This makes the following poll cleanups:
1) Add LWIP_ERROR in lwip_poll to check for invalid fds/nfds combinations.
This fixes a possible a NULL fds dereference in lwip_poll_scan()
2) Use has_ copies of the socket events in lwip_poll_should_wake() rather
passing the sock pointer and accessing socket after leaving the critical
section
This fixes the following warnings:
test_tcp.c:266:5: error: code will never be executed [-Werror,-Wunreachable-code]
pbuf_free(p);
^~~~~~~~~
- The check API 'fail' aborts the test, thus pbuf_free(p) will never be executed
pbuf.c:783:111: error: format specifies type 'unsigned short' but the argument has type 'u8_t' (aka 'unsigned char') [-Werror,-Wformat]
LWIP_DEBUGF( PBUF_DEBUG | LWIP_DBG_TRACE, ("pbuf_free: %p has ref %"U16_F", ending here.\n", (void *)p, ref));
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~
- LWIP_PBUF_REF_T is u8_t by default and doesn't match U16_F, so cast to u16_t. The cast and formatter will need to be changed
if ref is larger than 16 bits
ethernet.c:105:16: error: format specifies type 'unsigned char' but the argument has type 'unsigned int' [-Werror,-Wformat]
(unsigned)ethhdr->dest.addr[0], (unsigned)ethhdr->dest.addr[1], (unsigned)ethhdr->dest.addr[2],
~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- addr[] is type u8_t, formatter is X8_F which should be 8 bits. 'unsigned' is an int, so cast to unsighed char instead
Current newlib does not define O_NDELAY, but it needs to be the same
as O_NONBLOCK rather than using the lwip value of 1.
Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>
This makes two cleanups that follows the same organization as close/closesocket:
1) There is no lwip_ioctlsocket. Instead lwip_ioctl should redirect to ioctlsocket
2) With LWIP_POSIX_SOCKETS_IO_NAMES enabled, lwip_ioctl/ioctlsocket should redirect to ioctl
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>
Hop-by-Hop, Destination option header structures consist of 2 unsigned char; next option type and header length field.
And TLV(Type-Length-Value) option headers come by the number in header length field.
If the option type in TLV option header is not recognized and 2 MSB is not 0, it is handled as an exception.
Signed-off-by: goldsimon <goldsimon@gmx.de>
Hop-by-Hop, Destination option header structures consist of 2 unsigned char; next option type and header length field.
And TLV(Type-Length-Value) option headers come by the number in header length field.
If the option type in TLV option header is not recognized and 2 MSB is not 0, it is handled as an exception.
Signed-off-by: goldsimon <goldsimon@gmx.de>
For this, convert 'u8_t nexth' to a pointer and change 'icmp6_param_problem()' to take a pointer, not an offset number
Signed-off-by: goldsimon <goldsimon@gmx.de>
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>
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>
There were a couple cases in-between that could cause an exit from
tcp_output which don't use useg. With large send buffers, pcb->unacked
may be large and calculating useg is wasted in these exit cases
Some compilers may be re-ordering this already, but it doesn't hurt to
correctly arrange the code
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
netif_get_by_index() returns NULL if idx is NETIF_NO_INDEX.
So remove the superfluous NETIF_NO_INDEX checking for msg->msg.jl.if_idx
before calling netif_get_by_index().
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>
Interface indexes are u8_t internally so cast from sockets int representation to u8_t
This was found with MSVC 2013:
1>lwip\src\api\sockets.c(3190): warning C4242: 'function' : conversion from 'const unsigned int' to 'u8_t', possible loss of data
1>lwip\src\api\sockets.c(3698): warning C4242: 'function' : conversion from 'unsigned int' to 'u8_t', possible loss of data
lwip_itoa would output the number 0 as \0. This fixes the issue by
adding a special check before the normal conversion loop
This was found via shell cmd idxtoname and win32 port. "lo0" should
be returned for index 1
- 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)
There is already a guard in tcp_output_segment() for a pbuf still being
referenced by the netif driver due to deferred transmission, however the callers
are modifying state even when this gives up.
It seems cleaner to have the callers guard this case and avoid modifying their
state.
tcp_rexmit_rto() might better avoid re-transmission of any segments if any of
the unacked segments are deferred, to avoid loading the link further if it is
struggling to flush its buffered writes. Link level queues can be limited on
some devices and need spares for link management.
