Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20191126110659.GA14042@redhat.com> <20191126110758.GA14051@redhat.com>
 <CAHk-=whrhuNg_53wc3pBVToH-AUwKDbC5P_cb7=8bYfn=BYCJA@mail.gmail.com>
In-Reply-To: <CAHk-=whrhuNg_53wc3pBVToH-AUwKDbC5P_cb7=8bYfn=BYCJA@mail.gmail.com>
From:   Andy Lutomirski <luto@kernel.org>
Date:   Tue, 26 Nov 2019 21:46:14 -0800
Message-ID: <CALCETrWTjUJQVXmbhnAfpg=J1kV8XZnWrjR7Yt+a1Dd2GSTr5A@mail.gmail.com>
Subject: Re: [PATCH] ptrace/x86: introduce TS_COMPAT_RESTART to fix get_nr_restart_syscall()
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Oleg Nesterov <oleg@redhat.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Andy Lutomirski <luto@kernel.org>,
        Borislav Petkov <bp@alien8.de>, Ingo Molnar <mingo@kernel.org>,
        Jan Kratochvil <jan.kratochvil@redhat.com>,
        Pedro Alves <palves@redhat.com>, Peter Anvin <hpa@zytor.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        "the arch/x86 maintainers" <x86@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, Nov 26, 2019 at 8:04 PM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> On Tue, Nov 26, 2019 at 3:08 AM Oleg Nesterov <oleg@redhat.com> wrote:
> >
> > Alternatively we could add ->compat_restart into struct restart_block,
> > logically this is the same thing.
>
> That sounds like the better model to me. That's what the restart_block
> is about: it's supposed to contain the restart information.
>
> I'd much rather see the system call number added into the restart
> block (or just the "compat bit" - but we have that X32 case too, so
> why not put it all there). And then the get_nr_restart_syscall() hack
> goes away and is just "set state from the restart block".
>
> How painful would that be? I guess right now we always just set all
> the restart_block info manually in all the restart cases, and that
> could make it a bit painful to add this kind of architecture-specific
> flag, but it _sounds_ conceptually like the right thing to do.

How about we rename restart_block::fn to __fn, add fields
restart_syscall_nr and restart_syscall_arch, and do:

long restart_block_activate(long (*fn)(struct restart_block *))
{
  current->restart_block.__fn = fn;
  arch_restart_block_activate();
  return -ERESTART_RESTARTBLOCK;
  current->restart_block.syscall_nr
}

IMO the ideal solution would be to add a new syscall nr to restart a
syscall and make it the same on all architectures.  This has
unfortunate interactions with seccomp, though.

--Andy