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#ifndef __KERNEL__
35#include <stdint.h>
36#endif
37#include <linux/types.h>
38#include <linux/virtio_types.h>
39
40/* This marks a buffer as continuing via the next field. */
41#define VRING_DESC_F_NEXT 1
42/* This marks a buffer as write-only (otherwise read-only). */
43#define VRING_DESC_F_WRITE 2
44/* This means the buffer contains a list of buffer descriptors. */
45#define VRING_DESC_F_INDIRECT 4
46
47/*
48 * Mark a descriptor as available or used in packed ring.
49 * Notice: they are defined as shifts instead of shifted values.
50 */
51#define VRING_PACKED_DESC_F_AVAIL 7
52#define VRING_PACKED_DESC_F_USED 15
53
54/* The Host uses this in used->flags to advise the Guest: don't kick me when
55 * you add a buffer. It's unreliable, so it's simply an optimization. Guest
56 * will still kick if it's out of buffers. */
57#define VRING_USED_F_NO_NOTIFY 1
58/* The Guest uses this in avail->flags to advise the Host: don't interrupt me
59 * when you consume a buffer. It's unreliable, so it's simply an
60 * optimization. */
61#define VRING_AVAIL_F_NO_INTERRUPT 1
62
63/* Enable events in packed ring. */
64#define VRING_PACKED_EVENT_FLAG_ENABLE 0x0
65/* Disable events in packed ring. */
66#define VRING_PACKED_EVENT_FLAG_DISABLE 0x1
67/*
68 * Enable events for a specific descriptor in packed ring.
69 * (as specified by Descriptor Ring Change Event Offset/Wrap Counter).
70 * Only valid if VIRTIO_RING_F_EVENT_IDX has been negotiated.
71 */
72#define VRING_PACKED_EVENT_FLAG_DESC 0x2
73
74/*
75 * Wrap counter bit shift in event suppression structure
76 * of packed ring.
77 */
78#define VRING_PACKED_EVENT_F_WRAP_CTR 15
79
80/* We support indirect buffer descriptors */
81#define VIRTIO_RING_F_INDIRECT_DESC 28
82
83/* The Guest publishes the used index for which it expects an interrupt
84 * at the end of the avail ring. Host should ignore the avail->flags field. */
85/* The Host publishes the avail index for which it expects a kick
86 * at the end of the used ring. Guest should ignore the used->flags field. */
87#define VIRTIO_RING_F_EVENT_IDX 29
88
89/* Alignment requirements for vring elements.
90 * When using pre-virtio 1.0 layout, these fall out naturally.
91 */
92#define VRING_AVAIL_ALIGN_SIZE 2
93#define VRING_USED_ALIGN_SIZE 4
94#define VRING_DESC_ALIGN_SIZE 16
95
96/**
97 * struct vring_desc - Virtio ring descriptors,
98 * 16 bytes long. These can chain together via @next.
99 *
100 * @addr: buffer address (guest-physical)
101 * @len: buffer length
102 * @flags: descriptor flags
103 * @next: index of the next descriptor in the chain,
104 * if the VRING_DESC_F_NEXT flag is set. We chain unused
105 * descriptors via this, too.
106 */
107struct vring_desc {
108 __virtio64 addr;
109 __virtio32 len;
110 __virtio16 flags;
111 __virtio16 next;
112};
113
114struct vring_avail {
115 __virtio16 flags;
116 __virtio16 idx;
117 __virtio16 ring[];
118};
119
120/* u32 is used here for ids for padding reasons. */
121struct vring_used_elem {
122 /* Index of start of used descriptor chain. */
123 __virtio32 id;
124 /* Total length of the descriptor chain which was used (written to) */
125 __virtio32 len;
126};
127
128typedef struct vring_used_elem __attribute__((aligned(VRING_USED_ALIGN_SIZE)))
129 vring_used_elem_t;
130
131struct vring_used {
132 __virtio16 flags;
133 __virtio16 idx;
134 vring_used_elem_t ring[];
135};
136
137/*
138 * The ring element addresses are passed between components with different
139 * alignments assumptions. Thus, we might need to decrease the compiler-selected
140 * alignment, and so must use a typedef to make sure the aligned attribute
141 * actually takes hold:
142 *
143 * https://gcc.gnu.org/onlinedocs//gcc/Common-Type-Attributes.html#Common-Type-Attributes
144 *
145 * When used on a struct, or struct member, the aligned attribute can only
146 * increase the alignment; in order to decrease it, the packed attribute must
147 * be specified as well. When used as part of a typedef, the aligned attribute
148 * can both increase and decrease alignment, and specifying the packed
149 * attribute generates a warning.
