virtio: Add codes for virtual queue/ring management

This adds support for managing virtual queue/ring, the channel
for high performance I/O between host and guest.

Signed-off-by: Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>

drivers/virtio/Makefile

@@ -3,4 +3,4 @@
 # Copyright (C) 2018, Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi>
 # Copyright (C) 2018, Bin Meng <bmeng.cn@gmail.com>
 
-obj-y += virtio-uclass.o
+obj-y += virtio-uclass.o virtio_ring.o

drivers/virtio/virtio_ring.c

@@ -0,0 +1,358 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018, Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi>
+ * Copyright (C) 2018, Bin Meng <bmeng.cn@gmail.com>
+ *
+ * virtio ring implementation
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <malloc.h>
+#include <virtio_types.h>
+#include <virtio.h>
+#include <virtio_ring.h>
+
+int virtqueue_add(struct virtqueue *vq, struct virtio_sg *sgs[],
+		  unsigned int out_sgs, unsigned int in_sgs)
+{
+	struct vring_desc *desc;
+	unsigned int total_sg = out_sgs + in_sgs;
+	unsigned int i, n, avail, descs_used, uninitialized_var(prev);
+	int head;
+
+	WARN_ON(total_sg == 0);
+
+	head = vq->free_head;
+
+	desc = vq->vring.desc;
+	i = head;
+	descs_used = total_sg;
+
+	if (vq->num_free < descs_used) {
+		debug("Can't add buf len %i - avail = %i\n",
+		      descs_used, vq->num_free);
+		/*
+		 * FIXME: for historical reasons, we force a notify here if
+		 * there are outgoing parts to the buffer.  Presumably the
+		 * host should service the ring ASAP.
+		 */
+		if (out_sgs)
+			virtio_notify(vq->vdev, vq);
+		return -ENOSPC;
+	}
+
+	for (n = 0; n < out_sgs; n++) {
+		struct virtio_sg *sg = sgs[n];
+
+		desc[i].flags = cpu_to_virtio16(vq->vdev, VRING_DESC_F_NEXT);
+		desc[i].addr = cpu_to_virtio64(vq->vdev, (u64)(size_t)sg->addr);
+		desc[i].len = cpu_to_virtio32(vq->vdev, sg->length);
+
+		prev = i;
+		i = virtio16_to_cpu(vq->vdev, desc[i].next);
+	}
+	for (; n < (out_sgs + in_sgs); n++) {
+		struct virtio_sg *sg = sgs[n];
+
+		desc[i].flags = cpu_to_virtio16(vq->vdev, VRING_DESC_F_NEXT |
+						VRING_DESC_F_WRITE);
+		desc[i].addr = cpu_to_virtio64(vq->vdev,
+					       (u64)(uintptr_t)sg->addr);
+		desc[i].len = cpu_to_virtio32(vq->vdev, sg->length);
+
+		prev = i;
+		i = virtio16_to_cpu(vq->vdev, desc[i].next);
+	}
+	/* Last one doesn't continue */
+	desc[prev].flags &= cpu_to_virtio16(vq->vdev, ~VRING_DESC_F_NEXT);
+
+	/* We're using some buffers from the free list. */
+	vq->num_free -= descs_used;
+
+	/* Update free pointer */
+	vq->free_head = i;
+
+	/*
+	 * Put entry in available array (but don't update avail->idx
+	 * until they do sync).
+	 */
+	avail = vq->avail_idx_shadow & (vq->vring.num - 1);
+	vq->vring.avail->ring[avail] = cpu_to_virtio16(vq->vdev, head);
+
+	/*
+	 * Descriptors and available array need to be set before we expose the
+	 * new available array entries.
+	 */
+	virtio_wmb();
+	vq->avail_idx_shadow++;
+	vq->vring.avail->idx = cpu_to_virtio16(vq->vdev, vq->avail_idx_shadow);
+	vq->num_added++;
+
+	/*
+	 * This is very unlikely, but theoretically possible.
+	 * Kick just in case.
+	 */
+	if (unlikely(vq->num_added == (1 << 16) - 1))
+		virtqueue_kick(vq);
+
+	return 0;
+}
+
+static bool virtqueue_kick_prepare(struct virtqueue *vq)
+{
+	u16 new, old;
+	bool needs_kick;
+
+	/*
+	 * We need to expose available array entries before checking
+	 * avail event.
