Merge fc6d70f40b ("virtio_ring: check desc == NULL when using indirect with packed") into android-mainline

Steps on the way to 5.16-rc1

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I7196f49b7264d3f22de3b532aadf389daada8acd

drivers/block/Kconfig

@@ -371,6 +371,7 @@ config XEN_BLKDEV_BACKEND
 config VIRTIO_BLK
 	tristate "Virtio block driver"
 	depends on VIRTIO
+	select SG_POOL
 	help
 	  This is the virtual block driver for virtio.  It can be used with
           QEMU based VMMs (like KVM or Xen).  Say Y or M.

drivers/block/virtio_blk.c

@@ -24,6 +24,31 @@
 /* The maximum number of sg elements that fit into a virtqueue */
 #define VIRTIO_BLK_MAX_SG_ELEMS 32768
 
+#ifdef CONFIG_ARCH_NO_SG_CHAIN
+#define VIRTIO_BLK_INLINE_SG_CNT	0
+#else
+#define VIRTIO_BLK_INLINE_SG_CNT	2
+#endif
+
+static int virtblk_queue_count_set(const char *val,
+		const struct kernel_param *kp)
+{
+	return param_set_uint_minmax(val, kp, 1, nr_cpu_ids);
+}
+
+static const struct kernel_param_ops queue_count_ops = {
+	.set = virtblk_queue_count_set,
+	.get = param_get_uint,
+};
+
+static unsigned int num_request_queues;
+module_param_cb(num_request_queues, &queue_count_ops, &num_request_queues,
+		0644);
+MODULE_PARM_DESC(num_request_queues,
+		 "Limit the number of request queues to use for blk device. "
+		 "0 for no limit. "
+		 "Values > nr_cpu_ids truncated to nr_cpu_ids.");
+
 static int major;
 static DEFINE_IDA(vd_index_ida);
 
@@ -77,6 +102,7 @@ struct virtio_blk {
 struct virtblk_req {
 	struct virtio_blk_outhdr out_hdr;
 	u8 status;
+	struct sg_table sg_table;
 	struct scatterlist sg[];
 };
 
@@ -162,12 +188,92 @@ static int virtblk_setup_discard_write_zeroes(struct request *req, bool unmap)
 	return 0;
 }
 
+static void virtblk_unmap_data(struct request *req, struct virtblk_req *vbr)
+{
+	if (blk_rq_nr_phys_segments(req))
+		sg_free_table_chained(&vbr->sg_table,
+				      VIRTIO_BLK_INLINE_SG_CNT);
+}
+
+static int virtblk_map_data(struct blk_mq_hw_ctx *hctx, struct request *req,
+		struct virtblk_req *vbr)
+{
+	int err;
+
+	if (!blk_rq_nr_phys_segments(req))
+		return 0;
+
+	vbr->sg_table.sgl = vbr->sg;
+	err = sg_alloc_table_chained(&vbr->sg_table,
+				     blk_rq_nr_phys_segments(req),
+				     vbr->sg_table.sgl,
+				     VIRTIO_BLK_INLINE_SG_CNT);
+	if (unlikely(err))
+		return -ENOMEM;
+
+	return blk_rq_map_sg(hctx->queue, req, vbr->sg_table.sgl);
+}
+
+static void virtblk_cleanup_cmd(struct request *req)
+{
+	if (req->rq_flags & RQF_SPECIAL_PAYLOAD)
+		kfree(bvec_virt(&req->special_vec));
+}
+
+static int virtblk_setup_cmd(struct virtio_device *vdev, struct request *req,
+		struct virtblk_req *vbr)
+{
+	bool unmap = false;
+	u32 type;
+
+	vbr->out_hdr.sector = 0;
+
+	switch (req_op(req)) {
+	case REQ_OP_READ:
+		type = VIRTIO_BLK_T_IN;
+		vbr->out_hdr.sector = cpu_to_virtio64(vdev,
+						      blk_rq_pos(req));
+		break;
+	case REQ_OP_WRITE:
+		type = VIRTIO_BLK_T_OUT;
+		vbr->out_hdr.sector = cpu_to_virtio64(vdev,
+						      blk_rq_pos(req));
+		break;
+	case REQ_OP_FLUSH:
+		type = VIRTIO_BLK_T_FLUSH;
+		break;
+	case REQ_OP_DISCARD:
+		type = VIRTIO_BLK_T_DISCARD;
+		break;
+	case REQ_OP_WRITE_ZEROES:
+		type = VIRTIO_BLK_T_WRITE_ZEROES;
+		unmap = !(req->cmd_flags & REQ_NOUNMAP);
+		break;
+	case REQ_OP_DRV_IN:
+		type = VIRTIO_BLK_T_GET_ID;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return BLK_STS_IOERR;
+	}
+
+	vbr->out_hdr.type = cpu_to_virtio32(vdev, type);
+	vbr->out_hdr.ioprio = cpu_to_virtio32(vdev, req_get_ioprio(req));
+
+	if (type == VIRTIO_BLK_T_DISCARD || type == VIRTIO_BLK_T_WRITE_ZEROES) {
+		if (virtblk_setup_discard_write_zeroes(req, unmap))
+			return BLK_STS_RESOURCE;
+	}
+
+	return 0;
+}
+
 static inline void virtblk_request_done(struct request *req)
 {
 	struct virtblk_req *vbr = blk_mq_rq_to_pdu(req);
 
-	if (req->rq_flags & RQF_SPECIAL_PAYLOAD)
-		kfree(bvec_virt(&req->special_vec));
+	virtblk_unmap_data(req, vbr);
+	virtblk_cleanup_cmd(req);
 	blk_mq_end_request(req, virtblk_result(vbr));
 }
 
@@ -225,57 +331,23 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 	int qid = hctx->queue_num;
 	int err;
 	bool notify = false;
-	bool unmap = false;
-	u32 type;
 
 	BUG_ON(req->nr_phys_segments + 2 > vblk->sg_elems);
 
-	switch (req_op(req)) {
-	case REQ_OP_READ:
-	case REQ_OP_WRITE:
-		type = 0;
-		break;
-	case REQ_OP_FLUSH:
-		type = VIRTIO_BLK_T_FLUSH;
-		break;
-	case REQ_OP_DISCARD:
-		type = VIRTIO_BLK_T_DISCARD;
-		break;
-	case REQ_OP_WRITE_ZEROES:
-		type = VIRTIO_BLK_T_WRITE_ZEROES;
-		unmap = !(req->cmd_flags & REQ_NOUNMAP);
-		break;
-	case REQ_OP_DRV_IN:
-		type = VIRTIO_BLK_T_GET_ID;
-		break;
-	default:
-		WARN_ON_ONCE(1);
-		return BLK_STS_IOERR;
-	}
-
-	vbr->out_hdr.type = cpu_to_virtio32(vblk->vdev, type);
-	vbr->out_hdr.sector = type ?
-		0 : cpu_to_virtio64(vblk->vdev, blk_rq_pos(req));
-	vbr->out_hdr.ioprio = cpu_to_virtio32(vblk->vdev, req_get_ioprio(req));
+	err = virtblk_setup_cmd(vblk->vdev, req, vbr);
+	if (unlikely(err))
+		return err;
 
 	blk_mq_start_request(req);
 
-	if (type == VIRTIO_BLK_T_DISCARD || type == VIRTIO_BLK_T_WRITE_ZEROES) {
-		err = virtblk_setup_discard_write_zeroes(req, unmap);
-		if (err)
-			return BLK_STS_RESOURCE;
-	}
-
-	num = blk_rq_map_sg(hctx->queue, req, vbr->sg);
-	if (num) {
-		if (rq_data_dir(req) == WRITE)
-			vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_OUT);
-		else
-			vbr->out_hdr.type |= cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_IN);
+	num = virtblk_map_data(hctx, req, vbr);
+	if (unlikely(num < 0)) {
+		virtblk_cleanup_cmd(req);
+		return BLK_STS_RESOURCE;
 	}
 
 	spin_lock_irqsave(&vblk->vqs[qid].lock, flags);
-	err = virtblk_add_req(vblk->vqs[qid].vq, vbr, vbr->sg, num);
+	err = virtblk_add_req(vblk->vqs[qid].vq, vbr, vbr->sg_table.sgl, num);
 	if (err) {
 		virtqueue_kick(vblk->vqs[qid].vq);
 		/* Don't stop the queue if -ENOMEM: we may have failed to
@@ -284,6 +356,8 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 		if (err == -ENOSPC)
 			blk_mq_stop_hw_queue(hctx);
 		spin_unlock_irqrestore(&vblk->vqs[qid].lock, flags);
+		virtblk_unmap_data(req, vbr);
+		virtblk_cleanup_cmd(req);
 		switch (err) {
 		case -ENOSPC:
 			return BLK_STS_DEV_RESOURCE;
@@ -498,7 +572,9 @@ static int init_vq(struct virtio_blk *vblk)
 	if (err)
 		num_vqs = 1;
 
-	num_vqs = min_t(unsigned int, nr_cpu_ids, num_vqs);
+	num_vqs = min_t(unsigned int,
+			min_not_zero(num_request_queues, nr_cpu_ids),
+			num_vqs);
 
 	vblk->vqs = kmalloc_array(num_vqs, sizeof(*vblk->vqs), GFP_KERNEL);
 	if (!vblk->vqs)
@@ -660,16 +736,6 @@ static const struct attribute_group *virtblk_attr_groups[] = {
 	NULL,
 };
 
-static int virtblk_init_request(struct blk_mq_tag_set *set, struct request *rq,
-		unsigned int hctx_idx, unsigned int numa_node)
-{
-	struct virtio_blk *vblk = set->driver_data;
-	struct virtblk_req *vbr = blk_mq_rq_to_pdu(rq);
-
-	sg_init_table(vbr->sg, vblk->sg_elems);
-	return 0;
-}
-
 static int virtblk_map_queues(struct blk_mq_tag_set *set)
 {
 	struct virtio_blk *vblk = set->driver_data;
@@ -682,7 +748,6 @@ static const struct blk_mq_ops virtio_mq_ops = {
 	.queue_rq	= virtio_queue_rq,
 	.commit_rqs	= virtio_commit_rqs,
 	.complete	= virtblk_request_done,
-	.init_request	= virtblk_init_request,
 	.map_queues	= virtblk_map_queues,
 };
 
