[dpdk-dev,v5,3/8] virtio: enable use virtual address to fill desc

Message ID 1464605739-140761-4-git-send-email-jianfeng.tan@intel.com (mailing list archive)
State Superseded, archived
Delegated to: Yuanhan Liu
Headers

Commit Message

Jianfeng Tan May 30, 2016, 10:55 a.m. UTC
  This patch is related to how to calculate relative address for vhost
backend.

The principle is that: based on one or multiple shared memory regions,
vhost maintains a reference system with the frontend start address,
backend start address, and length for each segment, so that each
frontend address (GPA, Guest Physical Address) can be translated into
vhost-recognizable backend address. To make the address translation
efficient, we need to maintain as few regions as possible. In the case
of VM, GPA is always locally continuous. But for some other case, like
virtio-user, we use virtual address here.

It basically means:
  a. when set_base_addr, VA address is used;
  b. when preparing RX's descriptors, VA address is used;
  c. when transmitting packets, VA is filled in TX's descriptors;
  d. in TX and CQ's header, VA is used.

Signed-off-by: Huawei Xie <huawei.xie@intel.com>
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
---
 drivers/net/virtio/virtio_ethdev.c      | 21 ++++++++++++++++-----
 drivers/net/virtio/virtio_rxtx.c        |  5 ++---
 drivers/net/virtio/virtio_rxtx_simple.c | 13 +++++++------
 drivers/net/virtio/virtqueue.h          | 13 ++++++++++++-
 4 files changed, 37 insertions(+), 15 deletions(-)
  

Comments

Yuanhan Liu June 1, 2016, 8:03 a.m. UTC | #1
On Mon, May 30, 2016 at 10:55:34AM +0000, Jianfeng Tan wrote:
> This patch is related to how to calculate relative address for vhost
> backend.
> 
> The principle is that: based on one or multiple shared memory regions,
> vhost maintains a reference system with the frontend start address,
> backend start address, and length for each segment, so that each
> frontend address (GPA, Guest Physical Address) can be translated into
> vhost-recognizable backend address. To make the address translation
> efficient, we need to maintain as few regions as possible. In the case
> of VM, GPA is always locally continuous. But for some other case, like
> virtio-user, we use virtual address here.
> 
> It basically means:
>   a. when set_base_addr, VA address is used;
>   b. when preparing RX's descriptors, VA address is used;
>   c. when transmitting packets, VA is filled in TX's descriptors;
>   d. in TX and CQ's header, VA is used.
> 
> Signed-off-by: Huawei Xie <huawei.xie@intel.com>
> Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
> Acked-by: Neil Horman <nhorman@tuxdriver.com>
> ---
>  drivers/net/virtio/virtio_ethdev.c      | 21 ++++++++++++++++-----
>  drivers/net/virtio/virtio_rxtx.c        |  5 ++---
>  drivers/net/virtio/virtio_rxtx_simple.c | 13 +++++++------
>  drivers/net/virtio/virtqueue.h          | 13 ++++++++++++-
>  4 files changed, 37 insertions(+), 15 deletions(-)
> 
> @@ -419,8 +419,6 @@ int virtio_dev_queue_setup(struct rte_eth_dev *dev,
>  		vq->virtio_net_hdr_mem = hdr_mz->phys_addr;
>  
>  		memset(hdr_mz->addr, 0, hdr_mz_sz);
> -		vring_hdr_desc_init(vq);
> -
>  	} else if (queue_type == VTNET_CQ) {
>  		/* Allocate a page for control vq command, data and status */
>  		snprintf(vq_name, sizeof(vq_name), "port%d_cvq_hdrzone",
> @@ -441,6 +439,19 @@ int virtio_dev_queue_setup(struct rte_eth_dev *dev,
>  		memset(vq->virtio_net_hdr_mz->addr, 0, PAGE_SIZE);
>  	}
>  
> +	if (dev->pci_dev)
> +		vq->offset = offsetof(struct rte_mbuf, buf_physaddr);
> +	else {
> +		vq->vq_ring_mem = (phys_addr_t)vq->mz->addr;
> +		vq->offset = offsetof(struct rte_mbuf, buf_addr);
> +		if (vq->virtio_net_hdr_mz)
> +			vq->virtio_net_hdr_mem =
> +				(phys_addr_t)vq->virtio_net_hdr_mz->addr;
> +	}

I guess this piece of code deserves some comments. Say, for virtio-user
case (that is when dev->pci_dev is NULL), we use virtual address,
because, bala, bala ...


> @@ -165,6 +173,7 @@ struct virtqueue {
>  	void        *vq_ring_virt_mem;    /**< linear address of vring*/
>  	unsigned int vq_ring_size;
>  	phys_addr_t vq_ring_mem;          /**< physical address of vring */
> +					  /**< use virtual address for vdev. */

Replace vdev with "virtio-user" is better here?

