sergiotarxz/nuttx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

net/inet/inet_close.c: Last PR remove SO_LINGER socket option. SO_LINGER was always enabled is CONFIG_NET_SOLINGER was selected.

Browse Source
This commit is contained in:
Gregory Nutt 2019-07-01 07:37:41 -06:00
parent ed9fe70024
commit 5af5fc4409
2 changed files with 24 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
TODO
View File

@ -1,4 +1,4 @@
NuttX TODO List (Last updated April 29, 2019)
NuttX TODO List (Last updated July 1, 2019)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
This file summarizes known NuttX bugs, limitations, inconsistencies with
@ -19,7 +19,7 @@ nuttx/:
(9) Kernel/Protected Build
(3) C++ Support
(5) Binary loaders (binfmt/)
(18) Network (net/, drivers/net)
(17) Network (net/, drivers/net)
(4) USB (drivers/usbdev, drivers/usbhost)
(2) Other drivers (drivers/)
(9) Libraries (libs/libc/, libs/libm/)
@ -1653,23 +1653,6 @@ o Network (net/, drivers/net)
before, so I suspect it might not be so prevalent as one
might expect.
Title: TCP SOCKETS CLOSED TOO QUICKLY
Description: When a socket is closed, the resources are torn down
immediately (unless the SO_LINGER option is selected). As a
result, the socket does not send the FIN and this looks like
an unexpected, abnormal loss of connection to the remote peer.
Actually, it is worse than this: The is NO logic to send
FIN in when the file is close. This is pretty easy to do,
however:
- Wait for a TCP poll, then
- Call tcp_append with TCP_CLOSE in the flags. There is
already logic in tcp_appsend to send the FIN in this case,
it is just not being use.
Status: Open
Priority: Medium-Low.
Title: LOCAL DATAGRAM RECVFROM RETURNS WRONG SENDER ADDRESS
Description: The recvfrom logic for local datagram sockets returns the
incorrect sender "from" address. Instead, it returns the

34
net/inet/inet_close.c
View File

@ -213,7 +213,6 @@ static uint16_t tcp_close_eventhandler(FAR struct net_driver_s *dev,
pstate->cl_result = -ETIMEDOUT;
goto end_wait;
}
#endif /* CONFIG_NET_SOLINGER */
#ifdef CONFIG_NET_TCP_WRITE_BUFFERS
@ -327,6 +326,7 @@ static inline int tcp_close_disconnect(FAR struct socket *psock)
FAR struct tcp_conn_s *conn;
#ifdef CONFIG_NET_SOLINGER
struct timespec abstime;
bool linger;
#endif
int ret = OK;
@ -400,21 +400,31 @@ static inline int tcp_close_disconnect(FAR struct socket *psock)
/* Wait for the disconnect event */
#ifdef CONFIG_NET_SOLINGER
DEBUGVERIFY(clock_gettime(CLOCK_REALTIME, &abstime));
/* A non-NULL value of the priv field means that lingering is
* enabled.
*/
/* NOTE: s_linger's unit is deciseconds,
* so we don't need to update abstime.tv_nsec here.
*/
abstime.tv_sec += psock->s_linger / DSEC_PER_SEC;
if (-ETIMEDOUT == net_timedwait(&state.cl_sem, &abstime))
linger = _SO_GETOPT(psock->s_options, SO_LINGER);
if (linger)
{
state.cl_result = -ETIMEDOUT;
DEBUGVERIFY(clock_gettime(CLOCK_REALTIME, &abstime));
/* NOTE: s_linger's unit is deciseconds,
* so we don't need to update abstime.tv_nsec here.
*/
abstime.tv_sec += psock->s_linger / DSEC_PER_SEC;
if (-ETIMEDOUT == net_timedwait(&state.cl_sem, &abstime))
{
state.cl_result = -ETIMEDOUT;
}
}
else
#else
(void)net_timedwait(&state.cl_sem, NULL);
#endif
{
(void)net_timedwait(&state.cl_sem, NULL);
}
/* We are now disconnected */