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 #ifdef RTE_VIRTIO_USER
71 * Return the physical address (or virtual address in case of
72 * virtio-user) of mbuf data buffer.
74 #define VIRTIO_MBUF_ADDR(mb, vq) (*(uint64_t *)((uintptr_t)(mb) + (vq)->offset))
76 #define VIRTIO_MBUF_ADDR(mb, vq) ((mb)->buf_physaddr)
80 * Return the physical address (or virtual address in case of
81 * virtio-user) of mbuf data buffer, taking care of mbuf data offset
83 #define VIRTIO_MBUF_DATA_DMA_ADDR(mb, vq) \
84 (VIRTIO_MBUF_ADDR(mb, vq) + (mb)->data_off)
86 #define VTNET_SQ_RQ_QUEUE_IDX 0
87 #define VTNET_SQ_TQ_QUEUE_IDX 1
88 #define VTNET_SQ_CQ_QUEUE_IDX 2
90 enum { VTNET_RQ = 0, VTNET_TQ = 1, VTNET_CQ = 2 };
92 * The maximum virtqueue size is 2^15. Use that value as the end of
93 * descriptor chain terminator since it will never be a valid index
94 * in the descriptor table. This is used to verify we are correctly
95 * handling vq_free_cnt.
97 #define VQ_RING_DESC_CHAIN_END 32768
100 * Control the RX mode, ie. promiscuous, allmulti, etc...
101 * All commands require an "out" sg entry containing a 1 byte
102 * state value, zero = disable, non-zero = enable. Commands
103 * 0 and 1 are supported with the VIRTIO_NET_F_CTRL_RX feature.
104 * Commands 2-5 are added with VIRTIO_NET_F_CTRL_RX_EXTRA.
106 #define VIRTIO_NET_CTRL_RX 0
107 #define VIRTIO_NET_CTRL_RX_PROMISC 0
108 #define VIRTIO_NET_CTRL_RX_ALLMULTI 1
109 #define VIRTIO_NET_CTRL_RX_ALLUNI 2
110 #define VIRTIO_NET_CTRL_RX_NOMULTI 3
111 #define VIRTIO_NET_CTRL_RX_NOUNI 4
112 #define VIRTIO_NET_CTRL_RX_NOBCAST 5
117 * The MAC filter table is managed by the hypervisor, the guest should
118 * assume the size is infinite. Filtering should be considered
119 * non-perfect, ie. based on hypervisor resources, the guest may
120 * received packets from sources not specified in the filter list.
122 * In addition to the class/cmd header, the TABLE_SET command requires
123 * two out scatterlists. Each contains a 4 byte count of entries followed
124 * by a concatenated byte stream of the ETH_ALEN MAC addresses. The
125 * first sg list contains unicast addresses, the second is for multicast.
126 * This functionality is present if the VIRTIO_NET_F_CTRL_RX feature
129 * The ADDR_SET command requests one out scatterlist, it contains a
130 * 6 bytes MAC address. This functionality is present if the
131 * VIRTIO_NET_F_CTRL_MAC_ADDR feature is available.
133 struct virtio_net_ctrl_mac {
135 uint8_t macs[][ETHER_ADDR_LEN];
136 } __attribute__((__packed__));
138 #define VIRTIO_NET_CTRL_MAC 1
139 #define VIRTIO_NET_CTRL_MAC_TABLE_SET 0
140 #define VIRTIO_NET_CTRL_MAC_ADDR_SET 1
143 * Control VLAN filtering
145 * The VLAN filter table is controlled via a simple ADD/DEL interface.
146 * VLAN IDs not added may be filtered by the hypervisor. Del is the
147 * opposite of add. Both commands expect an out entry containing a 2
148 * byte VLAN ID. VLAN filtering is available with the
149 * VIRTIO_NET_F_CTRL_VLAN feature bit.
151 #define VIRTIO_NET_CTRL_VLAN 2
152 #define VIRTIO_NET_CTRL_VLAN_ADD 0
153 #define VIRTIO_NET_CTRL_VLAN_DEL 1
155 struct virtio_net_ctrl_hdr {
158 } __attribute__((packed));
160 typedef uint8_t virtio_net_ctrl_ack;
162 #define VIRTIO_NET_OK 0
163 #define VIRTIO_NET_ERR 1
165 #define VIRTIO_MAX_CTRL_DATA 2048
167 struct virtio_pmd_ctrl {
168 struct virtio_net_ctrl_hdr hdr;
169 virtio_net_ctrl_ack status;
170 uint8_t data[VIRTIO_MAX_CTRL_DATA];
173 struct vq_desc_extra {
179 struct virtio_hw *hw; /**< virtio_hw structure pointer. */
180 struct vring vq_ring; /**< vring keeping desc, used and avail */
182 * Last consumed descriptor in the used table,
183 * trails vq_ring.used->idx.
185 uint16_t vq_used_cons_idx;
186 uint16_t vq_nentries; /**< vring desc numbers */
187 uint16_t vq_free_cnt; /**< num of desc available */
188 uint16_t vq_avail_idx; /**< sync until needed */
189 uint16_t vq_free_thresh; /**< free threshold */
191 void *vq_ring_virt_mem; /**< linear address of vring*/
192 unsigned int vq_ring_size;
194 phys_addr_t vq_ring_mem; /**< physical address of vring,
195 * or virtual address for virtio_user. */
198 * Head of the free chain in the descriptor table. If
199 * there are no free descriptors, this will be set to
200 * VQ_RING_DESC_CHAIN_END.
