Re: [RFC 4/4] io_uring: implement futex wait

From: Pavel Begunkov
Date: Mon Jun 07 2021 - 08:15:54 EST

On 6/5/21 3:09 AM, Thomas Gleixner wrote:
> Here is _all_ the information you provided:
> 0) Cover letter:
> > Proof of concept for io_uring futex requests. The wake side does
> > FUTEX_WAKE_OP type of modify-compare operation but with a single
> > address. Wait reqs go through io-wq and so slow path.
> Describes WHAT it is supposed to do, but not at all WHY.
> Plus it describes it in terms which are maybe understandable for
> io-uring aware people, but certainly not for the general audience.

I actually agree with that and going to add it once I get details
I needed.

> > Should be interesting for a bunch of people, so we should first outline
> > API and capabilities it should give.
> You post patches which _should_ be interesting for a unspecified bunch
> of people, but you have no idea what the API and capabilities should
> be?

That's word carping. Some of the cases were known, but was more
interested atm in others I heard only a brief idea about, that's
why that person was CC'ed.

> IOW, this follows the design principle of: Throw stuff at the wall and
> see what sticks?

Exactly what it is *not*. Emails were chosen to clarify details,
nobody tells it wouldn't be reworked and adjusted. Do you imply
I should discuss ideas privately?

> But at the same time you want feedback from the people responsible for
> the subsystems you are modifying without providing the required
> information and worse:
> > As I almost never had to deal with futexes myself, would especially
> > love to hear use case, what might be lacking and other blind spots.
> So you came up with a solution with no use case and expect the futex
> people or whoever to figure out what you actually want to solve?

Again, not true. Where did you get that?

> Now let me quote Documentation/process/submitting-patches.rst:
> "Describe your problem. Whether your patch is a one-line bug fix or
> 5000 lines of a new feature, there must be an underlying problem that
> motivated you to do this work. Convince the reviewer that there is a
> problem worth fixing and that it makes sense for them to read past the
> first paragraph."
> Can you seriously point me to a single sentence in the above verbatim
> quotes from your cover letter and changelogs which complies with that rule?
> It does not matter whether this is RFC or not. You simply ignore well
> documented rules and then you get upset because I told you so:
> > 1) The proposed solution: I can't figure out from the changelogs or the
> > cover letter what kind of problems it solves and what the exact
> > semantics are. If you ever consider to submit futex patches, may I
> > recommend to study Documentation/process and get some inspiration
> > from git-log?
> And what's worse, you get impertinent about it:

Impertinent? Was just keeping up with your nice way of conveying
ideas. FWIW, it's not in particularly related to this small chunk
above at all.

> > I'm sorry you're incapable of grasping ideas quick
> Sure. I'm incompetent and stupid just because I can't figure out your
> brilliant ideas which are so well described - let me quote again:

That's your own interpretation, can't help you with that

> What's galling about that?
> - You wasted _my_ time by _not_ providing the information which I need
> to digest your submission.
> - I went way beyond what Documentation/process/ says and read past the
> first paragraph of useless information.
> - I provided you a detailed technical feedback nevertheless
> And as a result you attack me at a non-technical level. So where exactly
> is the "we" and who started galling?

If you think it was an attack, your response might have been interpreted
in a such way as well, even though it haven't by me. There are enough of
weird phrases and implications in your reply, but I have no intention
of going through it and picking up on every phrase, would be useless

>> Exactly why there was "we". I have my share of annoyance, which I would
>> readily put aside if that saves me time.
> I grant you to be annoyed as much as you want. But you are getting
> something fundamentaly wrong:
> "which I would readily put aside if that saves me time."
> As I told you above: You have been already wasting _my_ time by not
> providing the information which is required to actually look at what you
> propose.
>> Exactly why there was "we". I have my share of annoyance, which I would
>> readily put aside if that saves me time. And that's the suggestion
>> made
> In my first reply I made that a recommendation, so let me rephrease
> that:
> Read and comply to Documentation/process!
> It does not matter at all how brilliant the idea you have is and how
> stupid the reviewer at the other end might be. There are still rules to
> follow and they apply to the most brilliant people on the planet.
> So, as I told you before: Try again.

Pavel Begunkov