/0x260 [ 352.821934] ? _raw_spin_lock+0x81/0xd0 [ 352.822680] ? __handle_mm_fault+0xef3/0x1b20 [ 352.823549] ? rb_insert_color+0x2a/0x270 [ 352.824373] ? copy_page_range+0x16b0/0x16b0 [ 352.825209] ? perf_event_update_userpage+0x2d0/0x2d0 [ 352.826190] ? __fget_light+0xd9/0xf0 [ 352.826941] __sys_sendmsg+0xb3/0x130 [ 352.827613] ? __sys_sendmsg_sock+0x20/0x20 [ 352.828377] ? do_user_addr_fault+0x2c5/0x8a0 [ 352.829184] ? fpregs_assert_state_consistent+0x52/0x60 [ 352.830001] ? exit_to_user_mode_prepare+0x32/0x160 [ 352.830845] do_syscall_64+0x35/0x80 [ 352.831445] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.832331] RIP: 0033:0x7f7bee973c17 [ 352.833078] Code: 0c 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10 [ 352.836202] RSP: 002b:00007ffcbb368e28 EFLAGS: 00000246 ORIG_RAX: 000000000000002e [ 352.837524] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7bee973c17 [ 352.838715] RDX: 0000000000000000 RSI: 00007ffcbb368e50 RDI: 0000000000000003 [ 352.839838] RBP: 00007ffcbb36d090 R08: 00000000cea96d79 R09: 00007f7beea34a40 [ 352.841021] R10: 00000000004059bb R11: 0000000000000246 R12: 000000000046563f [ 352.842208] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffcbb36d088 [ 352.843784] Allocated by task 2960: [ 352.844451] kasan_save_stack+0x1b/0x40 [ 352.845173] __kasan_kmalloc+0x7c/0x90 [ 352.845873] fl_change+0x282/0x22db [cls_flower] [ 352.846696] tc_new_tfilter+0x6cf/0x1180 [ 352.847493] rtnetlink_rcv_msg+0x471/0x550 [ 352.848323] netlink_rcv_skb+0xc6/0x1f0 [ 352.849097] netlink_unicast+0x353/0x480 [ 352.849886] netlink_sendmsg+0x396/0x680 [ 352.850678] sock_sendmsg+0x6c/0x80 [ 352.851398] ____sys_sendmsg+0x3a5/0x3c0 [ 352.852202] ___sys_sendmsg+0xd8/0x140 [ 352.852967] __sys_sendmsg+0xb3/0x130 [ 352.853718] do_syscall_64+0x35/0x80 [ 352.854457] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.855830] Freed by task 7: [ 352.856421] kasan_save_stack+0x1b/0x40 [ 352.857139] kasan_set_track+0x1c/0x30 [ 352.857854] kasan_set_free_info+0x20/0x30 [ 352.858609] __kasan_slab_free+0xed/0x130 [ 352.859348] kfree+0xa7/0x3c0 [ 352.859951] process_one_work+0x44d/0x780 [ 352.860685] worker_thread+0x2e2/0x7e0 [ 352.861390] kthread+0x1f4/0x220 [ 352.862022] ret_from_fork+0x1f/0x30 [ 352.862955] Last potentially related work creation: [ 352.863758] kasan_save_stack+0x1b/0x40 [ 352.864378] kasan_record_aux_stack+0xab/0xc0 [ 352.865028] insert_work+0x30/0x160 [ 352.865617] __queue_work+0x351/0x670 [ 352.866261] rcu_work_rcufn+0x30/0x40 [ 352.866917] rcu_core+0x3b2/0xdb0 [ 352.867561] __do_softirq+0xf6/0x386 [ 352.868708] Second to last potentially related work creation: [ 352.869779] kasan_save_stack+0x1b/0x40 [ 352.870560] kasan_record_aux_stack+0xab/0xc0 [ 352.871426] call_rcu+0x5f/0x5c0 [ 352.872108] queue_rcu_work+0x44/0x50 [ 352.872855] __fl_put+0x17c/0x240 [cls_flower] [ 352.873733] fl_delete+0xc7/0x100 [cls_flower] [ 352.874607] tc_del_tfilter+0x510/0xb30 [ 352.886085] rtnetlink_rcv_msg+0x471/0x550 [ 352.886875] netlink_rcv_skb+0xc6/0x1f0 [ 352.887636] netlink_unicast+0x353/0x480 [ 352.888285] netlink_sendmsg+0x396/0x680 [ 352.888942] sock_sendmsg+0x6c/0x80 [ 352.889583] ____sys_sendmsg+0x3a5/0x3c0 [ 352.890311] ___sys_sendmsg+0xd8/0x140 [ 352.891019] __sys_sendmsg+0xb3/0x130 [ 352.891716] do_syscall_64+0x35/0x80 [ 352.892395] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.893666] The buggy address belongs to the object at ffff8881c8251000 which belongs to the cache kmalloc-2k of size 2048 [ 352.895696] The buggy address is located 1152 bytes inside of 2048-byte region [ffff8881c8251000, ffff8881c8251800) [ 352.897640] The buggy address belongs to the page: [ 352.898492] page:00000000213bac35 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1c8250 [ 352.900110] head:00000000213bac35 order:3 compound_mapcount:0 compound_pincount:0 [ 352.901541] flags: 0x2ffff800010200(slab|head|node=0|zone=2|lastcpupid=0x1ffff) [ 352.902908] raw: 002ffff800010200 0000000000000000 dead000000000122 ffff888100042f00 [ 352.904391] raw: 0000000000000000 0000000000080008 00000001ffffffff 0000000000000000 [ 352.905861] page dumped because: kasan: bad access detected [ 352.907323] Memory state around the buggy address: [ 352.908218] ffff8881c8251380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.909471] ffff8881c8251400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.910735] >ffff8881c8251480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.912012] ^ [ 352.912642] ffff8881c8251500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.913919] ffff8881c8251580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.915185] ================================================================== Fixes: d39d714969cd ("idr: introduce idr_for_each_entry_continue_ul()") Signed-off-by: Vlad Buslov <vladbu@nvidia.com> Acked-by: Cong Wang <cong.wang@bytedance.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-29 23:08:49 +08:00
rcu_read_lock();
}
net: sched: flower: protect fl_walk() with rcu Patch that refactored fl_walk() to use idr_for_each_entry_continue_ul() also removed rcu protection of individual filters which causes following use-after-free when filter is deleted concurrently. Fix fl_walk() to obtain rcu read lock while iterating and taking the filter reference and temporary release the lock while calling arg->fn() callback that can sleep. KASAN trace: [ 352.773640] ================================================================== [ 352.775041] BUG: KASAN: use-after-free in fl_walk+0x159/0x240 [cls_flower] [ 352.776304] Read of size 4 at addr ffff8881c8251480 by task tc/2987 [ 352.777862] CPU: 3 PID: 2987 Comm: tc Not tainted 5.15.0-rc2+ #2 [ 352.778980] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 [ 352.781022] Call Trace: [ 352.781573] dump_stack_lvl+0x46/0x5a [ 352.782332] print_address_description.constprop.0+0x1f/0x140 [ 352.783400] ? fl_walk+0x159/0x240 [cls_flower] [ 352.784292] ? fl_walk+0x159/0x240 [cls_flower] [ 352.785138] kasan_report.cold+0x83/0xdf [ 352.785851] ? fl_walk+0x159/0x240 [cls_flower] [ 352.786587] kasan_check_range+0x145/0x1a0 [ 352.787337] fl_walk+0x159/0x240 [cls_flower] [ 352.788163] ? fl_put+0x10/0x10 [cls_flower] [ 352.789007] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.790102] tcf_chain_dump+0x231/0x450 [ 352.790878] ? tcf_chain_tp_delete_empty+0x170/0x170 [ 352.791833] ? __might_sleep+0x2e/0xc0 [ 352.792594] ? tfilter_notify+0x170/0x170 [ 352.793400] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.794477] tc_dump_tfilter+0x385/0x4b0 [ 352.795262] ? tc_new_tfilter+0x1180/0x1180 [ 352.796103] ? __mod_node_page_state+0x1f/0xc0 [ 352.796974] ? __build_skb_around+0x10e/0x130 [ 352.797826] netlink_dump+0x2c0/0x560 [ 352.798563] ? netlink_getsockopt+0x430/0x430 [ 352.799433] ? __mutex_unlock_slowpath.constprop.0+0x220/0x220 [ 352.800542] __netlink_dump_start+0x356/0x440 [ 352.801397] rtnetlink_rcv_msg+0x3ff/0x550 [ 352.802190] ? tc_new_tfilter+0x1180/0x1180 [ 352.802872] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.803668] ? tc_new_tfilter+0x1180/0x1180 [ 352.804344] ? _copy_from_iter_nocache+0x800/0x800 [ 352.805202] ? kasan_set_track+0x1c/0x30 [ 