The flash configuration is something very board specific. So move the
file to a board specific location. Beneath that, not all controllers and
configurations need the flash config right at the address 0 of the
flash. For example on the i.MXRT11xx, the config has an offset for some
flash types.
The imx-gpio driver used in i.MX and i.MXRT BSPs generates a name based
on a fixed string. The original code only used one digit for the
controller. With the 13 GPIO controllers of the i.MXRT1166, that isn't
enough any more. This patch extends the name to two digits which should
be enough for the next controller generations.
CAVEAT: license change from BSD to Apache2 license!
Explanation:
The imported files come from CMSIS v5 project available on:
https://github.com/ARM-software/CMSIS_5/tree/develop
The files imported are located inside the CMSIS/Core/Include
project sub-directory.
The project does not provide any NOTICE file in its root directory nor
in the directory of the imported files.
The NOTICE file and its usage in the Apache 2 license was/is
so far the only issue mentioned in discussion of RTEMS developers/users
when considering inclusion of the code under Apache 2 license
into the RTEMS project.
Since the CMSIS v5 project is free from this legal hinder, we may freely
use it and update files to the latest version.
Technical: the patch replaces code from 2015 with the latest version
which brings quite a lot of bug fixes and most importantly opens
possibilities to support MCUs based on new ARM cores.
This adds support for the STM32H750B-DK discovery kit. This kit includes
a built-in STLINKv3 debugger which provides a USB serial bridge for
USART3. USART1 is routed to the Arduino header and USART2 is routed to
the STMOD connector. This BSP reuses what would otherwise be duplicated
files from the stm32h747i-disco BSP. Note that system_stm32h7xx.c has
been imported from the STM repository with two minor changes wrapped
with #if __rtems__. This hardware has been tested with hello and ticker.
rtems_task_wake_after takes a parameter in terms of a count of clock
ticks and not a measure in a subunit of seconds. This updates
documentation to reflect that. This also makes obvious the caveat about
the first tick wait not being a whole tick and points the user at a
replacement for better accuracy.
Updates #4772
Remove the BSP_POWER_DOWN_AT_FATAL_HALT BSP option. Applications should
do the customization of the system termination with an initial fatal
extension.
