Granted the user should really be using the Librem Key/phone to check
for tampering (since an attacker could control the Heads background
color) but this provides another visual queue for the user with
the GUI menu to catch less sophisticated tampering.
The HOTP counter isn't a secret but is just used to prevent replay
attacks (the time-based counter in TOTP isn't a secret either) so it
doesn't need to be protected in the TPM and storing it as a TPM
monotonic counter was causing conflicts with the Heads configuration
counter as TPM 1.2 can only increment one counter per reboot.
This change moves the HOTP counter into the file in /boot that was
previously keeping track of the TPM counter id.
TPM v1.2 has a limitation in that only a single monotonic counter can be
incremented between reboots [1]. So in the event we are using HOTP
monotonic counters, we need to reference those for the Heads rollback
counter when we update file signatures in /boot, otherwise the increment
stage at kexec-sign-config will fail since at each boot, the HOTP
monotonic counter has already been incremented.
[1] https://projects.csail.mit.edu/tc/tpmj/UsersGuide.html#inccounter
The Librem Key is a custom device USB-based security token Nitrokey is
producing for Purism and among other things it has custom firmware
created for use with Heads. In particular, when a board is configured
with CONFIG_LIBREMKEY, this custom firmware allows Heads to use the
sealed TOTP secret to also send an HOTP authentication to the Librem
Key. If the HOTP code is successful, the Librem Key will blink a green
LED, if unsuccessful it will blink red, thereby informing the user that
Heads has been tampered with without requiring them to use a phone to
validate the TOTP secret.
Heads will still use and show the TOTP secret, in case the user wants to
validate both codes (in case the Librem Key was lost or is no longer
trusted). It will also show the result of the HOTP verification (but not
the code itself), even though the user should trust only what the Librem
Key displays, so the user can confirm that both the device and Heads are
in sync. If HOTP is enabled, Heads will maintain a new TPM counter
separate from the Heads TPM counter that will increment each time HOTP
codes are checked.
This change also modifies the routines that update TOTP so that if
the Librem Key executables are present it will also update HOTP codes
and synchronize them with a Librem Key.
In addition to being able to flash a ROM from the GUI, it would also be
useful for a user to be able to add a GPG key to their keyring using the
flashing tool. This change adds the ability for a user to edit both a
ROM located on a USB key and also edit the running BIOS by using
flashrom to make a local copy of the running BIOS, edit it, then reflash
it. This also supports the upcoming delete feature in CBFS for
circumstances where keyring files already exist within CBFS.
If we want to modify a running BIOS we will need the ability to pull
down the current BIOS, modify it, and then reflash. This change adds a
read option to flash.sh and pulls down three versions of the BIOS and
only exists successfully if all three match.
To keep the flash logic simpler the GUI logic has been split into a
flash-gui.sh program so flash.sh behaves closer to the original flashrom
scripts it was based from. I've also removed the previous flashrom
scripts and incorporated their options into flash.sh. Finally I set
CONFIG_BOARD via the Makefile instead of setting a duplicate option in
each board's config.
Based on the conversation for PR #406, we decided to go with a more
generic script for general-purpose flashing instead of having individual
(and therefore very similar) flash scripts for each board type. This
script currently handles flashrom on Librem and X230 board types and
introduces a new CONFIG_BOARD option that sets specific flashrom
arguments based on the board.
It also adds support to gui-init to call this flash script.
Currently the TPM counter label is hard-coded. By changing it to a
variable in this function we can reuse all of the TPM counter functions
to create other monotonic counters in the TPM (if the hardware supports
it) with custom labels.
Wrapping text to 80 characters works but due to font size and padding
the maximum 80 character lines start to get truncated. Extending the
window to 90 characters will resolve this.
While the whiptail program wraps text appropriately based on column
size, the fbwhiptail program doesn't, leading to text that scrolls off
the window where it can no longer be read. This change wraps the longer
text output so it all fits.
Since fbwhiptail allows us to customize the background colors, we should
colorize warnings and error messages to provide a user with an
additional subtle cue that there might be a problem. I have added two
additional configuration options:
CONFIG_WARNING_BG_COLOR
CONFIG_ERROR_BG_COLOR
and in the librem13v2.config file you can see an example for how to set
them to be yellow and red gradients, respectively. I've also updated the
main two scripts that use whiptail to include those background colors.
If you decide to use regular whiptail, just don't set these config
options and it should behave as expected.
