On Wed, May 07, 2025 at 07:47:22PM -0400, Steven Rostedt wrote: > On Tue, 6 May 2025 17:47:57 +0100 > Vincent Donnefort <vdonnef...@google.com> wrote: > > > > diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h > > index 56e27263acf8..c0c7f8a0dcb3 100644 > > --- a/include/linux/ring_buffer.h > > +++ b/include/linux/ring_buffer.h > > @@ -248,4 +248,67 @@ int ring_buffer_map(struct trace_buffer *buffer, int > > cpu, > > struct vm_area_struct *vma); > > int ring_buffer_unmap(struct trace_buffer *buffer, int cpu); > > int ring_buffer_map_get_reader(struct trace_buffer *buffer, int cpu); > > + > > +#define meta_pages_lost(__meta) \ > > + ((__meta)->Reserved1) > > +#define meta_pages_touched(__meta) \ > > + ((__meta)->Reserved2) > > Hmm, I wonder if this would be worth adding to the user interface?
Would trace-cmd have any use for those fields? That said, even if it does not at the moment, it would mean the meta-page has a single version between kern/user-space and hyp/kern which is probably better? I can add an additional patch in this series to extend the meta-page. [...]
