goldsimon
|
59a5fb7ce8
|
Fixed bug #28054: Two segments with FIN flag on the out-of-sequence queue, also fixed PBUF_POOL leak in the out-of-sequence code
|
2009-11-29 13:23:21 +00:00 |
|
goldsimon
|
c5d2e536cf
|
Worked on tcp_oos unit tests, nearly all TCP_QUEUE_OOS code is covered now
|
2009-11-29 13:20:13 +00:00 |
|
goldsimon
|
aeef0a21f3
|
Fixed bug #28064: pbuf_alloc(PBUF_POOL) is not thread-safe by queueing a call into tcpip_thread to free ooseq-bufs if the pool is empty
|
2009-11-29 11:57:35 +00:00 |
|
goldsimon
|
902ad897b8
|
Initial check-in of some unit tests
|
2009-11-27 08:03:53 +00:00 |
|
goldsimon
|
478ccee5fc
|
tcp_rexmit(): no need to call tcp_output, since always called from tcp_input and thus tcp_output always returns without sending (prevent confusion by thinking data was sent while in tcp_rexmit!)
|
2009-11-26 16:44:16 +00:00 |
|
goldsimon
|
43fd5c28b5
|
Fixed bug #28098: Nagle can prevent fast retransmit from sending segment by basing the nagle-decision on TF_INFR also
|
2009-11-26 16:42:13 +00:00 |
|
goldsimon
|
4391463832
|
Got the tcp_nagle_*() defines wrong (inverted) :-(
|
2009-11-26 15:51:29 +00:00 |
|
goldsimon
|
c8d2d2a8ea
|
Fixed bug #28099 (API required to disable Nagle algorithm at PCB level): added tcp_nagele_*() function-like macros
|
2009-11-26 15:19:30 +00:00 |
|
goldsimon
|
6795aabecf
|
Fixed filename in last comment
|
2009-11-22 17:57:19 +00:00 |
|
goldsimon
|
a939c09a6b
|
Fixed bug #27905: FIN isn't combined with data on unsent
|
2009-11-22 17:55:03 +00:00 |
|
goldsimon
|
35d1c33e0a
|
tcp_alloc: prevent increasing stats.err for MEMP_TCP_PCB when reusing time-wait pcb as suggested by Bill 4 months ago
|
2009-11-22 17:52:48 +00:00 |
|
goldsimon
|
10edf64873
|
Fixed bug #27851 (TCP_EVENT_RECV(pcb, NULL,...) results in unreachable code warning) by calling tcp_recv_null if pcb->recv is null.
|
2009-11-22 16:17:55 +00:00 |
|
goldsimon
|
6d22c38e59
|
tcp_recv_null: call tcp_recved() if p != NULL to keep the window correct
|
2009-11-22 16:16:55 +00:00 |
|
goldsimon
|
8a81cb4ba0
|
Fixed bug #27955: netconn_close may nether return when LWIP_TCPIP_CORE_LOCKING enabled
|
2009-11-22 16:08:52 +00:00 |
|
goldsimon
|
6bb7f987f2
|
Fixed bug #28049 (Keep Alive timeout frees pcb then uses it, sometimes causing crash error) by calling tcp_rst() instead of tcp_abort() in tcp_slowtmr() when retransmit timer expires.
|
2009-11-22 15:44:12 +00:00 |
|
goldsimon
|
1c018caefe
|
move tcp_debug_print_state to prevent accessing pcb when it might already be deallocated due to recv_flags == TF_RESET or TF_CLOSED
|
2009-11-22 15:42:30 +00:00 |
|
goldsimon
|
0c136893e5
|
Fixed logging of timeout-function-pointer (pointer to local stack was logged instead of actual function pointer), minor coding style fix
|
2009-11-22 15:25:13 +00:00 |
|
goldsimon
|
f4c0655190
|
Fixed usage of logging levels (bug #27948: Incorrect logging levels used in various places)
|
2009-11-22 15:14:46 +00:00 |
|
goldsimon
|
49bdf32765
|
Fixed usage of logging levels (bug #27948: Incorrect logging levels used in various places)
|
2009-11-22 13:31:31 +00:00 |
|
goldsimon
|
88d02a624a
|
Clarified debug levels by renaming level OFF to ALL (the old define still exists for now to prevent breaking old code)
|
2009-11-22 13:14:57 +00:00 |
|
goldsimon
|
98a58f9e67
|
tcp_input: move tcp_debug_print_state to prevent accessing pcb when it might already be deallocated due to calling tcp_abort in tcp_process.
|
2009-11-20 16:56:25 +00:00 |
|
goldsimon
|
b30c6f8b9e
|
Fixed comment: tcp_pcb_remove does *not* deallocate the pcb
|
2009-11-20 16:43:57 +00:00 |
|
goldsimon
|
94a7fee8c8
|
Fixed bug #28062: Data received directly after accepting does not wake up select
|
2009-11-20 16:16:20 +00:00 |
|
goldsimon
|
6f38b63a47
|
Fixed bug #27994: incorrect define for freeaddrinfo(addrinfo)
|
2009-11-11 16:52:05 +00:00 |
|
goldsimon
|
70b05c8096
|
Increased default value for TCP_MSS to 536, updated default value for TCP_WND to 4*TCP_MSS to keep delayed ACK working.
