btattach command hanging when using HCI UART sample nrf52840

[samanthacho]

When flashing the chip with a build from the samples/bluetooth/hci_uart via west build -b nrf52840dk/nrf52840 samples/bluetooth/hci_uart --pristine and west flash off the current main of sdk-zephyr, a btattach command from a linux host connected to the Zephyr device over UART hangs indefinitely:

btattach -B /dev/ttyMSM` -S 1000000 -P h4

Attaching Primary controller to /dev/ttyMSM2
Switched line discipline from 0 to 15
= New Index: 00:00:00:00:00:00 (Primary,UART,hci1) #0 [hci1] 4311.006931
= Open Index: 00:00:00:00:00:00 [hci1] 4311.007519
< HCI Command: Reset (0x03|0x0003) plen 0 #1 [hci1] 4311.007671
Device index 1 attached

The bluez version is 5.54, zephyr-sdk is the latest.

Looking at the backtrace on the device via gdb shows the CPU as idle, and no attempts to access the UART are seen.

When attempting to disable the config value CONFIG_BT_CTLR as suggested here, the build fails with the below error. What must be done to enable HCI UART support on the Zephyr device?

-- Zephyr version: 4.0.0-rc2 (/Users/samantha.cho/clones/zephyrproject/zephyr), build: v4.0.0-rc2-53-gb9fc4cc4151f
[139/144] Linking C executable zephyr/zephyr_pre0.elf
FAILED: zephyr/zephyr_pre0.elf zephyr/zephyr_pre0.map /Users/samantha.cho/clones/zephyrproject/build/zephyr/zephyr_pre0.map
: && ccache /Users/samantha.cho/zephyr-sdk-0.17.0/arm-zephyr-eabi/bin/arm-zephyr-eabi-gcc  -gdwarf-4 -gdwarf-4 zephyr/CMakeFiles/zephyr_pre0.dir/misc/empty_file.c.obj -o zephyr/zephyr_pre0.elf  zephyr/CMakeFiles/offsets.dir/./arch/arm/core/offsets/offsets.c.obj  -T  zephyr/linker_zephyr_pre0.cmd  -Wl,-Map=/Users/samantha.cho/clones/zephyrproject/build/zephyr/zephyr_pre0.map  -Wl,--whole-archive  app/libapp.a  zephyr/libzephyr.a  zephyr/arch/common/libarch__common.a  zephyr/arch/arch/arm/core/libarch__arm__core.a  zephyr/arch/arch/arm/core/cortex_m/libarch__arm__core__cortex_m.a  zephyr/arch/arch/arm/core/mpu/libarch__arm__core__mpu.a  zephyr/lib/libc/picolibc/liblib__libc__picolibc.a  zephyr/lib/libc/common/liblib__libc__common.a  zephyr/lib/net_buf/liblib__net_buf.a  zephyr/soc/soc/nrf52840/libsoc__nordic.a  zephyr/subsys/bluetooth/common/libsubsys__bluetooth__common.a  zephyr/subsys/bluetooth/host/libsubsys__bluetooth__host.a  zephyr/drivers/clock_control/libdrivers__clock_control.a  zephyr/drivers/gpio/libdrivers__gpio.a  zephyr/drivers/pinctrl/libdrivers__pinctrl.a  zephyr/drivers/serial/libdrivers__serial.a  zephyr/drivers/timer/libdrivers__timer.a  modules/hal_nordic/nrfx/libmodules__hal_nordic__nrfx.a  modules/segger/libmodules__segger.a  -Wl,--no-whole-archive  zephyr/kernel/libkernel.a  -L/Users/samantha.cho/clones/zephyrproject/build/zephyr  zephyr/arch/common/libisr_tables.a  -mcpu=cortex-m4  -mthumb  -mabi=aapcs  -mfp16-format=ieee  -mtp=soft  -fuse-ld=bfd  -Wl,--gc-sections  -Wl,--build-id=none  -Wl,--sort-common=descending  -Wl,--sort-section=alignment  -Wl,-u,_OffsetAbsSyms  -Wl,-u,_ConfigAbsSyms  -nostdlib  -static  -Wl,-X  -Wl,-N  -Wl,--orphan-handling=warn  -Wl,-no-pie  -specs=picolibc.specs  -DPICOLIBC_LONG_LONG_PRINTF_SCANF -L"/Users/samantha.cho/zephyr-sdk-0.17.0/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/thumb/v7e-m/nofp" -lc -lgcc && cd /Users/samantha.cho/clones/zephyrproject/build/zephyr && /opt/homebrew/Cellar/cmake/3.30.5/bin/cmake -E true
/Users/samantha.cho/zephyr-sdk-0.17.0/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/../../../../arm-zephyr-eabi/bin/ld.bfd: zephyr/subsys/bluetooth/host/libsubsys__bluetooth__host.a(hci_raw.c.obj):/Users/samantha.cho/clones/zephyrproject/zephyr/subsys/bluetooth/host/hci_raw.c:60: undefined reference to `__device_dts_ord_133'
collect2: error: ld returned 1 exit status
ninja: build stopped: subcommand failed.