@@ -762,7 +827,7 @@ static int virtblk_probe(struct virtio_device *vdev)
 	vblk->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
 	vblk->tag_set.cmd_size =
 		sizeof(struct virtblk_req) +
-		sizeof(struct scatterlist) * sg_elems;
+		sizeof(struct scatterlist) * VIRTIO_BLK_INLINE_SG_CNT;
 	vblk->tag_set.driver_data = vblk;
 	vblk->tag_set.nr_hw_queues = vblk->num_vqs;
 

drivers/char/hw_random/virtio-rng.c

@@ -18,13 +18,20 @@ static DEFINE_IDA(rng_index_ida);
 struct virtrng_info {
 	struct hwrng hwrng;
 	struct virtqueue *vq;
-	struct completion have_data;
 	char name[25];
-	unsigned int data_avail;
 	int index;
-	bool busy;
 	bool hwrng_register_done;
 	bool hwrng_removed;
+	/* data transfer */
+	struct completion have_data;
+	unsigned int data_avail;
+	unsigned int data_idx;
+	/* minimal size returned by rng_buffer_size() */
+#if SMP_CACHE_BYTES < 32
+	u8 data[32];
+#else
+	u8 data[SMP_CACHE_BYTES];
+#endif
 };
 
 static void random_recv_done(struct virtqueue *vq)
@@ -35,54 +42,88 @@ static void random_recv_done(struct virtqueue *vq)
 	if (!virtqueue_get_buf(vi->vq, &vi->data_avail))
 		return;
 
+	vi->data_idx = 0;
+
 	complete(&vi->have_data);
 }
 
-/* The host will fill any buffer we give it with sweet, sweet randomness. */
-static void register_buffer(struct virtrng_info *vi, u8 *buf, size_t size)
+static void request_entropy(struct virtrng_info *vi)
 {
 	struct scatterlist sg;
 
-	sg_init_one(&sg, buf, size);
+	reinit_completion(&vi->have_data);
+	vi->data_avail = 0;
+	vi->data_idx = 0;
+
+	sg_init_one(&sg, vi->data, sizeof(vi->data));
 
 	/* There should always be room for one buffer. */
-	virtqueue_add_inbuf(vi->vq, &sg, 1, buf, GFP_KERNEL);
+	virtqueue_add_inbuf(vi->vq, &sg, 1, vi->data, GFP_KERNEL);
 
 	virtqueue_kick(vi->vq);
 }
 
+static unsigned int copy_data(struct virtrng_info *vi, void *buf,
+			      unsigned int size)
+{
+	size = min_t(unsigned int, size, vi->data_avail);
+	memcpy(buf, vi->data + vi->data_idx, size);
+	vi->data_idx += size;
+	vi->data_avail -= size;
+	if (vi->data_avail == 0)
+		request_entropy(vi);
+	return size;
+}
+
 static int virtio_read(struct hwrng *rng, void *buf, size_t size, bool wait)
 {
 	int ret;
 	struct virtrng_info *vi = (struct virtrng_info *)rng->priv;
+	unsigned int chunk;
+	size_t read;
 
 	if (vi->hwrng_removed)
 		return -ENODEV;
 
-	if (!vi->busy) {
-		vi->busy = true;
-		reinit_completion(&vi->have_data);
-		register_buffer(vi, buf, size);
+	read = 0;
+
+	/* copy available data */
+	if (vi->data_avail) {
+		chunk = copy_data(vi, buf, size);
+		size -= chunk;
+		read += chunk;
 	}
 
 	if (!wait)
-		return 0;
+		return read;
 
-	ret = wait_for_completion_killable(&vi->have_data);
-	if (ret < 0)
-		return ret;
+	/* We have already copied available entropy,
+	 * so either size is 0 or data_avail is 0
+	 */
+	while (size != 0) {
+		/* data_avail is 0 but a request is pending */
+		ret = wait_for_completion_killable(&vi->have_data);
+		if (ret < 0)
+			return ret;
+		/* if vi->data_avail is 0, we have been interrupted
+		 * by a cleanup, but buffer stays in the queue
+		 */
+		if (vi->data_avail == 0)
+			return read;
 
-	vi->busy = false;
+		chunk = copy_data(vi, buf + read, size);
+		size -= chunk;
+		read += chunk;
+	}
 
-	return vi->data_avail;
+	return read;
 }
 
 static void virtio_cleanup(struct hwrng *rng)
 {
 	struct virtrng_info *vi = (struct virtrng_info *)rng->priv;
 
-	if (vi->busy)
-		wait_for_completion(&vi->have_data);
+	complete(&vi->have_data);
 }
 
 static int probe_common(struct virtio_device *vdev)
@@ -118,6 +159,9 @@ static int probe_common(struct virtio_device *vdev)
 		goto err_find;
 	}
 
+	/* we always have a pending entropy request */
+	request_entropy(vi);
+
 	return 0;
 
 err_find:
@@ -133,9 +177,9 @@ static void remove_common(struct virtio_device *vdev)
 
 	vi->hwrng_removed = true;
 	vi->data_avail = 0;
+	vi->data_idx = 0;
 	complete(&vi->have_data);
 	vdev->config->reset(vdev);
-	vi->busy = false;
 	if (vi->hwrng_register_done)
 		hwrng_unregister(&vi->hwrng);
 	vdev->config->del_vqs(vdev);

drivers/net/virtio_net.c

@@ -408,12 +408,13 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
 	 * add_recvbuf_mergeable() + get_mergeable_buf_len()
 	 */
 	truesize = headroom ? PAGE_SIZE : truesize;
-	tailroom = truesize - len - headroom - (hdr_padded_len - hdr_len);
+	tailroom = truesize - headroom;
 	buf = p - headroom;
 
 	len -= hdr_len;
 	offset += hdr_padded_len;
 	p += hdr_padded_len;
+	tailroom -= hdr_padded_len + len;
 
 	shinfo_size = SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
 

drivers/vdpa/Kconfig

@@ -78,4 +78,12 @@ config VP_VDPA
 	help
 	  This kernel module bridges virtio PCI device to vDPA bus.
 
+config ALIBABA_ENI_VDPA
+	tristate "vDPA driver for Alibaba ENI"
+	select VIRTIO_PCI_LIB_LEGACY
+	depends on PCI_MSI && X86
+	help
+	  VDPA driver for Alibaba ENI (Elastic Network Interface) which is built upon
+	  virtio 0.9.5 specification.
+
 endif # VDPA

drivers/vdpa/Makefile

@@ -5,3 +5,4 @@ obj-$(CONFIG_VDPA_USER) += vdpa_user/
 obj-$(CONFIG_IFCVF)    += ifcvf/
 obj-$(CONFIG_MLX5_VDPA) += mlx5/
 obj-$(CONFIG_VP_VDPA)    += virtio_pci/
+obj-$(CONFIG_ALIBABA_ENI_VDPA) += alibaba/

drivers/vdpa/alibaba/Makefile

@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-$(CONFIG_ALIBABA_ENI_VDPA) += eni_vdpa.o
+

