Add null pointer checks for TCP, UDP and netif

Signed-off-by: Simon Goldschmidt <goldsimon@gmx.de>

tcp: add LWIP_HOOK_TCP_INPACKET_PCB

This hook is called from tcp_input() for all kinds of input pcbs when
selected to receive a pbuf (LISTEN, TIME_WAIT, rest). I can parse or
drop an rx pbuf.

Signed-off-by: goldsimon <goldsimon@gmx.de>

tcp: introduce ext_args per pcb

This introduces the concept of ext (external/extended) arguments per
tcp_pcb (also for listening pcbs) to store more data than just one
"void *arg" per pcb. The "arg" is for use to applications, whereas
the ext_args may be used by frameworks and leave "arg" untouched.

In addition to a void pointer, callbacks are added to help frameworks
migrate arguments from listen pcb to connection pcb and to free args
when the pcb is freed.

Signed-off-by: goldsimon <goldsimon@gmx.de>

tcp: centralize freeing tcp pcbs to memp_free

This should make it easier to add debugging messages or other hooks
to the point where tcp pcbs are deallocated.

Signed-off-by: goldsimon <goldsimon@gmx.de>

Work on task #14780: Add debug helper asserts to ensure threading/locking requirements are met
Add LWIP_ASSERT_CORE_LOCKED() in several places

[PATCH] Replace the OOSEQ max bytes and pbufs constants with functions.

Some systems need to take into account an RX buffer pool size when
advising an appropriate number of RX pbufs to queue on the ooseq
list. For some systems there is a practical hard limit beyond which
the rx pool becomes exhausted blocking reception of further buffers
until some are freed.

It also helps to be able to consider the available dynamic memory when
advising an appropriate maximum number of bytes to buffer on the ooseq
list.

These decisions can also benefit from knowing the number already
allocated on a particular pcb, so the ooseq tcp segement is passed to
these functions. For example, if the system only wants to allow the
total number of rx pbufs queued on all the ooseq lists to grow by one
and a pcb already has two then it can return three for this call, but
might return one for another call - supporting a greedy allocation
strategy.

Signed-off-by: goldsimon <goldsimon@gmx.de>

tcp_in: favour local variables over ifdef blocks

Make TCP_OOSEQ_MAX_BYTES and TCP_OOSEQ_MAX_PBUFS work independently

tcp_in: favour code block over #ifdef blocks for local variables only used in TCP_OOSEQ_MAX_BYTES || TCP_OOSEQ_MAX_PBUFS configs

Reformat core code using astylerc

Fixed removing unneeded TCP SACKs

TCP SACKs were removed after some changes in the ooseq queue,
but before all unneeded packets were removed from it.
Because of that, we would sometimes include SACKs
for data already delivered in-order.

Signed-off-by: goldsimon <goldsimon@gmx.de>

Fix bug #51937 by checking TF_CLOSED at the end of all pcb callbacks in tcp_input()

refactor tcp_input a bit in preparation of a fix for bug #51937

tcp_in: correct some loops continuing without maintaining a prev pointer.

This problem would appear to have only affected systems with multiple
interfaces. It was noted causing tcp resets when the pcb was lost, and there
might have been other associated problems.

Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>

Fix bug #51765 and fix commit 90873d6: call tcp_rexmit_fast() more than once on dupacks since the actual transmission is guarded by TF_INFR, not by pcb->dupacks

tcp: persist timer re-work (bug #50837)

This re-works the persist timer to have the following behavior:

1) Only start persist timer when a buffered segment doesn't fit within
the current window and there is no in-fligh data. Previously, the
persist timer was always started when the window went to zero even
if there was no buffered data (since timer was managed in receive
pathway rather than transmit pathway)
2) Upon first fire of persist timer, fill the remaining window if
non-zero by splitting the unsent segment. If split segment is sent,
persist timer is stopped, RTO timer is now ensuring reliable window
updates
3) If window is already zero when persist timer fires, send 1 byte probe
4) Persist timer and zero window probe should only be active when the
following are true:
* no in-flight data (pcb->unacked == NULL)
* when there is buffered data (pcb->unsent != NULL)
* when pcb->unsent->len > pcb->snd_wnd

TCP: simplify trimming left edge & use pbuf_remove_header() instead of pbuf_header()

tcp: use TCP_WND_INC for dupack > 3 case

TCP_WND_INC abstracts the rollover handling and allows the window to
reach it's maximum value

tcp: use tcp_set_flags() rather than manually setting flag

tcp: use tcp_ack_now rather than manually setting TF_ACK_NOW

Continue fixing the spirit of bug #51663: don't change pcb state if retransmission can't be done because segments are still queued for transmission

- add a better-documented static function tcp_output_segment_busy
- try to reduce the number of checks
- tcp_rexmit_rto: iterate pcb->unacked only once
- no need to check for ref==1 in tcp_rexmit_fast when tcp_rexmit does
- call tcp_rexmit_fast if dupacks >= 3 (not == 3) and use TF_INFR flag to guard the fast-rexmit case (that way, it's triggered again on the next dupack)

task #14597: cleanup pbuf_header usages (use pbuf_add_header/pbuf_remove_header instead)

tcp_getoptbyte() -> tcp_get_next_optbyte() (to make it clearer that this function modifies something - not only a getter)

