linux-kernel - Re: [PATCH v4 net-next 5/5] net: filter: split 'struct sk

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

Message-ID: <CAMEtUuy+iABzHagoono+CPB=Ex4Ew4vQBK65x1jDRC9_W9L7Dw@mail.gmail.com>
Date:	Thu, 31 Jul 2014 14:02:19 -0700
From:	Alexei Starovoitov <ast@...mgrid.com>
To:	Pablo Neira Ayuso <pablo@...filter.org>
Cc:	"David S. Miller" <davem@...emloft.net>,
	Daniel Borkmann <dborkman@...hat.com>,
	Willem de Bruijn <willemb@...gle.com>,
	Kees Cook <keescook@...omium.org>,
	Network Development <netdev@...r.kernel.org>,
	LKML <linux-kernel@...r.kernel.org>,
	netfilter-devel <netfilter-devel@...r.kernel.org>
Subject: Re: [PATCH v4 net-next 5/5] net: filter: split 'struct sk_filter'
 into socket and bpf parts

On Thu, Jul 31, 2014 at 12:40 PM, Pablo Neira Ayuso <pablo@...filter.org> wrote:
> On Wed, Jul 30, 2014 at 08:34:16PM -0700, Alexei Starovoitov wrote:
>> clean up names related to socket filtering and bpf in the following way:
>> - everything that deals with sockets keeps 'sk_*' prefix
>> - everything that is pure BPF is changed to 'bpf_*' prefix
>>
>> split 'struct sk_filter' into
>> struct sk_filter {
>>       atomic_t        refcnt;
>>       struct rcu_head rcu;
>>       struct bpf_prog *prog;
>> };
>
> I think you can use 'struct bpf_prog prog' instead so the entire
> sk_filter remains in the same memory blob (as it is before this
> patch).
>
> You can add an inline function to retrieve the bpg prog from the
> filter:
>
> static inline struct bpf_prog *sk_filter_bpf(struct sk_filter *)
>
> and use it whenever possible to fetch the bpf program. I'm suggesting
> this because we can use the zero array size in the socket filtering
> abstraction later on, if the function above is used, this just needs
> one line in that function to be updated to fetch the program from the
> placeholder.

correct. It would speed up SK_RUN_FILTER macro a little and I've
considered it, but decided to go with the pointer for two reasons:
1.In sk_attach_filter() the bpf_prog is allocated, then reallocated
as part of bpf_prepare_filter(). My patch #1 cleans up that part to
avoid 'struct sock *' dependency, so all bpf_* functions work
purely with bpf_prog... If bpf_prog is embedded inside sk_filter,
bpf_prepare_filter would need to have a callback to reallocate
the container struct and pass this callback through the chain
of calls, which is ugly.
2. having it as a pointer helps nft in the long run, since whole of
bpf_prog doesn't stay around inside sk_filter when it's not used.
So I think embedding pointer was a cleaner solution.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/