Re: [PATCH v3] fs: introduce pipe-only dump mode suid_dumpable=3

[Kees Cook]

On Fri, Jun 22, 2012 at 2:34 PM, Andrew Morton
<akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> On Fri, 22 Jun 2012 14:09:28 -0700
> Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>
>> On Fri, Jun 22, 2012 at 12:55 PM, Andrew Morton
>> <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
>> > On Fri, 22 Jun 2012 12:24:13 -0700
>> > Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>> >
>> >> The value
>> >> of suid_dumpable=2 is now historic, and attempting to set this sysctl
>> >> value returns -EINVAL.
>> >
>> > This sounds a bit harsh - will it not cause existing configurations to
>> > immediately break? __If so, would it not be better to retain the =2 mode
>> > for a while, and emit a nice warning when it is set?
>>
>> I view it as a security vulnerability, so I'd rather see it
>> eliminated. I see "=1" as a security vulnerability too, but at least
>> that's well-known to be a bad idea. The "=2" mode has been assumed to
>> be safe, but it isn't.
>
> But what will be the effects of the change?  People's initscripts do an
> "echo 2" which fails and the error message (if any) won't get logged
> anywhere where anyone looks.  So now their machine is bumbling along in
> the wrong mode and much later on, someone notices that coredumps are
> going awry?  This is not exactly a user-friendly way of rolling out
> kernel API changes!

Well, this is why I wanted to just change the meaning of "2" instead
of introducing "3". It seems much cleaner to me. Just stop "2" from
doing the dangerous thing and carry on.

> And how serious is the security vulnerability, in real-world terms?
> Serious enough to risk this amount of bustage?

If they're running in mode "2" and they do not have a coredump pipe
handler defined, local users can gain root access.

-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/