1 .. SPDX-License-Identifier: GPL-2.0
3 ===============================================
4 XFRM device - offloading the IPsec computations
5 ===============================================
7 Shannon Nelson <shannon.nelson@oracle.com>
8 Leon Romanovsky <leonro@nvidia.com>
14 IPsec is a useful feature for securing network traffic, but the
15 computational cost is high: a 10Gbps link can easily be brought down
16 to under 1Gbps, depending on the traffic and link configuration.
17 Luckily, there are NICs that offer a hardware based IPsec offload which
18 can radically increase throughput and decrease CPU utilization. The XFRM
19 Device interface allows NIC drivers to offer to the stack access to the
22 Right now, there are two types of hardware offload that kernel supports.
23 * IPsec crypto offload:
24 * NIC performs encrypt/decrypt
25 * Kernel does everything else
26 * IPsec packet offload:
27 * NIC performs encrypt/decrypt
28 * NIC does encapsulation
29 * Kernel and NIC have SA and policy in-sync
30 * NIC handles the SA and policies states
31 * The Kernel talks to the keymanager
33 Userland access to the offload is typically through a system such as
34 libreswan or KAME/raccoon, but the iproute2 'ip xfrm' command set can
35 be handy when experimenting. An example command might look something
36 like this for crypto offload:
38 ip x s add proto esp dst 14.0.0.70 src 14.0.0.52 spi 0x07 mode transport \
39 reqid 0x07 replay-window 32 \
40 aead 'rfc4106(gcm(aes))' 0x44434241343332312423222114131211f4f3f2f1 128 \
41 sel src 14.0.0.52/24 dst 14.0.0.70/24 proto tcp \
42 offload dev eth4 dir in
44 and for packet offload
46 ip x s add proto esp dst 14.0.0.70 src 14.0.0.52 spi 0x07 mode transport \
47 reqid 0x07 replay-window 32 \
48 aead 'rfc4106(gcm(aes))' 0x44434241343332312423222114131211f4f3f2f1 128 \
49 sel src 14.0.0.52/24 dst 14.0.0.70/24 proto tcp \
50 offload packet dev eth4 dir in
52 ip x p add src 14.0.0.70 dst 14.0.0.52 offload packet dev eth4 dir in
53 tmpl src 14.0.0.70 dst 14.0.0.52 proto esp reqid 10000 mode transport
55 Yes, that's ugly, but that's what shell scripts and/or libreswan are for.
59 Callbacks to implement
60 ======================
64 /* from include/linux/netdevice.h */
66 /* Crypto and Packet offload callbacks */
67 int (*xdo_dev_state_add) (struct xfrm_state *x, struct netlink_ext_ack *extack);
68 void (*xdo_dev_state_delete) (struct xfrm_state *x);
69 void (*xdo_dev_state_free) (struct xfrm_state *x);
70 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
71 struct xfrm_state *x);
72 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
74 /* Solely packet offload callbacks */
75 void (*xdo_dev_state_update_curlft) (struct xfrm_state *x);
76 int (*xdo_dev_policy_add) (struct xfrm_policy *x, struct netlink_ext_ack *extack);
77 void (*xdo_dev_policy_delete) (struct xfrm_policy *x);
78 void (*xdo_dev_policy_free) (struct xfrm_policy *x);
81 The NIC driver offering ipsec offload will need to implement callbacks
82 relevant to supported offload to make the offload available to the network
83 stack's XFRM subsystem. Additionally, the feature bits NETIF_F_HW_ESP and
84 NETIF_F_HW_ESP_TX_CSUM will signal the availability of the offload.
91 At probe time and before the call to register_netdev(), the driver should
92 set up local data structures and XFRM callbacks, and set the feature bits.
93 The XFRM code's listener will finish the setup on NETDEV_REGISTER.
97 adapter->netdev->xfrmdev_ops = &ixgbe_xfrmdev_ops;
98 adapter->netdev->features |= NETIF_F_HW_ESP;
99 adapter->netdev->hw_enc_features |= NETIF_F_HW_ESP;
101 When new SAs are set up with a request for "offload" feature, the
102 driver's xdo_dev_state_add() will be given the new SA to be offloaded
103 and an indication of whether it is for Rx or Tx. The driver should
105 - verify the algorithm is supported for offloads
106 - store the SA information (key, salt, target-ip, protocol, etc)
107 - enable the HW offload of the SA
108 - return status value:
110 =========== ===================================
112 -EOPNETSUPP offload not supported, try SW IPsec,
113 not applicable for packet offload mode
114 other fail the request
115 =========== ===================================
117 The driver can also set an offload_handle in the SA, an opaque void pointer
118 that can be used to convey context into the fast-path offload requests::
120 xs->xso.offload_handle = context;
123 When the network stack is preparing an IPsec packet for an SA that has
124 been setup for offload, it first calls into xdo_dev_offload_ok() with
125 the skb and the intended offload state to ask the driver if the offload
126 will serviceable. This can check the packet information to be sure the
127 offload can be supported (e.g. IPv4 or IPv6, no IPv4 options, etc) and
128 return true of false to signify its support.
131 When ready to send, the driver needs to inspect the Tx packet for the
132 offload information, including the opaque context, and set up the packet
135 xs = xfrm_input_state(skb);
136 context = xs->xso.offload_handle;
139 The stack has already inserted the appropriate IPsec headers in the
140 packet data, the offload just needs to do the encryption and fix up the
144 When a packet is received and the HW has indicated that it offloaded a
145 decryption, the driver needs to add a reference to the decoded SA into
146 the packet's skb. At this point the data should be decrypted but the
147 IPsec headers are still in the packet data; they are removed later up
148 the stack in xfrm_input().
150 find and hold the SA that was used to the Rx skb::
152 get spi, protocol, and destination IP from packet headers
153 xs = find xs from (spi, protocol, dest_IP)
156 store the state information into the skb::
158 sp = secpath_set(skb);
160 sp->xvec[sp->len++] = xs;
163 indicate the success and/or error status of the offload::
165 xo = xfrm_offload(skb);
166 xo->flags = CRYPTO_DONE;
167 xo->status = crypto_status;
169 hand the packet to napi_gro_receive() as usual
171 In ESN mode, xdo_dev_state_advance_esn() is called from xfrm_replay_advance_esn().
172 Driver will check packet seq number and update HW ESN state machine if needed.
175 HW adds and deletes XFRM headers. So in RX path, XFRM stack is bypassed if HW
176 reported success. In TX path, the packet lefts kernel without extra header
177 and not encrypted, the HW is responsible to perform it.
179 When the SA is removed by the user, the driver's xdo_dev_state_delete()
180 and xdo_dev_policy_delete() are asked to disable the offload. Later,
181 xdo_dev_state_free() and xdo_dev_policy_free() are called from a garbage
182 collection routine after all reference counts to the state and policy
183 have been removed and any remaining resources can be cleared for the
184 offload state. How these are used by the driver will depend on specific
187 As a netdev is set to DOWN the XFRM stack's netdev listener will call
188 xdo_dev_state_delete(), xdo_dev_policy_delete(), xdo_dev_state_free() and
189 xdo_dev_policy_free() on any remaining offloaded states.
191 Outcome of HW handling packets, the XFRM core can't count hard, soft limits.
192 The HW/driver are responsible to perform it and provide accurate data when
193 xdo_dev_state_update_curlft() is called. In case of one of these limits
194 occuried, the driver needs to call to xfrm_state_check_expire() to make sure
195 that XFRM performs rekeying sequence.