zephyr: Controller unresponsive, command opcode 0x2018 timeout with err -11

Describe the bug

When running Bluetooth qualification tests AutoPTS we notice some instability and random failures related to random failures when BTP command is executed. This turned out to be due to controller not responding to HCI command.

I was able to reproduce this locally with about 20% rate when executing eg CSIP/CL/CGGIT/CHA/BV-01-C.

This is on nRF5340 DK (PCA10095).

Regression

  • This is a regression.

Steps to reproduce

Run AutoPTS test in loop, eg ./autoptsclient_bot.py -c CSIP/CL/CGGIT/CHA/BV-01-C --rtt --debug --stress_test --retry=20

Relevant log output

[00:00:13.502,502] <dbg> bttester_gap: le_connected: C0:07:E8:76:0C:6B (public): 0x00
[00:00:13.502,532] <dbg> bttester: tester_event: service 0x01 opcode 0x82
[00:00:23.596,496] <dbg> bttester: cmd_handler: cmd service 0x01 opcode 0x11 index 0x00
ASSERTION FAIL [err == 0] @ WEST_TOPDIR/zephyr/subsys/bluetooth/host/hci_core.c:472
	Controller unresponsive, command opcode 0x2018 timeout with err -11
[00:00:33.596,771] <err> os: r0/a1:  0x00000003  r1/a2:  0x00000000  r2/a3:  0x00000002
[00:00:33.596,771] <err> os: r3/a4:  0x00000003 r12/ip:  0x2001f304 r14/lr:  0x00017901
[00:00:33.596,771] <err> os:  xpsr:  0x01000000
[00:00:33.596,801] <err> os: Faulting instruction address (r15/pc): 0x00017910
[00:00:33.596,832] <err> os: >>> ZEPHYR FATAL ERROR 3: Kernel oops on CPU 0
[00:00:33.596,862] <err> os: Current thread: 0x20006e60 (unknown)
[00:00:39.618,652] <err> os: Halting system

Impact

Bluetooth qualification (both weekly and daily smoke runs) tests give unreliable results.

Environment

Zephyr main branch, AutoPTS master branch

Logs

CSIP_CL_CGGIT_CHA_BV-01-C_2025_06_18_11_30_04.zip

About this issue

  • Original URL
  • State: open
  • Created a year ago
  • Comments: 15 (15 by maintainers)

Most upvoted comments

Could you increase BT_CTLR_RX_STACK_SIZE to 1280 bytes and check. Refer to: bc0b2837316f38a18af1a7ee1c0bb039d2ffb443

Could you also provide the console output of the second UART port of the nRF5340DK, this will have any assertion failure printed out from the network core. There will always be the boot banner at the minimum.