MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Mon, 25 Mar 2024 08:28:03 GMT

Thanks. This confirms that the crystal oscillator is not starting, which suggests that there is a problem with the crystal or decoupling caps. It could be a soldering or issue.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Mon, 25 Mar 2024 08:17:58 GMT

Hi,

with hello example code never exit while loop:

while(NRF_CLOCK->EVENTS_HFCLKSTARTED == 0);

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Mon, 25 Mar 2024 06:04:09 GMT

I would still like you to measure the crystal oscillator startup time from the nrf using the 'hello world' sample.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Sat, 23 Mar 2024 06:07:59 GMT

Hi,

any further thoughts, since I can't figure out the problem, except, nrf52832 malfunction or osc problem, but measurement shows pretty close to good one.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Fri, 22 Mar 2024 10:14:58 GMT

The purpose of this test is to measure the actual crystal startup time using a minimal project that does not already request the HFXO.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Fri, 22 Mar 2024 09:57:46 GMT

Hi,

we found some beacon sample and it is not working. It stops like our application. So I assume, that fw isn't a problem, but something else on hardware level. The most surprising is, that with your code sniped osc is started, but later we get the same assert like osc not working. Is there any status or register for nrf52832 we could check, which could provide more details about issue?

regards

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Fri, 22 Mar 2024 09:20:30 GMT

Hi,

hmm there are no advertising packets seen and only output is:

void main(void)
{
printk("Hello World! %s\n", CONFIG_BOARD);
}

I see no bt_init(); or bt_enable(); to be called, so I am not sure, if this sample is relevant or not. Also RTT looks like is OFF.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Fri, 22 Mar 2024 08:47:15 GMT

No worries. I see the startup time is only 1 us, which suggest that it was already started. Can you try the same code snippet in the "Hello world" sample?

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Thu, 21 Mar 2024 12:29:22 GMT

Sorry the wrong log. Here it is the log:

00> [00:00:00.002,716] [00:00:00.002,807] <inf> app: HF Clock has started. Startup time: 1 uS
00>
00> [00:00:00.011,474] <inf> fs_nvs: 2 Sectors of 4096 bytes
00> [00:00:00.011,505] <inf> fs_nvs: alloc wra: 0, fa8
00> [00:00:00.011,505] <inf> fs_nvs: data wra: 0, 50
00> [00:00:00.011,627] <inf> sdc_hci_driver: SoftDevice Controller build revision:
00> 0e e7 c5 66 67 18 3c ac b3 d2 cc 81 a3 dc f1 c0 |...fg.<. ........
00> c0 36 02 22 |.6."
00> [00:00:00.015,991] <inf> bt_hci_core: HW Platform: Nordic Semiconductor (0x0002)
00> [00:00:00.015,991] <inf> bt_hci_core: HW Variant: nRF52x (0x0002)
00> [00:00:00.016,021] <inf> bt_hci_core: Firmware: Standard Bluetooth controller (0x00) Version 14.50663 Build 1008232294
00> [00:00:00.016,479] <inf> bt_hci_core: No ID address. App must call settings_load()
00> [00:00:00.017,578] <inf> bt_0m
00> [0m
00> m
00> [00:00:00.032,714] <err> mpsl_init: MPSL ASSERT: 112, 2198
00> [00:00:00.032,714] <err> os: ***** HARD FAULT *****
00> [00:00:00.032,714] <err> os: Fault escalation (see below)
00> [00:00:00.032,714] <err> os: ARCH_EXCEPT with reason 3
00>
00> [00:00:00.032,745] <err> os: r0/a1: 0x00000003 r1/a2: 0x0000000e r2/a3: 0x0000000d
00> [00:00:00.032,745] <err> os: r3/a4: 0x20001af0 r12/ip: 0x20001244 r14/lr: 0x0002383f
00> [00:00:00.032,745] <err> os: xpsr: 0x61000018
00> [00:00:00.032,745] <err> os: Faulting instruction address (r15/pc): 0x00031a12
00> [00:00:00.032,745] <err> os: >>> ZEPHYR FATAL ERROR 3: Kernel oops on CPU 0
00> [00:00:00.032,775] <err> os: Fault during interrupt handling
00>
00> [00:00:00.032,775] <err> os: Current thread: 0x20002fb8 (unknown)
00> [00:00:01.731,231] <err> fatal_error: Resetting system

