[PATCH v3 2/2] nvme-tcp: send H2CData PDUs based on MAXH2CDATA
Sagi Grimberg
sagi at grimberg.me
Tue Nov 23 04:58:05 PST 2021
On 11/23/21 12:59 PM, Varun Prakash wrote:
> As per NVMe/TCP specification (revision 1.0a, section 3.6.2.3)
> Maximum Host to Controller Data length (MAXH2CDATA): Specifies the
> maximum number of PDU-Data bytes per H2CData PDU in bytes. This value
> is a multiple of dwords and should be no less than 4,096.
>
> Current code sets H2CData PDU data_length to r2t_length,
> it does not check MAXH2CDATA value. Fix this by setting H2CData PDU
> data_length to min(req->h2cdata_left, queue->maxh2cdata).
>
> Also validate MAXH2CDATA value returned by target in ICResp PDU,
> if it is not a multiple of dword or if it is less than 4096 return
> -EINVAL from nvme_tcp_init_connection().
>
> Signed-off-by: Varun Prakash <varun at chelsio.com>
> ---
>
> v3:
> - added h2cdata_left, h2cdata_offset
> - removed unnecessary local variables from nvme_tcp_try_send_data_pdu()
>
> v2:
> - removed nvme_tcp_update_h2c_data_pdu()
> - used sock_no_sendpage() instead of kernel_sendmsg()
>
> drivers/nvme/host/tcp.c | 63 +++++++++++++++++++++++++++++++++++++-----------
> include/linux/nvme-tcp.h | 1 +
> 2 files changed, 50 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
> index 5f8ad4d..0065892 100644
> --- a/drivers/nvme/host/tcp.c
> +++ b/drivers/nvme/host/tcp.c
> @@ -44,6 +44,8 @@ struct nvme_tcp_request {
> u32 data_len;
> u32 pdu_len;
> u32 pdu_sent;
> + u32 h2cdata_left;
> + u32 h2cdata_offset;
> u16 ttag;
> __le16 status;
> struct list_head entry;
> @@ -95,6 +97,7 @@ struct nvme_tcp_queue {
> struct nvme_tcp_request *request;
>
> int queue_size;
> + u32 maxh2cdata;
> size_t cmnd_capsule_len;
> struct nvme_tcp_ctrl *ctrl;
> unsigned long flags;
> @@ -572,23 +575,26 @@ static int nvme_tcp_handle_comp(struct nvme_tcp_queue *queue,
> return ret;
> }
>
> -static void nvme_tcp_setup_h2c_data_pdu(struct nvme_tcp_request *req,
> - struct nvme_tcp_r2t_pdu *pdu)
> +static void nvme_tcp_setup_h2c_data_pdu(struct nvme_tcp_request *req)
> {
> struct nvme_tcp_data_pdu *data = req->pdu;
> struct nvme_tcp_queue *queue = req->queue;
> struct request *rq = blk_mq_rq_from_pdu(req);
> + u32 h2cdata_sent = req->pdu_len;
> u8 hdgst = nvme_tcp_hdgst_len(queue);
> u8 ddgst = nvme_tcp_ddgst_len(queue);
>
> req->state = NVME_TCP_SEND_H2C_PDU;
> req->offset = 0;
> - req->pdu_len = le32_to_cpu(pdu->r2t_length);
> + req->pdu_len = min(req->h2cdata_left, queue->maxh2cdata);
> req->pdu_sent = 0;
> + req->h2cdata_left -= req->pdu_len;
> + req->h2cdata_offset += h2cdata_sent;
>
> memset(data, 0, sizeof(*data));
> data->hdr.type = nvme_tcp_h2c_data;
> - data->hdr.flags = NVME_TCP_F_DATA_LAST;
> + if (!req->h2cdata_left)
> + data->hdr.flags = NVME_TCP_F_DATA_LAST;
> if (queue->hdr_digest)
> data->hdr.flags |= NVME_TCP_F_HDGST;
> if (queue->data_digest)
> @@ -597,9 +603,9 @@ static void nvme_tcp_setup_h2c_data_pdu(struct nvme_tcp_request *req,
> data->hdr.pdo = data->hdr.hlen + hdgst;
> data->hdr.plen =
> cpu_to_le32(data->hdr.hlen + hdgst + req->pdu_len + ddgst);
> - data->ttag = pdu->ttag;
> + data->ttag = req->ttag;
> data->command_id = nvme_cid(rq);
> - data->data_offset = pdu->r2t_offset;
> + data->data_offset = cpu_to_le32(req->h2cdata_offset);
> data->data_length = cpu_to_le32(req->pdu_len);
> }
>
> @@ -609,6 +615,7 @@ static int nvme_tcp_handle_r2t(struct nvme_tcp_queue *queue,
> struct nvme_tcp_request *req;
> struct request *rq;
> u32 r2t_length = le32_to_cpu(pdu->r2t_length);
> + u32 r2t_offset = le32_to_cpu(pdu->r2t_offset);
>
> rq = nvme_find_rq(nvme_tcp_tagset(queue), pdu->command_id);
> if (!rq) {
> @@ -633,14 +640,19 @@ static int nvme_tcp_handle_r2t(struct nvme_tcp_queue *queue,
> return -EPROTO;
> }
>
> - if (unlikely(le32_to_cpu(pdu->r2t_offset) < req->data_sent)) {
> + if (unlikely(r2t_offset < req->data_sent)) {
> dev_err(queue->ctrl->ctrl.device,
> "req %d unexpected r2t offset %u (expected %zu)\n",
> - rq->tag, le32_to_cpu(pdu->r2t_offset), req->data_sent);
> + rq->tag, r2t_offset, req->data_sent);
> return -EPROTO;
> }
>
> - nvme_tcp_setup_h2c_data_pdu(req, pdu);
> + req->pdu_len = 0;
Why do you need to set this here? it is set in the initial command
execution...
