abc/lwip
1
0
Fork 0
mirror of https://git.savannah.nongnu.org/git/lwip.git synced 2025-12-12 01:36:39 +08:00
Code Issues Packages Projects Releases Wiki Activity
6,840 Commits 4 Branches 51 Tags
Commit Graph

985 Commits

Author SHA1 Message Date
goldsimon
4d6b90727f lwip_recvfrom: use LWIP_MIN instead of self-coded min 2017-04-20 22:07:08 +02:00
goldsimon
dbc969c139 Try to fix the socket API for sizeof(int) <= 2 by checking INT_MAX at 2 places 2017-04-20 22:06:36 +02:00
Dirk Ziegelmeier
d54240bdea One more try to fix compile without TCP 2017-04-19 08:44:05 +02:00
Dirk Ziegelmeier
302d84f5b2 Try to fix compile without TCP (reported by Axel Lin) 2017-04-19 08:18:15 +02:00
Tim Cussins
8e83e206f4 sockets.c: Make sock_inc_used overflow check robust. 
Before this patch, it was possible that the overflow check might
miss an overflow event.

e.g. Consider 2 threads, both executing this method. u8_t fd_used is on 255...

Thread A -> atomically increment fd_used (which is now 0)
Thread B -> atomically increment fd_used (which is now 1)
Thread A -> check overflow... sees everything ok
Thread B -> check overflow... sees everything ok

And the overflow is missed :(

Signed-off-by: goldsimon <goldsimon@gmx.de>
 2017-04-18 20:50:33 +02:00
Tim Cussins
7ac3056da9 sockets.c: Add missing lwip_select_dec_sockets_used within lwip_select. 
Signed-off-by: goldsimon <goldsimon@gmx.de>
 2017-04-18 20:50:33 +02:00
goldsimon
0952e618bd netconn_drain(): use lwip_netconn_is_err_msg() on acceptmbox, too. 2017-04-18 10:54:40 +02:00
Axel Lin
0b2a652317 api_msg: Use lwip_netconn_is_err_msg instead of NULL test in netconn_drain() 
The NULL test no longer work after commit e0a2472706ee, it needs to test with
lwip_netconn_is_err_msg() instead.

