Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <qk6hjuam54khlaikf2ssom6custxf5is2ekkaequf4hvode3ls@zgf7j5j4ubvw>
 <20230626-vorverlegen-setzen-c7f96e10df34@brauner> <4sdy3yn462gdvubecjp4u7wj7hl5aah4kgsxslxlyqfnv67i72@euauz57cr3ex>
 <20230626-fazit-campen-d54e428aa4d6@brauner> <qyohloajo5pvnql3iadez4fzgiuztmx7hgokizp546lrqw3axt@ui5s6kfizj3j>
 <CAHk-=wgmLd78uSLU9A9NspXyTM9s6C23OVDiN2YjA-d8_S0zRg@mail.gmail.com>
 <ZKinHejv+xBq+gti@casper.infradead.org> <CAHk-=wjJ8YP4wswYCC8X2o68vFeVzLesXbW-QdUgzzOZKaHJQw@mail.gmail.com>
In-Reply-To: <CAHk-=wjJ8YP4wswYCC8X2o68vFeVzLesXbW-QdUgzzOZKaHJQw@mail.gmail.com>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 7 Jul 2023 17:21:54 -0700
Message-ID: <CAHk-=widpMfo_Zh5zHqL0Ct_VNqsHiwFyLUdM8wm6NF8o6T5kQ@mail.gmail.com>
Subject: Re: Pending splice(file -> FIFO) excludes all other FIFO operations
 forever (was: ... always blocks read(FIFO), regardless of O_NONBLOCK on read side?)
To:     Matthew Wilcox <willy@infradead.org>
Cc:     =?UTF-8?Q?Ahelenia_Ziemia=C5=84ska?= 
        <nabijaczleweli@nabijaczleweli.xyz>,
        Christian Brauner <brauner@kernel.org>,
        Jens Axboe <axboe@kernel.dk>,
        David Howells <dhowells@redhat.com>,
        Alexander Viro <viro@zeniv.linux.org.uk>,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, 7 Jul 2023 at 17:07, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> > That is, do we want SIGWINCH or SIGALRM to result in a short read?
>
> Now, that's a different issue, and is actually handled by the signal
> layer: a signal that is ignored (where "ignored" includes the case of
> "default handler") will be dropped early, exactly because we don't
> want to interrupt things like tty or pipe reads when you resize the
> window.

In case you care, it's prepare_signal() -> sig_ignored() ->
sig_task_ignored() -> sig_handler_ignored() logic that does this
short-circuiting.

And while I don't think it's required by POSIX, this was definitely a
case where lots of programs that *don't* use any signal handlers at
all are very much not expecting to see -EINTR as a return value, and
used to break exactly on things like SIGWINCH when reading from a tty
or a pipe.

But that logic goes back to before linux-1.0.

In fact, I think it goes back to at least 0.99.10 (June '93).

                 Linus