150 */
151typedef struct vring_desc __attribute__((aligned(VRING_DESC_ALIGN_SIZE)))
152 vring_desc_t;
153typedef struct vring_avail __attribute__((aligned(VRING_AVAIL_ALIGN_SIZE)))
154 vring_avail_t;
155typedef struct vring_used __attribute__((aligned(VRING_USED_ALIGN_SIZE)))
156 vring_used_t;
157
158struct vring {
159 unsigned int num;
160
161 vring_desc_t *desc;
162
163 vring_avail_t *avail;
164
165 vring_used_t *used;
166};
167
168#ifndef VIRTIO_RING_NO_LEGACY
169
170/* The standard layout for the ring is a continuous chunk of memory which looks
171 * like this. We assume num is a power of 2.
172 *
173 * struct vring
174 * {
175 * // The actual descriptors (16 bytes each)
176 * struct vring_desc desc[num];
177 *
178 * // A ring of available descriptor heads with free-running index.
179 * __virtio16 avail_flags;
180 * __virtio16 avail_idx;
181 * __virtio16 available[num];
182 * __virtio16 used_event_idx;
183 *
184 * // Padding to the next align boundary.
185 * char pad[];
186 *
187 * // A ring of used descriptor heads with free-running index.
188 * __virtio16 used_flags;
189 * __virtio16 used_idx;
190 * struct vring_used_elem used[num];
191 * __virtio16 avail_event_idx;
192 * };
193 */
194/* We publish the used event index at the end of the available ring, and vice
195 * versa. They are at the end for backwards compatibility. */
196#define vring_used_event(vr) ((vr)->avail->ring[(vr)->num])
197#define vring_avail_event(vr) (*(__virtio16 *)&(vr)->used->ring[(vr)->num])
198
199static inline void vring_init(struct vring *vr, unsigned int num, void *p,
200 unsigned long align)
201{
202 vr->num = num;
203 vr->desc = p;
204 vr->avail = (struct vring_avail *)((char *)p + num * sizeof(struct vring_desc));
205 vr->used = (void *)(((uintptr_t)&vr->avail->ring[num] + sizeof(__virtio16)
206 + align-1) & ~(align - 1));
207}
208
209static inline unsigned vring_size(unsigned int num, unsigned long align)
210{
211 return ((sizeof(struct vring_desc) * num + sizeof(__virtio16) * (3 + num)
212 + align - 1) & ~(align - 1))
213 + sizeof(__virtio16) * 3 + sizeof(struct vring_used_elem) * num;
214}
215
216#endif /* VIRTIO_RING_NO_LEGACY */
217
218/* The following is used with USED_EVENT_IDX and AVAIL_EVENT_IDX */
219/* Assuming a given event_idx value from the other side, if
220 * we have just incremented index from old to new_idx,
221 * should we trigger an event? */
222static inline int vring_need_event(__u16 event_idx, __u16 new_idx, __u16 old)
223{
224 /* Note: Xen has similar logic for notification hold-off
225 * in include/xen/interface/io/ring.h with req_event and req_prod
226 * corresponding to event_idx + 1 and new_idx respectively.
227 * Note also that req_event and req_prod in Xen start at 1,
228 * event indexes in virtio start at 0. */
229 return (__u16)(new_idx - event_idx - 1) < (__u16)(new_idx - old);
230}
231
232struct vring_packed_desc_event {
233 /* Descriptor Ring Change Event Offset/Wrap Counter. */
234 __le16 off_wrap;
235 /* Descriptor Ring Change Event Flags. */
236 __le16 flags;
237};
238
239struct vring_packed_desc {
240 /* Buffer Address. */
241 __le64 addr;
242 /* Buffer Length. */
243 __le32 len;
244 /* Buffer ID. */
245 __le16 id;
246 /* The flags depending on descriptor type. */
247 __le16 flags;
248};
249
250#endif /* _UAPI_LINUX_VIRTIO_RING_H */
251