> + req->h2cdata_left = r2t_length;
> + req->h2cdata_offset = r2t_offset;
> + req->ttag = pdu->ttag;
> +
> + nvme_tcp_setup_h2c_data_pdu(req);
> nvme_tcp_queue_request(req, false, true);
>
> return 0;
> @@ -920,6 +932,7 @@ static int nvme_tcp_try_send_data(struct nvme_tcp_request *req)
> {
> struct nvme_tcp_queue *queue = req->queue;
> int req_data_len = req->data_len;
> + u32 h2cdata_left = req->h2cdata_left;
>
> while (true) {
> struct page *page = nvme_tcp_req_cur_page(req);
> @@ -964,7 +977,10 @@ static int nvme_tcp_try_send_data(struct nvme_tcp_request *req)
> req->state = NVME_TCP_SEND_DDGST;
> req->offset = 0;
> } else {
> - nvme_tcp_done_send_req(queue);
> + if (h2cdata_left)
> + nvme_tcp_setup_h2c_data_pdu(req);
> + else
> + nvme_tcp_done_send_req(queue);
> }
> return 1;
> }
> @@ -1022,9 +1038,14 @@ static int nvme_tcp_try_send_data_pdu(struct nvme_tcp_request *req)
> if (queue->hdr_digest && !req->offset)
> nvme_tcp_hdgst(queue->snd_hash, pdu, sizeof(*pdu));
>
> - ret = kernel_sendpage(queue->sock, virt_to_page(pdu),
> - offset_in_page(pdu) + req->offset, len,
> - MSG_DONTWAIT | MSG_MORE | MSG_SENDPAGE_NOTLAST);
> + if (!req->h2cdata_left)
> + ret = kernel_sendpage(queue->sock, virt_to_page(pdu),
> + offset_in_page(pdu) + req->offset, len,
> + MSG_DONTWAIT | MSG_MORE | MSG_SENDPAGE_NOTLAST);
> + else
> + ret = sock_no_sendpage(queue->sock, virt_to_page(pdu),
> + offset_in_page(pdu) + req->offset, len,
> + MSG_DONTWAIT | MSG_MORE | MSG_SENDPAGE_NOTLAST);
> if (unlikely(ret <= 0))
> return ret;
>
> @@ -1044,6 +1065,7 @@ static int nvme_tcp_try_send_ddgst(struct nvme_tcp_request *req)
> {
> struct nvme_tcp_queue *queue = req->queue;
> size_t offset = req->offset;
> + u32 h2cdata_left = req->h2cdata_left;
> int ret;
> struct msghdr msg = { .msg_flags = MSG_DONTWAIT };
> struct kvec iov = {
> @@ -1061,7 +1083,10 @@ static int nvme_tcp_try_send_ddgst(struct nvme_tcp_request *req)
> return ret;
>
> if (offset + ret == NVME_TCP_DIGEST_LENGTH) {
> - nvme_tcp_done_send_req(queue);
> + if (h2cdata_left)
> + nvme_tcp_setup_h2c_data_pdu(req);
> + else
> + nvme_tcp_done_send_req(queue);
> return 1;
> }
>
> @@ -1247,6 +1272,7 @@ static int nvme_tcp_init_connection(struct nvme_tcp_queue *queue)
> struct msghdr msg = {};
> struct kvec iov;
> bool ctrl_hdgst, ctrl_ddgst;
> + u32 maxh2cdata;
> int ret;
>
> icreq = kzalloc(sizeof(*icreq), GFP_KERNEL);
> @@ -1330,6 +1356,14 @@ static int nvme_tcp_init_connection(struct nvme_tcp_queue *queue)
> goto free_icresp;
> }
>
> + maxh2cdata = le32_to_cpu(icresp->maxdata);
> + if ((maxh2cdata % 4) || (maxh2cdata < NVME_TCP_MIN_MAXH2CDATA)) {
> + pr_err("queue %d: invalid maxh2cdata returned %u\n",
> + nvme_tcp_queue_id(queue), maxh2cdata);
> + goto free_icresp;
> + }
> + queue->maxh2cdata = maxh2cdata;
> +
> ret = 0;
> free_icresp:
> kfree(icresp);
> @@ -2314,6 +2348,7 @@ static blk_status_t nvme_tcp_setup_cmd_pdu(struct nvme_ns *ns,
> req->data_sent = 0;
> req->pdu_len = 0;
> req->pdu_sent = 0;
> + req->h2cdata_left = 0;
> req->data_len = blk_rq_nr_phys_segments(rq) ?
> blk_rq_payload_bytes(rq) : 0;
> req->curr_bio = rq->bio;
> diff --git a/include/linux/nvme-tcp.h b/include/linux/nvme-tcp.h
> index 959e0bd..7547015 100644
> --- a/include/linux/nvme-tcp.h
> +++ b/include/linux/nvme-tcp.h
> @@ -12,6 +12,7 @@
> #define NVME_TCP_DISC_PORT 8009
> #define NVME_TCP_ADMIN_CCSZ SZ_8K
> #define NVME_TCP_DIGEST_LENGTH 4
> +#define NVME_TCP_MIN_MAXH2CDATA 4096
>
> enum nvme_tcp_pfv {
> NVME_TCP_PFV_1_0 = 0x0,
>
More information about the Linux-nvme
mailing list