- 14 May, 2015 7 commits
-
-
David S. Miller authored
Pablo Neira Ayuso says: ==================== Netfilter ingress support (v4) This is the v4 round of patches to add the Netfilter ingress hook, it basically comes in two steps: 1) Add the CONFIG_NET_INGRESS switch to wrap the ingress static key around it. The idea is to use the same global static key to avoid adding more code to the hot path. 2) Add the Netfilter ingress hook after the tc ingress hook, under the global ingress_needed static key. As I said, the netfilter ingress hook also has its own static key, that is nested under the global static key. Please, see patch 5/5 for performance numbers and more information. I originally started this next round, as it was suggested, exploring the independent static key for netfilter ingress just after tc ingress, but the results that I gathered from that patch are not good for non-users: Result: OK: 6425927(c6425843+d83) usec, 100000000 (60byte,0frags) 15561955pps 7469Mb/sec (7469738400bps) errors: 100000000 this roughly means 500Kpps less performance wrt. the base numbers, so that's the reason why I discarded that approach and I focused on this. The idea of this patchset is to open the window to nf_tables, which comes with features that will work out-of-the-box (once the boiler plate code to support the 'netdev' table family is in place), to avoid repeating myself [1], the most relevant features are: 1) Multi-dimensional key dictionary lookups. 2) Arbitrary stateful flow tables. 3) Transactions and good support for dynamic updates. But there are also interest aspects to consider from userspace, such as the ability to support new layer 2 protocols without kernel updates, a well-defined netlink interface, userspace libraries and utilities for third party applications, among others. I hope we can be happy with this approach. Please, apply. Thanks. [1] http://marc.info/?l=netfilter-devel&m=143033337020328&w=2 ==================== Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Pablo Neira authored
This patch adds the Netfilter ingress hook just after the existing tc ingress hook, that seems to be the consensus solution for this. Note that the Netfilter hook resides under the global static key that enables ingress filtering. Nonetheless, Netfilter still also has its own static key for minimal impact on the existing handle_ing(). * Without this patch: Result: OK: 6216490(c6216338+d152) usec, 100000000 (60byte,0frags) 16086246pps 7721Mb/sec (7721398080bps) errors: 100000000 42.46% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core 25.92% kpktgend_0 [kernel.kallsyms] [k] kfree_skb 7.81% kpktgend_0 [pktgen] [k] pktgen_thread_worker 5.62% kpktgend_0 [kernel.kallsyms] [k] ip_rcv 2.70% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal 2.34% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk 1.44% kpktgend_0 [kernel.kallsyms] [k] __build_skb * With this patch: Result: OK: 6214833(c6214731+d101) usec, 100000000 (60byte,0frags) 16090536pps 7723Mb/sec (7723457280bps) errors: 100000000 41.23% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core 26.57% kpktgend_0 [kernel.kallsyms] [k] kfree_skb 7.72% kpktgend_0 [pktgen] [k] pktgen_thread_worker 5.55% kpktgend_0 [kernel.kallsyms] [k] ip_rcv 2.78% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal 2.06% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk 1.43% kpktgend_0 [kernel.kallsyms] [k] __build_skb * Without this patch + tc ingress: tc filter add dev eth4 parent ffff: protocol ip prio 1 \ u32 match ip dst 4.3.2.1/32 Result: OK: 9269001(c9268821+d179) usec, 100000000 (60byte,0frags) 10788648pps 5178Mb/sec (5178551040bps) errors: 100000000 40.99% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core 17.50% kpktgend_0 [kernel.kallsyms] [k] kfree_skb 11.77% kpktgend_0 [cls_u32] [k] u32_classify 5.62% kpktgend_0 [kernel.kallsyms] [k] tc_classify_compat 5.18% kpktgend_0 [pktgen] [k] pktgen_thread_worker 3.23% kpktgend_0 [kernel.kallsyms] [k] tc_classify 2.97% kpktgend_0 [kernel.kallsyms] [k] ip_rcv 1.83% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal 1.50% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk 0.99% kpktgend_0 [kernel.kallsyms] [k] __build_skb * With this patch + tc ingress: tc filter add dev eth4 parent ffff: protocol ip prio 1 \ u32 match ip dst 4.3.2.1/32 Result: OK: 9308218(c9308091+d126) usec, 100000000 (60byte,0frags) 10743194pps 5156Mb/sec (5156733120bps) errors: 100000000 42.01% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core 17.78% kpktgend_0 [kernel.kallsyms] [k] kfree_skb 11.70% kpktgend_0 [cls_u32] [k] u32_classify 5.46% kpktgend_0 [kernel.kallsyms] [k] tc_classify_compat 5.16% kpktgend_0 [pktgen] [k] pktgen_thread_worker 2.98% kpktgend_0 [kernel.kallsyms] [k] ip_rcv 2.84% kpktgend_0 [kernel.kallsyms] [k] tc_classify 1.96% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal 1.57% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk Note that the results are very similar before and after. I can see gcc gets the code under the ingress static key out of the hot path. Then, on that cold branch, it generates the code to accomodate the netfilter ingress static key. My explanation for this is that this reduces the pressure on the instruction cache for non-users as the new code is out of the hot path, and it comes with minimal impact for tc ingress users. Using gcc version 4.8.4 on: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 8 [...] L1d cache: 16K L1i cache: 64K L2 cache: 2048K L3 cache: 8192K Signed-off-by:Pablo Neira Ayuso <pablo@netfilter.org> Acked-by:
Alexei Starovoitov <ast@plumgrid.com> Signed-off-by:
-
Pablo Neira authored
This new config switch enables the ingress filtering infrastructure that is controlled through the ingress_needed static key. This prepares the introduction of the Netfilter ingress hook that resides under this unique static key. Note that CONFIG_SCH_INGRESS automatically selects this, that should be no problem since this also depends on CONFIG_NET_CLS_ACT. Signed-off-by:
-
Pablo Neira authored
In preparation to have netfilter ingress per-device hook list. Signed-off-by:
-
Pablo Neira authored
Signed-off-by:
-
Pablo Neira authored
Signed-off-by:
-
Dan Carpenter authored
The bitwise tests are always true here because it uses '|' where '&' is intended. Fixes: 98b5a0f4 ('net: macb: Add support for jumbo frames') Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
- 13 May, 2015 33 commits
-
-
Alexander Duyck authored
When I had inlined __alloc_rx_skb into __netdev_alloc_skb and __napi_alloc_skb I had overlooked the fact that there was a return in the __alloc_rx_skb. As a result we weren't reserving headroom or setting the skb->dev in certain cases. This change corrects that by adding a couple of jump labels to jump to depending on __alloc_skb either succeeding or failing. Fixes: 9451980a ("net: Use cached copy of pfmemalloc to avoid accessing page") Reported-by:
Felipe Balbi <balbi@ti.com> Signed-off-by:
Alexander Duyck <alexander.h.duyck@redhat.com> Tested-by:
Kevin Hilman <khilman@linaro.org> Signed-off-by:
-
David S. Miller authored
John W. Linville says: ==================== add GENEVE netdev tunnel driver This 5-patch kernel series adds a netdev implementation of a GENEVE tunnel driver, and the single iproute2 patch enables creation and such for those netdevs. This makes use of the existing GENEVE infrastructure already used by the OVS code. The net/ipv4/geneve.c file is renamed as net/ipv4/geneve_core.c as part of these changes. drivers/net/Kconfig | 14 + drivers/net/Makefile | 1 drivers/net/geneve.c | 503 +++++++++++++++++++++++++++++++++++++++++ include/net/geneve.h | 5 include/uapi/linux/if_link.h | 9 net/ipv4/Kconfig | 4 net/ipv4/Makefile | 2 net/ipv4/geneve.c | 6 net/ipv4/geneve_core.c | 4 net/openvswitch/Kconfig | 2 net/openvswitch/vport-geneve.c | 5 11 files changed, 538 insertions(+), 17 deletions(-) The overall structure of the GENEVE netdev driver is strongly influenced by the VXLAN netdev driver. This is not surprising, as the two drivers are intended to serve similar purposes. As development of the GENEVE driver continues, it is likely that those similarities will grow stronger. This will include both simple configuration options (e.g. TOS and TTL settings) and new control plane support. The current implementation is very simple, restricting itself to point to point links over IPv4. This is due only to the simplicity of the implementation, and no such limit is inherent to GENEVE in any way. Support for IPv6 links and more sophisticated control plane options are predictable enhancements. Using the included iproute2 patch, a GENEVE tunnel is created thusly: ip link add dev gnv0 type geneve remote 192.168.22.1 vni 1234 ip link set gnv0 up ip addr add 10.1.1.1/24 dev gnv0 After a corresponding tunnel interface is created at the link partner, traffic should proceed as expected. Please let me know if anyone has problems...thanks! ==================== Signed-off-by: -
