nuttx/net/pkt/pkt_recvmsg.c

504 lines
14 KiB
C
Raw Normal View History

/****************************************************************************
* net/pkt/pkt_recvmsg.c
*
* SPDX-License-Identifier: Apache-2.0
*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership. The
* ASF licenses this file to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance with the
* License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <nuttx/config.h>
#ifdef CONFIG_NET
#include <sys/types.h>
#include <sys/socket.h>
#include <stdint.h>
#include <string.h>
#include <errno.h>
#include <debug.h>
#include <assert.h>
#include <arch/irq.h>
#include <nuttx/semaphore.h>
#include <nuttx/net/net.h>
#include <nuttx/net/netdev.h>
#include "netdev/netdev.h"
#include "devif/devif.h"
#include "pkt/pkt.h"
#include "socket/socket.h"
#include <netpacket/packet.h>
/****************************************************************************
* Private Types
****************************************************************************/
struct pkt_recvfrom_s
{
FAR struct devif_callback_s *pr_cb; /* Reference to callback instance */
sem_t pr_sem; /* Semaphore signals recv completion */
size_t pr_buflen; /* Length of receive buffer */
FAR uint8_t *pr_buffer; /* Pointer to receive buffer */
ssize_t pr_recvlen; /* The received length */
int pr_result; /* Success:OK, failure:negated errno */
};
/****************************************************************************
* Private Functions
****************************************************************************/
/****************************************************************************
* Name: pkt_add_recvlen
*
* Description:
* Update information about space available for new data and update size
* of data in buffer, This logic accounts for the case where
* recvfrom_udpreadahead() sets state.pr_recvlen == -1 .
*
* Input Parameters:
* pstate recvfrom state structure
* recvlen size of new data appended to buffer
*
* Returned Value:
* None
*
****************************************************************************/
static inline void pkt_add_recvlen(FAR struct pkt_recvfrom_s *pstate,
size_t recvlen)
{
if (pstate->pr_recvlen < 0)
{
pstate->pr_recvlen = 0;
}
pstate->pr_recvlen += recvlen;
pstate->pr_buffer += recvlen;
pstate->pr_buflen -= recvlen;
}
/****************************************************************************
* Name: pkt_recvfrom_newdata
*
* Description:
* Copy the read data from the packet
*
* Input Parameters:
* dev The structure of the network driver that caused the event.
* pstate recvfrom state structure
*
* Returned Value:
* None.
*
* Assumptions:
* The network is locked.
*
****************************************************************************/
static void pkt_recvfrom_newdata(FAR struct net_driver_s *dev,
FAR struct pkt_recvfrom_s *pstate)
{
net/l2/l3/l4: add support of iob offload 1. Add new config CONFIG_NET_LL_GUARDSIZE to isolation of l2 stack, which will benefit l3(IP) layer for multi-MAC(l2) implementation, especially in some NICs such as celluler net driver. new configuration options: CONFIG_NET_LL_GUARDSIZE CONFIG_NET_LL_GUARDSIZE will reserved l2 buffer header size of network buffer to isolate the L2/L3 (MAC/IP) data on network layer, which will be beneficial to L3 network layer protocol transparent transmission and forwarding ------------------------------------------------------------ Layout of frist iob entry: iob_data (aligned by CONFIG_IOB_ALIGNMENT) | | io_offset(CONFIG_NET_LL_GUARDSIZE) | | ------------------------------------------------- iob | Reserved | io_len | ------------------------------------------------- ------------------------------------------------------------- Layout of different NICs implementation: iob_data (aligned by CONFIG_IOB_ALIGNMENT) | | io_offset(CONFIG_NET_LL_GUARDSIZE) | | ------------------------------------------------- Ethernet | Reserved | ETH_HDRLEN | io_len | ---------------------------------|--------------- 8021Q | Reserved | ETH_8021Q_HDRLEN | io_len | ---------------------------------|--------------- ipforward | Reserved | io_len | ------------------------------------------------- -------------------------------------------------------------------- 2. Support iob offload to l2 driver to avoid unnecessary memory copy Support send/receive iob vectors directly between the NICs and l3/l4 stack to avoid unnecessary memory copies, especially on hardware that supports Scatter/gather, which can greatly improve performance. new interface to support iob offload: ------------------------------------------ | IOB version | original | |----------------------------------------| | devif_iob_poll() | devif_poll() | | ... | ... | ------------------------------------------ -------------------------------------------------------------------- 1> NIC hardware support Scatter/gather transfer TX: tcp_poll()/udp_poll()/pkt_poll()/...(l3|l4) / \ / \ devif_poll_[l3|l4]_connections() devif_iob_send() (nocopy:udp/icmp/...) / \ (copy:tcp) / \ devif_iob_poll("NIC"_txpoll) callback() // "NIC"_txpoll | dev->d_iob: | --------------- --------------- io_data iob1 | | | iob3 | | | \ --------------- --------------- --------------- | --------------- | iob0 | | | | iob2 | | | | --------------- | --------------- | \ | / / \ | / / ---------------------------------------------- NICs io vector | | | | | | | | | | ---------------------------------------------- RX: [tcp|udp|icmp|...]ipv[4|6]_data_handler()(iob_concat/append to readahead) | | [tcp|udp|icmp|...]_ipv[4|6]_in()/... | | pkt/ipv[4/6]_input()/... | | NICs io vector receive(iov_base to each iobs) -------------------------------------------------------------------- 2> CONFIG_IOB_BUFSIZE is greater than MTU: TX: "(CONFIG_IOB_BUFSIZE) > (MAX_NETDEV_PKTSIZE + CONFIG_NET_GUARDSIZE + CONFIG_NET_LL_GUARDSIZE)" tcp_poll()/udp_poll()/pkt_poll()/...(l3|l4) / \ / \ devif_poll_[l3|l4]_connections() devif_iob_send() (nocopy:udp/icmp/...) / \ (copy:tcp) / \ devif_iob_poll("NIC"_txpoll) callback() // "NIC"_txpoll | "NIC"_send() (dev->d_iob->io_data[CONFIG_NET_LL_GUARDSIZE - NET_LL_HDRLEN(dev)]) RX: [tcp|udp|icmp|...]ipv[4|6]_data_handler()(iob_concat/append to readahead) | | [tcp|udp|icmp|...]_ipv[4|6]_in()/... | | pkt/ipv[4/6]_input()/... | | NICs io vector receive(iov_base to io_data) -------------------------------------------------------------------- 3> Compatible with all old flat buffer NICs TX: tcp_poll()/udp_poll()/pkt_poll()/...(l3|l4) / \ / \ devif_poll_[l3|l4]_connections() devif_iob_send() (nocopy:udp/icmp/...) / \ (copy:tcp) / \ devif_iob_poll(devif_poll_callback()) devif_poll_callback() /* new interface, gather iobs to flat buffer */ / \ / \ devif_poll("NIC"_txpoll) "NIC"_send()(dev->d_buf) RX: [tcp|udp|icmp|...]ipv[4|6]_data_handler()(iob_concat/append to readahead) | | [tcp|udp|icmp|...]_ipv[4|6]_in()/... | | netdev_input() /* new interface, Scatter/gather flat/iob buffer */ | | pkt/ipv[4|6]_input()/... | | NICs io vector receive(Orignal flat buffer) 3. Iperf passthrough on NuttX simulator: ------------------------------------------------- | Protocol | Server | Client | | |-----------------------------------------------| | TCP | 813 | 834 | Mbits/sec | | TCP(Offload) | 1720 | 1100 | Mbits/sec | | UDP | 22 | 757 | Mbits/sec | | UDP(Offload) | 25 | 1250 | Mbits/sec | ------------------------------------------------- Signed-off-by: chao an <anchao@xiaomi.com>
2022-11-23 12:39:40 +01:00
unsigned int offset;
size_t recvlen;
if (dev->d_len > pstate->pr_buflen)
{
recvlen = pstate->pr_buflen;
}
else
{
recvlen = dev->d_len;
}
/* Copy the new packet data into the user buffer */
net/l2/l3/l4: add support of iob offload 1. Add new config CONFIG_NET_LL_GUARDSIZE to isolation of l2 stack, which will benefit l3(IP) layer for multi-MAC(l2) implementation, especially in some NICs such as celluler net driver. new configuration options: CONFIG_NET_LL_GUARDSIZE CONFIG_NET_LL_GUARDSIZE will reserved l2 buffer header size of network buffer to isolate the L2/L3 (MAC/IP) data on network layer, which will be beneficial to L3 network layer protocol transparent transmission and forwarding ------------------------------------------------------------ Layout of frist iob entry: iob_data (aligned by