The previous commit had task_fork_test() also accept CONFIG_ARCH_HAVE_FORK,
because the nuttx symbols naming the three primitives did not exist yet. nuttx
now has them, so each test gates on the one primitive it tests and nothing
stands in for anything.
The OSTEST_HAVE_* indirection goes with the fallbacks, as does the
task_fork()->fork() shim in nand_sim_main.c. The whole-file #ifdef in
task_fork.c goes too: the build files already decide whether the file is
compiled, which is why fork.c and vfork.c do not carry one either.
task_fork_test() keys on CONFIG_TASK_FORK rather than the capability symbol.
ARCH_HAVE_TASK_FORK says the architecture can clone a task; TASK_FORK says this
build asked for it, and task_fork() is declared only under the latter, so
gating on the capability alone would fail to compile a TASK_FORK=n build.
TESTING_NAND_SIM depends on the same symbol for the same reason. vfork_test()
and fork_test() have no such split and key on ARCH_HAVE_VFORK and
ARCH_HAVE_FORK directly.
This must not be merged before the nuttx side.
Signed-off-by: Marco Casaroli <marco.casaroli@gmail.com>
Assisted-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
nuttx implements fork() and vfork() as the same function, and is gaining the
three separate primitives its issue #19540 describes: task_fork() (shares
memory, private stack copy, both running), vfork() (shares memory, parent
suspended) and POSIX fork() (child gets its own copy). This is the apps side
of that, and it lands first: it works against nuttx with or without the
split, so the tests keep running across the transition rather than silently
compiling out.
ostest's "vfork" test was never testing vfork(). It has the child write a
global and the parent observe the write -- which is the defining property of
*sharing*, not of vfork(), whose defining property is that the parent is
suspended and whose contract forbids the child to write anything at all. It
is renamed to task_fork.c, unchanged, because that is the primitive it has
always described.
vfork.c is rewritten to test what vfork() promises. The child does only what
POSIX permits -- it calls _exit(42), and nothing else, not even exit(), which
would run atexit handlers and flush stdio in the parent's address space. The
observable is therefore the child's exit status rather than a memory write.
Where child status is not retained -- ostest_main() sets SA_NOCLDWAIT for the
whole run, deliberately -- waitpid() returning ECHILD is accepted as equally
good evidence: it says the child was already gone when the parent asked.
fork.c is new and tests POSIX fork(): the child's writes to .data, .bss and
the heap are invisible to the parent and vice versa, a pointer to a stack
local taken before the fork names the same object in both, and the child does
everything a vfork() child may not -- calls malloc() and printf(), and returns
from the function that called fork().
All three run at the top of user_main() rather than in the middle. They
exercise the lowest-level machinery in the suite -- address environments,
stack setup, the architecture's register context -- so a fault in one takes
the process down instead of reporting a failure, and finding that out in
seconds rather than after everything else has passed is the difference
between a usable iteration and a coffee break when a port is being brought
up.
The other in-tree callers are audited for which primitive they actually
meant. nand_sim wants a daemon that outlives its caller and shares its
memory, which is task_fork(). bas's SHELL and EDIT statements, python's
_posixsubprocess and libwebsockets' feature macros want the fork-then-exec
path, which vfork() serves; python's os.fork() and libwebsockets'
LWS_HAVE_FORK stay on fork() proper. fdsantest's vfork case follows vfork().
Two third-party suites need their source lists narrowed, because they call
fork() from code that is compiled unconditionally:
* system/libuv -- test-fork.c and test-pipe-close-stdout-read-stdin.c are
filtered out of the test-*.c glob. Every test they define is already
excluded from the task list on NuttX by 0001-libuv-port-for-nuttx.patch --
the nine fork_* entries and pipe_close_stdout_read_stdin -- so they were
dead code being compiled only because fork() happened to be declared.
