STM32L4: fix trng clock setting

[bcostm]

Description

Remove the TRNG clock setting for STM32L4 devices in the trng_api.c file to solve the conflict with USB clock setting.

Fix #8864

Pull request type

[X] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Docs update
[ ] Test update
[ ] Breaking change

[ghost]

I can test on Nucleo-476 kit and a 475 custom board (runs the disco475iot01A target but is a custom board with a usb micro connector on usb signals PA10, PA11).
Will attach test report here when complete

[bcostm]

I have updated this PR and only removed the TRNG clock setting. Tests are on-going on my side on all STM32L4 devices. If there are ok then it will be not needed to change the system_clock.c files...

[ghost]

Tested on NUCLEO_L476RG board:

NUCLEO_L476RG_test.zip

[ghost]

Tested on custom DISCO_L475VG_IOT01A board:

DISCO_L475VG_IOT01A_test.zip

[bcostm]

Thanks for the reports. On my side the TRNG tests are also PASS:

+---------------------+-----------------+---------------------+-----------------+--------+--------+--------+--------------------+
| target              | platform_name   | test suite          | test case       | passed | failed | result | elapsed_time (sec) |
+---------------------+-----------------+---------------------+-----------------+--------+--------+--------+--------------------+
| NUCLEO_L432KC-ARM   | NUCLEO_L432KC   | tests-mbed_hal-trng | TRNG: trng_test | 1      | 0      | OK     | 0.22               |
| NUCLEO_L433RC_P-ARM | NUCLEO_L433RC_P | tests-mbed_hal-trng | TRNG: trng_test | 1      | 0      | OK     | 0.21               |
| NUCLEO_L476RG-ARM   | NUCLEO_L476RG   | tests-mbed_hal-trng | TRNG: trng_test | 1      | 0      | OK     | 0.22               |
| NUCLEO_L486RG-ARM   | NUCLEO_L486RG   | tests-mbed_hal-trng | TRNG: trng_test | 1      | 0      | OK     | 0.23               |
| NUCLEO_L496ZG-ARM   | NUCLEO_L496ZG   | tests-mbed_hal-trng | TRNG: trng_test | 1      | 0      | OK     | 0.21               |
+---------------------+-----------------+---------------------+-----------------+--------+--------+--------+--------------------+
+-------------------------+---------------------+---------------------+-----------------+--------+--------+--------+--------------------+
| target                  | platform_name       | test suite          | test case       | passed | failed | result | elapsed_time (sec) |
+-------------------------+---------------------+---------------------+-----------------+--------+--------+--------+--------------------+
| DISCO_L475VG_IOT01A-ARM | DISCO_L475VG_IOT01A | tests-mbed_hal-trng | TRNG: trng_test | 1      | 0      | OK     | 0.23               |
| DISCO_L476VG-ARM        | DISCO_L476VG        | tests-mbed_hal-trng | TRNG: trng_test | 1      | 0      | OK     | 0.24               |
+-------------------------+---------------------+---------------------+-----------------+--------+--------+--------+--------------------+

[jeromecoutant]

ST CI OK

[cmonr]

CI started

[mbed-ci]

Test run: SUCCESS

Summary: 11 of 11 test jobs passed
Build number : 1
Build artifacts

[0xc0170]

@bcostm We are currently investigating a test failure with L4 device and tls. From the logs, it seems like entropy is broken. Reviewing the changes between good and bad state, this PR was flagged. We will provide more details once we test reverting this one.

This is the commit range good and bad state (78d6018..fce4dc6)

[juhoeskeli]

@0xc0170 @bcostm tested with and without this PR using WISE_1570 (STM32L486). With the functions requiring trng will fail, and by reverting the merge it will work again.

[0xc0170]

OK, I'll open a revert for review (this is currently only on master as 5.11.1 label here, not yet released).