[2/4] node: add a node to receive pkts from kernel
Checks
Commit Message
Patch adds a node to receive packets from kernel
over a raw socket.
Signed-off-by: Vamsi Attunuru <vattunuru@marvell.com>
---
doc/guides/prog_guide/graph_lib.rst | 7 +
lib/node/kernel_recv.c | 277 ++++++++++++++++++++++++++++
lib/node/kernel_recv_priv.h | 74 ++++++++
lib/node/meson.build | 1 +
4 files changed, 359 insertions(+)
Comments
On Fri, Apr 21, 2023 at 11:33 AM Vamsi Attunuru <vattunuru@marvell.com> wrote:
>
> Patch adds a node to receive packets from kernel
> over a raw socket.
Subjection suggestion: node/kernel_rx: support receving packet from kernel
>
> Signed-off-by: Vamsi Attunuru <vattunuru@marvell.com>
> ---
> doc/guides/prog_guide/graph_lib.rst | 7 +
> lib/node/kernel_recv.c | 277 ++++++++++++++++++++++++++++
> lib/node/kernel_recv_priv.h | 74 ++++++++
> lib/node/meson.build | 1 +
> 4 files changed, 359 insertions(+)
>
> diff --git a/doc/guides/prog_guide/graph_lib.rst b/doc/guides/prog_guide/graph_lib.rst
> index b3b5b14827..1057f16de8 100644
> --- a/doc/guides/prog_guide/graph_lib.rst
> +++ b/doc/guides/prog_guide/graph_lib.rst
> @@ -402,3 +402,10 @@ on the raw socket.
>
> Aftering sending the burst of packets to kernel, this node redirects the same
> objects to pkt_drop node to free up the packet buffers.
> +
> +kernel_recv
Better to change node name to kernel_rx
> +~~~~~~~~~~~
> +This node receives packets from kernel over a raw socket interface. Uses ``poll``
> +function to poll on the socket fd for ``POLLIN`` events to read the packets from
> +raw socket to stream buffer and does ``rte_node_next_stream_move()`` when there
> +are received packets.
You can tell typical use case for punt and kernel_tx node. i.e
expection path handling.
Also, may consider chnage to kernel_tx for punt node.
> diff --git a/lib/node/kernel_recv.c b/lib/node/kernel_recv.c
> new file mode 100644
> index 0000000000..361dcc3b5f
> --- /dev/null
> +++ b/lib/node/kernel_recv.c
> @@ -0,0 +1,277 @@
> +/* SPDX-License-Identifier: BSD-3-Clause
> + * Copyright(C) 2023 Marvell International Ltd.
> + */
> +
> +#include <fcntl.h>
> +#include <poll.h>
> +#include <stdlib.h>
> +#include <sys/ioctl.h>
> +#include <sys/socket.h>
> +#include <unistd.h>
> +
> +#include <rte_debug.h>
> +#include <rte_ethdev.h>
> +#include <rte_graph.h>
> +#include <rte_graph_worker.h>
> +#include <rte_ip.h>
> +#include <rte_malloc.h>
> +#include <rte_mbuf.h>
> +#include <rte_mempool.h>
> +#include <rte_net.h>
> +
> +#include "ethdev_rx_priv.h"
> +#include "kernel_recv_priv.h"
> +#include "node_private.h"
> +
> +static struct kernel_recv_node_main kernel_recv_main;
Try to remove global varible.
> +#ifndef __INCLUDE_KERNEL_RECV_PRIV_H__
No need to add INCLUDE
> +#define __INCLUDE_KERNEL_RECV_PRIV_H__
> +
> +#define KERN_RECV_CACHE_COUNT 64
> +
> +typedef struct kernel_recv_info {
> + struct rte_mbuf *rx_bufs[KERN_RECV_CACHE_COUNT];
> + uint16_t cls_next;
> + uint16_t idx;
> + uint16_t cnt;
> + int sock;
> +} kernel_recv_info_t;
> +
> +/**
> + * @internal
> + *
> + * Kernel Recv node context structure.
Across the patch series, no need for Doxgen comments for internal functions.