	--yliu
  

Patch

diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c
index 781886d..1866afd 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -167,14 +167,14 @@  virtio_send_command(struct virtqueue *vq, struct virtio_pmd_ctrl *ctrl,
 	 * One RX packet for ACK.
 	 */
 	vq->vq_ring.desc[head].flags = VRING_DESC_F_NEXT;
-	vq->vq_ring.desc[head].addr = vq->virtio_net_hdr_mz->phys_addr;
+	vq->vq_ring.desc[head].addr = vq->virtio_net_hdr_mem;
 	vq->vq_ring.desc[head].len = sizeof(struct virtio_net_ctrl_hdr);
 	vq->vq_free_cnt--;
 	i = vq->vq_ring.desc[head].next;
 
 	for (k = 0; k < pkt_num; k++) {
 		vq->vq_ring.desc[i].flags = VRING_DESC_F_NEXT;
-		vq->vq_ring.desc[i].addr = vq->virtio_net_hdr_mz->phys_addr
+		vq->vq_ring.desc[i].addr = vq->virtio_net_hdr_mem
 			+ sizeof(struct virtio_net_ctrl_hdr)
 			+ sizeof(ctrl->status) + sizeof(uint8_t)*sum;
 		vq->vq_ring.desc[i].len = dlen[k];
@@ -184,7 +184,7 @@  virtio_send_command(struct virtqueue *vq, struct virtio_pmd_ctrl *ctrl,
 	}
 
 	vq->vq_ring.desc[i].flags = VRING_DESC_F_WRITE;
-	vq->vq_ring.desc[i].addr = vq->virtio_net_hdr_mz->phys_addr
+	vq->vq_ring.desc[i].addr = vq->virtio_net_hdr_mem
 			+ sizeof(struct virtio_net_ctrl_hdr);
 	vq->vq_ring.desc[i].len = sizeof(ctrl->status);
 	vq->vq_free_cnt--;
@@ -419,8 +419,6 @@  int virtio_dev_queue_setup(struct rte_eth_dev *dev,
 		vq->virtio_net_hdr_mem = hdr_mz->phys_addr;
 
 		memset(hdr_mz->addr, 0, hdr_mz_sz);
-		vring_hdr_desc_init(vq);
-
 	} else if (queue_type == VTNET_CQ) {
 		/* Allocate a page for control vq command, data and status */
 		snprintf(vq_name, sizeof(vq_name), "port%d_cvq_hdrzone",
@@ -441,6 +439,19 @@  int virtio_dev_queue_setup(struct rte_eth_dev *dev,
 		memset(vq->virtio_net_hdr_mz->addr, 0, PAGE_SIZE);
 	}
 
+	if (dev->pci_dev)
+		vq->offset = offsetof(struct rte_mbuf, buf_physaddr);
+	else {
+		vq->vq_ring_mem = (phys_addr_t)vq->mz->addr;
+		vq->offset = offsetof(struct rte_mbuf, buf_addr);
+		if (vq->virtio_net_hdr_mz)
+			vq->virtio_net_hdr_mem =
+				(phys_addr_t)vq->virtio_net_hdr_mz->addr;
+	}
+
+	if (queue_type == VTNET_TQ)
+		vring_hdr_desc_init(vq);
+
 	if (hw->vtpci_ops->setup_queue(hw, vq) < 0) {
 		PMD_INIT_LOG(ERR, "setup_queue failed");
 		virtio_dev_queue_release(vq);
diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index f326222..5b0c3df 100644
--- a/drivers/net/virtio/virtio_rxtx.c
+++ b/drivers/net/virtio/virtio_rxtx.c
@@ -193,8 +193,7 @@  virtqueue_enqueue_recv_refill(struct virtqueue *vq, struct rte_mbuf *cookie)
 
 	start_dp = vq->vq_ring.desc;
 	start_dp[idx].addr =
-		(uint64_t)(cookie->buf_physaddr + RTE_PKTMBUF_HEADROOM
-		- hw->vtnet_hdr_size);
+		MBUF_DATA_DMA_ADDR(cookie, vq->offset) - hw->vtnet_hdr_size;
 	start_dp[idx].len =
 		cookie->buf_len - RTE_PKTMBUF_HEADROOM + hw->vtnet_hdr_size;
 	start_dp[idx].flags =  VRING_DESC_F_WRITE;
@@ -265,7 +264,7 @@  virtqueue_enqueue_xmit(struct virtqueue *txvq, struct rte_mbuf *cookie,
 	}
 
 	do {
-		start_dp[idx].addr  = rte_mbuf_data_dma_addr(cookie);
+		start_dp[idx].addr  = MBUF_DATA_DMA_ADDR(cookie, txvq->offset);
 		start_dp[idx].len   = cookie->data_len;
 		start_dp[idx].flags = cookie->next ? VRING_DESC_F_NEXT : 0;
 		idx = start_dp[idx].next;
diff --git a/drivers/net/virtio/virtio_rxtx_simple.c b/drivers/net/virtio/virtio_rxtx_simple.c
index 8f5293d..83a794e 100644
--- a/drivers/net/virtio/virtio_rxtx_simple.c
+++ b/drivers/net/virtio/virtio_rxtx_simple.c
@@ -80,8 +80,8 @@  virtqueue_enqueue_recv_refill_simple(struct virtqueue *vq,
 	vq->sw_ring[desc_idx] = cookie;
 
