Python controller crashes with error: 'Aborted (Core Dumped)'

[Harshith-GRL]

Description:
When executing the TC_Darwin_pair.py script from Stress and stability tests repo, in numerous Occasions the python controller is crashing is crashing with error
free (): double free detected in tcache 2
Aborted (core dumped)

GitHub hash of the SDK that were being used are : 1b4c56c f8d457a

Platform
I Have performed the execution on two different DUT (Raspberrypi and Nordic board), using ble-wifi and ble-thread provisioning
Controller Platform - python controller running on Raspberrypi 4B -8GB,
Nordic Board as DUT - all-cluster-app running on Nordic nrf52840-DK board
Raspberrypi as DUT - all-cluster-app running on Raspberrypi 4B -8GB
Provisioning method for Raspberrypi DUT is Ble-wifi and nordic DUT is Ble-thread

Test scenario

1. Pair/Provision the DUT (All-cluster-app running on raspi) as per drawiin pairing flow using ble-wifi/ ble-thread.
2. Read device information after the device is provisioned
3. sleep for 2 secs
4. Unpair the DUT (i.e. removing the fabric)
5. Factory reset the DUT
6. repeat step1 - 6 for 500 iterations
7. Script reference: https://github.com/CHIP-Specifications/matter-qa/blob/main/src/matter_qa/scripts/reliability_scripts/TC_Darwin_Pair.py

Screenshot
This error will not appear in the python logs, Hence I am attaching a screenshot to represent how the error looks

Execution Details

Git hub hash id: 1b4c56c
Iterations scheduled to run: 500
Iterations executed before crash: 132
Python Log controller for Nordic as DUT:
Nordic_as_dut_test_log_1b4c56cf5a.zip

Git hub hash id: f8d457a
Iterations scheduled to run: 500
Iterations executed before crash: 263
Python Log controller for Raspberrypi as DUT:
rpi_ble-wifi_test_log_f8d457a44.zip

Git hub hash id: f8d457a
Iterations scheduled to run: 500
Iterations executed before crash: 25
Python Log controller for Nordic as DUT:
Nordic_as_dut_test_log_f8d457a44.zip

[cecille]

do you have a backtrace from the core?

[Harshith-GRL]

currently i do not have any backtrace captured.....
would you recommend any specific method or process to achieve this backtrace capture??

[bzbarsky-apple]

Run your python under a debugger such as gdb or lldb.

[Harshith-GRL]

I was able to capture the back-trace using the gdb with bt command when the code crashed

trace_back_file.txt

[Harshith-GRL]

@cecille and @bzbarsky-apple I have captured the core dump using generate-core-file Git is not allowing me to upload the file any suggestions on how we could share it to you folks?

[Harshith-GRL]

I have uploaded the core dump file here https://graniteriverlabs1-my.sharepoint.com/:f:/g/personal/qanderson_graniteriverlabs_com/ErKYTMry_iZFmhML6I7cVFIBLvCjJwL3aEdiJ2Q7_siM7Q

[bzbarsky-apple]

Looks like a double-free in MdnsAvahi? @cecille do you know who might know about that code?