drivers/vdpa/alibaba/eni_vdpa.c

@@ -0,0 +1,553 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * vDPA bridge driver for Alibaba ENI(Elastic Network Interface)
+ *
+ * Copyright (c) 2021, Alibaba Inc. All rights reserved.
+ * Author: Wu Zongyong <wuzongyong@linux.alibaba.com>
+ *
+ */
+
+#include "linux/bits.h"
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/vdpa.h>
+#include <linux/virtio.h>
+#include <linux/virtio_config.h>
+#include <linux/virtio_ring.h>
+#include <linux/virtio_pci.h>
+#include <linux/virtio_pci_legacy.h>
+#include <uapi/linux/virtio_net.h>
+
+#define ENI_MSIX_NAME_SIZE 256
+
+#define ENI_ERR(pdev, fmt, ...)	\
+	dev_err(&pdev->dev, "%s"fmt, "eni_vdpa: ", ##__VA_ARGS__)
+#define ENI_DBG(pdev, fmt, ...)	\
+	dev_dbg(&pdev->dev, "%s"fmt, "eni_vdpa: ", ##__VA_ARGS__)
+#define ENI_INFO(pdev, fmt, ...) \
+	dev_info(&pdev->dev, "%s"fmt, "eni_vdpa: ", ##__VA_ARGS__)
+
+struct eni_vring {
+	void __iomem *notify;
+	char msix_name[ENI_MSIX_NAME_SIZE];
+	struct vdpa_callback cb;
+	int irq;
+};
+
+struct eni_vdpa {
+	struct vdpa_device vdpa;
+	struct virtio_pci_legacy_device ldev;
+	struct eni_vring *vring;
+	struct vdpa_callback config_cb;
+	char msix_name[ENI_MSIX_NAME_SIZE];
+	int config_irq;
+	int queues;
+	int vectors;
+};
+
+static struct eni_vdpa *vdpa_to_eni(struct vdpa_device *vdpa)
+{
+	return container_of(vdpa, struct eni_vdpa, vdpa);
+}
+
+static struct virtio_pci_legacy_device *vdpa_to_ldev(struct vdpa_device *vdpa)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+
+	return &eni_vdpa->ldev;
+}
+
+static u64 eni_vdpa_get_features(struct vdpa_device *vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+	u64 features = vp_legacy_get_features(ldev);
+
+	features |= BIT_ULL(VIRTIO_F_ACCESS_PLATFORM);
+	features |= BIT_ULL(VIRTIO_F_ORDER_PLATFORM);
+
+	return features;
+}
+
+static int eni_vdpa_set_features(struct vdpa_device *vdpa, u64 features)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	if (!(features & BIT_ULL(VIRTIO_NET_F_MRG_RXBUF)) && features) {
+		ENI_ERR(ldev->pci_dev,
+			"VIRTIO_NET_F_MRG_RXBUF is not negotiated\n");
+		return -EINVAL;
+	}
+
+	vp_legacy_set_features(ldev, (u32)features);
+
+	return 0;
+}
+
+static u8 eni_vdpa_get_status(struct vdpa_device *vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	return vp_legacy_get_status(ldev);
+}
+
+static int eni_vdpa_get_vq_irq(struct vdpa_device *vdpa, u16 idx)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+	int irq = eni_vdpa->vring[idx].irq;
+
+	if (irq == VIRTIO_MSI_NO_VECTOR)
+		return -EINVAL;
+
+	return irq;
+}
+
+static void eni_vdpa_free_irq(struct eni_vdpa *eni_vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = &eni_vdpa->ldev;
+	struct pci_dev *pdev = ldev->pci_dev;
+	int i;
+
+	for (i = 0; i < eni_vdpa->queues; i++) {
+		if (eni_vdpa->vring[i].irq != VIRTIO_MSI_NO_VECTOR) {
+			vp_legacy_queue_vector(ldev, i, VIRTIO_MSI_NO_VECTOR);
+			devm_free_irq(&pdev->dev, eni_vdpa->vring[i].irq,
+				      &eni_vdpa->vring[i]);
+			eni_vdpa->vring[i].irq = VIRTIO_MSI_NO_VECTOR;
+		}
+	}
+
+	if (eni_vdpa->config_irq != VIRTIO_MSI_NO_VECTOR) {
+		vp_legacy_config_vector(ldev, VIRTIO_MSI_NO_VECTOR);
+		devm_free_irq(&pdev->dev, eni_vdpa->config_irq, eni_vdpa);
+		eni_vdpa->config_irq = VIRTIO_MSI_NO_VECTOR;
+	}
+
+	if (eni_vdpa->vectors) {
+		pci_free_irq_vectors(pdev);
+		eni_vdpa->vectors = 0;
+	}
+}
+
+static irqreturn_t eni_vdpa_vq_handler(int irq, void *arg)
+{
+	struct eni_vring *vring = arg;
+
+	if (vring->cb.callback)
+		return vring->cb.callback(vring->cb.private);
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t eni_vdpa_config_handler(int irq, void *arg)
+{
+	struct eni_vdpa *eni_vdpa = arg;
+
+	if (eni_vdpa->config_cb.callback)
+		return eni_vdpa->config_cb.callback(eni_vdpa->config_cb.private);
+
+	return IRQ_HANDLED;
+}
+
+static int eni_vdpa_request_irq(struct eni_vdpa *eni_vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = &eni_vdpa->ldev;
+	struct pci_dev *pdev = ldev->pci_dev;
+	int i, ret, irq;
+	int queues = eni_vdpa->queues;
+	int vectors = queues + 1;
+
+	ret = pci_alloc_irq_vectors(pdev, vectors, vectors, PCI_IRQ_MSIX);
+	if (ret != vectors) {
+		ENI_ERR(pdev,
+			"failed to allocate irq vectors want %d but %d\n",
+			vectors, ret);
+		return ret;
+	}
+
+	eni_vdpa->vectors = vectors;
+
+	for (i = 0; i < queues; i++) {
+		snprintf(eni_vdpa->vring[i].msix_name, ENI_MSIX_NAME_SIZE,
+			 "eni-vdpa[%s]-%d\n", pci_name(pdev), i);
+		irq = pci_irq_vector(pdev, i);
+		ret = devm_request_irq(&pdev->dev, irq,
+				       eni_vdpa_vq_handler,
+				       0, eni_vdpa->vring[i].msix_name,
+				       &eni_vdpa->vring[i]);
+		if (ret) {
+			ENI_ERR(pdev, "failed to request irq for vq %d\n", i);
+			goto err;
+		}
+		vp_legacy_queue_vector(ldev, i, i);
+		eni_vdpa->vring[i].irq = irq;
+	}
+
+	snprintf(eni_vdpa->msix_name, ENI_MSIX_NAME_SIZE, "eni-vdpa[%s]-config\n",
+		 pci_name(pdev));
+	irq = pci_irq_vector(pdev, queues);
+	ret = devm_request_irq(&pdev->dev, irq, eni_vdpa_config_handler, 0,
+			       eni_vdpa->msix_name, eni_vdpa);
+	if (ret) {
+		ENI_ERR(pdev, "failed to request irq for config vq %d\n", i);
+		goto err;
+	}
+	vp_legacy_config_vector(ldev, queues);
+	eni_vdpa->config_irq = irq;
+
+	return 0;
+err:
+	eni_vdpa_free_irq(eni_vdpa);
+	return ret;
+}
+
+static void eni_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+	struct virtio_pci_legacy_device *ldev = &eni_vdpa->ldev;
+	u8 s = eni_vdpa_get_status(vdpa);
+
+	if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
+	    !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
+		eni_vdpa_request_irq(eni_vdpa);
+	}
+
+	vp_legacy_set_status(ldev, status);
+
+	if (!(status & VIRTIO_CONFIG_S_DRIVER_OK) &&
+	    (s & VIRTIO_CONFIG_S_DRIVER_OK))
+		eni_vdpa_free_irq(eni_vdpa);
+}
+
+static int eni_vdpa_reset(struct vdpa_device *vdpa)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+	struct virtio_pci_legacy_device *ldev = &eni_vdpa->ldev;
+	u8 s = eni_vdpa_get_status(vdpa);
+
+	vp_legacy_set_status(ldev, 0);
+
+	if (s & VIRTIO_CONFIG_S_DRIVER_OK)
+		eni_vdpa_free_irq(eni_vdpa);
+
+	return 0;
+}
+
+static u16 eni_vdpa_get_vq_num_max(struct vdpa_device *vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	return vp_legacy_get_queue_size(ldev, 0);
+}
+
+static u16 eni_vdpa_get_vq_num_min(struct vdpa_device *vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	return vp_legacy_get_queue_size(ldev, 0);
+}
+
+static int eni_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid,
+				struct vdpa_vq_state *state)
+{
+	return -EOPNOTSUPP;
+}
+
+static int eni_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
+				 const struct vdpa_vq_state *state)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+	const struct vdpa_vq_state_split *split = &state->split;
+
+	/* ENI is build upon virtio-pci specfication which not support
+	 * to set state of virtqueue. But if the state is equal to the
+	 * device initial state by chance, we can let it go.
+	 */
+	if (!vp_legacy_get_queue_enable(ldev, qid)
+	    && split->avail_index == 0)
+		return 0;
+
+	return -EOPNOTSUPP;
+}
+
+
+static void eni_vdpa_set_vq_cb(struct vdpa_device *vdpa, u16 qid,
+			       struct vdpa_callback *cb)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+
+	eni_vdpa->vring[qid].cb = *cb;
+}
+
+static void eni_vdpa_set_vq_ready(struct vdpa_device *vdpa, u16 qid,
+				  bool ready)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	/* ENI is a legacy virtio-pci device. This is not supported
+	 * by specification. But we can disable virtqueue by setting
+	 * address to 0.
+	 */
+	if (!ready)
+		vp_legacy_set_queue_address(ldev, qid, 0);
+}
+
+static bool eni_vdpa_get_vq_ready(struct vdpa_device *vdpa, u16 qid)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	return vp_legacy_get_queue_enable(ldev, qid);
+}
+
+static void eni_vdpa_set_vq_num(struct vdpa_device *vdpa, u16 qid,
+			       u32 num)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+	struct pci_dev *pdev = ldev->pci_dev;
+	u16 n = vp_legacy_get_queue_size(ldev, qid);
+
+	/* ENI is a legacy virtio-pci device which not allow to change
+	 * virtqueue size. Just report a error if someone tries to
+	 * change it.
+	 */
+	if (num != n)
+		ENI_ERR(pdev,
+			"not support to set vq %u fixed num %u to %u\n",
+			qid, n, num);
+}
+
+static int eni_vdpa_set_vq_address(struct vdpa_device *vdpa, u16 qid,
+				   u64 desc_area, u64 driver_area,
+				   u64 device_area)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+	u32 pfn = desc_area >> VIRTIO_PCI_QUEUE_ADDR_SHIFT;
+
+	vp_legacy_set_queue_address(ldev, qid, pfn);
+
+	return 0;
+}
+
+static void eni_vdpa_kick_vq(struct vdpa_device *vdpa, u16 qid)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+
+	iowrite16(qid, eni_vdpa->vring[qid].notify);
+}
+
+static u32 eni_vdpa_get_device_id(struct vdpa_device *vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	return ldev->id.device;
+}
+
+static u32 eni_vdpa_get_vendor_id(struct vdpa_device *vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = vdpa_to_ldev(vdpa);
+
+	return ldev->id.vendor;
+}
+
+static u32 eni_vdpa_get_vq_align(struct vdpa_device *vdpa)
+{
+	return VIRTIO_PCI_VRING_ALIGN;
+}
+
+static size_t eni_vdpa_get_config_size(struct vdpa_device *vdpa)
+{
+	return sizeof(struct virtio_net_config);
+}
+
+
+static void eni_vdpa_get_config(struct vdpa_device *vdpa,
+				unsigned int offset,
+				void *buf, unsigned int len)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+	struct virtio_pci_legacy_device *ldev = &eni_vdpa->ldev;
+	void __iomem *ioaddr = ldev->ioaddr +
+		VIRTIO_PCI_CONFIG_OFF(eni_vdpa->vectors) +
+		offset;
+	u8 *p = buf;
+	int i;
+
+	for (i = 0; i < len; i++)
+		*p++ = ioread8(ioaddr + i);
+}
+
+static void eni_vdpa_set_config(struct vdpa_device *vdpa,
+				unsigned int offset, const void *buf,
+				unsigned int len)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+	struct virtio_pci_legacy_device *ldev = &eni_vdpa->ldev;
+	void __iomem *ioaddr = ldev->ioaddr +
+		VIRTIO_PCI_CONFIG_OFF(eni_vdpa->vectors) +
+		offset;
+	const u8 *p = buf;
+	int i;
+
+	for (i = 0; i < len; i++)
+		iowrite8(*p++, ioaddr + i);
+}
+
+static void eni_vdpa_set_config_cb(struct vdpa_device *vdpa,
+				   struct vdpa_callback *cb)
+{
+	struct eni_vdpa *eni_vdpa = vdpa_to_eni(vdpa);
+
+	eni_vdpa->config_cb = *cb;
+}
+
+static const struct vdpa_config_ops eni_vdpa_ops = {
+	.get_features	= eni_vdpa_get_features,
+	.set_features	= eni_vdpa_set_features,
+	.get_status	= eni_vdpa_get_status,
+	.set_status	= eni_vdpa_set_status,
+	.reset		= eni_vdpa_reset,
+	.get_vq_num_max	= eni_vdpa_get_vq_num_max,
+	.get_vq_num_min	= eni_vdpa_get_vq_num_min,
+	.get_vq_state	= eni_vdpa_get_vq_state,
+	.set_vq_state	= eni_vdpa_set_vq_state,
+	.set_vq_cb	= eni_vdpa_set_vq_cb,
+	.set_vq_ready	= eni_vdpa_set_vq_ready,
+	.get_vq_ready	= eni_vdpa_get_vq_ready,
+	.set_vq_num	= eni_vdpa_set_vq_num,
+	.set_vq_address	= eni_vdpa_set_vq_address,
+	.kick_vq	= eni_vdpa_kick_vq,
+	.get_device_id	= eni_vdpa_get_device_id,
+	.get_vendor_id	= eni_vdpa_get_vendor_id,
+	.get_vq_align	= eni_vdpa_get_vq_align,
+	.get_config_size = eni_vdpa_get_config_size,
+	.get_config	= eni_vdpa_get_config,
+	.set_config	= eni_vdpa_set_config,
+	.set_config_cb  = eni_vdpa_set_config_cb,
+	.get_vq_irq	= eni_vdpa_get_vq_irq,
+};
+
+
+static u16 eni_vdpa_get_num_queues(struct eni_vdpa *eni_vdpa)
+{
+	struct virtio_pci_legacy_device *ldev = &eni_vdpa->ldev;
+	u32 features = vp_legacy_get_features(ldev);
+	u16 num = 2;
+
+	if (features & BIT_ULL(VIRTIO_NET_F_MQ)) {
+		__virtio16 max_virtqueue_pairs;
+
+		eni_vdpa_get_config(&eni_vdpa->vdpa,
+			offsetof(struct virtio_net_config, max_virtqueue_pairs),
+			&max_virtqueue_pairs,
+			sizeof(max_virtqueue_pairs));
+		num = 2 * __virtio16_to_cpu(virtio_legacy_is_little_endian(),
+				max_virtqueue_pairs);
+	}
+
+	if (features & BIT_ULL(VIRTIO_NET_F_CTRL_VQ))
+		num += 1;
+
+	return num;
+}
+
+static void eni_vdpa_free_irq_vectors(void *data)
+{
+	pci_free_irq_vectors(data);
+}
+
+static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
+{
+	struct device *dev = &pdev->dev;
+	struct eni_vdpa *eni_vdpa;
+	struct virtio_pci_legacy_device *ldev;
+	int ret, i;
+
+	ret = pcim_enable_device(pdev);
+	if (ret)
+		return ret;
+
+	eni_vdpa = vdpa_alloc_device(struct eni_vdpa, vdpa,
+				     dev, &eni_vdpa_ops, NULL, false);
+	if (IS_ERR(eni_vdpa)) {
+		ENI_ERR(pdev, "failed to allocate vDPA structure\n");
+		return PTR_ERR(eni_vdpa);
+	}
+
+	ldev = &eni_vdpa->ldev;
+	ldev->pci_dev = pdev;
+
+	ret = vp_legacy_probe(ldev);
+	if (ret) {
+		ENI_ERR(pdev, "failed to probe legacy PCI device\n");
+		goto err;
+	}
+
+	pci_set_master(pdev);
+	pci_set_drvdata(pdev, eni_vdpa);
+
+	eni_vdpa->vdpa.dma_dev = &pdev->dev;
+	eni_vdpa->queues = eni_vdpa_get_num_queues(eni_vdpa);
+
+	ret = devm_add_action_or_reset(dev, eni_vdpa_free_irq_vectors, pdev);
+	if (ret) {
+		ENI_ERR(pdev,
+			"failed for adding devres for freeing irq vectors\n");
+		goto err;
+	}
+
+	eni_vdpa->vring = devm_kcalloc(&pdev->dev, eni_vdpa->queues,
+				      sizeof(*eni_vdpa->vring),
+				      GFP_KERNEL);
+	if (!eni_vdpa->vring) {
+		ret = -ENOMEM;
+		ENI_ERR(pdev, "failed to allocate virtqueues\n");
+		goto err;
+	}
+
+	for (i = 0; i < eni_vdpa->queues; i++) {
+		eni_vdpa->vring[i].irq = VIRTIO_MSI_NO_VECTOR;
+		eni_vdpa->vring[i].notify = ldev->ioaddr + VIRTIO_PCI_QUEUE_NOTIFY;
+	}
+	eni_vdpa->config_irq = VIRTIO_MSI_NO_VECTOR;
+
+	ret = vdpa_register_device(&eni_vdpa->vdpa, eni_vdpa->queues);
+	if (ret) {
+		ENI_ERR(pdev, "failed to register to vdpa bus\n");
+		goto err;
+	}
+
+	return 0;
+
+err:
+	put_device(&eni_vdpa->vdpa.dev);
+	return ret;
+}
+
+static void eni_vdpa_remove(struct pci_dev *pdev)
+{
+	struct eni_vdpa *eni_vdpa = pci_get_drvdata(pdev);
+
+	vdpa_unregister_device(&eni_vdpa->vdpa);
+	vp_legacy_remove(&eni_vdpa->ldev);
+}
+
+static struct pci_device_id eni_pci_ids[] = {
+	{ PCI_DEVICE_SUB(PCI_VENDOR_ID_REDHAT_QUMRANET,
+			 VIRTIO_TRANS_ID_NET,
+			 PCI_SUBVENDOR_ID_REDHAT_QUMRANET,
+			 VIRTIO_ID_NET) },
+	{ 0 },
+};
+
+static struct pci_driver eni_vdpa_driver = {
+	.name		= "alibaba-eni-vdpa",
+	.id_table	= eni_pci_ids,
+	.probe		= eni_vdpa_probe,
+	.remove		= eni_vdpa_remove,
+};
+
+module_pci_driver(eni_vdpa_driver);
+
+MODULE_AUTHOR("Wu Zongyong <wuzongyong@linux.alibaba.com>");
+MODULE_DESCRIPTION("Alibaba ENI vDPA driver");
+MODULE_LICENSE("GPL v2");