[Vidar Berg]

Hello,

btattach will not return when the device is successfully attached. Instead it will continue running to manage the connection with the HCI interface. You can open another terminal window or tab to use the interface. Alternatively, you can create a systemd service,etc to let btattach run in the background.

Best regards,

Vidar

[samanthacho]

Hi,

I believe the device is not running -- I don't see the hci init being called when adding a breakpoint via gdb, and on my host device, I see the device being reported as down. Is disabling the config CONFIG_BT_CTLR on the Zephyr device necessary, as suggested by the HCI UART instructions, and if so, how do I resolve the build failures seen when disabling it for the sample?

```

root@7f0311c2-29f3-400f-89ff-27f9b285608a:~# hciconfig
hci0: Type: Primary Bus: UART
BD Address: 00:00:00:00:00:00 ACL MTU: 0:0 SCO MTU: 0:0
DOWN
RX bytes:0 acl:0 sco:0 events:0 errors:0
TX bytes:4 acl:0 sco:0 commands:1 errors:0

```

[samanthacho]

when looking at the backtrace from gdb, it looks like the hci_uart_init is not being called, and the device is not moving past the PRE_KERNEL_2 stage:

```

arch_cpu_atomic_idle (key=key@entry=32) at /Users/samantha.cho/clones/zephyrproject/zephyr/arch/arm/core/cortex_m/cpu_idle.c:136
136 arch_irq_unlock(key);
(gdb) bt
#0 arch_cpu_atomic_idle (key=key@entry=32) at /Users/samantha.cho/clones/zephyrproject/zephyr/arch/arm/core/cortex_m/cpu_idle.c:136
#1 0x0001648a in k_cpu_atomic_idle (key=32) at /Users/samantha.cho/clones/zephyrproject/zephyr/include/zephyr/kernel.h:6017
#2 lfclk_spinwait (mode=CLOCK_CONTROL_NRF_LF_START_STABLE) at /Users/samantha.cho/clones/zephyrproject/zephyr/drivers/clock_control/clock_control_nrf.c:515
#3 z_nrf_clock_control_lf_on (start_mode=start_mode@entry=CLOCK_CONTROL_NRF_LF_START_STABLE) at /Users/samantha.cho/clones/zephyrproject/zephyr/drivers/clock_control/clock_control_nrf.c:571
#4 0x000173ac in sys_clock_driver_init () at /Users/samantha.cho/clones/zephyrproject/zephyr/drivers/timer/nrf_rtc_timer.c:766
#5 0x00018242 in z_sys_init_run_level (level=level@entry=INIT_LEVEL_PRE_KERNEL_2) at /Users/samantha.cho/clones/zephyrproject/zephyr/kernel/init.c:371
#6 0x0001840a in z_cstart () at /Users/samantha.cho/clones/zephyrproject/zephyr/kernel/init.c:779
#7 0x00003ab4 in z_prep_c () at /Users/samantha.cho/clones/zephyrproject/zephyr/arch/arm/core/cortex_m/prep_c.c:209
#8 0x000038fe in z_arm_reset () at /Users/samantha.cho/clones/zephyrproject/zephyr/arch/arm/core/cortex_m/reset.S:169
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

```

I also added a breakpoint to the hci_uart_init, and don't see it being called. Is there something needed to make the chip move into the application stage?

