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/* Virtio ring descriptors: 16 bytes. These can chain together via "next". */
90struct vring_desc {
91 /* Address (guest-physical). */
92 __virtio64 addr;
93 /* Length. */
94 __virtio32 len;
95 /* The flags as indicated above. */
96 __virtio16 flags;
97 /* We chain unused descriptors via this, too */
98 __virtio16 next;
99};
100
101struct vring_avail {
102 __virtio16 flags;
103 __virtio16 idx;
104 __virtio16 ring[];
105};
106
107/* u32 is used here for ids for padding reasons. */
108struct vring_used_elem {
109 /* Index of start of used descriptor chain. */
110 __virtio32 id;
111 /* Total length of the descriptor chain which was used (written to) */
112 __virtio32 len;
113};
114
115struct vring_used {
116 __virtio16 flags;
117 __virtio16 idx;
118 struct vring_used_elem ring[];
119};
120
121struct vring {
122 unsigned int num;
123
124 struct vring_desc *desc;
125
126 struct vring_avail *avail;
127
128 struct vring_used *used;
129};
130
131/* Alignment requirements for vring elements.
132 * When using pre-virtio 1.0 layout, these fall out naturally.
133 */
134#define VRING_AVAIL_ALIGN_SIZE 2
135#define VRING_USED_ALIGN_SIZE 4
136#define VRING_DESC_ALIGN_SIZE 16
137
138/* The standard layout for the ring is a continuous chunk of memory which looks
139 * like this. We assume num is a power of 2.
140 *
141 * struct vring
142 * {
143 * // The actual descriptors (16 bytes each)
144 * struct vring_desc desc[num];
145 *
146 * // A ring of available descriptor heads with free-running index.
147 * __virtio16 avail_flags;
148 * __virtio16 avail_idx;
149 * __virtio16 available[num];
150 * __virtio16 used_event_idx;
151 *
152 * // Padding to the next align boundary.
153 * char pad[];
154 *
155 * // A ring of used descriptor heads with free-running index.
156 * __virtio16 used_flags;
157 * __virtio16 used_idx;
158 * struct vring_used_elem used[num];
159 * __virtio16 avail_event_idx;
160 * };
161 */
162/* We publish the used event index at the end of the available ring, and vice
163 * versa. They are at the end for backwards compatibility. */
164#define vring_used_event(vr) ((vr)->avail->ring[(vr)->num])
165#define vring_avail_event(vr) (*(__virtio16 *)&(vr)->used->ring[(vr)->num])
166
167static inline void vring_init(struct vring *vr, unsigned int num, void *p,
168 unsigned long align)
169{
170 vr->num = num;
171 vr->desc = p;
172 vr->avail = p + num*sizeof(struct vring_desc);
173 vr->used = (void *)(((uintptr_t)&vr->avail->ring[num] + sizeof(__virtio16)
174 + align-1) & ~(align - 1));
175}
176
177static inline unsigned vring_size(unsigned int num, unsigned long align)
178{
179 return ((sizeof(struct vring_desc) * num + sizeof(__virtio16) * (3 + num)
180 + align - 1) & ~(align - 1))
181 + sizeof(__virtio16) * 3 + sizeof(struct vring_used_elem) * num;
182}
183
184/* The following is used with USED_EVENT_IDX and AVAIL_EVENT_IDX */
185/* Assuming a given event_idx value from the other side, if
186 * we have just incremented index from old to new_idx,
187 * should we trigger an event? */
188static inline int vring_need_event(__u16 event_idx, __u16 new_idx, __u16 old)
189{
190 /* Note: Xen has similar logic for notification hold-off
191 * in include/xen/interface/io/ring.h with req_event and req_prod
192 * corresponding to event_idx + 1 and new_idx respectively.
193 * Note also that req_event and req_prod in Xen start at 1,
194 * event indexes in virtio start at 0. */
195 return (__u16)(new_idx - event_idx - 1) < (__u16)(new_idx - old);
196}
197
198struct vring_packed_desc_event {
199 /* Descriptor Ring Change Event Offset/Wrap Counter. */
200 __le16 off_wrap;
201 /* Descriptor Ring Change Event Flags. */
202 __le16 flags;
203};
204
205struct vring_packed_desc {
206 /* Buffer Address. */
207 __le64 addr;
208 /* Buffer Length. */
209 __le32 len;
210 /* Buffer ID. */
211 __le16 id;
212 /* The flags depending on descriptor type. */
213 __le16 flags;
214};
215
216#endif /* _UAPI_LINUX_VIRTIO_RING_H */
217