We need parameters so that the module can read from conf file. Removing this from previous commit is my mistake.

We mainly talks about multi-processor.

It seems that we can't call register function right after unregistering.

This commit also adjusts some printed output and removes unused procfs item.

This commit also fixes a bug when removing linked list in local_irq_disable_trace.

This commit merges these two modules so we can manipulate irq_disable_detect through /proc/realtime_probe_tool.

Users now can clear the process_info list through the shell.

Please note that this commit change all tabs to 4 spaces.
Another important change is that we trace stacks in irqon_handler, avoid calling kfree. In fact, calling kfree in irqon_handler may end up in an infinite recursion because kfree calls local_irq_save and local_irq_restore.

We decide to choose tracepoint, which is presented here.

local_irq_disable_trace: Add initial support for multiprocessor systems and implement process info catching.
Note that currently there are two obvious bugs: wrong close time and memory leak. We will try to fix them.

This commit also fixes a bug which confuses function ktime_get_real_ts64 with ktime_get_ts64.

Users now can modify enable, irq, latence through the shell.
process_info output format through procfs is partially completed. Under development.

Note that currently it can't record files, and we need to limit the maximum length of the list.

Adjust the format and clarify a step

Still under development. On the strength of tracepoint our function can be called when IRQ status is altered.

This version is still under development; we need to store data into
linked lists and interact with procfs.

Check if the kernel linked list is empty in function 'proc_overwrite_switch_read'.