352.805900] netlink_rcv_skb+0xc6/0x1f0 [ 352.806587] ? rht_deferred_worker+0x6b0/0x6b0 [ 352.807455] ? rtnl_calcit.isra.0+0x1f0/0x1f0 [ 352.808324] ? netlink_ack+0x4d0/0x4d0 [ 352.809086] ? netlink_deliver_tap+0x62/0x3d0 [ 352.809951] netlink_unicast+0x353/0x480 [ 352.810744] ? netlink_attachskb+0x430/0x430 [ 352.811586] ? __alloc_skb+0xd7/0x200 [ 352.812349] netlink_sendmsg+0x396/0x680 [ 352.813132] ? netlink_unicast+0x480/0x480 [ 352.813952] ? __import_iovec+0x192/0x210 [ 352.814759] ? netlink_unicast+0x480/0x480 [ 352.815580] sock_sendmsg+0x6c/0x80 [ 352.816299] ____sys_sendmsg+0x3a5/0x3c0 [ 352.817096] ? kernel_sendmsg+0x30/0x30 [ 352.817873] ? __ia32_sys_recvmmsg+0x150/0x150 [ 352.818753] ___sys_sendmsg+0xd8/0x140 [ 352.819518] ? sendmsg_copy_msghdr+0x110/0x110 [ 352.820402] ? ___sys_recvmsg+0xf4/0x1a0 [ 352.821110] ? __copy_msghdr_from_user+0x260/0x260 [ 352.821934] ? _raw_spin_lock+0x81/0xd0 [ 352.822680] ? __handle_mm_fault+0xef3/0x1b20 [ 352.823549] ? rb_insert_color+0x2a/0x270 [ 352.824373] ? copy_page_range+0x16b0/0x16b0 [ 352.825209] ? perf_event_update_userpage+0x2d0/0x2d0 [ 352.826190] ? __fget_light+0xd9/0xf0 [ 352.826941] __sys_sendmsg+0xb3/0x130 [ 352.827613] ? __sys_sendmsg_sock+0x20/0x20 [ 352.828377] ? do_user_addr_fault+0x2c5/0x8a0 [ 352.829184] ? fpregs_assert_state_consistent+0x52/0x60 [ 352.830001] ? exit_to_user_mode_prepare+0x32/0x160 [ 352.830845] do_syscall_64+0x35/0x80 [ 352.831445] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.832331] RIP: 0033:0x7f7bee973c17 [ 352.833078] Code: 0c 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10 [ 352.836202] RSP: 002b:00007ffcbb368e28 EFLAGS: 00000246 ORIG_RAX: 000000000000002e [ 352.837524] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7bee973c17 [ 352.838715] RDX: 0000000000000000 RSI: 00007ffcbb368e50 RDI: 0000000000000003 [ 352.839838] RBP: 00007ffcbb36d090 R08: 00000000cea96d79 R09: 00007f7beea34a40 [ 352.841021] R10: 00000000004059bb R11: 0000000000000246 R12: 000000000046563f [ 352.842208] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffcbb36d088 [ 352.843784] Allocated by task 2960: [ 352.844451] kasan_save_stack+0x1b/0x40 [ 352.845173] __kasan_kmalloc+0x7c/0x90 [ 352.845873] fl_change+0x282/0x22db [cls_flower] [ 352.846696] tc_new_tfilter+0x6cf/0x1180 [ 352.847493] rtnetlink_rcv_msg+0x471/0x550 [ 352.848323] netlink_rcv_skb+0xc6/0x1f0 [ 352.849097] netlink_unicast+0x353/0x480 [ 352.849886] netlink_sendmsg+0x396/0x680 [ 352.850678] sock_sendmsg+0x6c/0x80 [ 352.851398] ____sys_sendmsg+0x3a5/0x3c0 [ 352.852202] ___sys_sendmsg+0xd8/0x140 [ 352.852967] __sys_sendmsg+0xb3/0x130 [ 352.853718] do_syscall_64+0x35/0x80 [ 352.854457] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.855830] Freed by task 7: [ 352.856421] kasan_save_stack+0x1b/0x40 [ 352.857139] kasan_set_track+0x1c/0x30 [ 352.857854] kasan_set_free_info+0x20/0x30 [ 352.858609] __kasan_slab_free+0xed/0x130 [ 352.859348] kfree+0xa7/0x3c0 [ 352.859951] process_one_work+0x44d/0x780 [ 352.860685] worker_thread+0x2e2/0x7e0 [ 352.861390] kthread+0x1f4/0x220 [ 352.862022] ret_from_fork+0x1f/0x30 [ 352.862955] Last potentially related work creation: [ 352.863758] kasan_save_stack+0x1b/0x40 [ 352.864378] kasan_record_aux_stack+0xab/0xc0 [ 352.865028] insert_work+0x30/0x160 [ 352.865617] __queue_work+0x351/0x670 [ 352.866261] rcu_work_rcufn+0x30/0x40 [ 352.866917] rcu_core+0x3b2/0xdb0 [ 352.867561] __do_softirq+0xf6/0x386 [ 352.868708] Second to last potentially related work creation: [ 352.869779] kasan_save_stack+0x1b/0x40 [ 352.870560] kasan_record_aux_stack+0xab/0xc0 [ 352.871426] call_rcu+0x5f/0x5c0 [ 352.872108] queue_rcu_work+0x44/0x50 [ 352.872855] __fl_put+0x17c/0x240 [cls_flower] [ 352.873733] fl_delete+0xc7/0x100 [cls_flower] [ 352.874607] tc_del_tfilter+0x510/0xb30 [ 352.886085] rtnetlink_rcv_msg+0x471/0x550 [ 352.886875] netlink_rcv_skb+0xc6/0x1f0 [ 352.887636] netlink_unicast+0x353/0x480 [ 352.888285] netlink_sendmsg+0x396/0x680 [ 352.888942] sock_sendmsg+0x6c/0x80 [ 352.889583] ____sys_sendmsg+0x3a5/0x3c0 [ 352.890311] ___sys_sendmsg+0xd8/0x140 [ 352.891019] __sys_sendmsg+0xb3/0x130 [ 352.891716] do_syscall_64+0x35/0x80 [ 352.892395] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 352.893666] The buggy address belongs to the object at ffff8881c8251000 which belongs to the cache kmalloc-2k of size 2048 [ 352.895696] The buggy address is located 1152 bytes inside of 2048-byte region [ffff8881c8251000, ffff8881c8251800) [ 352.897640] The buggy address belongs to the page: [ 352.898492] page:00000000213bac35 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1c8250 [ 352.900110] head:00000000213bac35 order:3 compound_mapcount:0 compound_pincount:0 [ 352.901541] flags: 0x2ffff800010200(slab|head|node=0|zone=2|lastcpupid=0x1ffff) [ 352.902908] raw: 002ffff800010200 0000000000000000 dead000000000122 ffff888100042f00 [ 352.904391] raw: 0000000000000000 0000000000080008 00000001ffffffff 0000000000000000 [ 352.905861] page dumped because: kasan: bad access detected [ 352.907323] Memory state around the buggy address: [ 352.908218] ffff8881c8251380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.909471] ffff8881c8251400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.910735] >ffff8881c8251480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.912012] ^ [ 352.912642] ffff8881c8251500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.913919] ffff8881c8251580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 352.915185] ================================================================== Fixes: d39d714969cd ("idr: introduce idr_for_each_entry_continue_ul()") Signed-off-by: Vlad Buslov <vladbu@nvidia.com> Acked-by: Cong Wang <cong.wang@bytedance.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-29 23:08:49 +08:00
rcu_read_unlock();
arg->cookie = id;
}
2019-04-24 14:53:31 +08:00
static struct cls_fl_filter *
fl_get_next_hw_filter(struct tcf_proto *tp, struct cls_fl_filter *f, bool add)
{
struct cls_fl_head *head = fl_head_dereference(tp);
spin_lock(&tp->lock);
if (list_empty(&head->hw_filters)) {
spin_unlock(&tp->lock);
return NULL;
}
if (!f)
f = list_entry(&head->hw_filters, struct cls_fl_filter,
hw_list);
list_for_each_entry_continue(f, &head->hw_filters, hw_list) {
if (!(add && f->deleted) && refcount_inc_not_zero(&f->refcnt)) {
spin_unlock(&tp->lock);
return f;
}
}
spin_unlock(&tp->lock);
return NULL;
}
static int fl_reoffload(struct tcf_proto *tp, bool add, flow_setup_cb_t *cb,
void *cb_priv, struct netlink_ext_ack *extack)
{
struct tcf_block *block = tp->chain->block;
struct flow_cls_offload cls_flower = {};
2019-04-24 14:53:31 +08:00
struct cls_fl_filter *f = NULL;
int err;
2019-04-24 14:53:31 +08:00
/* hw_filters list can only be changed by hw offload functions after
* obtaining rtnl lock. Make sure it is not changed while reoffload is
* iterating it.
*/
ASSERT_RTNL();
2019-04-24 14:53:31 +08:00
while ((f = fl_get_next_hw_filter(tp, f, add))) {
cls_flower.rule =
flow_rule_alloc(tcf_exts_num_actions(&f->exts));
if (!cls_flower.rule) {
__fl_put(f);
return -ENOMEM;
}
tc_cls_common_offload_init(&cls_flower.common, tp, f->flags,
extack);
cls_flower.command = add ?