+	 */
+	virtio_mb();
+
+	old = vq->avail_idx_shadow - vq->num_added;
+	new = vq->avail_idx_shadow;
+	vq->num_added = 0;
+
+	if (vq->event) {
+		needs_kick = vring_need_event(virtio16_to_cpu(vq->vdev,
+				vring_avail_event(&vq->vring)), new, old);
+	} else {
+		needs_kick = !(vq->vring.used->flags & cpu_to_virtio16(vq->vdev,
+				VRING_USED_F_NO_NOTIFY));
+	}
+
+	return needs_kick;
+}
+
+void virtqueue_kick(struct virtqueue *vq)
+{
+	if (virtqueue_kick_prepare(vq))
+		virtio_notify(vq->vdev, vq);
+}
+
+static void detach_buf(struct virtqueue *vq, unsigned int head)
+{
+	unsigned int i;
+	__virtio16 nextflag = cpu_to_virtio16(vq->vdev, VRING_DESC_F_NEXT);
+
+	/* Put back on free list: unmap first-level descriptors and find end */
+	i = head;
+
+	while (vq->vring.desc[i].flags & nextflag) {
+		i = virtio16_to_cpu(vq->vdev, vq->vring.desc[i].next);
+		vq->num_free++;
+	}
+
+	vq->vring.desc[i].next = cpu_to_virtio16(vq->vdev, vq->free_head);
+	vq->free_head = head;
+
+	/* Plus final descriptor */
+	vq->num_free++;
+}
+
+static inline bool more_used(const struct virtqueue *vq)
+{
+	return vq->last_used_idx != virtio16_to_cpu(vq->vdev,
+			vq->vring.used->idx);
+}
+
+void *virtqueue_get_buf(struct virtqueue *vq, unsigned int *len)
+{
+	unsigned int i;
+	u16 last_used;
+
+	if (!more_used(vq)) {
+		debug("(%s.%d): No more buffers in queue\n",
+		      vq->vdev->name, vq->index);
+		return NULL;
+	}
+
+	/* Only get used array entries after they have been exposed by host */
+	virtio_rmb();
+
+	last_used = (vq->last_used_idx & (vq->vring.num - 1));
+	i = virtio32_to_cpu(vq->vdev, vq->vring.used->ring[last_used].id);
+	if (len) {
+		*len = virtio32_to_cpu(vq->vdev,
+				       vq->vring.used->ring[last_used].len);
+		debug("(%s.%d): last used idx %u with len %u\n",
+		      vq->vdev->name, vq->index, i, *len);
+	}
+
+	if (unlikely(i >= vq->vring.num)) {
+		printf("(%s.%d): id %u out of range\n",
+		       vq->vdev->name, vq->index, i);
+		return NULL;
+	}
+
+	detach_buf(vq, i);
+	vq->last_used_idx++;
+	/*
+	 * If we expect an interrupt for the next entry, tell host
+	 * by writing event index and flush out the write before
+	 * the read in the next get_buf call.
