Several methods of disconnection from critical disconnection in linux

Several Methods for linux to break the critical disconnection-Linux general technology-Linux programming and kernel information. The following is a detailed description. When an interruption occurs, the global interruption is disabled to prevent the program from processing the interruption, and then to another interruption to disrupt the execution sequence, that is, to prevent interruption nesting. For example, in the irq_handler function, the interrupt should be disabled first. Alternatively, interruption is not allowed in some operation sequence. For example, a common driver method:

Unsigned int flag;
Local_irq_save (& flag );
............
Local_irq_restore (& flag );

Spin_loc_irqsave disallows interruption (only on the local processor) before obtaining the spin lock; the previous interruption status is saved in flags. if you are absolutely sure that the interrupt is not prohibited on your processor (or, in other words, you are sure that you should enable the interrupt when you release your spin lock), you can use the spin_lock_irq instead, and do not need to keep track of flags. finally, the software interruption is prohibited before the spin_lock_bh gets the lock, but the hardware interruption is reserved for open. I feel confused about the relationship between mutex and interruption. For example, after entering the critical section, an interruption occurs in the interrupt processing function, it also requires resources in the critical section, but it cannot be obtained. Is this interrupt processing function suspended into the waiting queue? Or is there a deadlock?