Update feature-nrf52-sdk15 for v5.12 #9999
Conversation
- top level files ported from TARGET_NORDIC/TARGET_NRF5x/ Also addressed: - fixed linking issue for gcc - added support for nRF52-DK builds, but reverted to using nRF52840 sdk_config.h (must be updated) - introduced "RTC" to targets.json (might need to be removed eventually)
Also addressed: - removed dependency on legacy config (excluded apply_old_config.h) - removed legacy pwm and saadc headers - Arm Compiler 5 linking issue (a band-aid for now... needs to be properly addressed for peripheral sharing) - added missing header in SoftDevice file
- Missed some NRFX defines that needed to changed - Set PWM base clock to 125kHz (needs to be reverted back to 1 MHz) - Updated sdk_config.h for nRF52_DK builds - Brought in updates from PR7779 (fix for nRF52 PWM issues)
Enable UART interfaces to enable serial debug prints.
Using the nrfx_get_irq_number only works with the handle. Since we know the IRQ numbers for UART0, RTC2 and EGU0, use them directly.
- Use new NRFX header file - Fix nRF52832 linker script/ld files (hardcoded addresses for now) - Temporarily remove DEVICE_TRNG for nRF52832 (which broke the build) - Improve serial_putc() fix so we don't rely on "extra" functions
- Added legacy nrf_drv_rng.c as there is no merit in fully adopting nrfx_rng.c - Added nrf_queue library component - Removed apply_old_config.h (unused for some time now) - Updated sdk_config.h for queue and RNG support for nRF52832 - Brought back RNG into targets.json for nRF52832 Note: nRF52840 still uses CryptoCell 310 for TRNG
When the SoftDevice (SD) is enabled we need to set the app_offset to 0x26000 to make room for the SoftDevice. If we let the build tools self-manage this, MBED_APP_START is coming out at 0x25000 which is not correct for the Nordic 15.0 SDK. The app_offset value is translated to MBED_APP_START by the build infrastructure. We were hard coding MBED_APP_START in the scatter and ld files as a temporary measure while testing. This now sets it properly if the SD is being brought in.
Ensure that vector table gets initialized properly. The table that we initialize in startup_nrf52840.S gets wiped out as the section is declared as noinit. Fix this by implementing the weak function mbed_sdk_init that inits the vector table.
* Remove NRF_SDH_CLOCK parameters from mbed_lib.json * Bring in QSPI for nRF52840 * Migrate legacy QSPI driver to SDK v15 (nrfx_qspi.h) * Remove outdated comment in i2c_api.c
The 15.0 SDK doesn't support the nRF51 so it must continue to use the legacy NRF drivers. Thus the original common_rtc.c, gpio_api.c and us_ticker.c are restored and placed under TARGET_NRF5x/TARGET_NRF51. Likewise the modified common_rtc.c, gpio_api.c and us_ticker.c are moved to TARGET_NRF5x/TARGET_NRF52 so they are excluded from nRF51 builds.
nRF5x PinNames.h never utilized PORT_SHIFT so removed for clarity after user commented on it.
- Add MBR, NONE and OTA SoftDevice build options for S132 and S140 - Add S112 SoftDevice (single build option) - Some folder restructuring in TARGET_SOFTDEVICE_COMMON was required
This reverts commit 3d2fa53. This was a breaking change for the "MBR" and "NONE" builds. After testing it was also determined that copying the vector table a second time wasn't required for the "FULL" build.
* Update TARGET_NRF5x/README.md to improve "Changing SoftDevice" section and added section on NRF52840 CryptoCell310 Support * Update the file list in TARGET_SDK_15_0/TARGET_SOFTDEVICE_COMMON/README.md * Add missing CR-LF to Nordic-provided SDK file * Rename a header file in the TARGET_SOFTDEVICE_S112 tree
- Brought in new nrfx APIs - Brought in PPI additions - Removed dead code for RTC
|
Any chance a WIP PR or issue can be created with current progress on this effort? |
|
@loverdeg-ep Could you elaborate on your question? This PR is just a minor update for the upcoming v5.12 release and it contains no new features. We currently have Nordic SDK v15 support here: https://github.com/ARMmbed/mbed-os/tree/feature-nrf52-sdk15 |
|
Per your comment here #7674 (comment) I was thinking an issue or work in progress merge request for the feature branch could be used to track the effort. I understand the reasoning for the current pace on this and believe it would be nice if the community could distribute this effort with you in Nordic's place. |
|
@loverdeg-ep OK - understood and I personally like that approach. The plan is to take this to mainline rather soon and then evaluate future work. FYI @dlfryar. |
* The following v14.2 files are removed from nrfx/legacy: nrf_drv_clock.h nrf_drv_common.h nrf_drv_gpiote.h nrf_drv_spi.h nrf_drv_twi.h nrf_drv_uarth.h * Remove unneeded references to "nrfx_glue.h" and "nrf_drv_common.h" * Remove the definition of PACKED from app_util_platform.h (already defined in mbed_toolchain.h)
|
Scheduled one CI run to test this rebase |
Test run: FAILEDSummary: 3 of 9 test jobs failed Failed test jobs:
|
|
Failures related, |
When removing CryptoCell310 we now need to define: NRFX_RNG_ENABLED=1 RNG_ENABLED=1 NRF_QUEUE_ENABLED=1 These are brought in via "macros_add".
|
@0xc0170 Builds for |
|
CI restarted |
Test run: SUCCESSSummary: 13 of 13 test jobs passed |
|
@TacoGrandeTX Does this still need work, or can the 'Do Not Merge' label be removed? |
|
@cmonr This is all set and it can be force-pushed as we discussed. When back from vacation I will bring in the USBD work from @AGlass0fMilk. |
|
Rebased |
|
|
|
+1 for NRF SDK 15 (current is |
Description
Keep feature-nrf52-sdk15 current with master for v5.12 release:
Do not merge - this is a feature rebase only
Pull request type
Reviewers
Release Notes
Successful greentea results attached. There are two failures for nRF52840_DK with AC6 but those tests pass with gcc so appears to be a GT problem with these tests on nRF52840 under AC6. There is one failure on nRF52_DK related to TRNG as that always times out on laptop (as before). @desmond-blue conducted BLE testing using standard BLE examples.
v5.12_tests.zip