[STM]Add RNG (random number generator) for STM32F4 STM32F7 families

[adustm]

These functions are to be tested in mbed-os environment.

The test was passed for DISCO_F429ZI and NUCLEO_F746ZG with IAR toolchain.
The compilation passes for DISCO_F429ZI / DISCO_F469NI / DISCO_F746NG / NUCLEO_F410RB / NUCLEO_F429ZI / NUCLEO_F746ZG / NUCLEO_F767ZI

Please note that not every STM32F4 platforms are supporting the RNG feature. Some defines have been added in hal/targets/hal/TARGET_STM/TARGET_STM32F4/entropy_hardware_poll.c to enable the relevant targets.
I did not find any code to test the function, thus I used the one below :

#include "test_env.h"
#include "entropy_poll.h"

#define BUFF_SIZE       500
unsigned char myrndbuff[BUFF_SIZE]={0};
size_t *olen;

int main() {
    int ret = 0;
    int result = false;
    size_t mysize=0;
    olen = &mysize;
    ret = mbedtls_hardware_poll( NULL, myrndbuff, BUFF_SIZE, olen );
    if (ret == 0 ) {
        printf("tls_ret_ok\n");
        printf("%i\n",(int)*olen);
        if (*olen == BUFF_SIZE) {
            printf("size ok\n");
            result = true;
        } else printf("size ko\n");
    } else printf("tls_ret_ko\n");
}

This result was the following :

tls_ret_ok
500
size ok

Another printf inside the code showed that the values seemed to be random (I trust the hardware :) )

[adustm]

@screamerbg @pjbakker
Could you please review and let me know what you think of it ?

[sg-]

left over debug info?

[sg-]

formatting is a bit funky (see: https://developer.mbed.org/teams/SDK-Development/wiki/mbed-sdk-coding-style) and updated docs coming :)

[adustm]

Hello Sam,
Thank you for the review.
Sorry for the coding style and docs (did you mean the doxygen comments ?), I copied it from the entropy done on the K64F platform ;)
Is it better now ?

[sg-]

/morph test

[mbed-bot]

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 591

Build failed!

[bridadan]

Looks like build failures for ARM toolchain for the following targets: NUCLEO_F429ZI , DISCO_F469NI, DISCO_F429ZI, NUCLEO_F410RB:

Building project test_setup_failure (NUCLEO_F410RB, ARM)
Scan: ARM
Scan: test_setup_failure
Compile: main.cpp
Link: test_setup_failure
Error: L6218E: Undefined symbol mbedtls_hardware_poll (referred from entropy.o).
Finished: 0 information, 0 warning and 1 error messages.

[0xc0170]

@adustm Please can you check those failures?

[adustm]

Hello everyone,
I just rebased the branch, merged the targets.json manually, and relaunched the build for the following list of targets : DISCO_F429ZI / DISCO_F469NI / DISCO_F746NG / NUCLEO_F410RB / NUCLEO_F429ZI / NUCLEO_F746ZG / NUCLEO_F767ZI
You are right, there was an issue with the labels for STM32F4 family. I have fixed those labels in the targets.json file.
I looks like those defines (for instance STM32F429xx) where present in my previous environment, and not anymore...

I am using the command : mbed test --compile -t ARM -m NUCLEO_F410RB and it compiles perfectly.

Let me know.
Kind regards
Armelle

[bridadan]

Thanks @adustm!

/morph test

[mbed-bot]

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 669

Test failed!

[adustm]

Hello, Could you give me the command + result of the morph test, please ? I can't read them

[adustm]

@0xc0170 , @bridadan
is it possible to get details on the failure, please ?

[bridadan]

Hi @adustm, sorry for the delay! There was a failure in the CI that occurred during this test, I will restart it.

/morph test

[mbed-bot]

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 685

Test failed!

[adustm]

Hello Brian,
I cannot access the result of your morph test.
Could you provide me a copy/paste of the failure now ? Could I have the command line + text of the result, so that I can test it, please ?
Kind regards
Armelle

[screamerbg]

/morph test

[mbed-bot]

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 710

Test failed!

[0xc0170]

The build 710 is without any failure

[screamerbg]

@0xc0170 any idea why it reports failure?

[bridadan]

Hi everyone, sorry for the delay on reporting back on this.

For build 710 (that last one), a test box died again which cause the whole thing to fail. I've since reorganized the boards so they are more balanced across the boxes, so that should help with stability.

I can confirm that build 685 looks ok, so the merge to master should be fine 👍

[adustm]

Hello, thanks for the analysis and merge. I will push the entropy for the rest of STM32 products soon.
Cheers
Armelle