Force inline critical_section_{enter,exit}

[tannewt]

Otherwise it may end up in flash when we want it to follow the caller's placement (maybe in RAM.)

[lurch]

Bazel CI checks are failing with:

In file included from src/common/pico_sync/critical_section.c:7:
src/common/pico_sync/include/pico/critical_section.h:62:15: error: expected ';' before 'static'
   62 | __force_inline static void critical_section_enter_blocking(critical_section_t *crit_sec) {
      |               ^~~~~~~
      |               ;
src/common/pico_sync/include/pico/critical_section.h:71:15: error: expected ';' before 'static'
   71 | __force_inline static void critical_section_exit(critical_section_t *crit_sec) {
      |               ^~~~~~~
      |               ;

https://github.com/search?q=repo%3Araspberrypi%2Fpico-sdk%20__force_inline&type=code shows a mix of __force_inline static ... and static __force_inline ... so this doesn't appear to be as simple as the keywords being listed in the wrong order?
ping @armandomontanez for suggestions.

[armandomontanez]

I'm pretty sure Bazel can't see the __force_inline symbol at all. __force_inline is defined in src/rp2_common/pico_platform_compiler/include/pico/platform/compiler.h, which notably is in rp2_common and doesn't appear to have a host port. Sounds like either CMake isn't building the pico_sync_critical_section library for host, or somehow the rp2_common header is slipping into the CMake host build?

Adding pico_platform_compiler as a special-case thing that can be directly referenced from common should fix this. Or I can reduce the scope of the Bazel host build to work around this if needed. I can get a diff sent over tomorrow.

[kilograham]

yeah; i'll fix that in the host headers

[kilograham]

yeah; i'll fix that in the host headers

don't seem to be able to push to your PR, so added this here #2396

I'm fine with the pico.h include (it should always be included first), however it is already included by pico/lock_core.h so not clear why you need it?

[tannewt]

Want me to merge in the #2396 changes?

[kilograham]

nah, i have merged it into develop, and re-kicked off CI for this branch

[tannewt]

Ok, I rebased it on develop and squashed the three commits. Hopefully it'll pass now