* testing/ltp -- the open_posix_testsuite is filtered through the existing
BLACKWORDS mechanism, which already drops tests for absent features and is
already conditioned on configuration symbols. Where fork() is not
provided this drops 278 of 1943 test files; the pattern is written to spare
vfork() and task_fork(), which remain available. Where fork() is provided
-- which today is everywhere -- nothing is dropped.
Compatibility: the nuttx symbols this keys on do not exist yet. task_fork.c
is built where CONFIG_TASK_FORK says task_fork() was built and, on a nuttx that
has no CONFIG_ARCH_HAVE_TASK_FORK at all -- which is the pre-split one -- where
CONFIG_ARCH_HAVE_FORK does. Today's fork() *is* task_fork(), so the test that
has always covered that primitive keeps running, under its own name, and no
coverage is lost across the transition. Both spellings are needed because
CONFIG_TASK_FORK is optional on the nuttx side: ARCH_HAVE_TASK_FORK says the
architecture can clone a task, TASK_FORK says this build asked for it. A
follow-up removes the fallback once the split has landed.
vfork_test() and fork_test() deliberately have no such fallback. Both check
semantics a pre-split nuttx does not describe -- the parent suspension and the
private copy -- and ARCH_HAVE_VFORK is the evidence that the split has landed.
Mapping them onto ARCH_HAVE_FORK would also add a test to configurations that
never had one, which is not free: vfork.c costs about 470 bytes of .text on
armv7-m at -Os, and that is what put lm3s6965-ek:qemu-protected over its 128
KiB user flash region.
Signed-off-by: Marco Casaroli <marco.casaroli@gmail.com>
Assisted-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Replace the bundled libtomcrypt chacha20-poly1305 implementation with an
adapter that drives the NuttX crypto device: the SSH construction maps onto
CRYPTO_CHACHA20_DJB (the original 64-bit counter/nonce ChaCha20
parameterization used by chacha20-poly1305@openssh.com) for the packet
length and payload streams, and onto CRYPTO_POLY1305 for the authentication
tag (plain MACs are driven in two steps through /dev/crypto: COP_FLAG_UPDATE
feeds the data, a final call retrieves the tag).
Signed-off-by: Felipe Moura <moura.fmo@gmail.com>
Replace the bundled libtomcrypt HMAC modules with an adapter that computes the hmac-sha2-256 packet MAC through the NuttX crypto device using CRYPTO_SHA2_256_HMAC sessions.
Require CRYPTO, CRYPTO_RANDOM_POOL, CRYPTO_CRYPTODEV and CRYPTO_CRYPTODEV_SOFTWARE_CRYPTO explicitly instead of selecting crypto support.
Signed-off-by: Felipe Moura <moura.fmo@gmail.com>
The Makefile build already forwards both values via PRIORITY /
STACKSIZE. Align the CMake build with the Makefile so that
CONFIG_NETUTILS_REXECD_PRIORITY and CONFIG_NETUTILS_REXECD_STACKSIZE
take effect under CMake as well.
Assisted-by: GitHubCopilot:claude-4.8-opus
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
Migrate passwd encrypt/verify to PBKDF2 modular crypt format using
kernel cryptodev (CRYPTO_PBKDF2_HMAC_SHA256 via /dev/crypto). Add
passwd_pbkdf2 wrapper, base64url helpers, complexity validation, and
pbkdf2_test for RFC 6070 vector coverage. FSUTILS_PASSWD selects
CRYPTO, ALLOW_BSD_COMPONENTS, and CRYPTO_CRYPTODEV so existing sim
defconfigs keep building. Change NSH_LOGIN_USERNAME default to root and
remove fixed-login password defaults.
BREAKING CHANGE: TEA-encoded /etc/passwd entries no longer verify.
Regenerate each entry after upgrading. Pair with the nuttx host mkpasswd
changes in apache/nuttx#19209. Boards must enable the appropriate
software or hardware crypto backend for PBKDF2 at runtime. When
CONFIG_NSH_LOGIN_FIXED=y, set CONFIG_NSH_LOGIN_PASSWORD in the board
defconfig or menuconfig; there is no default password.
