netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Sagi Grimberg <sagi@grimberg.me>
To: Boris Pismenny <borisp@mellanox.com>,
	dsahern@gmail.com, kuba@kernel.org, davem@davemloft.net,
	saeedm@nvidia.com, hch@lst.de, axboe@fb.com, kbusch@kernel.org,
	viro@zeniv.linux.org.uk, edumazet@google.com, smalin@marvell.com
Cc: boris.pismenny@gmail.com, linux-nvme@lists.infradead.org,
	netdev@vger.kernel.org, benishay@nvidia.com, ogerlitz@nvidia.com,
	yorayz@nvidia.com, Ben Ben-Ishay <benishay@mellanox.com>,
	Or Gerlitz <ogerlitz@mellanox.com>,
	Yoray Zack <yorayz@mellanox.com>
Subject: Re: [PATCH v3 net-next 07/21] nvme-tcp: Add DDP data-path
Date: Wed, 3 Feb 2021 00:51:45 -0800	[thread overview]
Message-ID: <9a8041aa-513f-09b9-63f4-3d12db19231c@grimberg.me> (raw)
In-Reply-To: <20210201100509.27351-8-borisp@mellanox.com>



On 2/1/21 2:04 AM, Boris Pismenny wrote:
> Introduce the NVMe-TCP DDP data-path offload.
> Using this interface, the NIC hardware will scatter TCP payload directly
> to the BIO pages according to the command_id in the PDU.
> To maintain the correctness of the network stack, the driver is expected
> to construct SKBs that point to the BIO pages.
> 
> The data-path interface contains two routines: tcp_ddp_setup/teardown.
> The setup provides the mapping from command_id to the request buffers,
> while the teardown removes this mapping.
> 
> For efficiency, we introduce an asynchronous nvme completion, which is
> split between NVMe-TCP and the NIC driver as follows:
> NVMe-TCP performs the specific completion, while NIC driver performs the
> generic mq_blk completion.
> 
> Signed-off-by: Boris Pismenny <borisp@mellanox.com>
> Signed-off-by: Ben Ben-Ishay <benishay@mellanox.com>
> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
> Signed-off-by: Yoray Zack <yorayz@mellanox.com>
> ---
>   drivers/nvme/host/tcp.c | 141 +++++++++++++++++++++++++++++++++++++---
>   1 file changed, 131 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
> index ea67caf9d326..5cb46deb56e0 100644
> --- a/drivers/nvme/host/tcp.c
> +++ b/drivers/nvme/host/tcp.c
> @@ -57,6 +57,11 @@ struct nvme_tcp_request {
>   	size_t			offset;
>   	size_t			data_sent;
>   	enum nvme_tcp_send_state state;
> +
> +	bool			offloaded;
> +	struct tcp_ddp_io	ddp;
> +	__le16			status;
> +	union nvme_result	result;
>   };
>   
>   enum nvme_tcp_queue_flags {
> @@ -232,10 +237,74 @@ static inline size_t nvme_tcp_pdu_last_send(struct nvme_tcp_request *req,
>   #ifdef CONFIG_TCP_DDP
>   
>   static bool nvme_tcp_resync_request(struct sock *sk, u32 seq, u32 flags);
> +static void nvme_tcp_ddp_teardown_done(void *ddp_ctx);
>   static const struct tcp_ddp_ulp_ops nvme_tcp_ddp_ulp_ops = {
>   	.resync_request		= nvme_tcp_resync_request,
> +	.ddp_teardown_done	= nvme_tcp_ddp_teardown_done,
>   };
>   
> +static
> +int nvme_tcp_teardown_ddp(struct nvme_tcp_queue *queue,
> +			  u16 command_id,
> +			  struct request *rq)
> +{
> +	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
> +	struct net_device *netdev = queue->ctrl->offloading_netdev;
> +	int ret;
> +
> +	if (unlikely(!netdev)) {
> +		pr_info_ratelimited("%s: netdev not found\n", __func__);
> +		return -EINVAL;
> +	}
> +
> +	ret = netdev->tcp_ddp_ops->tcp_ddp_teardown(netdev, queue->sock->sk,
> +						    &req->ddp, rq);
> +	sg_free_table_chained(&req->ddp.sg_table, SG_CHUNK_SIZE);
> +	req->offloaded = false;

Why is the offloaded = false needed here? you also clear it when
you setup.