FLOW_CLS_REPLACE : FLOW_CLS_DESTROY;
cls_flower.cookie = (unsigned long)f;
cls_flower.rule->match.dissector = &f->mask->dissector;
cls_flower.rule->match.mask = &f->mask->key;
cls_flower.rule->match.key = &f->mkey;
err = tc_setup_flow_action(&cls_flower.rule->action, &f->exts);
if (err) {
kfree(cls_flower.rule);
if (tc_skip_sw(f->flags)) {
NL_SET_ERR_MSG_MOD(extack, "Failed to setup flow action");
__fl_put(f);
return err;
}
goto next_flow;
}
cls_flower.classid = f->res.classid;
net: sched: refactor block offloads counter usage Without rtnl lock protection filters can no longer safely manage block offloads counter themselves. Refactor cls API to protect block offloadcnt with tcf_block->cb_lock that is already used to protect driver callback list and nooffloaddevcnt counter. The counter can be modified by concurrent tasks by new functions that execute block callbacks (which is safe with previous patch that changed its type to atomic_t), however, block bind/unbind code that checks the counter value takes cb_lock in write mode to exclude any concurrent modifications. This approach prevents race conditions between bind/unbind and callback execution code but allows for concurrency for tc rule update path. Move block offload counter, filter in hardware counter and filter flags management from classifiers into cls hardware offloads API. Make functions tcf_block_offload_{inc|dec}() and tc_cls_offload_cnt_update() to be cls API private. Implement following new cls API to be used instead: tc_setup_cb_add() - non-destructive filter add. If filter that wasn't already in hardware is successfully offloaded, increment block offloads counter, set filter in hardware counter and flag. On failure, previously offloaded filter is considered to be intact and offloads counter is not decremented. tc_setup_cb_replace() - destructive filter replace. Release existing filter block offload counter and reset its in hardware counter and flag. Set new filter in hardware counter and flag. On failure, previously offloaded filter is considered to be destroyed and offload counter is decremented. tc_setup_cb_destroy() - filter destroy. Unconditionally decrement block offloads counter. tc_setup_cb_reoffload() - reoffload filter to single cb. Execute cb() and call tc_cls_offload_cnt_update() if cb() didn't return an error. Refactor all offload-capable classifiers to atomically offload filters to hardware, change block offload counter, and set filter in hardware counter and flag by means of the new cls API functions. Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-26 21:44:59 +08:00
err = tc_setup_cb_reoffload(block, tp, add, cb,
TC_SETUP_CLSFLOWER, &cls_flower,
cb_priv, &f->flags,
&f->in_hw_count);
tc_cleanup_flow_action(&cls_flower.rule->action);
kfree(cls_flower.rule);
if (err) {
net: sched: refactor block offloads counter usage Without rtnl lock protection filters can no longer safely manage block offloads counter themselves. Refactor cls API to protect block offloadcnt with tcf_block->cb_lock that is already used to protect driver callback list and nooffloaddevcnt counter. The counter can be modified by concurrent tasks by new functions that execute block callbacks (which is safe with previous patch that changed its type to atomic_t), however, block bind/unbind code that checks the counter value takes cb_lock in write mode to exclude any concurrent modifications. This approach prevents race conditions between bind/unbind and callback execution code but allows for concurrency for tc rule update path. Move block offload counter, filter in hardware counter and filter flags management from classifiers into cls hardware offloads API. Make functions tcf_block_offload_{inc|dec}() and tc_cls_offload_cnt_update() to be cls API private. Implement following new cls API to be used instead: tc_setup_cb_add() - non-destructive filter add. If filter that wasn't already in hardware is successfully offloaded, increment block offloads counter, set filter in hardware counter and flag. On failure, previously offloaded filter is considered to be intact and offloads counter is not decremented. tc_setup_cb_replace() - destructive filter replace. Release existing filter block offload counter and reset its in hardware counter and flag. Set new filter in hardware counter and flag. On failure, previously offloaded filter is considered to be destroyed and offload counter is decremented. tc_setup_cb_destroy() - filter destroy. Unconditionally decrement block offloads counter. tc_setup_cb_reoffload() - reoffload filter to single cb. Execute cb() and call tc_cls_offload_cnt_update() if cb() didn't return an error. Refactor all offload-capable classifiers to atomically offload filters to hardware, change block offload counter, and set filter in hardware counter and flag by means of the new cls API functions. Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-26 21:44:59 +08:00
__fl_put(f);
return err;
}
next_flow:
__fl_put(f);
}
return 0;
}
static void fl_hw_add(struct tcf_proto *tp, void *type_data)
{
struct flow_cls_offload *cls_flower = type_data;
struct cls_fl_filter *f =
(struct cls_fl_filter *) cls_flower->cookie;
struct cls_fl_head *head = fl_head_dereference(tp);
spin_lock(&tp->lock);
list_add(&f->hw_list, &head->hw_filters);
spin_unlock(&tp->lock);
}
static void fl_hw_del(struct tcf_proto *tp, void *type_data)
{
struct flow_cls_offload *cls_flower = type_data;
struct cls_fl_filter *f =
(struct cls_fl_filter *) cls_flower->cookie;
spin_lock(&tp->lock);
if (!list_empty(&f->hw_list))
list_del_init(&f->hw_list);
spin_unlock(&tp->lock);
}
static int fl_hw_create_tmplt(struct tcf_chain *chain,
struct fl_flow_tmplt *tmplt)
{
struct flow_cls_offload cls_flower = {};
struct tcf_block *block = chain->block;
cls_flower.rule = flow_rule_alloc(0);
if (!cls_flower.rule)
return -ENOMEM;
cls_flower.common.chain_index = chain->index;
cls_flower.command = FLOW_CLS_TMPLT_CREATE;
cls_flower.cookie = (unsigned long) tmplt;
cls_flower.rule->match.dissector = &tmplt->dissector;
cls_flower.rule->match.mask = &tmplt->mask;
cls_flower.rule->match.key = &tmplt->dummy_key;
/* We don't care if driver (any of them) fails to handle this
* call. It serves just as a hint for it.
*/
net: sched: refactor block offloads counter usage Without rtnl lock protection filters can no longer safely manage block offloads counter themselves. Refactor cls API to protect block offloadcnt with tcf_block->cb_lock that is already used to protect driver callback list and nooffloaddevcnt counter. The counter can be modified by concurrent tasks by new functions that execute block callbacks (which is safe with previous patch that changed its type to atomic_t), however, block bind/unbind code that checks the counter value takes cb_lock in write mode to exclude any concurrent modifications. This approach prevents race conditions between bind/unbind and callback execution code but allows for concurrency for tc rule update path. Move block offload counter, filter in hardware counter and filter flags management from classifiers into cls hardware offloads API. Make functions tcf_block_offload_{inc|dec}() and tc_cls_offload_cnt_update() to be cls API private. Implement following new cls API to be used instead: tc_setup_cb_add() - non-destructive filter add. If filter that wasn't already in hardware is successfully offloaded, increment block offloads counter, set filter in hardware counter and flag. On failure, previously offloaded filter is considered to be intact and offloads counter is not decremented. tc_setup_cb_replace() - destructive filter replace. Release existing filter block offload counter and reset its in hardware counter and flag. Set new filter in hardware counter and flag. On failure, previously offloaded filter is considered to be destroyed and offload counter is decremented. tc_setup_cb_destroy() - filter destroy. Unconditionally decrement block offloads counter. tc_setup_cb_reoffload() - reoffload filter to single cb. Execute cb() and call tc_cls_offload_cnt_update() if cb() didn't return an error. Refactor all offload-capable classifiers to atomically offload filters to hardware, change block offload counter, and set filter in hardware counter and flag by means of the new cls API functions. Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-26 21:44:59 +08:00
tc_setup_cb_call(block, TC_SETUP_CLSFLOWER, &cls_flower, false, true);
kfree(cls_flower.rule);
return 0;
}
static void fl_hw_destroy_tmplt(struct tcf_chain *chain,
struct fl_flow_tmplt *tmplt)
{
struct flow_cls_offload cls_flower = {};
struct tcf_block *block = chain->block;
cls_flower.common.chain_index = chain->index;
cls_flower.command = FLOW_CLS_TMPLT_DESTROY;
cls_flower.cookie = (unsigned long) tmplt;
net: sched: refactor block offloads counter usage Without rtnl lock protection filters can no longer safely manage block offloads counter themselves. Refactor cls API to protect block offloadcnt with tcf_block->cb_lock that is already used to protect driver callback list and nooffloaddevcnt counter. The counter can be modified by concurrent tasks by new functions that execute block callbacks (which is safe with previous patch that changed its type to atomic_t), however, block bind/unbind code that checks the counter value takes cb_lock in write mode to exclude any concurrent modifications. This approach prevents race conditions between bind/unbind and callback execution code but allows for concurrency for tc rule update path. Move block offload counter, filter in hardware counter and filter flags management from classifiers into cls hardware offloads API. Make functions tcf_block_offload_{inc|dec}() and tc_cls_offload_cnt_update() to be cls API private. Implement following new cls API to be used instead: tc_setup_cb_add() - non-destructive filter add. If filter that wasn't already in hardware is successfully offloaded, increment block offloads counter, set filter in hardware counter and flag. On failure, previously offloaded filter is considered to be intact and offloads counter is not decremented. tc_setup_cb_replace() - destructive filter replace. Release existing filter block offload counter and reset its in hardware counter and flag. Set new filter in hardware counter and flag. On failure, previously offloaded filter is considered to be destroyed and offload counter is decremented. tc_setup_cb_destroy() - filter destroy. Unconditionally decrement block offloads counter. tc_setup_cb_reoffload() - reoffload filter to single cb. Execute cb() and call tc_cls_offload_cnt_update() if cb() didn't return an error. Refactor all offload-capable classifiers to atomically offload filters to hardware, change block offload counter, and set filter in hardware counter and flag by means of the new cls API functions. Signed-off-by: Vlad Buslov <vladbu@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-26 21:44:59 +08:00
