2.4.23 *_task_struct() observations ...

2.4.23 *_task_struct() observations ...

[Herbert Poetzl]

Hi All!

just wanted to get an opinion on the following ...

include/asm-i386/processor.h  defines ...

  #define alloc_task_struct() ((struct task_struct *) \
				__get_free_pages(GFP_KERNEL,1))
  #define free_task_struct(p)  free_pages((unsigned long) (p), 1)
  #define get_task_struct(tsk) atomic_inc(&virt_to_page(tsk)->count)

now there seems to be no put_task_struct(), but
there are some examples where get/free is used 
where I would expect a put_* ...

fs/proc/base.c for example does:

        read_lock(&tasklist_lock);
        task = find_task_by_pid(pid);
        if (task)
                get_task_struct(task);
        read_unlock(&tasklist_lock);
        if (!task)
                goto out;

        inode = proc_pid_make_inode(dir->i_sb, task, PROC_PID_INO);

        free_task_struct(task);

which is a little suspicious ...

please could anybody explain the logic behind that?
(if it isn't a widespread bug(tm))

TIA,
Herbert

Re: 2.4.23 *_task_struct() observations ...

[Russell King]

On Mon, Dec 01, 2003 at 06:26:20PM +0100, Herbert Poetzl wrote:
> just wanted to get an opinion on the following ...
> 
> include/asm-i386/processor.h  defines ...
> 
>   #define alloc_task_struct() ((struct task_struct *) \
> 				__get_free_pages(GFP_KERNEL,1))
>   #define free_task_struct(p)  free_pages((unsigned long) (p), 1)
>   #define get_task_struct(tsk) atomic_inc(&virt_to_page(tsk)->count)
> 
> now there seems to be no put_task_struct(), but
> there are some examples where get/free is used 
> where I would expect a put_* ...

put_* is free_*.  The reference count for the task struct is the page
count (virt_to_page(tsk)->count).

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
                 2.6 Serial core