This repo is for review of requests for signing shim. To create a request for review:
- clone this repo
- edit the template below
- add the shim.efi to be signed
- add build logs
- add any additional binaries/certificates/SHA256 hashes that may be needed
- commit all of that
- tag it with a tag of the form "myorg-shim-arch-YYYYMMDD"
- push that to github
- file an issue at https://github.com/rhboot/shim-review/issues with a link to your tag
- approval is ready when the "accepted" label is added to your issue
Note that we really only have experience with using GRUB2 on Linux, so asking us to endorse anything else for signing is going to require some convincing on your part.
Here's the template:
Canonical Ltd.
Ubuntu
What's the justification that this really does need to be signed for the whole world to be able to boot it?
We're a well-known Linux distro
We are big distro with ton of custom grub patches.
The security contacts need to be verified before the shim can be accepted. For subsequent requests, contact verification is only necessary if the security contacts or their PGP keys have changed since the last successful verification.
An authorized reviewer will initiate contact verification by sending each security contact a PGP-encrypted email containing random words.
You will be asked to post the contents of these mails in your shim-review
issue to prove ownership of the email addresses and PGP keys.
- Name: Julian Andres Klode
- Position: engineer
- Email address: [email protected]
- PGP key fingerprint: AEE1 C8AA AAF0 B768 4019 C546 021B 361B 6B03 1B00
(Key should be signed by the other security contacts, pushed to a keyserver like keyserver.ubuntu.com, and preferably have signatures that are reasonably well known in the Linux community.)
- Name: dann frazier
- Position: engineer
- Email address: [email protected]
- PGP key: dannf.pub
- PGP key fingerprint: 09F4 7DBF 2D32 EEDC 2443 EBEE 1BF8 3C5E 54FC 8640
(Key should be signed by the other security contacts, pushed to a keyserver like keyserver.ubuntu.com, and preferably have signatures that are reasonably well known in the Linux community.)
Please create your shim binaries starting with the 15.7 shim release tar file: https://github.com/rhboot/shim/releases/download/15.7/shim-15.7.tar.bz2
This matches https://github.com/rhboot/shim/releases/tag/15.7 and contains the appropriate gnu-efi source.
The shim-15.7.tar.bz2 is used as the original tarball.
https://code.launchpad.net/~ubuntu-core-dev/shim/+git/shim/+ref/master
Patches included also previous submission:
- debian/patches/ubuntu-no-addend-vendor-dbx.patch: Stop addending the vendor dbx to the MokListX, ours is too large. Our kernels don't read it anyway, and new ones that will can just embed it themselves.
No new patches.
If shim is loading GRUB2 bootloader what exact implementation of Secureboot in GRUB2 do you have? (Either Upstream GRUB2 shim_lock verifier or Downstream RHEL/Fedora/Debian/Canonical-like implementation)
2.06 with lockdown backports, shim_lock, with rhboot/linuxefi/Canonical like implementation.
If shim is loading GRUB2 bootloader and your previously released shim booted a version of grub affected by any of the CVEs in the July 2020 grub2 CVE list, the March 2021 grub2 CVE list, the June 7th 2022 grub2 CVE list, or the November 15th 2022 list, have fixes for all these CVEs been applied?
-
CVE-2020-14372
-
CVE-2020-25632
-
CVE-2020-25647
-
CVE-2020-27749
-
CVE-2020-27779
-
CVE-2021-20225
-
CVE-2021-20233
-
CVE-2020-10713
-
CVE-2020-14308
-
CVE-2020-14309
-
CVE-2020-14310
-
CVE-2020-14311
-
CVE-2020-15705
-
CVE-2021-3418 (if you are shipping the shim_lock module)
-
CVE-2021-3695
-
CVE-2021-3696
-
CVE-2021-3697
-
CVE-2022-28733
-
CVE-2022-28734
-
CVE-2022-28735
-
CVE-2022-28736
-
CVE-2022-28737
-
CVE-2022-2601
-
CVE-2022-3775
Yes.
Yes.
