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: <20200227132826.195669-1-brgerst@gmail.com> <20200227132826.195669-6-brgerst@gmail.com>
In-Reply-To: <20200227132826.195669-6-brgerst@gmail.com>
From:   Andy Lutomirski <luto@kernel.org>
Date:   Fri, 28 Feb 2020 10:46:55 -0800
Message-ID: <CALCETrXyYd=pqThrXQgbz5cqQTr8SKHZey4FGq5aV2_=CS4p9Q@mail.gmail.com>
Subject: Re: [PATCH v3 5/8] x86: Move 32-bit compat syscalls to common location
To:     Brian Gerst <brgerst@gmail.com>
Cc:     LKML <linux-kernel@vger.kernel.org>, X86 ML <x86@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@kernel.org>, Borislav Petkov <bp@alien8.de>,
        "H . Peter Anvin" <hpa@zytor.com>,
        Andy Lutomirski <luto@kernel.org>,
        Dominik Brodowski <linux@dominikbrodowski.net>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Feb 27, 2020 at 5:28 AM Brian Gerst <brgerst@gmail.com> wrote:
>
> Move the 32-bit wrappers for syscalls that take 64-bit arguments (loff_t)
> to a common location so that native 32-bit can use them in preparation for
> enabling pt_regs-based syscalls.

Can you clarify the purpose?  Even having read the series up to this
point, I have no idea what this has to do with pt_regs.

I think some renaming is in order.  Consider:

>
> -COMPAT_SYSCALL_DEFINE3(x86_ftruncate64, unsigned int, fd,
> -                      unsigned long, offset_low, unsigned long, offset_high)

It used to be at least a little bit clear what was going on.  There's
this compat-only mess that changes arguments for ftruncate64.  But
now:

> +SYSCALL_DEFINE3(x86_ftruncate64, unsigned int, fd,
> +               unsigned long, offset_low, unsigned long, offset_high)
>  {
>         return ksys_ftruncate(fd, ((loff_t) offset_high << 32) | offset_low);
>  }

What is this "x86" ftruncate64 thing?

Maybe call it ia32_ftructate?  Or at least do something to indicate
that this is for a specific ABI.

> diff --git a/arch/x86/um/sys_call_table_32.c b/arch/x86/um/sys_call_table_32.c
> index 9649b5ad2ca2..d5520e92f89d 100644
> --- a/arch/x86/um/sys_call_table_32.c
> +++ b/arch/x86/um/sys_call_table_32.c
> @@ -26,6 +26,16 @@
>
>  #define old_mmap sys_old_mmap
>
> +#define sys_x86_pread sys_pread64
> +#define sys_x86_pwrite sys_pwrite64
> +#define sys_x86_truncate64 sys_truncate64
> +#define sys_x86_ftruncate64 sys_ftruncate64
> +#define sys_x86_readahead sys_readahead
> +#define sys_x86_fadvise64 sys_fadvise64
> +#define sys_x86_fadvise64_64 sys_fadvise64_64
> +#define sys_x86_sync_file_range sys_sync_file_range
> +#define sys_x86_fallocate sys_fallocate

Can this not be killed by changing the table itself instead of adding
a bunch of defines?