Re: [patch] cli()/sti() cleanup, 2.5.27-A2

[Christoph Hellwig <hch@lst.de>]

On Mon, Jul 22, 2002 at 03:01:48PM +0200, Ingo Molnar wrote:
> So what i did in my tree was to introduce the following 5 core means of
> manipulating the local interrupt flags:
> 
> 	irq_off()
> 	irq_on()
> 	irq_save(flags)
> 	irq_save_off(flags)
> 	irq_restore(flags)

I'd prefer the following:

void irq_off(void);
void irq_on(void);

flags_t irq_save();		/* the old irq_save_off() */
void irq_restore(flags_t);

void __irq_save(void);		/* without saveing */

rational:  proper function-like API (should be inlines), irq save
without disableing is very uncommon, better make the API symmetric.