We occasionally have some public `foo.h` header that includes a private
`foo_arch.h` header. Users are expected to include the `foo.h` header
and not the `foo_arch.h`. However, clangd will claim that the `#include`
of `foo.h` is unused if only functions / macros/ types / ... from
`foor_arch.h` is used and nothing from `foo.h`.
This adds the `IWYU pragma: export` comment to the include of
`foo_arch.h` in `foo.h`, so that clangd treats functions / macros /
types provided by `foo_arch.h` as if they were instead provided by
`foo.h`, which fixes the false positives.
If a server has e.g. multiple global IPv6 addresses, it might
respond with a different address that the one it received the request on.
This is against the CoAP spec, but that doesn't stop existing implementations
from doing it.
It would seem that either we need to require the periph_pm module in shell
or make it optional... since we have many other optional modules here
and we still may want the RIOT_VERSION command, lets make it optional for now.
This has the following advantages:
- faster and leaner when C11 atomics are not efficient (e.g. on LLVM
this is almost always the case, as LLVM will only use efficient
atomics if it doesn't has to bail out to library calls even for
exotic things)
- Even for GCC e.g. on the nucleo-f429zi this safes 72 B of .text
for examples/filesystem despite runtime checks added for
over- and underflow
- less pain in the ass for C++ and rust users, as both C++ and
c2rust are incompatible with C11 atomics
- adds test for overflow of the open file counter for more robust
operation
- adds `assumes()` so that underflows are detected in non-production
code
The function gomach_vtdma_end is nearly identical to _no_vtdma_after_cp
expect the first few lines. This commit replaces the duplicating code
in gomach_vtdma_end by a call to _no_vtdma_after_cp.
Signed-off-by: Francois Berder <fberder@outlook.fr>
Both _rx_management_failed and _rx_management_success functions
attempt to sleep after handling the packet reception
failure/success. This commit extracts the sleep attempt in a
new _rx_management_attempt_sleep function.
Signed-off-by: Francois Berder <fberder@outlook.fr>
The Cargo.lock of rust_riotmodules_standalone should explicitly list all
versions of software that would ever be pulled in by enabling RIOT
modules backed by a Rust crate. That way, the Cargo.lock file will not
spontaneously change because the software is being built with a
different set of modules enabled.
This did not happen before, because while the optional dependencies
could all be selected on the command line, they were not visible as
possible features to `cargo update`, and were thus not encoded in the
Cargo.lock file.
Mixing address spaces is something LLVM doesn't like (for good
reason). This re-organized the code a bit so that this does not
happen anymore, even on AVR.
Split out of https://github.com/RIOT-OS/RIOT/pull/16924
This eagerly calls `__sinit()` instead of lazy initialization upon the
first call to stdio (e.g. `puts()`, `printf()`). The issue is that
without locking (as is currently the case for all RIOT platforms but
ESP) two concurrent "first calls" may result in concurrent
initialization of the same structure and data corruption.
Fixes https://github.com/RIOT-OS/RIOT/issues/20067
In order to properly handle an observe cancellation of a client, the server has to keep track of the notification MIDs (to be able to match an RST to a notification), see [RFC7641, 3.6 Cancellation](https://www.rfc-editor.org/rfc/rfc7641.html#section-3.6) for mor details. An alternative to this would be to make either the client send an explicit observe deregister request, or make the server send the next notification via CON (which hten allows matching of the RST due to the CON state).
This splits the _find_req_memo util function into multiple variants that match on different things. This is done in preparation of a feature that has to find a request based on a token value, without creating an artificial pdu for that. A nice side effect is that it also makes the calls to the find functions a bit more readable by not relying on an anonymous bool input.
Adds a separate board for native64 instead of the `NATIVE_64BIT` workaround.
The files in `boards/native64` are more or less dummy files and just include
the `boards/native` logic (similar to `openlabs-kw41z-mini-256kib`).
The main logic for native is in `makefiles/arch/native.inc.mk`, `cpu/native`
and `boards/native`.
The remaining changes concern the build system, and change native board checks
to native CPU checks to cover both boards.
This allows to define an extra string that will be output as part
of the version command.
e.g. an application may define
RIOT_VERSION_EXTRA += v1.33.7
RIOT_VERSION_EXTRA += flashed by $(shell whoami
This changes the implementation to be solely build upon `endian.h`
and `unaligned.h`.
This turns `byteorder.h` basically in syntactic sugar on top of the
`<endian.h>` API, reducing the complexity of the implementation and,
hence, the maintenance effort.
