Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 30 Mar 2023 15:48:48 +0100
From:   Vincent Donnefort <vdonnefort@google.com>
To:     Steven Rostedt <rostedt@goodmis.org>
Cc:     mhiramat@kernel.org, linux-kernel@vger.kernel.org,
        linux-trace-kernel@vger.kernel.org, kernel-team@android.com
Subject: Re: [PATCH v2 1/2] ring-buffer: Introducing ring-buffer mapping
 functions
Message-ID: <ZCWhUJkD0cgSjwkW@google.com>
References: <20230322102244.3239740-1-vdonnefort@google.com>
 <20230322102244.3239740-2-vdonnefort@google.com>
 <20230328224411.0d69e272@gandalf.local.home>
 <ZCQCsD9+nNwBYIyH@google.com>
 <20230329070353.1e1b443b@gandalf.local.home>
 <ZCQtpbyWrjliJkdg@google.com>
 <20230329084735.6c4a9229@rorschach.local.home>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230329084735.6c4a9229@rorschach.local.home>
Precedence: bulk

[...]

> > > struct ring_buffer_meta_page_header {
> > > #if __BITS_PER_LONG == 64
> > > 	__u64	entries;
> > > 	__u64	overrun;
> > > #else
> > > 	__u32	entries;
> > > 	__u32	overrun;
> > > #endif
> > > 	__u32	pages_touched;
> > > 	__u32	meta_page_size;
> > > 	__u32	reader_page;	/* page ID for the reader page */
> > > 	__u32	nr_data_pages;	/* doesn't take into account the reader_page */
> > > };
> > > 
> > > BTW, shouldn't the nr_data_pages take into account the reader page? As it
> > > is part of the array we traverse isn't it?  
> > 
> > It depends if the reader page has ever been swapped out. If yes, the reader
> > would have to start from reader_page and then switch to the data_pages.
> > Which sounds like a fiddly interface for the userspace.
> > 
> > So yeah, consuming-read only feels like a better start.
> > 
> 
> I agree. I'd like to get something in that can be extended, but simple
> enough that it's not too much of a barrier wrt getting the API correct.
> 
> -- Steve

Something I just realized though. In the event of being able to upstream the
hypervisor tracing based on the ring_buffer_meta_page, without non-consumming
support, we wouldn't have the "trace" file which is used to reset the buffers.

I'd guess we'd have to either create one that is read-only (a bit strange) or
let trace_pipe reset the buffer(s).