px4-firmware/nuttx/drivers/net/slip.c

1018 lines
27 KiB
C
Raw Normal View History

/****************************************************************************
* drivers/net/slip.c
*
* Copyright (C) 2011-2012 Gregory Nutt. All rights reserved.
* Author: Gregory Nutt <gnutt@nuttx.org>
*
* Reference: RFC 1055
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* 3. Neither the name NuttX nor the names of its contributors may be
* used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
* OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
* AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <nuttx/config.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <stdint.h>
#include <stdbool.h>
#include <stdio.h>
#include <time.h>
#include <string.h>
#include <fcntl.h>
#include <unistd.h>
#include <errno.h>
#include <debug.h>
#include <nuttx/irq.h>
#include <nuttx/net/net.h>
#include <nuttx/net/uip/uip.h>
#include <nuttx/net/uip/uip-arch.h>
#if defined(CONFIG_NET) && defined(CONFIG_NET_SLIP)
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
/* NOTE: Slip requires UART hardware handshake. If hardware handshake is
* not available with your UART, then you might try the 'slattach' option
* -L which enable "3-wire operation." That allows operation without the
* hardware handshake (but with the possibility of data overrun).
*/
/* Configuration ************************************************************/
#if UIP_LLH_LEN > 0
# error "UIP_LLH_LEN must be set to zero"
#endif
#ifndef CONFIG_NET_NOINTS
# warning "CONFIG_NET_NOINTS must be set"
#endif
#ifndef CONFIG_NET_MULTIBUFFER
# warning "CONFIG_NET_MULTIBUFFER must be set"
#endif
#ifndef CONFIG_SLIP_STACKSIZE
# define CONFIG_SLIP_STACKSIZE 2048
#endif
#ifndef CONFIG_SLIP_DEFPRIO
# define CONFIG_SLIP_DEFPRIO 128
#endif
/* The Linux slip module hard-codes its MTU size to 296 (40 bytes for the
* IP+TPC headers plus 256 bytes of data). So you might as well set
* CONFIG_NET_BUFSIZE to 296 as well.
*
* There may be an issue with this setting, however. I see that Linux uses
* a MTU of 296 and window of 256, but actually only sends 168 bytes of data:
* 40 + 128. I believe that is to allow for the 2x worst cast packet
* expansion. Ideally we would like to advertise the 256 MSS, but restrict
* uIP to 128 bytes (possibly by modifying the uip_mss() macro).
*/
#if CONFIG_NET_BUFSIZE < 296
# error "CONFIG_NET_BUFSIZE >= 296 is required"
#elif CONFIG_NET_BUFSIZE > 296
# warning "CONFIG_NET_BUFSIZE == 296 is optimal"
#endif
/* CONFIG_SLIP_NINTERFACES determines the number of physical interfaces
* that will be supported.
*/
#ifndef CONFIG_SLIP_NINTERFACES
# define CONFIG_SLIP_NINTERFACES 1
#endif
/* SLIP special character codes *******************************************/
#define SLIP_END 0300 /* Indicates end of packet */
#define SLIP_ESC 0333 /* Indicates byte stuffing */
#define SLIP_ESC_END 0334 /* ESC ESC_END means SLIP_END data byte */
#define SLIP_ESC_ESC 0335 /* ESC ESC_ESC means ESC data byte */
/* General driver definitions **********************************************/
/* TX poll delay = 1 second = 1000000 microseconds. */
#define SLIP_WDDELAY (1*1000000)
#define SLIP_POLLHSEC (1*2)
/* Statistics helper */
#ifdef CONFIG_NET_STATISTICS
# define SLIP_STAT(p,f) (p->stats.f)++
#else
# define SLIP_STAT(p,f)
#endif
/****************************************************************************
* Private Types
****************************************************************************/
/* Driver statistics */
#ifdef CONFIG_NET_STATISTICS
struct slip_statistics_s
{
uint32_t transmitted; /* Number of packets transmitted */
uint32_t received /* Number of packets received */
};
#endif
/* The slip_driver_s encapsulates all state information for a single hardware
* interface
*/
struct slip_driver_s
{
volatile bool bifup; /* true:ifup false:ifdown */
int fd; /* TTY file descriptor */
pid_t rxpid; /* Receiver thread ID */
pid_t txpid; /* Transmitter thread ID */
sem_t waitsem; /* Mutually exclusive access to uIP */
uint16_t rxlen; /* The number of bytes in rxbuf */
/* Driver statistics */
#ifdef CONFIG_NET_STATISTICS
struct slip_statistics_s stats;
#endif
/* This holds the information visible to uIP/NuttX */
struct uip_driver_s dev; /* Interface understood by uIP */
uint8_t rxbuf[CONFIG_NET_BUFSIZE + 2];
uint8_t txbuf[CONFIG_NET_BUFSIZE + 2];
};
/****************************************************************************
* Private Data
****************************************************************************/
/* We really should get rid of CONFIG_SLIP_NINTERFACES and, instead,
* kmalloc() new interface instances as needed.