Fixed bug #51528 (Bug when parsing WND_SCALE option)

Signed-off-by: goldsimon <goldsimon@gmx.de>

Fix two -Wconversion warnings

-Wconversion (still far from finished) and other minor compilation fixes...

remove invalid TODO comment

Fixed LWIP_TCP_SACK_VALID and its usage

Cleaned up a bit after merging SACK_OUT support (mostly coding style)

Partial SACK (RFC 2018) support

Adds partial support for selective acknowledgements (RFC 2018).
This change makes lwIP negotiate SACK support, and include SACK
data in outgoing empty ACK packets. It does not include it
in outgoing packets with data payload.
It also does not add support for handling incoming SACKs.

Signed-off-by: goldsimon <goldsimon@gmx.de>

work on -Wconversion...

Silence some unsed variable warnings.

tcp: fix cwnd rollover introduced by ABC

Changes for TCP Appropriate Byte Counting introduce a potential cwnd
rollover by not taking into account integer promotion on tcpwnd_size_t
during inequality comparisions

This fixes the issue by introducing a macro TCP_WND_INC which detects
the rollover correctly and now holds the tcpwnd_size_t at the maximum
value rather than not incrementing the window. This provides a slight
performance improvement by allowing full use of the tcpwnd_size_t number
space for the congestion window

bug #50837: add zero-window probe timeout

This commit adds a timeout to the zero-window probing (persist timer)
mechanism. LwIP has not historically had a timeout for the persist
timer, leading to unbounded blocking if connection drops during the
zero-window condition

This commit also adds two units test, one to check the RTO timeout
and a second to check the zero-window probe timeout

tcp: indentation fix in tcp_receive()

TCP_OVERSIZE block was moved out from inside the while loop and thus
should be on column 7

tcp_free_acked_segments: fix debug printf parameter errors, fix unused parameters for !LWIP_DEBUG and/or LWIP_NOASSERT

task #14433: TCP: combine unsent/unacked queue loops in tcp_receive()

Work on task #14494: Implement SO_BINDTODEVICE
TCP needs additional netif handling

Fixed copy&paste error found by travis-ci...

Start working task #14494: Implement SO_BINDTODEVICE
Implement binding TCP, UDP and RAW PCBs to a netif

doc: clarify RST handling comment

This clarifies the documentation around RST handling to document
the approach follows RFC 5961 and is not a standards violation

tcp: task #14128 - Appropriate Byte Counting support

This commit adds TCP Appropriate Byte Counting (ABC) support based on
RFC 3465

ABC replaces the previous congestion window growth mechanism and has been
configured with limit of 2 SMSS. See task #14128 for discussion on
defaults, but the goal is to mitigate the performance impact of delayed
ACKs on congestion window growth

This commit also introduces a mechanism to track when the stack is
undergoing a period following an RTO where data is being retransmitted.

Lastly, this adds a unit test to verify RTO period tracking and some
basic ABC cwnd checking

tcp: bug #50614: move unsent queue check/pcb->snd_buf update

This commit corrects what looks like an ancient incorrect organization
of the logic for processing an ACK which acks new data. Once moved,
we can also change to using TCP_SEQ_LEQ on ackno instead of TCP_BETWEEN
because ackno has already been checked against snd_nxt

The work of checking the unsent queue and updating pcb->snd_buf (both
steps required for new data ACK) should be located under the conditional
that checks TCP_SEQ_BETWEEN(ackno, pcb->lastack+1, pcb->snd_nxt)

The comment following the unsent queue check/pcb->snd_buf update even
indicates "End of ACK for new data processing" when the logic is clearly
outside of this check

From what I can tell, this mis-organization isn't causing any incorrect
behavior since the unsent queue checked that ackno was between start of
segment and snd_nxt and recv_acked would be 0 during pcb->snd_buf update.
Instead this is waisted work for duplicate ACKS (can be common) and other
old ACKs

tcp: do not keep sending SYNs when getting ACKs

If a locally generated TCP SYN packet is replied to with an ACK
packet, lwIP immediately sends a RST packet followed by resending the
SYN packet. This is expected, but on loopback interfaces the resent
SYN packet may immediately get another ACK reply, typically when the
other endpoint is in TIME_WAIT state (which ignores the RSTs). The
result is an endless loop of SYN, ACK, RST packets.

This patch applies the normal SYN retransmission limit in this
scenario, such that the endless loop is limited to a brief storm.

bug #50476: initialize ssthresh to TCP_SND_BUF

This commit changes ssthresh to be the largest effective congestion
window (amount of in-flight data). This follows the guidance of RFC
5681 which recommends setting ssthresh arbitrarily high.

