[Devel] Re: [PATCH 3/7] Containers (V8): Add generic multi-subsystem API to containers

Srivatsa Vaddagiri vatsa at in.ibm.com
Wed Apr 11 09:42:53 PDT 2007


Coming from a "simplify things" pov:

On Fri, Apr 06, 2007 at 04:32:24PM -0700, menage at google.com wrote:
>  struct container {
>  	unsigned long flags;		/* "unsigned long" so bitops work */
> 
>  	/*
> -	 * Count is atomic so can incr (fork) or decr (exit) without a lock.
> -	 */
> -	atomic_t count;			/* count tasks using this container */
> -
> -	/*
>  	 * We link our 'sibling' struct into our parent's 'children'.
>  	 * Our children link their 'sibling' into our 'children'.
>  	 */
> @@ -43,11 +106,13 @@ struct container {
>  	struct list_head children;	/* my children */
> 
>  	struct container *parent;	/* my parent */
> -	struct dentry *dentry;		/* container fs entry */
> +	struct dentry *dentry;	  	/* container fs entry */
> 
> -#ifdef CONFIG_CPUSETS
> -	struct cpuset *cpuset;
> -#endif
> +	/* Private pointers for each registered subsystem */
> +	struct container_subsys_state *subsys[CONTAINER_SUBSYS_COUNT];
> +
> +	struct containerfs_root *root;

Could this root pointer derived from dentry pointer
(cont->dentry->d_sb->s_fs_info)?

> +	struct container *top_container;

and this as well?

	cont->dentry->d_sb->s_fs_info->top_container

>  };

So we have the foward subsys pointer array being stored in both 
'struct container' and 'struct container_group' and reverse container pointer 
stored in struct container_subsys_state. Can we reduce this pointer maze by:

	struct container {
		/* All shared stuff like flags, parent/child pointers etc */
		..

		struct container_group *my_group;
		
	}

The forward mapping from 'struct container' to subsys objects is made via
'my_group'. This also lets 'struct container' be a placeholder strictly for
shared state. 

On further thoughts, perhaps even my_group can be avoided by having:

	dentry->d_fsdata point to my_group 

and cont->dentry->d_fsdata will point to my_group which we wanted to store
above.

I don't see distinct adv of doing this, but I suspect it will simplify
the structure relationship (and code) a bit.

-- 
Regards,
vatsa




More information about the Devel mailing list