4 * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
17 * * Neither the name of Intel Corporation nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
39 #include <rte_atomic.h>
40 #include <rte_memory.h>
41 #include <rte_memzone.h>
42 #include <rte_mempool.h>
44 #include "virtio_pci.h"
45 #include "virtio_ring.h"
46 #include "virtio_logs.h"
51 * Per virtio_config.h in Linux.
52 * For virtio_pci on SMP, we don't need to order with respect to MMIO
53 * accesses through relaxed memory I/O windows, so smp_mb() et al are
57 #define virtio_mb() rte_smp_mb()
58 #define virtio_rmb() rte_smp_rmb()
59 #define virtio_wmb() rte_smp_wmb()
61 #ifdef RTE_PMD_PACKET_PREFETCH
62 #define rte_packet_prefetch(p) rte_prefetch1(p)
64 #define rte_packet_prefetch(p) do {} while(0)
67 #define VIRTQUEUE_MAX_NAME_SZ 32
69 #define VTNET_SQ_RQ_QUEUE_IDX 0
70 #define VTNET_SQ_TQ_QUEUE_IDX 1
71 #define VTNET_SQ_CQ_QUEUE_IDX 2
73 enum { VTNET_RQ = 0, VTNET_TQ = 1, VTNET_CQ = 2 };
75 * The maximum virtqueue size is 2^15. Use that value as the end of
76 * descriptor chain terminator since it will never be a valid index
77 * in the descriptor table. This is used to verify we are correctly
78 * handling vq_free_cnt.
80 #define VQ_RING_DESC_CHAIN_END 32768
83 * Control the RX mode, ie. promiscuous, allmulti, etc...
84 * All commands require an "out" sg entry containing a 1 byte
85 * state value, zero = disable, non-zero = enable. Commands
86 * 0 and 1 are supported with the VIRTIO_NET_F_CTRL_RX feature.
87 * Commands 2-5 are added with VIRTIO_NET_F_CTRL_RX_EXTRA.
89 #define VIRTIO_NET_CTRL_RX 0
90 #define VIRTIO_NET_CTRL_RX_PROMISC 0
91 #define VIRTIO_NET_CTRL_RX_ALLMULTI 1
92 #define VIRTIO_NET_CTRL_RX_ALLUNI 2
93 #define VIRTIO_NET_CTRL_RX_NOMULTI 3
94 #define VIRTIO_NET_CTRL_RX_NOUNI 4
95 #define VIRTIO_NET_CTRL_RX_NOBCAST 5
100 * The MAC filter table is managed by the hypervisor, the guest should
101 * assume the size is infinite. Filtering should be considered
102 * non-perfect, ie. based on hypervisor resources, the guest may
103 * received packets from sources not specified in the filter list.
105 * In addition to the class/cmd header, the TABLE_SET command requires
106 * two out scatterlists. Each contains a 4 byte count of entries followed
107 * by a concatenated byte stream of the ETH_ALEN MAC addresses. The
108 * first sg list contains unicast addresses, the second is for multicast.
109 * This functionality is present if the VIRTIO_NET_F_CTRL_RX feature
112 * The ADDR_SET command requests one out scatterlist, it contains a
113 * 6 bytes MAC address. This functionality is present if the
114 * VIRTIO_NET_F_CTRL_MAC_ADDR feature is available.
116 struct virtio_net_ctrl_mac {
118 uint8_t macs[][ETHER_ADDR_LEN];
119 } __attribute__((__packed__));
121 #define VIRTIO_NET_CTRL_MAC 1
122 #define VIRTIO_NET_CTRL_MAC_TABLE_SET 0
123 #define VIRTIO_NET_CTRL_MAC_ADDR_SET 1
126 * Control VLAN filtering
128 * The VLAN filter table is controlled via a simple ADD/DEL interface.
129 * VLAN IDs not added may be filtered by the hypervisor. Del is the
130 * opposite of add. Both commands expect an out entry containing a 2
131 * byte VLAN ID. VLAN filtering is available with the
132 * VIRTIO_NET_F_CTRL_VLAN feature bit.
134 #define VIRTIO_NET_CTRL_VLAN 2
135 #define VIRTIO_NET_CTRL_VLAN_ADD 0
136 #define VIRTIO_NET_CTRL_VLAN_DEL 1
138 struct virtio_net_ctrl_hdr {
141 } __attribute__((packed));
143 typedef uint8_t virtio_net_ctrl_ack;
145 #define VIRTIO_NET_OK 0
146 #define VIRTIO_NET_ERR 1
148 #define VIRTIO_MAX_CTRL_DATA 2048
150 struct virtio_pmd_ctrl {
151 struct virtio_net_ctrl_hdr hdr;
152 virtio_net_ctrl_ack status;
153 uint8_t data[VIRTIO_MAX_CTRL_DATA];
156 struct vq_desc_extra {
162 struct virtio_hw *hw; /**< virtio_hw structure pointer. */
163 struct vring vq_ring; /**< vring keeping desc, used and avail */
165 * Last consumed descriptor in the used table,
166 * trails vq_ring.used->idx.
168 uint16_t vq_used_cons_idx;
169 uint16_t vq_nentries; /**< vring desc numbers */
170 uint16_t vq_free_cnt; /**< num of desc available */
171 uint16_t vq_avail_idx; /**< sync until needed */
172 uint16_t vq_free_thresh; /**< free threshold */
174 void *vq_ring_virt_mem; /**< linear address of vring*/
175 unsigned int vq_ring_size;
177 phys_addr_t vq_ring_mem; /**< physical address of vring */
180 * Head of the free chain in the descriptor table. If
181 * there are no free descriptors, this will be set to
182 * VQ_RING_DESC_CHAIN_END.