LwIP was previously using the receive window value at the end of the
3-way handshake and in the case of an active open where the receiver
used window scaling and/or window auto-tuning, this resulted in a very
small ssthresh value even though the window ramped up once the connection
was established

tcp: fixed bug #50418: LWIP_EVENT_API: fix invalid calbacks for SYN_RCVD pcb

minor: move assert to a slightly better place

tcp: fix accept event on closed listening PCBs

If LWIP_CALLBACK_API is not defined, but TCP_LISTEN_BACKLOG is, then
the LWIP_EVENT_ACCEPT TCP event may be triggered for closed listening
sockets. This case is just as disastrous for the event API as it is
for the callback API, as there is no way for the event hook to tell
whether the listening PCB is still around. Add the same protection
against this case for TCP_LISTEN_BACKLOG as was already in place for
LWIP_CALLBACK_API.

Also remove one NULL check for LWIP_CALLBACK_API that had already
become redundant for all callers, making the TCP_EVENT_ACCEPT code
for that callback wrapper more in line with the rest of the wrappers.

Add some parentheses for better code readability in tcp_in.c

bug #49631: handle zero-window probe and refused_data

This commit adds support for responding to a zero-window probe when
the refused_data pointer is set

A zero-window probe is a data segment received when rcv_ann_wnd
is 0. This corrects a standards violation where LwIP would not
respond to a zero-window probe with its current ACK value (RCV.NXT)
when it has refused data, thus leading to the probing TCP closing
out the connection

Add hook for TCP Initial Sequence Number generation

lwIP produces a TCP Initial Sequence Number (ISN) for each new TCP
connection. The current algorithm is simple and predictable however.
The result is that lwIP TCP connections may be the target of TCP
spoofing attacks. The problem of such attacks is well known, and a
recommended ISN generation algorithm is standardized in RFC 6528.
This algorithm requires a high-resolution timer and cryptographic
hashing function, though. The implementation (or best-effort
approximation) of both of these aspects is well beyond the scope of
lwIP itself.

For that reason, this patch adds LWIP_HOOK_TCP_ISN, a hook that
allows each platform to implement its own ISN generation using
locally available means. The hook provides full flexibility, in
that the hook may generate anything from a simple random number
(by being set to LWIP_RAND()) to a full RFC 6528 implementation.

Implementation note:

Users of the hook would typically declare the function prototype of
the hook function in arch/cc.h, as this is the last place where such
prototypes can be supplied. However, at that point, the ip_addr_t
type has not yet been defined. For that reason, this patch removes
the leading underscore from "struct _ip_addr", so that a prototype
of the hook function can use "struct ip_addr" instead of "ip_addr_t".

Signed-off-by: sg <goldsimon@gmx.de>

patch by Ambroz Bizjak: fixed bug #49717 (window size in received SYN and SYN-ACK assumed scaled)

Make lwIP compile with clang -Wdocumentation -> several documentation fixes

Cleanup hton*/ntoh* function handling and platform abstraction
Let lwip use functions/macros prefixed by lwip_ internally to avoid naming clashes with external #includes.
Remove over-complicated #define handling in def.h
Make functions easier to override in cc.h. The following is sufficient now (no more LWIP_PLATFORM_BYTESWAP):
#define lwip_htons(x) <your_htons>
#define lwip_htonl(x) <your_htonl>

tcp: fix FIN ACK handling with unsent data

TCP's snd_nxt represents the next sequence number after sent data, and
as such does not cover any unsent data queued on the connection. The
current implementation does not take the latter point into account
when processing FIN acknowledgments, mistakenly assuming that an
outgoing FIN is ACK'ed when the acknowledgment covers up to snd_nxt
while there is still unsent data. This patch adds a check for unsent
data to correct this, effectively preventing that TCP connections are
closed prematurely.

Remove duplicated include for lwip/inet_chksum.h

Include it once is enough.

Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>

fixed compiling TCP_OOSEQ_MAX_BYTES || TCP_OOSEQ_MAX_PBUFS (local variable 'p' was hidden)

minor: whitespace cleanups

fixed bug #48543 (TCP sent callback may prematurely report sent data when only part of a segment is acked) and don't include SYN/FIN in snd_buf counter (patch by Ambroz Bizjak)

Fixed bug #48539 (possible crash when packet received in SYN_SENT state)

change tcp_pcb->acked to be a global variable: used in one call stack only (idea by Ambroz Bizjak)

fixed bug #48476 (TCP sent callback called wrongly due to picking up old pcb->acked

fixed bug #48402 (Bug in skipping over TCP options)

Signed-off-by: goldsimon <goldsimon@gmx.de>

tcp: Avoid 3-second delay for a half-open connection

If lwIP encounters a half-open connection (e.g. due to a restarted
application reusing the same port numbers) it will correctly send a
RST but will not resend the SYN until one retransmission timeout later
(approximately three seconds). This can increase the time taken by
lpxelinux.0 to fetch its configuration file from a few milliseconds to
around 30 seconds.

Fix by immediately retransmitting the SYN whenever a half-open
connection is detected.