*/
static struct slip_driver_s g_slip[CONFIG_SLIP_NINTERFACES];
/****************************************************************************
* Private Function Prototypes
****************************************************************************/
static void slip_semtake(FAR struct slip_driver_s *priv);
/* Common TX logic */
static void slip_write(FAR struct slip_driver_s *priv, const uint8_t *buffer, int len);
static void slip_putc(FAR struct slip_driver_s *priv, int ch);
static int slip_transmit(FAR struct slip_driver_s *priv);
static int slip_uiptxpoll(struct uip_driver_s *dev);
static void slip_txtask(int argc, char *argv[]);
/* Packet receiver task */
static int slip_getc(FAR struct slip_driver_s *priv);
static inline void slip_receive(FAR struct slip_driver_s *priv);
static int slip_rxtask(int argc, char *argv[]);
/* NuttX callback functions */
static int slip_ifup(struct uip_driver_s *dev);
static int slip_ifdown(struct uip_driver_s *dev);
static int slip_txavail(struct uip_driver_s *dev);
#ifdef CONFIG_NET_IGMP
static int slip_addmac(struct uip_driver_s *dev, FAR const uint8_t *mac);
static int slip_rmmac(struct uip_driver_s *dev, FAR const uint8_t *mac);
#endif
/****************************************************************************
* Private Functions
****************************************************************************/
/****************************************************************************
* Name: slip_semtake
****************************************************************************/
static void slip_semtake(FAR struct slip_driver_s *priv)
{
/* Take the semaphore (perhaps waiting) */
while (sem_wait(&priv->waitsem) != 0)
{
/* The only case that an error should occur here is if
* the wait was awakened by a signal.
*/
ASSERT(errno == EINTR);
}
}
#define slip_semgive(p) sem_post(&(p)->waitsem);
/****************************************************************************
* Function: slip_write
*
* Description:
* Just an inline wrapper around fwrite with error checking.
*
* Parameters:
* priv - Reference to the driver state structure
* buffer - Buffer data to send
* len - Buffer length in bytes
*
****************************************************************************/
static inline void slip_write(FAR struct slip_driver_s *priv,
const uint8_t *buffer, int len)
{
/* Handle the case where the write is awakened by a signal */
while (write(priv->fd, buffer, len) < 0)
{
DEBUGASSERT(errno == EINTR);
}
}
/****************************************************************************
* Function: slip_putc
*
* Description:
* Just an inline wrapper around putc with error checking.
*
* Parameters:
* priv - Reference to the driver state structure
* ch - The character to send
*
****************************************************************************/
static inline void slip_putc(FAR struct slip_driver_s *priv, int ch)
{
uint8_t buffer = (uint8_t)ch;
slip_write(priv, &buffer, 1);
}
/****************************************************************************
* Function: slip_transmit
*
* Description:
* Start hardware transmission. Called either from the txdone interrupt
* handling or from watchdog based polling.
