On machines without a TPM, we'd still like some way for the BIOS to
attest that it has not been modified. With a Librem Key, we can have the
BIOS use its own ROM measurement converted to a SHA256sum and truncated
so it fits within an HOTP secret. Like with a TPM, a malicious BIOS with
access to the correct measurements can send pre-known good measurements
to the Librem Key.
This approach provides one big drawback in that we have to truncate the
SHA256sum to 20 characters so that it fits within the limitations of
HOTP secrets. This means the possibility of collisions is much higher
but again, an attacker could also capture and spoof an existing ROM's
measurements if they have prior access to it, either with this approach
or with a TPM.
Signed-off-by: Kyle Rankin <kyle.rankin@puri.sm>
TPM password must be 1-32 characters. Loop if the password is not
valid or the repeated password doesn't match, so the user can try
again.
Move prompt_new_owner_password to functions and use in both gui-init
and tpm-reset.
Fixes#1336
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
init must use busybox ash because it is used on legacy-flash boards.
Change shebang, move needed functions to ash_functions.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
Multiple traps overwrite each other. While no tpmr functions have more
than one trap right now, it is fragile, and the quoting is complex due
to double expansion. Use at_exit to add exit handlers that accumulate
and do not require special quoting.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
We just set the TPM owner password, so there's no need to make the user
enter it again. Eliminates some failure modes if the user mistypes it
or enters the wrong password.
Allow optionally passing in the TPM owner password in tpmr seal,
check_tpm_counter(), seal-totp, and generate_totp_htop(). The user is
still prompted if the password is needed but was not provided, so
existing uses in other contexts continue to work unchanged.
Prompt for the password in reset_tpm() and pass it down to each of the
above.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
gui-init: do not consume two unseal attempt to unseal both totp and hotp + cosmetic changes (slow down TPM DA lockout)
kexec-seal-key: Add DEBUG statement for PCR precalc
seal-totp: add DEBUG statements regarding skipping of PCR5 and PCR6 involvement into TOTP/HOTP sealing ops
seal-hotpkey: Add DEBUG statements related to reuse of TOTP sealed secret
tpmr: add DO_WITH_DEBUG calls to output pcrread and extend calls
tpmr: typo correction stating TRACE calls for tpm2 where it was for tpm1
tpmr: add DO_WITH_DEBUG calls for calcfuturepcr
functions: Cosmetic fix on pause_recovery asking user to press Enter to go to recovery shell on host console when board defines CONFIG_BOOT_RECOVERY_SERIAL
Not so related but part of output review and corrections:
kexec-insert-key: cosmetic changes prepending "+++" to disk related changes
kexec-save-default: cosmetic changes prepending "+++" to disk related changes
config/coreboot-qemu-tpm*.config: add ccache support for faster coreboot rebuild times
Provide mask_param() function to uniformly mask secret parameters,
while still indicating whether they are empty.
Extend DO_WITH_DEBUG to allow masking a password parameter by position,
using mask_param(). Move from ash_functions to functions (isn't used
by ash scripts).
Mask password parameters in kexec-unseal-key and tpmr seal. Use
mask_param() on existing masked params in tpmr.
Trim more troubleshooting output from tpm2_extend() in tpmr.
Clarify tpmr kexec_finalize echo; it's the TPM's platform heirarchy,
users might not know what this was referring to.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
Busybox no longer has CONFIG_BASH since we are deploying bash on most
boards. We also should clearly indicate which scripts cannot use
bashisms.
Change shebang in x230-flash.init, t430-flash.init, flash.sh to
/bin/ash. Execute /bin/sh for interactive shells.
Move key functions needed by those scripts to initrd/etc/ash_functions.
Source ash_functions instead of functions in those scripts, so any
bashisms in other functions won't break parsing of the script in ash.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
Use /bin/sh (ash in busybox builds) for interactive shells, not bash.
Preparation for trimming interactive features from bash to reduce size.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
Trace parameters to seal/unseal and some key tpm2 invocations. Trace
invocation of tpmr seal/unseal for disk unlock key.
Add DO_WITH_DEBUG() to trace a command and parameters, then execute it.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
- /tmp/debug.log is created and appended by all TRACE and DEBUG calls in code
- fix some logic errors seen when no DEBUG entry were outputted in /tmp/debug.log
Most logic throughout Heads doesn't need to know TPM1 versus TPM2 (and
shouldn't, the differences should be localized). Some checks were
incorrect and are fixed by this change. Most checks are now unchanged
relative to master.
There are not that many places outside of tpmr that need to
differentiate TPM1 and TPM2. Some of those are duplicate code that
should be consolidated (seal-hotpkey, unseal-totp, unseal-hotp), and
some more are probably good candidates for abstracting in tpmr so the
business logic doesn't have to know TPM1 vs. TPM2.