CONFIG_IOB_ALIGNMENT) | | io_offset(CONFIG_NET_LL_GUARDSIZE) | | ------------------------------------------------- iob | Reserved | io_len | ------------------------------------------------- ------------------------------------------------------------- Layout of different NICs implementation: iob_data (aligned by CONFIG_IOB_ALIGNMENT) | | io_offset(CONFIG_NET_LL_GUARDSIZE) | | ------------------------------------------------- Ethernet | Reserved | ETH_HDRLEN | io_len | ---------------------------------|--------------- 8021Q | Reserved | ETH_8021Q_HDRLEN | io_len | ---------------------------------|--------------- ipforward | Reserved | io_len | ------------------------------------------------- -------------------------------------------------------------------- 2. Support iob offload to l2 driver to avoid unnecessary memory copy Support send/receive iob vectors directly between the NICs and l3/l4 stack to avoid unnecessary memory copies, especially on hardware that supports Scatter/gather, which can greatly improve performance. new interface to support iob offload: ------------------------------------------ | IOB version | original | |----------------------------------------| | devif_iob_poll() | devif_poll() | | ... | ... | ------------------------------------------ -------------------------------------------------------------------- 1> NIC hardware support Scatter/gather transfer TX: tcp_poll()/udp_poll()/pkt_poll()/...(l3|l4) / \ / \ devif_poll_[l3|l4]_connections() devif_iob_send() (nocopy:udp/icmp/...) / \ (copy:tcp) / \ devif_iob_poll("NIC"_txpoll) callback() // "NIC"_txpoll | dev->d_iob: | --------------- --------------- io_data iob1 | | | iob3 | | | \ --------------- --------------- --------------- | --------------- | iob0 | | | | iob2 | | | | --------------- | --------------- | \ | / / \ | / / ---------------------------------------------- NICs io vector | | | | | | | | | | ---------------------------------------------- RX: [tcp|udp|icmp|...]ipv[4|6]_data_handler()(iob_concat/append to readahead) | | [tcp|udp|icmp|...]_ipv[4|6]_in()/... | | pkt/ipv[4/6]_input()/... | | NICs io vector receive(iov_base to each iobs) -------------------------------------------------------------------- 2> CONFIG_IOB_BUFSIZE is greater than MTU: TX: "(CONFIG_IOB_BUFSIZE) > (MAX_NETDEV_PKTSIZE + CONFIG_NET_GUARDSIZE + CONFIG_NET_LL_GUARDSIZE)" tcp_poll()/udp_poll()/pkt_poll()/...(l3|l4) / \ / \ devif_poll_[l3|l4]_connections() devif_iob_send() (nocopy:udp/icmp/...) / \ (copy:tcp) / \ devif_iob_poll("NIC"_txpoll) callback() // "NIC"_txpoll | "NIC"_send() (dev->d_iob->io_data[CONFIG_NET_LL_GUARDSIZE - NET_LL_HDRLEN(dev)]) RX: [tcp|udp|icmp|...]ipv[4|6]_data_handler()(iob_concat/append to readahead) | | [tcp|udp|icmp|...]_ipv[4|6]_in()/... | | pkt/ipv[4/6]_input()/... | | NICs io vector receive(iov_base to io_data) -------------------------------------------------------------------- 3> Compatible with all old flat buffer NICs TX: tcp_poll()/udp_poll()/pkt_poll()/...(l3|l4) / \ / \ devif_poll_[l3|l4]_connections() devif_iob_send() (nocopy:udp/icmp/...) / \ (copy:tcp) / \ devif_iob_poll(devif_poll_callback()) devif_poll_callback() /* new interface, gather iobs to flat buffer */ / \ / \ devif_poll("NIC"_txpoll) "NIC"_send()(dev->d_buf) RX: [tcp|udp|icmp|...]ipv[4|6]_data_handler()(iob_concat/append to readahead) | | [tcp|udp|icmp|...]_ipv[4|6]_in()/... | | netdev_input() /* new interface, Scatter/gather flat/iob buffer */ | | pkt/ipv[4|6]_input()/... | | NICs io vector receive(Orignal flat buffer) 3. Iperf passthrough on NuttX simulator: ------------------------------------------------- | Protocol | Server | Client | | |-----------------------------------------------| | TCP | 813 | 834 | Mbits/sec | | TCP(Offload) | 1720 | 1100 | Mbits/sec | | UDP | 22 | 757 | Mbits/sec | | UDP(Offload) | 25 | 1250 | Mbits/sec | ------------------------------------------------- Signed-off-by: chao an <anchao@xiaomi.com>
2022-11-23 12:39:40 +01:00
offset = (dev->d_appdata - dev->d_iob->io_data) - dev->d_iob->io_offset;
recvlen = iob_copyout(pstate->pr_buffer, dev->d_iob, recvlen, offset);