> -----Original Message-----
> From: Jerin Jacob <jerinjacobk@gmail.com>
> Sent: Monday, May 29, 2023 11:10 PM
> To: Vamsi Krishna Attunuru <vattunuru@marvell.com>
> Cc: dev@dpdk.org; Jerin Jacob Kollanukkaran <jerinj@marvell.com>; Nithin
> Kumar Dabilpuram <ndabilpuram@marvell.com>
> Subject: [EXT] Re: [PATCH 2/4] node: add a node to receive pkts from kernel
>
> External Email
>
> ----------------------------------------------------------------------
> On Fri, Apr 21, 2023 at 11:33 AM Vamsi Attunuru <vattunuru@marvell.com>
> wrote:
> >
> > Patch adds a node to receive packets from kernel over a raw socket.
>
> Subjection suggestion: node/kernel_rx: support receving packet from kernel
> >
> > Signed-off-by: Vamsi Attunuru <vattunuru@marvell.com>
> > ---
> > doc/guides/prog_guide/graph_lib.rst | 7 +
> > lib/node/kernel_recv.c | 277 ++++++++++++++++++++++++++++
> > lib/node/kernel_recv_priv.h | 74 ++++++++
> > lib/node/meson.build | 1 +
> > 4 files changed, 359 insertions(+)
> >
> > diff --git a/doc/guides/prog_guide/graph_lib.rst
> > b/doc/guides/prog_guide/graph_lib.rst
> > index b3b5b14827..1057f16de8 100644
> > --- a/doc/guides/prog_guide/graph_lib.rst
> > +++ b/doc/guides/prog_guide/graph_lib.rst
> > @@ -402,3 +402,10 @@ on the raw socket.
> >
> > Aftering sending the burst of packets to kernel, this node redirects
> > the same objects to pkt_drop node to free up the packet buffers.
> > +
> > +kernel_recv
>
> Better to change node name to kernel_rx
>
> > +~~~~~~~~~~~
> > +This node receives packets from kernel over a raw socket interface.
> > +Uses ``poll`` function to poll on the socket fd for ``POLLIN`` events
> > +to read the packets from raw socket to stream buffer and does
> > +``rte_node_next_stream_move()`` when there are received packets.
>
> You can tell typical use case for punt and kernel_tx node. i.e expection path
> handling.
> Also, may consider chnage to kernel_tx for punt node.
>
ack
>
> > diff --git a/lib/node/kernel_recv.c b/lib/node/kernel_recv.c new file
> > mode 100644 index 0000000000..361dcc3b5f
> > --- /dev/null
> > +++ b/lib/node/kernel_recv.c
> > @@ -0,0 +1,277 @@
> > +/* SPDX-License-Identifier: BSD-3-Clause
> > + * Copyright(C) 2023 Marvell International Ltd.
> > + */
> > +
> > +#include <fcntl.h>
> > +#include <poll.h>
> > +#include <stdlib.h>
> > +#include <sys/ioctl.h>
> > +#include <sys/socket.h>
> > +#include <unistd.h>
> > +
> > +#include <rte_debug.h>
> > +#include <rte_ethdev.h>
> > +#include <rte_graph.h>
> > +#include <rte_graph_worker.h>
> > +#include <rte_ip.h>
> > +#include <rte_malloc.h>
> > +#include <rte_mbuf.h>
> > +#include <rte_mempool.h>
> > +#include <rte_net.h>
> > +
> > +#include "ethdev_rx_priv.h"
> > +#include "kernel_recv_priv.h"
> > +#include "node_private.h"
> > +
> > +static struct kernel_recv_node_main kernel_recv_main;
>
> Try to remove global varible.
>
> > +#ifndef __INCLUDE_KERNEL_RECV_PRIV_H__
>
> No need to add INCLUDE
>
> > +#define __INCLUDE_KERNEL_RECV_PRIV_H__
> > +
> > +#define KERN_RECV_CACHE_COUNT 64
> > +
> > +typedef struct kernel_recv_info {
> > + struct rte_mbuf *rx_bufs[KERN_RECV_CACHE_COUNT];
> > + uint16_t cls_next;
> > + uint16_t idx;
> > + uint16_t cnt;
> > + int sock;
> > +} kernel_recv_info_t;
> > +
> > +/**
> > + * @internal
> > + *
> > + * Kernel Recv node context structure.
