Re: [RFD] linux-firmware key arrangement for firmware signing

[Andy Lutomirski]

On Tue, May 19, 2015 at 5:39 PM, Luis R. Rodriguez <mcgrof@xxxxxxxx> wrote:
> On Tue, May 19, 2015 at 04:42:05PM -0700, Andy Lutomirski wrote:
>> On Tue, May 19, 2015 at 4:30 PM, Julian Calaby <julian.calaby@xxxxxxxxx> wrote:
>> > Hi All,
>> >
>> > On Wed, May 20, 2015 at 6:59 AM, Andy Lutomirski <luto@xxxxxxxxxx> wrote:
>> >> [added cc's from the other thread]
>> >>
>> >> On 05/19/2015 01:02 PM, Luis R. Rodriguez wrote:
>> >>>
>> >>> David Howells has posted v4 of his series of supporting PKCS#7 for module
>> >>> signing. I'm in my v3 series now on RFCs for firmware PKCS#7 support, and
>> >>> after
>> >>> some review and patch shuffling I think this is ready for patch form.  My
>> >>> own
>> >>> series however depend on quite a bit of other pending changes, one series
>> >>> which
>> >>> will go through Rusty's tree, another series of fixes on firmware_class
>> >>> which
>> >>> should go through Greg's tree. I'll wait until all this and David's own
>> >>> patches
>> >>> get merged before posting firmware PKCS#7 support. Before all this though
>> >>> in
>> >>> preparation for fw signing one thing we should start to talk about more
>> >>> broadly
>> >>> however is how linux-firmware binary file signing would work in practice
>> >>> and
>> >>> what we need, and make sure folks are OK with all this.
>> >>>
>> >>> First, firmware signing will be completely optional as with module
>> >>> signing.
>> >>>
>> >>
>> >> ...
>> >>
>> >>> Other than this last nitpick, any other concerns or recommendations ?
>> >>
>> >>
>> >> A couple.  Some of these are general concerns with the existing
>> >> infrastructure, but #1 is a specific problem that gets much worse if we add
>> >> firmware signing.  Feel free to ignore 2-4.
>> >>
>> >> 1. We should get the signature semantics right.  I think that, for modules,
>> >> we currently sign literally the module payload.  For modules, in my
>> >> semi-amateurish crypto universe [1], this is fine *as long as the key in
>> >> question is used for no other purpose*.  For firmware, it's dangerous, since
>> >> it would be vulnerable to substitution attacks in which the adversary
>> >> convinces us to interpret one firmware file as firmware for another device
>> >> or purpose entirely.
>> >>
>> >> We should be signing something that's semantically equivalent to "This is a
>> >> valid module: xyz", "This is a valid 'regulatory.bin': xyz", or "This is a
>> >> valid kexec image: xyz".
>> >
>> > Something that occurred to me (as a complete bystander) was: would it
>> > make sense to have keys able to be restricted to particular "types" of
>> > signable data? I.e. the key that can sign a valid regulatory.bin file
>> > cannot be used to sign a module or a kexec image. - This could remove
>> > the need to have multiple keyrings. (Also, UEFI keys unless otherwise
>> > tagged could be restricted to only signing bootloaders or kernels)
>>
>> Seems sensible to me.
>
> As for having keys for fw signing be specific to fw data without a keyring,
> if that is desirable I think we can devise a way to do that. For instance
> if we wanted to we can have FW_SIG by default trust first keys on
> system_trusted_keyring just as module signature works -- or if we wanted to
> just trust, say a Kyle key. Not sure if the later is possible yet, but htat
> would require some changes. Then as an evolution if we wanted to enable a
> specific request fw to be mapped to a specific fw file the new APIs I was
> looking to add could easily enable this provided that we first decide we
> do want to trust say one key perhaps not on system_trusted_keyring for fw
> signing. That'd need to be decided first.
>
> As for the UEFI stuff -- from what I gather its too late there. We could
> certainly go with something else for fw signing though, just lemme hear it
> hard and clear.
>
>> FWIW, I'm starting to think that UEFI-based validation of kexec images
>> should be totally separate.  It uses a nasty PE format with a hideous
>> PKCS #7 formatted signature.  Maybe that should be a completely
>> separate piece of code.
>
> LSM'ify it I guess? Again, if that's reasonable then I think we'll need
> stacking and that's still not merged.

Isn't stacking backwards for this, though?  The semantics we'd want is
accept if any verifiers accept, not accept if all verifiers accept,
right?

--Andy
--
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/