Note that yields a small ROM reduction as well *yeah!*
```
make BOARD=nrf52840dk RIOT_CI_BUILD=1 BUILD_IN_DOCKER=1 -C tests/unittests
```
Yields before this commit:
```
text data bss dec hex filename
417788 2200 28640 448628 6d874 /data/riotbuild/riotbase/tests/unittests/bin/nrf52840dk/tests_unittests.elf
```
And with this commit:
```
text data bss dec hex filename
417756 2200 28640 448596 6d854 /data/riotbuild/riotbase/tests/unittests/bin/nrf52840dk/tests_unittests.elf
```
This pulls in embedded-nal 0.7 implementations, and (on both crates)
alterations that decouple riot-wrappers from riot-sys and simplify build
system integration, and do away with the need for passing
RIOT_USEMODULE.
Older versions of newlib already provide the magic endian numbers
via `machine/endian.h`, which may be indirectly included. This changes
the header to only provide the macros if the are not provided otherwise.
For sanity, it checks if the values are indeed the expected magic
numbers, even if provided from other sources.
The constants BIG_ENDIAN etc. were not consistently defined. This
bug was not caught by the unit tests, as the preprocessor would
treat undefined macros as being `0` in numeric comparisons, hence
the following test did select the correct implementation:
```C
#if BYTE_ORDER == LITTLE_ENDIAN
/* correct implementation for all currently supported boards */
#endif
```
This adds now an explicit test to the unit tests to ensure that the
magic numbers are consistently the correct values. Hence, this bug
should no longer be able to sneak in.
Co-authored-by: Teufelchen <9516484+Teufelchen1@users.noreply.github.com>
This provides glibc, NetBSD, FreeBSD compatible endian.h header with a
lean and simple API to convert between host byte order to little endian
and big endian and the other way around.
* priority_queue_t: Replaced `-1U` literal with PRIORITY_QUEUE_DATA_SIGNALING define
* architecture.h: Added 64-bit
* bloom.h: Fixed typedef for the hashfp_t function pointer
* vfs.h: Increased default vfs buffer sizes for 64 bit
* bytes.h: Check if socklen_t is already defined
* ztimer: Use PRIxPTR format specifier
This removes the static (thread-unsafe) variables from sha256 and
hmac_sha256 to remove a potential footgun. The static variable is only
used when the caller does not supply a pointer to store the digest and
it is returned via the (undocumented) return value.
This commit removes this option and makes the digest argument mandatory.
When riotboot/slot.h is included, but riotboot is not enabled in the
build, a compiler error occures. This is because SLOT0_LEN is not
defined. This patch fixes this by surrounding the offending macrro with
a conditional compile. `riotboot_slot_size()` may still be called, in
which case it returns 0 for the size of any slot given.
This patch updates shell.c's doc that undersells the shells complexity.
The comment seems to have been written prior to the shell's ability to
parse command args and handle quoting sequences.
This patch adds the missing NULL terminator to the argv passed to shell
command handlers. Without it, Newlib's getop() was intermittently
causing hardfaults. Closer inspection of NewLib's code revealed that it
relies in this NULL termination. ANSI-C also requires it of the argv
passed to main().
sock_udp_recv_buf_aux() sometimes will return -ETIMEDOUT before
the given timeout has expired (e.g. 28798µs instead of 160000µs).
This messes with many assumptions and breaks protocols that rely
on the timeout.
Until we have a proper fix, add this workaround.
The function signature in the Doxygen generated doc was incorrect.
Given that the implementation does have the correct signature, the
impact may not be that large. Still, incorrect doc is confusing, so
let's fix it.
20003: coap: create typedef for CoAP methods r=benpicco a=benpicco
20028: nanocoap: fix incorrect uses of coap_get_code() r=benpicco a=benpicco
20036: boards/nucleo-l4r5zi : add PWM configuration r=benpicco a=krzysztof-cabaj
### Contribution description
This PR adds to the Nucleo-l4r5zi PWM configuration
### Testing procedure
Flash the board using `tests/periph/pwm` program. Check if you could, for example, change LED
intensity using PWM.
### Issues/PRs references
None.
20041: pkg/littlefs2: bump to v2.8.1 r=benpicco a=bergzand
### Contribution description
Minor version number bump.
### Testing procedure
- Check if I copied the correct commit hash.
- Run the tests.
