1 #ifndef _UAPI_LINUX_VIRTIO_RING_H 2 #define _UAPI_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 <linux/types.h> 35 #include <linux/virtio_types.h> 36 37 /* This marks a buffer as continuing via the next field. */ 38 #define VRING_DESC_F_NEXT 1 39 /* This marks a buffer as write-only (otherwise read-only). */ 40 #define VRING_DESC_F_WRITE 2 41 /* This means the buffer contains a list of buffer descriptors. */ 42 #define VRING_DESC_F_INDIRECT 4 43 44 /* 45 * Mark a descriptor as available or used in packed ring. 46 * Notice: they are defined as shifts instead of shifted values. 47 */ 48 #define VRING_PACKED_DESC_F_AVAIL 7 49 #define VRING_PACKED_DESC_F_USED 15 50 51 /* The Host uses this in used->flags to advise the Guest: don't kick me when 52 * you add a buffer. It's unreliable, so it's simply an optimization. Guest 53 * will still kick if it's out of buffers. */ 54 #define VRING_USED_F_NO_NOTIFY 1 55 /* The Guest uses this in avail->flags to advise the Host: don't interrupt me 56 * when you consume a buffer. It's unreliable, so it's simply an 57 * optimization. */ 58 #define VRING_AVAIL_F_NO_INTERRUPT 1 59 60 /* Enable events in packed ring. */ 61 #define VRING_PACKED_EVENT_FLAG_ENABLE 0x0 62 /* Disable events in packed ring. */ 63 #define VRING_PACKED_EVENT_FLAG_DISABLE 0x1 64 /* 65 * Enable events for a specific descriptor in packed ring. 66 * (as specified by Descriptor Ring Change Event Offset/Wrap Counter). 67 * Only valid if VIRTIO_RING_F_EVENT_IDX has been negotiated. 68 */ 69 #define VRING_PACKED_EVENT_FLAG_DESC 0x2 70 71 /* 72 * Wrap counter bit shift in event suppression structure 73 * of packed ring. 74 */ 75 #define VRING_PACKED_EVENT_F_WRAP_CTR 15 76 77 /* We support indirect buffer descriptors */ 78 #define VIRTIO_RING_F_INDIRECT_DESC 28 79 80 /* The Guest publishes the used index for which it expects an interrupt 81 * at the end of the avail ring. Host should ignore the avail->flags field. */ 82 /* The Host publishes the avail index for which it expects a kick 83 * at the end of the used ring. Guest should ignore the used->flags field. */ 84 #define VIRTIO_RING_F_EVENT_IDX 29 85 86 /* Alignment requirements for vring elements. 87 * When using pre-virtio 1.0 layout, these fall out naturally. 88 */ 89 #define VRING_AVAIL_ALIGN_SIZE 2 90 #define VRING_USED_ALIGN_SIZE 4 91 #define VRING_DESC_ALIGN_SIZE 16 92 93 /** 94 * struct vring_desc - Virtio ring descriptors, 95 * 16 bytes long. These can chain together via @next. 96 * 97 * @addr: buffer address (guest-physical) 98 * @len: buffer length 99 * @flags: descriptor flags 100 * @next: index of the next descriptor in the chain, 101 * if the VRING_DESC_F_NEXT flag is set. We chain unused 102 * descriptors via this, too. 103 */ 104 struct vring_desc { 105 __virtio64 addr; 106 __virtio32 len; 107 __virtio16 flags; 108 __virtio16 next; 109 }; 110 111 struct vring_avail { 112 __virtio16 flags; 113 __virtio16 idx; 114 __virtio16 ring[]; 115 }; 116 117 /* u32 is used here for ids for padding reasons. */ 118 struct vring_used_elem { 119 /* Index of start of used descriptor chain. */ 120 __virtio32 id; 121 /* Total length of the descriptor chain which was used (written to) */ 122 __virtio32 len; 123 }; 124 125 typedef struct vring_used_elem __attribute__((aligned(VRING_USED_ALIGN_SIZE))) 126 vring_used_elem_t; 127 128 struct vring_used { 129 __virtio16 flags; 130 __virtio16 idx; 131 vring_used_elem_t ring[]; 132 }; 133 134 /* 135 * The ring element addresses are passed between components with different 136 * alignments assumptions. Thus, we might need to decrease the compiler-selected 137 * alignment, and so must use a typedef to make sure the aligned attribute 138 * actually takes hold: 139 * 140 * https://gcc.gnu.org/onlinedocs//gcc/Common-Type-Attributes.html#Common-Type-Attributes 141 * 142 * When used on a struct, or struct member, the aligned attribute can only 143 * increase the alignment; in order to decrease it, the packed attribute must 144 * be specified as well. When used as part of a typedef, the aligned attribute 145 * can both increase and decrease alignment, and specifying the packed 146 * attribute generates a warning. 