1 (0.5.3) Changes since version 0.5.2
5 * memp_malloc(MEMP_API_MSG) could fail with multiple application
6 threads because it wasn't protected by semaphores.
10 * struct ip_addr now packed.
12 * The name of the time variable in arp.c has been changed to ctime
13 to avoid conflicts with the time() function.
15 (0.5.2) Changes since version 0.5.1
19 * A new TCP function, tcp_tmr(), now handles both TCP timers.
23 * A bug in tcp_parseopt() could cause the stack to hang because of a
26 * The address of new connections in the accept() function in the BSD
27 socket library was not handled correctly.
29 * pbuf_dechain() did not update the ->tot_len field of the tail.
31 * Aborted TCP connections were not handled correctly in all
36 * All protocol header structs are now packed.
38 * The ->len field in the tcp_seg structure now counts the actual
39 amount of data, and does not add one for SYN and FIN segments.
41 (0.5.1) Changes since version 0.5.0
45 * Possible to run as a user process under Linux.
47 * Preliminary support for cross platform packed structs.
49 * ARP timer now implemented.
53 * TCP output queue length was badly initialized when opening
56 * TCP delayed ACKs were not sent correctly.
58 * Explicit initialization of BSS segment variables.
60 * read() in BSD socket library could drop data.
62 * Problems with memory alignment.
64 * Situations when all TCP buffers were used could lead to
67 * TCP MSS option wasn't parsed correctly.
69 * Problems with UDP checksum calculation.
71 * IP multicast address tests had endianess problems.
73 * ARP requests had wrong destination hardware address.
77 * struct eth_addr changed from u16_t[3] array to u8_t[6].
79 * A ->linkoutput() member was added to struct netif.
81 * TCP and UDP ->dest_* struct members where changed to ->remote_*.
83 * ntoh* macros are now null definitions for big endian CPUs.
85 (0.5.0) Changes since version 0.4.2
89 * Redesigned operating system emulation layer to make porting easier.
91 * Better control over TCP output buffers.
97 * Locking issues in buffer management.
99 * Bugfixes in the sequential API.
101 * IP forwarding could cause memory leakage. This has been fixed.
105 * Directory structure somewhat changed; the core/ tree has been
108 (0.4.2) Changes since version 0.4.1
112 * Experimental ARP implementation added.
114 * Skeleton Ethernet driver added.
116 * Experimental BSD socket API library added.
120 * In very intense situations, memory leakage could occur. This has
125 * Variables named "data" and "code" have been renamed in order to
126 avoid name conflicts in certain compilers.
128 * Variable++ have in appliciable cases been translated to ++variable
129 since some compilers generate better code in the latter case.
131 (0.4.1) Changes since version 0.4
135 * TCP: Connection attempts time out earlier than data
136 transmissions. Nagle algorithm implemented. Push flag set on the
137 last segment in a burst.
139 * UDP: experimental support for UDP-Lite extensions.
143 * TCP: out of order segments were in some cases handled incorrectly,
144 and this has now been fixed. Delayed acknowledgements was broken
145 in 0.4, has now been fixed. Binding to an address that is in use
146 now results in an error. Reset connections sometimes hung an
147 application; this has been fixed.
149 * Checksum calculation sometimes failed for chained pbufs with odd
150 lengths. This has been fixed.
152 * API: a lot of bug fixes in the API. The UDP API has been improved
153 and tested. Error reporting and handling has been
154 improved. Logical flaws and race conditions for incoming TCP
155 connections has been found and removed.
157 * Memory manager: alignment issues. Reallocating memory sometimes
158 failed, this has been fixed.
160 * Generic library: bcopy was flawed and has been fixed.
164 * API: all datatypes has been changed from generic ones such as
165 ints, to specified ones such as u16_t. Functions that return
166 errors now have the correct type (err_t).
168 * General: A lot of code cleaned up and debugging code removed. Many
169 portability issues have been fixed.
171 * The license was changed; the advertising clause was removed.
175 * Thanks: Huge thanks go to Dagan Galarneau, Horst Garnetzke, Petri
176 Kosunen, Mikael Caleres, and Frits Wilmink for reporting and
179 (0.4) Changes since version 0.3.1
181 * Memory management has been radically changed; instead of
182 allocating memory from a shared heap, memory for objects that are
183 rapidly allocated and deallocated is now kept in pools. Allocation
184 and deallocation from those memory pools is very fast. The shared
185 heap is still present but is used less frequently.
187 * The memory, memory pool, and packet buffer subsystems now support
188 4-, 2-, or 1-byte alignment.
190 * "Out of memory" situations are handled in a more robust way.
192 * Stack usage has been reduced.
194 * Easier configuration of lwIP parameters such as memory usage,
195 TTLs, statistics gathering, etc. All configuration parameters are
196 now kept in a single header file "lwipopts.h".
198 * The directory structure has been changed slightly so that all
199 architecture specific files are kept under the src/arch
202 * Error propagation has been improved, both in the protocol modules
205 * The code for the RTXC architecture has been implemented, tested
208 * Bugs have been found and corrected in the TCP, UDP, IP, API, and
209 the Internet checksum modules.
211 * Bugs related to porting between a 32-bit and a 16-bit architecture
212 have been found and corrected.
214 * The license has been changed slightly to conform more with the
215 original BSD license, including the advertisement clause.
217 (0.3.1) Changes since version 0.3
219 * Fix of a fatal bug in the buffer management. Pbufs with allocated
220 RAM never returned the RAM when the pbuf was deallocated.
222 * TCP congestion control, window updates and retransmissions did not
223 work correctly. This has now been fixed.
225 * Bugfixes in the API.
227 (0.3) Changes since version 0.2
229 * New and improved directory structure. All include files are now
230 kept in a dedicated include/ directory.
232 * The API now has proper error handling. A new function,
233 netconn_err(), now returns an error code for the connection in
236 * Improvements in the memory management subsystem. The system now
237 keeps a pointer to the lowest free memory block. A new function,
238 mem_malloc2() tries to allocate memory once, and if it fails tries
239 to free some memory and retry the allocation.
241 * Much testing has been done with limited memory
242 configurations. lwIP now does a better job when overloaded.
244 * Some bugfixes and improvements to the buffer (pbuf) subsystem.
246 * Many bugfixes in the TCP code:
248 - Fixed a bug in tcp_close().
250 - The TCP receive window was incorrectly closed when out of
251 sequence segments was received. This has been fixed.
253 - Connections are now timed-out of the FIN-WAIT-2 state.
255 - The initial congestion window could in some cases be too
256 large. This has been fixed.
258 - The retransmission queue could in some cases be screwed up. This
261 - TCP RST flag now handled correctly.
263 - Out of sequence data was in some cases never delivered to the
264 application. This has been fixed.
266 - Retransmitted segments now contain the correct acknowledgment
267 number and advertised window.
269 - TCP retransmission timeout backoffs are not correctly computed
270 (ala BSD). After a number of retransmissions, TCP now gives up
273 * TCP connections now are kept on three lists, one for active
274 connections, one for listening connections, and one for
275 connections that are in TIME-WAIT. This greatly speeds up the fast
276 timeout processing for sending delayed ACKs.
278 * TCP now provides proper feedback to the application when a
279 connection has been successfully set up.
281 * More comments have been added to the code. The code has also been
284 (0.2) Initial public release.