tc_setup_cb_call(block, TC_SETUP_CLSFLOWER, &cls_flower, false, true);
}
static void *fl_tmplt_create(struct net *net, struct tcf_chain *chain,
struct nlattr **tca,
struct netlink_ext_ack *extack)
{
struct fl_flow_tmplt *tmplt;
struct nlattr **tb;
int err;
if (!tca[TCA_OPTIONS])
return ERR_PTR(-EINVAL);
tb = kcalloc(TCA_FLOWER_MAX + 1, sizeof(struct nlattr *), GFP_KERNEL);
if (!tb)
return ERR_PTR(-ENOBUFS);
netlink: make validation more configurable for future strictness We currently have two levels of strict validation: 1) liberal (default) - undefined (type >= max) & NLA_UNSPEC attributes accepted - attribute length >= expected accepted - garbage at end of message accepted 2) strict (opt-in) - NLA_UNSPEC attributes accepted - attribute length >= expected accepted Split out parsing strictness into four different options: * TRAILING - check that there's no trailing data after parsing attributes (in message or nested) * MAXTYPE - reject attrs > max known type * UNSPEC - reject attributes with NLA_UNSPEC policy entries * STRICT_ATTRS - strictly validate attribute size The default for future things should be *everything*. The current *_strict() is a combination of TRAILING and MAXTYPE, and is renamed to _deprecated_strict(). The current regular parsing has none of this, and is renamed to *_parse_deprecated(). Additionally it allows us to selectively set one of the new flags even on old policies. Notably, the UNSPEC flag could be useful in this case, since it can be arranged (by filling in the policy) to not be an incompatible userspace ABI change, but would then going forward prevent forgetting attribute entries. Similar can apply to the POLICY flag. We end up with the following renames: * nla_parse -> nla_parse_deprecated * nla_parse_strict -> nla_parse_deprecated_strict * nlmsg_parse -> nlmsg_parse_deprecated * nlmsg_parse_strict -> nlmsg_parse_deprecated_strict * nla_parse_nested -> nla_parse_nested_deprecated * nla_validate_nested -> nla_validate_nested_deprecated Using spatch, of course: @@ expression TB, MAX, HEAD, LEN, POL, EXT; @@ -nla_parse(TB, MAX, HEAD, LEN, POL, EXT) +nla_parse_deprecated(TB, MAX, HEAD, LEN, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse_strict(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated_strict(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression TB, MAX, NLA, POL, EXT; @@ -nla_parse_nested(TB, MAX, NLA, POL, EXT) +nla_parse_nested_deprecated(TB, MAX, NLA, POL, EXT) @@ expression START, MAX, POL, EXT; @@ -nla_validate_nested(START, MAX, POL, EXT) +nla_validate_nested_deprecated(START, MAX, POL, EXT) @@ expression NLH, HDRLEN, MAX, POL, EXT; @@ -nlmsg_validate(NLH, HDRLEN, MAX, POL, EXT) +nlmsg_validate_deprecated(NLH, HDRLEN, MAX, POL, EXT) For this patch, don't actually add the strict, non-renamed versions yet so that it breaks compile if I get it wrong. Also, while at it, make nla_validate and nla_parse go down to a common __nla_validate_parse() function to avoid code duplication. Ultimately, this allows us to have very strict validation for every new caller of nla_parse()/nlmsg_parse() etc as re-introduced in the next patch, while existing things will continue to work as is. In effect then, this adds fully strict validation for any new command. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-26 20:07:28 +08:00
err = nla_parse_nested_deprecated(tb, TCA_FLOWER_MAX,
tca[TCA_OPTIONS], fl_policy, NULL);
if (err)
goto errout_tb;
tmplt = kzalloc(sizeof(*tmplt), GFP_KERNEL);
if (!tmplt) {
err = -ENOMEM;
goto errout_tb;
}
tmplt->chain = chain;
err = fl_set_key(net, tb, &tmplt->dummy_key, &tmplt->mask, extack);
if (err)
goto errout_tmplt;
fl_init_dissector(&tmplt->dissector, &tmplt->mask);
err = fl_hw_create_tmplt(chain, tmplt);
if (err)
goto errout_tmplt;
kfree(tb);
return tmplt;
errout_tmplt:
kfree(tmplt);
errout_tb:
kfree(tb);
return ERR_PTR(err);
}
static void fl_tmplt_destroy(void *tmplt_priv)
{
struct fl_flow_tmplt *tmplt = tmplt_priv;
fl_hw_destroy_tmplt(tmplt->chain, tmplt);
kfree(tmplt);
}
static int fl_dump_key_val(struct sk_buff *skb,
void *val, int val_type,
void *mask, int mask_type, int len)
{
int err;
if (!memchr_inv(mask, 0, len))
return 0;
err = nla_put(skb, val_type, len, val);
if (err)
return err;
if (mask_type != TCA_FLOWER_UNSPEC) {
err = nla_put(skb, mask_type, len, mask);
if (err)
return err;
}
return 0;
}
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
static int fl_dump_key_port_range(struct sk_buff *skb, struct fl_flow_key *key,
struct fl_flow_key *mask)
{
cls_flower: Fix the behavior using port ranges with hw-offload The recent commit 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") had added filtering based on port ranges to tc flower. However the commit missed necessary changes in hw-offload code, so the feature gave rise to generating incorrect offloaded flow keys in NIC. One more detailed example is below: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 100-200 action drop With the setup above, an exact match filter with dst_port == 0 will be installed in NIC by hw-offload. IOW, the NIC will have a rule which is equivalent to the following one. $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 ingress protocol ip flower ip_proto tcp \ dst_port 0 action drop The behavior was caused by the flow dissector which extracts packet data into the flow key in the tc flower. More specifically, regardless of exact match or specified port ranges, fl_init_dissector() set the FLOW_DISSECTOR_KEY_PORTS flag in struct flow_dissector to extract port numbers from skb in skb_flow_dissect() called by fl_classify(). Note that device drivers received the same struct flow_dissector object as used in skb_flow_dissect(). Thus, offloaded drivers could not identify which of these is used because the FLOW_DISSECTOR_KEY_PORTS flag was set to struct flow_dissector in either case. This patch adds the new FLOW_DISSECTOR_KEY_PORTS_RANGE flag and the new tp_range field in struct fl_flow_key to recognize which filters are applied to offloaded drivers. At this point, when filters based on port ranges passed to drivers, drivers return the EOPNOTSUPP error because they do not support the feature (the newly created FLOW_DISSECTOR_KEY_PORTS_RANGE flag). Fixes: 5c72299fba9d ("net: sched: cls_flower: Classify packets using port ranges") Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-03 18:40:12 +08:00
if (fl_dump_key_val(skb, &key->tp_range.tp_min.dst,
TCA_FLOWER_KEY_PORT_DST_MIN,
&mask->tp_range.tp_min.dst, TCA_FLOWER_UNSPEC,
sizeof(key->tp_range.tp_min.dst)) ||
fl_dump_key_val(skb, &key->tp_range.tp_max.dst,
TCA_FLOWER_KEY_PORT_DST_MAX,
&mask->tp_range.tp_max.dst, TCA_FLOWER_UNSPEC,
sizeof(key->tp_range.tp_max.dst)) ||
fl_dump_key_val(skb, &key->tp_range.tp_min.src,
TCA_FLOWER_KEY_PORT_SRC_MIN,
&mask->tp_range.tp_min.src, TCA_FLOWER_UNSPEC,
sizeof(key->tp_range.tp_min.src)) ||
fl_dump_key_val(skb, &key->tp_range.tp_max.src,
TCA_FLOWER_KEY_PORT_SRC_MAX,
&mask->tp_range.tp_max.src, TCA_FLOWER_UNSPEC,
sizeof(key->tp_range.tp_max.src)))
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
return -1;
return 0;
}
static int fl_dump_key_mpls_opt_lse(struct sk_buff *skb,
struct flow_dissector_key_mpls *mpls_key,
struct flow_dissector_key_mpls *mpls_mask,
u8 lse_index)
{
struct flow_dissector_mpls_lse *lse_mask = &mpls_mask->ls[lse_index];
struct flow_dissector_mpls_lse *lse_key = &mpls_key->ls[lse_index];
int err;
err = nla_put_u8(skb, TCA_FLOWER_KEY_MPLS_OPT_LSE_DEPTH,
lse_index + 1);
if (err)
return err;
if (lse_mask->mpls_ttl) {
err = nla_put_u8(skb, TCA_FLOWER_KEY_MPLS_OPT_LSE_TTL,
lse_key->mpls_ttl);
if (err)
return err;
}
if (lse_mask->mpls_bos) {
err = nla_put_u8(skb, TCA_FLOWER_KEY_MPLS_OPT_LSE_BOS,
lse_key->mpls_bos);
if (err)
return err;
}
if (lse_mask->mpls_tc) {
err = nla_put_u8(skb, TCA_FLOWER_KEY_MPLS_OPT_LSE_TC,
lse_key->mpls_tc);
if (err)
return err;
}
if (lse_mask->mpls_label) {
err = nla_put_u32(skb, TCA_FLOWER_KEY_MPLS_OPT_LSE_LABEL,
lse_key->mpls_label);
if (err)
return err;
}
return 0;
}
static int fl_dump_key_mpls_opts(struct sk_buff *skb,
struct flow_dissector_key_mpls *mpls_key,
struct flow_dissector_key_mpls *mpls_mask)
{
struct nlattr *opts;
struct nlattr *lse;
u8 lse_index;
int err;
opts = nla_nest_start(skb, TCA_FLOWER_KEY_MPLS_OPTS);
if (!opts)
return -EMSGSIZE;
for (lse_index = 0; lse_index < FLOW_DIS_MPLS_MAX; lse_index++) {
if (!(mpls_mask->used_lses & 1 << lse_index))
continue;
lse = nla_nest_start(skb, TCA_FLOWER_KEY_MPLS_OPTS_LSE);
if (!lse) {
err = -EMSGSIZE;
goto err_opts;
}
err = fl_dump_key_mpls_opt_lse(skb, mpls_key, mpls_mask,
lse_index);
if (err)
goto err_opts_lse;
nla_nest_end(skb, lse);
}
nla_nest_end(skb, opts);
return 0;
err_opts_lse:
nla_nest_cancel(skb, lse);
err_opts:
nla_nest_cancel(skb, opts);
return err;
}
static int fl_dump_key_mpls(struct sk_buff *skb,
struct flow_dissector_key_mpls *mpls_key,
struct flow_dissector_key_mpls *mpls_mask)
{
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
struct flow_dissector_mpls_lse *lse_mask;
struct flow_dissector_mpls_lse *lse_key;
int err;
if (!mpls_mask->used_lses)
return 0;
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_mask = &mpls_mask->ls[0];
lse_key = &mpls_key->ls[0];
/* For backward compatibility, don't use the MPLS nested attributes if
* the rule can be expressed using the old attributes.
