19242: usbus/msc: add initial Mass Storage Class support r=benpicco a=dylad
### Contribution description
This PR adds the initial support for Mass Storage Class in USBUS. This PR relies on the RIOT MTD implementation to implement the Mass Storage Class support. With the provided test application, a MTD device will be accessible as a normal storage device on your host computer.
Read and Write operations are allowed.
Multiple LUNs are supported so several MTD devices can be exported through USB.
The MSC relies on SCSI protocol to operate.
Currently there are some limitations:
Supported host : Linux & Windows (macOS is untested)
MSC cannot be used if MTD page size > 4096
MTD device must have at least 512 bytes of memory to be exported.
Please be aware that performance are not so great.
### Testing procedure
Flash `tests/usbus_msc` application on a board with at least one MTD device.
Once the shell has started, prepare one or several MTD devices to be exported using `add_lun` command.
Once ready, start the USB connection with `usb_attach`
All MTD exported should appear as` /dev/sdX` on Linux.
### Issues/PRs references
Supersede #15941
Co-authored-by: Dylan Laduranty <dylan.laduranty@mesotic.com>
19331: pkg/tinydtls: Adjust defaults r=miri64 a=chrysn
### Contribution description
This adjusts two defaults in tinydtls:
* Default verbosity is set to warning. At the info level, this module produces way more output (several lines per new connection, and even per message) than is common in RIOT.
* If gcoap is used, the buffer size is adjusted to the gcoap buffer size plus overhead. Otherwise, CoAP-over-DTLS works fine until one happens to request larger resources.
### Testing procedure
* Run examples/gcoap_dtls
* Send a CoAP request from outside, eg. with `aiocoap-client 'coaps://[fe80::3c63:beff:fe85:ca96%tapbr0]/.well-known/core' --credentials testserver.json` (where testserver.json is `{"coaps://[fe80::3c63:beff:fe85:ca96%tapbr0]/*": {"dtls": {"psk": {"ascii": "secretPSK"}, "client-identity": {"ascii": "Client_identity"}}}}`).
Before, there are messages shown for every request; now there are none.
Modify `examples/gcoap/server.c` as follows:
```patch
diff --git a/examples/gcoap/server.c b/examples/gcoap/server.c
index bf2315cd01..28e1faac27 100644
--- a/examples/gcoap/server.c
+++ b/examples/gcoap/server.c
`@@` -68,7 +68,7 `@@` static const coap_resource_t _resources[] = {
};
static const char *_link_params[] = {
- ";ct=0;rt=\"count\";obs",
+ ";ct=0;rt=\"count\";obs;looooooooooooooooooooooong-attribute=\"loooooooooooooooooooooooooooooong\"",
NULL
};
```
The request passes; without this patch, it is stuck in retransmissions until "Network error: Retransmissions exceeded".
### Issues/PRs references
This contributes to making #19289 usable with a minimum level of security. (That module fills up the gcoap buffer to the brim). While the module handles the verbosity as well as it can (occasionally admitting that it lost bytes of output), the previous verbosity produces an infinite stream of stdout data. (But the default should be quiet immaterial of that particular PR).
Co-authored-by: chrysn <chrysn@fsfe.org>
19321: examples/gnrc_border_router: add BLE as downlink option r=benpicco a=benpicco
19325: esptools/install.sh: Fix shellcheck issues r=benpicco a=bergzand
### Contribution description
Quote all the things!
### Testing procedure
The script should still work as before
### Issues/PRs references
None
19327: shell/cmds: GNRC: replace puts() with printf() r=benpicco a=benpicco
19328: pkg/u8g2: bump version r=benpicco a=benpicco
Co-authored-by: Benjamin Valentin <benjamin.valentin@ml-pa.com>
Co-authored-by: Koen Zandberg <koen@bergzand.net>
Co-authored-by: Benjamin Valentin <benpicco@beuth-hochschule.de>
17091: USBUS: Add URB support r=benpicco a=bergzand
### Contribution description
This commit adds support for URBs (USB Request/Response Blocks). These
allow for submitting multi-transfer sized buffers with USBUS handling
the individual usbdev xmits. Multiple URBs can be queued at once for a
single endpoint and USBUS will handle them in the order of submission.
OUT endpoint URBs must always consist of a whole number of full-sized
transfers (N x MaxEndpointSize). They will automatically finish after
the endpoint received a transfer less than the endpoint size.
IN endpoints can be arbitrary-sized and do not have to consist of a
whole number of full-sized transmissions. They support a flag to
indicate that the last transfer in the sequence must be less than a full
sized transfer (USBUS_URB_FLAG_AUTO_ZLP) and this adds a zero length
transfer at the end of the transmissions if the last transfer was equal
to the maximum transfer size.
URBs can be cancelled, but if the URB is already being processed it will
be cancelled after the current transmission within the URB is finished.
If it is still in the queue it will immediately be removed from the
queue.
### Testing procedure
- `tests/usbus_cdc_ecm` should still work. Testing one of the usbdev-supported platform should be sufficient here.