Fixes: e0a2472706ee ("netconn/sockets: remove fatal error handling, fix asynchronous error handling, ensure data before RST can be received")
Signed-off-by: Axel Lin <axel.lin@ingics.com>
 2017-04-18 10:47:42 +02:00
goldsimon
5310d8f13c event_callback: separated ++ from if 2017-04-12 12:40:50 +02:00
goldsimon
e0a2472706 netconn/sockets: remove fatal error handling, fix asynchronous error handling, ensure data before RST can be received 2017-04-12 12:37:16 +02:00
goldsimon
064044eeae sockets: speed up event_callback: don't iterate select_cb_list for NETCONN_EVT_RCVPLUS/SENDPLUS if the socket has been readable/writable before already 2017-04-12 09:12:55 +02:00
goldsimon
58c21eb13c sockets: speed up event_callback: don't iterate select_cb_list for NETCONN_EVT_RCVMINUS/SENDMINUS 2017-04-11 20:59:00 +02:00
goldsimon
b6fdb61654 sockets.c: refactor event_callback() a bit 2017-04-11 20:55:47 +02:00
goldsimon
b0af09ec45 select_cb_ctr no needed any more for LWIP_TCPIP_CORE_LOCKING==1 2017-04-11 12:50:25 +02:00
goldsimon
92997756d1 task #14420 (Remove sys_sem_signal from inside SYS_ARCH_PROTECT crit section) done for LWIP_TCPIP_CORE_LOCKING==1 2017-04-11 12:40:44 +02:00
goldsimon
856b49a057 lwip_sendmsg: fix check for unsupported flags 2017-03-29 22:38:16 +02:00
goldsimon
ca9f57c09f lwip_fcntl/F_GETFL: no need to check sock->conn != NULL: already checked by get_socket() 2017-03-29 22:26:34 +02:00
goldsimon
5b6c654dd1 make F_GETFL return file access mode for !CORE_LOCKING as well 2017-03-29 20:54:24 +02:00
Joan Lledó
172dab1289 lwip_fcntl() returns access modes 2017-03-29 20:46:30 +02:00
goldsimon
d71653f049 sockets.c: fix compiling in release mode 2017-03-23 10:44:28 +01:00
goldsimon
a49e040072 sockets: remove select_cb definitions #if !LWIP_SOCKET_SELECT 2017-03-23 08:34:56 +01:00
goldsimon
ba20cd229e api_msg:accept_function(): fix typo 2017-03-22 21:12:09 +01:00
Mikhail Lappo
eba1b971c0 Possible null-pointer dereference 
In assertion the pointer that is potentialy
null is dereferenced. The check for null was
located after.
 2017-03-22 15:50:34 +01:00
Dirk Ziegelmeier
9e20fe2cfb Apply [patch #9287] Possible null-pointer dereference from Mikhail Lappo in a modified way 
Moved the debug output to a location where we know that newpcb != NULL
Fixes possible NULL pointer dereferencing in debug message output
 2017-03-22 14:58:13 +01:00
goldsimon
d9a738d85f sockets: fix lwip_getsockname/lwip_getpeername for dual-stack: ip_addr_t type "any" (dual) has to be converted to AF_INET6 2017-03-17 08:54:51 +01:00
goldsimon
fc47f846ed Fix and improve sockets unit test and unit test sys_arch (with a little help of tcpip.c) 2017-03-16 22:49:38 +01:00
goldsimon
eb1aadb218 implement udp/raw recvmsg() by moving recvfrom() into a common function taking an array of iovecs... (IP_PKTINFO/in_pktinfo still missing) 2017-03-13 22:31:17 +01:00
goldsimon
cb1a271c61 lwip_recvmsg: MSG_PEEK must be limited to the first iov for TCP 2017-03-11 20:42:59 +01:00
goldsimon
5752b24d38 fix bug #50503: LWIP_NETCONN_FULLDUPLEX: some LWIP_ERROR paths don't call done_socket() 2017-03-10 11:52:59 +01:00
goldsimon
1b14c2e7b0 sockets: fix printf warning in gcc 2017-03-10 07:56:14 +01:00
goldsimon
2c77560870 My first try at 'recvmsg()', TCP only, for now... 2017-03-09 21:49:55 +01:00
Joel Cunningham
fd9ac30062 Fix dual-stack build failure in lwip_sendmsg 
This corrects a typo introduced in c9d0192b4aa5753a1a71fa23fc910dcd82d64eb9
that broke the build for dual-stack (IPv4 and IPv6)
 2017-03-08 16:34:49 -06:00
goldsimon
c797222407 lwip_sendmsg: implement EMSGSIZE checks 2017-03-07 21:48:59 +01:00
goldsimon
c9d0192b4a lwip_sendmsg: small performance improvement: netbuf can be allocated on the stack as it is used internally only (see lwip_sendto) 2017-03-07 21:43:23 +01:00
goldsimon
270fdfff07 netconn_write_vectors_partly() watch out for overflow of data to send (must fit into INT_MAX for sockets) 2017-03-07 21:30:03 +01:00
goldsimon
4dd378b126 socket sendto: gracefully handle 'size' not fitting into an u16_t (return EMSGSIZE error) see task #14378 2017-03-07 20:56:37 +01:00
goldsimon
5c33efe430 minor: indentation fix 2017-03-07 20:37:52 +01:00
goldsimon
194803a3a7 netbuf: correctly reset netbuf checksum for LWIP_CHECKSUM_ON_COPY==1 2017-03-07 20:37:06 +01:00
goldsimon
c6c693923e One more try to fix the build... 2017-03-07 09:45:01 +01:00
goldsimon
53f717338b sockets: fixed printf format (not reported by mingw port due to -Wno-format) 2017-03-07 08:59:26 +01:00
goldsimon
b71d4477ea socket tcp performance tweak: handle window update (call into tcpipi_thread) only once per recv for multi-segment receives 2017-03-06 22:22:22 +01:00
goldsimon
f0bc2fa968 netconn_tcp_recvd should take u32_t, not u16_t (as used in msg) 2017-03-06 22:20:41 +01:00
goldsimon
f02119af62 lwip_recvfrom: fix tcp socket error handling 2017-03-06 22:13:58 +01:00
goldsimon
04bff63f49 sockets: netconn event callback is only used for select (nonblocking sockets are now implemented at netconn layer) -> add option LWIP_SOCKET_SELECT to reflect this 2017-03-06 22:06:53 +01:00
goldsimon
c77a7fe824 lwip_recv_tcp: fix full-duplex: remove invalid calls to done_socket() 2017-03-06 22:03:02 +01:00
goldsimon
30be7b582e sockets: remove sock->lastoffset and free unused pbufs instead (using pbuf_free_header) 2017-03-06 21:54:50 +01:00
goldsimon
c295717ce7 sockets: clean up all the if/else in lwip_recvfrom and use new nonblocking features in netconn API: I'm OK with code duplication if it gets more readable :-/ 2017-03-06 21:50:42 +01:00
goldsimon
1945582c10 netconn: added receive flag NETCONN_NOAUTORCVD and netconn_tcp_recvd() to delay rx window updates (e.g. when receiving more data as performance improvement) 2017-03-06 21:33:35 +01:00
goldsimon
b86787c39c Improve code readability by moving tcp-specific functionality from netconn_recv_data() to new static function netconn_recv_data_tcp() since all tcp code paths are different anyway 2017-03-06 21:27:33 +01:00
Joel Cunningham
dd4ded3978 do_writemore: fix blocking bug 
A bug was introduced in the atomic vector feature for blocking netconns
where if we couldn't write the entire vector due to send buffer being
full (write_more is 0), we would not update the vector state and then
when sent_tcp() is called, it would actually re-send the previous vector
and if additional calls were required to finish the write, msg.w.offset
would eventually exceed msg.w.len, This was found by testing "stats"
from the shell and hitting the LWIP_ASSERT in do_writemore() that
checks offset < len

The fix simply updates the vector state after every ERR_OK return from
tcp_write().  While not all cases (non-blocking sockets) need to update
the state in this case, it keeps the logic simple and also makes
debugging simpler because you don't have stale vector state at any
point
 2017-03-04 12:34:44 -06:00