>
> Across the patch series, no need for Doxgen comments for internal
> functions.
Sure Jerin.
@@ -402,3 +402,10 @@ on the raw socket.
Aftering sending the burst of packets to kernel, this node redirects the same
objects to pkt_drop node to free up the packet buffers.
+
+kernel_recv
+~~~~~~~~~~~
+This node receives packets from kernel over a raw socket interface. Uses ``poll``
+function to poll on the socket fd for ``POLLIN`` events to read the packets from
+raw socket to stream buffer and does ``rte_node_next_stream_move()`` when there
+are received packets.
new file mode 100644
@@ -0,0 +1,277 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Marvell International Ltd.
+ */
+
+#include <fcntl.h>
+#include <poll.h>
+#include <stdlib.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#include <unistd.h>
+
+#include <rte_debug.h>
+#include <rte_ethdev.h>
+#include <rte_graph.h>
+#include <rte_graph_worker.h>
+#include <rte_ip.h>
+#include <rte_malloc.h>
+#include <rte_mbuf.h>
+#include <rte_mempool.h>
+#include <rte_net.h>
+
+#include "ethdev_rx_priv.h"
+#include "kernel_recv_priv.h"
+#include "node_private.h"
+
+static struct kernel_recv_node_main kernel_recv_main;
+
+static inline struct rte_mbuf *
+alloc_rx_mbuf(kernel_recv_node_ctx_t *ctx)
+{
+ kernel_recv_info_t *rx = ctx->recv_info;
+
+ if (rx->idx >= rx->cnt) {
+ uint16_t cnt;
+
+ rx->idx = 0;
+ rx->cnt = 0;
+
+ cnt = rte_pktmbuf_alloc_bulk(ctx->pktmbuf_pool, rx->rx_bufs, KERN_RECV_CACHE_COUNT);
+ if (cnt <= 0)
+ return NULL;
+
+ rx->cnt = cnt;
+ }
+
+ return rx->rx_bufs[rx->idx++];
+}
+
+static inline void
+mbuf_update(struct rte_mbuf **mbufs, uint16_t nb_pkts)
+{
+ struct rte_net_hdr_lens hdr_lens;
+ struct rte_mbuf *m;
+ int i;
+
+ for (i = 0; i < nb_pkts; i++) {
+ m = mbufs[i];
+
+ m->packet_type = rte_net_get_ptype(m, &hdr_lens, RTE_PTYPE_ALL_MASK);
+
+ m->ol_flags = 0;
+ m->tx_offload = 0;
+
+ m->l2_len = hdr_lens.l2_len;
+ m->l3_len = hdr_lens.l3_len;
+ m->l4_len = hdr_lens.l4_len;
+ }
+}
+
+static uint16_t
+recv_pkt_parse(void **objs, uint16_t nb_pkts)
+{
+ uint16_t pkts_left = nb_pkts;
+ struct rte_mbuf **pkts;
+ int i;
+
+ pkts = (struct rte_mbuf **)objs;
+
+ if (pkts_left >= 4) {
+ for (i = 0; i < 4; i++)
+ rte_prefetch0(rte_pktmbuf_mtod(pkts[i], void *));
+ }
+
+ while (pkts_left >= 12) {
+ /* Prefetch next-next mbufs */
+ rte_prefetch0(pkts[8]);
+ rte_prefetch0(pkts[9]);
+ rte_prefetch0(pkts[10]);
+ rte_prefetch0(pkts[11]);
+
+ /* Prefetch next mbuf data */
+ rte_prefetch0(rte_pktmbuf_mtod(pkts[4], void *));
+ rte_prefetch0(rte_pktmbuf_mtod(pkts[5], void *));
+ rte_prefetch0(rte_pktmbuf_mtod(pkts[6], void *));
+ rte_prefetch0(rte_pktmbuf_mtod(pkts[7], void *));
+
+ /* Extract ptype of mbufs */
+ mbuf_update(pkts, 4);
+
+ pkts += 4;
+ pkts_left -= 4;
+ }
+
