Does 'create_workqueue()' create a new worker thread? - linux-kernel

Question

I want to create a new work queue using create_workqueue()
The author of the book I'm reading about kernel development says "This function creates all the worker threads (one for each processor in the system) and prepares them to handle work."
My code shown at the end of this question, creates a kernel and schedules work on two queues. One is a default workqueue, and the other uses customized workqueue. They should by handled by different worker threads.
However, I see in the results (shown below) both are handled by same process (PID 42501), which is the thread kworker2 in my virtual machine.
Run result:
#include <linux/workqueue.h>
#include "kn_common.h"
#include <linux/sched.h>
MODULE_LICENSE("Dual BSD/GPL");
static void my_work_func(struct work_struct *work){
long ID;
printk(KERN_ALERT"=============\n");
print_current_time(0);
ID = current->pid;
printk(KERN_ALERT"my workqueue function is called!.... pid = %ld\n", ID);
printk(KERN_ALERT"=============\n");
}
static void my_work_custom_func(struct work_struct *work){
long ID;
printk(KERN_ALERT"=============\n");
print_current_time(0);
ID = current->pid;
printk(KERN_ALERT"my customize workqueue is called!... pid = %ld\n", ID);
printk(KERN_ALERT"=============\n");
}
DECLARE_WORK(mywork, my_work_func);
static int testworkqueue_init(void){
struct workqueue_struct *myworkqueue = create_workqueue("myworkqueue");
// init a work_struct dynamically use pointer
struct work_struct *mywork2;
mywork2 = kmalloc(sizeof(struct work_struct), GFP_KERNEL);
INIT_WORK(mywork2, my_work_custom_func);
flush_scheduled_work();
// schedule work
schedule_work(&mywork);
// flush customized workqueue
flush_workqueue(myworkqueue);
queue_work(myworkqueue, mywork2);
return 0;
}
static void testworkqueue_exit(void){
printk(KERN_ALERT"*************\n");
print_current_time(0);
printk(KERN_ALERT"testworkqueu exit\n");
printk(KERN_ALERT"*************\n");
}
module_init(testworkqueue_init);
module_exit(testworkqueue_exit);
And Makefile
obj-m += myworkqueue.o
myworkqueue-objs := testworkqueue.o kn_common.o
CURRENT_PATH := $(shell pwd)
LINUX_KERNEL := $(shell uname -r)
# you may change this to your own kernel src path
LINUX_KERNEL_PATH := /lib/modules/$(LINUX_KERNEL)/build
all:
make -C $(LINUX_KERNEL_PATH) M=$(CURRENT_PATH) modules
rm -rf modules.order Module.symvers .*.cmd *.o *.mod.c .tmp_versions *.unsigned
clean:
rm -rf modules.order Module.symvers .*.cmd *.o *.mod.c *.ko .tmp_versions *.unsigned

Answer 1

Originally implemented Multi thread(MT) Workqueue wasted a lot of resource, the level of concurrency provided was unsatisfactory.
New design has introduced to get high level of concurrency. Functions "create_*workqueue()" are deprecated and scheduled for removal.
Please read this for latest implementation of workqueue.