Previously, CONFIG_TPM could be variously 'y', 'n', or empty. Now it
is always 'y' or 'n', and 'y' means "any TPM". Board configs are
unchanged, setting CONFIG_TPM2_TOOLS=y implies CONFIG_TPM=y so this
doesn't have to be duplicated and can't be mistakenly mismatched.
There were a few checks for CONFIG_TPM = n that only coincidentally
worked for TPM2 because CONFIG_TPM was empty (not 'n'). This test is
now OK, but the checks were also cleaned up to '!= "y"' for robustness.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
-coreboot support of TPM v2.0 (shared config for TPM2 support across all 4 previous variations)
-swtpm set to be launched under TPM v2.0 mode under board config
-Documentation file under each board.md softlinks to qemu-coreboot-fbwhiptail-tpm1.md (which has been generalized)
This is skeleton for TPM v2 integration under Heads
-------------
WiP
TODO:
- libcurl cannot be built as a tpm2-tools dependency as of now not sure why. curl currently needs to be added in board config to be built
- Note: tpm-reset (master and here) needs some review, no handle of no tpm use case. Caller is responsible to not call it otherwise does nothing
- init tries to bind fd and fails currently
- Note: Check if whiptail is different of fbwhiptail in clearing screen. As of now every clear seems to be removed, still whiptail clears previous console output
- When no OS' /boot can be mounted, do not try to TPM reset (will fail)
- seal-hotpkey is not working properly
- setting disk unlock key asks for TPM ownership passphrase (sealing in NV requires ownership, but text is misleading user as if reowning TPM)
- We should cache input, feed tpm behind the scene and wipe passphrase and state clearly that this is TPM disk unlock kye passphrase.
- primary key from TPM2 is invalid most of the time from kexec-select-boot and verifying global hashes but is setuped correctly at disk unlock key setup
- would be nice to take advantage of bash function tracing to understand where we are for debugging purposes, code takes ash in consideration only
- tpmr says it implements nv calls but actually doesn't. Removing those falsely wrapped functions would help.
- Implementing them would be better
- REVIEW TODOS IN CODE
- READD CIRCLECI CONFIG
Current state:
- TPM unseal works without disk unlock key and generates TOTP properly (was missing die condition at unseal to not produce always good TOTP even if invalid)
- TPM disk encryption key fails. Hypothesis is that sealing with USB drivers loaded and measures in inconsistent with sealed with/without.
- TPM disk unsealing happens without USB modules being loaded in non-HOTP setup. This fails.
- Current tests are with fbwhiptail (no clear called so having traces on command line of what happens)
- Testing with HOTP implementation for sealing/unsealing since that forces USB module loads on each boot to remove this from failing possibilities
- Add TRACE function tracing output under etc/functions, depending on CONFIG_ENABLE_FUNCTION_TRACING_OUTPUT enabled in board configs
- Replace current DEBUG to TRACE calls in code, reserving DEBUG calls for more verbose debugging later on (output of variables etc)
- add 'export CONFIG_ENABLE_FUNCTION_TRACING_OUTPUT=y' in qemu-coreboot(fb)whiptail-tpm1(-hotp) boards to see it in action
Adds check to detect device formatted as fat32 without partition table.
With fat32 fdisk does not print message about invalid partition table
and instead it'll print an empty table with header.
In both cases total output has the same length of 5 lines: 3 about
device info, 1 empty line and the 5th will be the table header or
invalid partition message.
Signed-off-by: Daniel Pineda <daniel.pineda@puri.sm>
Since it's not supposed to be shell safe, just display safe
inside double quotes, we can allow some more characters.
Also fix the escape character not being escaped.
busybox sha256sum will create a checksum file for uncommon file names
(e.g. /boot/foo"$\n"bar), but fail to verify that exact file.
https://bugs.busybox.net/show_bug.cgi?id=14226
Thus disallow all files in /boot/ with strange file names at the time of
signing for now. Verifying in the presence of new files with such file
names in /boot/ is no issue for the kexec_tree verification due to the
previously implemented escaping mechanism.
Attempt to fix the following issues:
1. unescaped file names may let an attacker display arbitrary
whiptail prompts --> escape, original code by @JonathonHall-Purism
2. whiptail itself allows escape characters such as \n
--> use an escape character not used by whiptail, i.e. #
3. performance issues caused by diff'ing too early -->
only generate a diff to display to the user, if an actual issue is
found
Extract exclusion for unpartitioned block device of partitioned media
to gui_functions, and exclude them even if kernel hasn't listed the
partitions yet. (Fixes flash/USB boot prompts incorrectly trying to
use the whole device for partitioned media the first time.)
Ignore block devices of size 0, like empty USB SD card readers.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
bin/kexec-parse-boot: test 2bb1f52bf5 that fix correctly comma seperated arguments.