184 uint16_t vq_desc_head_idx;
185 uint16_t vq_desc_tail_idx;
186 uint16_t vq_queue_index; /**< PCI queue index */
187 uint16_t *notify_addr;
189 struct rte_mbuf **sw_ring; /**< RX software ring. */
190 struct vq_desc_extra vq_descx[0];
193 /* If multiqueue is provided by host, then we suppport it. */
194 #define VIRTIO_NET_CTRL_MQ 4
195 #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET 0
196 #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN 1
197 #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MAX 0x8000
199 #define VIRTIO_NET_CTRL_MAC_ADDR_SET 1
202 * This is the first element of the scatter-gather list. If you don't
203 * specify GSO or CSUM features, you can simply ignore the header.
205 struct virtio_net_hdr {
206 #define VIRTIO_NET_HDR_F_NEEDS_CSUM 1 /**< Use csum_start,csum_offset*/
208 #define VIRTIO_NET_HDR_GSO_NONE 0 /**< Not a GSO frame */
209 #define VIRTIO_NET_HDR_GSO_TCPV4 1 /**< GSO frame, IPv4 TCP (TSO) */
210 #define VIRTIO_NET_HDR_GSO_UDP 3 /**< GSO frame, IPv4 UDP (UFO) */
211 #define VIRTIO_NET_HDR_GSO_TCPV6 4 /**< GSO frame, IPv6 TCP */
212 #define VIRTIO_NET_HDR_GSO_ECN 0x80 /**< TCP has ECN set */
214 uint16_t hdr_len; /**< Ethernet + IP + tcp/udp hdrs */
215 uint16_t gso_size; /**< Bytes to append to hdr_len per frame */
216 uint16_t csum_start; /**< Position to start checksumming from */
217 uint16_t csum_offset; /**< Offset after that to place checksum */
221 * This is the version of the header to use when the MRG_RXBUF
222 * feature has been negotiated.
224 struct virtio_net_hdr_mrg_rxbuf {
225 struct virtio_net_hdr hdr;
226 uint16_t num_buffers; /**< Number of merged rx buffers */
229 /* Region reserved to allow for transmit header and indirect ring */
230 #define VIRTIO_MAX_TX_INDIRECT 8
231 struct virtio_tx_region {
232 struct virtio_net_hdr_mrg_rxbuf tx_hdr;
233 struct vring_desc tx_indir[VIRTIO_MAX_TX_INDIRECT]
234 __attribute__((__aligned__(16)));
237 /* Chain all the descriptors in the ring with an END */
239 vring_desc_init(struct vring_desc *dp, uint16_t n)
243 for (i = 0; i < n - 1; i++)
244 dp[i].next = (uint16_t)(i + 1);
245 dp[i].next = VQ_RING_DESC_CHAIN_END;
249 * Tell the backend not to interrupt us.
251 void virtqueue_disable_intr(struct virtqueue *vq);
253 * Dump virtqueue internal structures, for debug purpose only.
255 void virtqueue_dump(struct virtqueue *vq);
257 * Get all mbufs to be freed.
259 struct rte_mbuf *virtqueue_detatch_unused(struct virtqueue *vq);
262 virtqueue_full(const struct virtqueue *vq)
264 return vq->vq_free_cnt == 0;
267 #define VIRTQUEUE_NUSED(vq) ((uint16_t)((vq)->vq_ring.used->idx - (vq)->vq_used_cons_idx))
270 vq_update_avail_idx(struct virtqueue *vq)
273 vq->vq_ring.avail->idx = vq->vq_avail_idx;
277 vq_update_avail_ring(struct virtqueue *vq, uint16_t desc_idx)
281 * Place the head of the descriptor chain into the next slot and make
282 * it usable to the host. The chain is made available now rather than
283 * deferring to virtqueue_notify() in the hopes that if the host is
284 * currently running on another CPU, we can keep it processing the new
287 avail_idx = (uint16_t)(vq->vq_avail_idx & (vq->vq_nentries - 1));
288 if (unlikely(vq->vq_ring.avail->ring[avail_idx] != desc_idx))
289 vq->vq_ring.avail->ring[avail_idx] = desc_idx;
294 virtqueue_kick_prepare(struct virtqueue *vq)
296 return !(vq->vq_ring.used->flags & VRING_USED_F_NO_NOTIFY);
300 virtqueue_notify(struct virtqueue *vq)
303 * Ensure updated avail->idx is visible to host.
304 * For virtio on IA, the notificaiton is through io port operation
305 * which is a serialization instruction itself.
307 vq->hw->vtpci_ops->notify_queue(vq->hw, vq);
310 #ifdef RTE_LIBRTE_VIRTIO_DEBUG_DUMP
311 #define VIRTQUEUE_DUMP(vq) do { \
312 uint16_t used_idx, nused; \
313 used_idx = (vq)->vq_ring.used->idx; \
314 nused = (uint16_t)(used_idx - (vq)->vq_used_cons_idx); \
315 PMD_INIT_LOG(DEBUG, \
316 "VQ: - size=%d; free=%d; used=%d; desc_head_idx=%d;" \
317 " avail.idx=%d; used_cons_idx=%d; used.idx=%d;" \
318 " avail.flags=0x%x; used.flags=0x%x", \
319 (vq)->vq_nentries, (vq)->vq_free_cnt, nused, \
320 (vq)->vq_desc_head_idx, (vq)->vq_ring.avail->idx, \
321 (vq)->vq_used_cons_idx, (vq)->vq_ring.used->idx, \
322 (vq)->vq_ring.avail->flags, (vq)->vq_ring.used->flags); \
325 #define VIRTQUEUE_DUMP(vq) do { } while (0)
328 #endif /* _VIRTQUEUE_H_ */