Re: [PATCH 3/7] ring-buffer: make moving the tail page a separatefunction

[Ingo Molnar]

* Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:

> 
> On Thu, 7 May 2009, Ingo Molnar wrote:
> 
> > 
> > * Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> > >  
> > > +static struct ring_buffer_event *
> > > +__rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
> > > +		  unsigned type, unsigned long length, u64 *ts)
> > > +{
> > > +	struct buffer_page *tail_page, *commit_page;
> > > +	struct ring_buffer_event *event;
> > > +	unsigned long tail, write;
> > > +
> > > +	commit_page = cpu_buffer->commit_page;
> > > +	/* we just need to protect against interrupts */
> > > +	barrier();
> > > +	tail_page = cpu_buffer->tail_page;
> > > +	write = local_add_return(length, &tail_page->write);
> > > +	tail = write - length;
> > > +
> > > +	/* See if we shot pass the end of this buffer page */
> > > +	if (write > BUF_PAGE_SIZE)
> > > +		return rb_move_tail(cpu_buffer, length, tail,
> > > +				    commit_page, tail_page, ts);
> > 
> > Nice! The __rb_reserve_next() fast-path logic became a lot clearer.
> 
> Thanks!
> 
> > 
> > The above branch might be unlikely(), right? With usual record sizes 
> > of around 40 bytes, we'll have a 100 records for every page 
> > overflow. That's i think within the reach of unlikely().
> > 
> > Depends on how much of a mess GCC makes of it though.
> 
> I looked at the assembly that gcc generates, and it is fine. gcc 
> inlines the function and puts it at the end, thus it already 
> treats it as an unlikely. I would like to avoid adding unlikely 
> annotations when possible.

Fair enough!

	Ingo
--
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/