93 lines
3.7 KiB
Plaintext
93 lines
3.7 KiB
Plaintext
------------------------------------------------------------------------------
|
|
HACKING
|
|
------------------------------------------------------------------------------
|
|
|
|
Coding style
|
|
------------
|
|
|
|
The whole library is programmed using the Linux kernel coding style, see
|
|
http://lxr.linux.no/linux/Documentation/CodingStyle for details.
|
|
|
|
Please use the same style for any code contributions, thanks!
|
|
|
|
Amendments to the Linux kernel coding style
|
|
-------------------------------------------
|
|
|
|
1) We use the stdint types. The linux kernel accepts the abbreviated types (u8,
|
|
s8, u16 and so on) for legacy reasons. We should in general not introduce
|
|
things like types ourselves as long as they are not necessary to make our
|
|
job possible of refining the hardware and make it easier to be used. stdint
|
|
is a standard and it is not in the scope of our project to introduce a new
|
|
type standard.
|
|
|
|
2) Based on the same logic as in (1) we do not use __packed and __aligned
|
|
definitions, it is not our job to add compiler extensions. If we need to
|
|
deal with compiler incompatibility we will do that the same way we are
|
|
dealing with the depricated attribute by introducing a normal macro that is
|
|
not in the compiler reserved keyword space.
|
|
|
|
3) We accept to write an empty body busy waiting while loop like this:
|
|
while (1);
|
|
there is no need to put the colon on the next line as per linux kernel
|
|
style.
|
|
|
|
4) We always add brackets around bodies of if, while and for statements, even
|
|
if the body contains only one expression. It is dangerous to not have them
|
|
as it easily happens that one adds a second expression and is hunting for
|
|
hours why the code is not working just because of a missing bracket pair.
|
|
|
|
Development guidelines
|
|
----------------------
|
|
|
|
- Every new file added must have the usual license header, see the
|
|
existing files for examples.
|
|
|
|
- In general, please try to keep the register and bit naming as close
|
|
as possible to the official vendor datasheets. Among other reasons, this
|
|
makes it easier for users to find what they're looking for in the
|
|
datasheets, programming manuals, and application notes.
|
|
|
|
- All register definitions should follow the following naming conventions:
|
|
|
|
- The #define names should be all-caps, parts are separated by
|
|
an underscore.
|
|
|
|
- The name should be of the form SUBSYSTEM_REGISTER_BIT, e.g.
|
|
ADC_CR2_DMA, where ADC is the subsystem name, CR2 is the register NAME,
|
|
and DMA is the name of the bit in the register that is defined.
|
|
|
|
- All subsystem-specific function names should be prefixed with the
|
|
subsystem name. For example, gpio_set_mode() or rcc_osc_on().
|
|
|
|
- Please consistently use the short form types from <cm3/common.h>,
|
|
e.g. u8, u16, u32, and so on.
|
|
|
|
- Variables that are used to store register values read from registers or
|
|
to be stored in a register should be named reg8, reg16, reg32 etc.
|
|
|
|
- In the examples directory, the following structure should be used:
|
|
|
|
- One (or more) subdirectories for the type of microcontroller, e.g.
|
|
lm3s, lpc13xx, stm32/f1, stm32/f2, stm32/f4.
|
|
|
|
- One subdirectory in there for each eval board or piece of hardware, e.g.
|
|
stm32-h103, lisa-m, stm32vl-discovery, stm32f4-discovery, etc.
|
|
|
|
- One subdirectory in there for each example, e.g.
|
|
miniblink, button, usart, usb_dfu, etc.
|
|
|
|
Tips and tricks
|
|
---------------
|
|
|
|
SublimeText users:
|
|
|
|
- The project contains a sublime project description file with some basic
|
|
settings provided to make hacking on libopencm3 easier.
|
|
|
|
- Recommended SublimeText plugins when hacking on libopencm3:
|
|
|
|
- TrailingSpaces: Show and trim trailing line spaces.
|
|
|
|
- SublimeLinter: Run checkpatch.pl in the background while you write your
|
|
code and indicate possible coding style issues on the fly.
|