Still TODO: when booting, Heads tries to find where the ISO with /dev/disk/by-label/ARCH_202202 wich is never brought up. uuids could, not sure why the label is not brought up correctly. Maybe an issue in the way Arch makes the ISO.
@tslilc : Any idea to continue #584 or modify #762?
Some devices, like older Librem laptops, use BAT vs BAT0 for the
primary battery. Accomodate this by using BAT* instead of BAT0.
Could break on multi-battery devices, but none supported in tree
currently so more of a theoretical problem.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Grepping on just 'Disk' can lead to disk UUID identifier strings
being added to /tmp/disklist, which then fail to parse later on.
Avoid this by grepping on 'Disk /dev' instead.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Modify update_checksums() to return 0/1 on success/failure, and modify
callers to notify the user on error with a GUI msg.
Set background color as appropriate in gui-init.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Persist the background color (and error state) through
the main menu and all submenus. Use warning
background color for destructive operations, error color
for errors.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Set and export currently-used defaults in gui-init, but still
allow for inidividual boards to override via config if desired.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Update_checksum was already changing files in /boot, befor checking for
gpg card. If no card is present, the user will end up in the recovery
next time instead of getting the same dialog again. Therefore, the
confirm_gpg_card should be checked before altering files.
The dead -u flag/$update_counter is used to mark the necessisty to
update the hash files now.
Currently, exiting the recovery shell results in a kernel panic,
necessitating a hard reset / power cycle. As this is less than ideal,
drop the exec and add a loop to restart the shell.
Addresses issue #833
Tested under qemu-coreboot-fbwhiptail
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Using 'let' in these scripts fails when evaluating to zero
for some reason, so replace with '$(())' which works as intended.
Test: Boot device selection menu shown properly when
new/unpartitioned drive installed.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Currently, /media is mounted once per boot, which causes issues
if a user need to change USB sticks, or unknowning performs an
operation that mounts /media and then needs to access a different
USB stick later (eg, updating the firmware).
To mitigate this, always unmount /media if mounted before scanning
for USB devices, so the user can choose the correct device at the
time of its use.
Additionally, add a unique exit code for user abort so we're not
treating it the same as a failure, and use it to prevent unnecessary
GUI prompts when cancelling selection of a USB device.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
For the handful of operations which need to be done with /boot
as the pwd, encapsulate them in subshells to ensure the pwd
doesn't unexpectedly change for other operations, as functions
which need to mount/unmount /boot may fail if the pwd isn't root.
Also, set the pwd to root at the start of detect_boot_device as an
added safety measure.
Test: run oem-factory-reset function, ensure it doesn't fail to
detect boot device due to incorrect working directory.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Move code duplicated across several GUI scripts into a common
gui_functions file and include/use that.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
If kexec-sign-config fails due to GPG key not present,
the double die() results in a kernel panic (and if it didn't,
/boot would be left mounted RW). Fix this by removing call to
die() and ensuring /boot remounted RO regardless checksum
update success or failure.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Add function to detect boot device. Start by checking
CONFIG_BOOT_DEV, then iterate thru all bootable partitions.
Check if partition is mountable, contains grub directory.
Update CONFIG_BOOT_DEV and mount on /boot if successful.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Initial card detection can sometimes fail, so prompt the user
to remove/reinsert their GPG card before retrying. Since
errexit is likely set, disable it prior to calling --card-status so
we can handle the error ourself, then re-set if necessary when done.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
Move the non-GUI component of update_checksums() to
/etc/functions so it can be reused outside of gui-init.
Add check that /boot/kexec_default_hashes.txt exists before parsing
it, since doesn't exist if there's no default boot target set yet.
Eliminates spurious error text and/or premature exit depending on
state of errexit.
Signed-off-by: Matt DeVillier <matt.devillier@puri.sm>
There are cases when grepping for an option in the config file where
grep will not find it, which is fine in this case, but without adjusting
the exit code in that case it can make an entire script bail out.
We need to handle the case where the specific config file doesn't exist,
or else grep fails, so we touch the file ahead of time. Mounting the usb
storage caused problems when you re-enter the menu a second time, so we
will just load the storage module.
As part of the config gui we want to be able to have the system define
new config options without them being lost if the user makes their own
changes in CBFS. To allow that this change creates a function initiated
in init that combines all /etc/config* files into /tmp/config. All
existing scripts have been changed to source /tmp/config instead of
/etc/config. The config-gui.sh script now uses /etc/config.user to hold
user configuration options but the combine_configs function will allow
that to expand as others want to split configuration out further.
As it stands here are the current config files:
/etc/config -- Compiled-in configuration options
/etc/config.user -- User preferences that override /etc/config
/tmp/config -- Running config referenced by the BIOS, combination
of existing configs
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.
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.
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.
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.
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.
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.
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 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.