+	 */
+	if (!(vq->avail_flags_shadow & VRING_AVAIL_F_NO_INTERRUPT))
+		virtio_store_mb(&vring_used_event(&vq->vring),
+				cpu_to_virtio16(vq->vdev, vq->last_used_idx));
+
+	return (void *)(uintptr_t)virtio64_to_cpu(vq->vdev,
+						  vq->vring.desc[i].addr);
+}
+
+static struct virtqueue *__vring_new_virtqueue(unsigned int index,
+					       struct vring vring,
+					       struct udevice *udev)
+{
+	unsigned int i;
+	struct virtqueue *vq;
+	struct virtio_dev_priv *uc_priv = dev_get_uclass_priv(udev);
+	struct udevice *vdev = uc_priv->vdev;
+
+	vq = malloc(sizeof(*vq));
+	if (!vq)
+		return NULL;
+
+	vq->vdev = vdev;
+	vq->index = index;
+	vq->num_free = vring.num;
+	vq->vring = vring;
+	vq->last_used_idx = 0;
+	vq->avail_flags_shadow = 0;
+	vq->avail_idx_shadow = 0;
+	vq->num_added = 0;
+	list_add_tail(&vq->list, &uc_priv->vqs);
+
+	vq->event = virtio_has_feature(vdev, VIRTIO_RING_F_EVENT_IDX);
+
+	/* Tell other side not to bother us */
+	vq->avail_flags_shadow |= VRING_AVAIL_F_NO_INTERRUPT;
+	if (!vq->event)
+		vq->vring.avail->flags = cpu_to_virtio16(vdev,
+				vq->avail_flags_shadow);
+
+	/* Put everything in free lists */
+	vq->free_head = 0;
+	for (i = 0; i < vring.num - 1; i++)
+		vq->vring.desc[i].next = cpu_to_virtio16(vdev, i + 1);
+
+	return vq;
+}
+
+struct virtqueue *vring_create_virtqueue(unsigned int index, unsigned int num,
+					 unsigned int vring_align,
+					 struct udevice *udev)
+{
+	struct virtqueue *vq;
+	void *queue = NULL;
+	struct vring vring;
+
+	/* We assume num is a power of 2 */
+	if (num & (num - 1)) {
+		printf("Bad virtqueue length %u\n", num);
+		return NULL;
+	}
+
+	/* TODO: allocate each queue chunk individually */
+	for (; num && vring_size(num, vring_align) > PAGE_SIZE; num /= 2) {
+		queue = memalign(PAGE_SIZE, vring_size(num, vring_align));
+		if (queue)
+			break;
+	}
+
+	if (!num)
+		return NULL;
+
+	if (!queue) {
+		/* Try to get a single page. You are my only hope! */
+		queue = memalign(PAGE_SIZE, vring_size(num, vring_align));
+	}
+	if (!queue)
+		return NULL;
+
+	memset(queue, 0, vring_size(num, vring_align));
+	vring_init(&vring, num, queue, vring_align);
+
+	vq = __vring_new_virtqueue(index, vring, udev);
+	if (!vq) {
+		free(queue);
+		return NULL;
+	}
+	debug("(%s): created vring @ %p for vq @ %p with num %u\n", udev->name,
+	      queue, vq, num);
+
+	return vq;
+}
+
+void vring_del_virtqueue(struct virtqueue *vq)
+{
+	free(vq->vring.desc);
+	list_del(&vq->list);
+	free(vq);
+}
+
+unsigned int virtqueue_get_vring_size(struct virtqueue *vq)
+{
+	return vq->vring.num;
+}
+
+ulong virtqueue_get_desc_addr(struct virtqueue *vq)
+{
+	return (ulong)vq->vring.desc;
+}
+
+ulong virtqueue_get_avail_addr(struct virtqueue *vq)
+{
+	return (ulong)vq->vring.desc +
+	       ((char *)vq->vring.avail - (char *)vq->vring.desc);
+}
+
+ulong virtqueue_get_used_addr(struct virtqueue *vq)
+{
+	return (ulong)vq->vring.desc +
+	       ((char *)vq->vring.used - (char *)vq->vring.desc);
+}
+
+bool virtqueue_poll(struct virtqueue *vq, u16 last_used_idx)
+{
+	virtio_mb();
+
+	return last_used_idx != virtio16_to_cpu(vq->vdev, vq->vring.used->idx);
+}
+
+void virtqueue_dump(struct virtqueue *vq)
+{
+	unsigned int i;
+
+	printf("virtqueue %p for dev %s:\n", vq, vq->vdev->name);
+	printf("\tindex %u, phys addr %p num %u\n",
+	       vq->index, vq->vring.desc, vq->vring.num);
+	printf("\tfree_head %u, num_added %u, num_free %u\n",
