Sub-idle thread priority class

1 . A method comprising:

assigning, by a scheduler of a computer system, a sub-idle execution priority class to a processing thread associated with a request queue of an input/output (I/O) device;

identifying, by a processing device, a work completion request in the request queue; and

responsive to predicting, for a processor of the computer system, an idle time exceeding a threshold idle period and to identifying the work completion request in the request queue, running the processing thread to process the work completion request.

2 . The method of claim 1 , wherein the threshold idle period is an average period of time for receiving responses to requests sent by threads assigned to a priority class higher than the sub-idle execution priority class.

3 . The method of claim 1 , wherein the threshold idle period is a period of time associated with the processing thread, wherein the period of time exceeds a threshold period of time associated with the sub-idle execution priority class.

4 . The method of claim 1 , further comprising:

freeing a portion of memory associated with the work completion request.

5 . The method of claim 1 , wherein predicting, for the processor of the computer system, the idle time comprises:

monitoring the processor to detect a transition of the processor from an idle state to an active state;

recording an actual idle time of the processor; and

calculating, based on the actual idle time, the idle time.

6 . The method of claim 1 , wherein predicting, for the processor of the computer system, the idle time is based on an output of a trained machine learning model.

7 . The method of claim 1 , wherein the work completion request polls a completion queue for an entry associated with a work request in the request queue.

8 . A system comprising:

a memory; and

a processing device communicatively coupled to the memory to:

assign, by a scheduler of a computer system, a sub-idle execution priority class to a processing thread associated with a request queue of an input/output (I/O) device;

identify, by the processing device, a work completion request in the request queue; and

responsive to predicting, for a processor of the computer system, an idle time exceeding a threshold idle period and to identifying the work completion request in the request queue, running the processing thread to process the work completion request.

9 . The system of claim 8 , wherein the threshold idle period is an average period of time for receiving responses to requests sent by threads assigned to a priority class higher than the sub-idle execution priority class.

10 . The system of claim 8 , wherein the threshold idle period is a period of time associated with the processing thread, wherein the period of time exceeds the threshold idle period of time associated with the sub-idle execution priority class.

11 . The system of claim 8 , wherein the processing device is further to:

free a portion of memory associated with the work completion request.

12 . The system of claim 8 , wherein predicting, for the processor of the computer system, the idle time comprises:

monitoring the processor to detect a transition of the processor from an idle state to an active state;

recording an actual idle time of the processor; and

calculating, based on the actual idle time, the idle time.

13 . The system of claim 8 , wherein predicting, for the processor of the computer system, the idle time is based on an output of a trained machine learning model.

14 . The system of claim 8 , wherein the work completion request polls a completion queue for an entry associated with a work request in the request queue.

15 . A non-transitory computer readable storage medium including instructions that, when executed by a processing device, cause the processing device to perform a method comprising:

assigning, by a scheduler of a computer system, a sub-idle execution priority class to a processing thread associated with a request queue of an input/output (I/O) device;

identifying, by the processing device, a work completion request in the request queue; and

responsive to predicting, for a processor of the computer system, an idle time exceeding a threshold idle period and to identifying the work completion request in the request queue, running the processing thread to process the work completion request.

16 . The non-transitory computer readable storage medium of claim 15 , wherein the threshold idle period is an average period of time for receiving responses to requests sent by threads assigned to a priority class higher than the sub-idle execution priority class.

17 . The non-transitory computer readable storage medium of claim 15 , wherein the threshold idle period is a period of time associated with the processing thread, wherein the period of time exceeds a threshold period of time associated with the sub-idle execution priority class.

18 . The non-transitory computer readable storage medium of claim 15 , further comprising:

freeing a portion of memory associated with the work completion request.

19 . The non-transitory computer readable storage medium of claim 15 , further predicting, for the processor of the computer system, the idle time comprises:

monitoring the processor to detect a transition of the processor from an idle state to an active state;

recording an actual idle time of the processor; and

calculating, based on the actual idle time, the idle time.

20 . The non-transitory computer readable storage medium of claim 15 , wherein predicting, for the processor of the computer system, the idle time is based on an output of a trained machine learning model.