Bus Fault while running crytpo sample examples on nRF5340 DK

[BoscoJacinto]

Hi,

I tried running the crypto sample example as part of the nRF SDK project on my nRF5340 DK.

I successfully completed as few tests and then fails while trying to run the AES ECB MAC test suite.

It throws a BUS FAULT at BFAR Address: 0x50008158. The entire log is attached here.

Is anyone else facing the same problem, Ive used the project as it was in an unmodified state.

384: ChaChaPoly Invalid nonce_len=12 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:124]

385: ChaChaPoly message_len=114 ad_len=12 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:86]

386: ChaChaPoly message_len=265 ad_len=12 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:51]

Test suite AEAD CHACHAPOLY succeeded

Running test suite AEAD CHACHAPOLY simple

387: ChaChaPoly Decrypt Invalid mac -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:218]

388: ChaChaPoly Encrypt Invalid mac -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:199]

389: ChaChaPoly Decrypt Invalid key -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:180]

390: ChaChaPoly Encrypt Invalid key -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:161]

391: ChaChaPoly Valid ad_len=0 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:106]

Test suite AEAD CHACHAPOLY simple succeeded

Running test suite AES ECB MAC

392: CMAC 256 message_len=64 -- [00152177] <err> os: Exception occurred in Secure State

[00152338] <err> os: ***** HARD FAULT *****

[00152467] <err> os:   Fault escalation (see below)

[00152619] <err> os: ***** BUS FAULT *****

[00152745] <err> os:   Precise data bus error

[00152880] <err> os:   BFAR Address: 0x50008158

[00153021] <err> os: r0/a1:  0x00000000  r1/a2:  0x00000004  r2/a3:  0x0006f37c

[00153253] <err> os: r3/a4:  0x2000ff10 r12/ip:  0x00000000 r14/lr:  0x00025ca9

[00153485] <err> os:  xpsr:  0x61000000

[00153602] <err> os: Faulting instruction address (r15/pc): 0x00027008

[00153809] <err> os: >>> ZEPHYR FATAL ERROR 0: CPU exception on CPU 0

[00154012] <err> os: Current thread: 0x2000a438 (unknown)

[00154181] <err> os: Halting system


CTRL-A Z for help | 115200 8N1 | NOR | Minicom 2.7.1 | VT102 | Offline | ttyACM2                                                                                                                                  

Regards,

Bosco Jacinto

[Carl Richard]

Hello, Bosco!

Based on your earlier cases I assume that your are building for the non-secure partition of the nRF5340(nrf5340dk_nrf5340_cpuappns as target). Correct? This test sample requires the CryptoCell peripheral by default, which is not accessible from the non-secure domain. That may be the origin of the issue you're seeing here.

Best regards,
Carl Richard

[BoscoJacinto]

Hi Carl,

Thanks for your quick response. Sorry I dint mention the target, yes was building for nrf5340dk_nrf5340_cpuappns.

Ok so I do need to build for nrf5340dk_nrf5340_cpuapp ?

I did try with target nrf5340dk_nrf5340_cpuapp, I got a null pointer exception this time at the same AES ECB MAC test suite.

387: state_reset -- PASS -- [../src/main.c:48]

388: ChaChaPoly Invalid mac_len=15 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:142]

389: ChaChaPoly Invalid nonce_len=12 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:124]

390: ChaChaPoly message_len=114 ad_len=12 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:86]

391: ChaChaPoly message_len=265 ad_len=12 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:51]

Test suite AEAD CHACHAPOLY succeeded

Running test suite AEAD CHACHAPOLY simple

392: state_reset -- PASS -- [../src/main.c:48]

393: ChaChaPoly Decrypt Invalid mac -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:218]

394: ChaChaPoly Encrypt Invalid mac -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:199]

395: ChaChaPoly Decrypt Invalid key -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:180]

396: ChaChaPoly Encrypt Invalid key -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:161]

397: ChaChaPoly Valid ad_len=0 -- PASS -- [../test_cases/test_vectors_aead_chachapoly.c:106]

Test suite AEAD CHACHAPOLY simple succeeded

Running test suite AES ECB MAC

398: state_reset -- PASS -- [../src/main.c:48]

399: CMAC 256 message_len=64 -- [00214230] <err> os: ***** Debug monitor exception *****

Null-pointer exception?