- added `altcp_tls_free_config()`.
- added `altcp_tls_context()` function to allow mbedtls parameter tweak.
Since state structure isn't exported, this allow application to get
internal context (port dependent) to tweak it.
- free altcp_pcb when lower error callback called.
- expose `altcp_tcp_setup()` so we can wrap altcp over existing tcp pcb.
- avoid calling tcp_close() with NULL pcb.
- free altcp_pcb struct when error callback called.
According to `mqtt_tcp_err_cb()` in src/apps/mqtt/mqtt.c, altcp socket should
work the same way than raw tcp socket. So freeing altcp_pcb ensure this.
The new functions both take size_t as increment/decrement argument instead of s16_t (which needed to be range-checked before conversion everywhere) - in most places, the direction (increment or decrement) is known anyway, so no need to encode it in a sign bit
In tcp_output() there were a number of blocks of code performing
duplicate checks of 'if (seg == NULL)'. This combines them together
to reduce duplicate checks
TCP_OUTPUT_DEBUG and TCP_CWND_DEBUG also don't need to be guarded
by #if/#endif since the LWIP_DEBUGF infrastructure already compiles them
out when LWIP_DEBUG is not defined
The ip6_frag.drop counter is updated before all the code paths calling
goto nullreturn, so let's move updating ip6_frag.drop stats to nullreturn.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
This fixes a couple of occurrences where the src and dst parameters to
ip4_route_src() were swapped. This was most likely due to confusion between
ip_route(src, dst) and ip4_route_src(dst, src)
This was found in a system where LWIP_IPV4_SRC_ROUTING is 0
The UDP case was an application socket bound to INADDR_ANY with
IP_MULTICAST_IF set. Transmits would result in calling ip4_route(dst) where
dst was pcb->local_addr (which was INADDR_ANY) instead of pcb->mcast_ip4.
This resulted in a routing failure
The ICMP issue was found through code analysis only
There were uses of dhcp_release() followed immediately by dhcp_discover() but
dhcp_release() now stops dhcp so discovery would fail, so call dhcp_start()
after release which restarts discovery.
Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>
According to commit 1f780e86d5 ("PPP timeouts required depend on the number of allowed PPP sessions"):
Per PPP needs 6 timeouts (AUTH + PAP|CHAP|EAP + LCP + IPCP + IP6CP + PPPoE).
So update the minimal MEMP_NUM_SYS_TIMEOUT setting check accordingly.
Since we have LWIP_NUM_SYS_TIMEOUT_INTERNAL so just switch to use it.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: goldsimon <goldsimon@gmx.de>
This refactors event_callback() to separate updating socket event
state from processing the select list (to apply socket event change)
Refactoring changes:
1) select_list_cb processing has been moved to a new local function called
select_check_waiters()
2) goto no_select_wakeup has been removed and now we use a flag
to track whether to call select_check_waiters()
3) There is a small functional change for !LWIP_TCPIP_CORE_LOCKING.
We call SYS_ARCH_UNPROTECT after saving events but before calling
select_check_waiters() (which now calls PROTECT before starting the loop).
Before the code held the PROTECT across saving the events and the first
loop iteration, but this didn't protect against anything because each loop
iteration we do an UNPROTECT/PROTECT
4) Better documentation for both LWIP_TCPIP_CORE_LOCKING and
!LWIP_TCPIP_CORE_LOCKING
These are now defined to return != SYS_ARCH_TIMEOUT on success rather than the time
waiting. The returned times were unused by lwip and this simplifies at
least some implementations.
Signed-off-by: goldsimon <goldsimon@gmx.de>
This fixes a bug where when writing IP_PKTINFO to msg_control, the
msg_controllen field was not updated with the length written
This bug is exposed by applications that provide a msg_control buffer large
enough for multiple control messages. Then when calling CMSG_NXTHDR, it
returned a next cmsg pointer even though was no additional message
The ip6_addr_t structure may have an addition slot so is not necessarily
the size of an ipv6 address, so some uses of sizeof(ip6_addr_t) were not
correct.
Signed-off-by: goldsimon <goldsimon@gmx.de>
No need to have additional if statement for PBUF_REF/PBUF_ROM.
