Date:   Thu, 15 Dec 2022 14:10:14 -0800
From:   "Paul E. McKenney" <paulmck@...nel.org>
To:     Joel Fernandes <joel@...lfernandes.org>
Cc:     Frederic Weisbecker <frederic@...nel.org>, boqun.feng@...il.com,
        neeraj.iitr10@...il.com, urezki@...il.com, rcu@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH RFC] srcu: Yet more detail for
 srcu_readers_active_idx_check() comments

On Thu, Dec 15, 2022 at 04:42:15PM -0500, Joel Fernandes wrote:
> 
> 
> > On Dec 15, 2022, at 4:39 PM, Paul E. McKenney <paulmck@...nel.org> wrote:
> > 
> > On Thu, Dec 15, 2022 at 03:33:39PM -0500, Joel Fernandes wrote:
> >>> On Thu, Dec 15, 2022 at 3:03 PM Joel Fernandes <joel@...lfernandes.org> wrote:
> >>> 
> >>> Hi Paul,
> >>> 
> >>>> On Thu, Dec 15, 2022 at 2:58 PM Paul E. McKenney <paulmck@...nel.org> wrote:
> >>> [...]
> >>>>> If the first read section's srcu_read_unlock() and its corresponding
> >>>>> smp_mb()  happened before the flip, then the increment of old idx
> >>>>> would happen only once. The next srcu_read_lock() will read the new
> >>>>> index. If the srcu_read_unlock() and it's corresponding smp_mb()
> >>>>> happened after the flip, the old_idx will be sampled again and can be
> >>>>> incremented twice. So it depends on how the flip races with
> >>>>> srcu_read_unlock().
> >>>> 
> >>>> I do understand that a number of people like reasoning about
> >>>> memory-barrier ordering, courtesy of the sequentially consistent portions
> >>>> of the C and C++ memory models, but thinking in terms of the accesses
> >>>> surrounding the memory barriers has been far less error-prone.
> >>> 
> >>> Sure, but we are already talking in terms of the access to idx right?
> >>> That's what we're saying is visible by memory barriers and we are
> >>> trying to reason here about the ordering (flip does the write to idx
> >>> and followed by smp_mb(), and there is corresponding read of idx on
> >>> the srcu_read_lock() side. So we are indeed talking in terms of
> >>> access, but let me know if I missed something.
> >>> 
> >>>>> Also, since this is all hard to reason about I started making some
> >>>>> diagrams, LOL. For your amusement, here is why need to scan both idx
> >>>>> during grace period detection: https://i.imgur.com/jz4bNKd.png
> >>>> 
> >>>> Nice!
> >>>> 
> >>>> I suggest placing a gap between GP 2 and GP 3.  That way, you can make it
> >>>> very clear that Reader 1's critical section starts after the end of GP 2
> >>>> (thus clearly never blocking GP 2) and before GP 3 (thus possibly having
> >>>> a reference to some data that is going to be freed at the end of GP 3).
> >>>> 
> >>>> I also suggest coloring Reader 1 red and Reader 2 green, given that the
> >>>> color red generally indicates danger.
> >>> 
> >>> Thanks for these suggestions! I will make the update. I am planning to
> >>> make a number of diagrams for other scenarios as well, as it helps
> >>> visualize. Google drawing is nice for these. I am happy to share these
> >>> with you all if there is interest :).
> >> 
> >> I made these updates, please see: https://i.imgur.com/hoKLvtt.png
> >> 
> >> Feel free to use the image for any purpose and thanks ;-)
> > 
> > Very good, thank you!
> > 
> > Would it be possible to have an arrow marked "X" or "reference to X"
> > from the beginning of the 'Mark "x" for GC' box to the box labeled
> > 'Enter RSCS (access "X")'?
> 
> I am currently away from desk. I shared the google drawing with you. Could you check and make the change, if that’s ok with you?
> 
> Thank you so much,

I took a cut at it.  Thoughts?

							Thanx, Paul

Powered by blists - more mailing lists