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;
Date:   Fri, 29 Mar 2019 14:39:56 +0100
From:   Borislav Petkov <bp@alien8.de>
To:     Jann Horn <jannh@google.com>
Cc:     Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>,
        "H. Peter Anvin" <hpa@zytor.com>, x86@kernel.org,
        linux-kernel@vger.kernel.org,
        Andrew Morton <akpm@linux-foundation.org>,
        Qiaowei Ren <qiaowei.ren@intel.com>,
        Ben Dooks <ben.dooks@codethink.co.uk>
Subject: Re: [PATCH 2/2] x86: fix __user annotations
Message-ID: <20190329133956.GE21152@zn.tnic>
References: <20190328212321.92463-1-jannh@google.com>
 <20190328212321.92463-2-jannh@google.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20190328212321.92463-2-jannh@google.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Mar 28, 2019 at 10:23:21PM +0100, Jann Horn wrote:
> Fix __user annotations in various places across the x86 tree:
> 
>  - cast to wrong pointer type in __user_atomic_cmpxchg_inatomic()
>  - generic_load_microcode() deals with a pointer that can be either a
>    kernel pointer or a user pointer; change the code to pass it around as
>    a __user pointer, and add explicit casts to convert between __user and
>    __kernel
>  - save_xstate_epilog() has missing __user in explicit casts
>  - setup_sigcontext() and x32_setup_rt_frame() rely on the cast performed
>    by put_user_ex() on its first argument, but sparse requires __force for
>    casting __user pointers to unsigned long
>  - xen_hvm_config() has missing __user
> 
> This patch removes all sparse warnings about the asn:1 address space
> (__user) in arch/x86/ for my kernel config.
> 
> Signed-off-by: Jann Horn <jannh@google.com>
> ---
> This patch requires the previous one, "[PATCH 1/2] kernel.h: use
> parentheses around argument in u64_to_user_ptr()", otherwise
> xen_hvm_config() breaks. Can we take both together through the x86 tree,
> or does the first one have to go through akpm's tree?

I don't see why not, unless akpm has objections.

However,

>  arch/x86/include/asm/uaccess.h        |  3 +--
>  arch/x86/include/asm/uaccess_64.h     |  2 +-

This chunk is being discussed here already:

https://lkml.kernel.org/r/20190228185027.2480-1-ben.dooks@codethink.co.uk

and I'd like to take Ben's v2 when Ben adds Linus' explanation.

Then, it would be probably easier if you could split that patch into:

>  arch/x86/kernel/cpu/microcode/intel.c | 20 ++++++++++++--------

microcode

>  arch/x86/kernel/fpu/signal.c          |  6 +++---
>  arch/x86/kernel/signal.c              |  4 ++--

fpu patch

>  arch/x86/kvm/x86.c                    |  8 ++++----

kvm patch

which would make review/merging/etc considerably easier.

Thx.

-- 
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.