drivers/vdpa/mlx5/core/mlx5_vdpa.h

@@ -63,7 +63,7 @@ struct mlx5_control_vq {
 	unsigned short head;
 };
 
-struct mlx5_ctrl_wq_ent {
+struct mlx5_vdpa_wq_ent {
 	struct work_struct work;
 	struct mlx5_vdpa_dev *mvdev;
 };

drivers/vdpa/mlx5/net/mlx5_vnet.c

@@ -159,8 +159,9 @@ struct mlx5_vdpa_net {
 	struct mlx5_fc *rx_counter;
 	struct mlx5_flow_handle *rx_rule;
 	bool setup;
-	u16 mtu;
 	u32 cur_num_vqs;
+	struct notifier_block nb;
+	struct vdpa_callback config_cb;
 };
 
 static void free_resources(struct mlx5_vdpa_net *ndev);
@@ -1557,14 +1558,14 @@ static void mlx5_cvq_kick_handler(struct work_struct *work)
 {
 	virtio_net_ctrl_ack status = VIRTIO_NET_ERR;
 	struct virtio_net_ctrl_hdr ctrl;
-	struct mlx5_ctrl_wq_ent *wqent;
+	struct mlx5_vdpa_wq_ent *wqent;
 	struct mlx5_vdpa_dev *mvdev;
 	struct mlx5_control_vq *cvq;
 	struct mlx5_vdpa_net *ndev;
 	size_t read, write;
 	int err;
 
-	wqent = container_of(work, struct mlx5_ctrl_wq_ent, work);
+	wqent = container_of(work, struct mlx5_vdpa_wq_ent, work);
 	mvdev = wqent->mvdev;
 	ndev = to_mlx5_vdpa_ndev(mvdev);
 	cvq = &mvdev->cvq;
@@ -1616,7 +1617,7 @@ static void mlx5_vdpa_kick_vq(struct vdpa_device *vdev, u16 idx)
 	struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev);
 	struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
 	struct mlx5_vdpa_virtqueue *mvq;
-	struct mlx5_ctrl_wq_ent *wqent;
+	struct mlx5_vdpa_wq_ent *wqent;
 
 	if (!is_index_valid(mvdev, idx))
 		return;
@@ -1852,6 +1853,7 @@ static u64 mlx5_vdpa_get_features(struct vdpa_device *vdev)
 	ndev->mvdev.mlx_features |= BIT_ULL(VIRTIO_NET_F_CTRL_VQ);
 	ndev->mvdev.mlx_features |= BIT_ULL(VIRTIO_NET_F_CTRL_MAC_ADDR);
 	ndev->mvdev.mlx_features |= BIT_ULL(VIRTIO_NET_F_MQ);
+	ndev->mvdev.mlx_features |= BIT_ULL(VIRTIO_NET_F_STATUS);
 
 	print_features(mvdev, ndev->mvdev.mlx_features, false);
 	return ndev->mvdev.mlx_features;
@@ -1942,16 +1944,16 @@ static int mlx5_vdpa_set_features(struct vdpa_device *vdev, u64 features)
 		return err;
 
 	ndev->mvdev.actual_features = features & ndev->mvdev.mlx_features;
-	ndev->config.mtu = cpu_to_mlx5vdpa16(mvdev, ndev->mtu);
-	ndev->config.status |= cpu_to_mlx5vdpa16(mvdev, VIRTIO_NET_S_LINK_UP);
 	update_cvq_info(mvdev);
 	return err;
 }
 
 static void mlx5_vdpa_set_config_cb(struct vdpa_device *vdev, struct vdpa_callback *cb)
 {
-	/* not implemented */
-	mlx5_vdpa_warn(to_mvdev(vdev), "set config callback not supported\n");
+	struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev);
+	struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
+
+	ndev->config_cb = *cb;
 }
 
