stress_mt failures on Linux and stuck on macOS

[tormodvolden]

We have previously seen errors on macOS on virtual machines, but here we have something I can reproduce. Testing on a Lenovo L13 with Intel i5-1135G7 (8 cores), Ubuntu 24.04 and its kernel 6.8.1, I get with 1.0.27 or latest git (last tested at 2a138c6):

Running multithreaded init/exit test...
Starting 8 threads
Running multithreaded init/exit test with enumeration...
Starting 8 threads
No access to device 1d6b:0003, skipping transfer tests.
No access to device 06cb:00bd, skipping transfer tests.
No access to device 13d3:56bc, skipping transfer tests.
No access to device 192f:0416, skipping transfer tests.
No access to device 04d9:1829, skipping transfer tests.
No access to device 05e3:0608, skipping transfer tests.
No access to device 05e3:0608, skipping transfer tests.
No access to device 046d:0a56, skipping transfer tests.
No access to device 8087:0026, skipping transfer tests.
No access to device 1d6b:0002, skipping transfer tests.
No access to device 1d6b:0003, skipping transfer tests.
No access to device 1d6b:0002, skipping transfer tests.
Thread 2 failed (iteration 1): LIBUSB_ERROR_IO
Thread 3 failed (iteration 1): LIBUSB_ERROR_IO
Thread 5 failed (iteration 1): LIBUSB_ERROR_IO
All threads discovered 13 devices (12 not opened)
All done, 3 errors

Or

Running multithreaded init/exit test...
Starting 8 threads
Running multithreaded init/exit test with enumeration...
Starting 8 threads
No access to device 1d6b:0003, skipping transfer tests.
No access to device 06cb:00bd, skipping transfer tests.
No access to device 13d3:56bc, skipping transfer tests.
No access to device 192f:0416, skipping transfer tests.
No access to device 04d9:1829, skipping transfer tests.
No access to device 05e3:0608, skipping transfer tests.
No access to device 05e3:0608, skipping transfer tests.
No access to device 046d:0a56, skipping transfer tests.
No access to device 8087:0026, skipping transfer tests.
No access to device 1d6b:0002, skipping transfer tests.
No access to device 1d6b:0003, skipping transfer tests.
No access to device 1d6b:0002, skipping transfer tests.
Thread 0 failed (iteration 1): LIBUSB_ERROR_IO
Device count mismatch: Thread 1 discovered 13 devices instead of 0
Thread 2 failed (iteration 1): LIBUSB_ERROR_IO
Thread 7 failed (iteration 1): LIBUSB_ERROR_IO
All done, 4 errors

[tormodvolden]

On the macOS CI it can also get into an infinite loop it seems like: https://github.com/libusb/libusb/actions/runs/8991096355/job/24697895520 (been running 2 hours now).

[mcuee]

For Windows, we have two more ignored errors, I tend to think LIBUSB_ERROR_NOT_FOUND is not a normal error as well. But I need to review my past test results.

#if defined(PLATFORM_WINDOWS)
				    || open_err == LIBUSB_ERROR_NOT_SUPPORTED
				    || open_err == LIBUSB_ERROR_NOT_FOUND
#endif

[mcuee]

On the macOS CI it can also get into an infinite loop it seems like: https://github.com/libusb/libusb/actions/runs/8991096355/job/24697895520 (been running 2 hours now).

Same for this one (running for more than 5 hours)
https://github.com/libusb/libusb/actions/runs/9276043843/job/25522153514

[tormodvolden]

I can reproduce it also with debug logging:

[ 2.949947] [0001427f] libusb: debug [usbi_wait_for_events] poll() returned 1
[ 2.949967] [0001427f] libusb: debug [reap_for_handle] urb type=2 status=-71 transferred=0
[ 2.949952] [0001427d] libusb: error [get_usbfs_fd] libusb couldn't open USB device /dev/bus/usb/003/003, errno=13
[ 2.949972] [0001427f] libusb: debug [handle_control_completion] handling completion status -71
[ 2.949975] [0001427d] libusb: error [get_usbfs_fd] libusb requires write access to USB device nodes
[ 2.949996] [0001427d] libusb: debug [libusb_open] open 3.3 returns -3
No access to device 13d3:56bc, skipping transfer tests.
[ 2.950006] [0001427d] libusb: debug [libusb_get_device_descriptor]  
[ 2.950010] [0001427d] libusb: debug [libusb_open] open 3.11
[ 2.949978] [0001427f] libusb: debug [handle_control_completion] low-level bus error -71

Error -71 is -EPROTO (see also https://www.kernel.org/doc/html/latest/driver-api/usb/error-codes.html)

It turned out this was always related to a "Alcor Micro Corp. AU9540 Smartcard Reader". I think it gets stressed by all the descriptor requests. When I disable the descriptor requests for this device, I don't get any errors any longer.