*
* Parameters:
* priv - Reference to the driver state structure
*
* Returned Value:
* OK on success; a negated errno on failure
*
****************************************************************************/
static int slip_transmit(FAR struct slip_driver_s *priv)
{
uint8_t *src;
uint8_t *start;
uint8_t esc;
int remaining;
int len;
/* Increment statistics */
nvdbg("Sending packet size %d\n", priv->dev.d_len);
SLIP_STAT(priv, transmitted);
/* Send an initial END character to flush out any data that may have
* accumulated in the receiver due to line noise
*/
slip_putc(priv, SLIP_END);
/* For each byte in the packet, send the appropriate character sequence */
src = priv->dev.d_buf;
remaining = priv->dev.d_len;
start = src;
len = 0;
while (remaining-- > 0)
{
switch (*src)
{
/* If it's the same code as an END character, we send a special two
* character code so as not to make the receiver think we sent an
* END
*/
case SLIP_END:
esc = SLIP_ESC_END;
goto escape;
/* If it's the same code as an ESC character, we send a special two
* character code so as not to make the receiver think we sent an
* ESC
*/
case SLIP_ESC:
esc = SLIP_ESC_ESC;
escape:
{
/* Flush any unsent data */
if (len > 0)
{
slip_write(priv, start, len);
/* Reset */
start = src + 1;
len = 0;
}
/* Then send the escape sequence */
slip_putc(priv, SLIP_ESC);
slip_putc(priv, esc);
}
break;
/* otherwise, just bump up the count */
default:
len++;
break;
}
/* Point to the next character in the packet */
src++;
}
/* We have looked at every character in the packet. Now flush any unsent
* data
*/
if (len > 0)
{
slip_write(priv, start, len);
}
/* And send the END token */
slip_putc(priv, SLIP_END);
return OK;
}
/****************************************************************************
* Function: slip_uiptxpoll
*
* Description:
* Check if uIP has any outgoing packets ready to send. This is a
* callback from uip_poll(). uip_poll() may be called:
*
* 1. When the preceding TX packet send is complete, or
* 2. When the preceding TX packet send times o ]ut and the interface is reset
* 3. During normal TX polling
*
* Parameters:
* dev - Reference to the NuttX driver state structure
*
* Returned Value:
* OK on success; a negated errno on failure
*
* Assumptions:
* The initiator of the poll holds the priv->waitsem;
*
****************************************************************************/
static int slip_uiptxpoll(struct uip_driver_s *dev)
{
FAR struct slip_driver_s *priv = (FAR struct slip_driver_s *)dev->d_private;
/* If the polling resulted in data that should be sent out on the network,
* the field d_len is set to a value > 0.
*/
if (priv->dev.d_len > 0)
{
slip_transmit(priv);
}
/* If zero is returned, the polling will continue until all connections have
* been examined.
*/
return 0;
}
/****************************************************************************
* Function: slip_txtask
*
* Description:
* Polling and transmission is performed on tx thread.
*
* Parameters:
* arg - Reference to the NuttX driver state structure
*
* Returned Value:
* None
*
****************************************************************************/
static void slip_txtask(int argc, char *argv[])
{
FAR struct slip_driver_s *priv;
unsigned int index = *(argv[1]) - '0';
uip_lock_t flags;
ndbg("index: %d\n", index);
DEBUGASSERT(index < CONFIG_SLIP_NINTERFACES);
/* Get our private data structure instance and wake up the waiting
* initialization logic.
*/
priv = &g_slip[index];
slip_semgive(priv);
/* Loop forever */
for (;;)
{
/* Wait for the timeout to expire (or until we are signaled by by */
usleep(SLIP_WDDELAY);
/* Is the interface up? */
if (priv->bifup)
{
/* Get exclusive access to uIP (if it it is already being used
* slip_rxtask, then we have to wait).
*/
slip_semtake(priv);
/* Poll uIP for new XMIT data. BUG: We really need to calculate
* the number of hsecs! When we are awakened by slip_txavail, the
* number will be smaller; when we have to wait for the semaphore
* (above), it may be larger.