*/
if (mpls_mask->used_lses & ~1 ||
(!lse_mask->mpls_ttl && !lse_mask->mpls_bos &&
!lse_mask->mpls_tc && !lse_mask->mpls_label))
return fl_dump_key_mpls_opts(skb, mpls_key, mpls_mask);
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
if (lse_mask->mpls_ttl) {
err = nla_put_u8(skb, TCA_FLOWER_KEY_MPLS_TTL,
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_key->mpls_ttl);
if (err)
return err;
}
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
if (lse_mask->mpls_tc) {
err = nla_put_u8(skb, TCA_FLOWER_KEY_MPLS_TC,
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_key->mpls_tc);
if (err)
return err;
}
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
if (lse_mask->mpls_label) {
err = nla_put_u32(skb, TCA_FLOWER_KEY_MPLS_LABEL,
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_key->mpls_label);
if (err)
return err;
}
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
if (lse_mask->mpls_bos) {
err = nla_put_u8(skb, TCA_FLOWER_KEY_MPLS_BOS,
flow_dissector: Parse multiple MPLS Label Stack Entries The current MPLS dissector only parses the first MPLS Label Stack Entry (second LSE can be parsed too, but only to set a key_id). This patch adds the possibility to parse several LSEs by making __skb_flow_dissect_mpls() return FLOW_DISSECT_RET_PROTO_AGAIN as long as the Bottom Of Stack bit hasn't been seen, up to a maximum of FLOW_DIS_MPLS_MAX entries. FLOW_DIS_MPLS_MAX is arbitrarily set to 7. This should be enough for many practical purposes, without wasting too much space. To record the parsed values, flow_dissector_key_mpls is modified to store an array of stack entries, instead of just the values of the first one. A bit field, "used_lses", is also added to keep track of the LSEs that have been set. The objective is to avoid defining a new FLOW_DISSECTOR_KEY_MPLS_XX for each level of the MPLS stack. TC flower is adapted for the new struct flow_dissector_key_mpls layout. Matching on several MPLS Label Stack Entries will be added in the next patch. The NFP and MLX5 drivers are also adapted: nfp_flower_compile_mac() and mlx5's parse_tunnel() now verify that the rule only uses the first LSE and fail if it doesn't. Finally, the behaviour of the FLOW_DISSECTOR_KEY_MPLS_ENTROPY key is slightly modified. Instead of recording the first Entropy Label, it now records the last one. This shouldn't have any consequences since there doesn't seem to have any user of FLOW_DISSECTOR_KEY_MPLS_ENTROPY in the tree. We'd probably better do a hash of all parsed MPLS labels instead (excluding reserved labels) anyway. That'd give better entropy and would probably also simplify the code. But that's not the purpose of this patch, so I'm keeping that as a future possible improvement. Signed-off-by: Guillaume Nault <gnault@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2020-05-26 20:29:00 +08:00
lse_key->mpls_bos);
if (err)
return err;
}
return 0;
}
static int fl_dump_key_ip(struct sk_buff *skb, bool encap,
struct flow_dissector_key_ip *key,
struct flow_dissector_key_ip *mask)
{
int tos_key = encap ? TCA_FLOWER_KEY_ENC_IP_TOS : TCA_FLOWER_KEY_IP_TOS;
int ttl_key = encap ? TCA_FLOWER_KEY_ENC_IP_TTL : TCA_FLOWER_KEY_IP_TTL;
int tos_mask = encap ? TCA_FLOWER_KEY_ENC_IP_TOS_MASK : TCA_FLOWER_KEY_IP_TOS_MASK;
int ttl_mask = encap ? TCA_FLOWER_KEY_ENC_IP_TTL_MASK : TCA_FLOWER_KEY_IP_TTL_MASK;
if (fl_dump_key_val(skb, &key->tos, tos_key, &mask->tos, tos_mask, sizeof(key->tos)) ||
fl_dump_key_val(skb, &key->ttl, ttl_key, &mask->ttl, ttl_mask, sizeof(key->ttl)))
return -1;
return 0;
}
static int fl_dump_key_vlan(struct sk_buff *skb,
int vlan_id_key, int vlan_prio_key,
struct flow_dissector_key_vlan *vlan_key,
struct flow_dissector_key_vlan *vlan_mask)
{
int err;
if (!memchr_inv(vlan_mask, 0, sizeof(*vlan_mask)))
return 0;
if (vlan_mask->vlan_id) {
err = nla_put_u16(skb, vlan_id_key,
vlan_key->vlan_id);
if (err)
return err;
}
if (vlan_mask->vlan_priority) {
err = nla_put_u8(skb, vlan_prio_key,
vlan_key->vlan_priority);
if (err)
return err;
}
return 0;
}
static void fl_get_key_flag(u32 dissector_key, u32 dissector_mask,
u32 *flower_key, u32 *flower_mask,
u32 flower_flag_bit, u32 dissector_flag_bit)
{
if (dissector_mask & dissector_flag_bit) {
*flower_mask |= flower_flag_bit;
if (dissector_key & dissector_flag_bit)
*flower_key |= flower_flag_bit;
}
}
static int fl_dump_key_flags(struct sk_buff *skb, u32 flags_key, u32 flags_mask)
{
u32 key, mask;
__be32 _key, _mask;
int err;
if (!memchr_inv(&flags_mask, 0, sizeof(flags_mask)))
return 0;
key = 0;
mask = 0;
fl_get_key_flag(flags_key, flags_mask, &key, &mask,
TCA_FLOWER_KEY_FLAGS_IS_FRAGMENT, FLOW_DIS_IS_FRAGMENT);
fl_get_key_flag(flags_key, flags_mask, &key, &mask,
TCA_FLOWER_KEY_FLAGS_FRAG_IS_FIRST,
FLOW_DIS_FIRST_FRAG);
_key = cpu_to_be32(key);
_mask = cpu_to_be32(mask);
err = nla_put(skb, TCA_FLOWER_KEY_FLAGS, 4, &_key);
if (err)
return err;
return nla_put(skb, TCA_FLOWER_KEY_FLAGS_MASK, 4, &_mask);
}
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
static int fl_dump_key_geneve_opt(struct sk_buff *skb,
struct flow_dissector_key_enc_opts *enc_opts)
{
struct geneve_opt *opt;
struct nlattr *nest;
int opt_off = 0;
nest = nla_nest_start_noflag(skb, TCA_FLOWER_KEY_ENC_OPTS_GENEVE);
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
if (!nest)
goto nla_put_failure;
while (enc_opts->len > opt_off) {
opt = (struct geneve_opt *)&enc_opts->data[opt_off];
if (nla_put_be16(skb, TCA_FLOWER_KEY_ENC_OPT_GENEVE_CLASS,
opt->opt_class))
goto nla_put_failure;
if (nla_put_u8(skb, TCA_FLOWER_KEY_ENC_OPT_GENEVE_TYPE,
opt->type))
goto nla_put_failure;
if (nla_put(skb, TCA_FLOWER_KEY_ENC_OPT_GENEVE_DATA,
opt->length * 4, opt->opt_data))
goto nla_put_failure;
opt_off += sizeof(struct geneve_opt) + opt->length * 4;
}
nla_nest_end(skb, nest);
return 0;
nla_put_failure:
nla_nest_cancel(skb, nest);
return -EMSGSIZE;
}
static int fl_dump_key_vxlan_opt(struct sk_buff *skb,
struct flow_dissector_key_enc_opts *enc_opts)
{
struct vxlan_metadata *md;
struct nlattr *nest;
nest = nla_nest_start_noflag(skb, TCA_FLOWER_KEY_ENC_OPTS_VXLAN);
if (!nest)
goto nla_put_failure;
md = (struct vxlan_metadata *)&enc_opts->data[0];
if (nla_put_u32(skb, TCA_FLOWER_KEY_ENC_OPT_VXLAN_GBP, md->gbp))
goto nla_put_failure;
nla_nest_end(skb, nest);
return 0;
nla_put_failure:
nla_nest_cancel(skb, nest);
return -EMSGSIZE;
}
static int fl_dump_key_erspan_opt(struct sk_buff *skb,
struct flow_dissector_key_enc_opts *enc_opts)
{
struct erspan_metadata *md;
struct nlattr *nest;
nest = nla_nest_start_noflag(skb, TCA_FLOWER_KEY_ENC_OPTS_ERSPAN);
if (!nest)
goto nla_put_failure;
md = (struct erspan_metadata *)&enc_opts->data[0];
if (nla_put_u8(skb, TCA_FLOWER_KEY_ENC_OPT_ERSPAN_VER, md->version))
goto nla_put_failure;
if (md->version == 1 &&
nla_put_be32(skb, TCA_FLOWER_KEY_ENC_OPT_ERSPAN_INDEX, md->u.index))
goto nla_put_failure;
if (md->version == 2 &&
(nla_put_u8(skb, TCA_FLOWER_KEY_ENC_OPT_ERSPAN_DIR,
md->u.md2.dir) ||
nla_put_u8(skb, TCA_FLOWER_KEY_ENC_OPT_ERSPAN_HWID,
