Re: [PATCH 0/3] v4 Improve task->comm locking situation

[KOSAKI Motohiro]

(2011/05/18 12:16), Tetsuo Handa wrote:
> What about replacing
> 	char comm[TASK_COMM_LEN];
> with
> 	char *rcu_commname;
> and switching it atomically using RCU?
> 
> Advantages:
>    Readers can use RCU read lock rather than spinlock.
>    Makes task_struct smaller.
> 
> Disadvantages:
>    Need to allocate TASK_COMM_LEN bytes upon dup_task_struct() and set_task_comm().
>    Need to rewrite all task_struct.comm readers (steps shown below).
> 
> Steps to rewrite task_struct.comm readers
> 
> (1) Introduce a temporary accessor (say, task_comm).
> 
>      #define task_comm(tsk) (tsk)->comm
> 
> (2) Rewrite all tsk->comm users to task_comm(tsk).
> 
> (3) Replace sizeof(tsk->comm) with TASK_COMM_LEN.

The problem is, they aren't a few. But if you have a enough brave,
you can rewrite all. I'm not against it.




--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/