Re: [PATCH v8 08/12] user_events: Optimize writing events by only copying data once

[Beau Belgrave <beaub@linux.microsoft.com>]

On Thu, Dec 23, 2021 at 12:11:42AM +0900, Masami Hiramatsu wrote:
> On Thu, 16 Dec 2021 09:35:07 -0800
> Beau Belgrave <beaub@linux.microsoft.com> wrote:
> 
> > Pass iterator through to probes to allow copying data directly to the
> > probe buffers instead of taking multiple copies. Enables eBPF user and
> > raw iterator types out to programs for no-copy scenarios.
> 
> This looks good to me, except for 1 nitpick. See below.
> 
> Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
> 
> > 
> > Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
> > ---
> >  kernel/trace/trace_events_user.c | 113 +++++++++++++++++++++++--------
> >  1 file changed, 85 insertions(+), 28 deletions(-)
> > 
> > diff --git a/kernel/trace/trace_events_user.c b/kernel/trace/trace_events_user.c
> > index cc30d1fcbb63..fa3e26281fc3 100644
> > --- a/kernel/trace/trace_events_user.c
> > +++ b/kernel/trace/trace_events_user.c
> > @@ -41,6 +41,9 @@
> >  #define MAX_FIELD_ARRAY_SIZE 1024
> >  #define MAX_FIELD_ARG_NAME 256
> >  

[..]

> > -static void user_event_ftrace(struct user_event *user, void *data, u32 datalen,
> > +static void user_event_ftrace(struct user_event *user, struct iov_iter *i,
> >  			      void *tpdata)
> >  {
> >  	struct trace_event_file *file;
> > @@ -540,41 +556,85 @@ static void user_event_ftrace(struct user_event *user, void *data, u32 datalen,
> >  
> >  	/* Allocates and fills trace_entry, + 1 of this is data payload */
> >  	entry = trace_event_buffer_reserve(&event_buffer, file,
> > -					   sizeof(*entry) + datalen);
> > +					   sizeof(*entry) + i->count);
> >  
> >  	if (unlikely(!entry))
> >  		return;
> >  
> > -	memcpy(entry + 1, data, datalen);
> > +	if (unlikely(!copy_nofault(entry + 1, i->count, i))) {
> > +		__trace_event_discard_commit(event_buffer.buffer,
> > +					     event_buffer.event);
> > +		return;
> > +	}
> >  
> >  	trace_event_buffer_commit(&event_buffer);
> 
> Will this be 
> 
> 	if (unlikely(!copy_nofault(entry + 1, i->count, i)))
> 		__trace_event_discard_commit(event_buffer.buffer,
> 					     event_buffer.event);
> 	else
> 		trace_event_buffer_commit(&event_buffer);
> 
> ?

Sure, why not.

Thanks,
-Beau