What approach allows you to move error handling to a separate file?

Question

Software design

Hello,
I understand that I asked a bit obliquely, I will try to explain with an example. When we work, for example, with libusb, we write something like this

ret  = libusb_init(&ctx); //Инициализация библиотеки
if (ret < 0) { //Что-то странное, libusb работать не хочет
    goto init_err_mark; 
}
devh = libusb_open_device_with_vid_pid(ctx, 1234, 5285); 
if(devh == NULL) { //Устройство не открылось
    goto devh_err_mark;
}
ret = libusb_claim_interface(devh, 0); 
if(ret < 0) { //Интерфейс не открывается
    goto iface_err_mark;
}
... //Собственно работаем с устройством
//Освобождаем ресурсы
    libusb_release_interface(devh, 0);
iface_err_mark: 
    libusb_close(devh);
devh_err_mark:
    libusb_exit(ctx);
init_err_mark:
    return ret;

those. First, we initialize the resource, and an error at any initialization step means termination of work, and then we release the resources in the reverse order. Such code is a little inconvenient to read and write, you get distracted by the "something went wrong" branches. I ask you to suggest a way to write scripts for correct operation and error handling separately, I have monads, future and promise mixed up in my head.