Signed-off-by: Abhishek Mishra <mishra.abhishek2808@gmail.com>
Rename base64_encode, base64_decode, and ecc_make_key in bundled
libtomcrypt to avoid duplicate symbol link errors when dropbear is built
alongside netutils/codecs and NuttX crypto (e.g. sim:dropbear).
Signed-off-by: Abhishek Mishra <mishra.abhishek2808@gmail.com>
The per-session NSH task is reaped with waitpid() in
dropbear_nshsession.c. Without CONFIG_SCHED_CHILD_STATUS the kernel does
not retain the child exit status, so waitpid() returns ECHILD right after
authentication and the interactive session never starts
("NSH session wait failed: Unknown error 10").
Depend on SCHED_HAVE_PARENT and SCHED_CHILD_STATUS (following the project's
depends-on-over-select convention) so the requirement is explicit; boards
enabling Dropbear must set these in their defconfig.
Signed-off-by: Felipe Moura <moura.fmo@gmail.com>
Dropbear gathers its entropy from /dev/urandom (libtomcrypt's
rng_get_bytes tries /dev/urandom before /dev/random), so requiring
ARCH_HAVE_RNG and DEV_RANDOM shuts the port out of targets without a
hardware TRNG -- including the simulator, where a dropbear
configuration silently loses NETUTILS_DROPBEAR at configure time.
Keep only the DEV_URANDOM requirement.
Signed-off-by: Felipe Moura <moura.fmo@gmail.com>
recv() returns zero when the peer closes the TCP connection cleanly. The SMTP response path only treated negative returns as errors, so EOF could leave stale or empty response data to be checked as if a server reply had arrived.
Return ERROR on zero-length receives at each SMTP response point and keep the SPDX copyright header within the project line length limit.
Signed-off-by: Old-Ding <ai.neo.ae86@gmail.com>
The current distclean implementation has two issues:
1. Application.mk does not remove .kconfig on distclean, so stale
config fragments from earlier builds can pollute subsequent ones.
2. Several external-library Makefiles (lame, libshvc, libulut) run
"make -C <subdir> distclean" without checking whether the
subdirectory exists. When the library was never downloaded (common
in CI partial-build environments), distclean fails with "No such
file or directory".
Fix:
- Application.mk: add $(call DELFILE, .kconfig) to the distclean
target so that .kconfig is cleaned along with .built, .depend, etc.
- audioutils/lame/Makefile: guard the distclean recipe with
"if [ -d $(DST_PATH) ]; then ...; fi" and use $(MAKE) instead
of bare make.
- netutils/libshvc/Makefile, netutils/libulut/Makefile: same
directory-existence guard for their distclean recipes.
- crypto/wolfssl/Makefile: change "distclean:" to "distclean::"
(double-colon) so it does not override the distclean:: rule
inherited from Application.mk.
Signed-off-by: hanzhijian <hanzhijian@zepp.com>
By default rexecd spawns a detached worker thread per accepted connection
to allow concurrent sessions. Add a "-t" runtime option to instead handle
each connection inline in the main task, avoiding the per-connection
worker stack (CONFIG_NETUTILS_REXECD_STACKSIZE) allocation from the heap.
This matters on low-memory targets with limited free heap.
With "-t", connections are served strictly one at a time: a long-running
or interactive command blocks the accept loop until it finishes. Without
it, the existing thread-per-connection behaviour is unchanged.
Assisted-by: GitHubCopilot:claude-4.8-opus
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
popen() now returns a fopencookie-backed FILE* whose fileno() yields the
cookie pointer rather than a real descriptor, so poll() silently ignores
the negative fd and the relay loop blocks forever waiting for command
completion. See https://github.com/apache/nuttx-apps/pull/3511
Use dpopen()/dpclose() to obtain the raw bidirectional socket descriptor
directly, restoring the EOF/POLLHUP command-completion signal.