[samanthacho]

By commenting out `z_nrf_clock_control_lf_on` in `zephyr/drivers/timer/nrf_rtc_timer.c`, the device moves into the application stage, and I see the uart init function being called. However, the device is still down and cannot be communicated with.

```

 btattach -B /dev/ttyMSM2 -S 1000000 -P h4 &

```

```

root@7f0311c2-29f3-400f-89ff-27f9b285608a:~# hciconfig
hci0: Type: Primary Bus: UART
BD Address: 00:00:00:00:00:00 ACL MTU: 0:0 SCO MTU: 0:0
DOWN
RX bytes:0 acl:0 sco:0 events:0 errors:0
TX bytes:4 acl:0 sco:0 commands:1 errors:0

```

When I initiate from the host device (running bluez) the `btattach` command, or an `hciconfig hci0 up`, I see `bt_uart_isr` being called on the device, but the host device eventually reports the command times out 
```

root:~# hciconfig hci0 up
Can't init device hci0: Connection timed out (110)

```

[Vidar Berg]

It looks like you’re not using the 52840 DK board with the optional 32 kHz crystal mounted. In that case, you need to change the low frequency clock source from the crystal oscillator to the internal RC oscillator (LFCLK controller. To do this, add CONFIG_CLOCK_CONTROL_NRF_K32SRC_RC=y to your project configuration file. 

I would also recommend you consider using the nRF Connect SDK (west init -m https://github.com/nrfconnect/sdk-nrf  --mr v2.7.0) which includes a qualified Bluetooth controller (Bluetooth QDIDs, ). 

[samanthacho]

Yes I am not using the 32kHz crystal, I have made the suggested change. Thank you!

I rebuilt with with the suggested SDK -- what I am seeing is that the chip seems to be receiving messages over UART from the host device (I can see the interrupt RX commands being called), and I can see the TX interrupt being called, but the host device does not seem to be receiving any of the transmitted data over UART, thus all the commands to bring up the UART interface are timing out.

[samanthacho]

To confirm is the baudrate correct at `1000000` for the btattach command?

[Vidar Berg]

Yes, the baudrate should be 1M if you build for the nrf52840dk/nrf52840 because of the Devicetree overlay in https://github.com/nrfconnect/sdk-zephyr/blob/main/samples/bluetooth/hci_uart/boards/nrf52840dk_nrf52840.overlay. The overlay also sets the 'hw-flow-control' property which enables the CTS and RTS handshake signals. Are those connected to your Linux host?

EDIT: I tested this sample on an nRF52840 DK using the default build configuration in SDK 2.7.0 with an Ubuntu 22.04 host, but was not able to reproduce the issue.

[samanthacho]

Yes, I verified the pins are connected for the HW flow control.

```

[00:02:03.923,492] <wrn> hci_uart: rx is ready
[00:02:03.923,522] <wrn> hci_uart: read 1 req 1
[00:02:03.923,522] <wrn> hci_uart: read 3 req 3
[00:02:03.923,553] <wrn> hci_uart: read 0 req 0
[00:02:03.923,583] <wrn> hci_uart: putting RX packet in queue.
[00:02:03.923,614] <wrn> hci_uart: spurious interrupt
[00:02:03.923,889] <wrn> hci_uart: buf 0x2000e620 type 1 len 7

```

I am seeing when using the RTT debugger that when the `hciconfig hci0 up` command is initiated from the host, although the DK receives packages over the UART, it never attempts to transmit via `tx_isr`. I also do not see the device moving the tx to enabled.

[samanthacho]

Upon further debugging efforts, what I am seeing is that it seems like the BT controller is not responding the way we are expected.
- On the first issue of the `btattach` command from the host, the DK responds with `0e 04 01 03 0c 00`, which I believe to be the successful completion of a reset of the BT controller.

- the `tx_isr` function is triggered via interrupt, and the transmit buffer currently contains 7 bytes, `04 0e 04 01 03 0c 00`, which is the reset response.

- the buffer is left with 6 bytes (the value of the rest of the reset command) and these are never transmitted

- subsequent requests are propagated to the BT controller, but the interrupt to attempt to send them is never triggered, thus the tx_isr is never called and no data is attempted to be transmitted over uart.

Should the whole message be sent in tx_isr via the `uart_fifo_fill` function? If not, when are the rest of the bytes supposed to be transmitted? Also, when data is added via `h4_send`, why is the interrupt to send the data via tx_isr not being triggered?

Also is this statement in hci_uart/src/main.c correct?
```

if (!(uart_irq_rx_ready(hci_uart_dev) ||

uart_irq_tx_ready(hci_uart_dev))) {

LOG_DBG("spurious interrupt");

}

```

I am wondering if it should be `if (!(uart_irq_rx_ready(hci_uart_dev) || uart_irq_tx_ready(hci_uart_dev))`?