> +	return ret;
> +}
> +
> +static void nvme_tcp_ddp_teardown_done(void *ddp_ctx)
> +{
> +	struct request *rq = ddp_ctx;
> +	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
> +
> +	if (!nvme_try_complete_req(rq, cpu_to_le16(req->status << 1), req->result))
> +		nvme_complete_rq(rq);

Why is the status shifted here? it was taken from the cqe as is..

> +}
> +
> +static
> +int nvme_tcp_setup_ddp(struct nvme_tcp_queue *queue,
> +		       u16 command_id,
> +		       struct request *rq)
> +{
> +	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
> +	struct net_device *netdev = queue->ctrl->offloading_netdev;
> +	int ret;
> +
> +	req->offloaded = false;
> +
> +	if (unlikely(!netdev)) {
> +		pr_info_ratelimited("%s: netdev not found\n", __func__);

dev_info_ratelimited please.

> +		return -EINVAL;
> +	}
> +
> +	req->ddp.command_id = command_id;
> +	req->ddp.sg_table.sgl = req->ddp.first_sgl;
> +	ret = sg_alloc_table_chained(&req->ddp.sg_table, blk_rq_nr_phys_segments(rq),
> +				     req->ddp.sg_table.sgl, SG_CHUNK_SIZE);
> +	if (ret)
> +		return -ENOMEM;
> +	req->ddp.nents = blk_rq_map_sg(rq->q, rq, req->ddp.sg_table.sgl);
> +
> +	ret = netdev->tcp_ddp_ops->tcp_ddp_setup(netdev,
> +						 queue->sock->sk,
> +						 &req->ddp);
> +	if (!ret)
> +		req->offloaded = true;
> +	return ret;
> +}
> +
>   static
>   int nvme_tcp_offload_socket(struct nvme_tcp_queue *queue)
>   {
> @@ -377,6 +446,25 @@ bool nvme_tcp_resync_request(struct sock *sk, u32 seq, u32 flags)
>   
>   #else
>   
> +static
> +int nvme_tcp_setup_ddp(struct nvme_tcp_queue *queue,
> +		       u16 command_id,
> +		       struct request *rq)
> +{
> +	return -EINVAL;
> +}
> +
> +static
> +int nvme_tcp_teardown_ddp(struct nvme_tcp_queue *queue,
> +			  u16 command_id,
> +			  struct request *rq)
> +{
> +	return -EINVAL;
> +}
> +
> +static void nvme_tcp_ddp_teardown_done(void *ddp_ctx)
> +{}
> +
>   static
>   int nvme_tcp_offload_socket(struct nvme_tcp_queue *queue)
>   {
> @@ -665,6 +753,7 @@ static void nvme_tcp_error_recovery(struct nvme_ctrl *ctrl)
>   static int nvme_tcp_process_nvme_cqe(struct nvme_tcp_queue *queue,
>   		struct nvme_completion *cqe)
>   {
> +	struct nvme_tcp_request *req;
>   	struct request *rq;
>   
>   	rq = blk_mq_tag_to_rq(nvme_tcp_tagset(queue), cqe->command_id);
> @@ -676,8 +765,15 @@ static int nvme_tcp_process_nvme_cqe(struct nvme_tcp_queue *queue,
>   		return -EINVAL;
>   	}
>   
> -	if (!nvme_try_complete_req(rq, cqe->status, cqe->result))
> -		nvme_complete_rq(rq);
> +	req = blk_mq_rq_to_pdu(rq);
> +	if (req->offloaded) {
> +		req->status = cqe->status;
> +		req->result = cqe->result;
> +		nvme_tcp_teardown_ddp(queue, cqe->command_id, rq);
> +	} else {
> +		if (!nvme_try_complete_req(rq, cqe->status, cqe->result))
> +			nvme_complete_rq(rq);
> +	}

Maybe move this to nvme_tcp_complete_request as it is called from two
code paths.