[00214466] <err> os: r0/a1:  0x00000000  r1/a2:  0x00000004  r2/a3:  0x00063734

[00214698] <err> os: r3/a4:  0x000000a0 r12/ip:  0x00000000 r14/lr:  0x0000d621

[00214931] <err> os:  xpsr:  0x61000000

[00215048] <err> os: Faulting instruction address (r15/pc): 0x0000eace

[00215255] <err> os: >>> ZEPHYR FATAL ERROR 0: CPU exception on CPU 0

[00215459] <err> os: Current thread: 0x200002d8 (unknown)

[00215628] <err> os: Halting system

The following is the call stack, am not sure if the resolution of this picture is ok

The following is the project config

#
# Copyright (c) 2019 Nordic Semiconductor ASA
#
# SPDX-License-Identifier: LicenseRef-Nordic-5-Clause
#

# TEST
CONFIG_ZTEST=y
CONFIG_ZTEST_FAIL_FAST=y
CONFIG_ZTEST_STACKSIZE=4096
CONFIG_MAIN_STACK_SIZE=8192
CONFIG_ZTEST_TC_UTIL_USER_OVERRIDE=y

# LOG
CONFIG_TEST_LOGGING_DEFAULTS=n
CONFIG_LOG=y
CONFIG_LOG_PRINTK=y
CONFIG_LOG_MODE_IMMEDIATE=y
CONFIG_LOG_DEFAULT_LEVEL=3
CONFIG_LOG_BACKEND_UART=y
CONFIG_LOG_BACKEND_SHOW_COLOR=n
CONFIG_LOG_BACKEND_FORMAT_TIMESTAMP=n

# MISC
CONFIG_DK_LIBRARY=y
CONFIG_SPEED_OPTIMIZATIONS=y

# CRYPTO
CONFIG_NORDIC_SECURITY_BACKEND=y
CONFIG_NRF_SECURITY_RNG=y
CONFIG_ENTROPY_GENERATOR=y

CONFIG_MBEDTLS=y
CONFIG_MBEDTLS_LIBRARY=y
CONFIG_MBEDTLS_ENABLE_HEAP=y
CONFIG_MBEDTLS_HEAP_SIZE=65536
CONFIG_MBEDTLS_INSTALL_PATH="DUMMY"

CONFIG_MBEDTLS_ECJPAKE_C=y

# Quick fix to ensure the crypto tests can execute directly on nRF9160
CONFIG_SRAM_SIZE=128

Regards,

Bosco jacinto

[Carl Richard]

Hi again!

For this specific sample you should build for the secure target, yes. However, there are methods for using CryptoCell from the non-secure domain. You can read more about this in the documentation.

Based on the error it seems like there may be some issue with the debugging. Have you tried to run the sample while not debugging in SES?

In addition you could search for 200002d8 in the <project_root>/build/zephyr/zephyr.map file and see which thread that is the culprit. This way you could check the program counter (PC) as well.

Best regards,
Carl Richard

[BoscoJacinto]

Hi Carl,

I tried running it without the debugger as well, the result is the same.

The PC is at the fault handler (fault_s)

And hex 200002d8 is in ztest_thread in ztest.c itself

I am using master i guess because i had not checked out any specific tag during  "west init"

Regards,

Bosco Jacinto

[Carl Richard]

Hi again, Bosco!

I can verify that there is an issue with the Crypto tests for nRF5340 on the master branch. However, I've verified that succeeds on the v1.5.0 release so I suggest switching to that. Use the following procedure

cd nrf
git checkout v1.5.0
west update

We recommend that customers to develop on the release tags, as the master branch is known to be unreliable, but I'll report the issue of course.

Best regards,
Carl Richard

[BoscoJacinto]

Hi Carl,

Thanks a lot.  I will try version 1.5.0

Regards,

Bosco Jacinto

[Carl Richard]

My pleasure. Please report back if you face any other issues.

Best regards,
Carl Richard