Re: [PATCH] tty: vt: make do_con_write() no-op if IRQ is disabled

From: Linus Torvalds <torvalds@linux-foundation.org>
To: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: "Fabio M. De Francesco" <fmdefrancesco@gmail.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Peter Hurley <peter@hurleysoftware.com>,
	Jiri Slaby <jirislaby@kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH] tty: vt: make do_con_write() no-op if IRQ is disabled
Date: Wed, 1 Dec 2021 11:05:09 -0800	[thread overview]
Message-ID: <CAHk-=wjVL_CLm-+=7qf2obF6f8D+ujysmqp5dKdAb7UEyo1cZg@mail.gmail.com> (raw)
In-Reply-To: <3add7ade-9c9b-2839-5a0c-0a38c4be0e34@i-love.sakura.ne.jp>

On Wed, Dec 1, 2021 at 5:41 AM Tetsuo Handa
<penguin-kernel@i-love.sakura.ne.jp> wrote:
>
> What do you think? Can we apply this?

I think this patch is only papering over the problem, and the issue goes deeper.

It may be that "papering over the issue" successfully hides it
completely, but it's still a horribly bad approach.

> > -     if (in_interrupt())
> > +     if (in_interrupt() || irqs_disabled())
> >               return count;

This kind of stuff is broken. Pretty much always.

And in this case, it's still broken, because things like "called under
a non-irq spinlock" would still not show up.

And no, I do *not* mean that the code should try to figure that out. I
mean that the problem goes further up, and that the fact that we get
to do_con_write() in the first place when we're in an invalid context
is wrong, wrong, wrong.

How the heck do we get here from just an ioctl?

Looking at the backtrace, I see

   n_hdlc_send_frames+0x24b/0x490 drivers/tty/n_hdlc.c:290
   tty_wakeup+0xe1/0x120 drivers/tty/tty_io.c:534
   __start_tty drivers/tty/tty_io.c:806 [inline]
   __start_tty+0xfb/0x130 drivers/tty/tty_io.c:799

and apparently it's that hdlc line discipline (and
n_hdlc_send_frames() in particular) that is the problem here.

I think that's where the fix should be.

                Linus