From: Chaoyong He <chaoyong.he@corigine.com>
To: dev@dpdk.org
Cc: oss-drivers@corigine.com, niklas.soderlund@corigine.com,
Chaoyong He <chaoyong.he@corigine.com>
Subject: [PATCH 06/10] net/nfp: add the xmit function of NFDk for ctrl VNIC
Date: Wed, 7 Jun 2023 09:57:05 +0800 [thread overview]
Message-ID: <20230607015709.336420-7-chaoyong.he@corigine.com> (raw)
In-Reply-To: <20230607015709.336420-1-chaoyong.he@corigine.com>
Add the xmit function of NFDk for ctrl VNIC.
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@corigine.com>
---
drivers/net/nfp/flower/nfp_flower_ctrl.c | 138 ++++++++++++++++++++++-
drivers/net/nfp/nfdk/nfp_nfdk.h | 2 +
drivers/net/nfp/nfdk/nfp_nfdk_dp.c | 2 +-
3 files changed, 140 insertions(+), 2 deletions(-)
diff --git a/drivers/net/nfp/flower/nfp_flower_ctrl.c b/drivers/net/nfp/flower/nfp_flower_ctrl.c
index bafcf5b017..0be44a68aa 100644
--- a/drivers/net/nfp/flower/nfp_flower_ctrl.c
+++ b/drivers/net/nfp/flower/nfp_flower_ctrl.c
@@ -12,6 +12,7 @@
#include "../nfp_ctrl.h"
#include "../nfp_rxtx.h"
#include "../nfd3/nfp_nfd3.h"
+#include "../nfdk/nfp_nfdk.h"
#include "nfp_flower.h"
#include "nfp_flower_ctrl.h"
#include "nfp_flower_cmsg.h"
@@ -224,14 +225,149 @@ nfp_flower_ctrl_vnic_nfd3_xmit(struct nfp_app_fw_flower *app_fw_flower,
return cnt;
}
+static uint16_t
+nfp_flower_ctrl_vnic_nfdk_xmit(struct nfp_app_fw_flower *app_fw_flower,
+ struct rte_mbuf *mbuf)
+{
+ int nop_descs;
+ uint32_t type;
+ uint32_t dma_len;
+ uint32_t tmp_dlen;
+ uint64_t dma_addr;
+ uint32_t dlen_type;
+ uint32_t used_descs;
+ uint32_t free_descs;
+ struct rte_mbuf **lmbuf;
+ struct nfp_net_txq *txq;
+ uint32_t issued_descs = 0;
+ struct rte_eth_dev *ctrl_dev;
+ struct nfp_net_nfdk_tx_desc *ktxds;
+
+ ctrl_dev = app_fw_flower->ctrl_hw->eth_dev;
+
+ /* Flower ctrl vNIC only has a single tx queue */
+ txq = ctrl_dev->data->tx_queues[0];
+
+ if (unlikely(mbuf->nb_segs > 1)) {
+ PMD_TX_LOG(ERR, "Multisegment packet not supported");
+ return 0;
+ }
+
+ if (nfp_net_nfdk_free_tx_desc(txq) < NFDK_TX_DESC_PER_SIMPLE_PKT ||
+ nfp_net_nfdk_txq_full(txq))
+ nfp_net_tx_free_bufs(txq);
+
+ free_descs = nfp_net_nfdk_free_tx_desc(txq);
+ if (unlikely(free_descs < NFDK_TX_DESC_PER_SIMPLE_PKT)) {
+ PMD_TX_LOG(ERR, "ctrl dev no free descs");
+ return 0;
+ }
+
+ nop_descs = nfp_net_nfdk_tx_maybe_close_block(txq, mbuf);
+ if (nop_descs < 0)
+ return 0;
+
+ issued_descs += nop_descs;
+ ktxds = &txq->ktxds[txq->wr_p];
+
+ /*
+ * Checksum and VLAN flags just in the first descriptor for a
+ * multisegment packet, but TSO info needs to be in all of them.