get_hwid(&md->u.md2))))
goto nla_put_failure;
nla_nest_end(skb, nest);
return 0;
nla_put_failure:
nla_nest_cancel(skb, nest);
return -EMSGSIZE;
}
static int fl_dump_key_ct(struct sk_buff *skb,
struct flow_dissector_key_ct *key,
struct flow_dissector_key_ct *mask)
{
if (IS_ENABLED(CONFIG_NF_CONNTRACK) &&
fl_dump_key_val(skb, &key->ct_state, TCA_FLOWER_KEY_CT_STATE,
&mask->ct_state, TCA_FLOWER_KEY_CT_STATE_MASK,
sizeof(key->ct_state)))
goto nla_put_failure;
if (IS_ENABLED(CONFIG_NF_CONNTRACK_ZONES) &&
fl_dump_key_val(skb, &key->ct_zone, TCA_FLOWER_KEY_CT_ZONE,
&mask->ct_zone, TCA_FLOWER_KEY_CT_ZONE_MASK,
sizeof(key->ct_zone)))
goto nla_put_failure;
if (IS_ENABLED(CONFIG_NF_CONNTRACK_MARK) &&
fl_dump_key_val(skb, &key->ct_mark, TCA_FLOWER_KEY_CT_MARK,
&mask->ct_mark, TCA_FLOWER_KEY_CT_MARK_MASK,
sizeof(key->ct_mark)))
goto nla_put_failure;
if (IS_ENABLED(CONFIG_NF_CONNTRACK_LABELS) &&
fl_dump_key_val(skb, &key->ct_labels, TCA_FLOWER_KEY_CT_LABELS,
&mask->ct_labels, TCA_FLOWER_KEY_CT_LABELS_MASK,
sizeof(key->ct_labels)))
goto nla_put_failure;
return 0;
nla_put_failure:
return -EMSGSIZE;
}
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
static int fl_dump_key_options(struct sk_buff *skb, int enc_opt_type,
struct flow_dissector_key_enc_opts *enc_opts)
{
struct nlattr *nest;
int err;
if (!enc_opts->len)
return 0;
nest = nla_nest_start_noflag(skb, enc_opt_type);
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
if (!nest)
goto nla_put_failure;
switch (enc_opts->dst_opt_type) {
case TUNNEL_GENEVE_OPT:
err = fl_dump_key_geneve_opt(skb, enc_opts);
if (err)
goto nla_put_failure;
break;
case TUNNEL_VXLAN_OPT:
err = fl_dump_key_vxlan_opt(skb, enc_opts);
if (err)
goto nla_put_failure;
break;
case TUNNEL_ERSPAN_OPT:
err = fl_dump_key_erspan_opt(skb, enc_opts);
if (err)
goto nla_put_failure;
break;
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
default:
goto nla_put_failure;
}
nla_nest_end(skb, nest);
return 0;
nla_put_failure:
nla_nest_cancel(skb, nest);
return -EMSGSIZE;
}
static int fl_dump_key_enc_opt(struct sk_buff *skb,
struct flow_dissector_key_enc_opts *key_opts,
struct flow_dissector_key_enc_opts *msk_opts)
{
int err;
err = fl_dump_key_options(skb, TCA_FLOWER_KEY_ENC_OPTS, key_opts);
if (err)
return err;
return fl_dump_key_options(skb, TCA_FLOWER_KEY_ENC_OPTS_MASK, msk_opts);
}
static int fl_dump_key(struct sk_buff *skb, struct net *net,
struct fl_flow_key *key, struct fl_flow_key *mask)
{
if (mask->meta.ingress_ifindex) {
struct net_device *dev;
dev = __dev_get_by_index(net, key->meta.ingress_ifindex);
if (dev && nla_put_string(skb, TCA_FLOWER_INDEV, dev->name))
goto nla_put_failure;
}
if (fl_dump_key_val(skb, key->eth.dst, TCA_FLOWER_KEY_ETH_DST,
mask->eth.dst, TCA_FLOWER_KEY_ETH_DST_MASK,
sizeof(key->eth.dst)) ||
fl_dump_key_val(skb, key->eth.src, TCA_FLOWER_KEY_ETH_SRC,
mask->eth.src, TCA_FLOWER_KEY_ETH_SRC_MASK,
sizeof(key->eth.src)) ||
fl_dump_key_val(skb, &key->basic.n_proto, TCA_FLOWER_KEY_ETH_TYPE,
&mask->basic.n_proto, TCA_FLOWER_UNSPEC,
sizeof(key->basic.n_proto)))
goto nla_put_failure;
if (fl_dump_key_mpls(skb, &key->mpls, &mask->mpls))
goto nla_put_failure;
if (fl_dump_key_vlan(skb, TCA_FLOWER_KEY_VLAN_ID,
TCA_FLOWER_KEY_VLAN_PRIO, &key->vlan, &mask->vlan))
goto nla_put_failure;
if (fl_dump_key_vlan(skb, TCA_FLOWER_KEY_CVLAN_ID,
TCA_FLOWER_KEY_CVLAN_PRIO,
&key->cvlan, &mask->cvlan) ||
(mask->cvlan.vlan_tpid &&
nla_put_be16(skb, TCA_FLOWER_KEY_VLAN_ETH_TYPE,
key->cvlan.vlan_tpid)))
goto nla_put_failure;
if (mask->basic.n_proto) {
if (mask->cvlan.vlan_tpid) {
if (nla_put_be16(skb, TCA_FLOWER_KEY_CVLAN_ETH_TYPE,
key->basic.n_proto))
goto nla_put_failure;
} else if (mask->vlan.vlan_tpid) {
if (nla_put_be16(skb, TCA_FLOWER_KEY_VLAN_ETH_TYPE,
key->basic.n_proto))
goto nla_put_failure;
}
}
if ((key->basic.n_proto == htons(ETH_P_IP) ||
key->basic.n_proto == htons(ETH_P_IPV6)) &&
(fl_dump_key_val(skb, &key->basic.ip_proto, TCA_FLOWER_KEY_IP_PROTO,
&mask->basic.ip_proto, TCA_FLOWER_UNSPEC,
sizeof(key->basic.ip_proto)) ||
fl_dump_key_ip(skb, false, &key->ip, &mask->ip)))
goto nla_put_failure;
if (key->control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS &&
(fl_dump_key_val(skb, &key->ipv4.src, TCA_FLOWER_KEY_IPV4_SRC,
&mask->ipv4.src, TCA_FLOWER_KEY_IPV4_SRC_MASK,
sizeof(key->ipv4.src)) ||
fl_dump_key_val(skb, &key->ipv4.dst, TCA_FLOWER_KEY_IPV4_DST,
&mask->ipv4.dst, TCA_FLOWER_KEY_IPV4_DST_MASK,
sizeof(key->ipv4.dst))))
goto nla_put_failure;
else if (key->control.addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS &&
(fl_dump_key_val(skb, &key->ipv6.src, TCA_FLOWER_KEY_IPV6_SRC,
&mask->ipv6.src, TCA_FLOWER_KEY_IPV6_SRC_MASK,
sizeof(key->ipv6.src)) ||
fl_dump_key_val(skb, &key->ipv6.dst, TCA_FLOWER_KEY_IPV6_DST,
&mask->ipv6.dst, TCA_FLOWER_KEY_IPV6_DST_MASK,
sizeof(key->ipv6.dst))))
goto nla_put_failure;
if (key->basic.ip_proto == IPPROTO_TCP &&
(fl_dump_key_val(skb, &key->tp.src, TCA_FLOWER_KEY_TCP_SRC,
&mask->tp.src, TCA_FLOWER_KEY_TCP_SRC_MASK,
sizeof(key->tp.src)) ||
fl_dump_key_val(skb, &key->tp.dst, TCA_FLOWER_KEY_TCP_DST,
&mask->tp.dst, TCA_FLOWER_KEY_TCP_DST_MASK,
sizeof(key->tp.dst)) ||
fl_dump_key_val(skb, &key->tcp.flags, TCA_FLOWER_KEY_TCP_FLAGS,
&mask->tcp.flags, TCA_FLOWER_KEY_TCP_FLAGS_MASK,
sizeof(key->tcp.flags))))
goto nla_put_failure;
else if (key->basic.ip_proto == IPPROTO_UDP &&
(fl_dump_key_val(skb, &key->tp.src, TCA_FLOWER_KEY_UDP_SRC,
&mask->tp.src, TCA_FLOWER_KEY_UDP_SRC_MASK,
sizeof(key->tp.src)) ||
fl_dump_key_val(skb, &key->tp.dst, TCA_FLOWER_KEY_UDP_DST,
&mask->tp.dst, TCA_FLOWER_KEY_UDP_DST_MASK,
sizeof(key->tp.dst))))
goto nla_put_failure;
else if (key->basic.ip_proto == IPPROTO_SCTP &&
(fl_dump_key_val(skb, &key->tp.src, TCA_FLOWER_KEY_SCTP_SRC,
&mask->tp.src, TCA_FLOWER_KEY_SCTP_SRC_MASK,
sizeof(key->tp.src)) ||
fl_dump_key_val(skb, &key->tp.dst, TCA_FLOWER_KEY_SCTP_DST,
&mask->tp.dst, TCA_FLOWER_KEY_SCTP_DST_MASK,
sizeof(key->tp.dst))))
goto nla_put_failure;
else if (key->basic.n_proto == htons(ETH_P_IP) &&
key->basic.ip_proto == IPPROTO_ICMP &&
(fl_dump_key_val(skb, &key->icmp.type,
TCA_FLOWER_KEY_ICMPV4_TYPE, &mask->icmp.type,
TCA_FLOWER_KEY_ICMPV4_TYPE_MASK,
sizeof(key->icmp.type)) ||
fl_dump_key_val(skb, &key->icmp.code,
TCA_FLOWER_KEY_ICMPV4_CODE, &mask->icmp.code,
TCA_FLOWER_KEY_ICMPV4_CODE_MASK,
sizeof(key->icmp.code))))
goto nla_put_failure;
else if (key->basic.n_proto == htons(ETH_P_IPV6) &&
key->basic.ip_proto == IPPROTO_ICMPV6 &&
(fl_dump_key_val(skb, &key->icmp.type,
TCA_FLOWER_KEY_ICMPV6_TYPE, &mask->icmp.type,
TCA_FLOWER_KEY_ICMPV6_TYPE_MASK,
sizeof(key->icmp.type)) ||
fl_dump_key_val(skb, &key->icmp.code,
TCA_FLOWER_KEY_ICMPV6_CODE, &mask->icmp.code,
TCA_FLOWER_KEY_ICMPV6_CODE_MASK,
sizeof(key->icmp.code))))
goto nla_put_failure;
else if ((key->basic.n_proto == htons(ETH_P_ARP) ||
key->basic.n_proto == htons(ETH_P_RARP)) &&
(fl_dump_key_val(skb, &key->arp.sip,
TCA_FLOWER_KEY_ARP_SIP, &mask->arp.sip,
TCA_FLOWER_KEY_ARP_SIP_MASK,
sizeof(key->arp.sip)) ||
