Re: [PATCH 1/9] Known exploit detection

[Kees Cook]

On Fri, Dec 13, 2013 at 10:14 AM, Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> On Fri, Dec 13, 2013 at 9:58 AM, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>>
>> These locations tend to be very hard to reach accidentally
>
> Not necessarily.
>
> Don't get me wrong - I think that it's a good idea to at least have
> the option to complain about certain errors, and leave markers in the
> logs about things that look suspicious.
>
> But looking through the recent list of commits that explicitly mention
> a CVE, the only one I find where a syslog message would make sense is
> the HID validation ones. There, adding a warning about malicious HID
> devices sounds like a good idea.
>
> But a *lot* of the rest is just checking ranges or making sure we have
> proper string handling etc that just wouldn't be practical to check.
> So the error itself may be "hard to reach accidentally", but
> *checking* it would be so complex/painful that it would likely just
> introduce more room for bugs.
>
> So I think the "WARNING" thing is a good idea, but I think it is a
> good idea if it's used very judiciously. IOW, not for "random CVE"
> (because quite frankly, most of them seem to be utter shit), but for
> serious known issues. And for those issues *only*.
>
> If I start seeing patches adding warnings "just because there's a
> CVE", then I'm not in the least interested. But if there is some known
> root-kit or similar, then by all means..

Yeah, totally agreed. Doing it for all CVEs (or even most) would be a
disaster. Stuff like memory content leak CVEs are usually on common
paths that userspace uses all the time. Vegard proposed only doing it
for serious privilege escalation issues, and I couldn't agree more.

-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/