John W. Linville authored
This is an initial implementation of a netdev driver for GENEVE tunnels. This implementation uses a fixed UDP port, and only supports point-to-point links with specific partner endpoints. Only IPv4 links are supported at this time. Signed-off-by:
John W. Linville <linville@tuxdriver.com> Signed-off-by:
-
John W. Linville authored
Signed-off-by:
-
John W. Linville authored
net/ipv4/geneve.c -> net/ipv4/geneve_core.c This name better reflects the purpose of the module. Signed-off-by:
-
John W. Linville authored
This is a static inline with identical definitions in multiple places... Signed-off-by:
-
John W. Linville authored
This file is essentially a library for implementing the geneve encapsulation protocol. The file does not register any rtnl_link_ops, so the MODULE_ALIAS_RTNL_LINK macro is inappropriate here. Signed-off-by:
-
Pablo Neira authored
This fixes 4577139b ("net: use jump label patching for ingress qdisc in __netif_receive_skb_core"). The only client of this is sch_ingress and it depends on NET_CLS_ACT. So there is no way these definition can be of any help. Cc: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by:
Daniel Borkmann <daniel@iogearbox.net> Acked-by:
-
David S. Miller authored
Willem de Bruijn says: ==================== refine packet socket rollover: 1. mitigate a case of lock contention 2. avoid exporting resource exhaustion to other sockets, by migrating only to a victim socket that has ample room 3. avoid reordering of most flows on the socket, by migrating first the flow responsible for load imbalance 4. help processes detect load imbalance, by exporting rollover counters Context: rollover implements flow migration in packet socket fanout groups in case of extreme load imbalance. It is a specific implementation of migration that minimizes reordering by selecting the same victim socket when possible (and by selecting subsequent victims in a round robin fashion, from which its name derives). Changes: v2 -> v3: - statistics: replace unsigned long with __aligned_u64 v1 -> v2: - huge flow detection: run lockless - huge flow detection: replace stored index with random - contention avoidance: test in packet_poll while lock held - contention avoidance: clear pressure sooner packet_poll and packet_recvmsg would clear only if the sock is empty to avoid taking the necessary lock. But, * packet_poll already holds this lock, so a lockless variant __packet_rcv_has_room is cheap. * packet_recvmsg is usually called only for non-ring sockets, which also runs lockless. - preparation: drop "single return" patch packet_rcv_has_room is now a locked wrapper around __packet_rcv_has_room, achieving the same (single footer). The benchmark mentioned in the patches is at https://github.com/wdebruij/kerneltools/blob/master/tests/bench_rollover.c ==================== Signed-off-by: -
Willem de Bruijn authored
Rollover indicates exceptional conditions. Export a counter to inform socket owners of this state. If no socket with sufficient room is found, rollover fails. Also count these events. Finally, also count when flows are rolled over early thanks to huge flow detection, to validate its correctness. Tested: Read counters in bench_rollover on all other tests in the patchset Signed-off-by:
Willem de Bruijn <willemb@google.com> Signed-off-by:
-
Willem de Bruijn authored