*/
flags = uip_lock();
priv->dev.d_buf = priv->txbuf;
(void)uip_timer(&priv->dev, slip_uiptxpoll, SLIP_POLLHSEC);
uip_unlock(flags);
slip_semgive(priv);
}
}
}
/****************************************************************************
* Function: slip_getc
*
* Description:
* Get one byte from the serial input.
*
* Parameters:
* priv - Reference to the driver state structure
*
* Returned Value:
* The returned byte
*
****************************************************************************/
static inline int slip_getc(FAR struct slip_driver_s *priv)
{
uint8_t ch;
while (read(priv->fd, &ch, 1) < 0)
{
DEBUGASSERT(errno == EINTR);
}
return (int)ch;
}
/****************************************************************************
* Function: slip_receive
*
* Description:
* Read a packet from the serial input
*
* Parameters:
* priv - Reference to the driver state structure
*
* Returned Value:
* None
*
****************************************************************************/
static inline void slip_receive(FAR struct slip_driver_s *priv)
{
uint8_t ch;
/* Copy the data data from the hardware to to the RX buffer until we
* put together a whole packet. Make sure not to copy them into the
* packet if we run out of room.
*/
nvdbg("Receiving packet\n");
for (;;)
{
/* Get the next character in the stream. */
ch = slip_getc(priv);
/* Handle bytestuffing if necessary */
switch (ch)
{
/* If it's an END character then we're done with the packet.
* (OR we are just starting a packet)
*/
case SLIP_END:
nvdbg("END\n");
/* A minor optimization: if there is no data in the packet, ignore
* it. This is meant to avoid bothering IP with all the empty
* packets generated by the duplicate END characters which are in
* turn sent to try to detect line noise.
*/
if (priv->rxlen > 0)
{
nvdbg("Received packet size %d\n", priv->rxlen);
return;
}
break;
/* if it's the same code as an ESC character, wait and get another
* character and then figure out what to store in the packet based
* on that.
*/
case SLIP_ESC:
nvdbg("ESC\n");
ch = slip_getc(priv);
/* if "ch" is not one of these two, then we have a protocol
* violation. The best bet seems to be to leave the byte alone
* and just stuff it into the packet
*/
switch (ch)
{
case SLIP_ESC_END:
nvdbg("ESC-END\n");
ch = SLIP_END;
break;
case SLIP_ESC_ESC:
nvdbg("ESC-ESC\n");
ch = SLIP_ESC;
break;
default:
ndbg("ERROR: Protocol violation: %02x\n", ch);
break;
}
/* Here we fall into the default handler and let it store the
* character for us
*/
default:
if (priv->rxlen < CONFIG_NET_BUFSIZE+2)
{
priv->rxbuf[priv->rxlen++] = ch;
}
break;
}
}
}
/****************************************************************************
* Function: slip_rxtask
*
* Description:
* Wait for incoming data.
*
* Parameters:
* argc
* argv
*
* Returned Value:
* (Does not return)
*
* Assumptions:
*
****************************************************************************/
static int slip_rxtask(int argc, char *argv[])
{
FAR struct slip_driver_s *priv;
unsigned int index = *(argv[1]) - '0';
uip_lock_t flags;
int ch;
ndbg("index: %d\n", index);
DEBUGASSERT(index < CONFIG_SLIP_NINTERFACES);
/* Get our private data structure instance and wake up the waiting
* initialization logic.
*/
priv = &g_slip[index];
slip_semgive(priv);
/* Loop forever */
for (;;)
{
/* Wait for the next character to be available on the input stream. */
nvdbg("Waiting...\n");
ch = slip_getc(priv);
/* Ignore any input that we receive before the interface is up. */
if (!priv->bifup)
{
continue;
}
/* We have something...
*
* END characters may appear at packet boundaries BEFORE as well as
* after the beginning of the packet. This is normal and expected.
*/
if (ch == SLIP_END)
{
priv->rxlen = 0;
}
/* Otherwise, we are in danger of being out-of-sync. Apparently the
* leading END character is optional. Let's try to continue.
