[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20221203231122.GZ4001@paulmck-ThinkPad-P17-Gen-1>
Date: Sat, 3 Dec 2022 15:11:22 -0800
From: "Paul E. McKenney" <paulmck@...nel.org>
To: "stern@...land.harvard.edu" <stern@...land.harvard.edu>
Cc: Jonas Oberhauser <jonas.oberhauser@...wei.com>,
Boqun Feng <boqun.feng@...il.com>,
"parri.andrea@...il.com" <parri.andrea@...il.com>,
"will@...nel.org" <will@...nel.org>,
"peterz@...radead.org" <peterz@...radead.org>,
"npiggin@...il.com" <npiggin@...il.com>,
"dhowells@...hat.com" <dhowells@...hat.com>,
"j.alglave@....ac.uk" <j.alglave@....ac.uk>,
"luc.maranget@...ia.fr" <luc.maranget@...ia.fr>,
"akiyks@...il.com" <akiyks@...il.com>,
"dlustig@...dia.com" <dlustig@...dia.com>,
"joel@...lfernandes.org" <joel@...lfernandes.org>,
"urezki@...il.com" <urezki@...il.com>,
"quic_neeraju@...cinc.com" <quic_neeraju@...cinc.com>,
"frederic@...nel.org" <frederic@...nel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] tools: memory-model: Make plain accesses carry
dependencies
On Sat, Dec 03, 2022 at 04:32:19PM -0500, stern@...land.harvard.edu wrote:
> On Sat, Dec 03, 2022 at 12:44:05PM -0800, Paul E. McKenney wrote:
> > On Sat, Dec 03, 2022 at 03:34:20PM -0500, stern@...land.harvard.edu wrote:
> > > On Sat, Dec 03, 2022 at 11:02:26AM -0800, Paul E. McKenney wrote:
> > > > On Sat, Dec 03, 2022 at 11:58:36AM +0000, Jonas Oberhauser wrote:
> > > > >
> > > > >
> > > > > -----Original Message-----
> > > > > From: Boqun Feng [mailto:boqun.feng@...il.com]
> > > > > Sent: Friday, December 2, 2022 7:50 PM
> > > >
> > > > > > I wonder is this patch a first step to solve the OOTA problem you reported in OSS:
> > > > > > https://static.sched.com/hosted_files/osseu2022/e1/oss-eu22-jonas.pdf
> > > > > > If so maybe it's better to put the link in the commit log I think.
> > > > >
> > > > > It's not directly related to that specific problem, it does solve some other OOTA issues though.
> > > > > If you think we should link to the talk, there's also a video with slightly more updated slides from the actual talk: https://www.youtube.com/watch?v=iFDKhIxKhoQ
> > > > > do you think I should link to both then?
> > > >
> > > > It is not hard for me to add that in if people believe that it should be
> > > > included. But default is lazy in this case. ;-)
> > >
> > > I don't think there's any need to mention that video in the commit log.
> > > It's an introductory talk, and it's pretty safe to assume that anyone
> > > reading the commit because they are interested in the LKMM in great
> > > detail is already beyond the introductory level.
> > >
> > > On the other hand, it wouldn't hurt to include a Link: tag to an
> > > appropriate message in this email thread. (I leave it up to Paul to
> > > decide which message is most "appropriate" -- there may not be a good
> > > candidate, because a lot of the messages were not CC'ed to LKML.)
> >
> > For this approach, I would add this:
> >
> > Link: https://lore.kernel.org/all/4262e55407294a5989e766bc4dc48293@huawei.com/
>
> There's no point including that link; it merely points to messages
> containing or commenting on early versions of the commit. It adds very
> little information not already present in the commit itself. (Have you
> read any of Linus's criticisms of the Link: tags that people tend to
> include in patches they send him? It's the same principle.)
>
> I was thinking of the discussion which led up to the commit being
> written, where Jonas first brought up the idea that plain accesses
> should be able to carry dependencies just like accesses to registers.
> That's the sort of thing which would give readers some context and
> understanding of the reasoning behind the commit. They were part of the
> thread with the subject "RE: Interesting LKMM litmus test".
>
> But I can't find those messages on lore.kernel.org (which isn't
> surprising, as they weren't CC'ed to any mailing lists).
I am OK with them being made public, maybe in a Google document or
some such.
We would of course also need the consent of everyone else on that thread.
> > I could of course do both the extra paragraph -and- the Link:. ;-)
> >
> > Thoughts?
>
> My advice: Omit them both.
It would be good to reference something or another. ;-)
Thanx, Paul
Powered by blists - more mailing lists