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, 20 Mar 2020 16:46:33 +0100
From:   Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To:     Yonghong Song <yhs@fb.com>
Cc:     Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>,
        Andrii Nakryiko <andriin@fb.com>, netdev@vger.kernel.org,
        bpf@vger.kernel.org, linux-kernel@vger.kernel.org,
        Maciej =?utf-8?Q?=C5=BBenczykowski?= <maze@google.com>,
        John Stultz <john.stultz@linaro.org>,
        Alexander Potapenko <glider@google.com>,
        Alistair Delva <adelva@google.com>
Subject: Re: [PATCH] bpf: explicitly memset the bpf_attr structure
Message-ID: <20200320154633.GB765793@kroah.com>
References: <20200320094813.GA421650@kroah.com>
 <d0dfc2ff-8323-12ed-3c7e-e8c6a118b890@fb.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <d0dfc2ff-8323-12ed-3c7e-e8c6a118b890@fb.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Mar 20, 2020 at 08:23:57AM -0700, Yonghong Song wrote:
> 
> 
> On 3/20/20 2:48 AM, Greg Kroah-Hartman wrote:
> > For the bpf syscall, we are relying on the compiler to properly zero out
> > the bpf_attr union that we copy userspace data into.  Unfortunately that
> > doesn't always work properly, padding and other oddities might not be
> > correctly zeroed, and in some tests odd things have been found when the
> > stack is pre-initialized to other values.
> 
> Maybe add more contexts about the failure itself so it could be clear
> why we need this patch.

I didn't have the full details, I think Maciej has them though.

> As far as I know from the link below, the failure happens in
> CHECK_ATTR() which checks any unused *area* for a particular subcommand
> must be 0, and this patch tries to provide this guarantee beyond
> area beyond min(uattr_size, sizeof(attr)).

That macro also will get tripped up if padding is not zeroed out as
well, so this is good to fix up.

thanks for the review.

greg k-h