>   	queue->nr_cqe++;
>   
>   	return 0;
> @@ -871,9 +967,18 @@ static int nvme_tcp_recv_pdu(struct nvme_tcp_queue *queue, struct sk_buff *skb,
>   static inline void nvme_tcp_end_request(struct request *rq, u16 status)
>   {
>   	union nvme_result res = {};
> +	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
> +	struct nvme_tcp_queue *queue = req->queue;
> +	struct nvme_tcp_data_pdu *pdu = (void *)queue->pdu;
>   
> -	if (!nvme_try_complete_req(rq, cpu_to_le16(status << 1), res))
> -		nvme_complete_rq(rq);
> +	if (req->offloaded) {
> +		req->status = cpu_to_le16(status << 1);
> +		req->result = res;
> +		nvme_tcp_teardown_ddp(queue, pdu->command_id, rq);
> +	} else {
> +		if (!nvme_try_complete_req(rq, cpu_to_le16(status << 1), res))
> +			nvme_complete_rq(rq);
> +	}
>   }
>   
>   static int nvme_tcp_recv_data(struct nvme_tcp_queue *queue, struct sk_buff *skb,
> @@ -920,12 +1025,22 @@ static int nvme_tcp_recv_data(struct nvme_tcp_queue *queue, struct sk_buff *skb,
>   		recv_len = min_t(size_t, recv_len,
>   				iov_iter_count(&req->iter));
>   
> -		if (queue->data_digest)
> -			ret = skb_copy_and_hash_datagram_iter(skb, *offset,
> -				&req->iter, recv_len, queue->rcv_hash);
> -		else
> -			ret = skb_copy_datagram_iter(skb, *offset,
> -					&req->iter, recv_len);
> +		if (test_bit(NVME_TCP_Q_OFF_DDP, &queue->flags)) {
> +			if (queue->data_digest)
> +				ret = skb_ddp_copy_and_hash_datagram_iter(skb, *offset,
> +						&req->iter, recv_len, queue->rcv_hash);
> +			else
> +				ret = skb_ddp_copy_datagram_iter(skb, *offset,
> +						&req->iter, recv_len);
> +		} else {
> +			if (queue->data_digest)
> +				ret = skb_copy_and_hash_datagram_iter(skb, *offset,
> +						&req->iter, recv_len, queue->rcv_hash);
> +			else
> +				ret = skb_copy_datagram_iter(skb, *offset,
> +						&req->iter, recv_len);
> +		}
> +

Maybe move this hunk to nvme_tcp_consume_skb or something?

>   		if (ret) {
>   			dev_err(queue->ctrl->ctrl.device,
>   				"queue %d failed to copy request %#x data",
> @@ -1149,6 +1264,7 @@ static int nvme_tcp_try_send_cmd_pdu(struct nvme_tcp_request *req)
>   	bool inline_data = nvme_tcp_has_inline_data(req);
>   	u8 hdgst = nvme_tcp_hdgst_len(queue);
>   	int len = sizeof(*pdu) + hdgst - req->offset;
> +	struct request *rq = blk_mq_rq_from_pdu(req);
>   	int flags = MSG_DONTWAIT;
>   	int ret;
>   
> @@ -1157,6 +1273,10 @@ static int nvme_tcp_try_send_cmd_pdu(struct nvme_tcp_request *req)
>   	else
>   		flags |= MSG_EOR;
>   
> +	if (test_bit(NVME_TCP_Q_OFF_DDP, &queue->flags) &&
> +	    blk_rq_nr_phys_segments(rq) && rq_data_dir(rq) == READ)
> +		nvme_tcp_setup_ddp(queue, pdu->cmd.common.command_id, rq);
> +
>   	if (queue->hdr_digest && !req->offset)
>   		nvme_tcp_hdgst(queue->snd_hash, pdu, sizeof(*pdu));
>   
> @@ -2464,6 +2584,7 @@ static blk_status_t nvme_tcp_setup_cmd_pdu(struct nvme_ns *ns,
>   	req->data_len = blk_rq_nr_phys_segments(rq) ?
>   				blk_rq_payload_bytes(rq) : 0;
>   	req->curr_bio = rq->bio;
> +	req->offloaded = false;

offloaded is being cleared lots of times, and I'm not clear what are
the lifetime rules here.