### Issues/PRs references
Needs #17064
18148: sys/flash_utils: helpers to store data in flash r=benpicco a=maribu
### Contribution description
This helpers that allow storing, accessing, and working with data in flash that works for both classical Harvard architectures (which do not map flash also into the address space) as well as modern Harvard architectures and von-Neumann architectures.
With this, `examples/default` again runs on the Arduino Uno / Nano. Since this board is still the "entry kit" for many people to embedded hardware, it would be nice to support it with our default example.
### Testing procedure
`examples/default` should run and work on ATmega boards (especially ATmega328P and ATmega32U4 based boards) as well on all other boards now.
### Issues/PRs references
None
Co-authored-by: Koen Zandberg <koen@bergzand.net>
Co-authored-by: Marian Buschsieweke <marian.buschsieweke@ovgu.de>
This allows automatically moving format strings to flash, provided that
code previously compiled fine with `-Wformat-nonliteral` (which in RIOT
is the case due to `-Wformat=2`).
This adds a layer of convenience abstraction over classical Harvard
architectures (like most AVRs) that do not map the flash memory into
the data address space and modern Harvard architectures or von-Neumann
architectures that do so. The motivation is to safe a lot of RAM for
AVR by storing constant strings into flash.
This commit adds support for URBs (USB Request/Response Blocks). These
allow for submitting multi-transfer sized buffers with USBUS handling
the individual usbdev xmits. Multiple URBs can be queued at once for a
single endpoint and USBUS will handle them in the order of submission.
OUT endpoint URBs must always consist of a whole number of full-sized
transfers (N x MaxEndpointSize). They will automatically finish after
the endpoint received a transfer less than the endpoint size.
IN endpoints can be arbitrary-sized and do not have to consist of a
whole number of full-sized transmissions. They support a flag to
indicate that the last transfer in the sequence must be less than a full
sized transfer (USBUS_URB_FLAG_AUTO_ZLP) and this adds a zero length
transfer at the end of the transmissions if the last transfer was equal
to the maximum transfer size.
URBs can be cancelled, but if the URB is already being processed it will
be cancelled after the current transmission within the URB is finished.
If it is still in the queue it will immediately be removed from the
queue.
18682: pkg/lwext4: add lightweight implementation of the ext2/3/4 filesystem r=benpicco a=benpicco
Co-authored-by: Benjamin Valentin <benjamin.valentin@ml-pa.com>
19119: makefiles/tools/serial.inc.mk: Allow detection of debug adapter r=benpicco a=maribu
### Contribution description
This PR adds the ability to automatically detect the debug adapter for boards with an integrated programmer/debugger, if that debugger also provides the TTY.
This extends the TTY detection that can be enabled with `MOST_RECENT_PORT=1` to set `DEBUG_ADAPTER_ID` to the TTY's serial, but only if `DEBUG_ADAPTER_ID_IS_TTY_SERIAL` is set to `1` by the board (as not all boards have an integrated programmer/debugger).
### Testing procedure
Connect a HiFive-1B and a nRF52840DK at the same time and try `make BOARD=<nrf52840dk|hifive1b> MOST_RECENT_PORT=1 -C examples/default flash term` for both. The programmer will not reliably select the correct programmer in `master`. With this PR, it will.
### Issues/PRs references
None
19313: gnrc_static: don't parse address as prefix r=benpicco a=benpicco
Co-authored-by: Marian Buschsieweke <marian.buschsieweke@ovgu.de>
Co-authored-by: Benjamin Valentin <benpicco@beuth-hochschule.de>
19292: sys/phydat: Fix unit confusion r=miri64 a=maribu
### Contribution description
Previously, `UNIT_G` was used for g-force with the correct symbol `g`, `UNIT_GR` for gram (as in kilogram) with the incorrect symbol `G` (which would be correct for Gauss), and `UNIT_GS` for Gauss with symbol `Gs` (which is an alternative correct symbol).
To avoid confusion between G-Force, Gauss, and Gram the units have been renamed to `UNIT_G_FORCE`, `UNIT_GRAM`, and `UNIT_GAUSS`. In addition, gram now uses the correct symbol `g`; which sadly is the same as for g-force. But usually there is enough context to tell them apart.
### Testing procedure
Green CI
### Issues/PRs references
None
19307: nanocoap_link_format: fix off-by-one error r=miri64 a=benpicco
Co-authored-by: Marian Buschsieweke <marian.buschsieweke@ovgu.de>
Co-authored-by: Benjamin Valentin <benpicco@beuth-hochschule.de>
18746: sys/clif: Fixing out of bounds read under certain conditions r=maribu a=Teufelchen1
Hi 😈
This fixes a potential out of bounds read in clif_encode_link. There is no code in RIOT that can be exploited.
The fix does not break the current API but alters the behaviour slightly. Before the change, the length attributes of `clif_attr_t` where optional. If missing, the length was deduced using `strlen()`. This fix makes those parameters required and if they are `0` it operates as if the length really is `0`. This might not be ideal but it is the only non api breaking fix I could think off.