147 */ 148 typedef struct vring_desc __attribute__((aligned(VRING_DESC_ALIGN_SIZE))) 149 vring_desc_t; 150 typedef struct vring_avail __attribute__((aligned(VRING_AVAIL_ALIGN_SIZE))) 151 vring_avail_t; 152 typedef struct vring_used __attribute__((aligned(VRING_USED_ALIGN_SIZE))) 153 vring_used_t; 154 155 struct vring { 156 unsigned int num; 157 158 vring_desc_t *desc; 159 160 vring_avail_t *avail; 161 162 vring_used_t *used; 163 }; 164 165 #ifndef VIRTIO_RING_NO_LEGACY 166 167 /* The standard layout for the ring is a continuous chunk of memory which looks 168 * like this. We assume num is a power of 2. 169 * 170 * struct vring 171 * { 172 * // The actual descriptors (16 bytes each) 173 * struct vring_desc desc[num]; 174 * 175 * // A ring of available descriptor heads with free-running index. 176 * __virtio16 avail_flags; 177 * __virtio16 avail_idx; 178 * __virtio16 available[num]; 179 * __virtio16 used_event_idx; 180 * 181 * // Padding to the next align boundary. 182 * char pad[]; 183 * 184 * // A ring of used descriptor heads with free-running index. 185 * __virtio16 used_flags; 186 * __virtio16 used_idx; 187 * struct vring_used_elem used[num]; 188 * __virtio16 avail_event_idx; 189 * }; 190 */ 191 /* We publish the used event index at the end of the available ring, and vice 192 * versa. They are at the end for backwards compatibility. */ 193 #define vring_used_event(vr) ((vr)->avail->ring[(vr)->num]) 194 #define vring_avail_event(vr) (*(__virtio16 *)&(vr)->used->ring[(vr)->num]) 195 196 static inline void vring_init(struct vring *vr, unsigned int num, void *p, 197 unsigned long align) 198 { 199 vr->num = num; 200 vr->desc = p; 201 vr->avail = (struct vring_avail *)((char *)p + num * sizeof(struct vring_desc)); 202 vr->used = (void *)(((unsigned long)&vr->avail->ring[num] + sizeof(__virtio16) 203 + align-1) & ~(align - 1)); 204 } 205 206 static inline unsigned vring_size(unsigned int num, unsigned long align) 207 { 208 return ((sizeof(struct vring_desc) * num + sizeof(__virtio16) * (3 + num) 209 + align - 1) & ~(align - 1)) 210 + sizeof(__virtio16) * 3 + sizeof(struct vring_used_elem) * num; 211 } 212 213 #endif /* VIRTIO_RING_NO_LEGACY */ 214 215 /* The following is used with USED_EVENT_IDX and AVAIL_EVENT_IDX */ 216 /* Assuming a given event_idx value from the other side, if 217 * we have just incremented index from old to new_idx, 218 * should we trigger an event? */ 219 static inline int vring_need_event(__u16 event_idx, __u16 new_idx, __u16 old) 220 { 221 /* Note: Xen has similar logic for notification hold-off 222 * in include/xen/interface/io/ring.h with req_event and req_prod 223 * corresponding to event_idx + 1 and new_idx respectively. 224 * Note also that req_event and req_prod in Xen start at 1, 225 * event indexes in virtio start at 0. */ 226 return (__u16)(new_idx - event_idx - 1) < (__u16)(new_idx - old); 227 } 228 229 struct vring_packed_desc_event { 230 /* Descriptor Ring Change Event Offset/Wrap Counter. */ 231 __le16 off_wrap; 232 /* Descriptor Ring Change Event Flags. */ 233 __le16 flags; 234 }; 235 236 struct vring_packed_desc { 237 /* Buffer Address. */ 238 __le64 addr; 239 /* Buffer Length. */ 240 __le32 len; 241 /* Buffer ID. */ 242 __le16 id; 243 /* The flags depending on descriptor type. */ 244 __le16 flags; 245 }; 246 247 #endif /* _UAPI_LINUX_VIRTIO_RING_H */ 248