Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210729222635.2937453-1-sspatil@android.com> <20210729222635.2937453-2-sspatil@android.com>
 <CAHk-=wh-DWvsFykwAy6uwyv24nasJ39d7SHT+15x+xEXBtSm_Q@mail.gmail.com>
 <cee514d6-8551-8838-6d61-098d04e226ca@android.com> <CAHk-=wjStQurUzSAPVajL6Rj=CaPuSSgwaMO=0FJzFvSD66ACw@mail.gmail.com>
 <b1688f32-cb0e-04e1-3c91-aa8cddbcf41d@android.com>
In-Reply-To: <b1688f32-cb0e-04e1-3c91-aa8cddbcf41d@android.com>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 30 Jul 2021 15:06:00 -0700
Message-ID: <CAHk-=witY33b-vqqp=ApqyoFDpx9p+n4PwG9N-TvF8bq7-tsHw@mail.gmail.com>
Subject: Re: [PATCH 1/1] fs: pipe: wakeup readers everytime new data written
 is to pipe
To:     Sandeep Patil <sspatil@android.com>
Cc:     linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        David Howells <dhowells@redhat.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        stable <stable@vger.kernel.org>,
        Android Kernel Team <kernel-team@android.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Jul 30, 2021 at 12:47 PM Sandeep Patil <sspatil@android.com> wrote:
>
> aren't we supposed to wakeup on each write in level-triggered (default)
> case though?

No.

The thing about level triggered is that if the condition was already
true, it would not need a wakeup in the first place.

Put another way: select() and poll() are both fundamentally
level-triggered. If the condition was already true, they will return
success immediately, and don't need any extraneous wakeups.

This is literally an epoll() confusion about what an "edge" is.

An edge is not "somebody wrote more data". An edge is "there was no
data, now there is data".

And a level triggered event is *also* not "somebody wrote more data".
A level-triggered signal is simply "there is data".

Notice how neither edge nor level are about "more data". One is about
the edge of "no data" -> "some data", and the other is just a "data is
available".

Sadly, it seems that our old "we'll wake things up whether needed or
not" implementation ended up being something that people thought was
edge-triggered semantics.

But we have the policy that regressions aren't about documentation or
even sane behavior.

Regressions are about whether a user application broke in a noticeable way.

                     Linus