Issue:
DHCP server will send DHCP offer/ack via broadcast packet if broadcast
flag is enabled in DHCP discover/request.
There's no retransmition mechanism for broadcast frame in 802.11 MAC
layer, thus transmit fail rate might be very high. This will increase DHCP fail rate.
Solution:
Disable broadcast flag by default
Signed-off-by: liangchaozhong <liangchaozhong@xiaomi.com>
recv request can only get maximum 1500 bytes payload in single response in
earlier design, this patch aggregrate more payloads（maximum 16k bytes）
into one response to make sure usrsock’s req/response mechanism is not
bottleneck during usrsock’s throughput test.
Signed-off-by: liangchaozhong <liangchaozhong@xiaomi.com>
Error: discover.c:210:13: error: format '%lx' expects argument of type 'long unsigned int', but argument 3 has type 'in_addr_t' {aka 'unsigned int'} [-Werror=format=]
210 | ninfo("Received discover from %08lx'\n", srcaddr.sin_addr.s_addr);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~
| |
| in_addr_t {aka unsigned int}
discover.c:210:41: note: format string is defined here
210 | ninfo("Received discover from %08lx'\n", srcaddr.sin_addr.s_addr);
| ~~~~^
| |
| long unsigned int
| %08x
In file included from discover.c:25:
discover.c: In function 'discover_openlistener':
Error: discover.c:359:9: error: format '%lx' expects argument of type 'long unsigned int', but argument 3 has type 'uint32_t' {aka 'unsigned int'} [-Werror=format=]
359 | ninfo("serverip: %08lx\n", ntohl(g_state.serverip));
| ^~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~
| |
| uint32_t {aka unsigned int}
discover.c:359:24: note: format string is defined here
359 | ninfo("serverip: %08lx\n", ntohl(g_state.serverip));
| ~~~~^
| |
| long unsigned int
| %08x
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
This option, which resolves to -w when CONFIG_CYGWIN_WINTOOL is
configured, is now appended to INCDIR in tools/Config.mk.
See git commit # 5eae32577e5d5226e5d3027c169eeb369f83f77d in the main
Summary:
- I noticed that the webserver can not handle a directory.
- This commit fixes this issue.
Impact:
- None
Testing:
- Tested with spresense:wifi_smp
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
High performance websocket client/server
The goal of cwebsocket is to provide a portable,
high performance websocket client/server,
especially on low power embedded systems.
cwebsocket is currently in a development state. You may encounter bugs.
Report them for a timely fix.
Successful tests have been conducted on the following architectures:
x86
x86_64
ARM
cwebsocket is compliant with the following standards:
ANSI C
POSIX
RFC 6455
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
A return code of EINTR is perfectly normal, and isn't an error as such.
It's an indication that your program may need to do something because
a signal occurred, but if not, should re-call recv().
Signed-off-by: chao an <anchao@xiaomi.com>
dhcp6c.c:222:3: error: unknown type name ‘pthread_t’
222 | pthread_t thread;
| ^~~~~~~~~
dhcp6c.c: In function ‘dhcp6c_get_result’:
dhcp6c.c:415:7: warning: implicit declaration of function ‘netlib_prefix2ipv6netmask’ [-Wimplicit-function-declaration]
415 | netlib_prefix2ipv6netmask(presult->pl, &presult->netmask);
| ^~~~~~~~~~~~~~~~~~~~~~~~~
dhcp6c.c: In function ‘dhcp6c_parse_ia’:
dhcp6c.c:1046:30: warning: unused variable ‘pdhcp6c’ [-Wunused-variable]
1046 | FAR struct dhcp6c_state_s *pdhcp6c = (FAR struct dhcp6c_state_s *)handle;
| ^~~~~~~
dhcp6c.c: In function ‘dhcp6c_precise_open’:
dhcp6c.c:1702:3: warning: missing braces around initializer [-Wmissing-braces]
1702 | {
| ^
......
