Re: [PATCH] removes filp_count_lock and changes nr_files type to atomic_t

[Arjan van de Ven <arjan@infradead.org>]

On Thu, 2005-08-25 at 10:45 +0200, Eric Dumazet wrote:
> This patch removes filp_count_lock spinlock, used to protect files_stat.nr_files.
> 
> Just use atomic_t type and atomic_inc()/atomic_dec() operations.
> 
> This patch assumes that atomic_read() is a plain {return v->counter;} on all 
> architectures. (keywords : sysctl, /proc/sys/fs/file-nr, proc_dointvec)
> 

this patch adds atomic ops where there were none before
>  static inline void file_free(struct file *f)
> @@ -70,7 +62,7 @@
>  	/*
>  	 * Privileged users can go above max_files
>  	 */
> -	if (files_stat.nr_files >= files_stat.max_files &&
> +	if (atomic_read(&files_stat.nr_files) >= files_stat.max_files &&
>  				!capable(CAP_SYS_ADMIN))
>  		goto over;
>  

here 

> @@ -94,10 +86,10 @@
>  
>  over:
>  	/* Ran out of filps - report that */
> -	if (files_stat.nr_files > old_max) {
> +	if (atomic_read(&files_stat.nr_files) > old_max) {
>  		printk(KERN_INFO "VFS: file-max limit %d reached\n",
>  					files_stat.max_files);
> -		old_max = files_stat.nr_files;
> +		old_max = atomic_read(&files_stat.nr_files);
>  	}
>  	goto fail;

and here

for those architectures that need atomics for read (parisc? arm?)

however.. wouldn't it be better to make this a per cpu variable for
write, and for read iterate or do something smart otherwise?