linux-hardening - Re: [PATCH][next] net: ethernet: rmnet: Replace zero-length array with DECLARE_FLEX

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 for Android: free password hash cracker in your pocket

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

Date:   Tue, 27 Sep 2022 07:07:02 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Kees Cook <keescook@...omium.org>
Cc:     "Gustavo A. R. Silva" <gustavoars@...nel.org>,
        Subash Abhinov Kasiviswanathan <quic_subashab@...cinc.com>,
        Sean Tranchetti <quic_stranche@...cinc.com>,
        "David S. Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>,
        Paolo Abeni <pabeni@...hat.com>, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-hardening@...r.kernel.org,
        Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [PATCH][next] net: ethernet: rmnet: Replace zero-length array
 with DECLARE_FLEX_ARRAY() helper

On Mon, 26 Sep 2022 19:22:30 -0700 Kees Cook wrote:
> > Not directly related to this patch, but I just had to look at pahole
> > output for sk_buff and the struct_group() stuff makes is really painful
> > to read :/ Offsets for the members are relative to the "group" and they
> > are all repeated.
> > 
> > Is there any chance you could fix that? Before we sprinkle more pixie
> > dust around, perhaps?  
> 
> Unfortunately I don't see a way around it until we can make changes to
> the C language spec, and that's measured in decades. :(

I think BPF folks have had some success adding C extensions, like
tagging. Some form of attribute would really look so much better than
this DECLARE_FLEX_ARRAY() macro indirection. Maybe it's just me :(

> Perhaps we could add some kind of heuristic to pahole to "hide" one of
> the internal struct_group() copies, and to hide the empty flexible-array
> wrapper structs? (pahole already can't tell the difference between a
> 0-length array and a flexible-array.) Would that be workable?

That'd be my knee-jerk fix, too. Or at least render the offsets for 
the anonymous side of the union as absolute rather than relative.