1706 | {0},
| -
| {{0}}
dhcp6c.c:1796:43: error: ‘UDP_BINDTODEVICE’ undeclared (first use in this function); did you mean ‘SO_BINDTODEVICE’?
1796 | setsockopt(pdhcp6c->sockfd, SOL_SOCKET, UDP_BINDTODEVICE, ifname,
| ^~~~~~~~~~~~~~~~
| SO_BINDTODEVICE
dhcp6c.c:1796:43: note: each undeclared identifier is reported only once for each function it appears in
dhcp6c.c: In function ‘dhcp6c_request_async’:
dhcp6c.c:1886:9: warning: implicit declaration of function ‘pthread_create’; did you mean ‘timer_create’? [-Wimplicit-function-declaration]
1886 | ret = pthread_create(&pdhcp6c->thread, NULL, dhcp6c_run, pdhcp6c);
| ^~~~~~~~~~~~~~
| timer_create
Signed-off-by: chao an <anchao@xiaomi.com>
rename the UDP_BINDTODEVICE to SO_BINDTODEVICE to follow the linux
style to be compatible with non-UDP protocol binding requirements
Signed-off-by: chao.an <anchao@xiaomi.com>
Summary:
- This commit makes telnetd_daemon() in public so that we
can call it from applications.
- Also, adds new configs to support posix_spawnp()
Impact:
- telnetd only
Testing:
- Tested with sabre-6quad:netknsh (will be updated later)
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
NXStyle fixes
Apply suggestions from code review
Co-authored-by: Petro Karashchenko <petro.karashchenko@gmail.com>
API fixes and NXStyle fixes
Apply suggestions from code review
Co-authored-by: Petro Karashchenko <petro.karashchenko@gmail.com>
Use thread-safe ntoa functions
NXStyle fixes
Apply suggestions from code review
Co-authored-by: Petro Karashchenko <petro.karashchenko@gmail.com>
API fixes and NXStyle fixes
Apply suggestions from code review
Co-authored-by: Petro Karashchenko <petro.karashchenko@gmail.com>
NETUTILS_DHCPC_RECV_TIMEOUT -> NETUTILS_DHCPC_RECV_TIMEOUT_MS
Recently its unit has been changed from second to millisecond. [1]
Using the same Kconfig name for a different meaning is a pitfall
for an upgrade. I was using 3ms timeout because of this.
This commit renames it so that at least the default value
is used for a careless user like me.
[1]
```
commit 8fd4b6105b
Author: zrrong <zrrong@bouffalolab.com>
Date: Fri May 27 11:35:23 2022 +0800
netutils/dhcpc: Change the timeout unit to milliseconds
```
My primary motivation at this point is to use it for basic proxy auth.
(specify "Proxy-Authorization" header)
But there can be other use cases for proxy-specific extra headers.
If/when we want to support other non-trivial auth methods, probably
a callback-based mechanism will be necessary. But at this point,
this serves my purpose well.
Use a separate webclient_context for tunnel establishment.
I chose this way (instead of having tunnelling steps in
the state machine of a single webclient_context) because
I want to allow tunnelling of non-HTTP protocols sooner or later.
Add a primitive API for tunnel establishment.
(WEBCLIENT_FLAG_TUNNEL and webclient_get_tunnel)
I plan to use this to implement https proxy support.
