netdev - Re: [PATCH bpf 1/3] bpf, sockmap: Check for any of tcp_bpf

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] [day] [month] [year] [list]

Date:   Mon, 16 Jan 2023 14:53:09 +0300
From:   Dan Carpenter <error27@...il.com>
To:     Jakub Sitnicki <jakub@...udflare.com>
Cc:     oe-kbuild@...ts.linux.dev, netdev@...r.kernel.org, lkp@...el.com,
        oe-kbuild-all@...ts.linux.dev, kernel-team@...udflare.com,
        John Fastabend <john.fastabend@...il.com>,
        Eric Dumazet <edumazet@...gle.com>,
        Daniel Borkmann <daniel@...earbox.net>,
        Alexei Starovoitov <ast@...nel.org>,
        Andrii Nakryiko <andrii@...nel.org>,
        syzbot+04c21ed96d861dccc5cd@...kaller.appspotmail.com
Subject: Re: [PATCH bpf 1/3] bpf, sockmap: Check for any of tcp_bpf_prots
 when cloning a listener

On Mon, Jan 16, 2023 at 12:31:11PM +0100, Jakub Sitnicki wrote:
> >> Clang can do pointer arithmetic on 2D arrays just fine :-)
> >
> > Heh.  I must have an older version of Clang.
> >
> >   CC      net/ipv4/tcp_bpf.o
> > net/ipv4/tcp_bpf.c:644:41: warning: array index 2 is past the end of the array (that has type 'struct proto[2][4]') [-Warray-bounds]
> >         if (tcp_bpf_prots[0] <= prot && prot < tcp_bpf_prots[ARRAY_SIZE(tcp_bpf_prots)])
> >                                                ^             ~~~~~~~~~~~~~~~~~~~~~~~~~
> > net/ipv4/tcp_bpf.c:544:1: note: array 'tcp_bpf_prots' declared here
> > static struct proto tcp_bpf_prots[TCP_BPF_NUM_PROTS][TCP_BPF_NUM_CFGS];
> > ^
> > 1 warning generated.
> 
> FWIW, I've checked against:
> 
> $ clang --version
> clang version 15.0.6 (Fedora 15.0.6-2.fc37)
> 
> Gotta keep it fresh to be able to build bpf selftests ;-)
> But I sure don't want to break builds with older Clangs.

I'm actually on a newer 16.x something version from git.

Btw, it made me outrageously happy that Clang was one for one bug
compatible with Smatch on this.

With this kind of warning you could either print a warning when there is
a read but that's not what either Smatch or Clang do.  Smatch looks at
the offset and then checks to see if the code is just doing pointer
math to find the &(array + 1) address.

So Smatch checks is the offset known to be exactly ARRAY_SIZE() and are
we taking the address of that.  I have updated that check.

regards,
dan carpenter