Assisted-by: GitHubCopilot:claude-4.8-opus
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
When the relay loop exits early without entering any branch, ret would be
returned uninitialized, tripping -Werror=maybe-uninitialized. Initialize
it to 0.
Assisted-by: GitHubCopilot:claude-4.8-opus
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
Integrated SSH daemon authenticating against FSUTILS_PASSWD, with an
ECDSA P-256 host key and an NSH session over a PTY per connection. Built
from the upstream Dropbear tarball (pinned commit) and patched for
NuttX, using Dropbear's bundled libtomcrypt for all crypto. setsid()
(apache/nuttx#19184) and link() now come from NuttX, not local stubs.
Signed-off-by: Felipe Moura <moura.fmo@gmail.com>
Prior to this change, if an empty folder existed instead of the
actual libwebsockets source, the build would fail. This commit
checks for an actual file instead, avoid such kind of errors.
Signed-off-by: Tiago Medicci <tiago.medicci@espressif.com>
libwebsockets can be built without TLS support. To allow this, it
was necessary to create a specific Kconfig option that enables TLS
support if OPENSSL_MBEDTLS_WRAPPER is enabled. Otherwise, TLS is
not supported (but libwebsockets can still be used with plain ws://
connections).
Signed-off-by: Tiago Medicci <tiago.medicci@espressif.com>
Extend the allocation-failure check to httpd_strdup(), the thttpd-local wrapper that has the same failure contract as strdup().
This is logically separable from the direct strdup()/asprintf() fixes: it prevents using hs->hostname before checking whether the wrapped string allocation succeeded, and releases the partially allocated server state on failure.
The scope intentionally does not extend to malloc(), calloc(), or other raw allocators because #1727 specifically calls out strdup()/asprintf(), and covering all allocation APIs would make this PR much broader.
Signed-off-by: Nightt <87569709+nightt5879@users.noreply.github.com>
Add missing failure handling for direct strdup() and asprintf() calls where the allocated result is consumed locally before any NULL/error check.
This keeps the scope to the functions named in #1727 and avoids changing pass-through return sites where callers already receive NULL on allocation failure.
Signed-off-by: Nightt <87569709+nightt5879@users.noreply.github.com>
Now that time_t is unconditionally 64-bit (signed int64_t) and the
struct timespec fields tv_sec / tv_nsec are wide enough on their own,
the explicit (uint64_t)/(int64_t)/(int) casts that used to guard the
multiplications and subtractions in *_us / *_ms / *_ns helpers are no
longer needed. Drop them to keep the timekeeping math readable.
In the same spirit, this commit also normalises the printf-style format
specifiers and casts used to print tv_sec / tv_nsec / tv_usec values.
The prior code was a mix of "%d"/"%u"/"%ld"/"%lu"/"%lld" with matching
(int)/(unsigned long)/(long long) casts; some formats truncated time_t
on 32-bit hosts, others mismatched signedness or width. Replace all
such cases with the portable POSIX-recommended forms:
- tv_sec (time_t, signed, impl-defined width) -> %jd + (intmax_t)
- tv_nsec (long, signed) -> %ld (no cast)
- tv_usec (suseconds_t / long) -> %ld (no cast)
Also drop two stale `(FAR const time_t *)&ts.tv_sec` casts that are
unnecessary now that ts.tv_sec is plain time_t.
