diff mbox series

[v20,06/20] nvme-tcp: Add DDP data-path

Message ID 20231122134833.20825-7-aaptel@nvidia.com (mailing list archive)
State Deferred
Headers show
Series nvme-tcp receive offloads | expand

Checks

Context Check Description
netdev/series_format fail Series longer than 15 patches (and no cover letter)
netdev/codegen success Generated files up to date
netdev/tree_selection success Guessed tree name to be net-next, async
netdev/fixes_present success Fixes tag not required for -next series
netdev/header_inline success No static functions without inline keyword in header files
netdev/build_32bit success Errors and warnings before: 1127 this patch: 1127
netdev/cc_maintainers success CCed 5 of 5 maintainers
netdev/build_clang success Errors and warnings before: 1154 this patch: 1154
netdev/verify_signedoff success Signed-off-by tag matches author and committer
netdev/deprecated_api success None detected
netdev/check_selftest success No net selftest shell script
netdev/verify_fixes success No Fixes tag
netdev/build_allmodconfig_warn success Errors and warnings before: 1154 this patch: 1154
netdev/checkpatch success total: 0 errors, 0 warnings, 0 checks, 166 lines checked
netdev/build_clang_rust success No Rust files in patch. Skipping build
netdev/kdoc success Errors and warnings before: 0 this patch: 0
netdev/source_inline success Was 0 now: 0

Commit Message

Aurelien Aptel Nov. 22, 2023, 1:48 p.m. UTC
From: Boris Pismenny <borisp@nvidia.com>

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: 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@nvidia.com>
Signed-off-by: Ben Ben-Ishay <benishay@nvidia.com>
Signed-off-by: Or Gerlitz <ogerlitz@nvidia.com>
Signed-off-by: Yoray Zack <yorayz@nvidia.com>
Signed-off-by: Shai Malin <smalin@nvidia.com>
Signed-off-by: Aurelien Aptel <aaptel@nvidia.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
---
 drivers/nvme/host/tcp.c | 111 ++++++++++++++++++++++++++++++++++++++--
 1 file changed, 106 insertions(+), 5 deletions(-)

Comments

Sagi Grimberg Nov. 28, 2023, 10:40 a.m. UTC | #1
> +static void nvme_tcp_complete_request(struct request *rq,
> +				      __le16 status,
> +				      union nvme_result result,
> +				      __u16 command_id)
> +{
> +#ifdef CONFIG_ULP_DDP
> +	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
> +
> +	if (req->offloaded) {
> +		req->ddp_status = status;

unless this is really a ddp_status, don't name it as such. afiact
it is the nvme status, so lets stay consistent with the naming.

btw, for making the code simpler we can promote the request
status/result capture out of CONFIG_ULP_DDP to the general logic
and then I think the code will look slightly simpler.

This will be consistent with what we do in nvme-rdma and PI.

> +		req->result = result;
> +		nvme_tcp_teardown_ddp(req->queue, rq);
> +		return;
> +	}
> +#endif
> +
> +	if (!nvme_try_complete_req(rq, status, result))
> +		nvme_complete_rq(rq);
> +}
> +
>   static int nvme_tcp_process_nvme_cqe(struct nvme_tcp_queue *queue,
>   		struct nvme_completion *cqe)
>   {
> @@ -772,10 +865,9 @@ static int nvme_tcp_process_nvme_cqe(struct nvme_tcp_queue *queue,
>   	if (req->status == cpu_to_le16(NVME_SC_SUCCESS))
>   		req->status = cqe->status;
>   
> -	if (!nvme_try_complete_req(rq, req->status, cqe->result))
> -		nvme_complete_rq(rq);
> +	nvme_tcp_complete_request(rq, req->status, cqe->result,
> +				  cqe->command_id);
>   	queue->nr_cqe++;
> -
>   	return 0;
>   }
>   
> @@ -973,10 +1065,13 @@ 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)
>   {
> +	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;
>   	union nvme_result res = {};
>   
> -	if (!nvme_try_complete_req(rq, cpu_to_le16(status << 1), res))
> -		nvme_complete_rq(rq);
> +	nvme_tcp_complete_request(rq, cpu_to_le16(status << 1), res,
> +				  pdu->command_id);
>   }
>   
>   static int nvme_tcp_recv_data(struct nvme_tcp_queue *queue, struct sk_buff *skb,
> @@ -1283,6 +1378,9 @@ static int nvme_tcp_try_send_cmd_pdu(struct nvme_tcp_request *req)
>   	else
>   		msg.msg_flags |= MSG_EOR;
>   
> +	if (test_bit(NVME_TCP_Q_OFF_DDP, &queue->flags))
> +		nvme_tcp_setup_ddp(queue, blk_mq_rq_from_pdu(req));
> +