 #define MLX5_VDPA_MAX_VQ_ENTRIES 256
@@ -2404,6 +2406,82 @@ struct mlx5_vdpa_mgmtdev {
 	struct mlx5_vdpa_net *ndev;
 };
 
+static u8 query_vport_state(struct mlx5_core_dev *mdev, u8 opmod, u16 vport)
+{
+	u32 out[MLX5_ST_SZ_DW(query_vport_state_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_vport_state_in)] = {};
+	int err;
+
+	MLX5_SET(query_vport_state_in, in, opcode, MLX5_CMD_OP_QUERY_VPORT_STATE);
+	MLX5_SET(query_vport_state_in, in, op_mod, opmod);
+	MLX5_SET(query_vport_state_in, in, vport_number, vport);
+	if (vport)
+		MLX5_SET(query_vport_state_in, in, other_vport, 1);
+
+	err = mlx5_cmd_exec_inout(mdev, query_vport_state, in, out);
+	if (err)
+		return 0;
+
+	return MLX5_GET(query_vport_state_out, out, state);
+}
+
+static bool get_link_state(struct mlx5_vdpa_dev *mvdev)
+{
+	if (query_vport_state(mvdev->mdev, MLX5_VPORT_STATE_OP_MOD_VNIC_VPORT, 0) ==
+	    VPORT_STATE_UP)
+		return true;
+
+	return false;
+}
+
+static void update_carrier(struct work_struct *work)
+{
+	struct mlx5_vdpa_wq_ent *wqent;
+	struct mlx5_vdpa_dev *mvdev;
+	struct mlx5_vdpa_net *ndev;
+
+	wqent = container_of(work, struct mlx5_vdpa_wq_ent, work);
+	mvdev = wqent->mvdev;
+	ndev = to_mlx5_vdpa_ndev(mvdev);
+	if (get_link_state(mvdev))
+		ndev->config.status |= cpu_to_mlx5vdpa16(mvdev, VIRTIO_NET_S_LINK_UP);
+	else
+		ndev->config.status &= cpu_to_mlx5vdpa16(mvdev, ~VIRTIO_NET_S_LINK_UP);
+
+	if (ndev->config_cb.callback)
+		ndev->config_cb.callback(ndev->config_cb.private);
+
+	kfree(wqent);
+}
+
+static int event_handler(struct notifier_block *nb, unsigned long event, void *param)
+{
+	struct mlx5_vdpa_net *ndev = container_of(nb, struct mlx5_vdpa_net, nb);
+	struct mlx5_eqe *eqe = param;
+	int ret = NOTIFY_DONE;
+	struct mlx5_vdpa_wq_ent *wqent;
+
+	if (event == MLX5_EVENT_TYPE_PORT_CHANGE) {
+		switch (eqe->sub_type) {
+		case MLX5_PORT_CHANGE_SUBTYPE_DOWN:
+		case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE:
+			wqent = kzalloc(sizeof(*wqent), GFP_ATOMIC);
+			if (!wqent)
+				return NOTIFY_DONE;
+
+			wqent->mvdev = &ndev->mvdev;
+			INIT_WORK(&wqent->work, update_carrier);
+			queue_work(ndev->mvdev.wq, &wqent->work);
+			ret = NOTIFY_OK;
+			break;
+		default:
+			return NOTIFY_DONE;
+		}
+		return ret;
+	}
+	return ret;
+}
+
 static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name)
 {
 	struct mlx5_vdpa_mgmtdev *mgtdev = container_of(v_mdev, struct mlx5_vdpa_mgmtdev, mgtdev);
@@ -2413,6 +2491,7 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name)
 	struct mlx5_vdpa_net *ndev;
 	struct mlx5_core_dev *mdev;
 	u32 max_vqs;
+	u16 mtu;
 	int err;
 
 	if (mgtdev->ndev)
@@ -2440,14 +2519,22 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name)
 	init_mvqs(ndev);
 	mutex_init(&ndev->reslock);
 	config = &ndev->config;
-	err = query_mtu(mdev, &ndev->mtu);
+	err = query_mtu(mdev, &mtu);
 	if (err)
 		goto err_mtu;
 
+	ndev->config.mtu = cpu_to_mlx5vdpa16(mvdev, mtu);
+	ndev->config.status |= cpu_to_mlx5vdpa16(mvdev, VIRTIO_NET_S_LINK_UP);
+
 	err = mlx5_query_nic_vport_mac_address(mdev, 0, 0, config->mac);
 	if (err)
 		goto err_mtu;
 
+	if (get_link_state(mvdev))
+		ndev->config.status |= cpu_to_mlx5vdpa16(mvdev, VIRTIO_NET_S_LINK_UP);
+	else
+		ndev->config.status &= cpu_to_mlx5vdpa16(mvdev, ~VIRTIO_NET_S_LINK_UP);
+
 	if (!is_zero_ether_addr(config->mac)) {
 		pfmdev = pci_get_drvdata(pci_physfn(mdev->pdev));
 		err = mlx5_mpfs_add_mac(pfmdev, config->mac);
@@ -2473,12 +2560,14 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name)
 	if (err)
 		goto err_mr;
 
-	mvdev->wq = create_singlethread_workqueue("mlx5_vdpa_ctrl_wq");
+	mvdev->wq = create_singlethread_workqueue("mlx5_vdpa_wq");
 	if (!mvdev->wq) {
 		err = -ENOMEM;
 		goto err_res2;
 	}
 
+	ndev->nb.notifier_call = event_handler;
+	mlx5_notifier_register(mdev, &ndev->nb);
 	ndev->cur_num_vqs = 2 * mlx5_vdpa_max_qps(max_vqs);
 	mvdev->vdev.mdev = &mgtdev->mgtdev;
 	err = _vdpa_register_device(&mvdev->vdev, ndev->cur_num_vqs + 1);
@@ -2509,7 +2598,9 @@ static void mlx5_vdpa_dev_del(struct vdpa_mgmt_dev *v_mdev, struct vdpa_device *
 {
 	struct mlx5_vdpa_mgmtdev *mgtdev = container_of(v_mdev, struct mlx5_vdpa_mgmtdev, mgtdev);
 	struct mlx5_vdpa_dev *mvdev = to_mvdev(dev);
+	struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
 
+	mlx5_notifier_unregister(mvdev->mdev, &ndev->nb);
 	destroy_workqueue(mvdev->wq);
 	_vdpa_unregister_device(dev);
 	mgtdev->ndev = NULL;

drivers/vdpa/vdpa.c

@@ -26,8 +26,16 @@ static int vdpa_dev_probe(struct device *d)
 {
 	struct vdpa_device *vdev = dev_to_vdpa(d);
 	struct vdpa_driver *drv = drv_to_vdpa(vdev->dev.driver);
+	const struct vdpa_config_ops *ops = vdev->config;
+	u32 max_num, min_num = 1;
 	int ret = 0;
 
+	max_num = ops->get_vq_num_max(vdev);
+	if (ops->get_vq_num_min)
+		min_num = ops->get_vq_num_min(vdev);
+	if (max_num < min_num)
+		return -EINVAL;
+
 	if (drv && drv->probe)
 		ret = drv->probe(vdev);
 
@@ -492,6 +500,7 @@ vdpa_dev_fill(struct vdpa_device *vdev, struct sk_buff *msg, u32 portid, u32 seq
 	      int flags, struct netlink_ext_ack *extack)
 {
 	u16 max_vq_size;
+	u16 min_vq_size = 1;
 	u32 device_id;
 	u32 vendor_id;
 	void *hdr;
@@ -508,6 +517,8 @@ vdpa_dev_fill(struct vdpa_device *vdev, struct sk_buff *msg, u32 portid, u32 seq
 	device_id = vdev->config->get_device_id(vdev);
 	vendor_id = vdev->config->get_vendor_id(vdev);
 	max_vq_size = vdev->config->get_vq_num_max(vdev);
+	if (vdev->config->get_vq_num_min)
+		min_vq_size = vdev->config->get_vq_num_min(vdev);
 
 	err = -EMSGSIZE;
 	if (nla_put_string(msg, VDPA_ATTR_DEV_NAME, dev_name(&vdev->dev)))
@@ -520,6 +531,8 @@ vdpa_dev_fill(struct vdpa_device *vdev, struct sk_buff *msg, u32 portid, u32 seq
 		goto msg_err;
 	if (nla_put_u16(msg, VDPA_ATTR_DEV_MAX_VQ_SIZE, max_vq_size))
 		goto msg_err;
+	if (nla_put_u16(msg, VDPA_ATTR_DEV_MIN_VQ_SIZE, min_vq_size))
+		goto msg_err;
 
 	genlmsg_end(msg, hdr);
 	return 0;

drivers/vdpa/virtio_pci/vp_vdpa.c

@@ -76,6 +76,17 @@ static u8 vp_vdpa_get_status(struct vdpa_device *vdpa)
 	return vp_modern_get_status(mdev);
 }
 
+static int vp_vdpa_get_vq_irq(struct vdpa_device *vdpa, u16 idx)
+{
+	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
+	int irq = vp_vdpa->vring[idx].irq;
+
+	if (irq == VIRTIO_MSI_NO_VECTOR)
+		return -EINVAL;
+
+	return irq;
+}
+
 static void vp_vdpa_free_irq(struct vp_vdpa *vp_vdpa)
 {
 	struct virtio_pci_modern_device *mdev = &vp_vdpa->mdev;
@@ -427,6 +438,7 @@ static const struct vdpa_config_ops vp_vdpa_ops = {
 	.get_config	= vp_vdpa_get_config,
 	.set_config	= vp_vdpa_set_config,
 	.set_config_cb  = vp_vdpa_set_config_cb,
+	.get_vq_irq	= vp_vdpa_get_vq_irq,
 };
 
 static void vp_vdpa_free_irq_vectors(void *data)

drivers/virtio/Kconfig

@@ -20,6 +20,15 @@ config VIRTIO_PCI_LIB
 	  PCI device with possible vendor specific extensions. Any
 	  module that selects this module must depend on PCI.
 
