Re: [PATCH 1/9] Known exploit detection

[Sasha Levin]

Hi Vegard,

On 12/12/2013 11:52 AM, vegard.nossum@xxxxxxxxxx wrote:
> +#ifdef CONFIG_EXPLOIT_DETECTION
> +extern void _exploit(const char *id);

So right now the on/off switch is a kernel config option. I suggest we should add another
dynamic switch (maybe in the form of jump labels) to add an additional level of control:

 - It will allow having an opt-in option. Right now users are forced into
having this feature if the distro maintainers enable it.
	- Which means that distro maintainers are less likely to enable it.

 - If the SHTF and there's something wrong we would want a way to disable it
without having to re-compile the kernel.


<bikeshedding>
Also,

Maybe in the future we could enable/disable specific exploits based on severity or certainty
(how likely that this specific activity is an exploit attempt).

</bikeshedding>

On 12/12/2013 11:52 AM, vegard.nossum@xxxxxxxxxx wrote:
> +#define exploit_on(cond, id) \
> +	do { \
> +		if (unlikely(cond)) \
> +			_exploit(id); \
> +	} while (0)

What if we make exploit_on() something like this:

	#define exploit_on(cond, id) ({			\
		int __ret_exploit_on = !!(cond);	\
		if (unlikely(__ret_exploit_on))		\
			_exploit(id);			\
		unlikely(__ret_exploit_on);		\
	})

That way we can use it within if() conditionals similar to WARN_ON:

	if (exploit_on(srclen > HFS_NAMELEN, "CVE-2011-4330"))
		srclen = HFS_NAMELEN;



Thanks,
Sasha
--
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/