fl_dump_key_val(skb, &key->arp.tip,
TCA_FLOWER_KEY_ARP_TIP, &mask->arp.tip,
TCA_FLOWER_KEY_ARP_TIP_MASK,
sizeof(key->arp.tip)) ||
fl_dump_key_val(skb, &key->arp.op,
TCA_FLOWER_KEY_ARP_OP, &mask->arp.op,
TCA_FLOWER_KEY_ARP_OP_MASK,
sizeof(key->arp.op)) ||
fl_dump_key_val(skb, key->arp.sha, TCA_FLOWER_KEY_ARP_SHA,
mask->arp.sha, TCA_FLOWER_KEY_ARP_SHA_MASK,
sizeof(key->arp.sha)) ||
fl_dump_key_val(skb, key->arp.tha, TCA_FLOWER_KEY_ARP_THA,
mask->arp.tha, TCA_FLOWER_KEY_ARP_THA_MASK,
sizeof(key->arp.tha))))
goto nla_put_failure;
net: sched: cls_flower: Classify packets using port ranges Added support in tc flower for filtering based on port ranges. Example: 1. Match on a port range: ------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower ip_proto tcp dst_port range 20-30 skip_hw\ action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 filter protocol ip pref 1 flower chain 0 handle 0x1 eth_type ipv4 ip_proto tcp dst_port range 20-30 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 1 ref 1 bind 1 installed 85 sec used 3 sec Action statistics: Sent 460 bytes 10 pkt (dropped 10, overlimits 0 requeues 0) backlog 0b 0p requeues 0 2. Match on IP address and port range: -------------------------------------- $ tc filter add dev enp4s0 protocol ip parent ffff:\ prio 1 flower dst_ip 192.168.1.1 ip_proto tcp dst_port range 100-200\ skip_hw action drop $ tc -s filter show dev enp4s0 parent ffff: filter protocol ip pref 1 flower chain 0 handle 0x2 eth_type ipv4 ip_proto tcp dst_ip 192.168.1.1 dst_port range 100-200 skip_hw not_in_hw action order 1: gact action drop random type none pass val 0 index 2 ref 1 bind 1 installed 58 sec used 2 sec Action statistics: Sent 920 bytes 20 pkt (dropped 20, overlimits 0 requeues 0) backlog 0b 0p requeues 0 v4: 1. Added condition before setting port key. 2. Organized setting and dumping port range keys into functions and added validation of input range. v3: 1. Moved new fields in UAPI enum to the end of enum. 2. Removed couple of empty lines. v2: Addressed Jiri's comments: 1. Added separate functions for dst and src comparisons. 2. Removed endpoint enum. 3. Added new bit TCA_FLOWER_FLAGS_RANGE to decide normal/range lookup. 4. Cleaned up fl_lookup function. Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-11-13 08:15:55 +08:00
if ((key->basic.ip_proto == IPPROTO_TCP ||
key->basic.ip_proto == IPPROTO_UDP ||
key->basic.ip_proto == IPPROTO_SCTP) &&
fl_dump_key_port_range(skb, key, mask))
goto nla_put_failure;
if (key->enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS &&
(fl_dump_key_val(skb, &key->enc_ipv4.src,
TCA_FLOWER_KEY_ENC_IPV4_SRC, &mask->enc_ipv4.src,
TCA_FLOWER_KEY_ENC_IPV4_SRC_MASK,
sizeof(key->enc_ipv4.src)) ||
fl_dump_key_val(skb, &key->enc_ipv4.dst,
TCA_FLOWER_KEY_ENC_IPV4_DST, &mask->enc_ipv4.dst,
TCA_FLOWER_KEY_ENC_IPV4_DST_MASK,
sizeof(key->enc_ipv4.dst))))
goto nla_put_failure;
else if (key->enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS &&
(fl_dump_key_val(skb, &key->enc_ipv6.src,
TCA_FLOWER_KEY_ENC_IPV6_SRC, &mask->enc_ipv6.src,
TCA_FLOWER_KEY_ENC_IPV6_SRC_MASK,
sizeof(key->enc_ipv6.src)) ||
fl_dump_key_val(skb, &key->enc_ipv6.dst,
TCA_FLOWER_KEY_ENC_IPV6_DST,
&mask->enc_ipv6.dst,
TCA_FLOWER_KEY_ENC_IPV6_DST_MASK,
sizeof(key->enc_ipv6.dst))))
goto nla_put_failure;
if (fl_dump_key_val(skb, &key->enc_key_id, TCA_FLOWER_KEY_ENC_KEY_ID,
&mask->enc_key_id, TCA_FLOWER_UNSPEC,
sizeof(key->enc_key_id)) ||
fl_dump_key_val(skb, &key->enc_tp.src,
TCA_FLOWER_KEY_ENC_UDP_SRC_PORT,
&mask->enc_tp.src,
TCA_FLOWER_KEY_ENC_UDP_SRC_PORT_MASK,
sizeof(key->enc_tp.src)) ||
fl_dump_key_val(skb, &key->enc_tp.dst,
TCA_FLOWER_KEY_ENC_UDP_DST_PORT,
&mask->enc_tp.dst,
TCA_FLOWER_KEY_ENC_UDP_DST_PORT_MASK,
sizeof(key->enc_tp.dst)) ||
net/sched: allow flower to match tunnel options Allow matching on options in Geneve tunnel headers. This makes use of existing tunnel metadata support. The options can be described in the form CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is represented as a 16bit hexadecimal value, TYPE as an 8bit hexadecimal value and DATA as a variable length hexadecimal value. e.g. # ip link add name geneve0 type geneve dstport 0 external # tc qdisc add dev geneve0 ingress # tc filter add dev geneve0 protocol ip parent ffff: \ flower \ enc_src_ip 10.0.99.192 \ enc_dst_ip 10.0.99.193 \ enc_key_id 11 \ geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \ ip_proto udp \ action mirred egress redirect dev eth1 This patch adds support for matching Geneve options in the order supplied by the user. This leads to an efficient implementation in the software datapath (and in our opinion hardware datapaths that offload this feature). It is also compatible with Geneve options matching provided by the Open vSwitch kernel datapath which is relevant here as the Flower classifier may be used as a mechanism to program flows into hardware as a form of Open vSwitch datapath offload (sometimes referred to as OVS-TC). The netlink Kernel/Userspace API may be extended, for example by adding a flag, if other matching options are desired, for example matching given options in any order. This would require an implementation in the TC software datapath. And be done in a way that drivers that facilitate offload of the Flower classifier can reject or accept such flows based on hardware datapath capabilities. This approach was discussed and agreed on at Netconf 2017 in Seoul. Signed-off-by: Simon Horman <simon.horman@netronome.com> Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-08-07 23:36:01 +08:00
fl_dump_key_ip(skb, true, &key->enc_ip, &mask->enc_ip) ||
fl_dump_key_enc_opt(skb, &key->enc_opts, &mask->enc_opts))
goto nla_put_failure;
if (fl_dump_key_ct(skb, &key->ct, &mask->ct))
goto nla_put_failure;
if (fl_dump_key_flags(skb, key->control.flags, mask->control.flags))
goto nla_put_failure;
if (fl_dump_key_val(skb, &key->hash.hash, TCA_FLOWER_KEY_HASH,
&mask->hash.hash, TCA_FLOWER_KEY_HASH_MASK,
sizeof(key->hash.hash)))
goto nla_put_failure;
return 0;
nla_put_failure:
return -EMSGSIZE;
}
static int fl_dump(struct net *net, struct tcf_proto *tp, void *fh,
struct sk_buff *skb, struct tcmsg *t, bool rtnl_held)
{
struct cls_fl_filter *f = fh;
struct nlattr *nest;
struct fl_flow_key *key, *mask;
bool skip_hw;
if (!f)
return skb->len;
t->tcm_handle = f->handle;
nest = nla_nest_start_noflag(skb, TCA_OPTIONS);
if (!nest)
goto nla_put_failure;
spin_lock(&tp->lock);
if (f->res.classid &&
nla_put_u32(skb, TCA_FLOWER_CLASSID, f->res.classid))
goto nla_put_failure_locked;
key = &f->key;
mask = &f->mask->key;
skip_hw = tc_skip_hw(f->flags);
if (fl_dump_key(skb, net, key, mask))
goto nla_put_failure_locked;
if (f->flags && nla_put_u32(skb, TCA_FLOWER_FLAGS, f->flags))
goto nla_put_failure_locked;
spin_unlock(&tp->lock);
if (!skip_hw)
fl_hw_update_stats(tp, f, rtnl_held);
if (nla_put_u32(skb, TCA_FLOWER_IN_HW_COUNT, f->in_hw_count))
goto nla_put_failure;
if (tcf_exts_dump(skb, &f->exts))
goto nla_put_failure;