```c
typedef struct {
char *value;
unsigned value_len; NO LONGER OPTIONAL
const char *key;
unsigned key_len; NO LONGER OPTIONAL
} clif_attr_t;
```
Depends on #18744
cc `@leandrolanzieri`
19161: bors.yaml: re-activate labels check + add block_labels r=miri64 a=miri64
Co-authored-by: Teufelchen1 <bennet.blischke@haw-hamburg.de>
Co-authored-by: Martine Lenders <m.lenders@fu-berlin.de>
Previously, `UNIT_G` was used for g-force with the correct symbol `g`,
`UNIT_GR` for gram (as in kilogram) with the incorrect symbol `G` (which
would be correct for Gauss), and `UNIT_GS` for Gauss with symbol `Gs`
(which is an uncommon but correct symbol).
To avoid confusion between G-Force, Gauss, and Gram the units have been
renamed to `UNIT_G_FORCE`, `UNIT_GRAM`, and `UNIT_GAUSS`. In addition,
gram now uses the correct symbol `g` and Gauss uses `G`.
19299: Add missing newline to output of telnet example r=benpicco a=Enoch247
### Contribution description
From the commit msg:
> The telnet example prints a line to the console, but it is not ended with a newline. When using pyterm, the last line is then never shown as it reads the console line by line and is waiting for the end of the line.
>
> This patch swaps use of `printf` for `puts` for the last line printed. This means the missing newline character gets added. This is also done to be consistent with the rest of the file, where puts is used whenever possible instead of printf.
### Testing procedure
1. `cd examples/telnet`
2. `make all flash term`
3. observe the final line "Local shell disabled" is printed, but would not have without this patch
### Issues/PRs references
- none known
19301: fib: document unused state r=benpicco a=benpicco
Co-authored-by: Joshua DeWeese <jdeweese@primecontrols.com>
Co-authored-by: Benjamin Valentin <benjamin.valentin@bht-berlin.de>
19294: sys/shell: don't include suit command by default r=benpicco a=benpicco
19295: gcoap: Finish the gcoap_get_resource_list_tl -> gcoap_get_resource_list renaming r=benpicco a=chrysn
### Contribution description
In #16688, an argument was added to the `gcoap_get_resource_list` function by creating a new function `gcoap_get_resource_list_tl` with a deprecation and roll-over plan.
This plan has not been acted on so far.
This PR shortens the original plan by just adding the argument to `gcoap_get_resource_list` and removing `gcoap_get_resource_list_tl` in a single go. The rationale for this deviation is that while it's a public API, its only two practical consumers are the (built-in) well-known/core implementation, and the (built-in) CoRE Resource Directory (cord) endpoint. Moreover, a further change to this API (switching over to `coap_block_slicer_t`) is expected to happen within this release cycle, which would take something like 4 total releases to get through otherwise, which is unrealistic for an API that there are no known external users of.
A second commit clean up ToDo items (in the changed function's documentation) that referred to a IETF draft that has long been abandoned by the CoRE WG.
### Testing procedure
Plain inspection and CI passing should suffice.
### AOB
There is a second analogous pair left over from #16688, `gcoap_req_send` / `gcoap_req_send_tl`. As that *is* expected to be used widely, I prefer not to mix these two concerns, and get the present one through without unnecessary hold-up.
Co-authored-by: Benjamin Valentin <benjamin.valentin@bht-berlin.de>
Co-authored-by: chrysn <chrysn@fsfe.org>
This is an API change in the latter, which would typically now take an
extra argument GCOAP_SOCKET_TYPE_UNDEF.
Follow-Up-For: https://github.com/RIOT-OS/RIOT/pull/16688
19278: gnrc_dhcpv6_client_simple_pd: select upstream based on type/index r=benpicco a=benpicco
19290: sys/crypto: make AES_KEY struct private & rename it r=benpicco a=benpicco
Co-authored-by: Benjamin Valentin <benjamin.valentin@ml-pa.com>
Co-authored-by: Benjamin Valentin <benjamin.valentin@bht-berlin.de>
* riot-wrappers:
* Fix infinite loop when using a Mutex
* Make ValueInThread Copy/Clone
* riot-sys:
* Export xxx_DEV (eg. I2C_DEV) C macros as functions
* Add auto_init_utils.h
19142: sock_dtls: move common code into sock_dtls_establish_session() r=benpicco a=benpicco
Co-authored-by: Benjamin Valentin <benpicco@beuth-hochschule.de>
19027: sys/fmt: optimize scn_u32_dec scn_u32_hex r=benpicco a=kfessel
### Contribution description
Improves the compilation result for `scn_u32_dec` `scn_u32_hex` especially on `cortex-m` reducing either stack usage and or code size.
This makes use of unsigned int overflow (slightly less better without doing that `hexn`).
See godbolt (original versions got an `o` attached, modified versions got `k`s) all functions are marked `_S_` defined to `static`
by assigning the global at end the compiled function can be changed (`deco deck hexo hexk hexkk hexn`)
this PR is `hexkk` and `deck`
### Testing procedure
run unit-test/test-fmt
```
<RIOT>/tests/unittests$ make tests-fmt
<RIOT>/tests/unittests$ make term
```
### Issues/PRs references
[godbolt](https://godbolt.org/z/MzT1zh4q1)
Co-authored-by: Karl Fessel <karl.fessel@ovgu.de>
