netdev - Re: [PATCH net-next 2/6] net: sched: Add centralized RED flag checking

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: <20200310160052.72e7e09b@kicinski-fedora-PC1C0HJN>
Date:   Tue, 10 Mar 2020 16:00:52 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Petr Machata <petrm@...lanox.com>
Cc:     Ido Schimmel <idosch@...sch.org>, netdev@...r.kernel.org,
        davem@...emloft.net, jiri@...lanox.com, jhs@...atatu.com,
        xiyou.wangcong@...il.com, mlxsw@...lanox.com,
        Ido Schimmel <idosch@...lanox.com>
Subject: Re: [PATCH net-next 2/6] net: sched: Add centralized RED flag
 checking

On Tue, 10 Mar 2020 23:23:23 +0100 Petr Machata wrote:
> Jakub Kicinski <kuba@...nel.org> writes:
> 
> > On Tue, 10 Mar 2020 10:48:24 +0100 Petr Machata wrote:  
> >> > The only flags which are validated today are the gRED per-vq ones, which
> >> > are a recent addition and were validated from day one.  
> >>
> >> Do you consider the validation as such to be a problem? Because that
> >> would mean that the qdiscs that have not validated flags this way
> >> basically cannot be extended ever ("a buggy userspace used to get a
> >> quiet slicing of flags, and now they mean something").  
> >
> > I just remember leaving it as is when I was working on GRED, because
> > of the potential breakage. The uAPI policy is what it is, then again
> > we probably lose more by making the code of these ancient Qdiscs ugly
> > than we win :(
> >
> > I don't feel like I can ack it with clear conscience tho.  
> 
> Just to make sure -- are you opposed to adding a new flag, or to
> validation? 

They are both uABI changes, so both.

> At least the adaptative flag was added years after the
> others in 2011. I wasn't paying much attention to kernel back then, but
> I think the ABI rules are older than that.

Yes, but some (e.g. TC subsystem) didn't really care much about those
rules until more recently.

The alternative to validation/adding flag in place is obviously to add 
a new netlink attribute which would be validated from the start. Can you
give it a try and see how ugly it gets?