Re: [patch 1/3] dynamic printk - core infrastructure

[Joe Perches <joe@perches.com>]

On Fri, 2008-04-04 at 14:13 -0400, Jason Baron wrote:
> here is a re-spun patch incorporating comments from the thread.

> +#define dev_dbg(dev, format, arg...) do {				    \
> +	if (unlikely(imv_read(dynamic_debug_on)))			    \

I think the use of unlikely here and in pr_debug is unnecessary.

> +static struct hlist_head file_table[FILE_TABLE_SIZE] =
> +	{ [0 ... FILE_TABLE_SIZE-1] = HLIST_HEAD_INIT };
> +static DECLARE_MUTEX(debug_list_mutex);
> +static int nr_entries;

Shouldn't this be a module_table?

I think that hashing may be an unnecessary optimization and
that there's no real need to use a limited FILE_TABLE_SIZE
at all.  I think the most likely use cases for this facility
has either one or two modules being debugged or all modules
or no modules.

cheers, Joe