Migrate flows from a socket to another socket in the fanout group not only when the socket is full. Start migrating huge flows early, to divert possible 4-tuple attacks without affecting normal traffic. Introduce fanout_flow_is_huge(). This detects huge flows, which are defined as taking up more than half the load. It does so cheaply, by storing the rxhashes of the N most recent packets. If over half of these are the same rxhash as the current packet, then drop it. This only protects against 4-tuple attacks. N is chosen to fit all data in a single cache line. Tested: Ran bench_rollover for 10 sec with 1.5 Mpps of single flow input. lpbb5:/export/hda3/willemb# ./bench_rollover -l 1000 -r -s cpu rx rx.k drop.k rollover r.huge r.failed 0 14 14 0 0 0 0 1 20 20 0 0 0 0 2 16 16 0 0 0 0 3 6168824 6168824 0 4867721 4867721 0 4 4867741 4867741 0 0 0 0 5 12 12 0 0 0 0 6 15 15 0 0 0 0 7 17 17 0 0 0 0 Signed-off-by: -
Willem de Bruijn authored
Rollover has to call packet_rcv_has_room on sockets in the fanout group to find a socket to migrate to. This operation is expensive especially if the packet sockets use rings, when a lock has to be acquired. Avoid pounding on the lock by all sockets by temporarily marking a socket as "under memory pressure" when such pressure is detected. While set, only the socket owner may call packet_rcv_has_room on the socket. Once it detects normal conditions, it clears the flag. The socket is not used as a victim by any other socket in the meantime. Under reasonably balanced load, each socket writer frequently calls packet_rcv_has_room and clears its own pressure field. As a backup for when the socket is rarely written to, also clear the flag on reading (packet_recvmsg, packet_poll) if this can be done cheaply (i.e., without calling packet_rcv_has_room). This is only for edge cases. Tested: Ran bench_rollover: a process with 8 sockets in a single fanout group, each pinned to a single cpu that receives one nic recv interrupt. RPS and RFS are disabled. The benchmark uses packet rx_ring, which has to take a lock when determining whether a socket has room. Sent 3.5 Mpps of UDP traffic with sufficient entropy to spread uniformly across the packet sockets (and inserted an iptables rule to drop in PREROUTING to avoid protocol stack processing). Without this patch, all sockets try to migrate traffic to neighbors, causing lock contention when searching for a non- empty neighbor. The lock is the top 9 entries. perf record -a -g sleep 5 - 17.82% bench_rollover [kernel.kallsyms] [k] _raw_spin_lock - _raw_spin_lock - 99.00% spin_lock + 81.77% packet_rcv_has_room.isra.41 + 18.23% tpacket_rcv + 0.84% packet_rcv_has_room.isra.41 + 5.20% ksoftirqd/6 [kernel.kallsyms] [k] _raw_spin_lock + 5.15% ksoftirqd/1 [kernel.kallsyms] [k] _raw_spin_lock + 5.14% ksoftirqd/2 [kernel.kallsyms] [k] _raw_spin_lock + 5.12% ksoftirqd/7 [kernel.kallsyms] [k] _raw_spin_lock + 5.12% ksoftirqd/5 [kernel.kallsyms] [k] _raw_spin_lock + 5.10% ksoftirqd/4 [kernel.kallsyms] [k] _raw_spin_lock + 4.66% ksoftirqd/0 [kernel.kallsyms] [k] _raw_spin_lock + 4.45% ksoftirqd/3 [kernel.kallsyms] [k] _raw_spin_lock + 1.55% bench_rollover [kernel.kallsyms] [k] packet_rcv_has_room.isra.41 On net-next with this patch, this lock contention is no longer a top entry. Most time is spent in the actual read function. Next up are other locks: + 15.52% bench_rollover bench_rollover [.] reader + 4.68% swapper [kernel.kallsyms] [k] memcpy_erms + 2.77% swapper [kernel.kallsyms] [k] packet_lookup_frame.isra.51 + 2.56% ksoftirqd/1 [kernel.kallsyms] [k] memcpy_erms + 2.16% swapper [kernel.kallsyms] [k] tpacket_rcv + 1.93% swapper [kernel.kallsyms] [k] mlx4_en_process_rx_cq Looking closer at the remaining _raw_spin_lock, the cost of probing in rollover is now comparable to the cost of taking the lock later in tpacket_rcv. - 1.51% swapper [kernel.kallsyms] [k] _raw_spin_lock - _raw_spin_lock + 33.41% packet_rcv_has_room + 28.15% tpacket_rcv + 19.54% enqueue_to_backlog + 6.45% __free_pages_ok + 2.78% packet_rcv_fanout + 2.13% fanout_demux_rollover + 2.01% netif_receive_skb_internal Signed-off-by: -
Willem de Bruijn authored
