An undersized e_phentsize made the last program-header loop iteration
read past the validated table (and past the image buffer when the
table ends at the image end). Reject such images with -3.
Verification: tools/unit-tests/unit-elf-phentsize fails pre-fix
(undersized ELF accepted, ret=0), passes post-fix (2/2); full
make -C tools/unit-tests run green.
wolfBoot_seal_blob() and wolfBoot_unseal_blob() unload the
policy_session TPM handle on cleanup but leave the stack session object
(SRK-derived session key material) in the frame; ForceZero it in the
common cleanup of both.
wolfBoot_seal_auth()/wolfBoot_unseal_auth() copy the caller authValue
into a stack WOLFTPM2_KEYBLOB and the NV helpers copy it into a stack
WOLFTPM2_NV; none were cleared before return, inconsistent with the
existing TPM2_ForceZero(&unsealOut) in the same file. ForceZero each
stack object at the single exit.
wolfBoot_tpm2_deinit() unloads the session and SRK handles and cleans
up the device, but the file-scope wolftpm_session (HMAC/parameter-
encryption session key) and wolftpm_srk (SRK authValue) stayed in .bss
SRAM for the booted OS to read. ForceZero both after Cleanup.
wolfBoot_verify_signature_ecc() imports the raw signature into mp_int
r/s and passes them to wc_ecc_verify_hash_ex() without clearing them;
mp_clear() scrubs the digit memory before the stack frame retires.
The small-stack crypto pool hands out workspace for hash blocks and
signature verification state; the slot was released back to the pool
still holding that data. Scrub it in XFREE before marking the slot
free (wc_ForceZero, same helper the rest of the crypto path uses).
The CTR counter block built on the stack is a full copy of the secret
disk-encryption nonce plus counter; both in-tree sibling IV helpers
already scrub their stack copy, this was the only one that did not.
A 0 return (no base clock / ICS never stabilized) proceeded with
an unconfigured card clock. Fail sdhci_init, emmc_card_full_init
and the UHS-I 50MHz step when the clock is not programmed.
The do/while fell out with status==0 when the card stayed busy
until timeout, indistinguishable from supported, so the SWITCH
command went out for a function the card never cleared. Set
status=-1 when timeout reaches 0.
ERASE_TIMEOUT was defined but never used; the final erase ACK used
the plain short WAIT_CYCLES budget, so a slow remote flash made
ext_flash_erase() fail even when the erase succeeded. Factor
wait_ack_cycles() out of wait_ack() and use WAIT_CYCLES *
ERASE_TIMEOUT for the erase ACK, like uart_rx_timeout() does with
READ_TIMEOUT. Add unit tests for the success and timeout paths.
The signer (sign.c) accepts total_img_sz == max_img_sz (the exact fit:
header + payload + trailer == partition size), but wolfBoot_update
rejected fw_size == MAX_UPDATE_SIZE via '> MAX_UPDATE_SIZE - 1', leaving
a 1-byte gap and an off-by-one between the two components. Drop the '- 1'
so the updater accepts the exact fit the signer produces. Rename
test_update_max_size_rejected to _accepted (version 2 now stages) and add
test_update_max_size_plus_one_rejected (MAX_UPDATE_SIZE + 1 still rejected).
Check ata_async_info.in_progress before preparing the slot or touching
the shared buffer, so a second security command cannot clobber the
passphrase still being DMA'd. With that guard an async call always
means in flight (exec_cmd_slot_ex sets in_progress and returns BUSY),
so the post-command scrub is unconditional for the async path.
c3f9717f guarded the success/trigger/sector-flag declarations behind
WOLFBOOT_FIXED_PARTITIONS, but wolfboot.h includes target.h only under
__WOLFBOOT, so test-apps (compiled without __WOLFBOOT) never saw the
macro and failed with implicit-declaration errors on imx95_m7, mcxa,
mcxw and mcxw-tz. Unguard the header declarations; keep the definitions
guarded and move wolfBoot_success out of the fixed-partition guard
since it only uses the partition-state API.
Called from the RAM-resident update path (update_ram.c) under
EXT_FLASH; place the body in RAM so it runs after flash remap
instead of faulting.
Verification: built unit-custom-trailer-nopart (EXT_FLASH) - compiles
clean; imx95_m7 (EXT_FLASH, CI repro) - compiles clean.
- libwolfboot.c: keep the partition state APIs (set/get_partition_state)
available to CUSTOM_PARTITION_TRAILER / WOLFBOOT_NO_PARTITIONS builds;
guard the fixed-partition APIs (sector flags, erase, trigger, success)
behind WOLFBOOT_FIXED_PARTITIONS. Add the unit-custom-trailer-nopart
compile target to lock in that configuration.
- unit-p1021-fcm-bytes.c: add program tests for status 0x81 (fail) and
0x00 (write-protected), asserting no later page is programmed.
- unit-p1021-erase-advance.c: add an erase test for status 0x00
(write-protected), asserting failure before advancing to the next block.