+config VIRTIO_PCI_LIB_LEGACY
+	tristate
+	help
+	  Legacy PCI device (Virtio PCI Card 0.9.x Draft and older device)
+	  implementation.
+	  This module implements the basic probe and control for devices
+	  which are based on legacy PCI device. Any module that selects this
+	  module must depend on PCI.
+
 menuconfig VIRTIO_MENU
 	bool "Virtio drivers"
 	default y
@@ -43,6 +52,7 @@ config VIRTIO_PCI_LEGACY
 	bool "Support for legacy virtio draft 0.9.X and older devices"
 	default y
 	depends on VIRTIO_PCI
+	select VIRTIO_PCI_LIB_LEGACY
 	help
           Virtio PCI Card 0.9.X Draft (circa 2014) and older device support.
 

drivers/virtio/Makefile

@@ -1,6 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-$(CONFIG_VIRTIO) += virtio.o virtio_ring.o
 obj-$(CONFIG_VIRTIO_PCI_LIB) += virtio_pci_modern_dev.o
+obj-$(CONFIG_VIRTIO_PCI_LIB_LEGACY) += virtio_pci_legacy_dev.o
 obj-$(CONFIG_VIRTIO_MMIO) += virtio_mmio.o
 obj-$(CONFIG_VIRTIO_PCI) += virtio_pci.o
 virtio_pci-y := virtio_pci_modern.o virtio_pci_common.o

drivers/virtio/virtio_pci_common.c

@@ -549,6 +549,8 @@ static int virtio_pci_probe(struct pci_dev *pci_dev,
 
 	pci_set_master(pci_dev);
 
+	vp_dev->is_legacy = vp_dev->ldev.ioaddr ? true : false;
+
 	rc = register_virtio_device(&vp_dev->vdev);
 	reg_dev = vp_dev;
 	if (rc)
@@ -557,10 +559,10 @@ static int virtio_pci_probe(struct pci_dev *pci_dev,
 	return 0;
 
 err_register:
-	if (vp_dev->ioaddr)
-	     virtio_pci_legacy_remove(vp_dev);
+	if (vp_dev->is_legacy)
+		virtio_pci_legacy_remove(vp_dev);
 	else
-	     virtio_pci_modern_remove(vp_dev);
+		virtio_pci_modern_remove(vp_dev);
 err_probe:
 	pci_disable_device(pci_dev);
 err_enable_device:
@@ -587,7 +589,7 @@ static void virtio_pci_remove(struct pci_dev *pci_dev)
 
 	unregister_virtio_device(&vp_dev->vdev);
 
-	if (vp_dev->ioaddr)
+	if (vp_dev->is_legacy)
 		virtio_pci_legacy_remove(vp_dev);
 	else
 		virtio_pci_modern_remove(vp_dev);

drivers/virtio/virtio_pci_common.h

@@ -25,6 +25,7 @@
 #include <linux/virtio_config.h>
 #include <linux/virtio_ring.h>
 #include <linux/virtio_pci.h>
+#include <linux/virtio_pci_legacy.h>
 #include <linux/virtio_pci_modern.h>
 #include <linux/highmem.h>
 #include <linux/spinlock.h>
@@ -44,16 +45,14 @@ struct virtio_pci_vq_info {
 struct virtio_pci_device {
 	struct virtio_device vdev;
 	struct pci_dev *pci_dev;
+	struct virtio_pci_legacy_device ldev;
 	struct virtio_pci_modern_device mdev;
 
-	/* In legacy mode, these two point to within ->legacy. */
+	bool is_legacy;
+
 	/* Where to read and clear interrupt */
 	u8 __iomem *isr;
 
-	/* Legacy only field */
-	/* the IO mapping for the PCI config space */
-	void __iomem *ioaddr;
-
 	/* a list of queues so we can dispatch IRQs */
 	spinlock_t lock;
 	struct list_head virtqueues;

drivers/virtio/virtio_pci_legacy.c

@@ -14,6 +14,7 @@
  *  Michael S. Tsirkin <mst@redhat.com>
  */
 
+#include "linux/virtio_pci_legacy.h"
 #include "virtio_pci_common.h"
 
 /* virtio config->get_features() implementation */
@@ -23,7 +24,7 @@ static u64 vp_get_features(struct virtio_device *vdev)
 
 	/* When someone needs more than 32 feature bits, we'll need to
 	 * steal a bit to indicate that the rest are somewhere else. */
-	return ioread32(vp_dev->ioaddr + VIRTIO_PCI_HOST_FEATURES);
+	return vp_legacy_get_features(&vp_dev->ldev);
 }
 
 /* virtio config->finalize_features() implementation */
@@ -38,7 +39,7 @@ static int vp_finalize_features(struct virtio_device *vdev)
 	BUG_ON((u32)vdev->features != vdev->features);
 
 	/* We only support 32 feature bits. */
-	iowrite32(vdev->features, vp_dev->ioaddr + VIRTIO_PCI_GUEST_FEATURES);
+	vp_legacy_set_features(&vp_dev->ldev, vdev->features);
 
 	return 0;
 }
@@ -48,7 +49,7 @@ static void vp_get(struct virtio_device *vdev, unsigned offset,
 		   void *buf, unsigned len)
 {
 	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
-	void __iomem *ioaddr = vp_dev->ioaddr +
+	void __iomem *ioaddr = vp_dev->ldev.ioaddr +
 			VIRTIO_PCI_CONFIG_OFF(vp_dev->msix_enabled) +
 			offset;
 	u8 *ptr = buf;
@@ -64,7 +65,7 @@ static void vp_set(struct virtio_device *vdev, unsigned offset,
 		   const void *buf, unsigned len)
 {
 	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
-	void __iomem *ioaddr = vp_dev->ioaddr +
+	void __iomem *ioaddr = vp_dev->ldev.ioaddr +
 			VIRTIO_PCI_CONFIG_OFF(vp_dev->msix_enabled) +
 			offset;
 	const u8 *ptr = buf;
@@ -78,7 +79,7 @@ static void vp_set(struct virtio_device *vdev, unsigned offset,
 static u8 vp_get_status(struct virtio_device *vdev)
 {
 	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
-	return ioread8(vp_dev->ioaddr + VIRTIO_PCI_STATUS);
+	return vp_legacy_get_status(&vp_dev->ldev);
 }
 
 static void vp_set_status(struct virtio_device *vdev, u8 status)
@@ -86,28 +87,24 @@ static void vp_set_status(struct virtio_device *vdev, u8 status)
 	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
 	/* We should never be setting status to 0. */
 	BUG_ON(status == 0);
-	iowrite8(status, vp_dev->ioaddr + VIRTIO_PCI_STATUS);
+	vp_legacy_set_status(&vp_dev->ldev, status);
 }
 
 static void vp_reset(struct virtio_device *vdev)
 {
 	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
 	/* 0 status means a reset. */
-	iowrite8(0, vp_dev->ioaddr + VIRTIO_PCI_STATUS);
+	vp_legacy_set_status(&vp_dev->ldev, 0);
 	/* Flush out the status write, and flush in device writes,
 	 * including MSi-X interrupts, if any. */
-	ioread8(vp_dev->ioaddr + VIRTIO_PCI_STATUS);
+	vp_legacy_get_status(&vp_dev->ldev);
 	/* Flush pending VQ/configuration callbacks. */
 	vp_synchronize_vectors(vdev);
 }
 
 static u16 vp_config_vector(struct virtio_pci_device *vp_dev, u16 vector)
 {
-	/* Setup the vector used for configuration events */
-	iowrite16(vector, vp_dev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR);
-	/* Verify we had enough resources to assign the vector */
-	/* Will also flush the write out to device */
-	return ioread16(vp_dev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR);
+	return vp_legacy_config_vector(&vp_dev->ldev, vector);
 }
 
 static struct virtqueue *setup_vq(struct virtio_pci_device *vp_dev,
@@ -123,12 +120,9 @@ static struct virtqueue *setup_vq(struct virtio_pci_device *vp_dev,
 	int err;
 	u64 q_pfn;
 
-	/* Select the queue we're interested in */
-	iowrite16(index, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_SEL);
-
 	/* Check if queue is either not available or already active. */
-	num = ioread16(vp_dev->ioaddr + VIRTIO_PCI_QUEUE_NUM);
-	if (!num || ioread32(vp_dev->ioaddr + VIRTIO_PCI_QUEUE_PFN))
+	num = vp_legacy_get_queue_size(&vp_dev->ldev, index);
+	if (!num || vp_legacy_get_queue_enable(&vp_dev->ldev, index))
 		return ERR_PTR(-ENOENT);
 
 	info->msix_vector = msix_vec;
@@ -151,13 +145,12 @@ static struct virtqueue *setup_vq(struct virtio_pci_device *vp_dev,
 	}
 
 	/* activate the queue */
-	iowrite32(q_pfn, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_PFN);
+	vp_legacy_set_queue_address(&vp_dev->ldev, index, q_pfn);
 
-	vq->priv = (void __force *)vp_dev->ioaddr + VIRTIO_PCI_QUEUE_NOTIFY;
+	vq->priv = (void __force *)vp_dev->ldev.ioaddr + VIRTIO_PCI_QUEUE_NOTIFY;
 
 	if (msix_vec != VIRTIO_MSI_NO_VECTOR) {
-		iowrite16(msix_vec, vp_dev->ioaddr + VIRTIO_MSI_QUEUE_VECTOR);
-		msix_vec = ioread16(vp_dev->ioaddr + VIRTIO_MSI_QUEUE_VECTOR);
+		msix_vec = vp_legacy_queue_vector(&vp_dev->ldev, index, msix_vec);
 		if (msix_vec == VIRTIO_MSI_NO_VECTOR) {
 			err = -EBUSY;
 			goto out_deactivate;
@@ -167,7 +160,7 @@ static struct virtqueue *setup_vq(struct virtio_pci_device *vp_dev,
 	return vq;
 
 out_deactivate:
-	iowrite32(0, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_PFN);
+	vp_legacy_set_queue_address(&vp_dev->ldev, index, 0);
 out_del_vq:
 	vring_del_virtqueue(vq);
 	return ERR_PTR(err);
@@ -178,17 +171,15 @@ static void del_vq(struct virtio_pci_vq_info *info)
 	struct virtqueue *vq = info->vq;
 	struct virtio_pci_device *vp_dev = to_vp_device(vq->vdev);
 