Ubuntu shim uses a self-managed CA certificate as the VENDOR_CERT. It remains unchanged.
This version revokes all previously used certificates, so is a clean slate that doesn't trust any existing binary.
Pre-SBAT shim was revoked in dbx update
Is upstream commit 1957a85b0032a81e6482ca4aab883643b8dae06e "efi: Restrict efivar_ssdt_load when the kernel is locked down" applied?
Is upstream commit 75b0cea7bf307f362057cc778efe89af4c615354 "ACPI: configfs: Disallow loading ACPI tables when locked down" applied?
Is upstream commit eadb2f47a3ced5c64b23b90fd2a3463f63726066 "lockdown: also lock down previous kgdb use" applied?
All Ubuntu kernels in all currently supported series have the above applied.
All vulnerable kernels are disallowed to boot by VENDOR_DBX by their signing cert being revoked in vendor dbx.
Yes but there are like hundred patches and like 80 different kernels, so it's a bit much to include here. There's additional secure boot enforcing patches, hardware enablement, and zfs is built alongside.
Most interesting things are:
df8b92624f UBUNTU: SAUCE: (lockdown) security: lockdown: expose a hook to lock the kernel down
fede732054 UBUNTU: SAUCE: (lockdown) efi: Add an EFI_SECURE_BOOT flag to indicate secure boot mode
438296a598 UBUNTU: SAUCE: (lockdown) efi: Lock down the kernel if booted in secure boot mode
03deb74301 UBUNTU: SAUCE: (lockdown) s390: Lock down the kernel when the IPL secure flag is set
c2952ca438 UBUNTU: SAUCE: (lockdown) KEYS: Make use of platform keyring for module signature verify
9ba951d4e7 UBUNTU: SAUCE: (lockdown) arm64: Allow locking down the kernel under EFI secure boot
01f96e4abc UBUNTU: SAUCE: (lockdown) security: lockdown: Make CONFIG_LOCK_DOWN_IN_EFI_SECURE_BOOT more generic
59a69f2418 UBUNTU: SAUCE: (lockdown) powerpc: lock down kernel in secure boot mode
0db545033f UBUNTU: SAUCE: integrity: Load mokx certs from the EFI MOK config table
7482fcc79c UBUNTU: SAUCE: integrity: add informational messages when revoking certs
9075b83ae9 UBUNTU: [Packaging] Revoke 2012 UEFI signing certificate as built-in
The above ensure that lockdown is enforced when booted with secureboot, MOKX keys are imported into kernel .blacklist keyring, and thus revoked kernels are prohibited from kexec/kdump.
Notable features of our config options:
CONFIG_MODULE_SIG=y
CONFIG_MODULE_SIG_ALL=y
# CONFIG_MODULE_SIG_FORCE is not set
CONFIG_MODULE_SIG_FORMAT=y
CONFIG_MODULE_SIG_HASH="sha512"
CONFIG_MODULE_SIG_KEY="certs/signing_key.pem"
CONFIG_MODULE_SIG_KEY_TYPE_RSA=y
CONFIG_MODULE_SIG_SHA512=y
CONFIG_SYSTEM_REVOCATION_KEYS="debian/canonical-revoked-certs.pem"
CONFIG_SYSTEM_TRUSTED_KEYS="debian/canonical-certs.pem"
The above settings ensure that all drivers are signed with built-time ephemeral signing key. In addition, we trust livepatch & 3rd-party driver signing key for signing modules post kernel build.
Drivers signed with built-in kernel signing key:
-
CONFIG_STAGING=y
that are listed in./drivers/staging/signature-inclusion
, currently exfat, realtek wifi drivers only. NB! most importantly android ashmem/binder are not signed -
Vendored at build-time dkms modules listed in
debian/dkms-versions
, currently these arezfs-linux
,v4l2loopback
,backport-iwlwifi-dkms
for ZFS, webcam and wifi support.