Only migrate flows to sockets that have sufficient headroom, where sufficient is defined as having at least 25% empty space. The kernel has three different buffer types: a regular socket, a ring with frames (TPACKET_V[12]) or a ring with blocks (TPACKET_V3). The latter two do not expose a read pointer to the kernel, so headroom is not computed easily. All three needs a different implementation to estimate free space. Tested: Ran bench_rollover for 10 sec with 1.5 Mpps of single flow input. bench_rollover has as many sockets as there are NIC receive queues in the system. Each socket is owned by a process that is pinned to one of the receive cpus. RFS is disabled. RPS is enabled with an identity mapping (cpu x -> cpu x), to count drops with softnettop. lpbb5:/export/hda3/willemb# ./bench_rollover -r -l 1000 -s Press [Enter] to exit cpu rx rx.k drop.k rollover r.huge r.failed 0 16 16 0 0 0 0 1 21 21 0 0 0 0 2 5227502 5227502 0 0 0 0 3 18 18 0 0 0 0 4 6083289 6083289 0 5227496 0 0 5 22 22 0 0 0 0 6 21 21 0 0 0 0 7 9 9 0 0 0 0 Signed-off-by: -
Willem de Bruijn authored
Replace rollover state per fanout group with state per socket. Future patches will add fields to the new structure. Signed-off-by:
-
Willem de Bruijn authored
packet_rcv_fanout calls fanout_demux_rollover twice. Move all rollover logic into the callee to simplify these callsites, especially with upcoming changes. The main differences between the two callsites is that the FLAG variant tests whether the socket previously selected by another mode (RR, RND, HASH, ..) has room before migrating flows, whereas the rollover mode has no original socket to test. Signed-off-by:
-
Eric Dumazet authored
__ip_local_out_sk() is only used from net/ipv4/ip_output.c net/ipv4/ip_output.c:94:5: warning: symbol '__ip_local_out_sk' was not declared. Should it be static? Fixes: 7026b1dd ("netfilter: Pass socket pointer down through okfn().") Signed-off-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
-
Eric Dumazet authored
tw_timer_handler() is only used from net/ipv4/inet_timewait_sock.c Fixes: 789f558c ("tcp/dccp: get rid of central timewait timer") Signed-off-by:
-
David S. Miller authored
Jiri Pirko says: ==================== introduce programable flow dissector and cls_flower Per Davem's request, I prepared this patchset which introduces programmable flow dissector. For current users of flow_keys, there is a wrapper skb_flow_dissect_flow_keys which maintains the previous behaviour. For purposes of cls_flower, couple of new dissection keys were introduced. Note that this dissector can be also eventually used by openvswitch code. Also, as a next step, I plan to get rid of *skb_flow_get_ports(export) and *__skb_get_poff as their functionality can be now implemented by skb_flow_dissect as well. v2->v3: - remove TCA_FLOWER_POLICE attr suggested by Jamal v1->v2: - move __skb_tx_hash rather to dev.c as suggested by Alex ==================== Signed-off-by: -
Jiri Pirko authored
This patch introduces a flow-based filter. So far, the very essential packet fields are supported. This patch is only the first step. There is a lot of potential performance improvements possible to implement. Also a lot of features are missing now. They will be addressed in follow-up patches. Signed-off-by:
Jiri Pirko <jiri@resnulli.us> Acked-by:
Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
So far, only hashes made out of ipv6 addresses could be dissected. This patch introduces support for dissection of full ipv6 addresses. Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
Introduce dissector infrastructure which allows user to specify which parts of skb he wants to dissect. Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
next to its user. No relation to flow_dissector so it makes no sense to have it in flow_dissector.c Signed-off-by:
-
Jiri Pirko authored
__skb_tx_hash function has no relation to flow_dissect so just move it to dev.c Signed-off-by:
-
Jiri Pirko authored
Since the definition of the function is in flow_dissector.c, it makes sense to have the declaration in flow_dissector.h Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
Since these functions are defined in flow_dissector.c, move header declarations from skbuff.h into flow_dissector.h Signed-off-by:
-
Jiri Pirko authored
commit 56193d1b ("net: Add function for parsing the header length out of linear ethernet frames") added this function declaration but it is defined nowhere. Signed-off-by:
-
Jiri Pirko authored
add couple of empty lines on the way. Signed-off-by:
-