When a user gets confirmation of their boot menu choice, that's largely
to give them the option of making their boot choice the default. In the
case of "force mode" there's no reason for the user to be presented with
that dialog so this change skips right ahead to the boot once they have
In the event a user does pick the insecure "force" boot option that
bypasses checksum and signing checks in Heads, it would be nice to
provide a clear visual warning during the boot process that they are in
this state. This change will add a kernel argument that changes the boot
console background to be red and removes any boot splash that might
obscure it, in the event the user picks the insecure boot mode.
Since a user should only boot into this mode during emergencies, having
it be apparent that it's an unsafe mode helps ensure the user doesn't
pick this boot option needlessly.
Currently when the boot entries change, kexec-select-boot dies. Given
the normal loop is set up to catch this event and display a regular boot
menu at the next iteration of the loop, instead of dying it would be
better to just warn and then return from that function back into the
main loop. In addition to that I added a GUI menu for the same warning
when in GUI mode.
Part of the Heads workflow involves handling legitimate changes to /boot
as part of the package manager. This is a challenging workflow to handle
as package managers on many systems work in a completely unattended way
(and some even reboot first, apply updates, and then reboot again).
We need to be able to detect changes that are potentially caused by a
package manager so to do that I've set up a trigger within the OS
(currently just for Debian) that runs both before and after package
updates. It verifies the signatures in /boot and if they fail before
package updates it creates a log file in
/boot/kexec_package_trigger_pre.txt. If they fail after package updates
run /boot/kexec_package_trigger_post.txt is created. These files contain
the following fields:
CHANGED_FILES: A list of files in /boot that failed the sha256sum check
UPDATE_INITRAMFS_PACKAGE: An (optional) list of packages known to
trigger initramfs changes
Following those fields is a list of log output from the last package
manager run which contains its own formatted fields (I'm pulling from
/var/lib/dpkg/info).
When a user selects a boot option, gui-init first verifies the
checksums just to catch errors before calling kexec-select-boot. If
there are any errors it looks for these package logs and if they exist,
it displays appropriate warnings. If the files are absent it displays a
more generic warning. The user is also given an opportunity to re-sign
the /boot hashes.
There was a bug in the "force" boot mode where it would still fail if
signatures didn't match. This was because the check_config function
validates the signatures for kexec files. I've added a few conditionals
here so that in the case of a forced boot mode, we can bypass those
signature checks that would prevent boot and error out to a recovery
console.
The number of options we want in the menu is starting to get large
enough that it's worth slimming things down in the main menu and move
options to nested menus. Along with this nested menu change is the
option to re-sign and re-hash files in /boot directly from the menu.
One of the other core functions a user needs when bootstrapping is
taking over the TPM. I've added a new option in the menu for this and it
revealed that some of the menus needed more space so I've widened all
the menus and also made the main menu longer so the options don't
scroll.
The point of this change is to provide a failsafe (failunsafe?) mode for
less technically-savvy users who will ultimately be using Heads by
default on Librem laptops.
There are some scenarios where an end user might forget to update hashes
in /boot after an initrd change or might have some other hash mismatch.
Currently that user would then be stuck in a recovery console in Heads
not knowing what to do within that limited shell environment to fix the
situation.
This change adds a 'force' mode to kexec-select-boot that goes straight
into a boot menu and bypasses the hash checks so the user could more
easily get back into their system to attempt to repair it. It adds
appropriate warnings about why this is a risky option and moves it down
toward the bottom of the menu. The goal would be to just have this be an
emergency option our support could guide a user to if they ended up in
this situation.
Guarded linuxboot specific init entries
Removed Makefile entries into separate file (conflicts with srcing /etc/config)
Added CONFIG_BOOT_LOCAL/_REMOTE to control interface setup
Fixed CONFIG_TPM usage
if "CONFIG_TPM=y" is not present in the config file, functionalities
needing TPM could be disabled, while leaving other functionalities intact.
This will make Heads a more general-usage bootloader payload atop coreboot.
In particular I added a GUI menu to instruct the user if there is no
TOTP code registered (as is the case upon first flash) and also added
better handling of the case the user selects 'default boot' when there
is no default boot set yet. Apart from that where there were text-only
menus left in gui-init I've replaced them with GUI menus.
When selecting the boot menu option (m) in the gui-init you call out to
kexec-select-boot. To better maintain the graphical menu experience,
I've added a -g option to kexec-select-boot that, when set, will use a
graphical whiptail menu for the most common menu selection modes.
This is a modified version of the generic-init script that uses whiptail
to generate a graphical menu. I changed two of the options so that the
user can refresh the menu to get an updated TOTP code if needed.