Drivers signed with SYSTEM_TRUSTED_KEYS
:
-
Canonical Livepatch Service modules for livepatching security vulnerabilities
-
Detached reproducible builds NVIDIA proprietary driver signatures
Certificates present in CONFIG_SYSTEM_REVOCATION_KEYS
:
- The certificates in
CONFIG_SYSTEM_REVOCATION_KEYS
are the same as shim'sVENDOR_DBX
discussed below. This is to ensure that kernel prohibits kexec/kdump of kernels that are distrusted by the shim to boot. This works, even if MOKX mirroring facility fails at runtime, due to shim/platform deficiencies.
If you use vendor_db functionality of providing multiple certificates and/or hashes please briefly describe your certificate setup.
If there are allow-listed hashes please provide exact binaries for which hashes are created via file sharing service, available in public with anonymous access for verification.
VENDOR_DB is not used.
If you are re-using a previously used (CA) certificate, you will need to add the hashes of the previous GRUB2 binaries exposed to the CVEs to vendor_dbx in shim in order to prevent GRUB2 from being able to chainload those older GRUB2 binaries. If you are changing to a new (CA) certificate, this does not apply.
We are shipping vendor_dbx that includes all previously used certificates.
What OS and toolchain must we use to reproduce this build? Include where to find it, etc. We're going to try to reproduce your build as closely as possible to verify that it's really a build of the source tree you tell us it is, so these need to be fairly thorough. At the very least include the specific versions of gcc, binutils, and gnu-efi which were used, and where to find those binaries.
If the shim binaries can't be reproduced using the provided Dockerfile, please explain why that's the case and what the differences would be.
Ubuntu 22.10 (kinetic kudo):
binutils (= 2.39-3ubuntu1),
binutils-aarch64-linux-gnu (= 2.39-3ubuntu1),
binutils-common (= 2.39-3ubuntu1),
binutils-x86-64-linux-gnu (= 2.39-3ubuntu1),
gcc-12 (= 12.2.0-3ubuntu1),
gcc-12-base (= 12.2.0-3ubuntu1),
gcc (= 4:12.2.0-1ubuntu1),
libc6-dev (= 2.36-0ubuntu4),
To build:
Use included Dockerfiles or just check the GitHub workflow which does it for you.
This should include logs for creating the buildroots, applying patches, doing the build, creating the archives, etc.
The .log files
Rebased against 15.7
$ sha256sum shim*.efi d02697ef3c6a2a4980e3b2ceedce0fd6591ff10e163fe6f7678553723a366254 shimaa64.efi ab1b96c04a3898253c5f8b381cc21e0d03c5e5f674fd102070afefb8469ed8fc shimx64.efi
The CA certificate used as VENDOR_CERT is always stored offline, split using Shamir's Secret Sharing into 7 fragments distributed globally, 3 of which are required to assemble the cert.
Thus we require international travel to be available to assemble it and issue new certificates.
No
Do you add a vendor-specific SBAT entry to the SBAT section in each binary that supports SBAT metadata ( grub2, fwupd, fwupdate, shim + all child shim binaries )?
Please provide exact SBAT entries for all SBAT binaries you are booting or planning to boot directly through shim.
Where your code is only slightly modified from an upstream vendor's, please also preserve their SBAT entries to simplify revocation.
