include/linux/virtio_ring.h

   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  * and lguest, 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
  36 /* This marks a buffer as continuing via the next field. */
  37 #define VRING_DESC_F_NEXT       1
  38 /* This marks a buffer as write-only (otherwise read-only). */
  39 #define VRING_DESC_F_WRITE      2
  40 /* This means the buffer contains a list of buffer descriptors. */
  41 #define VRING_DESC_F_INDIRECT   4
  42
  43 /* The Host uses this in used->flags to advise the Guest: don't kick me when
  44  * you add a buffer.  It's unreliable, so it's simply an optimization.  Guest
  45  * will still kick if it's out of buffers. */
  46 #define VRING_USED_F_NO_NOTIFY  1
  47 /* The Guest uses this in avail->flags to advise the Host: don't interrupt me
  48  * when you consume a buffer.  It's unreliable, so it's simply an
  49  * optimization.  */
  50 #define VRING_AVAIL_F_NO_INTERRUPT      1
  51
  52 /* We support indirect buffer descriptors */
  53 #define VIRTIO_RING_F_INDIRECT_DESC     28
  54
  55 /* The Guest publishes the used index for which it expects an interrupt
  56  * at the end of the avail ring. Host should ignore the avail->flags field. */
  57 /* The Host publishes the avail index for which it expects a kick
  58  * at the end of the used ring. Guest should ignore the used->flags field. */
  59 #define VIRTIO_RING_F_EVENT_IDX         29
  60
  61 /* Virtio ring descriptors: 16 bytes.  These can chain together via "next". */
  62 struct vring_desc {
  63         /* Address (guest-physical). */
  64         __u64 addr;
  65         /* Length. */
  66         __u32 len;
  67         /* The flags as indicated above. */
  68         __u16 flags;
  69         /* We chain unused descriptors via this, too */
  70         __u16 next;
  71 };
  72
  73 struct vring_avail {
  74         __u16 flags;
  75         __u16 idx;
  76         __u16 ring[];
  77 };
  78
  79 /* u32 is used here for ids for padding reasons. */
  80 struct vring_used_elem {
  81         /* Index of start of used descriptor chain. */
  82         __u32 id;
  83         /* Total length of the descriptor chain which was used (written to) */
  84         __u32 len;
  85 };
  86
  87 struct vring_used {
  88         __u16 flags;
  89         __u16 idx;
  90         struct vring_used_elem ring[];
  91 };
  92
  93 struct vring {
  94         unsigned int num;
  95
  96         struct vring_desc *desc;
  97
  98         struct vring_avail *avail;
  99
 100         struct vring_used *used;
 101 };
 102
 103 /* The standard layout for the ring is a continuous chunk of memory which looks
 104  * like this.  We assume num is a power of 2.
 105  *
 106  * struct vring
 107  * {
 108  *      // The actual descriptors (16 bytes each)
 109  *      struct vring_desc desc[num];
 110  *
 111  *      // A ring of available descriptor heads with free-running index.
 112  *      __u16 avail_flags;
 113  *      __u16 avail_idx;
 114  *      __u16 available[num];
 115  *      __u16 used_event_idx;
 116  *
 117  *      // Padding to the next align boundary.
 118  *      char pad[];
 119  *
 120  *      // A ring of used descriptor heads with free-running index.
 121  *      __u16 used_flags;
 122  *      __u16 used_idx;
 123  *      struct vring_used_elem used[num];
 124  *      __u16 avail_event_idx;
 125  * };
 126  */
 127 /* We publish the used event index at the end of the available ring, and vice
 128  * versa. They are at the end for backwards compatibility. */
 129 #define vring_used_event(vr) ((vr)->avail->ring[(vr)->num])
 130 #define vring_avail_event(vr) (*(__u16 *)&(vr)->used->ring[(vr)->num])
 131
 132 static inline void vring_init(struct vring *vr, unsigned int num, void *p,
 133                               unsigned long align)
 134 {
 135         vr->num = num;
 136         vr->desc = p;
 137         vr->avail = p + num*sizeof(struct vring_desc);
 138         vr->used = (void *)(((unsigned long)&vr->avail->ring[num] + sizeof(__u16)
 139                 + align-1) & ~(align - 1));
 140 }
 141
 142 static inline unsigned vring_size(unsigned int num, unsigned long align)
 143 {
 144         return ((sizeof(struct vring_desc) * num + sizeof(__u16) * (3 + num)
 145                  + align - 1) & ~(align - 1))
 146                 + sizeof(__u16) * 3 + sizeof(struct vring_used_elem) * num;
 147 }
 148
 149 /* The following is used with USED_EVENT_IDX and AVAIL_EVENT_IDX */
 150 /* Assuming a given event_idx value from the other size, if
 151  * we have just incremented index from old to new_idx,
 152  * should we trigger an event? */
 153 static inline int vring_need_event(__u16 event_idx, __u16 new_idx, __u16 old)
 154 {
 155         /* Note: Xen has similar logic for notification hold-off
 156          * in include/xen/interface/io/ring.h with req_event and req_prod
 157          * corresponding to event_idx + 1 and new_idx respectively.
 158          * Note also that req_event and req_prod in Xen start at 1,
 159          * event indexes in virtio start at 0. */
 160         return (__u16)(new_idx - event_idx - 1) < (__u16)(new_idx - old);
 161 }
 162
 163 #ifdef __KERNEL__
 164 #include <linux/irqreturn.h>
 165 struct virtio_device;
 166 struct virtqueue;
 167
 168 struct virtqueue *vring_new_virtqueue(unsigned int num,
 169                                       unsigned int vring_align,
 170                                       struct virtio_device *vdev,
 171                                       bool weak_barriers,
 172                                       void *pages,
 173                                       void (*notify)(struct virtqueue *vq),
 174                                       void (*callback)(struct virtqueue *vq),
 175                                       const char *name);
 176 void vring_del_virtqueue(struct virtqueue *vq);
 177 /* Filter out transport-specific feature bits. */
 178 void vring_transport_features(struct virtio_device *vdev);
 179
 180 irqreturn_t vring_interrupt(int irq, void *_vq);
 181 #endif /* __KERNEL__ */
 182 #endif /* _LINUX_VIRTIO_RING_H */