- hal/nxp_p1021.c: fix ONFI status polarity (bit 0 set = fail, bit 7
clear = protected; was checking bits 0+1 set = success)
- src/libwolfboot.c: move partition state APIs under HAVE_PARTITION_TRAILERS
so CUSTOM_PARTITION_TRAILER without WOLFBOOT_FIXED_PARTITIONS links
- src/x86/ata.c: do not zeroize DMA buffer when another async op is
in progress (buffer still owned by first command's transfer)
- unit-p1021-fcm-bytes/erase-advance: update mocks to ONFI status bytes
- unit-update-ram-custom-trailer: add BOOT selection assertion via
mock_max_read_addr, PID-suffixed temp files, unlink in cleanup
- unit-update-ram-nofixed-noramboot: unlink PID-specific temp files
- src/x86/ata.c: verify PxCI is clear in the TFES path before scrubbing
the DMA buffer (symmetric with the success path; prevents scrubbing
while the HBA may still reference the buffer)
- unit-update-ram-nofixed-noramboot.c: append PID to /tmp filenames to
avoid collisions under parallel make -j or shared runners
Add HAVE_PARTITION_TRAILERS macro (set when WOLFBOOT_FIXED_PARTITIONS or
CUSTOM_PARTITION_TRAILER is defined) and use it to gate the TESTING state
transition in update_ram.c. Previously the gate was WOLFBOOT_FIXED_PARTITIONS
only, so targets using custom partition trailers never transitioned from
UPDATING to TESTING, allowing rollback to older versions.
Add unit-update-ram-custom-trailer test covering the CUSTOM_PARTITION_TRAILER
path with mocked get/set_trailer_at callbacks.
The no-base-clock path disables the clock and returns 0 (error),
unlike the already-set path above which returns last_clock_khz.
The old NOTE claimed callers could not tell the two apart.
security_command_passphrase() only wiped the static DMA buffer on the
synchronous path; in async mode (ata_security_erase_unit) the
passphrase stayed resident in the file-static buffer for the rest of
the boot, since ata_cmd_complete_async() never scrubbed it.
Add a scrub_buffer flag to the async state, set it when the command
goes in flight (or wipe immediately if it never started), and scrub in
ata_cmd_complete_async() on both the success and task-file-error exits,
once the HBA has retired the command.
Covered by two new unit tests: scrub after async completion and after
an async port error; the in-flight state is asserted unscrubbed.
Reported by Fenrir.
ARM_TEE_PS_SET on an existing object XMEMCPY'd the new value over the
old one without clearing the buffer first, so a SET that stores less
data (or zero data) left the tail of the previous object readable
through ARM_TEE_PS_GET.
Zero the full data area with wc_ForceZero after every validation check
passes and before the copy, matching the DELETE path.
Reported by Fenrir.
pkcs11_crypto_init() tore down the session on failure but left the
pre-populated pkcs11_pin credential in retained bootloader memory
after the handoff. Wipe it in the failure path, same as the deinit
path (F-12114). Regression test in unit-pkcs11-pin-zeroize: login
rejected -> init fails -> every pin byte is zero.
The version gate in the retry loop panicked whenever the selected
candidate failed verification and the fallback partition carried a
lower version, even though that image is valid and signed. A single
corrupted higher-version image bricked the device instead of falling
back. The gate only ever fired on the failure path (candidate
selection already prefers the higher version), so remove it; the
TESTING-state anti-rollback check in wolfBoot_dualboot_candidate()
still guards the confirmed-update path.
wolfBoot_start() zeroed os_image once, before the retry loop, so the
fallback iteration kept the stale img->hdr of the failed partition and
wolfBoot_open_image_address() (which only adopts load_address when
hdr is NULL) re-verified the wrong image. The external header cache
had the same problem: it kept the first image opened.
memset the struct and invalidate the EXT_FLASH header cache at the
top of each loop iteration, per the documented precondition of
wolfBoot_open_image_address().
Add unit-update-ram-nofixed-noramboot: the non-fixed-partition,
non-RAMBOOT (XIP) layout where load_address varies per retry, with
fallback tests both directions plus candidate-selection checks.
The weak no-op sat inside NVM_FLASH_WRITEONCE, so pkcs11_store.c failed to
link on the nrf5340/nrf54l TrustZone configs. STM32F4 and STM32G4 enable the
flash instruction/data caches and never reset them, so give them a real one.
The store commits sectors with hal_flash_erase()/hal_flash_write() and
then reads them back through the memory map - sector_ptr(),
cache_get_sector()'s refill, and the raw magic reads in check_vault().
On a part that caches flash reads (STM32 ICACHE) those reads can return
pre-erase bytes. check_vault() is the worst case: a stale magic there
does not merely read wrong, it triggers restore_backup() or a full vault
re-initialisation, losing the token.
Adds power-fail injection to the flash mock and a test that cuts power at
every flash operation of a rewrite window, asserting the object always reads
back as the whole old payload, the whole new payload, or empty. Fails at
op 3 without this fix, passes with it.
unit-pkcs11_store 10/10; full unit-tests suite green.
Daniele Lacamera
Aidan Garske
David Garske