Re: [PATCH RFC 0/5] rcu doc updates for whatisRCU and checklist

[Joel Fernandes]

On Fri, Oct 05, 2018 at 07:46:28PM -0400, Theodore Y. Ts'o wrote:
> On Fri, Oct 05, 2018 at 04:18:09PM -0700, Joel Fernandes (Google) wrote:
> > 
> > Here are this week's rcu doc updates based on combing through whatisRCU and
> > checklists. Hopefully you agree with them. I left several old _bh and _sched
> > API references as is, since I don't think its a good idea to remove them till
> > the APIs themselves are removed, however I did remove several of them as well
> > (like in the first patch in this series) since I feel its better to "encourage"
> > new users not to use the old API.
> 
> Hi Joel,

Hi Ted,

> 
> As it so happens, I just recently wrote my first RCU patch[1] (file
> systems, especially on-disk data structures, generally tend not to be
> good candidates for RCU semantics).
> 
> [1] http://patchwork.ozlabs.org/patch/979779/
> 
> So if you are working on improving RCU documentation, I thought I
> would give two comments on the RCU docs from the perspective of a
> developer trying to use RCU for the first time.
> 
> * whatisRCU is great, but one the example in Section 3 uses
>   rcu_dereference_protected() without explaining it.  Given that using
>   that function seems to be considered best practice, maybe a few more
>   words there would be in order?  That function isn't mentioned in
>   rcu.txt either, BTW.

I actually felt the same about rcu_dereference_protected while reading and
then looked at the comment above the implementation. The code comments are
pretty detailed, but I agree the example should mention a few words about it
since it uses it. I could look into improving that, no problem.

> * lockdep.txt *does* explain what rcu_dereference_protected() does,
>   but it doesn't really describe lockdep_is_held().  You can mostly
>   figure it out from context, but it wasn't obvious to me what locks
>   it could be used against, and in the case of a rw_semaphore, whether
>   it applied to shared as well as exclusive locks.  That's a lockdep
>   abstraction, and not a RCU abstraction, but lockdep isn't
>   particularly well documented, so I ended up spending 20-30 minutes
>   or so looking at the lockdep implementation before I was sure it
>   actually worked the way I thought it was going to.

Ok, makes sense to improve it. Since I haven't yet looked through lockdep.txt
yet (as a part of my broader documentation effort for the RCU consolidation),
I can take up improving that based on your suggestions since I have to look
into it anyway :). I'll look into that next week and CC you on this.

> Anyway, I was going to put submitting a patch to improve whatisRCU on
> my (vastly over-long) TODO list, but when I saw your patch set, I
> couldn't resist trying to see if I could fob it off on you.  If you
> don't think that's fair (and it probably isn't really), just let me
> know, and I'll put it back on my todo list.  :-)

Its Ok :) I'm happy to help! thanks for letting me know.

Best,

 - Joel