-	iowrite16(vq->index, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_SEL);
-
 	if (vp_dev->msix_enabled) {
-		iowrite16(VIRTIO_MSI_NO_VECTOR,
-			  vp_dev->ioaddr + VIRTIO_MSI_QUEUE_VECTOR);
+		vp_legacy_queue_vector(&vp_dev->ldev, vq->index,
+				VIRTIO_MSI_NO_VECTOR);
 		/* Flush the write out to device */
-		ioread8(vp_dev->ioaddr + VIRTIO_PCI_ISR);
+		ioread8(vp_dev->ldev.ioaddr + VIRTIO_PCI_ISR);
 	}
 
 	/* Select and deactivate the queue */
-	iowrite32(0, vp_dev->ioaddr + VIRTIO_PCI_QUEUE_PFN);
+	vp_legacy_set_queue_address(&vp_dev->ldev, vq->index, 0);
 
 	vring_del_virtqueue(vq);
 }
@@ -211,51 +202,18 @@ static const struct virtio_config_ops virtio_pci_config_ops = {
 /* the PCI probing function */
 int virtio_pci_legacy_probe(struct virtio_pci_device *vp_dev)
 {
+	struct virtio_pci_legacy_device *ldev = &vp_dev->ldev;
 	struct pci_dev *pci_dev = vp_dev->pci_dev;
 	int rc;
 
-	/* We only own devices >= 0x1000 and <= 0x103f: leave the rest. */
-	if (pci_dev->device < 0x1000 || pci_dev->device > 0x103f)
-		return -ENODEV;
+	ldev->pci_dev = pci_dev;
 
-	if (pci_dev->revision != VIRTIO_PCI_ABI_VERSION) {
-		printk(KERN_ERR "virtio_pci: expected ABI version %d, got %d\n",
-		       VIRTIO_PCI_ABI_VERSION, pci_dev->revision);
-		return -ENODEV;
-	}
-
-	rc = dma_set_mask(&pci_dev->dev, DMA_BIT_MASK(64));
-	if (rc) {
-		rc = dma_set_mask_and_coherent(&pci_dev->dev, DMA_BIT_MASK(32));
-	} else {
-		/*
-		 * The virtio ring base address is expressed as a 32-bit PFN,
-		 * with a page size of 1 << VIRTIO_PCI_QUEUE_ADDR_SHIFT.
-		 */
-		dma_set_coherent_mask(&pci_dev->dev,
-				DMA_BIT_MASK(32 + VIRTIO_PCI_QUEUE_ADDR_SHIFT));
-	}
-
-	if (rc)
-		dev_warn(&pci_dev->dev, "Failed to enable 64-bit or 32-bit DMA.  Trying to continue, but this might not work.\n");
-
-	rc = pci_request_region(pci_dev, 0, "virtio-pci-legacy");
+	rc = vp_legacy_probe(ldev);
 	if (rc)
 		return rc;
 
-	rc = -ENOMEM;
-	vp_dev->ioaddr = pci_iomap(pci_dev, 0, 0);
-	if (!vp_dev->ioaddr)
-		goto err_iomap;
-
-	vp_dev->isr = vp_dev->ioaddr + VIRTIO_PCI_ISR;
-
-	/* we use the subsystem vendor/device id as the virtio vendor/device
-	 * id.  this allows us to use the same PCI vendor/device id for all
-	 * virtio devices and to identify the particular virtio driver by
-	 * the subsystem ids */
-	vp_dev->vdev.id.vendor = pci_dev->subsystem_vendor;
-	vp_dev->vdev.id.device = pci_dev->subsystem_device;
+	vp_dev->isr = ldev->isr;
+	vp_dev->vdev.id = ldev->id;
 
 	vp_dev->vdev.config = &virtio_pci_config_ops;
 
@@ -264,16 +222,11 @@ int virtio_pci_legacy_probe(struct virtio_pci_device *vp_dev)
 	vp_dev->del_vq = del_vq;
 
 	return 0;
-
-err_iomap:
-	pci_release_region(pci_dev, 0);
-	return rc;
 }
 
 void virtio_pci_legacy_remove(struct virtio_pci_device *vp_dev)
 {
-	struct pci_dev *pci_dev = vp_dev->pci_dev;
+	struct virtio_pci_legacy_device *ldev = &vp_dev->ldev;
 
-	pci_iounmap(pci_dev, vp_dev->ioaddr);
-	pci_release_region(pci_dev, 0);
+	vp_legacy_remove(ldev);
 }

drivers/virtio/virtio_pci_legacy_dev.c

@@ -0,0 +1,220 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include "linux/virtio_pci.h"
+#include <linux/virtio_pci_legacy.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+
+/*
+ * vp_legacy_probe: probe the legacy virtio pci device, note that the
+ * caller is required to enable PCI device before calling this function.
+ * @ldev: the legacy virtio-pci device
+ *
+ * Return 0 on succeed otherwise fail
+ */
+int vp_legacy_probe(struct virtio_pci_legacy_device *ldev)
+{
+	struct pci_dev *pci_dev = ldev->pci_dev;
+	int rc;
+
+	/* We only own devices >= 0x1000 and <= 0x103f: leave the rest. */
+	if (pci_dev->device < 0x1000 || pci_dev->device > 0x103f)
+		return -ENODEV;
+
+	if (pci_dev->revision != VIRTIO_PCI_ABI_VERSION)
+		return -ENODEV;
+
+	rc = dma_set_mask(&pci_dev->dev, DMA_BIT_MASK(64));
+	if (rc) {
+		rc = dma_set_mask_and_coherent(&pci_dev->dev, DMA_BIT_MASK(32));
+	} else {
+		/*
+		 * The virtio ring base address is expressed as a 32-bit PFN,
+		 * with a page size of 1 << VIRTIO_PCI_QUEUE_ADDR_SHIFT.
+		 */
+		dma_set_coherent_mask(&pci_dev->dev,
+				DMA_BIT_MASK(32 + VIRTIO_PCI_QUEUE_ADDR_SHIFT));
+	}
+
+	if (rc)
+		dev_warn(&pci_dev->dev, "Failed to enable 64-bit or 32-bit DMA.  Trying to continue, but this might not work.\n");
+
+	rc = pci_request_region(pci_dev, 0, "virtio-pci-legacy");
+	if (rc)
+		return rc;
+
+	ldev->ioaddr = pci_iomap(pci_dev, 0, 0);
+	if (!ldev->ioaddr)
+		goto err_iomap;
+
+	ldev->isr = ldev->ioaddr + VIRTIO_PCI_ISR;
+
+	ldev->id.vendor = pci_dev->subsystem_vendor;
+	ldev->id.device = pci_dev->subsystem_device;
+
+	return 0;
+err_iomap:
+	pci_release_region(pci_dev, 0);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(vp_legacy_probe);
+
+/*
+ * vp_legacy_probe: remove and cleanup the legacy virtio pci device
+ * @ldev: the legacy virtio-pci device
+ */
+void vp_legacy_remove(struct virtio_pci_legacy_device *ldev)
+{
+	struct pci_dev *pci_dev = ldev->pci_dev;
+
+	pci_iounmap(pci_dev, ldev->ioaddr);
+	pci_release_region(pci_dev, 0);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_remove);
+
+/*
+ * vp_legacy_get_features - get features from device
+ * @ldev: the legacy virtio-pci device
+ *
+ * Returns the features read from the device
+ */
+u64 vp_legacy_get_features(struct virtio_pci_legacy_device *ldev)
+{
+
+	return ioread32(ldev->ioaddr + VIRTIO_PCI_HOST_FEATURES);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_get_features);
+
+/*
+ * vp_legacy_get_driver_features - get driver features from device
+ * @ldev: the legacy virtio-pci device
+ *
+ * Returns the driver features read from the device
+ */
+u64 vp_legacy_get_driver_features(struct virtio_pci_legacy_device *ldev)
+{
+	return ioread32(ldev->ioaddr + VIRTIO_PCI_GUEST_FEATURES);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_get_driver_features);
+
+/*
+ * vp_legacy_set_features - set features to device
+ * @ldev: the legacy virtio-pci device
+ * @features: the features set to device
+ */
+void vp_legacy_set_features(struct virtio_pci_legacy_device *ldev,
+			    u32 features)
+{
+	iowrite32(features, ldev->ioaddr + VIRTIO_PCI_GUEST_FEATURES);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_set_features);
+
+/*
+ * vp_legacy_get_status - get the device status
+ * @ldev: the legacy virtio-pci device
+ *
+ * Returns the status read from device
+ */
+u8 vp_legacy_get_status(struct virtio_pci_legacy_device *ldev)
+{
+	return ioread8(ldev->ioaddr + VIRTIO_PCI_STATUS);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_get_status);
+
+/*
+ * vp_legacy_set_status - set status to device
+ * @ldev: the legacy virtio-pci device
+ * @status: the status set to device
+ */
+void vp_legacy_set_status(struct virtio_pci_legacy_device *ldev,
+				 u8 status)
+{
+	iowrite8(status, ldev->ioaddr + VIRTIO_PCI_STATUS);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_set_status);
+
+/*
+ * vp_legacy_queue_vector - set the MSIX vector for a specific virtqueue
+ * @ldev: the legacy virtio-pci device
+ * @index: queue index
+ * @vector: the config vector
+ *
+ * Returns the config vector read from the device
+ */
+u16 vp_legacy_queue_vector(struct virtio_pci_legacy_device *ldev,
+			   u16 index, u16 vector)
+{
+	iowrite16(index, ldev->ioaddr + VIRTIO_PCI_QUEUE_SEL);
+	iowrite16(vector, ldev->ioaddr + VIRTIO_MSI_QUEUE_VECTOR);
+	/* Flush the write out to device */
+	return ioread16(ldev->ioaddr + VIRTIO_MSI_QUEUE_VECTOR);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_queue_vector);
+
+/*
+ * vp_legacy_config_vector - set the vector for config interrupt
+ * @ldev: the legacy virtio-pci device
+ * @vector: the config vector
+ *
+ * Returns the config vector read from the device
+ */
+u16 vp_legacy_config_vector(struct virtio_pci_legacy_device *ldev,
+			    u16 vector)
+{
+	/* Setup the vector used for configuration events */
+	iowrite16(vector, ldev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR);
+	/* Verify we had enough resources to assign the vector */
+	/* Will also flush the write out to device */
+	return ioread16(ldev->ioaddr + VIRTIO_MSI_CONFIG_VECTOR);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_config_vector);
+
+/*
+ * vp_legacy_set_queue_address - set the virtqueue address
+ * @ldev: the legacy virtio-pci device
+ * @index: the queue index
+ * @queue_pfn: pfn of the virtqueue
+ */
+void vp_legacy_set_queue_address(struct virtio_pci_legacy_device *ldev,
+			     u16 index, u32 queue_pfn)
+{
+	iowrite16(index, ldev->ioaddr + VIRTIO_PCI_QUEUE_SEL);
+	iowrite32(queue_pfn, ldev->ioaddr + VIRTIO_PCI_QUEUE_PFN);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_set_queue_address);
+
+/*
+ * vp_legacy_get_queue_enable - enable a virtqueue
+ * @ldev: the legacy virtio-pci device
+ * @index: the queue index
+ *
+ * Returns whether a virtqueue is enabled or not
+ */
+bool vp_legacy_get_queue_enable(struct virtio_pci_legacy_device *ldev,
+				u16 index)
+{
+	iowrite16(index, ldev->ioaddr + VIRTIO_PCI_QUEUE_SEL);
+	return ioread32(ldev->ioaddr + VIRTIO_PCI_QUEUE_PFN);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_get_queue_enable);
+
+/*
+ * vp_legacy_get_queue_size - get size for a virtqueue
+ * @ldev: the legacy virtio-pci device
+ * @index: the queue index
+ *
+ * Returns the size of the virtqueue
+ */
+u16 vp_legacy_get_queue_size(struct virtio_pci_legacy_device *ldev,
+			     u16 index)
+{
+	iowrite16(index, ldev->ioaddr + VIRTIO_PCI_QUEUE_SEL);
+	return ioread16(ldev->ioaddr + VIRTIO_PCI_QUEUE_NUM);
+}
+EXPORT_SYMBOL_GPL(vp_legacy_get_queue_size);
+
+MODULE_VERSION("0.1");
+MODULE_DESCRIPTION("Legacy Virtio PCI Device");
+MODULE_AUTHOR("Wu Zongyong <wuzongyong@linux.alibaba.com>");
+MODULE_LICENSE("GPL");

