"ASSERTION FAIL [err == 0] k_sem_take failed with err -11" when HCI is used with LP-UART

[Anirudh Agarkhed]

Test setup -

1. nRF52840 DK - 1 as hci host, 1 as hci controller

2. HCI Host runs nrf/samples/bluetooth/peripheral_hr_coded with modified configuration (attached as conf_hci_host.zip)

3. HCI Controller runs nrf/samples/bluetooth/hci_lpuart with overlay and conf for nrf52840 dk (attached as conf_hci_controller.zip)

4. Connections are made tx-pin <-> rx-pin, req-pin <-> rdy-pin with each other.

Issues observed -

[2021-03-24 12:10:46.108] ASSERTION FAIL [err == 0] @ WEST_TOPDIR/zephyr/subsys/bluetooth/host/hci_core.c:333
[2021-03-24 12:10:46.115]   k_sem_take failed with err -11

This results in a continuous restart of the hci host firmware.

Question -

1. What should be done to make HCI over LP-UART work?conf_hci_controller.zipconf_hci_host.zip

[Carl Richard]

Hi, Anirudh!

I'm looking into this. Have you made sure that the hci_host has the correct configuration, including CONFIG_BT_H4=y for enabling HCI over UART? You can look at the samples/nrf9160/lte_ble_gateway sample for reference.

Best regards,
Carl Richard

[Anirudh Agarkhed]

Hello Carl,

Yes, that option is used, and also the example that you have mentioned is being used as a reference. Just as an additional question, can HCI function over 2 Wire UART, and not mandate H4 which depends on 4 wire UART - CTS/RTS?

Regards,

Anirudh Agarkhed

[Carl Richard]

Understood! Then I will do some testing tomorrow and see if I can reproduce. 

Anirudh Agarkhed said:

Just as an additional question, can HCI function over 2 Wire UART, and not mandate H4 which depends on 4 wire UART - CTS/RTS?

It can. This is a restriction related to the hci_uart sample of Zephyr, and does not limit hci_lp_uart.

By the way, have you tried using 1000000 baud rate?

Best regards,
Carl Richard

[Anirudh Agarkhed]

Yes, have used both 1000000 and 115200. When I was zipping the file, the config was set to 115200.

Regards,

Anirudh Agarkhed

[Carl Richard]

Hi!

I've spent some time looking into this now and while I haven't found a fix yet, I think I've found the source of the issue. The host and controller must be synchronized somehow, this is, from what I can see, done by resetting the controller when the host is ready. On the nRF9160DK there is a dedicated reset line between the nRF52840 SoC and nRF9160 SiP, as well as firmware handling the hci setup. This is not implemented for the nRF52840DK.

I suggest that you study the nrf9160dk_nrf9160 and nrf9160dk_nrf52840 board definitions in the Zephyr, as well as the file nRF52840_reset.c, and see if you can reverse engineer the nRF9160DK setup to work for the two nRF52840 DKs. I'll have a look at this myself next week. 

Bes regards,
Carl Richard