+	       vq->free_head, vq->num_added, vq->num_free);
+	printf("\tlast_used_idx %u, avail_flags_shadow %u, avail_idx_shadow %u\n",
+	       vq->last_used_idx, vq->avail_flags_shadow, vq->avail_idx_shadow);
+
+	printf("Descriptor dump:\n");
+	for (i = 0; i < vq->vring.num; i++) {
+		printf("\tdesc[%u] = { 0x%llx, len %u, flags %u, next %u }\n",
+		       i, vq->vring.desc[i].addr, vq->vring.desc[i].len,
+		       vq->vring.desc[i].flags, vq->vring.desc[i].next);
+	}
+
+	printf("Avail ring dump:\n");
+	printf("\tflags %u, idx %u\n",
+	       vq->vring.avail->flags, vq->vring.avail->idx);
+	for (i = 0; i < vq->vring.num; i++) {
+		printf("\tavail[%u] = %u\n",
+		       i, vq->vring.avail->ring[i]);
+	}
+
+	printf("Used ring dump:\n");
+	printf("\tflags %u, idx %u\n",
+	       vq->vring.used->flags, vq->vring.used->idx);
+	for (i = 0; i < vq->vring.num; i++) {
+		printf("\tused[%u] = { %u, %u }\n", i,
+		       vq->vring.used->ring[i].id, vq->vring.used->ring[i].len);
+	}
+}

include/virtio_ring.h

@@ -0,0 +1,320 @@
+/* SPDX-License-Identifier: BSD-3-Clause */
+/*
+ * Copyright (C) 2018, Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi>
+ * Copyright (C) 2018, Bin Meng <bmeng.cn@gmail.com>
+ *
+ * From Linux kernel include/uapi/linux/virtio_ring.h
+ */
+
+#ifndef _LINUX_VIRTIO_RING_H
+#define _LINUX_VIRTIO_RING_H
+
+#include <virtio_types.h>
+
+/* This marks a buffer as continuing via the next field */
+#define VRING_DESC_F_NEXT		1
+/* This marks a buffer as write-only (otherwise read-only) */
+#define VRING_DESC_F_WRITE		2
+/* This means the buffer contains a list of buffer descriptors */
+#define VRING_DESC_F_INDIRECT		4
+
+/*
+ * The Host uses this in used->flags to advise the Guest: don't kick me when
+ * you add a buffer. It's unreliable, so it's simply an optimization. Guest
+ * will still kick if it's out of buffers.
+ */
+#define VRING_USED_F_NO_NOTIFY		1
+
+/*
+ * The Guest uses this in avail->flags to advise the Host: don't interrupt me
+ * when you consume a buffer. It's unreliable, so it's simply an optimization.
+ */
+#define VRING_AVAIL_F_NO_INTERRUPT	1
+
+/* We support indirect buffer descriptors */
+#define VIRTIO_RING_F_INDIRECT_DESC	28
+
+/*
+ * The Guest publishes the used index for which it expects an interrupt
+ * at the end of the avail ring. Host should ignore the avail->flags field.
+ *
+ * The Host publishes the avail index for which it expects a kick
+ * at the end of the used ring. Guest should ignore the used->flags field.
+ */
+#define VIRTIO_RING_F_EVENT_IDX		29
+
+/* Virtio ring descriptors: 16 bytes. These can chain together via "next". */
+struct vring_desc {
+	/* Address (guest-physical) */
+	__virtio64 addr;
+	/* Length */
+	__virtio32 len;
+	/* The flags as indicated above */
+	__virtio16 flags;
+	/* We chain unused descriptors via this, too */
+	__virtio16 next;
+};
+
+struct vring_avail {
+	__virtio16 flags;
+	__virtio16 idx;
+	__virtio16 ring[];
+};
+
+struct vring_used_elem {
+	/* Index of start of used descriptor chain */
+	__virtio32 id;
+	/* Total length of the descriptor chain which was used (written to) */
+	__virtio32 len;
+};
+
+struct vring_used {
+	__virtio16 flags;
+	__virtio16 idx;
+	struct vring_used_elem ring[];
+};
+
+struct vring {
+	unsigned int num;
+	struct vring_desc *desc;
+	struct vring_avail *avail;
+	struct vring_used *used;
+};
+
+/**
+ * virtqueue - a queue to register buffers for sending or receiving.