+ */
+ dma_len = mbuf->data_len;
+ if (dma_len <= NFDK_TX_MAX_DATA_PER_HEAD)
+ type = NFDK_DESC_TX_TYPE_SIMPLE;
+ else
+ type = NFDK_DESC_TX_TYPE_GATHER;
+
+ /* Implicitly truncates to chunk in below logic */
+ dma_len -= 1;
+
+ /*
+ * We will do our best to pass as much data as we can in descriptor
+ * and we need to make sure the first descriptor includes whole
+ * head since there is limitation in firmware side. Sometimes the
+ * value of 'dma_len & NFDK_DESC_TX_DMA_LEN_HEAD' will be less
+ * than packet head len.
+ */
+ if (dma_len > NFDK_DESC_TX_DMA_LEN_HEAD)
+ dma_len = NFDK_DESC_TX_DMA_LEN_HEAD;
+ dlen_type = dma_len | (NFDK_DESC_TX_TYPE_HEAD & (type << 12));
+ ktxds->dma_len_type = rte_cpu_to_le_16(dlen_type);
+ dma_addr = rte_mbuf_data_iova(mbuf);
+ ktxds->dma_addr_hi = rte_cpu_to_le_16(dma_addr >> 32);
+ ktxds->dma_addr_lo = rte_cpu_to_le_32(dma_addr & 0xffffffff);
+ ktxds++;
+
+ /*
+ * Preserve the original dlen_type, this way below the EOP logic
+ * can use dlen_type.
+ */
+ tmp_dlen = dlen_type & NFDK_DESC_TX_DMA_LEN_HEAD;
+ dma_len -= tmp_dlen;
+ dma_addr += tmp_dlen + 1;
+
+ /*
+ * The rest of the data (if any) will be in larger DMA descriptors
+ * and is handled with the dma_len loop.
+ */
+ lmbuf = &txq->txbufs[txq->wr_p].mbuf;
+ if (*lmbuf != NULL)
+ rte_pktmbuf_free_seg(*lmbuf);
+ *lmbuf = mbuf;
+ while (dma_len > 0) {
+ dma_len -= 1;
+ dlen_type = NFDK_DESC_TX_DMA_LEN & dma_len;
+
+ ktxds->dma_len_type = rte_cpu_to_le_16(dlen_type);
+ ktxds->dma_addr_hi = rte_cpu_to_le_16(dma_addr >> 32);
+ ktxds->dma_addr_lo = rte_cpu_to_le_32(dma_addr & 0xffffffff);
+ ktxds++;
+
+ dma_len -= dlen_type;
+ dma_addr += dlen_type + 1;
+ }
+
+ (ktxds - 1)->dma_len_type = rte_cpu_to_le_16(dlen_type | NFDK_DESC_TX_EOP);
+
+ ktxds->raw = rte_cpu_to_le_64(NFDK_DESC_TX_CHAIN_META);
+ ktxds++;
+
+ used_descs = ktxds - txq->ktxds - txq->wr_p;
+ if (RTE_ALIGN_FLOOR(txq->wr_p, NFDK_TX_DESC_BLOCK_CNT) !=
+ RTE_ALIGN_FLOOR(txq->wr_p + used_descs - 1, NFDK_TX_DESC_BLOCK_CNT)) {
+ PMD_TX_LOG(INFO, "Used descs cross block boundary");
+ return 0;
+ }
+
+ txq->wr_p = D_IDX(txq, txq->wr_p + used_descs);
+ if (txq->wr_p % NFDK_TX_DESC_BLOCK_CNT)
+ txq->data_pending += mbuf->pkt_len;
+ else
+ txq->data_pending = 0;
+
+ issued_descs += used_descs;
+
+ /* Increment write pointers. Force memory write before we let HW know */
+ rte_wmb();
+ nfp_qcp_ptr_add(txq->qcp_q, NFP_QCP_WRITE_PTR, issued_descs);
+
+ return 1;
+}
+
void
nfp_flower_ctrl_vnic_xmit_register(struct nfp_app_fw_flower *app_fw_flower)
{
+ struct nfp_net_hw *hw;
struct nfp_flower_nfd_func *nfd_func;
+ hw = app_fw_flower->pf_hw;
