netdev - Re: [PATCH net] tipc: not enable tipc when ipv6 works as a module

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: <CAM_iQpUEjZzW-e=h30KZVvg02ZZMRHZn9JExxgn6E=XyWsjzNQ@mail.gmail.com>
Date:   Mon, 17 Aug 2020 13:59:46 -0700
From:   Cong Wang <xiyou.wangcong@...il.com>
To:     Randy Dunlap <rdunlap@...radead.org>
Cc:     Xin Long <lucien.xin@...il.com>,
        network dev <netdev@...r.kernel.org>,
        David Miller <davem@...emloft.net>,
        Jon Maloy <jmaloy@...hat.com>,
        Ying Xue <ying.xue@...driver.com>,
        tipc-discussion@...ts.sourceforge.net
Subject: Re: [PATCH net] tipc: not enable tipc when ipv6 works as a module

On Mon, Aug 17, 2020 at 1:43 PM Randy Dunlap <rdunlap@...radead.org> wrote:
>
> On 8/17/20 1:29 PM, Cong Wang wrote:
> > On Mon, Aug 17, 2020 at 12:55 PM Randy Dunlap <rdunlap@...radead.org> wrote:
> >>
> >> TIPC=m and IPV6=m builds just fine.
> >>
> >> Having tipc autoload ipv6 is a different problem. (IMO)
> >>
> >>
> >> This Kconfig entry:
> >>  menuconfig TIPC
> >>         tristate "The TIPC Protocol"
> >>         depends on INET
> >> +       depends on IPV6 || IPV6=n
> >>
> >> says:
> >> If IPV6=n, TIPC can be y/m/n.
> >> If IPV6=y/m, TIPC is limited to whatever IPV6 is set to.
> >
> > Hmm, nowadays we _do_ have IPV6=y on popular distros.
> > So this means TIPC would have to be builtin after this patch??
>
> No, it does not mean that. We can still have IPV6=y and TIPC=m.
>
> Hm, maybe I should have said this instead:
>
>   If IPV6=y/m, TIPC is limited _by_ whatever IPV6 is set to.
>                 (instead of    _to_ )
>
> Does that help any?
>
> The "limited" in Kconfig rules is a "less than or equal to"
> limit, where 'm' < 'y'.

Yeah, this is more clear now. If so, that means all the symbols
we have in ipv6_stub can be gone now, assuming module
dependency is automatically solved when both are modules.

Is this a new Kconfig feature? ipv6_stub was introduced for
VXLAN, at that time I don't remember we have such kind of
Kconfig rules, otherwise it would not be needed.

I also glanced at Documentation/kbuild/kconfig-language.txt,
I do not see such a rule, I guess the doc is not updated.


> > Still sounds harsh, right?
> >
> > At least on my OpenSUSE I have CONFIG_IPV6=y and
> > CONFIG_TIPC=m.
> >
> >> TIPC cannot be =y unless IPV6=y.
> >
> > Interesting, I never correctly understand that "depends on"
> > behavior.
> >
> > But even if it builds, how could TIPC module find and load
> > IPV6 module? Does IPV6 module automatically become its
> > dependency now I think?
>
> Sorry, I don't know about this.

You can check `modinfo tipc` output after compiling both as
modules. (Sorry that I only have CONFIG_MODULES=n here.)

Thanks.