*/
else
{
priv->rxbuf[0] = (uint8_t)ch;
priv->rxlen = 1;
}
/* Copy the data data from the hardware to priv->rxbuf until we put
* together a whole packet.
*/
slip_receive(priv);
SLIP_STAT(priv, received);
/* All packets are assumed to be IP packets (we don't have a choice..
* there is no Ethernet header containing the EtherType). So pass the
* received packet on for IP processing -- but only if it is big
* enough to hold an IP header.
*/
if (priv->rxlen >= UIP_IPH_LEN)
{
/* Handle the IP input. Get exclusive access to uIP. */
slip_semtake(priv);
priv->dev.d_buf = priv->rxbuf;
priv->dev.d_len = priv->rxlen;
flags = uip_lock();
uip_input(&priv->dev);
/* If the above function invocation resulted in data that should
* be sent out on the network, the field d_len will set to a
* value > 0. NOTE that we are transmitting using the RX buffer!
*/
if (priv->dev.d_len > 0)
{
slip_transmit(priv);
}
uip_unlock(flags);
slip_semgive(priv);
}
else
{
SLIP_STAT(priv, rxsmallpacket);
}
}
/* We won't get here */
return OK;
}
/****************************************************************************
* Function: slip_ifup
*
* Description:
* NuttX Callback: Bring up the Ethernet interface when an IP address is
* provided
*
* Parameters:
* dev - Reference to the NuttX driver state structure
*
* Returned Value:
* None
*
* Assumptions:
*
****************************************************************************/
static int slip_ifup(struct uip_driver_s *dev)
{
FAR struct slip_driver_s *priv = (FAR struct slip_driver_s *)dev->d_private;
ndbg("Bringing up: %d.%d.%d.%d\n",
dev->d_ipaddr & 0xff, (dev->d_ipaddr >> 8) & 0xff,
(dev->d_ipaddr >> 16) & 0xff, dev->d_ipaddr >> 24 );
/* Mark the interface up */
priv->bifup = true;
return OK;
}
/****************************************************************************
* Function: slip_ifdown
*
* Description:
* NuttX Callback: Stop the interface.
*
* Parameters:
* dev - Reference to the NuttX driver state structure
*
* Returned Value:
* None
*
* Assumptions:
*
****************************************************************************/
static int slip_ifdown(struct uip_driver_s *dev)
{
FAR struct slip_driver_s *priv = (FAR struct slip_driver_s *)dev->d_private;
/* Mark the device "down" */
priv->bifup = false;
return OK;
}
/****************************************************************************
* Function: slip_txavail
*
* Description:
* Driver callback invoked when new TX data is available. This is a
* stimulus perform an out-of-cycle poll and, thereby, reduce the TX
* latency.
*
* Parameters:
* dev - Reference to the NuttX driver state structure
*
* Returned Value:
* None
*
****************************************************************************/
static int slip_txavail(struct uip_driver_s *dev)
{
FAR struct slip_driver_s *priv = (FAR struct slip_driver_s *)dev->d_private;
/* Ignore the notification if the interface is not yet up */
if (priv->bifup)
{
/* Wake up the TX polling thread */
kill(priv->txpid, SIGALRM);
}
return OK;
}
/****************************************************************************
* Function: slip_addmac
*
* Description:
* NuttX Callback: Add the specified MAC address to the hardware multicast
* address filtering
*
* Parameters:
* dev - Reference to the NuttX driver state structure
* mac - The MAC address to be added
*
* Returned Value:
* None
*
* Assumptions:
*
****************************************************************************/
#ifdef CONFIG_NET_IGMP
static int slip_addmac(struct uip_driver_s *dev, FAR const uint8_t *mac)
{
FAR struct slip_driver_s *priv = (FAR struct slip_driver_s *)dev->d_private;
/* Add the MAC address to the hardware multicast routing table */
return OK;
}
#endif
/****************************************************************************
* Function: slip_rmmac
*
* Description:
* NuttX Callback: Remove the specified MAC address from the hardware multicast
* address filtering
*
* Parameters:
* dev - Reference to the NuttX driver state structure
* mac - The MAC address to be removed
*
* Returned Value:
* None
*
* Assumptions:
*
****************************************************************************/
#ifdef CONFIG_NET_IGMP
static int slip_rmmac(struct uip_driver_s *dev, FAR const uint8_t *mac)
{
FAR struct slip_driver_s *priv = (FAR struct slip_driver_s *)dev->d_private;
/* Add the MAC address to the hardware multicast routing table */
return OK;
}
#endif
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Function: slip_initialize
*
* Description:
* Instantiate a SLIP network interface.
