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: <20201220095830.17843-1-hca@linux.ibm.com> <CAK8P3a1NPMbKnuZ7-b0qbVVtLHT2YG6eEtzKUOf4AZkcu5VO4g@mail.gmail.com>
 <CA+FuTSeKfSouk7SV0cfuZ122J3fAc_ddy_a9unyCRq4F7f6+pw@mail.gmail.com> <CAHk-=wjXFfeCHdhAwH8ciXhMJwAc3dOnCAug4G8=eb=+zFnkTQ@mail.gmail.com>
In-Reply-To: <CAHk-=wjXFfeCHdhAwH8ciXhMJwAc3dOnCAug4G8=eb=+zFnkTQ@mail.gmail.com>
From:   Arnd Bergmann <arnd@kernel.org>
Date:   Sun, 20 Dec 2020 21:14:34 +0100
Message-ID: <CAK8P3a37qym_RKvCcKn1o53qBuA1gEk4FwPJ644LcPSJW-DrfA@mail.gmail.com>
Subject: Re: [PATCH] epoll: fix compat syscall wire up of epoll_pwait2
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Willem de Bruijn <willemdebruijn.kernel@gmail.com>,
        Andy Lutomirski <luto@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        Peter Anvin <hpa@zytor.com>,
        Heiko Carstens <hca@linux.ibm.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        Al Viro <viro@zeniv.linux.org.uk>,
        Arnd Bergmann <arnd@arndb.de>,
        Matthew Wilcox <willy@infradead.org>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>,
        Thomas Bogendoerfer <tsbogend@alpha.franken.de>,
        Vasily Gorbik <gor@linux.ibm.com>,
        Christian Borntraeger <borntraeger@de.ibm.com>,
        "David S. Miller" <davem@davemloft.net>,
        "the arch/x86 maintainers" <x86@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Sun, Dec 20, 2020 at 7:51 PM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> On Sun, Dec 20, 2020 at 10:22 AM Willem de Bruijn
> <willemdebruijn.kernel@gmail.com> wrote:
> >
> > Slightly tangential, it's not immediately clear to me why in
> > arch/x86/entry/syscalls/syscall_32.tbl epoll_pwait does not need a
> > compat entry, unlike on other architectures and unlike signalfd.
>
> Hmm. Good question. That looks like a bug to me. Probably nobody
> noticed because it's so rarely used.
>
> Or maybe I'm missing something too.
>
> Adding x86 entry code people to the participants.

The sigset_t argument is actually compatible between x86-32 and x86-64
because

- The bits are in the same order on little-endian machines
- _NSIG is the same as _COMPAT_NSIG (unlike old sparc kernels)
- accessing a 64-bit with 32-bit alignment is always allowed on x86

All other architectures with compat mode support big-endian
code at least as an option, so they have to use the compat
version.

         Arnd