USB smart card readers are most full speed devices, and there is no
"rate-matching hubs" beneath the root hub on older (e.g. GM45) plat-
forms, which has companion OHCI or UHCI controllers and needs cor-
responding drivers to communicate with card readers directly plugged
into the motherboard, otherwise a discrete USB hub should be inserted
between the motherboard and the reader.
This time I make inserting linux modules for OHCI and UHCI controllable
with option CONFIG_LINUX_USB_COMPANION_CONTROLLER.
A linux config for x200 is added as an example.
Tested on my x200s and elitebook revolve 810g1.
if "CONFIG_TPM=y" is not present in the config file, functionalities
needing TPM could be disabled, while leaving other functionalities intact.
This will make Heads a more general-usage bootloader payload atop coreboot.
This development branch builds a NERF firmware for the Dell R630
server. It does not use coreboot; instead it branches directly
from the vendor's PEI core into Linux and the Heads runtime
that is setup to be run as an EFI executable.
Closes issue #226
Also changed to procedure to show LVM volume groups and block
device ids to aid in choosing the right combination during the
TPM LUKS key sealing process.
Changed the checking of required hashes or required rollback state
to be right before boot, allowing the user to sign/set defaults
in interactive mode.
Also cleaned up usages of recovery and fixed iso parameter
regression.
Similar to qubes-update, it will save then verify the hashes of
the kexec files. Once TOTP is verified, a normal boot will verify
that the file hashes and all the kexec params match and if
successful, boot directly to OS.
Also added a config option to require hash verification for
non-recovery boots, failing to recovery not met.
Refactored boot parsing code and applied that in local-init to
scan /boot for grub options and allow the user to unsafely boot
anything. This goes a long way to addressing #196.
Optionally the user can customize those boot parameters or enforce
arbitrary hashes on the boot device by creating and signing config
files in /boot/ or /media/ or /media/kexec_iso/ISO_FILENAME/.
usb-boot automatically uses internal xen binary / command line
when multiboot is detected.
also tweaked to evaluate/remove variable refs in kexec arguments
Supports booting from USB media using either the root device or
a signed ISO as the boot device. Boot options are parsed with
quick/dirty shell scripts to infer kexec params.
Closes#195 and begins to address #196
This adds support for seamless booting of Qubes with a TPM disk key,
as well as signing of qubes files in /boot with a Yubikey.
The signed hashes also includes a TPM counter, which is incremented
when new hashes are signed. This prevents rollback attacks against
the /boot filesystem.
The TPMTOTP value is presented to the user at the time of entering
the disk encryption keys. Hitting enter will generate a new code.
The LUKS headers are included in the TPM sealing of the disk
encryption keys.
Issue #123: This streamline Qubes startup experience by
making it possible to have a single-password decryption.
Issue #29: The disk keys in `/secret.key` are passed to the systemd
in initramfs through `/etc/crypttab`, which is generated on each boot.
This is slow; need to look at alternate ways.
Issue #110: By using LVM instead of partitions it is now
possible to find the root filesystem in a consistent way.
Issue #80: LVM is now included in the ROM.
This also adds a set of files in the qubes/ directory that
are meant to be copied to the /boot partition.
Issue #154: for ease of upgrading Qubes, the script should
live on /boot instead of in the ROM. This requires a GPG
signature on the startup script to avoid attacks by modifying
the boot script.
Issue #123: this streamlines the boot process for Qubes, although
the disk password is still not passed in correctly to the initrd
(issue #29).
This does not address issues #110 of how to find the root device.
The best approach is probably disk labels, which will require
special installation instructions.
This addresses multiple issues:
* Issue #63: initrd is build fresh each time, so tracked files do not matter.
* Issue #144: build time configuration
* Issue #123: allows us to customize the startup experience
* Issue #122: manual start-xen will go away
* Issue #25: tpmtotp PCRs are updated after reading the secret
* Issue #16: insmod now meaures modules
Reduce the size of flashrom by commenting out most flash chips,
boards and programmers.
Wrapper script to make it easier to rewrite the ROM on the x230
using the flashrom layout.
Keep the entire 12 MB ROM for flashing.
This is a step towards unifying the server and laptop config (issue #139)
and also makes it possible to later remove the USB modules from the
normal boot path.
Replace the expired key with my updated key, although users
should add their own keys to sign their own firmware images.
Todo: document how to add/replace public keys.
Longer term todo: remove trusted key from the initrd image
so that there is nothing variable between different users'
builds.