It can be merged to the existing swtich(type) cases.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
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>
Use vj_uncompress_err() instead of duplicating the same code.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Sylvain Rochet <gradator@gradator.net>
Current code does not correctly update ifoutoctets counter because nb->tot_len
is always 0. Fix it by setting nb->tot_len to actual payload length so we can
update ifoutoctets correctly.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Sylvain Rochet <gradator@gradator.net>
This changes tcpflags_t to be a u16_t for all cases. The TCP Appropriate
Byte Count support added a new flag that used a bit past 8 and since this
flag is now required, tcpflags_t can no longer be a u8_t
This does not increase the size of struct tcp_pcb due to padding that
already existed (see bug #51326 for details)
Signed-off-by: goldsimon <goldsimon@gmx.de>
The function previously returned after posting a message, which is a short operation. Now it actually waits until the operation has completed - which may take a long time. This may break user programs. So all that remains is the cleanup separation in tcpip_callback() and tcpip_try_callback() :-(
Created two new functions for API cleanup:
tcpip_callback() that blocks until message is posted, cannot be called from IRQs.
tcpip_try_callback() that does not block and just tries to post a message. Can be called from IRQs.
Add compatibility #define tcpip_callback_with_block() that maps to these two functions according to "block" parameter.
This switches netconn_gethostbyname to use tcpip_send_msg_wait_sem to
take advantage of core locking support when enabled.
tcpip_send_msg_wait_sem handles blocking for the non-core locking case,
so we can remove the manual blocking in netconn_gethostbyname. For the
core locking case, we need to block if waiting on DNS callback. To
achieve this, we unlock the core and wait in lwip_netconn_do_gethostbyname.
This is the similar approach that netconn_write takes when it needs to
block to continue the write (see lwip_netconn_do_write)
This improves performance in the core locking case and is no change
for the non-core locking case
TCP timestamps were only sent if the remote side
requested it first. This enables the use of timestamps
for outgoing connections as well.
Signed-off-by: goldsimon <goldsimon@gmx.de>
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
etharp_query() queues packets, instead of sending, if a relevant arp-request is
pending.
Code walks the packet (a pbuf chain) to determine whether any pbufs are marked
'volatile': If so, we cannot simply enqueue the packet, and instead allocate a
new pbuf from RAM, copying the original packet, and enqueueing this new pbuf.
The bug here is that the allocation refers to the tot_len field of a temp pbuf*,
'p', instead of the head, 'q'.
In the case where the first pbuf of the chain is non-volatile but the second pbuf
*is* volatile, then we'll request an allocation that uses the tot_len field of
the second pbuf. If the first pbuf is non-zero length, the allocated pbuf (chain)
will be too small to allow the copy.
Signed-off-by: goldsimon <goldsimon@gmx.de>
All callers pass pbuf_type to pbuf_init_alloced_pbuf(), so make it take
pbuf_type instead of u8_t.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: goldsimon <goldsimon@gmx.de>
netif->output and etharp_output are only available when LWIP_IPV4=1.
Fix the skeleton file.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: goldsimon <goldsimon@gmx.de>
The lwip/arch.h already provides a default implentation of LWIP_PLATFORM_ASSERT
and LWIP_PLATFORM_DIAG.
So both LWIP_PLATFORM_ASSERT and LWIP_PLATFORM_DIAG are never undefined here.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Call smtp_free_struct(s) in all smtp_send_mail_alloced error paths to ensure
no memory leak.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>
The pcb is "struct altcp_pcb *" so we cannot call tcp_sndbuf/tcp_write here.
Use altcp_sndbuf/altcp_write instead.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>
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
We don't have to keep a helper function just for the sake of a PBUF_RAW
constant. Inline ppp_singlebuf function.
Signed-off-by: Sylvain Rochet <gradator@gradator.net>
pbuf_coalesce() creates a single pbuf out of a chain of pbufs, which is
exactly what ppp_singlebuf() need.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Sylvain Rochet <gradator@gradator.net>
Fix below build error if LWIP_IPV4 == 0.