nfd_func = &app_fw_flower->nfd_func;
- nfd_func->ctrl_vnic_xmit_t = nfp_flower_ctrl_vnic_nfd3_xmit;
+ if (hw->ver.extend == NFP_NET_CFG_VERSION_DP_NFD3)
+ nfd_func->ctrl_vnic_xmit_t = nfp_flower_ctrl_vnic_nfd3_xmit;
+ else
+ nfd_func->ctrl_vnic_xmit_t = nfp_flower_ctrl_vnic_nfdk_xmit;
}
uint16_t
diff --git a/drivers/net/nfp/nfdk/nfp_nfdk.h b/drivers/net/nfp/nfdk/nfp_nfdk.h
index 0e3c6cc90d..898454102d 100644
--- a/drivers/net/nfp/nfdk/nfp_nfdk.h
+++ b/drivers/net/nfp/nfdk/nfp_nfdk.h
@@ -237,5 +237,7 @@ int nfp_net_nfdk_tx_queue_setup(struct rte_eth_dev *dev,
uint16_t nb_desc,
unsigned int socket_id,
const struct rte_eth_txconf *tx_conf);
+int nfp_net_nfdk_tx_maybe_close_block(struct nfp_net_txq *txq,
+ struct rte_mbuf *pkt);
#endif /* _NFP_NFDK_H_ */
diff --git a/drivers/net/nfp/nfdk/nfp_nfdk_dp.c b/drivers/net/nfp/nfdk/nfp_nfdk_dp.c
index 59c507e03b..dbb15581c4 100644
--- a/drivers/net/nfp/nfdk/nfp_nfdk_dp.c
+++ b/drivers/net/nfp/nfdk/nfp_nfdk_dp.c
@@ -63,7 +63,7 @@ nfp_net_nfdk_tx_close_block(struct nfp_net_txq *txq,
txq->wr_p = D_IDX(txq, wr_p + nop_slots);
}
-static int
+int
nfp_net_nfdk_tx_maybe_close_block(struct nfp_net_txq *txq,
struct rte_mbuf *pkt)
{
--
2.39.1
next prev parent reply other threads:[~2023-06-07 1:58 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-07 1:56 [PATCH 00/10] support rte_flow for flower firmware with NFDk Chaoyong He
2023-06-07 1:57 ` [PATCH 01/10] net/nfp: fix the problem of rtsym Chaoyong He
2023-06-07 1:57 ` [PATCH 02/10] net/nfp: add a check function for the NFD version Chaoyong He
2023-06-07 1:57 ` [PATCH 03/10] net/nfp: refactor the insert metadata logic of NFD3 Chaoyong He
2023-06-07 1:57 ` [PATCH 04/10] net/nfp: add the insert metadata logic of NFDk Chaoyong He
2023-06-07 1:57 ` [PATCH 05/10] net/nfp: add a dispatch layer for Tx function of ctrl VNIC Chaoyong He
2023-06-07 1:57 ` Chaoyong He [this message]
2023-06-07 1:57 ` [PATCH 07/10] net/nfp: add a dispatch layer for Tx function of representor Chaoyong He
2023-06-07 1:57 ` [PATCH 08/10] net/nfp: merge the xmit function of NFD3 Chaoyong He
2023-06-07 1:57 ` [PATCH 09/10] net/nfp: add the xmit function of NFDk for PF representor Chaoyong He
2023-06-07 1:57 ` [PATCH 10/10] net/nfp: merge the recv function Chaoyong He
2023-06-07 14:23 ` [PATCH 00/10] support rte_flow for flower firmware with NFDk Ferruh Yigit
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=20230607015709.336420-7-chaoyong.he@corigine.com \
--to=chaoyong.he@corigine.com \
--cc=dev@dpdk.org \
--cc=niklas.soderlund@corigine.com \
--cc=oss-drivers@corigine.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).