That is, the primary user will be webclient itself.
nsh> ping 192.168.31.162
PING 192.168.31.162 56 bytes of data
56 bytes from 192.168.31.162: icmp_seq=0 time=8 ms
56 bytes from 192.168.31.162: icmp_seq=1 time=27 ms
56 bytes from 192.168.31.162: icmp_seq=2 time=17 ms
56 bytes from 192.168.31.162: icmp_seq=3 time=14 ms
56 bytes from 192.168.31.162: icmp_seq=4 time=16 ms
56 bytes from 192.168.31.162: icmp_seq=5 time=14 ms
56 bytes from 192.168.31.162: icmp_seq=6 time=8 ms
56 bytes from 192.168.31.162: icmp_seq=7 time=87 ms
WARNING: Received after timeout
56 bytes from 192.168.31.162: icmp_seq=8 time=1003 ms
56 bytes from 192.168.31.162: icmp_seq=8 time=6 ms
56 bytes from 192.168.31.162: icmp_seq=9 time=5 ms
10 packets transmitted, 11 received, 4294967287% packet loss, time 10011 ms
Signed-off-by: zhouliang3 <zhouliang3@xiaomi.com>
This reverts commit bd539651cdf1e5067dc5d21349ef835e98e262eb.
Reason for revert: Change the implementation method
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
1、Round trip times in the ping command range from millisecond to subtle
2、Add statistics on RTT related min/avg/Max/mdev in ping program
3、The ping command supports ctrl+c interruption operations
Signed-off-by: xuewenliang <xuewenliang@xiaomi.com>
Gregory Nutt is the copyright holder for those files and he has submitted the
SGA as a result we can migrate the licenses to Apache.
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
Gregory Nutt has submitted the SGA
Pierre-Noel Bouteville has submitted the ICLA
as a result we can migrate the licenses to Apache.
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
When making the following change, I haven't noticed that
g_httpuseragentfields contains a "Connection" header.
```
commit 092ce81444
Author: YAMAMOTO Takashi <yamamoto@midokura.com>
Date: Mon Mar 7 09:30:23 2022 +0900
webclient: Always use "connection: close" for HTTP 1.1 for now
* This matches the HTTP 1.0 behavior.
* Persistent connection doesn't make much sense with the current API.
```
It seems that some servers are not happy with the duplicated header
and ignore them. (and do the default keep-alive for HTTP 1.1)
eg. Azure Blob (global)
Gregory Nutt is the copyright holder for those files and he has submitted the
SGA as a result we can migrate the licenses to Apache.
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
httpd_dirlist.c: In function 'httpd_dirlist':
Error: httpd_dirlist.c:199:40: error: '%s' directive output may be truncated writing up to 255 bytes into a region of size 128 [-Werror=format-truncation=]
199 | snprintf(path, CONFIG_NAME_MAX, "%s/%s",
| ^~
httpd_dirlist.c:199:7: note: 'snprintf' output between 2 and 385 bytes into a destination of size 128
199 | snprintf(path, CONFIG_NAME_MAX, "%s/%s",
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
200 | file->path, dent->d_name);
| ~~~~~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
Sometimes netcat in server mode crashed with "psock_accept: ERROR: si_accept failed: -9".
And sometimes it crashed with "up_assert: Assertion failed at file:inet/inet_sockif.c line: 841 task: netcat".
This option enables using sendfile() in ftpc binary transfer mode of PUT operation.
If the option is enabled but ASCII transfer mode is activated,
ftpc falls back to the combination of read() and write().
Using sendfile() provides a higher performance compared to
the combination of read() and write().
Also this option is useful for testing / debugging tcp_sendfile()
functionality of NuttX TCP/IP stack.
This option enables using sendfile() in netcat client mode
if a normal file (not stdin) is sent. If the option is enabled
but stdin is sent rather than a normal file, netcat falls back
to the combination of read() and write().
Using sendfile() provides a higher performance compared to
the combination of read() and write().
Also this option is useful for testing / debugging tcp_sendfile()
functionality of NuttX TCP/IP stack.
The logic changes are as follows:
1、Loop sending the DISCOVER up to CONFIG_NETUTILS_DHCPC_RETRIES times
and exit if failure
2、Loop sending the REQUEST up to CONFIG_NETUTILS_DHCPC_RETRIES times
and exit if failure
Signed-off-by: songlinzhang <songlinzhang@xiaomi.com>