Arithmetic-cleanup files (existing scope):
- benchmarks/cyclictest/cyclictest.c: timediff_us()
- benchmarks/sd_bench/sd_bench_main.c: get_time_delta_us()
- examples/oneshot/oneshot_main.c: maxus computation
- examples/watchdog/watchdog_main.c: current_time_ms (x2)
- industry/nxmodbus/nxmb_internal.h: nxmb_util_clock_ms()
- netutils/ntpclient/ntpclient.c: timespec2ntp()
- netutils/ptpd/ptpd.c: ptp_adjtime()
- system/dd/dd_main.c: elapsed accounting
- testing/drivers/drivertest/drivertest_posix_timer.c:
get_timestamp()
- testing/drivers/sd_stress/sd_stress_main.c:get_time_delta()
- testing/sched/getprime/getprime_main.c: elapsed accounting
- testing/sched/pthread_mutex_perf/pthread_mutex_perf.c:
timespec_avg()
Printf-format-fix files (new in this revision):
- examples/adjtime/adjtime_main.c
- examples/charger/charger_main.c
- examples/netpkt/netpkt_ethercat.c
- fsutils/mkfatfs/mkfatfs.c
- graphics/tiff/tiff_initialize.c
- netutils/ptpd/ptpd.c
- nshlib/nsh_timcmds.c
- system/coredump/coredump.c
- system/ptpd/ptpd_main.c
- testing/drivers/drivertest/drivertest_oneshot.c
- testing/mm/kasantest/kasantest.c
- testing/ostest/semtimed.c
- testing/sched/pthread_mutex_perf/pthread_mutex_perf.c
- testing/sched/timerjitter/timerjitter.c
- testing/testsuites/kernel/time/cases/clock_test_clock01.c
- testing/testsuites/kernel/time/cases/clock_test_smoke.c
No behavioural change.
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
Refactor netlib_check_ipconnectivity() to use gateway/router ping instead
of DNS server ping for network connectivity checks. Add IPv6 support.
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
Update the column ruler and field labels in route_ipv6_entry comment
to match the actual output: lowercase to uppercase, remove colons.
Consistent with Nuttx definition
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
Align with the upstream nuttx change that drops the sclock_t alias now
that clock_t is itself a signed 64-bit type (int64_t). Beyond the pure
sclock_t -> clock_t rename, this commit also removes the
now-unnecessary (time_t)-1 / (time_t)0 / (off_t)-1 casts that were
papering over the previously-unsigned time_t.
Files touched:
- examples/alarm/alarm_main.c: sclock_t -> clock_t
- netutils/dhcpd/dhcpd.c: sclock_t -> clock_t
- netutils/ftpd/ftpd.c: sclock_t -> clock_t
- netutils/thttpd/libhttpd.c,
netutils/thttpd/tdate_parse.c: drop (time_t)-1 / (time_t)0 /
(off_t)-1 sentinel casts
- system/resmonitor/filldisk.c,
system/zmodem/zm_receive.c: sclock_t -> clock_t
- testing/ostest/wdog.c: sclock_t -> clock_t (the bulk
of the rename, 30 sites)
- testing/testsuites/kernel/syscall/cases/{clock_nanosleep_test,
fsync_test,time_test}.c: drop (time_t)-1 casts in the
new signed-time_t world
No behavioural change.
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
NuttX always uses a 64-bit system clock now (time_t/clock_t are
always 64-bit). Remove the obsolete CONFIG_SYSTEM_TIME64 #ifdef
branches and Kconfig dependency.
- examples/dronecan: drop SYSTEM_TIME64 dependency
- examples/netlink_route: always use PRIx64
- netutils/netinit: always use the 1-hour LONG_TIME_SEC
- netutils/ptpd: always pack the 48-bit seconds field and apply
the 64-bit overflow guard in timespec_delta_ns()
- testing/ostest/semtimed: always validate tv_sec >= 0
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
The matching nuttx commit removes CONFIG_HAVE_LONG_LONG; clean up the
remaining users in nuttx-apps so that "long long" is always assumed:
- examples/noteprintf, examples/nxscope:
unconditionally exercise the %lld / long long branch.
- fsutils/mkfatfs/configfat.c:
drop the 32-bit fall-back paths in mkfatfs_nfatsect12/16/32;
always use uint64_t for the FAT-size arithmetic.
