1#ifndef _LINUX_VIRTIO_RING_H 2#define _LINUX_VIRTIO_RING_H 3/* An interface for efficient virtio implementation, currently for use by KVM, 4 * but hopefully others soon. Do NOT change this since it will 5 * break existing servers and clients. 6 * 7 * This header is BSD licensed so anyone can use the definitions to implement 8 * compatible drivers/servers. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. Neither the name of IBM nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * Copyright Rusty Russell IBM Corporation 2007. */ 34#include <stdint.h> 35#include "standard-headers/linux/types.h" 36#include "standard-headers/linux/virtio_types.h" 37 38/* This marks a buffer as continuing via the next field. */ 39#define VRING_DESC_F_NEXT 1 40/* This marks a buffer as write-only (otherwise read-only). */ 41#define VRING_DESC_F_WRITE 2 42/* This means the buffer contains a list of buffer descriptors. */ 43#define VRING_DESC_F_INDIRECT 4 44 45/* 46 * Mark a descriptor as available or used in packed ring. 47 * Notice: they are defined as shifts instead of shifted values. 48 */ 49#define VRING_PACKED_DESC_F_AVAIL 7 50#define VRING_PACKED_DESC_F_USED 15 51 52/* The Host uses this in used->flags to advise the Guest: don't kick me when 53 * you add a buffer. It's unreliable, so it's simply an optimization. Guest 54 * will still kick if it's out of buffers. */ 55#define VRING_USED_F_NO_NOTIFY 1 56/* The Guest uses this in avail->flags to advise the Host: don't interrupt me 57 * when you consume a buffer. It's unreliable, so it's simply an 58 * optimization. */ 59#define VRING_AVAIL_F_NO_INTERRUPT 1 60 61/* Enable events in packed ring. */ 62#define VRING_PACKED_EVENT_FLAG_ENABLE 0x0 63/* Disable events in packed ring. */ 64#define VRING_PACKED_EVENT_FLAG_DISABLE 0x1 65/* 66 * Enable events for a specific descriptor in packed ring. 67 * (as specified by Descriptor Ring Change Event Offset/Wrap Counter). 68 * Only valid if VIRTIO_RING_F_EVENT_IDX has been negotiated. 69 */ 70#define VRING_PACKED_EVENT_FLAG_DESC 0x2 71 72/* 73 * Wrap counter bit shift in event suppression structure 74 * of packed ring. 75 */ 76#define VRING_PACKED_EVENT_F_WRAP_CTR 15 77 78/* We support indirect buffer descriptors */ 79#define VIRTIO_RING_F_INDIRECT_DESC 28 80 81/* The Guest publishes the used index for which it expects an interrupt 82 * at the end of the avail ring. Host should ignore the avail->flags field. */ 83/* The Host publishes the avail index for which it expects a kick 84 * at the end of the used ring. Guest should ignore the used->flags field. */ 85#define VIRTIO_RING_F_EVENT_IDX 29 86 87/* Alignment requirements for vring elements. 88 * When using pre-virtio 1.0 layout, these fall out naturally. 89 */ 90#define VRING_AVAIL_ALIGN_SIZE 2 91#define VRING_USED_ALIGN_SIZE 4 92#define VRING_DESC_ALIGN_SIZE 16 93 94/** 95 * struct vring_desc - Virtio ring descriptors, 96 * 16 bytes long. These can chain together via @next. 97 * 98 * @addr: buffer address (guest-physical) 99 * @len: buffer length 100 * @flags: descriptor flags 101 * @next: index of the next descriptor in the chain, 102 * if the VRING_DESC_F_NEXT flag is set. We chain unused 103 * descriptors via this, too. 104 */ 105struct vring_desc { 106 __virtio64 addr; 107 __virtio32 len; 108 __virtio16 flags; 109 __virtio16 next; 110}; 111 112struct vring_avail { 113 __virtio16 flags; 114 __virtio16 idx; 115 __virtio16 ring[]; 116}; 117 118/* uint32_t is used here for ids for padding reasons. */ 119struct vring_used_elem { 120 /* Index of start of used descriptor chain. */ 121 __virtio32 id; 122 /* Total length of the descriptor chain which was used (written to) */ 123 __virtio32 len; 124}; 125 126typedef struct vring_used_elem __attribute__((aligned(VRING_USED_ALIGN_SIZE))) 127 vring_used_elem_t; 128 129struct vring_used { 130 __virtio16 flags; 131 __virtio16 idx; 132 vring_used_elem_t ring[]; 133}; 134 135/* 136 * The ring element addresses are passed between components with different 