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Thu, 21 Mar 2024 11:57:18 GMT

Are you outputting the startup time as shown in my code snippet? I don't see it in the log you posted.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Thu, 21 Mar 2024 11:48:43 GMT

Measured clock with oscilloscope is 32MHz and stable.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Thu, 21 Mar 2024 11:48:02 GMT

Hi,

we did this, but code is stuck inside while(); loop and does not exit when using debugger. If check power on run I get similar ouput:

00> [00:00:00.002,71[00:00:00.011,444] <inf> fs_nvs: 2 Sectors of 4096 bytes
00> [00:00:00.011,444] <inf> fs_nvs: alloc wra: 0, fa8
00> [00:00:00.011,444] <inf> fs_nvs: data wra: 0, 50
00> [00:00:00.011,566] <inf> sdc_hci_driver: SoftDevice Controller build revision:
00> 0e e7 c5 66 67 18 3c ac b3 d2 cc 81 a3 dc f1 c0 |...fg.<. ........
00> c0 36 02 22 |.6."
00> [00:00:00.015,960] <inf> bt_hci_core: HW Platform: Nordic Semiconductor (0x0002)
00> [00:00:00.015,960] <inf> bt_hci_core: HW Variant: nRF52x (0x0002)
00> [00:00:00.015,991] <inf> bt_hci_core: Firmware: Standard Bluetooth controller (0x00) Version 14.50663 Build 1008232294
00> [00:00:00.016,448] <inf> bt_hci_core: No ID address. App must call settings_load()
00> [00:00:00.017,517] <inf> bt_hci_core: Identity: F9:E3:ED:5D:18:35 (random)
00> [00:00:00.017,547] <inf> bt_hci_core: HCI: version 5.2 (0x0b) revision 0x12fe, manufacturer 0x0059
00> [00:00:00.017,578] <inf> bt_hci_core: LMP: version 5.2 (0x0b) subver 0x12fe
00> [00:00:00.030,303] <err> mpsl_init: MPSL ASSERT: 112, 2198
00> [00:00:00.030,334] <err> os: ***** HARD FAULT *****
00> [00:00:00.030,334] <err> os: Fault escalation (see below)
00> [00:00:00.030,334] <err> os: ARCH_EXCEPT with reason 3
00>
00> [00:00:00.030,334] <err> os: r0/a1: 0x00000003 r1/a2: 0x0000000c r2/a3: 0x0000000b
00> [00:00:00.030,334] <err> os: r3/a4: 0x20001af0 r12/ip: 0x20001244 r14/lr: 0x0002382f
00> [00:00:00.030,364] <err> os: xpsr: 0x61000018
00> [00:00:00.030,364] <err> os: Faulting instruction address (r15/pc): 0x00031a02
00> [00:00:00.030,364] <err> os: >>> ZEPHYR FATAL ERROR 3: Kernel oops on CPU 0
00> [00:00:00.030,364] <err> os: Fault during interrupt handling
00>
00> [00:00:00.030,364] <err> os: Current thread: 0x20001b70 (unknown)
00> [00:00:01.093,322] <err> fatal_error: Resetting system

I am not sure, why I am seeing this kind of a log, since the mcu should stuck inside while I presume.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Thu, 21 Mar 2024 11:09:03 GMT

[quote user="Aedev"]Also in Config we have selected "Clock startup policy (Wait for stability)[/quote]

This applies to the LFXO.