+ if (pkts_left > 0)
+ mbuf_update(pkts, pkts_left);
+
+ return nb_pkts;
+}
+
+static uint16_t
+kernel_recv_node_do(struct rte_graph *graph, struct rte_node *node, kernel_recv_node_ctx_t *ctx)
+{
+ kernel_recv_info_t *rx;
+ uint16_t next_index;
+ int fd;
+
+ rx = ctx->recv_info;
+ next_index = rx->cls_next;
+
+ fd = rx->sock;
+ if (fd > 0) {
+ struct rte_mbuf **mbufs;
+ uint16_t len = 0, count = 0;
+ int nb_cnt, i;
+
+ nb_cnt = (node->size >= RTE_GRAPH_BURST_SIZE) ? RTE_GRAPH_BURST_SIZE : node->size;
+
+ mbufs = (struct rte_mbuf **)node->objs;
+ for (i = 0; i < nb_cnt; i++) {
+ struct rte_mbuf *m = alloc_rx_mbuf(ctx);
+
+ if (!m)
+ break;
+
+ len = read(fd, rte_pktmbuf_mtod(m, char *), rte_pktmbuf_tailroom(m));
+ if (len == 0 || len == 0xFFFF) {
+ rte_pktmbuf_free(m);
+ if (rx->idx <= 0)
+ node_dbg("kernel_recv", "rx_mbuf array is empty\n");
+ rx->idx--;
+ break;
+ }
+ *mbufs++ = m;
+
+ m->port = node->id;
+ rte_pktmbuf_data_len(m) = len;
+
+ count++;
+ }
+
+ if (count) {
+ recv_pkt_parse(node->objs, count);
+ node->idx = count;
+
+ /* Enqueue to next node */
+ rte_node_next_stream_move(graph, node, next_index);
+ }
+
+ return count;
+ }
+
+ return 0;
+}
+
+static uint16_t
+kernel_recv_node_process(struct rte_graph *graph, struct rte_node *node, void **objs,
+ uint16_t nb_objs)
+{
+ kernel_recv_node_ctx_t *ctx = (kernel_recv_node_ctx_t *)node->ctx;
+ int fd;
+
+ RTE_SET_USED(objs);
+ RTE_SET_USED(nb_objs);
+
+ if (!ctx)
+ return 0;
+
+ fd = ctx->recv_info->sock;
+ if (fd > 0) {
+ struct pollfd fds = {.fd = fd, .events = POLLIN};
+
+ if (poll(&fds, 1, 0) > 0) {
+ if (fds.revents & POLLIN)
+ return kernel_recv_node_do(graph, node, ctx);
+ }
+ }
+
+ return 0;
+}
+
+static int
+kernel_recv_node_init(const struct rte_graph *graph, struct rte_node *node)
+{
+ kernel_recv_node_ctx_t *ctx = (kernel_recv_node_ctx_t *)node->ctx;
+ kernel_recv_node_elem_t *elem = kernel_recv_main.head;
+ kernel_recv_info_t *recv_info;
+ int sock;
+
+ while (elem) {
+ if (elem->nid == node->id) {
+ /* Update node specific context */
+ memcpy(ctx, &elem->ctx, sizeof(kernel_recv_node_ctx_t));
+ break;
+ }
+ elem = elem->next;
+ }
+
+ RTE_VERIFY(elem != NULL);
+
+ if (ctx->pktmbuf_pool == NULL) {
+ node_err("kernel_recv", "Invalid mbuf pool on graph %s\n", graph->name);
+ return -EINVAL;
+ }
+
+ recv_info =
+ rte_zmalloc("kernel_recv_info", sizeof(kernel_recv_info_t), RTE_CACHE_LINE_SIZE);
+ if (!recv_info) {
+ node_err("kernel_recv", "Kernel recv_info is NULL\n");
+ return -ENOMEM;
+ }
+
+ sock = socket(AF_INET, SOCK_RAW, IPPROTO_RAW);
+ if (sock < 0) {
+ node_err("kernel_recv", "Unable to open RAW socket\n");
+ return sock;
+ }
+
+ recv_info->cls_next = ETHDEV_RX_NEXT_PKT_CLS;
+ recv_info->sock = sock;
+ ctx->recv_info = recv_info;
+
+ return 0;
+}
+
+static void