202 uint16_t vq_desc_head_idx;
203 uint16_t vq_desc_tail_idx;
204 uint16_t vq_queue_index; /**< PCI queue index */
205 uint16_t offset; /**< relative offset to obtain addr in mbuf */
206 uint16_t *notify_addr;
208 struct rte_mbuf **sw_ring; /**< RX software ring. */
209 struct vq_desc_extra vq_descx[0];
212 /* If multiqueue is provided by host, then we suppport it. */
213 #define VIRTIO_NET_CTRL_MQ 4
214 #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET 0
215 #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN 1
216 #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MAX 0x8000
218 #define VIRTIO_NET_CTRL_MAC_ADDR_SET 1
221 * This is the first element of the scatter-gather list. If you don't
222 * specify GSO or CSUM features, you can simply ignore the header.
224 struct virtio_net_hdr {
225 #define VIRTIO_NET_HDR_F_NEEDS_CSUM 1 /**< Use csum_start,csum_offset*/
226 #define VIRTIO_NET_HDR_F_DATA_VALID 2 /**< Checksum is valid */
228 #define VIRTIO_NET_HDR_GSO_NONE 0 /**< Not a GSO frame */
229 #define VIRTIO_NET_HDR_GSO_TCPV4 1 /**< GSO frame, IPv4 TCP (TSO) */
230 #define VIRTIO_NET_HDR_GSO_UDP 3 /**< GSO frame, IPv4 UDP (UFO) */
231 #define VIRTIO_NET_HDR_GSO_TCPV6 4 /**< GSO frame, IPv6 TCP */
232 #define VIRTIO_NET_HDR_GSO_ECN 0x80 /**< TCP has ECN set */
234 uint16_t hdr_len; /**< Ethernet + IP + tcp/udp hdrs */
235 uint16_t gso_size; /**< Bytes to append to hdr_len per frame */
236 uint16_t csum_start; /**< Position to start checksumming from */
237 uint16_t csum_offset; /**< Offset after that to place checksum */
241 * This is the version of the header to use when the MRG_RXBUF
242 * feature has been negotiated.
244 struct virtio_net_hdr_mrg_rxbuf {
245 struct virtio_net_hdr hdr;
246 uint16_t num_buffers; /**< Number of merged rx buffers */
249 /* Region reserved to allow for transmit header and indirect ring */
250 #define VIRTIO_MAX_TX_INDIRECT 8
251 struct virtio_tx_region {
252 struct virtio_net_hdr_mrg_rxbuf tx_hdr;
253 struct vring_desc tx_indir[VIRTIO_MAX_TX_INDIRECT]
254 __attribute__((__aligned__(16)));
257 /* Chain all the descriptors in the ring with an END */
259 vring_desc_init(struct vring_desc *dp, uint16_t n)
263 for (i = 0; i < n - 1; i++)
264 dp[i].next = (uint16_t)(i + 1);
265 dp[i].next = VQ_RING_DESC_CHAIN_END;
269 * Tell the backend not to interrupt us.
271 void virtqueue_disable_intr(struct virtqueue *vq);
273 * Dump virtqueue internal structures, for debug purpose only.
275 void virtqueue_dump(struct virtqueue *vq);
277 * Get all mbufs to be freed.
279 struct rte_mbuf *virtqueue_detatch_unused(struct virtqueue *vq);
282 virtqueue_full(const struct virtqueue *vq)
284 return vq->vq_free_cnt == 0;
287 #define VIRTQUEUE_NUSED(vq) ((uint16_t)((vq)->vq_ring.used->idx - (vq)->vq_used_cons_idx))
290 vq_update_avail_idx(struct virtqueue *vq)
293 vq->vq_ring.avail->idx = vq->vq_avail_idx;
297 vq_update_avail_ring(struct virtqueue *vq, uint16_t desc_idx)
301 * Place the head of the descriptor chain into the next slot and make
302 * it usable to the host. The chain is made available now rather than
303 * deferring to virtqueue_notify() in the hopes that if the host is
304 * currently running on another CPU, we can keep it processing the new
307 avail_idx = (uint16_t)(vq->vq_avail_idx & (vq->vq_nentries - 1));
308 if (unlikely(vq->vq_ring.avail->ring[avail_idx] != desc_idx))
309 vq->vq_ring.avail->ring[avail_idx] = desc_idx;
314 virtqueue_kick_prepare(struct virtqueue *vq)
316 return !(vq->vq_ring.used->flags & VRING_USED_F_NO_NOTIFY);
320 virtqueue_notify(struct virtqueue *vq)
323 * Ensure updated avail->idx is visible to host.
324 * For virtio on IA, the notificaiton is through io port operation
325 * which is a serialization instruction itself.
327 vq->hw->vtpci_ops->notify_queue(vq->hw, vq);
330 #ifdef RTE_LIBRTE_VIRTIO_DEBUG_DUMP
331 #define VIRTQUEUE_DUMP(vq) do { \
332 uint16_t used_idx, nused; \
333 used_idx = (vq)->vq_ring.used->idx; \
334 nused = (uint16_t)(used_idx - (vq)->vq_used_cons_idx); \
335 PMD_INIT_LOG(DEBUG, \
336 "VQ: - size=%d; free=%d; used=%d; desc_head_idx=%d;" \
337 " avail.idx=%d; used_cons_idx=%d; used.idx=%d;" \
338 " avail.flags=0x%x; used.flags=0x%x", \
339 (vq)->vq_nentries, (vq)->vq_free_cnt, nused, \
340 (vq)->vq_desc_head_idx, (vq)->vq_ring.avail->idx, \
341 (vq)->vq_used_cons_idx, (vq)->vq_ring.used->idx, \
342 (vq)->vq_ring.avail->flags, (vq)->vq_ring.used->flags); \
345 #define VIRTQUEUE_DUMP(vq) do { } while (0)
348 #endif /* _VIRTQUEUE_H_ */