We keep coming back to this. Why isn't setup done at setup time?
Aurelien Aptel Nov. 29, 2023, 1:55 p.m. UTC | #2
Sagi Grimberg <sagi@grimberg.me> writes:
>> +static void nvme_tcp_complete_request(struct request *rq,
>> +                                   __le16 status,
>> +                                   union nvme_result result,
>> +                                   __u16 command_id)
>> +{
>> +#ifdef CONFIG_ULP_DDP
>> +     struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
>> +
>> +     if (req->offloaded) {
>> +             req->ddp_status = status;
>
> unless this is really a ddp_status, don't name it as such. afiact
> it is the nvme status, so lets stay consistent with the naming.
>
> btw, for making the code simpler we can promote the request
> status/result capture out of CONFIG_ULP_DDP to the general logic
> and then I think the code will look slightly simpler.
>
> This will be consistent with what we do in nvme-rdma and PI.

Ok, we will rename satuts to nvme_status and move it and result out of
the ifdef.

>> @@ -1283,6 +1378,9 @@ static int nvme_tcp_try_send_cmd_pdu(struct nvme_tcp_request *req)
>>       else
>>               msg.msg_flags |= MSG_EOR;
>>
>> +     if (test_bit(NVME_TCP_Q_OFF_DDP, &queue->flags))
>> +             nvme_tcp_setup_ddp(queue, blk_mq_rq_from_pdu(req));
>> +
>
> We keep coming back to this. Why isn't setup done at setup time?

Sorry, this is a left-over from previous tests, we will move it as we
agreed last time [1].

1: https://lore.kernel.org/all/ef66595c-95cd-94c4-7f51-d3d7683a188a@grimberg.me/
diff mbox series

Patch

diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
index 7ad6a4854fce..680d909eb3fb 100644
--- a/drivers/nvme/host/tcp.c
+++ b/drivers/nvme/host/tcp.c
@@ -127,6 +127,13 @@  struct nvme_tcp_request {
 	size_t			offset;
 	size_t			data_sent;
 	enum nvme_tcp_send_state state;
+
+#ifdef CONFIG_ULP_DDP
+	bool			offloaded;
+	struct ulp_ddp_io	ddp;
+	__le16			ddp_status;
+	union nvme_result	result;
+#endif
 };
 
 enum nvme_tcp_queue_flags {
@@ -354,10 +361,72 @@  nvme_tcp_get_ddp_netdev_with_limits(struct nvme_tcp_ctrl *ctrl)
 }
 
 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 ulp_ddp_ulp_ops nvme_tcp_ddp_ulp_ops = {
 	.resync_request		= nvme_tcp_resync_request,
+	.ddp_teardown_done	= nvme_tcp_ddp_teardown_done,
 };
 
+static void nvme_tcp_teardown_ddp(struct nvme_tcp_queue *queue,
+				  struct request *rq)
+{
+	struct net_device *netdev = queue->ctrl->ddp_netdev;
+	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
+
+	ulp_ddp_teardown(netdev, queue->sock->sk, &req->ddp, rq);
+	sg_free_table_chained(&req->ddp.sg_table, SG_CHUNK_SIZE);
+}
+
+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, req->ddp_status, req->result))
+		nvme_complete_rq(rq);
+}
+
+static void nvme_tcp_setup_ddp(struct nvme_tcp_queue *queue,
+			       struct request *rq)
+{
+	struct net_device *netdev = queue->ctrl->ddp_netdev;
+	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
+	int ret;
+
+	if (rq_data_dir(rq) != READ ||
+	    queue->ctrl->ddp_threshold > blk_rq_payload_bytes(rq))
+		return;
+
+	/*
+	 * DDP offload is best-effort, errors are ignored.
+	 */
+
+	req->ddp.command_id = nvme_cid(rq);
+	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)
+		goto err;
+	req->ddp.nents = blk_rq_map_sg(rq->q, rq, req->ddp.sg_table.sgl);
+
+	ret = ulp_ddp_setup(netdev, queue->sock->sk, &req->ddp);
+	if (ret) {
+		sg_free_table_chained(&req->ddp.sg_table, SG_CHUNK_SIZE);
+		goto err;
+	}
+
+	/* if successful, sg table is freed in nvme_tcp_teardown_ddp() */
+	req->offloaded = true;
+
+	return;
+err:
+	WARN_ONCE(ret, "ddp setup failed (queue 0x%x, cid 0x%x, ret=%d)",
+		  nvme_tcp_queue_id(queue),
+		  nvme_cid(rq),
+		  ret);
+}
+
 static int nvme_tcp_offload_socket(struct nvme_tcp_queue *queue)
 {
 	struct ulp_ddp_config config = {.type = ULP_DDP_NVME};
@@ -478,6 +547,10 @@  static int nvme_tcp_offload_socket(struct nvme_tcp_queue *queue)
 static void nvme_tcp_unoffload_socket(struct nvme_tcp_queue *queue)
 {}
 