- logging/nxscope/nxscope_chan.c:
drop the CONFIG_HAVE_LONG_LONG guard around the 64-bit
sample helpers.
- netutils/ftpd, netutils/ntpclient:
always emit the 64-bit format strings.
- testing/fs/fstest/fstest_main.c:
drop the parallel 32-bit verification path; keep only the
64-bit (long long) random pattern generator.
- system/zmodem/host/nuttx/compiler.h:
remove the host-side CONFIG_HAVE_LONG_LONG stub macro.
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
Use the DHCP-learned NTP server list as the default ntpc server source when DHCP option 42 is available.
Fall back to CONFIG_NETUTILS_NTPCLIENT_SERVER only when DHCP does not provide any NTP servers, and restart ntpc when the DHCP-provided server list changes.
Signed-off-by: Jerry Ma <masc2008@gmail.com>
Request DHCP option 42 and store the returned IPv4 NTP server
addresses in struct dhcpc_state for later consumers.
Signed-off-by: Jerry Ma <masc2008@gmail.com>
Replace app-side includes of <debug.h> with <nuttx/debug.h> to use the
header from the NuttX tree explicitly after the header move.
Signed-off-by: Piyush Patle <piyushpatle228@gmail.com>
Fix mismatch on apps directory naming when it is cloned under 'nuttx-apps' instead of 'apps'.
Signed-off-by: Filipe Cavalcanti <filipe.cavalcanti@espressif.com>
Modified CMakeLists.txt, Kconfig, and Makefile to change the cJSON download URL to include 'refs/tags/' for proper version fetching. This ensures compatibility with GitHub's archive download requirements.
Signed-off-by: Arjav Patel <arjav1528@gmail.com>
Modified the CMakeLists.txt and Makefile to ensure the correct format for downloading the cJSON tarball from GitHub by including 'refs/tags/' in the URL. This change addresses issues with fetching the correct version of the library.
Signed-off-by: Arjav Patel <arjav1528@gmail.com>
removes redundant type casts in ARP-related network library functions.
The casts from req.arp_dev (already a char array) to (FAR char *)
are unnecessary and can be safely removed.
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
When using AF_RPMSG (-r option), rexecd was setting the rp_name
to the raw port number, while rexec was using htons(port). This
caused a mismatch where rexec couldn't connect to rexecd.
This commit fixes the issue by applying htons() to REXECD_PORT
when setting rp_name in the AF_RPMSG case, making it consistent
with the rexec client implementation.
Signed-off-by: fangpeina <fangpeina@xiaomi.com>
Fix cast from pointer to integer of different size to ensure
proper type conversion and avoid compilation warnings.
Signed-off-by: hongfengchen <hongfengchen@xiaomi.com>
Add missing header files in netutils and lsan modules to fix
implicit function declarations and improve compilation compatibility.
Signed-off-by: hongfengchen <hongfengchen@xiaomi.com>
BARE is a simple binary representation for structured application data.
cbare, the BARE implementation ported by this commit, is an I/O agnostic
C implementation of the BARE message encoding.
Signed-off-by: Jiri Vlasak <jvlasak@elektroline.cz>
NNG depends on options that, if not enabled, makes compilation or
runtime crashes. Noting these options in help helps to find out which
are such options.
Signed-off-by: Jiri Vlasak <jvlasak@elektroline.cz>
standardize the implementation of ping6 to better comply with socket
permissions on certain systems.
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
tests only supports tests in non-encrypted mode. when we open tests
compilation in other modes, there will be many compilation warning
with mismatched parameter types, and it will not run correctly.
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
In some scenarios, dynamic memory allocation is not allowed, so it is modified to a static allocation method.
Signed-off-by: gaohedong <gaohedong@xiaomi.com>
Improve the behavior of the ARP table so that the manually configured
ARP table has the highest priority, can only be manually modified to
other values, otherwise will never change again, and will not time out.
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