ninfo("Received %d bytes (of %d)\n", (int)recvlen, (int)dev->d_len);
/* Update the accumulated size of the data read */
pkt_add_recvlen(pstate, recvlen);
}
/****************************************************************************
* Name: pkt_recvfrom_sender
*
* Description:
*
* Input Parameters:
*
* Returned Value:
*
* Assumptions:
*
****************************************************************************/
static inline void pkt_recvfrom_sender(FAR struct net_driver_s *dev,
FAR struct pkt_recvfrom_s *pstate)
{
}
/****************************************************************************
* Name: pkt_recvfrom_eventhandler
*
* Description:
*
* Input Parameters:
*
* Returned Value:
*
* Assumptions:
*
****************************************************************************/
static uint16_t pkt_recvfrom_eventhandler(FAR struct net_driver_s *dev,
FAR void *pvpriv, uint16_t flags)
{
struct pkt_recvfrom_s *pstate = pvpriv;
ninfo("flags: %04x\n", flags);
/* 'priv' might be null in some race conditions (?) */
if (pstate)
{
/* If a new packet is available, then complete the read action. */
if ((flags & PKT_NEWDATA) != 0)
{
/* Copy the packet */
pkt_recvfrom_newdata(dev, pstate);
/* We are finished. */
ninfo("PKT done\n");
/* Don't allow any further call backs. */
pstate->pr_cb->flags = 0;
pstate->pr_cb->priv = NULL;
pstate->pr_cb->event = NULL;
/* Save the sender's address in the caller's 'from' location */
pkt_recvfrom_sender(dev, pstate);
/* indicate that the data has been consumed */
flags &= ~PKT_NEWDATA;
/* Wake up the waiting thread, returning the number of bytes
* actually read.
*/
nxsem_post(&pstate->pr_sem);
}
}
return flags;
}
/****************************************************************************
* Name: pkt_recvfrom_initialize
*
* Description:
* Initialize the state structure
*
* Input Parameters:
* psock Pointer to the socket structure for the socket
* buf Buffer to receive data
* len Length of buffer
* pstate A pointer to the state structure to be initialized
*
* Returned Value:
* None
*
* Assumptions:
*
****************************************************************************/
static void pkt_recvfrom_initialize(FAR struct socket *psock, FAR void *buf,
size_t len, FAR struct sockaddr *infrom,
FAR socklen_t *fromlen,
FAR struct pkt_recvfrom_s *pstate)
{
/* Initialize the state structure. */
memset(pstate, 0, sizeof(struct pkt_recvfrom_s));
nxsem_init(&pstate->pr_sem, 0, 0); /* Doesn't really fail */
pstate->pr_buflen = len;
pstate->pr_buffer = buf;
}
/* The only un-initialization that has to be performed is destroying the
* semaphore.
*/
#define pkt_recvfrom_uninitialize(s) nxsem_destroy(&(s)->pr_sem)
/****************************************************************************
* Name: pkt_recvfrom_result
*
* Description:
* Evaluate the result of the recv operations
*
* Input Parameters:
* result The result of the net_sem_wait operation (may indicate EINTR)
* pstate A pointer to the state structure to be initialized
*
* Returned Value:
* The result of the recv operation with errno set appropriately
*
* Assumptions:
*
****************************************************************************/
static ssize_t pkt_recvfrom_result(int result,
FAR struct pkt_recvfrom_s *pstate)
{
/* Check for a error/timeout detected by the event handler. Errors are
* signaled by negative errno values for the rcv length
*/
if (pstate->pr_result < 0)
{
/* This might return EAGAIN on a timeout or ENOTCONN on loss of
* connection (TCP only)
*/
return pstate->pr_result;
}
/* If net_sem_wait failed, then we were probably reawakened by a signal.
* In this case, net_sem_wait will have returned negated errno
* appropriately.
*/
if (result < 0)
{
return result;
}
return pstate->pr_recvlen;
}
/****************************************************************************
* Name: pkt_readahead
*
* Description:
* Copy the buffered read-ahead data to the user buffer.
*
* Input Parameters:
* conn - PKT socket connection structure containing the read-
* ahead data.
* buf target buffer.
*
* Returned Value:
* Number of bytes copied to the user buffer
*
* Assumptions:
* The network is locked.