+static void nvme_tcp_setup_ddp(struct nvme_tcp_queue *queue,
+			       struct request *rq)
+{}
+
 static void nvme_tcp_resync_response(struct nvme_tcp_queue *queue,
 				     struct sk_buff *skb, unsigned int offset)
 {}
@@ -753,6 +826,26 @@  static void nvme_tcp_error_recovery(struct nvme_ctrl *ctrl)
 	queue_work(nvme_reset_wq, &to_tcp_ctrl(ctrl)->err_work);
 }
 
+static void nvme_tcp_complete_request(struct request *rq,
+				      __le16 status,
+				      union nvme_result result,
+				      __u16 command_id)
+{
+#ifdef CONFIG_ULP_DDP
+	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
+
+	if (req->offloaded) {
+		req->ddp_status = status;
+		req->result = result;
+		nvme_tcp_teardown_ddp(req->queue, rq);
+		return;
+	}
+#endif
+
+	if (!nvme_try_complete_req(rq, status, result))
+		nvme_complete_rq(rq);
+}
+
 static int nvme_tcp_process_nvme_cqe(struct nvme_tcp_queue *queue,
 		struct nvme_completion *cqe)
 {
@@ -772,10 +865,9 @@  static int nvme_tcp_process_nvme_cqe(struct nvme_tcp_queue *queue,
 	if (req->status == cpu_to_le16(NVME_SC_SUCCESS))
 		req->status = cqe->status;
 
-	if (!nvme_try_complete_req(rq, req->status, cqe->result))
-		nvme_complete_rq(rq);
+	nvme_tcp_complete_request(rq, req->status, cqe->result,
+				  cqe->command_id);
 	queue->nr_cqe++;
-
 	return 0;
 }
 
@@ -973,10 +1065,13 @@  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)
 {
+	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;
 	union nvme_result res = {};
 
-	if (!nvme_try_complete_req(rq, cpu_to_le16(status << 1), res))
-		nvme_complete_rq(rq);
+	nvme_tcp_complete_request(rq, cpu_to_le16(status << 1), res,
+				  pdu->command_id);
 }
 
 static int nvme_tcp_recv_data(struct nvme_tcp_queue *queue, struct sk_buff *skb,
@@ -1283,6 +1378,9 @@  static int nvme_tcp_try_send_cmd_pdu(struct nvme_tcp_request *req)
 	else
 		msg.msg_flags |= MSG_EOR;
 
+	if (test_bit(NVME_TCP_Q_OFF_DDP, &queue->flags))
+		nvme_tcp_setup_ddp(queue, blk_mq_rq_from_pdu(req));
+
 	if (queue->hdr_digest && !req->offset)
 		nvme_tcp_hdgst(queue->snd_hash, pdu, sizeof(*pdu));
 
@@ -2724,6 +2822,9 @@  static blk_status_t nvme_tcp_setup_cmd_pdu(struct nvme_ns *ns,
 	if (ret)
 		return ret;
 
+#ifdef CONFIG_ULP_DDP
+	req->offloaded = false;
+#endif
 	req->state = NVME_TCP_SEND_CMD_PDU;
 	req->status = cpu_to_le16(NVME_SC_SUCCESS);
 	req->offset = 0;