On Tue, 6 Oct 2020 at 10:06, Ard Biesheuvel ardb@kernel.org wrote:
On Tue, 6 Oct 2020 at 10:00, François Ozog francois.ozog@linaro.org wrote:
Le mar. 6 oct. 2020 à 09:21, Ard Biesheuvel ardb@kernel.org a écrit :
On Tue, 6 Oct 2020 at 06:35, Heinrich Schuchardt xypron.glpk@gmx.de wrote:
Am 6. Oktober 2020 00:37:58 MESZ schrieb Grant Likely <
grant.likely@arm.com>:
On 03/10/2020 09:51, Heinrich Schuchardt wrote:
Hello Ilias, hello Christian,
with commit ec80b4735a59 ("efi_loader: Implement FileLoad2 for
initramfs
loading") Ilias provided the possibility to specify a device path (CONFIG_EFI_INITRD_FILESPEC) from which an initial RAM disk can be served via the EFI_FILE_LOAD2_PROTOCOL.
Ard extended the Linux EFI stub to allow loading the initial RAM
disk
via the EFI_FILE_LOAD2_PROTOCOL with the utmost priority.
With commit ecc7fdaa9ef1 ("bootm: Add a bootm command for type IH_OS_EFI") Cristian enabled signed FIT images that contain a device tree and a UEFI binary (enabled by CONFIG_BOOTM_EFI=y).
In the DTE calls we have discussed that it is unfortunate that we do
not
have a method to validate initial RAM images in the UEFI context.
To me it would look like a good path forward to combine the two
ideas:
- Let the signed FIT image (of type IH_OS_EFI) contain a RAM disk
- Pass location and size to the UEFI subsystem and serve them via the EFI_FILE_LOAD2_PROTOCOL.
We could also extend the bootefi command to be callable as
bootefi $kernel_addr_r $ramdisk_addr_r:$filesize $fdt_addr_rlike the booti command to serve an initial RAM disk.
What are your thoughts?
Hi Heinrich,
I've got concerns about this approach. Even though it uses the UEFI infrastructure, images deployed in this way are U-Boot specific and won't ever be applicable on EDK2 or other UEFI implementations.
However there is another way to approach it which I think Francois touched on. If instead a UEFI stub was added to the FIT image, in the same way that the kernel has a UEFI stub, then the logic of decoding the FIT and choosing the correct DTB & initrd can be part of the image and it becomes applicable to any UEFI implementation. It would also
address
Ard's concern of loading the FIT into memory, and then copying due to the EFI_FILE_LOAD2 path. The FIT stub would already know the image is in RAM, that is is reserved correctly, and just pass the correct
addresses
to the kernel as part of the normal boot flow.
Signing would also be taken care of because the whole FIT can be signed, and that signature would be checked when it gets loaded.
Thoughts?
The gain of a fit image in U-Boot used for calling the Linux kernel
via the EFI stub vs calling the legacy entry point comes down to providing the EFI_RNG_PROTOCOL to be used for KASLR.
For initrd a stub UEFI binary will work. But if you want to provide a
kernel specific dtb with the same stub binary it will require a new service for device-tree fixups.
Both approaches are on Francois' DTE slidedeck.
When thinking of security what really is unclear to me is how we can
safely provide the decryption key for the root partition. Without such a means secure boot stops being secure once Linux starts the init process. I would assume that only the secure world can provide the key.
Secure boot only deals with authentication, which does not require any secret keys.
Encrypted file systems are a completely separate matter. Typically, this is based on TPM-based local attestation, where the decryption key has been sealed into the TPM, and is only unsealed when all the boot components check out (based on their 'measurements' [aka hashes] that are cumulatively recorded in TPM hardware registers called PCRs)
In general, keeping secrets on a device is much more difficult than authenticating executable images, and typically involves some user provided secret, or h/w support. UEFI secure boot only reasons about authentication.
If SecureBoot is involving Microsoft certificate, a US governmental agency could get its code signed and insert itself in a stealth mode (efi apps or drivers -> rootkit?)
If the set of keys must include this certificate , then to achieve SecureBoot you actually need the TPM to control what signed code you accept.
I am sorry, but this makes no sense at all. Either you trust the MS certificate or you don't -
I trust the certificate technology but I may not trust the signature process at Microsoft: as a consumer I am fine with it, if I were a French nuclear plant security officer, I wouldn't trust the MS certification process (because NSA can get Microsoft to sign a piece of code I don't want). So trustworthiness or trustworthiness target levels should be defined (an effort is below and was presented at DTE call)
if you don't, then you cannot rely on the TPM either, as you are relying on the boot chain not to lie about the measurements it sends to the TPM. IOW, you can trick the TPM into releasing its secrets by sending the 'correct' values that it expects, rather than measurements that were taken from the actual boot stages.
Indeed. And I think there are countermeasures. In the same spirit, there are hardware attacks: Riscure proved SecureBoot can be defeated by injecting 12V on a USB port of a platform: the pseudo code "if <signature_verified> then <continue_secure_boot> else <stop_boot_with_alert>" went on <continue_secure_boot> while signature was bad.
From the above, and based on Arm defined certification levels, I
postulate that there are three classes of "trustworthiness target levels": PSA level 1: limited security requirements, self certification PSA level 2: software attacks (local or remote) resilience, lab certified PSA level 3: hardware attacks resilience
In this context, it may be defined that: - a PSA level 2 with basic trustworthiness target level is OK with trusting the Msft process of certification (basically that is all datacenter, home computing world) - a PSA level 3 with high end trustworthiness target level is not OK with Msft certificate, will require TPM with mitigation of evasion techniques you mention, provide counter measures (hardware or software) to ensure guaranteed behavior in key decisions such as "if <signature_verified>", have non repudiable auditability of events, and a continuous assessment of measured boot in RAM (this was proposed by OpenFog group - probably by Arm BTW - while not digging in how this could actually be done).
At the end of the day, trustworthiness target level is about matching the use case with costs and complexity of management.
There will be thus many trustworthiness target levels. we should define them and their implications on technologies, processes (OTA...) and software design patterns (mitigation of hardware attacks when hardware mitigation is not available). And it may be possible that the high end trustworthiness levels will never be open source. In any case we shall define them and state what can be expected from the technologies we build so that product vendors can characterize the trustworthiness level of their products.
If the certificate can be omitted I would either sign the shim or add the
sha256 of trustable ones.
The unsealing if keys for rootfs decryption is yet another capability that is offered by TPM to protect against different nature of attacks (conuterfeit products, confidentiality...)
--
François-Frédéric Ozog | *Director Linaro Edge & Fog Computing Group* T: +33.67221.6485 francois.ozog@linaro.org | Skype: ffozog