Unified Firmware

[antoinevg]

This PR introduces a new BOARD target called UNIVERSAL that can be used to compile a combined firmware that supports both HackRF One and HackRF Pro hardware.

Conventions

It introduces a new pattern to the code base for working with board support.

Previously, we used a mix of patterns of the form #ifdef HACKRF_ONE or #ifndef PRALINE in conjunction with platform_detect.detected_platform() to manage board differences.

This PR applies the following new patterns across the codebase:

• Conditional directives now only serve as guards for what code gets linked into the firmware for a given board.
• To control code execution always use if (detected_platform() == BOARD_ID_x) or switch (detected_platform())
• Conditional directives of the form #ifdef SOME_BOARD become #if defined(SOME_BOARD) or #ifndef SOME_BOARD becomes #if !defined(SOME_BOARD). This keeps usage consistent with the next point.
• Specifically, for HACKRF_ONE and PRALINE always include the UNIVERSAL target to ensure the code will be present in the combined firmware binary. For example: #if defined(PRALINE) || defined(UNIVERSAL)
• Be careful of #else, this caught me a few times:

#if defined(PRALINE) || defined(UNIVERSAL)`
    ... some code ...
#else
    ... this code now gets left out of UNIVERSAL ...
#endif

Make sure to use this form instead:

#if defined(PRALINE) || defined(UNIVERSAL)`
    ... some code ...
#endif
#if !defined(PRALINE) || defined(UNIVERSAL)`
    ... this code is now also included in UNIVERSAL ...
#endif

Compilation

mkdir firmware/build
cd firmware/build
cmake .. -DBOARD=UNIVERSAL
make

Size impact

Single-board firmware are slightly bigger as a result:

PRALINE:     was: 240288  => now:  243336  difference: 3048
HACKRF_ONE:  was:  45920  => now:   48672  difference: 2752
RAD1O:       was:  75244  => now:   77548  difference: 2304
JAWBREAKER:  was:  38060  => now:   39988  difference: 1928
JELLYBEAN:   was:    n/a  => now:   39204

The size increase is mostly due to:

• platform_scu: 968 bytes
• platform_gpio: 1044 bytes

…with the remaining being due to runtime platform detection across the rest of the code-base.

The combined firmware comes to 258376 bytes.

Testing

This PR has been tested with both single-board and combined firmware using hackrf.git/ci-scripts/hackrf-test.py and praline-factory-test.git/bin/hackrf_test.py in the following configurations:

EUT: Hackrf One r4, TESTER: HackRF One r10c Portapack H4M clone from SDRStore

python hackrf_test.py --unattended --ci --eut $EUT --tester $TESTER --rev any -f $FIRMWARE -L /tmp/log

EUT: Hackrf Pro, TESTER: HackRF One r4

python hackrf_test.py --factory -p --eut $EUT --tester $TESTER --rev any -f $FIRMWARE -L /tmp/log --tcxo

Future Work

Now that conditional compilation directives are no longer used to control code execution we have the opportunity to consider whether we should follow the example of firmware/common/rad1o/ and add firmware/common/hackrf_one/ and firmware/common/praline/ directories.

---

• Finish up proof of concept for rf_path.c
• Add max2831 support to max283x
• Refactor remaining gpio's into platform_gpio.c
• Refactor remaining scu's into platform_scu.c
• Figure out how to handle config & drivers that need to setup early
• Figure out how to handle LED's for platform_detect.c
• Refactor remaining #ifdefs
• Get PRALINE binary booting on HACKRF_ONE
• Test. Test. Test.
• Clean up #includes
• Fix MCU LED not lighting up on HackRF Pro
• Fix supported platforms text in hackrf_info
• Add platform compilation conditionals back