drivers/virtio/virtio_ring.c

@@ -1050,12 +1050,12 @@ static struct vring_packed_desc *alloc_indirect_packed(unsigned int total_sg,
 }
 
 static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
-				       struct scatterlist *sgs[],
-				       unsigned int total_sg,
-				       unsigned int out_sgs,
-				       unsigned int in_sgs,
-				       void *data,
-				       gfp_t gfp)
+					 struct scatterlist *sgs[],
+					 unsigned int total_sg,
+					 unsigned int out_sgs,
+					 unsigned int in_sgs,
+					 void *data,
+					 gfp_t gfp)
 {
 	struct vring_packed_desc *desc;
 	struct scatterlist *sg;
@@ -1065,6 +1065,8 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
 
 	head = vq->packed.next_avail_idx;
 	desc = alloc_indirect_packed(total_sg, gfp);
+	if (!desc)
+		return -ENOMEM;
 
 	if (unlikely(vq->vq.num_free < 1)) {
 		pr_debug("Can't add buf len 1 - avail = 0\n");
@@ -1176,6 +1178,7 @@ static inline int virtqueue_add_packed(struct virtqueue *_vq,
 	unsigned int i, n, c, descs_used, err_idx;
 	__le16 head_flags, flags;
 	u16 head, id, prev, curr, avail_used_flags;
+	int err;
 
 	START_USE(vq);
 
@@ -1191,9 +1194,14 @@ static inline int virtqueue_add_packed(struct virtqueue *_vq,
 
 	BUG_ON(total_sg == 0);
 
-	if (virtqueue_use_indirect(_vq, total_sg))
-		return virtqueue_add_indirect_packed(vq, sgs, total_sg,
-				out_sgs, in_sgs, data, gfp);
+	if (virtqueue_use_indirect(_vq, total_sg)) {
+		err = virtqueue_add_indirect_packed(vq, sgs, total_sg, out_sgs,
+						    in_sgs, data, gfp);
+		if (err != -ENOMEM)
+			return err;
+
+		/* fall back on direct */
+	}
 
 	head = vq->packed.next_avail_idx;
 	avail_used_flags = vq->packed.avail_used_flags;

drivers/virtio/virtio_vdpa.c

@@ -145,7 +145,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
 	/* Assume split virtqueue, switch to packed if necessary */
 	struct vdpa_vq_state state = {0};
 	unsigned long flags;
-	u32 align, num;
+	u32 align, max_num, min_num = 1;
+	bool may_reduce_num = true;
 	int err;
 
 	if (!name)
@@ -163,16 +164,21 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
 	if (!info)
 		return ERR_PTR(-ENOMEM);
 
-	num = ops->get_vq_num_max(vdpa);
-	if (num == 0) {
+	max_num = ops->get_vq_num_max(vdpa);
+	if (max_num == 0) {
 		err = -ENOENT;
 		goto error_new_virtqueue;
 	}
 
+	if (ops->get_vq_num_min)
+		min_num = ops->get_vq_num_min(vdpa);
+
+	may_reduce_num = (max_num == min_num) ? false : true;
+
 	/* Create the vring */
 	align = ops->get_vq_align(vdpa);
-	vq = vring_create_virtqueue(index, num, align, vdev,
-				    true, true, ctx,
+	vq = vring_create_virtqueue(index, max_num, align, vdev,
+				    true, may_reduce_num, ctx,
 				    virtio_vdpa_notify, callback, name);
 	if (!vq) {
 		err = -ENOMEM;

include/linux/vdpa.h

@@ -171,6 +171,9 @@ struct vdpa_map_file {
  * @get_vq_num_max:		Get the max size of virtqueue
  *				@vdev: vdpa device
  *				Returns u16: max size of virtqueue
+ * @get_vq_num_min:		Get the min size of virtqueue (optional)
+ *				@vdev: vdpa device
+ *				Returns u16: min size of virtqueue
  * @get_device_id:		Get virtio device id
  *				@vdev: vdpa device
  *				Returns u32: virtio device id
@@ -257,7 +260,7 @@ struct vdpa_config_ops {
 	struct vdpa_notification_area
 	(*get_vq_notification)(struct vdpa_device *vdev, u16 idx);
 	/* vq irq is not expected to be changed once DRIVER_OK is set */
-	int (*get_vq_irq)(struct vdpa_device *vdv, u16 idx);
+	int (*get_vq_irq)(struct vdpa_device *vdev, u16 idx);
 
 	/* Device ops */
 	u32 (*get_vq_align)(struct vdpa_device *vdev);
@@ -266,6 +269,7 @@ struct vdpa_config_ops {
 	void (*set_config_cb)(struct vdpa_device *vdev,
 			      struct vdpa_callback *cb);
 	u16 (*get_vq_num_max)(struct vdpa_device *vdev);
+	u16 (*get_vq_num_min)(struct vdpa_device *vdev);
 	u32 (*get_device_id)(struct vdpa_device *vdev);
 	u32 (*get_vendor_id)(struct vdpa_device *vdev);
 	u8 (*get_status)(struct vdpa_device *vdev);

include/linux/virtio_pci_legacy.h

@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_VIRTIO_PCI_LEGACY_H
+#define _LINUX_VIRTIO_PCI_LEGACY_H
+
+#include "linux/mod_devicetable.h"
+#include <linux/pci.h>
+#include <linux/virtio_pci.h>
+
+struct virtio_pci_legacy_device {
+	struct pci_dev *pci_dev;
+
+	/* Where to read and clear interrupt */
+	u8 __iomem *isr;
+	/* The IO mapping for the PCI config space (legacy mode only) */
+	void __iomem *ioaddr;
+
+	struct virtio_device_id id;
+};
+
+u64 vp_legacy_get_features(struct virtio_pci_legacy_device *ldev);
+u64 vp_legacy_get_driver_features(struct virtio_pci_legacy_device *ldev);
+void vp_legacy_set_features(struct virtio_pci_legacy_device *ldev,
+			u32 features);
+u8 vp_legacy_get_status(struct virtio_pci_legacy_device *ldev);
+void vp_legacy_set_status(struct virtio_pci_legacy_device *ldev,
+			u8 status);
+u16 vp_legacy_queue_vector(struct virtio_pci_legacy_device *ldev,
+			   u16 idx, u16 vector);
+u16 vp_legacy_config_vector(struct virtio_pci_legacy_device *ldev,
+		     u16 vector);
+void vp_legacy_set_queue_address(struct virtio_pci_legacy_device *ldev,
+			     u16 index, u32 queue_pfn);
+bool vp_legacy_get_queue_enable(struct virtio_pci_legacy_device *ldev,
+				u16 idx);
+void vp_legacy_set_queue_size(struct virtio_pci_legacy_device *ldev,
+			      u16 idx, u16 size);
+u16 vp_legacy_get_queue_size(struct virtio_pci_legacy_device *ldev,
+			     u16 idx);
+int vp_legacy_probe(struct virtio_pci_legacy_device *ldev);
+void vp_legacy_remove(struct virtio_pci_legacy_device *ldev);
+
+#endif

include/uapi/linux/vdpa.h

@@ -32,6 +32,7 @@ enum vdpa_attr {
 	VDPA_ATTR_DEV_VENDOR_ID,		/* u32 */
 	VDPA_ATTR_DEV_MAX_VQS,			/* u32 */
 	VDPA_ATTR_DEV_MAX_VQ_SIZE,		/* u16 */
+	VDPA_ATTR_DEV_MIN_VQ_SIZE,		/* u16 */
 
 	/* new attributes must be added above here */
 	VDPA_ATTR_MAX,