*
* Parameters:
* intf - In the case where there are multiple SLIP interfaces, this value
* identifies which is to be initialized. The network name will be,
* for example, "/dev/slip5" for intf == 5
*
* Returned Value:
* OK on success; Negated errno on failure.
*
* Assumptions:
*
****************************************************************************/
int slip_initialize(int intf, const char *devname)
{
struct slip_driver_s *priv;
char buffer[8];
const char *argv[2];
/* Get the interface structure associated with this interface number. */
DEBUGASSERT(intf < CONFIG_SLIP_NINTERFACES);
priv = &g_slip[intf];
/* Initialize the driver structure */
memset(priv, 0, sizeof(struct slip_driver_s));
priv->dev.d_ifup = slip_ifup; /* I/F up (new IP address) callback */
priv->dev.d_ifdown = slip_ifdown; /* I/F down callback */
priv->dev.d_txavail = slip_txavail; /* New TX data callback */
#ifdef CONFIG_NET_IGMP
priv->dev.d_addmac = slip_addmac; /* Add multicast MAC address */
priv->dev.d_rmmac = slip_rmmac; /* Remove multicast MAC address */
#endif
priv->dev.d_private = priv; /* Used to recover private state from dev */
/* Open the device */
priv->fd = open(devname, O_RDWR, 0666);
if (priv->fd < 0)
{
ndbg("ERROR: Failed to open %s: %d\n", devname, errno);
return -errno;
}
/* Initialize the wait semaphore */
sem_init(&priv->waitsem, 0, 0);
/* Put the interface in the down state. This usually amounts to resetting
* the device and/or calling slip_ifdown().
*/
slip_ifdown(&priv->dev);
/* Start the SLIP receiver task */
snprintf(buffer, 8, "%d", intf);
argv[0] = buffer;
argv[1] = NULL;
#ifndef CONFIG_CUSTOM_STACK
priv->rxpid = task_create("rxslip", CONFIG_SLIP_DEFPRIO,
CONFIG_SLIP_STACKSIZE, (main_t)slip_rxtask, argv);
#else
priv->rxpid = task_create("rxslip", CONFIG_SLIP_DEFPRIO,
(main_t)slip_rxtask, argv);
#endif
if (priv->rxpid < 0)
{
ndbg("ERROR: Failed to start receiver task\n");
return -errno;
}
/* Wait and make sure that the receive task is started. */
slip_semtake(priv);
/* Start the SLIP transmitter task */
#ifndef CONFIG_CUSTOM_STACK
priv->txpid = task_create("txslip", CONFIG_SLIP_DEFPRIO,
CONFIG_SLIP_STACKSIZE, (main_t)slip_txtask, argv);
#else
priv->txpid = task_create("txslip", CONFIG_SLIP_DEFPRIO,
(main_t)slip_txtask, argv);
#endif
if (priv->txpid < 0)
{
ndbg("ERROR: Failed to start receiver task\n");
return -errno;
}
/* Wait and make sure that the transmit task is started. */
slip_semtake(priv);
/* Bump the semaphore count so that it can now be used as a mutex */
slip_semgive(priv);
/* Register the device with the OS so that socket IOCTLs can be performed */
(void)netdev_register(&priv->dev);
/* When the RX and TX tasks were created, the TTY file descriptor was
* dup'ed for each task. This task no longer needs the file descriptor
* and we can safely close it.
*/
close(priv->fd);
return OK;
}
#endif /* CONFIG_NET && CONFIG_NET_SLIP */