The clock adjustment API only used a 16 bit integer for speed correction. This
is to course grained to allow compensating clock drifts at high accuracy.
Using a 32 bit integer instead would allow to fix for a drift of up to
about 1 nanosecond drift per each 5 seconds.
That ought to be enough for anyone! (*cough* *cough*)
FEATURES_CONFLICTING previously was declared prior to the function it is
calling, resulting in empty output during the first dependency resolution
iteration. This fixes the order so that the conflicting features are detected
right from the first recursion.
This allows using the macro inside the periph_conf.h board files since the
periph/gpio.h header can't be included on the peripheral configuration.
Signed-off-by: Jean Pierre Dudey <me@jeandudey.tech>
Currently, when the prompt is read in `pyterm` the space after it is
ignored for the prompt and the output command just adds its own prompt.
This leads to the next output always having a leading space, see e.g.
this output from `tests/shell` using `RIOT_TERMINAL=pyterm`:
```
make: Entering directory '/home/mlenders/Repositories/RIOT-OS/RIOT2/tests/shell'
/home/mlenders/Repositories/RIOT-OS/RIOT2/dist/tools/pyterm/pyterm -p "/dev/ttyUSB1" -b "500000"
Twisted not available, please install it if you want to use pyterm's JSON capabilities
2021-02-09 14:47:15,071 # Connect to serial port /dev/ttyUSB1
Welcome to pyterm!
Type '/exit' to exit.
bufsize
2021-02-09 14:47:19,712 # bufsize
2021-02-09 14:47:19,712 # 128
> bufsize
2021-02-09 14:47:21,535 # bufsize
2021-02-09 14:47:21,536 # 128
>
```
While this isn't necessarily a problem in most cases, it becomes a
problem when the prompt is expected and the output of a command is
empty. In that case, the space is added to the empty output, making it
" ", so the prompt output command is never triggered and the prompt is
added to the next command in the log output. To demonstrate I added a
command `empty` to `tests/shell` that just does nothing and deactivated
the command echoing using `CFLAGS=-DCONFIG_SHELL_NO_ECHO=1`:
```
empty
> empty
empty
bufsize
2021-02-09 14:54:33,753 # > > 128
>
```
This fixes that problem by also reading the assumed space (we already
assume the prompt, so I don't see no harm in that) and if it is not a
space to skip the reading of the next char in the next iteration of the
reader loop.
The USB bootloader can take a while to be ready / Linux takes a while
to set up the interface.
1s is right at the edge, leading to races and often failed flashing.
Bump the delay to 2s for a wider safety margin, with this flashing is
a lot more reliable.