Signed-off-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: goldsimon <goldsimon@gmx.de>

fixed bug #48170 (Vulnerable to TCP RST spoofing) (original patch by Fabian Koch)

Revert "fixed bug #48170 (patch by Fabian Koch)"

This reverts commit 236bc194222b604f251cd86d4715cd05d9e40f94.
The fix introduced bug #48328 -> reverted

fixed bug #48170 (patch by Fabian Koch)

Fix several CLANG format string warnings

some minor improvements to the tcp options checking code

[PATCH] Drop instead of ASSERT in tcp_input header parsing

Since allowing input validation to trip the ASSERT handler is bad,
let's just drop the packets instead if validation fails.

Signed-off-by: sg <goldsimon@gmx.de>

Fixed possible problems with tcp_backlog_delayed/tcp_backlog_accepted

removed unused return values of tcp_listen_input & tcp_timewait_input; fixed unused variable warning

call accept-callback with ERR_MEM when allocating a pcb fails on passive open to inform the application about this error; ATTENTION: applications have to handle NULL pcb in accept callback!

tcp: changed accept handling to be done internally: the application does not have to call tcp_accepted() any more. Instead, when delaying accept (e.g. sockets do), call tcp_backlog_delayed()/tcp_backlog_accepted() (fixes bug #46696)

TCP: Implement dual stack support using new IP_ANY_TYPE introduced at UDP

Apply patch from Simon to fix buffer overflow handling tcp_in.c which was discovered by Coverity

minor: Cleanups in raw/tcp/udp code by using macros and reducing #ifdefs

Change signature of ip_output, ip_output_if, ip_output_if_src, ip_output_hinted, ip_route, ip_netif_get_local_ip, ip_route_get_local_ip and tcp_eff_send_mss not to take an isipv6 parameter. Use the IP version of the destination address instead.

Together with Simon: Eliminate last instances of isipv6 member in pcbs;
Don't allow NULL pointers (IPv4 ANY) in connect() and sendto() functions as destinations since this does not make sense.

Reduce PCB_ISIPV6 macro usage some more

Eliminate IP_PCB_IPVER_INPUT_MATCH macro

Reduce usage of PCB IP version flag in raw and tcp code. Maybe we can figure out a good way for dual-stack UDP when we manage to remove the flag (nearly) entirely from the code.

Drop #if TCP_DEBUG guard around tcp_debug_print_state calls

The #if TCP_DEBUG guard can be removed because tcp_debug_print_state()
print nothing if !TCP_DEBUG.

Signed-off-by: Axel Lin <axel.lin@ingics.com>

started to move "private" header files containing implementation details to "lwip/priv/" include directory to seperate the API from the implementation.

minor: fixed coding style (lwip style)

minor: coding style

minor/coding style: removed spaces before line ending (from file header)

minor/coding style: removed spaces before line ending

fixed some printf formatters (mainly for window scaling code)

Hook up TCP cachehit stat

This commit hooks up the TCP cachehit stat to the PCB locality feature
so that when a PCB is moved to the head of the list and a segment comes
in, we consider this a cache hit

This also matches the usage of the cachehit stat in UDP

Separate mib2 counter/table callbacks from snmp agent. This both cleans up the code and should allow integration of a 3rd party agent/mib2.

task #12178: hardware checksum capabilities can be configured per netif (use NETIF_SET_CHECKSUM_CTRL() in your netif's init function)

fixed bug #44023: TCP ssthresh value is unclear: ssthresh is set to the full send window for active open, too, and is updated once after SYN to ensure the correct send window is used

TCP window scaling: don't parse window scaling option on retransmission

fixed bug #45559: Window scaling casts u32_t to u16_t without checks

fixed bug #45120: Broadcast & multiple interfaces handling

Fixed ~ warnings in a different way

Fixed warnings about NULL check not required (ip_addr_isany) and implicit conversion (~)

fixed some more missing casts...

fixed compiler warning (signed/unsigned comparison) introduced when fixing bug #44766

task #12722 (improve IPv4/v6 address handling): renamed ip_addr_t to ip4_addr_t, renamed ipX_addr_t to ip_addr_t and added IP version;
ip_addr_t is used for all generic IP addresses for the API, ip(4/6)_addr_t are only used internally or when initializing netifs or when calling version-related functions

fixed bug #44766 (LWIP_WND_SCALE: tcphdr->wnd was not scaled in two places)

fixed bug #41318 (Bad memory ref in tcp_input() after tcp_close())

fixed bug #38468 (tcp_sent() not called on half-open connection for data ACKed with the same ack as FIN)

tcp_in: typo fix. Fixes bug #4458944589.

tcp: Fix ooseq processing when seqno is near 2^32

An edge case in out-of-sequence TCP packet processing did not use the
appropriate macro to check if the sender overran the recieve window.

Consequently, this case sometimes evaluated to true when it shouldn't, which
resulted in various bad behavior, including trying to resize a TCP buffer to 4
gigabytes.

patch #8423 "arch/perf.h" should be made an optional item

fixed bug #20506 "Initial congestion window is very small" again by implementing the calculation formula from RFC3390

Parts of patch #8397 Typos corrected in comments and text outputs

fixed bug #37969 SYN packet dropped as short packet in tcp_input function

Fixed bug #36210 lwIP does not elicit an empty ACK when received unacceptable ACK

Fixed typo.