 	start_dp = vq->vq_ring.desc;
-	start_dp[desc_idx].addr = (uint64_t)((uintptr_t)cookie->buf_physaddr +
-		RTE_PKTMBUF_HEADROOM - vq->hw->vtnet_hdr_size);
+	start_dp[desc_idx].addr = MBUF_DATA_DMA_ADDR(cookie, vq->offset) -
+				  vq->hw->vtnet_hdr_size;
 	start_dp[desc_idx].len = cookie->buf_len -
 		RTE_PKTMBUF_HEADROOM + vq->hw->vtnet_hdr_size;
 
@@ -119,8 +119,8 @@  virtio_rxq_rearm_vec(struct virtqueue *rxvq)
 		*(uint64_t *)p = rxvq->mbuf_initializer;
 
 		start_dp[i].addr =
-			(uint64_t)((uintptr_t)sw_ring[i]->buf_physaddr +
-			RTE_PKTMBUF_HEADROOM - rxvq->hw->vtnet_hdr_size);
+			MBUF_DATA_DMA_ADDR(sw_ring[i], rxvq->offset) -
+			rxvq->hw->vtnet_hdr_size;
 		start_dp[i].len = sw_ring[i]->buf_len -
 			RTE_PKTMBUF_HEADROOM + rxvq->hw->vtnet_hdr_size;
 	}
@@ -366,7 +366,7 @@  virtio_xmit_pkts_simple(void *tx_queue, struct rte_mbuf **tx_pkts,
 			txvq->vq_descx[desc_idx + i].cookie = tx_pkts[i];
 		for (i = 0; i < nb_tail; i++) {
 			start_dp[desc_idx].addr =
-				rte_mbuf_data_dma_addr(*tx_pkts);
+				MBUF_DATA_DMA_ADDR(*tx_pkts, txvq->offset);
 			start_dp[desc_idx].len = (*tx_pkts)->pkt_len;
 			tx_pkts++;
 			desc_idx++;
@@ -377,7 +377,8 @@  virtio_xmit_pkts_simple(void *tx_queue, struct rte_mbuf **tx_pkts,
 	for (i = 0; i < nb_commit; i++)
 		txvq->vq_descx[desc_idx + i].cookie = tx_pkts[i];
 	for (i = 0; i < nb_commit; i++) {
-		start_dp[desc_idx].addr = rte_mbuf_data_dma_addr(*tx_pkts);
+		start_dp[desc_idx].addr =
+			MBUF_DATA_DMA_ADDR(*tx_pkts, txvq->offset);
 		start_dp[desc_idx].len = (*tx_pkts)->pkt_len;
 		tx_pkts++;
 		desc_idx++;
diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h
index 4e543d2..8ffc366 100644
--- a/drivers/net/virtio/virtqueue.h
+++ b/drivers/net/virtio/virtqueue.h
@@ -66,6 +66,14 @@  struct rte_mbuf;
 
 #define VIRTQUEUE_MAX_NAME_SZ 32
 
+#ifdef RTE_VIRTIO_VDEV
+#define MBUF_DATA_DMA_ADDR(mb, offset) \
+	((uint64_t)((uintptr_t)(*(void **)((uintptr_t)mb + offset)) \
+			+ (mb)->data_off))
+#else /* RTE_VIRTIO_VDEV */
+#define MBUF_DATA_DMA_ADDR(mb, offset) rte_mbuf_data_dma_addr(mb)
+#endif /* RTE_VIRTIO_VDEV */
+
 #define VTNET_SQ_RQ_QUEUE_IDX 0
 #define VTNET_SQ_TQ_QUEUE_IDX 1
 #define VTNET_SQ_CQ_QUEUE_IDX 2
@@ -165,6 +173,7 @@  struct virtqueue {
 	void        *vq_ring_virt_mem;    /**< linear address of vring*/
 	unsigned int vq_ring_size;
 	phys_addr_t vq_ring_mem;          /**< physical address of vring */
+					  /**< use virtual address for vdev. */
 
 	struct vring vq_ring;    /**< vring keeping desc, used and avail */
 	uint16_t    vq_free_cnt; /**< num of desc available */
@@ -183,8 +192,10 @@  struct virtqueue {
 	 */
 	uint16_t vq_used_cons_idx;
 	uint16_t vq_avail_idx;
+	uint16_t offset; /**< relative offset to obtain addr in mbuf */
 	uint64_t mbuf_initializer; /**< value to init mbufs. */
-	phys_addr_t virtio_net_hdr_mem; /**< hdr for each xmit packet */
+	phys_addr_t virtio_net_hdr_mem; /**< phys addr of hdr memzone */
+					/**< use virtual address for vdev */
 
 	struct rte_mbuf **sw_ring; /**< RX software ring. */
 	/* dummy mbuf, for wraparound when processing RX ring. */