Re: v5 of seccomp filter c/r patches

[Kees Cook]

On Fri, Oct 2, 2015 at 3:57 PM, Daniel Borkmann <daniel@xxxxxxxxxxxxx> wrote:
> On 10/03/2015 12:44 AM, Tycho Andersen wrote:
>>
>> On Fri, Oct 02, 2015 at 02:10:24PM -0700, Kees Cook wrote:
>
> ...
>>
>> Ok, how about,
>>
>> struct sock_filter insns[BPF_MAXINSNS];
>> insn_cnt = ptrace(PTRACE_SECCOMP_GET_FILTER, pid, insns, i);
>
>
> Would also be good that when the storage buffer (insns) is NULL,
> it just returns you the number of sock_filter insns (or 0 when
> nothing attached).
>
> That would be consistent with classic socket filters (see
> sk_get_filter()), and user space could allocate a specific
> size instead of always passing in max insns.

Yes please. :)

>> when asking for the ith filter? It returns either the number of
>> instructions, -EINVAL if something was wrong (i, pid,
>> CONFIG_CHECKPOINT_RESTORE isn't enabled). While it would always
>> succeed now, if/when the underlying filter was not created from a bpf
>> classic filter, we can return -EMEDIUMTYPE? (Suggestions welcome, I
>> picked this mostly based on what sounds nice.)

We can bikeshed the non-classic case when we need it, but I think
EINVAL is "not under seccomp", and ENOENT is "no such index".

-Kees

-- 
Kees Cook
Chrome OS Security
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/