+kernel_recv_node_fini(const struct rte_graph *graph __rte_unused, struct rte_node *node)
+{
+ kernel_recv_node_ctx_t *ctx = (kernel_recv_node_ctx_t *)node->ctx;
+
+ if (ctx->recv_info) {
+ close(ctx->recv_info->sock);
+ ctx->recv_info->sock = -1;
+ rte_free(ctx->recv_info);
+ }
+
+ ctx->recv_info = NULL;
+}
+
+struct kernel_recv_node_main *
+kernel_recv_node_data_get(void)
+{
+ return &kernel_recv_main;
+}
+
+static struct rte_node_register kernel_recv_node_base = {
+ .process = kernel_recv_node_process,
+ .flags = RTE_NODE_SOURCE_F,
+ .name = "kernel_recv",
+
+ .init = kernel_recv_node_init,
+ .fini = kernel_recv_node_fini,
+
+ .nb_edges = KERNEL_RECV_NEXT_MAX,
+ .next_nodes = {
+ /* Default pkt classification node */
+ [KERNEL_RECV_NEXT_PKT_CLS] = "pkt_cls",
+ [KERNEL_RECV_NEXT_IP4_LOOKUP] = "ip4_lookup",
+ },
+};
+
+struct rte_node_register *
+kernel_recv_node_get(void)
+{
+ return &kernel_recv_node_base;
+}
+
+RTE_NODE_REGISTER(kernel_recv_node_base);
new file mode 100644
@@ -0,0 +1,74 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Marvell International Ltd.
+ */
+
+#ifndef __INCLUDE_KERNEL_RECV_PRIV_H__
+#define __INCLUDE_KERNEL_RECV_PRIV_H__
+
+#define KERN_RECV_CACHE_COUNT 64
+
+typedef struct kernel_recv_info {
+ struct rte_mbuf *rx_bufs[KERN_RECV_CACHE_COUNT];
+ uint16_t cls_next;
+ uint16_t idx;
+ uint16_t cnt;
+ int sock;
+} kernel_recv_info_t;
+
+/**
+ * @internal
+ *
+ * Kernel Recv node context structure.
+ */
+typedef struct kernel_recv_node_ctx {
+ struct rte_mempool *pktmbuf_pool;
+ kernel_recv_info_t *recv_info;
+} kernel_recv_node_ctx_t;
+
+/**
+ * @internal
+ *
+ * Kernel Recv node list element structure.
+ */
+typedef struct kernel_recv_node_elem {
+ struct kernel_recv_node_elem *next; /**< Pointer to the next node element. */
+ struct kernel_recv_node_ctx ctx; /**< Kernel Recv node context. */
+ rte_node_t nid; /**< Node identifier of the Kernel Recv node. */
+} kernel_recv_node_elem_t;
+
+enum kernel_recv_next_nodes {
+ KERNEL_RECV_NEXT_IP4_LOOKUP,
+ KERNEL_RECV_NEXT_PKT_CLS,
+ KERNEL_RECV_NEXT_MAX,
+};
+
+/**
+ * @internal
+ *
+ * Kernel Recv node main structure.
+ */
+struct kernel_recv_node_main {
+ kernel_recv_node_elem_t *head; /**< Pointer to the head node element. */
+};
+
+/**
+ * @internal
+ *
+ * Get the Kernel Recv node data.
+ *
+ * @return
+ * Pointer to Kernel Recv node data.
+ */
+struct kernel_recv_node_main *kernel_recv_node_data_get(void);
+
+/**
+ * @internal
+ *
+ * Get the Kernel Recv node.
+ *
+ * @return
+ * Pointer to the Kernel Recv node.
+ */
+struct rte_node_register *kernel_recv_node_get(void);
+
+#endif /* __INCLUDE_KERNEL_RECV_PRIV_H__ */
@@ -13,6 +13,7 @@ sources = files(
'ethdev_tx.c',
'ip4_lookup.c',
'ip4_rewrite.c',
+ 'kernel_recv.c',
'log.c',
'null.c',
'pkt_cls.c',