shim, fb, mm:
sbat,1,SBAT Version,sbat,1,https://github.com/rhboot/shim/blob/main/SBAT.md
shim,3,UEFI shim,shim,1,https://github.com/rhboot/shim
shim.ubuntu,1,Ubuntu,shim,15.7-0ubuntu1,https://www.ubuntu.com/
grub: (template)
sbat,1,SBAT Version,sbat,1,https://github.com/rhboot/shim/blob/main/SBAT.md
grub,3,Free Software Foundation,grub,@UPSTREAM_VERSION@,https://www.gnu.org/software/grub/
grub.ubuntu,1,Ubuntu,grub2,@DEB_VERSION@,https://www.ubuntu.com/
fwupd:
sbat,1,UEFI shim,sbat,1,https://github.com/rhboot/shim/blob/main/SBAT.md
fwupd,1,Firmware update daemon,fwupd,1.5.11,https://github.com/fwupd/fwupd
fwupd.ubuntu,1,Ubuntu,fwupd,1.5.11-0ubuntu2,https://launchpad.net/ubuntu/+source/fwupd
kernel.efi:
sbat,1,SBAT Version,sbat,1,https://github.com/rhboot/shim/blob/main/SBAT.md
systemd,1,The systemd Developers,systemd,245,https://www.freedesktop.org/wiki/Software/systemd
systemd.ubuntu,1,Ubuntu,systemd,245.4-4ubuntu3.6,https://bugs.launchpad.net/ubuntu/
basic: all_video boot btrfs cat chain configfile echo efifwsetup efinet ext2 fat font gettext gfxmenu gfxterm gfxterm_background gzio halt help hfsplus iso9660 jpeg keystatus loadenv loopback linux ls lsefi lsefimmap lsefisystab lssal memdisk minicmd normal ntfs part_apple part_msdos part_gpt password_pbkdf2 png probe reboot regexp search search_fs_uuid search_fs_file search_label sleep smbios squash4 test true video xfs zfs zfscrypt zfsinfo "
amd64-only: cpuid linuxefi play tpm
installed grub: cryptodisk gcry_arcfour gcry_blowfish gcry_camellia gcry_cast5 gcry_crc gcry_des gcry_dsa gcry_idea gcry_md4 gcry_md5 gcry_rfc2268 gcry_rijndael gcry_rmd160 gcry_rsa gcry_seed gcry_serpent gcry_sha1 gcry_sha256 gcry_sha512 gcry_tiger gcry_twofish gcry_whirlpool luks lvm mdraid09 mdraid1x raid5rec raid6rec
network grub image: http tftp
Building / Publishing https://launchpad.net/ubuntu/+source/grub2-unsigned - same signed grub binaries for all series
currently building next one (first one signed with it in):
https://launchpad.net/~ubuntu-uefi-team/+archive/ubuntu/ppa/+packages
Git managed source code https://code.launchpad.net/~ubuntu-core-dev/grub/+git/ubuntu/+ref/ubuntu
Note patches debian/patches
We load various UKIs which use systemd-boot stub to combine kernels and initrds into a single binary.
fwupd of course.
If your GRUB2 launches any other binaries that are not the Linux kernel in SecureBoot mode, please provide further details on what is launched and how it enforces Secureboot lockdown.
GRUB2 may launch Windows Bootmgr on dual boot systems. Nebooted shim+grub2 may chainloader load shim+grub2 again from disk, which will verify things again as usual. (https://maas.io usecase).
fwupd verifies capsule signatures; kernel implements lockdown.
Our kernels also check MokListXRT for revocations for kexec.
No, our grub enforces lockdown & uses shim protocol (rhboot linuxefi sb patches) to verify next component.
linux, various versions. They include lockdown patches & ACPI patches, lockdown is enforced when booted with SecureBoot, config enforces kernel module signatures under lockdown.
VENDOR_DBX file is included as canonical-dbx-20221103.esl
One can unpack them using sig-list-to-certs
utility, and
finds as the changelog states:
This vendor dbx revokes all certificates that have been used
so far.
- CN = Canonical Ltd. Secure Boot Signing
- CN = Canonical Ltd. Secure Boot Signing (2017)
- CN = Canonical Ltd. Secure Boot Signing (ESM 2018)
- CN = Canonical Ltd. Secure Boot Signing (2019)
- CN = Canonical Ltd. Secure Boot Signing (Ubuntu Core 2019)
- CN = Canonical Ltd. Secure Boot Signing (2021 v1)
- CN = Canonical Ltd. Secure Boot Signing (2021 v2)
- CN = Canonical Ltd. Secure Boot Signing (2021 v3)
-
we have disabled ExitBootServices check, to allow chainloading a second shim from disk, from netbooted shim+grub. All shims these days require signature validation thus this is safe to do. We need this to support secureboot in https://maas.io which by default netboots & recovers bare metal machines.
-
we have disabled the unacceptable 5s boot delay in fallback when TPM is present, as it impacts bootspeed for the noninteractive cloud instances that have vTPM & SecureBoot.