Another fix to window scaling: support queueing more than 64 KByte in ooseq data

Added comments to clarify the LWIP_TCP_TIMESTAMPS option.

TCP window scaling: change tcp_pcb::acked to u32_t when window scaling is enabled to prevent an overflow when more than 64 Kbytes are ACKed

patch #6537/#7858: TCP window scaling support (OOS queueing still needs special handling)

partly fixed bug #25882: TCP hangs on MSS > pcb->snd_wnd (by not creating segments bigger than half the window)

fixed bug #35305: pcb may be freed too early on shutdown(WR)

fixed bug #31177: tcp timers can corrupt tcp_active_pcbs in some cases

fixed bug #34638: Dead code in tcp_receive - pcb->dupacks

- moved processing of refused_data to an own function (used from tcp_fasttmr and tcp_input);
- improved readability of tcp_slowtmr by using defines to access keepalive variables

fixed bug #34517 (persist timer is started although no zero window is received) by starting the persist timer when a zero window is received, not when we have more data queued for sending than fits into the window

slightly rearranged freeing an acked segment to prevent keeping the reference too long

Tried to fix bug #32417 ("TCP_OVERSIZE seems to have problems with (fast-)retransmission"): Reset pcb->unsent_oversize in 2 more places...

Implemented limiting data on ooseq queue (task #9989) (define TCP_OOSEQ_MAX_BYTES / TCP_OOSEQ_MAX_PBUFS in lwipopts.h), added unit test for this new feature

fixed bug #28288: Data after FIN in oos queue

fixed bug #33871: rejecting TCP_EVENT_RECV() for the last packet including FIN can lose data

fixed bug #33952 PUSH flag in incoming packet is lost when packet is aggregated and sent to application

fixed bug #34111 RST for ACK to listening pcb has wrong seqno

Fixed complier error for CHECKSUM_CHECK_TCP==0

use PCB_IS_IPV6(pcb) instead of pcb->isipv6 everywhere

Combined IPv4 and IPv6 code where possible, added defines to access IPv4/IPv6 in non-IP code so that the code is more readable.

... and finally, we got a first working version of a dual-stack lwIP runnin IPv4 and IPv6 in parallel - big thanks to Ivan Delamer! (this is work in progress, so please beware, test a lot and report problems!)

Removed 'dataptr' from 'struct tcp_seg' and calculate it in tcp_zero_window_probe (the only place where it was used).

Mreged back changes that were lost during the savannah hack 3 weeks ago (using the sources from http://git.infradead.org/users/dwmw2/lwip.git)

bug #27352: removed packing from ip_addr_t, the packed version is now only used in protocol headers. Added global storage for current src/dest IP address while in input functions.

Free recv_data when TF_RXCLOSED is set

Let FIN come through although TF_RXCLOSED is set, send RST when data is received although TF_RXCLOSED is set, added TCP_EVENT_CLOSED for clearer code

SO_REUSE: tcp_input: correctly handle multiple pcbs listening on the same port (but different address): first search for a specific address an only pass to ANY if no specific address has been found listening

Remove uused SOF_* flags, define inherited flags in ip.h, not in tcp_in.c

Fixed bug #29080: Correctly handle remote side overrunning our rcv_wnd in ooseq case.

task #7040 (Work on tcp_enqueue): Don't waste memory when chaining segments, added option TCP_OVERSIZE to prevent creating many small pbufs when calling tcp_write with many small blocks of data. Instead, pbufs are allocated larger than needed and the space is used for later calls to tcp_write.

Added comment after #endif

Moved stack-internal parts of tcp.h to tcp_impl.h, tcp.h now only contains the actual application programmer's API

Task #10088: Correctly implement close() vs. shutdown(). Now the application does not get any more recv callbacks after calling tcp_close(). Added tcp_shutdown().

Use new macro ip_addr_copy where applicable

Fixed wrong printf formatter

Replaced struct ip_addr by typedef ip_addr_t to make changing the actual implementation behind the typedef easier.

Cleanly separate the portability file inet.h and its contents from the stack: moved htonX- functions to def.h (and the new def.c - they are not ipv4 dependent), let inet.h depend on ip_addr.h and not the other way round. This fixes bug #28732.