cc -g -Wall -DLWIP_DEBUG -pedantic -Werror -Wparentheses -Wsequence-point -Wswitch-default -Wextra -Wundef -Wshadow -Wpointer-arith -Wcast-qual -Wc++-compat -Wwrite-strings -Wold-style-definition -Wcast-align -Wmissing-prototypes -Wredundant-decls -Wnested-externs -Wno-address -Wunreachable-code -Wuninitialized -Wlogical-op -I. -I../../.. -I../../../../lwip/src/include -I../../../ports/unix/port/include -I../../../../mbedtls/include -Wno-redundant-decls -DLWIP_HAVE_MBEDTLS=1 -c ../../../../lwip/src/core/netif.c
../../../../lwip/src/core/netif.c: In function ‘netif_add’:
../../../../lwip/src/core/netif.c:284:7: error: ‘ipaddr’ undeclared (first use in this function)
if (ipaddr == NULL) {
^~~~~~
../../../../lwip/src/core/netif.c:284:7: note: each undeclared identifier is reported only once for each function it appears in
../../../../lwip/src/core/netif.c:285:14: error: implicit declaration of function ‘ip_2_ip4’ [-Werror=implicit-function-declaration]
ipaddr = ip_2_ip4(IP4_ADDR_ANY);
^~~~~~~~
../../../../lwip/src/core/netif.c:285:5: error: nested extern declaration of ‘ip_2_ip4’ [-Werror=nested-externs]
ipaddr = ip_2_ip4(IP4_ADDR_ANY);
^~~~~~
../../../../lwip/src/core/netif.c:285:23: error: ‘IP4_ADDR_ANY’ undeclared (first use in this function)
ipaddr = ip_2_ip4(IP4_ADDR_ANY);
^~~~~~~~~~~~
../../../../lwip/src/core/netif.c:287:7: error: ‘netmask’ undeclared (first use in this function)
if (netmask == NULL) {
^~~~~~~
../../../../lwip/src/core/netif.c:290:7: error: ‘gw’ undeclared (first use in this function)
if (gw == NULL) {
^~
cc1: all warnings being treated as errors
../../Common.allports.mk:94: recipe for target 'netif.o' failed
make: *** [netif.o] Error 1
Fixes: 5967380c20 ("netif_add: avoid passing NULL pointers to subsequent functions")
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Dirk Ziegelmeier <dirk@ziegelmeier.net>
Current code fails to allocate zero length pbuf (e.g. for PBUF_RAW PBUF_POOL),
fix it.
Fixes: eb269e61b5 ("First step to clean up pbuf implementation: add pbuf_alloc_reference() to allocate pbufs referencing external payload; move member initialization to common function; simplify PBUF_POOL chain allocator")
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: goldsimon <goldsimon@gmx.de>
This changes the CMSG alignment macros to ensure struct cmsghdr and data
are on a word (double word on 16-bit arch) aligned boundary
We need to ensure at least 32-bit alignment for 16-bit systems because
socklen_t could be 32-bit due to our definition
This fixes build error if LWIP_NETIF_TX_SINGLE_PBUF==1.
Fixes: dd811bca06 ("Fix bug #50694 (TX exist more pbufs after enable LWIP_NETIF_TX_SINGLE_PBUF) by not executing phase 2 for LWIP_NETIF_TX_SINGLE_PBUF==1")
Signed-off-by: Axel Lin <axel.lin@ingics.com>
This commit adds CMSG infrastructure (currently used with recvmsg) and
the IP_PKTINFO socket option.
In order to use IP_PKTINFO, set LWIP_NETBUF_RECVINFO to 1
Unit test is added to verify this feature
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
while ((q != NULL) && (options[offset] != DHCP_OPTION_END) && (offset < offset_max)) {
should be
while ((q != NULL) && (offset < offset_max) && (options[offset] != DHCP_OPTION_END)) {
See https://jira.reactos.org/browse/CORE-8978 for more info.
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>
The NULL test no longer work after commit e0a2472706, it needs to test with
lwip_netconn_is_err_msg() instead.
Fixes: e0a2472706 ("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>
This sets the pbuf's if_idx during the loopif poll function (the
equivalent netif input function). This was found during IP_PKTINFO
development where p->if_idx is read and was uninitialized
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
Introduce a new configuration define SNTP_COMP_ROUNDTRIP to compensate
for network round-trip delays when setting the system clock from SNTP.
Note that this feature requires compiler support for 64-bit arithmetic.
Avoid piecemeal invocation of pbuf_copy_partial() for each SNTP header
field. The new code may sometimes copy more than necessary, but the
complexity is reduced.
Generalize the NTP timestamp conversion arithmetic, and provide hooks
for using native NTP timestamps when setting or getting the system
clock time. Convert microseconds to a fraction as needed when getting
the system time.