+ *
+ * @list: the chain of virtqueues for this device
+ * @vdev: the virtio device this queue was created for
+ * @index: the zero-based ordinal number for this queue
+ * @num_free: number of elements we expect to be able to fit
+ * @vring: actual memory layout for this queue
+ * @event: host publishes avail event idx
+ * @free_head: head of free buffer list
+ * @num_added: number we've added since last sync
+ * @last_used_idx: last used index we've seen
+ * @avail_flags_shadow: last written value to avail->flags
+ * @avail_idx_shadow: last written value to avail->idx in guest byte order
+ */
+struct virtqueue {
+	struct list_head list;
+	struct udevice *vdev;
+	unsigned int index;
+	unsigned int num_free;
+	struct vring vring;
+	bool event;
+	unsigned int free_head;
+	unsigned int num_added;
+	u16 last_used_idx;
+	u16 avail_flags_shadow;
+	u16 avail_idx_shadow;
+};
+
+/*
+ * Alignment requirements for vring elements.
+ * When using pre-virtio 1.0 layout, these fall out naturally.
+ */
+#define VRING_AVAIL_ALIGN_SIZE		2
+#define VRING_USED_ALIGN_SIZE		4
+#define VRING_DESC_ALIGN_SIZE		16
+
+/*
+ * We publish the used event index at the end of the available ring,
+ * and vice versa. They are at the end for backwards compatibility.
+ */
+#define vring_used_event(vr)	((vr)->avail->ring[(vr)->num])
+#define vring_avail_event(vr)	(*(__virtio16 *)&(vr)->used->ring[(vr)->num])
+
+static inline void vring_init(struct vring *vr, unsigned int num, void *p,
+			      unsigned long align)
+{
+	vr->num = num;
+	vr->desc = p;
+	vr->avail = p + num * sizeof(struct vring_desc);
+	vr->used = (void *)(((uintptr_t)&vr->avail->ring[num] +
+		   sizeof(__virtio16) + align - 1) & ~(align - 1));
+}
+
+static inline unsigned int vring_size(unsigned int num, unsigned long align)
+{
+	return ((sizeof(struct vring_desc) * num +
+		sizeof(__virtio16) * (3 + num)  + align - 1) & ~(align - 1)) +
+		sizeof(__virtio16) * 3 + sizeof(struct vring_used_elem) * num;
+}
+
+/*
+ * The following is used with USED_EVENT_IDX and AVAIL_EVENT_IDX.
+ * Assuming a given event_idx value from the other side, if we have just
+ * incremented index from old to new_idx, should we trigger an event?
+ */
+static inline int vring_need_event(__u16 event_idx, __u16 new_idx, __u16 old)
+{
+	/*
+	 * Note: Xen has similar logic for notification hold-off
+	 * in include/xen/interface/io/ring.h with req_event and req_prod
+	 * corresponding to event_idx + 1 and new_idx respectively.
+	 * Note also that req_event and req_prod in Xen start at 1,
+	 * event indexes in virtio start at 0.
+	 */
+	return (__u16)(new_idx - event_idx - 1) < (__u16)(new_idx - old);
+}
+
+struct virtio_sg;
+
+/**
+ * virtqueue_add - expose buffers to other end
+ *
+ * @vq:		the struct virtqueue we're talking about
+ * @sgs:	array of terminated scatterlists
+ * @out_sgs:	the number of scatterlists readable by other side
+ * @in_sgs:	the number of scatterlists which are writable
+ *		(after readable ones)
+ *
+ * Caller must ensure we don't call this with other virtqueue operations
+ * at the same time (except where noted).
+ *
+ * Returns zero or a negative error (ie. ENOSPC, ENOMEM, EIO).
+ */
+int virtqueue_add(struct virtqueue *vq, struct virtio_sg *sgs[],
+		  unsigned int out_sgs, unsigned int in_sgs);
+
+/**
+ * virtqueue_kick - update after add_buf
+ *
+ * @vq:		the struct virtqueue
+ *
+ * After one or more virtqueue_add() calls, invoke this to kick
+ * the other side.
+ *
+ * Caller must ensure we don't call this with other virtqueue
+ * operations at the same time (except where noted).
+ */
+void virtqueue_kick(struct virtqueue *vq);
+
+/**
+ * virtqueue_get_buf - get the next used buffer
+ *
+ * @vq:		the struct virtqueue we're talking about
+ * @len:	the length written into the buffer
+ *
+ * If the device wrote data into the buffer, @len will be set to the
+ * amount written. This means you don't need to clear the buffer
+ * beforehand to ensure there's no data leakage in the case of short
+ * writes.