|
2009-10-30 10:02:06 +00:00 |
|
goldsimon
|
28ac4347d4
|
Fixed wrong filename
|
2009-10-29 17:10:19 +00:00 |
|
kieranm
|
dee1d82c11
|
re-work the fast retransmission code to follow algorithm from TCP/IP
Illustrated
|
2009-10-29 15:48:57 +00:00 |
|
kieranm
|
6d73f82f41
|
Update version for 1.3.2 rc 1
|
2009-10-28 15:17:16 +00:00 |
|
kieranm
|
71ddff4964
|
Fix BUG#27445: grow cwnd with every duplicate ACK
|
2009-10-28 15:13:51 +00:00 |
|
goldsimon
|
af3b796488
|
Corrected parameter spelling in doc
|
2009-10-27 20:30:44 +00:00 |
|
goldsimon
|
e2de2c6bb2
|
Added netifapi_netif_set_addr()
|
2009-10-27 20:29:16 +00:00 |
|
goldsimon
|
2ff0ce2d0a
|
Minor: Improved memory layout/alignment of struct dhcp for 32-bit platforms
|
2009-10-26 09:59:54 +00:00 |
|
goldsimon
|
b09b8a0ccc
|
Tiny code size improvement using goto instead of duplicating code
|
2009-10-26 09:55:46 +00:00 |
|
goldsimon
|
db259c3557
|
Improved heap usage of lwip_getaddrinfo by only allocating one block of memory per call.
|
2009-10-26 09:30:50 +00:00 |
|
goldsimon
|
bcc87ef851
|
bug-fix in the TCP_EVENT_RECV macro (has to call tcp_recved if pcb->recv is NULL to keep rcv_wnd correct)
|
2009-10-25 18:47:44 +00:00 |
|
goldsimon
|
650f16b6d9
|
Minor: fixed typo
|
2009-10-25 16:47:41 +00:00 |
|
goldsimon
|
ae2dd38e0d
|
Another fix for bug #26251: RST process in TIME_WAIT TCP state
|
2009-10-25 11:45:11 +00:00 |
|
kieranm
|
0319c1cf90
|
Update link to wiki in README file
|
2009-10-23 15:23:23 +00:00 |
|
goldsimon
|
81f9442ac7
|
Fixed bug #27783: Silly window avoidance for small window sizes
|
2009-10-23 13:17:18 +00:00 |
|
goldsimon
|
18ab274af3
|
Fixed bug #26251: RST process in TIME_WAIT TCP state
|
2009-10-22 13:37:44 +00:00 |
|
goldsimon
|
65d1f52423
|
Changed fix for bug #27215 (TCP sent() callback gives leagin and trailing 1 byte len (SYN/FIN)) by decreasing pcb->acked appropriately
|
2009-10-21 15:42:14 +00:00 |
|
goldsimon
|
f1b82e0e9a
|
bug #27315: zero window probe and FIN: only send pure FIN if the enqueued FIN segment doesn't contain any other data
|
2009-10-21 15:15:34 +00:00 |
|
goldsimon
|
67411c4299
|
Minor code simplification (don't store received pbuf, change conditional code to assert where applicable), check pbuf length before testing for valid reply
|
2009-10-19 20:06:01 +00:00 |
|
goldsimon
|
a37e62b7d0
|
Removed most calls to udp_connect since they aren't necessary when using udp_sendto_if() - always stay connected to IP_ADDR_ANY.
|
2009-10-19 16:43:50 +00:00 |
|
goldsimon
|
e2c1f7d5b5
|
dhcp_unfold_reply: NULL memory might have been freed after mem_malloc returned NULL
|
2009-10-18 09:26:27 +00:00 |
|
goldsimon
|
ec97fbd101
|
dhcp_inform(): netif->dhcp pointed to unallocated memory when udp_new() failed
|
2009-10-18 09:13:47 +00:00 |
|
goldsimon
|
b7d7559cc9
|
Fixed bug #27215: TCP sent() callback gives leadin and trailing 1 byte len (SYN/FIN)
|
2009-10-18 08:30:44 +00:00 |
|
goldsimon
|
b4404ff0c8
|
Corrected typo
|
2009-10-18 08:23:05 +00:00 |
|
goldsimon
|
502e89f4ad
|
Fixed bug #27315: zero window probe and FIN
|
2009-10-16 21:07:48 +00:00 |
|
goldsimon
|
d8d8cf7e98
|
Fixed bug #27390: Source IP check in ip_input() causes it to drop valid DHCP packets -> allow 0.0.0.0 as source address when LWIP_DHCP is enabled
|
2009-10-16 12:39:24 +00:00 |
|