Fixed bug #27871: Calling tcp_abort() in recv callback can lead to accessing unallocated memory. As a consequence, ERR_ABRT means the application has called tcp_abort()!

bug #28659: Missing casts

Added comments

Another fix for bug #28241 (ooseq processing) and adapted corresponding unit test

BUG28241: improve oos sequence processing with patches from Oleg Tyshev.

tcp_process: add state info to log message that is used for multiple states ("TCP connection closed")

Fixed bug #28106: dup ack for fast retransmit could have non-zero length

Fixed bug #27904: TCP sends too many ACKs: delay resetting tcp_input_pcb until after calling the pcb's callbacks

tcp_receive(): removed unnecessary return value, added comment

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

move tcp_debug_print_state to prevent accessing pcb when it might already be deallocated due to recv_flags == TF_RESET or TF_CLOSED

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.

re-work the fast retransmission code to follow algorithm from TCP/IP
Illustrated

Fix BUG#27445: grow cwnd with every duplicate ACK

Another fix for bug #26251: RST process in TIME_WAIT TCP state

Fixed bug #26251: RST process in TIME_WAIT TCP state

Changed fix for bug #27215 (TCP sent() callback gives leagin and trailing 1 byte len (SYN/FIN)) by decreasing pcb->acked appropriately

Fixed bug #27215: TCP sent() callback gives leadin and trailing 1 byte len (SYN/FIN)

Fixed bug #27329: dupacks by unidirectional data transmit

BUG27209: handle trimming of segments when out of window or out of
order properly

BUG27199: use snd_wl2 instead of snd_wl1

Added debug print when rejecting incoming connections due to exceeded listen backlog

BUG26301 and BUG26267: correct simultaneous close behaviour, and make
snd_nxt have the same meaning as in the RFCs.

Patch #6823: tcp_in.c - small optimization

tcp_listen_input: format code for #if LWIP_TCP_TIMESTAMPS that calls tcp_enqueue like the other calls.

BUG20515: rework way TCP window updates are calculated and sent

TASK9218: add support for TCP timestamp options

BUG20779: cope with SYN packets received during established states,
and retransmission of initial SYN.

BUG25622: handle return code of tcp_enqueue in tcp_listen_input()

patch #6699: fixed some warnings on platform where sizeof(int) == 2

fixed bug ##24596: Vulnerability on faulty TCP options length in tcp_parseopt

* tcp_in.c: Fix for bug #23693 as suggested by Art R. Ensure cseg is unused
if tcp_seg_copy fails.

Remove redundant "if" statement, and use real rcv_wnd
rather than rcv_ann_wnd when deciding if packets are in-window.
Contributed by <arasmussen@consultant.datasys.swri.edu>

fixed bug #22249: division by zero could occur if a remote host sent a zero mss as TCP option.

tcp.c, tcp_in.c, tcp.h, opt.h: Rename LWIP_CALCULATE_EFF_SEND_MSS in TCP_CALCULATE_EFF_SEND_MSS to have coherent TCP options names.

rawapi.txt, api_msg.c, tcp.c, tcp_in.c, tcp.h: changes for task #7675 "Enable to refuse data on a TCP_EVENT_RECV call". Important, behavior changes for the tcp_recv callback (see rawapi.txt).

rawapi.txt, api.h, api_lib.c, api_msg.h, api_msg.c, sockets.c, tcp.h, tcp.c, tcp_in.c, init.c, opt.h: rename backlog options with TCP_ prefix, limit the "backlog" parameter in an u8_t, 0 is interpreted as "smallest queue", add documentation in the rawapi.txt file.

Add persist timer

tcp.h, opt.h, api.h, api_msg.h, tcp.c, tcp_in.c, api_lib.c, api_msg.c, sockets.c, init.c: task #7252: Implement TCP listen backlog: Warning: raw API applications have to call 'tcp_accepted(pcb)' in their accept callback to keep accepting new connections.

2007-12-20 Kieran Mansley (based on patch from Oleg Tyshev)
* tcp_in.c: fix bug# 21535 (nrtx not reset correctly in SYN_SENT stat)e

2007-12-20 Kieran Mansley (based on patch from Per-Henrik Lundbolm)
* tcp.c, tcp_in.c, tcp_out.c, tcp.h: fix bug #20199 (better handling
of silly window avoidance and prevent lwIP from shrinking the window)

Removed debug code

Fix bug #21699: segment leak in ooseq processing when last data packet was lost (free last segment if FIN is already received)

fix bug #21699 (segment leak in ooseq): add assert that all segment lists are empty in tcp_pcb_remove before setting pcb to CLOSED state; don't directly set CLOSED state from LAST_ACK in tcp_process

Fixed bug #21582: pcb->acked accounting can be wrong when ACKs arrive out-of-order

Implemented calculating the effective send-mss based on the MTU of the netif used to send. Enabled by default. Disable by setting LWIP_CALCULATE_EFF_SEND_MSS to 0. This fixes bug #21535.

Moved inet files into ipv4/ipv6 directory; splitted inet.c/inet.h into inet and chksum part; changed includes in all lwIP files as appropriate

Fixed bug #21494: The send mss (pcb->mss) is set to 536 (or TCP_MSS if that is smaller) as long as no MSS option is received from the remote host.

Fixed bug #21491: The MSS option sent (with SYN) is now based on TCP_MSS instead of pcb->mss (on passive open now effectively sending our configured TCP_MSS instead of the one received).

Fixed bug #21181: On active open, the initial congestion window was calculated based on the configured TCP_MSS, not on the MSS option received with SYN+ACK.

Added some explaining comments.

Minor changes (but in lot of files): add #if/#endif for options where they could miss. #if LWIP_xxx if always put after #include "lwip/opt.h" (note this one indirectly include cc.h). Move others includes inside #if/#endif block.

