netdev - Re: [PATCH] net/mlx4_en: protect ring->xdp_prog with rcu_read

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CALx6S35+ndZ9+dPdWm7sOfQw75jrcdjSomzGZKd0JLJ19su2GA@mail.gmail.com>
Date:   Fri, 2 Sep 2016 12:14:20 -0700
From:   Tom Herbert <tom@...bertland.com>
To:     Brenden Blanco <bblanco@...mgrid.com>
Cc:     Tariq Toukan <tariqt@...lanox.com>,
        "David S. Miller" <davem@...emloft.net>,
        Linux Kernel Network Developers <netdev@...r.kernel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Alexei Starovoitov <alexei.starovoitov@...il.com>,
        Tariq Toukan <ttoukan.linux@...il.com>,
        Or Gerlitz <gerlitz.or@...il.com>
Subject: Re: [PATCH] net/mlx4_en: protect ring->xdp_prog with rcu_read_lock

On Fri, Sep 2, 2016 at 11:13 AM, Brenden Blanco <bblanco@...mgrid.com> wrote:
> On Mon, Aug 29, 2016 at 10:46:38AM -0700, Tom Herbert wrote:
> [...]
>> Brenden, tracking down how the structure is freed needed a few steps,
>> please make sure the RCU requirements are well documented. Also, I'm
> Really? It's just bpf_prog_put->call_rcu(__bpf_prog_put_rcu). I suppose
> what's missing is a general guideline for which functions new consumers
> of bpf should use, but I wouldn't trust myself to write such holistic
> documentation accurately (e.g. interacting with nmi probes and such).
>> still not a fan of using xchg to set the program, seems that a lock
>> could be used in that path.
> Where would such a lock go? Everything in mlx4/en_netdev.c relies on
> rtnl, which seems sufficient and obvious...adding some new field
> specific lock would be distracting and unneeded.

Just use the same mutex_lock that is already being taken in case when
priv->xdp_ring_num != xdp_ring_num. Then use normal rcu functions to
dereference and assign pointers.

Tom

>>
>> Thanks,
>> Tom