nla_nest_end(skb, nest);
if (tcf_exts_dump_stats(skb, &f->exts) < 0)
goto nla_put_failure;
return skb->len;
nla_put_failure_locked:
spin_unlock(&tp->lock);
nla_put_failure:
nla_nest_cancel(skb, nest);
return -1;
}
static int fl_terse_dump(struct net *net, struct tcf_proto *tp, void *fh,
struct sk_buff *skb, struct tcmsg *t, bool rtnl_held)
{
struct cls_fl_filter *f = fh;
struct nlattr *nest;
bool skip_hw;
if (!f)
return skb->len;
t->tcm_handle = f->handle;
nest = nla_nest_start_noflag(skb, TCA_OPTIONS);
if (!nest)
goto nla_put_failure;
spin_lock(&tp->lock);
skip_hw = tc_skip_hw(f->flags);
if (f->flags && nla_put_u32(skb, TCA_FLOWER_FLAGS, f->flags))
goto nla_put_failure_locked;
spin_unlock(&tp->lock);
if (!skip_hw)
fl_hw_update_stats(tp, f, rtnl_held);
if (tcf_exts_terse_dump(skb, &f->exts))
goto nla_put_failure;
nla_nest_end(skb, nest);
return skb->len;
nla_put_failure_locked:
spin_unlock(&tp->lock);
nla_put_failure:
nla_nest_cancel(skb, nest);
return -1;
}
static int fl_tmplt_dump(struct sk_buff *skb, struct net *net, void *tmplt_priv)
{
struct fl_flow_tmplt *tmplt = tmplt_priv;
struct fl_flow_key *key, *mask;
struct nlattr *nest;
nest = nla_nest_start_noflag(skb, TCA_OPTIONS);
if (!nest)
goto nla_put_failure;
key = &tmplt->dummy_key;
mask = &tmplt->mask;
if (fl_dump_key(skb, net, key, mask))
goto nla_put_failure;
nla_nest_end(skb, nest);
return skb->len;
nla_put_failure:
nla_nest_cancel(skb, nest);
return -EMSGSIZE;
}
static void fl_bind_class(void *fh, u32 classid, unsigned long cl, void *q,
unsigned long base)
{
struct cls_fl_filter *f = fh;
if (f && f->res.classid == classid) {
if (cl)
__tcf_bind_filter(q, &f->res, base);
else
__tcf_unbind_filter(q, &f->res);
}
}
net/sched: add delete_empty() to filters and use it in cls_flower Revert "net/sched: cls_u32: fix refcount leak in the error path of u32_change()", and fix the u32 refcount leak in a more generic way that preserves the semantic of rule dumping. On tc filters that don't support lockless insertion/removal, there is no need to guard against concurrent insertion when a removal is in progress. Therefore, for most of them we can avoid a full walk() when deleting, and just decrease the refcount, like it was done on older Linux kernels. This fixes situations where walk() was wrongly detecting a non-empty filter, like it happened with cls_u32 in the error path of change(), thus leading to failures in the following tdc selftests: 6aa7: (filter, u32) Add/Replace u32 with source match and invalid indev 6658: (filter, u32) Add/Replace u32 with custom hash table and invalid handle 74c2: (filter, u32) Add/Replace u32 filter with invalid hash table id On cls_flower, and on (future) lockless filters, this check is necessary: move all the check_empty() logic in a callback so that each filter can have its own implementation. For cls_flower, it's sufficient to check if no IDRs have been allocated. This reverts commit 275c44aa194b7159d1191817b20e076f55f0e620. Changes since v1: - document the need for delete_empty() when TCF_PROTO_OPS_DOIT_UNLOCKED is used, thanks to Vlad Buslov - implement delete_empty() without doing fl_walk(), thanks to Vlad Buslov - squash revert and new fix in a single patch, to be nice with bisect tests that run tdc on u32 filter, thanks to Dave Miller Fixes: 275c44aa194b ("net/sched: cls_u32: fix refcount leak in the error path of u32_change()") Fixes: 6676d5e416ee ("net: sched: set dedicated tcf_walker flag when tp is empty") Suggested-by: Jamal Hadi Salim <jhs@mojatatu.com> Suggested-by: Vlad Buslov <vladbu@mellanox.com> Signed-off-by: Davide Caratti <dcaratti@redhat.com> Reviewed-by: Vlad Buslov <vladbu@mellanox.com> Tested-by: Jamal Hadi Salim <jhs@mojatatu.com> Acked-by: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-28 23:36:58 +08:00
static bool fl_delete_empty(struct tcf_proto *tp)
{
struct cls_fl_head *head = fl_head_dereference(tp);
spin_lock(&tp->lock);
tp->deleting = idr_is_empty(&head->handle_idr);
spin_unlock(&tp->lock);
return tp->deleting;
}
static struct tcf_proto_ops cls_fl_ops __read_mostly = {
.kind = "flower",
.classify = fl_classify,
.init = fl_init,
.destroy = fl_destroy,
.get = fl_get,
.put = fl_put,
.change = fl_change,
.delete = fl_delete,
net/sched: add delete_empty() to filters and use it in cls_flower Revert "net/sched: cls_u32: fix refcount leak in the error path of u32_change()", and fix the u32 refcount leak in a more generic way that preserves the semantic of rule dumping. On tc filters that don't support lockless insertion/removal, there is no need to guard against concurrent insertion when a removal is in progress. Therefore, for most of them we can avoid a full walk() when deleting, and just decrease the refcount, like it was done on older Linux kernels. This fixes situations where walk() was wrongly detecting a non-empty filter, like it happened with cls_u32 in the error path of change(), thus leading to failures in the following tdc selftests: 6aa7: (filter, u32) Add/Replace u32 with source match and invalid indev 6658: (filter, u32) Add/Replace u32 with custom hash table and invalid handle 74c2: (filter, u32) Add/Replace u32 filter with invalid hash table id On cls_flower, and on (future) lockless filters, this check is necessary: move all the check_empty() logic in a callback so that each filter can have its own implementation. For cls_flower, it's sufficient to check if no IDRs have been allocated. This reverts commit 275c44aa194b7159d1191817b20e076f55f0e620. Changes since v1: - document the need for delete_empty() when TCF_PROTO_OPS_DOIT_UNLOCKED is used, thanks to Vlad Buslov - implement delete_empty() without doing fl_walk(), thanks to Vlad Buslov - squash revert and new fix in a single patch, to be nice with bisect tests that run tdc on u32 filter, thanks to Dave Miller Fixes: 275c44aa194b ("net/sched: cls_u32: fix refcount leak in the error path of u32_change()") Fixes: 6676d5e416ee ("net: sched: set dedicated tcf_walker flag when tp is empty") Suggested-by: Jamal Hadi Salim <jhs@mojatatu.com> Suggested-by: Vlad Buslov <vladbu@mellanox.com> Signed-off-by: Davide Caratti <dcaratti@redhat.com> Reviewed-by: Vlad Buslov <vladbu@mellanox.com> Tested-by: Jamal Hadi Salim <jhs@mojatatu.com> Acked-by: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-28 23:36:58 +08:00
.delete_empty = fl_delete_empty,
.walk = fl_walk,
.reoffload = fl_reoffload,
.hw_add = fl_hw_add,
.hw_del = fl_hw_del,
.dump = fl_dump,
.terse_dump = fl_terse_dump,
.bind_class = fl_bind_class,
.tmplt_create = fl_tmplt_create,
.tmplt_destroy = fl_tmplt_destroy,
.tmplt_dump = fl_tmplt_dump,
.owner = THIS_MODULE,
.flags = TCF_PROTO_OPS_DOIT_UNLOCKED,
};
static int __init cls_fl_init(void)
{
return register_tcf_proto_ops(&cls_fl_ops);
}
static void __exit cls_fl_exit(void)
{
unregister_tcf_proto_ops(&cls_fl_ops);
}
module_init(cls_fl_init);
module_exit(cls_fl_exit);
MODULE_AUTHOR("Jiri Pirko <jiri@resnulli.us>");
MODULE_DESCRIPTION("Flower classifier");
MODULE_LICENSE("GPL v2");