Rename pbuf's "flgs" in "flags" (see in [lwip-devel] Last changes by Marc)

Minor changes (tabs, ident, coding style...)

Split pbuf flags in pbuf type and flgs.
Improved lwip_recvfrom(). TCP push now propagated.

* tcp_in.c: Fix for bug #19953 - correct TCP state machine when
* receiving FIN+ACK in SYN_RECV state

Minor change: add in tcp_in.c in tcp_input(), some calls to handle error statistics (stats & snmp)

Comments Fix for Doxygen documentation

Minor fix (warnings on unused args, wrong type, cast u32_t to u16_t...)

Fix bug #20506: Slow start / initial congestion window starts with 2 * mss (instead of 1 * mss previously) to comply with some newer RFCs and other stacks.

Fix minor patch #5741 build error...

2007-07-13 Kieran Mansley
* src/core/tcp_in.c Apply patch#5741 from Oleg Tyshev to fix bug in
out of sequence processing of received packets

Added some ASSERTS and casts to prevent warnings when assigning to smaller types.

Changed tcp_pcb->snd_queuelen from u8_t to u16_t to prevent overflowing when sending many small packets with big send buffer, added assertions and oveflow checks for snd_queuelen.

Removed warning "empty body in an if statement" in tcp_receive() for LWIP_DEBUG=0

tcp_in.c: Fix bug #20126 : Zero ssthresh bug (by Per-Henrik Lundblom and Kieran Mansley).

Done some work on task #1549 (function documentation)

* tcp.c, tcp_in.c, tcp_out.c, tcp.h: Modify way the retransmission
timer is reset to fix bug#19434, with help from Oleg Tyshev.

* Fix all uses of pbuf_header to check the return value. In some
cases just assert if it fails as I'm not sure how to fix them, but
this is no worse than before when they would carry on regardless
of the failure.

Keepalive values can be configured at run time with LWIP_TCP_KEEPALIVE, but don't change this unless you know what you're doing (default are RFC1122 compliant). Note that TCP_KEEPIDLE and TCP_KEEPINTVL have to be set in seconds.

Old TCP_KEEPALIVE can always be used, and its only difference with TCP_KEEPIDLE is this last use seconds and not milliseconds with setsockopt.

Some documentation (not perfect, but...):

http://publib.boulder.ibm.com/infocenter/pseries/v5r3/index.jsp?topic=/com.ibm.aix.commtechref/doc/commtrf2/setsockopt.htm

Added SNMP counters.

Fix for out-of-sequence FIN handling, patch by Oleg Tychev.

Dangling inseg.p pointer fix by Pedro Alves.

Removed 'even sndbuf' fix in TCP, this fix was a ugly hack for the incomplete checksum routine that does not work now that the checksum routine is fixed.

Duplicate FIN ACK race condition fix by Kelvin Lawson.

Introduced cc.h formatters and removed SO_REUSE from transport layers.

2005-06-08 Leon Woestenberg <leon.woestenberg@gmx.net>
* tcp_in.c: Send reset on unexpected ACK in two cases. Fix submitted by olaf.peters@technolution.

4 February 2004, Leon Woestenberg <leon.woestenberg@gmx.net>
tcp_in.c: Applied fix patch for bug #2679.
tcp_out.c: Applied fix patch for bug #2679.
http://savannah.nongnu.org/bugs/?func=detailitem&item_id=2679

Kieran Mansley - kjm25@cam.ac.uk - 24th Nov 2004
* Increase pcb->snd_buf by 1 when an ACK is received in SYN_SENT state to ensure correct operation

Kieran Mansley - kjm25@cam.ac.uk - 20th September 2004
* Corrected "out by one" error on one of the TCP_SEQ_BETWEEN macro calls introduced recently

Kieran Mansley - kjm25@cam.ac.uk - 12th September 2004

Applied patch from Sam Jansen as detailed in
http://lists.gnu.org/archive/html/lwip-users/2004-07/msg00106.html
to correctly handle retransmission after a retransmission timeout

Fixed typo (missing "{") in previous checkin

Add TCP_SEQ_BETWEEN macro for comparing a range of sequence numbers

Ingress TCP keep-alive with garbage byte support.

pbuf_free() assert triggered by NULLified inseg.p. Reported by Karl Jeacle on 25-7-2004 on lwip-users.

Disabled some code by #if 0 #endif rather than /* */.

Kieran Mansley - 14th July 2004

* Fixed whitespace indenting in parts of tcp_in.c
* Changed adjustment of ssthresh in response to fast retransmit
* Commented out iteration of unsent list when new ACK received as we no longer put all unacked data on unsent list when retransmitting

Kieran Mansley - kjm25@cam.ac.uk - 26th May 2004

* Fixed bug 9076 (changes to pbuf tot_len and seg dataptr when trimming front edge of a received packet)

Clean-up source comment documentation for Doxygen.

Kieran Mansley - kjm25@cam.ac.uk - 23rd March 2003
- Changed all #ifdef SO_REUSE into #if SO_REUSE

Made #ifdef's into #if's for SO_REUSE.