[quote user="Aedev"]that I can't find: mpsl_clock_hfclk_latency_set(MPSL_CLOCK_HF_LATENCY_WORST_CASE);[/quote]

This option did not exist in SDK v1.9.x. This is the code snippet I was referring to:

	NRF_TIMER1->BITMODE = TIMER_BITMODE_BITMODE_32Bit << TIMER_BITMODE_BITMODE_Pos;
	NRF_TIMER1->TASKS_CLEAR = 1;
	NRF_CLOCK->EVENTS_HFCLKSTARTED = 0;
	NRF_TIMER1->TASKS_START = 1;
	NRF_CLOCK->TASKS_HFCLKSTART = 1;
	while(NRF_CLOCK->EVENTS_HFCLKSTARTED == 0);
	NRF_TIMER1->TASKS_CAPTURE[0] = 1;

	printk("HF Clock has started. Startup time: %d uS\n", NRF_TIMER1->CC[0]);

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Thu, 21 Mar 2024 11:03:05 GMT

Also in Config we have selected "Clock startup policy (Wait for stability)", which may also means some wait until clock is stable.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Thu, 21 Mar 2024 10:58:37 GMT

Hi,

the problem is, that I can't find: mpsl_clock_hfclk_latency_set(MPSL_CLOCK_HF_LATENCY_WORST_CASE); in used SDK. Is this newer function or it belongs to some other library?

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Thu, 21 Mar 2024 10:32:54 GMT

Hi,

I'm a bit surprised that the startup and oscillation appear to be the same on the good and bad device. Please try to add the code snippet I proposed earlier to the beginning of main to see if the HF clock start event is triggered on the bad device.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Thu, 21 Mar 2024 09:53:06 GMT

Hi,

I managed to measure some crystal startup times on GOOD and BAD device. Since I do not have active probe, I can't tell if measurement is 100% OK or not. If I measure crystal frequency I get stable 31.9999Mhz on both samples with similar amplitude. Comparing both device I can't see any obviously difference between them, except of exception.

Any further thoughts to investigate the cause?

Thanks

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Thu, 21 Mar 2024 08:07:03 GMT

MCUBoot doesn't use HFXO. Also, the log shows that the assert is raised when you call bt_enable.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Thu, 21 Mar 2024 07:32:39 GMT

Hi,

on your suggested link above I can see we could use: mpsl_clock_hfclk_latency_set(MPSL_CLOCK_HF_LATENCY_WORST_CASE), to somehow compensate startup times. Since we use MCUboot could you suggest if this could solve an issue and what is the proper way to call this function. The problem is, that this fault happens before main() is called.

thanks

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Wed, 20 Mar 2024 14:51:50 GMT

The Bluetooth stack will not use the RADIO without the HFXO running. The HFINT is not accurate enough to create the carrier frequency.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Wed, 20 Mar 2024 12:48:16 GMT

Hi,

is there an option to bypass this external HF crstal and use HFINT without and switching between HFINT and HFXO? I will measure startup time on faulty board, but any crystal service is almost impossible to do.

Thanks.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Wed, 20 Mar 2024 12:34:52 GMT

There aren't any parameters to configure, unfortunately. You can try to measure the startup time on a failing board: RE: MPSL ASSERT: 112, 2134 if using multiple BLE communication to see if it starts up at all.

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Wed, 20 Mar 2024 12:32:26 GMT

Hi,

yes we are observing this issues across 5-10% boards in a single batch. Is there any option to set up this HF crystal with different parameters to see if this solve the problem?

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Vidar Berg — Wed, 20 Mar 2024 12:16:57 GMT

Hi,

Thanks. This assertion is triggered when the HF crystal oscillator is not running when it is supposed to. Are you seeing this across multiple boards?

RE: MPSL ASSERT: 112, 2198 issue and mpsl_init() does not return!

Aedev — Wed, 20 Mar 2024 12:13:40 GMT

Hi,

firmware was build with NCS 1.9.1.

Thanks