+ *
+ * Caller must ensure we don't call this with other virtqueue
+ * operations at the same time (except where noted).
+ *
+ * Returns NULL if there are no used buffers, or the memory buffer
+ * handed to virtqueue_add_*().
+ */
+void *virtqueue_get_buf(struct virtqueue *vq, unsigned int *len);
+
+/**
+ * vring_create_virtqueue - create a virtqueue for a virtio device
+ *
+ * @index:	the index of the queue
+ * @num:	number of elements of the queue
+ * @vring_align:the alignment requirement of the descriptor ring
+ * @udev:	the virtio transport udevice
+ * @return: the virtqueue pointer or NULL if failed
+ *
+ * This creates a virtqueue and allocates the descriptor ring for a virtio
+ * device. The caller should query virtqueue_get_ring_size() to learn the
+ * actual size of the ring.
+ *
+ * This API is supposed to be called by the virtio transport driver in the
+ * virtio find_vqs() uclass method.
+ */
+struct virtqueue *vring_create_virtqueue(unsigned int index, unsigned int num,
+					 unsigned int vring_align,
+					 struct udevice *udev);
+
+/**
+ * vring_del_virtqueue - destroy a virtqueue
+ *
+ * @vq:		the struct virtqueue we're talking about
+ *
+ * This destroys a virtqueue. If created with vring_create_virtqueue(),
+ * this also frees the descriptor ring.
+ *
+ * This API is supposed to be called by the virtio transport driver in the
+ * virtio del_vqs() uclass method.
+ */
+void vring_del_virtqueue(struct virtqueue *vq);
+
+/**
+ * virtqueue_get_vring_size - get the size of the virtqueue's vring
+ *
+ * @vq:		the struct virtqueue containing the vring of interest
+ * @return: the size of the vring in a virtqueue.
+ */
+unsigned int virtqueue_get_vring_size(struct virtqueue *vq);
+
+/**
+ * virtqueue_get_desc_addr - get the vring descriptor table address
+ *
+ * @vq:		the struct virtqueue containing the vring of interest
+ * @return: the descriptor table address of the vring in a virtqueue.
+ */
+ulong virtqueue_get_desc_addr(struct virtqueue *vq);
+
+/**
+ * virtqueue_get_avail_addr - get the vring available ring address
+ *
+ * @vq:		the struct virtqueue containing the vring of interest
+ * @return: the available ring address of the vring in a virtqueue.
+ */
+ulong virtqueue_get_avail_addr(struct virtqueue *vq);
+
+/**
+ * virtqueue_get_used_addr - get the vring used ring address
+ *
+ * @vq:		the struct virtqueue containing the vring of interest
+ * @return: the used ring address of the vring in a virtqueue.
+ */
+ulong virtqueue_get_used_addr(struct virtqueue *vq);
+
+/**
+ * virtqueue_poll - query pending used buffers
+ *
+ * @vq:			the struct virtqueue we're talking about
+ * @last_used_idx:	virtqueue last used index
+ *
+ * Returns "true" if there are pending used buffers in the queue.
+ */
+bool virtqueue_poll(struct virtqueue *vq, u16 last_used_idx);
+
+/**
+ * virtqueue_dump - dump the virtqueue for debugging
+ *
+ * @vq:		the struct virtqueue we're talking about
+ *
+ * Caller must ensure we don't call this with other virtqueue operations
+ * at the same time (except where noted).
+ */
+void virtqueue_dump(struct virtqueue *vq);
+
+/*
+ * Barriers in virtio are tricky. Since we are not in a hyperviosr/guest
+ * scenario, having these as nops is enough to work as expected.
+ */
+
+static inline void virtio_mb(void)
+{
+}
+
+static inline void virtio_rmb(void)
+{
+}
+
+static inline void virtio_wmb(void)
+{
+}
+
+static inline void virtio_store_mb(__virtio16 *p, __virtio16 v)
+{
+	WRITE_ONCE(*p, v);
+}
+
+#endif /* _LINUX_VIRTIO_RING_H */