Fix cyclic dependencies by careful re-ordering of #includes.
This is really nasty. Can we forward-declare pointers to structs?

Replaced ip_addr_isbroadcast() macro by function.
Overrides patch #2679, as this must be solved inside ip_addr_isbroadcast(), inspired by BSD.

23/02/2003 - Kieran Mansley - kieranm@gtemail.net
* Applied patch #2679

* Adds check to line 133 for NETIF_FLAG_BROADCAST

Added ability to disable checksum generation and checking at compile-time

Applied Ian Wienand's patches (4-2-2004). Was partly applied earlier.

Merged from DEVEL, except for the API change in etharp.c.

Applied alias.diff by Ian Wienand on lwip-devel on February 4th 2004.
Using union to make explicit either tcp_listen_pcb or tcp_pcb.

Merged from DEVEL to main. Two TCP fixes and two NULL reference fixes.

Merge from DEVEL. Assert used field not available without callback API.

Merged from DEVEL into main tree.

Fixes wrongly cast LWIP_DEBUGF arguments (patch #1596 and more).

Changed DEBUGF to LWIP_DEBUGF

Replaced all tabs with two spaces (regardless of indentation is correct).

put LWIP_MAX and LWIP_MIN macros in def.h and use these instead of locally declaring them in .c files

Applied patches posted by Marc Boucher on lwip-users May 18th 2003.

Major stylo search/replace for "One space between keyword and opening bracket."

Patch #1183 applied. This drops short UDP/TCP packets.

Fix memory leaks.

When pbufs are chained, the second pbuf must be freed, if no higher level
application refers to it any longer.

etharp_query() has error return type now. Matched dhcp.c with this change.
Added debug messages in other places.

Added documentation and comments.

if LWIP_TCP is 0 do not link in TCP code.putting ugly ifdefs in api and core :(.Also only udp_init if LWIP_UDP is on

Using a negated unsigned value as func. arg. fails for Keil C166 compilers.

byte-order handling functions are in inet.c now and the uperrcase counterparts are gone. opt.h has all the
configurable items debug does not need to be directly included.

Add the following features and bugfixes:

Added select() functionality to sockets library.
Support for errno in sockets library.
Byte ordering fixes.
basic lwip_ioctl(), FIONREAD, get/setsockopt() etc. support

- added additional argument to netif_add to pass state pointer so that the
if_init function has access to context information before
the interface is added, without accessing globals.

- added netif_remove()

- to conserve cpu load the tcpip_tcp_timer should only be active
when tcbs that need it exist.

- pass length of available data to callbacks for NETCONN_EVT_RCV events

- added tcpip_link_input(), a hack to allow processing of PPP
packets in tcpip_thread() context. This saves threads and context
switches.

- renamed incompatible ASSERT() macro to LWIP_ASSERT() to avoid name
collision.

- changed a bunch of %d's to %u's in format strings for unsigned values.

- added ip_frag to lwip_stats.

- changed IP_REASS_MAXAGE and IP_REASS_TMO defaults to more realistic
values.

- added sys_timeout_remove() function to cancel timeouts (needed by PPP
amongst other things).

- tolerate NULL returns from sys_arch_timeouts() since some threads might
not need to use or have timeouts.

- added sys_sem_wait_timeout()

- moved mem_malloc() function to end of mem.c to work around tasking
compiler bug.

- automatically bind to local tcp port if 0.

- allow customization of port ranges for automatic local bindings.

- corrected various typos, spelling errors, etc..

Thanks to Marc Boucher for many of these changes.

ack duplicate segments too to prevent connection hanging when a single transmitted ACK gets lost

Fixed some typo's in the comments.

seqno is u32_t so use %lu instead of %ld when printing it

Updated lwIP module copyright years to include 2003. Committers must check theirs.

sizeof(u8_t) is 1 at temperatures > 0 K

renamed (hopefully everywhere) stats to lwip_stats.closes bug #1901

Use C style comments.In debug stataments cast various struct pointers to void* to
avoid printf warnings.misc warnings in etharp.

Zero pcb->acked if ack does not acknowledge new data
- prevents the sent function being called multiple times when no new data has been acknowledged

Added debug messages for corner cases of TCP states.

Fix function names in DEBUGF stataments.Remove reference to "reclaimed" field in stat.

Fixed a few C style issues to make it compatible with the rest of lwIP.

Added check on entire sequence number of received packet being less than the rcv_nxt variable.
(see http://www.sics.se/mailing-lists/lwip.html/msg01409.html)

Fixed bug in duplicate ack handling
(see http://www.sics.se/mailing-lists/lwip.html/msg01460.html)
Solution adds check on right window edge when duplicate ack received, if it is different it is not a duplicate.

Fixed congestion window bug where the pcb->cwnd variable overflowed when increased.

Updated TCP out-of-sequence handling to use recv_data variable instead of old pcb->recv_data.

New file structure towards SLIP/PPP. Renamed tcp_*.c for 8.3 sake.
See lwip@sics.se Mon, 21 Oct 2002 magnus.ivarsson