*
****************************************************************************/
static inline ssize_t pkt_readahead(FAR struct pkt_conn_s *conn,
FAR void *buf, size_t buflen)
{
FAR struct iob_s *iob;
ssize_t ret = -ENODATA;
/* Check there is any packets already buffered in a read-ahead buffer. */
if ((iob = iob_peek_queue(&conn->readahead)) != NULL)
{
DEBUGASSERT(iob->io_pktlen > 0);
/* Copy to user */
ret = iob_copyout(buf, iob, buflen, 0);
ninfo("Received %zd bytes (of %u)\n", ret, iob->io_pktlen);
/* Remove the I/O buffer chain from the head of the read-ahead
* buffer queue.
*/
iob_remove_queue(&conn->readahead);
/* And free the I/O buffer chain */
iob_free_chain(iob);
}
return ret;
}
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Name: pkt_recvmsg
*
* Description:
* Implements the socket recvmsg interface for the case of the AF_INET
* and AF_INET6 address families. pkt_recvmsg() receives messages from
* a socket, and may be used to receive data on a socket whether or not it
* is connection-oriented.
*
* If 'msg_name' is not NULL, and the underlying protocol provides the
* source address, this source address is filled in. The argument
* 'msg_namelen' is initialized to the size of the buffer associated with
* msg_name, and modified on return to indicate the actual size of the
* address stored there.
*
* Input Parameters:
* psock A pointer to a NuttX-specific, internal socket structure
* msg Buffer to receive the message
* flags Receive flags
*
* Returned Value:
* On success, returns the number of characters received. If no data is
* available to be received and the peer has performed an orderly shutdown,
* recvmsg() will return 0. Otherwise, on errors, a negated errno value is
* returned (see recvmsg() for the list of appropriate error values).
*
****************************************************************************/
ssize_t pkt_recvmsg(FAR struct socket *psock, FAR struct msghdr *msg,
int flags)
{
FAR void *buf = msg->msg_iov->iov_base;
size_t len = msg->msg_iov->iov_len;
FAR struct sockaddr *from = msg->msg_name;
FAR socklen_t *fromlen = &msg->msg_namelen;
FAR struct pkt_conn_s *conn = psock->s_conn;
FAR struct net_driver_s *dev;
struct pkt_recvfrom_s state;
ssize_t ret;
/* If a 'from' address has been provided, verify that it is large
* enough to hold this address family.
*/
if (from != NULL && *fromlen < sizeof(sa_family_t))
{
return -EINVAL;
}
if (psock->s_type != SOCK_RAW)
{
nerr("ERROR: Unsupported socket type: %d\n", psock->s_type);
ret = -ENOSYS;
}
/* Perform the packet recvfrom() operation */
/* Initialize the state structure. This is done with the network
* locked because we don't want anything to happen until we are ready.
*/
net_lock();
/* Check if there is buffered read-ahead data for this socket. We may have
* already received the response to previous command.
*/
if (!IOB_QEMPTY(&conn->readahead))
{
ret = pkt_readahead(conn, buf, len);
}
else if (_SS_ISNONBLOCK(conn->sconn.s_flags) ||
(flags & MSG_DONTWAIT) != 0)
{
/* Handle non-blocking PKT sockets */
ret = -EAGAIN;
}
else
{
pkt_recvfrom_initialize(psock, buf, len, from, fromlen, &state);
/* Get the device driver that will service this transfer */
dev = pkt_find_device(conn);
if (dev == NULL)
{
ret = -ENODEV;
goto errout_with_state;
}
/* TODO pkt_recvfrom_initialize() expects from to be of type
* sockaddr_in, but in our case is sockaddr_ll
*/
#if 0
ret = pkt_connect(conn, NULL);
if (ret < 0)
{
goto errout_with_state;
}
#endif
/* Set up the callback in the connection */
state.pr_cb = pkt_callback_alloc(dev, conn);
if (state.pr_cb)
{
state.pr_cb->flags = (PKT_NEWDATA | PKT_POLL);
state.pr_cb->priv = (FAR void *)&state;
state.pr_cb->event = pkt_recvfrom_eventhandler;
/* Wait for either the receive to complete or for an error/timeout
* to occur. NOTES: (1) net_sem_wait will also terminate if a
* signal is received, (2) the network is locked! It will be
* un-locked while the task sleeps and automatically re-locked when
* the task restarts.
*/
ret = net_sem_wait(&state.pr_sem);
/* Make sure that no further events are processed */
pkt_callback_free(dev, conn, state.pr_cb);
ret = pkt_recvfrom_result(ret, &state);
}
else
{
ret = -EBUSY;
}
errout_with_state:
pkt_recvfrom_uninitialize(&state);
}
net_unlock();
return ret;
}
#endif /* CONFIG_NET */