### Issues/PRs references
None
Co-authored-by: Benjamin Valentin <benjamin.valentin@ml-pa.com>
Co-authored-by: krzysztof-cabaj <kcabaj@gmail.com>
Co-authored-by: Koen Zandberg <koen@bergzand.net>
19999: nib/_nib-6ln: bail out early if address is no longer assigned r=miri64 a=benpicco
Co-authored-by: Benjamin Valentin <benjamin.valentin@ml-pa.com>
19990: sys/psa_crypto: allow repeated initialization r=benpicco a=mguetschow
### Contribution description
- simple unit test which calls `psa_crypto_init()` twice
- fix to no re-initialize key slots (which left them in a broken state)
### Testing procedure
- `make -C tests/sys/psa_crypto all test` succeeds
- `git checkout HEAD~1 && make -C tests/sys/psa_crypto all test` fails
Co-authored-by: Mikolai Gütschow <mikolai.guetschow@tu-dresden.de>
19465: drivers/mtd: use XFA for pointers to defined MTDs r=benpicco a=gschorcht
### Contribution description
This PR provides the support to hold pointers to defined MTDs within a XFA. The XFA allows
- to access MTDs of different types (`mtd_flashpage`, `mtd_sdcard`, `mtd_emulated`, ...) by an index
- to determine the number of MTDs defined in the system.
### Testing procedure
To be defined once PR #19443 is merged because emulated MTDs will allow to test this PR on arbitrary boards.
### Porting Guide
For external boards:
- remove the `MTD_NUMOF` definition from `board.h`
- add `MTD_XFA_ADD(<mtd_dev>, <idx>);` to the definition of `<mtd_dev>`.
- `MTD_0`, `MTD_1`, … defines are no longer needed.
### Issues/PRs references
Related to PR #19443
19981: Fletcher32: Add incremental API r=benpicco a=bergzand
### Contribution description
This PR extends the current fletcher32 checksum with an incremental API mode. This way the bytes to be checksummed can be supplied via multiple successive calls and do not have to be provided in a single consecutive buffer.
I've also rephrased the warning with the original function a bit as that function uses an `unaligned_get_u16` to access the data. The data thus does not require alignment, but the length does need to be supplied as number of 16 bit words.
### Testing procedure
The test has been extended
### Issues/PRs references
None
19995: sys/psa_crypto: Fix macro for public key max size and SE example r=benpicco a=Einhornhool
### Contribution description
#### 1. Wrong public key size when using secure elements, introduced by #19954
Fixed conditions for key size macros in `crypto_sizes.h`.
#### 2. EdDSA and ECDSA examples fail when using a secure element because of unsopported changes introduced by #19954
Updated `example/psa_crypto` to use only supported functions for secure elements.
### Testing procedure
Build `example/psa_crypto` for secure elements and run application
Output on master:
```
2023-10-19 14:33:24,372 # main(): This is RIOT! (Version: 2019.07-devel-22378-gb6772)
2023-10-19 14:33:24,372 # HMAC SHA256 took 56393 us
2023-10-19 14:33:24,372 # Cipher AES 128 took 68826 us
2023-10-19 14:33:24,372 # *** RIOT kernel panic:
2023-10-19 14:33:24,373 # HARD FAULT HANDLER
2023-10-19 14:33:24,373 #
2023-10-19 14:33:24,373 # *** rebooting...
```
Output with fixes:
```
2023-10-19 13:35:24,715 # main(): This is RIOT! (Version: 2019.07-devel-22384-g8ef66-dev/psa-crypto-fixes)
2023-10-19 13:35:24,715 # HMAC SHA256 took 56374 us
2023-10-19 13:35:24,715 # Cipher AES 128 took 68805 us
2023-10-19 13:35:24,715 # ECDSA took 281164 us
2023-10-19 13:35:24,715 # All Done
```
Co-authored-by: Gunar Schorcht <gunar@schorcht.net>
Co-authored-by: Koen Zandberg <koen@bergzand.net>
Co-authored-by: Lena Boeckmann <lena.boeckmann@haw-hamburg.de>
19963: sys/event/timeout: add event_timeout_is_pending() r=benpicco a=benpicco
19971: sys/shell/gnrc_netif: Fix a few blockers for sharing ifconfig shell r=benpicco a=yarrick
19974: gnrc_ipv6_ext_frag: _completed: Add comment why list head is not checked for NULL pointer dereference r=benpicco a=miri64
19975: pkg/nanocbor: Bump to latest commit r=benpicco a=bergzand
### Contribution description
Not much to see here
Important changes:
- Add stream-like interface for encoder
- Separate functions for number of items left in arrays and maps
### Testing procedure
The usual test should still work
### Issues/PRs references
None
19976: core: Express -1 as ~0 in thread_status_t cast r=benpicco a=SimonIT
Co-authored-by: Benjamin Valentin <benjamin.valentin@ml-pa.com>
Co-authored-by: Erik Ekman <eekman@google.com>
Co-authored-by: Martine Lenders <m.lenders@fu-berlin.de>
Co-authored-by: Koen Zandberg <koen@bergzand.net>
Co-authored-by: SimonIT <simonit.orig@gmail.com>