137 * alignments assumptions. Thus, we might need to decrease the compiler-selected 138 * alignment, and so must use a typedef to make sure the aligned attribute 139 * actually takes hold: 140 * 141 * https://gcc.gnu.org/onlinedocs//gcc/Common-Type-Attributes.html#Common-Type-Attributes 142 * 143 * When used on a struct, or struct member, the aligned attribute can only 144 * increase the alignment; in order to decrease it, the packed attribute must 145 * be specified as well. When used as part of a typedef, the aligned attribute 146 * can both increase and decrease alignment, and specifying the packed 147 * attribute generates a warning. 148 */ 149typedef struct vring_desc __attribute__((aligned(VRING_DESC_ALIGN_SIZE))) 150 vring_desc_t; 151typedef struct vring_avail __attribute__((aligned(VRING_AVAIL_ALIGN_SIZE))) 152 vring_avail_t; 153typedef struct vring_used __attribute__((aligned(VRING_USED_ALIGN_SIZE))) 154 vring_used_t; 155 156struct vring { 157 unsigned int num; 158 159 vring_desc_t *desc; 160 161 vring_avail_t *avail; 162 163 vring_used_t *used; 164}; 165 166#ifndef VIRTIO_RING_NO_LEGACY 167 168/* The standard layout for the ring is a continuous chunk of memory which looks 169 * like this. We assume num is a power of 2. 170 * 171 * struct vring 172 * { 173 * // The actual descriptors (16 bytes each) 174 * struct vring_desc desc[num]; 175 * 176 * // A ring of available descriptor heads with free-running index. 177 * __virtio16 avail_flags; 178 * __virtio16 avail_idx; 179 * __virtio16 available[num]; 180 * __virtio16 used_event_idx; 181 * 182 * // Padding to the next align boundary. 183 * char pad[]; 184 * 185 * // A ring of used descriptor heads with free-running index. 186 * __virtio16 used_flags; 187 * __virtio16 used_idx; 188 * struct vring_used_elem used[num]; 189 * __virtio16 avail_event_idx; 190 * }; 191 */ 192/* We publish the used event index at the end of the available ring, and vice 193 * versa. They are at the end for backwards compatibility. */ 194#define vring_used_event(vr) ((vr)->avail->ring[(vr)->num]) 195#define vring_avail_event(vr) (*(__virtio16 *)&(vr)->used->ring[(vr)->num]) 196 197static inline void vring_init(struct vring *vr, unsigned int num, void *p, 198 unsigned long align) 199{ 200 vr->num = num; 201 vr->desc = p; 202 vr->avail = (struct vring_avail *)((char *)p + num * sizeof(struct vring_desc)); 203 vr->used = (void *)(((uintptr_t)&vr->avail->ring[num] + sizeof(__virtio16) 204 + align-1) & ~(align - 1)); 205} 206 207static inline unsigned vring_size(unsigned int num, unsigned long align) 208{ 209 return ((sizeof(struct vring_desc) * num + sizeof(__virtio16) * (3 + num) 210 + align - 1) & ~(align - 1)) 211 + sizeof(__virtio16) * 3 + sizeof(struct vring_used_elem) * num; 212} 213 214#endif /* VIRTIO_RING_NO_LEGACY */ 215 216/* The following is used with USED_EVENT_IDX and AVAIL_EVENT_IDX */ 217/* Assuming a given event_idx value from the other side, if 218 * we have just incremented index from old to new_idx, 219 * should we trigger an event? */ 220static inline int vring_need_event(uint16_t event_idx, uint16_t new_idx, uint16_t old) 221{ 222 /* Note: Xen has similar logic for notification hold-off 223 * in include/xen/interface/io/ring.h with req_event and req_prod 224 * corresponding to event_idx + 1 and new_idx respectively. 225 * Note also that req_event and req_prod in Xen start at 1, 226 * event indexes in virtio start at 0. */ 227 return (uint16_t)(new_idx - event_idx - 1) < (uint16_t)(new_idx - old); 228} 229 230struct vring_packed_desc_event { 231 /* Descriptor Ring Change Event Offset/Wrap Counter. */ 232 uint16_t off_wrap; 233 /* Descriptor Ring Change Event Flags. */ 234 uint16_t flags; 235}; 236 237struct vring_packed_desc { 238 /* Buffer Address. */ 239 uint64_t addr; 240 /* Buffer Length. */ 241 uint32_t len; 242 /* Buffer ID. */ 243 uint16_t id; 244 /* The flags depending on descriptor type. */ 245 uint16_t flags; 246}; 247 248#endif /* _LINUX_VIRTIO_RING_H */ 249