Re: [PATCH] net: sched: Fix memory exposure from short TCA_U32_SEL

From: Cong Wang <xiyou.wangcong@gmail.com>
To: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>,
	Kees Cook <keescook@chromium.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Jiri Pirko <jiri@resnulli.us>, David Miller <davem@davemloft.net>,
	Linux Kernel Network Developers <netdev@vger.kernel.org>
Subject: Re: [PATCH] net: sched: Fix memory exposure from short TCA_U32_SEL
Date: Mon, 27 Aug 2018 14:31:41 -0700	[thread overview]
Message-ID: <CAM_iQpVEyq9hR3bbOtLFKoLo6nHCtiL6A__uEz3JdDO79GF_8A@mail.gmail.com> (raw)
In-Reply-To: <cb9d4e74-3498-48bd-45e0-05e925dbdb5b@mojatatu.com>

On Mon, Aug 27, 2018 at 4:58 AM Jamal Hadi Salim <jhs@mojatatu.com> wrote:
>
> On 2018-08-26 6:57 p.m., Al Viro wrote:
> > On Sun, Aug 26, 2018 at 06:32:37PM +0100, Al Viro wrote:
> >
> >> As far as I can tell, the solution is
> > [snip long and painful reasoning]
> >> pointers, and not in provably opaque fashion.  Theoretically, the three tcf_...
> >> inlines above need another look; fortunately, they don't use ->next at all, not to
> >> mention not being used anywhere outside of net/sched/*.c
> >>
> >>      The 80 lines above prove that we only need to grep net/sched/*.c for
> >> tcf_proto_ops method calls.  And only because we don't have (thank $DEITY)
> >> anything that could deconstruct types - as soon as some bastard grows means
> >> to say "type of the second argument of the function pointed to by p", this
> >> kind of analysis, painful as it is, goes out of window.  Even as it is,
> >> do you really like the idea of newbies trying to get through the exercises
> >> like the one above?
> >
> > BTW, would there be any problem if we took the definitions of tcf_proto and
> > tcf_proto_ops to e.g. net/sched/tcf_proto.h (along with the three inlines in
> > in pkt_cls.h), left forwards in sch_generic.h and added includes of "tcf_proto.h"
> > where needed in net/sched/*.c?
> >
>
> I cant think of any challenges. Cong/Jiri? Would it require development
> time classifiers/actions/qdiscs to sit in that directory (I suspect you
> dont want them in include/net).
> BTW, the idea of improving grep-ability of the code by prefixing the
> ops appropriately makes sense. i.e we should have ops->cls_init,
> ops->act_init etc.

Hmm? Isn't struct tcf_proto_ops used and must be provided
by each tc filter module? How does it work if you move it into
net/sched/* for out-of-tree modules? Are they supposed to
include "..../net/sched/tcf_proto.h"?? Or something else?

BTW, we need some grep tool that really understands C syntax,
not making each variable friendly to plain grep.