  parent reply	other threads:[~2021-02-03  8:53 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-02-01 10:04 [PATCH v3 net-next 00/21] nvme-tcp receive offloads Boris Pismenny
2021-02-01 10:04 ` [PATCH v3 net-next 01/21] iov_iter: Introduce new procedures for copy to iter/pages Boris Pismenny
2021-02-01 17:35   ` Christoph Hellwig
2021-02-02 18:00     ` Or Gerlitz
2021-02-03 16:56       ` Christoph Hellwig
2021-02-03 19:34         ` Ira Weiny
2021-02-07 14:13           ` Boris Pismenny
2021-02-07 14:24         ` Boris Pismenny
2021-02-01 10:04 ` [PATCH v3 net-next 02/21] net: Introduce direct data placement tcp offload Boris Pismenny
2021-02-02 10:06   ` Tom Parkin
2021-02-01 10:04 ` [PATCH v3 net-next 03/21] net: Introduce crc offload for tcp ddp ulp Boris Pismenny
2021-02-01 10:04 ` [PATCH v3 net-next 04/21] net: SKB copy(+hash) iterators for DDP offloads Boris Pismenny
2021-02-01 10:04 ` [PATCH v3 net-next 05/21] net/tls: expose get_netdev_for_sock Boris Pismenny
2021-02-01 10:04 ` [PATCH v3 net-next 06/21] nvme-tcp: Add DDP offload control path Boris Pismenny
2021-02-01 17:37   ` Christoph Hellwig
2021-02-02 18:09     ` Or Gerlitz
2021-02-03  9:17   ` Sagi Grimberg
2021-02-01 10:04 ` [PATCH v3 net-next 07/21] nvme-tcp: Add DDP data-path Boris Pismenny
2021-02-01 17:37   ` Christoph Hellwig
2021-02-02 18:14     ` Or Gerlitz
2021-02-03  8:56       ` Sagi Grimberg
2021-02-03 10:02         ` Christoph Hellwig
2021-02-03 10:21           ` Sagi Grimberg
2021-02-03  8:51   ` Sagi Grimberg [this message]
2021-02-04 19:20     ` Or Gerlitz
2021-02-01 10:04 ` [PATCH v3 net-next 08/21] nvme-tcp : Recalculate crc in the end of the capsule Boris Pismenny
2021-02-03  9:06   ` Sagi Grimberg
2021-02-04 18:36     ` Or Gerlitz
2021-02-07 16:40     ` Or Gerlitz
2021-02-01 10:04 ` [PATCH v3 net-next 09/21] nvme-tcp: Deal with netdevice DOWN events Boris Pismenny
2021-02-03  9:09   ` Sagi Grimberg
2021-02-04 18:29     ` Or Gerlitz
2021-02-01 10:04 ` [PATCH v3 net-next 10/21] net/mlx5: Header file changes for nvme-tcp offload Boris Pismenny
2021-02-01 10:04 ` [PATCH v3 net-next 11/21] net/mlx5: Add 128B CQE for NVMEoTCP offload Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 12/21] net/mlx5e: TCP flow steering for nvme-tcp Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 13/21] net/mlx5e: NVMEoTCP offload initialization Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 14/21] net/mlx5e: KLM UMR helper macros Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 15/21] net/mlx5e: NVMEoTCP use KLM UMRs Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 16/21] net/mlx5e: NVMEoTCP queue init/teardown Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 17/21] net/mlx5e: NVMEoTCP async ddp invalidation Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 18/21] net/mlx5e: NVMEoTCP ddp setup and resync Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 19/21] net/mlx5e: NVMEoTCP, data-path for DDP+CRC offload Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 20/21] net/mlx5e: NVMEoTCP statistics Boris Pismenny
2021-02-01 10:05 ` [PATCH v3 net-next 21/21] Documentation: add TCP DDP offload documentation Boris Pismenny

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9a8041aa-513f-09b9-63f4-3d12db19231c@grimberg.me \
    --to=sagi@grimberg.me \
    --cc=axboe@fb.com \
    --cc=benishay@mellanox.com \
    --cc=benishay@nvidia.com \
    --cc=boris.pismenny@gmail.com \
    --cc=borisp@mellanox.com \
    --cc=davem@davemloft.net \
    --cc=dsahern@gmail.com \
    --cc=edumazet@google.com \
    --cc=hch@lst.de \
    --cc=kbusch@kernel.org \
    --cc=kuba@kernel.org \
    --cc=linux-nvme@lists.infradead.org \
    --cc=netdev@vger.kernel.org \
    --cc=ogerlitz@mellanox.com \
    --cc=ogerlitz@nvidia.com \
    --cc=saeedm@nvidia.com \
    --cc=smalin@marvell.com \
    --cc=viro@zeniv.linux.org.uk \
    --cc=yorayz@mellanox.com \
    --cc=yorayz@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).