Re: [PATCH v10 1/2] ring-buffer: Introducing ring-buffer mapping functions

[Vincent Donnefort]

On Tue, Jan 09, 2024 at 06:58:13PM -0500, Steven Rostedt wrote:
> On Wed, 10 Jan 2024 08:42:05 +0900
> Masami Hiramatsu (Google) <mhiramat@xxxxxxxxxx> wrote:
> 
> > On Tue, 9 Jan 2024 15:13:51 +0000
> > Vincent Donnefort <vdonnefort@xxxxxxxxxx> wrote:
> > 
> > > > > @@ -388,6 +389,7 @@ struct rb_irq_work {
> > > > >  	bool				waiters_pending;
> > > > >  	bool				full_waiters_pending;
> > > > >  	bool				wakeup_full;
> > > > > +	bool				is_cpu_buffer;  
> > > > 
> > > > I think 'is_cpu_buffer' is a bit unclear (or generic),
> > > > what about 'meta_page_update'?  
> > > 
> > > Hum not sure about that change. This was really to identify if parent of
> > > rb_irq_work is a cpu_buffer or a trace_buffer. It can be a cpu_buffer regardless
> > > of the need to update the meta-page.  
> > 
> > Yeah, I just meant that is "for_cpu_buffer", not "rb_irq_work is_cpu_buffer".
> > So when reading the code, I just felt uncomfortable.
> > 
> 
> How about "in_cpu_buffer" as that is what it is.
> 
> struct ring_buffer_per_cpu {
> 	struct rb_irq_work {
> 		bool	in_cpu_buffer;
> 	}
> }
> 
> Would that make you feel more comfortable? ;-)
> 
> -- Steve

I'll actually solve that by moving that update from the rb_irq_work to
ring_buffer_map_get_reader().

Reason is the rb_irq_work is only triggered when !O_NONBLOCK is set.

> 
> -- 
> To unsubscribe from this group and stop receiving emails from it, send an email to kernel-team+unsubscribe@xxxxxxxxxxx.
>