Use loader protocol for image verification where available, otherwise
fall back to the old shim lock protocol.
Signed-off-by: Mate Kukri <mate.kukri@canonical.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
These can be used to register a different implementation later,
for example, when shim provides a protocol with those functions.
Signed-off-by: Julian Andres Klode <julian.klode@canonical.com>
Signed-off-by: Mate Kukri <mate.kukri@canonical.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
The Xen image is loaded with an alignment, not always at "start".
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
More readable, could consume less space.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
We need to use FOR_LIST_ELEMENTS_SAFE() instead of FOR_LIST_ELEMENTS()
as single_binary_unload(), called during the loop, is changing the list
using grub_list_remove(). Given the environment probably the old code
simply removed only the first module on the list not freeing all the others.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
When attempting to build grub-emu the compilation failed with the
following error message:
include/grub/dl.h: In function ‘grub_dl_is_persistent’:
include/grub/dl.h:262:1: error: no return statement in function returning non-void [-Werror=return-type]
To avoid the error make the function always return 0.
Fixes: ba8eadde6be1 (dl: Provide a fake grub_dl_set_persistent() and grub_dl_is_persistent() for the emu target)
Signed-off-by: Gary Lin <glin@suse.com>
Cc: Daniel Axtens <dja@axtens.net>
Cc: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In grub_cmd_initrd(), memory is allocated for variable initrd_ctx
before calling grub_relocator_alloc_chunk_align_safe(). When the
function call fails, initrd_ctx should be freed before exiting
grub_cmd_initrd().
Fixes: CID 473852
Signed-off-by: Lidong Chen <lidong.chen@oracle.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Unload previous Linux kernel/initrd before updating the global variable
kernel_size. Otherwise the previous Linux kernel gets deallocated with
the kernel_size of the Linux kernel that is being currently loaded.
Signed-off-by: Adriano Cordova <adriano.cordova@canonical.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
If a simple string for arguments are passed it should be NUL terminated.
This is true for other code but not for "linux" command.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This is more consistent with the above code using sizeof(grub_efi_char16_t).
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Although the length should not exceed 2^31 grub_size_t is more
suitable for that variable. len is used to compute the size
of buffers which in C is a size_t, not a int. It is used
for GRUB_EFI_BYTES_TO_PAGES which expects unsigned values.
It is assigned to load_options_size which is unsigned, not signed.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
The size passed to grub_utf8_to_utf16() for the source string is
used as a limit for the string if NUL character is not encountered.
However, len, which is "strlen(src) * 2 + 2" is surely greater than
strlen(src). Pass the exact correct length.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
If the following allocation fails this would leave load_options NULL
while load_options_size not valid. If the allocation succeed
load_options_size is overwritten.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
(grub_size_t) -1 is never returned, the function always return
a not negative values. This is important for overflows considerations.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Commit 40e261b89b71 (lib/LzmaEnc: Validate "len" before subtracting)
ensures that the variable len is at least 2. As a result, GetLenToPosState(len)
never returns a value greater than or equal to kNumLenToPosStates,
making the changes introduced in the commit 16c0dbf4bc6a (lzma: Make
sure we don't dereference past array) unreachable and no longer necessary.
This reverts commit 16c0dbf4bc6a (lzma: Make sure we don't dereference past array).
Fixes: CID 481982
Signed-off-by: Lidong Chen <lidong.chen@oracle.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Correct a test failure in netboot_test and file_filter_test caused by an
issue cleaning up the tmp directory created for netboot. Netboot creates
a subdirectory in the tmp folder that causes the rmdir to fail - so
cleanup the subdirectory first.
Fixes: 1d59f39b5f1b (tests/util/grub-shell: Remove the work directory on successful run and debug is not on)
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Tested-by: Leo Sandoval <lsandova@redhat.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In bidi_line_wrap(), "kk - 1" in the for loop init, "i = kk - 1",
underflows when "kk" (unsigned int) is 0. Assigning the result of
"kk - 1" to signed int "i" may cause overflow. To address both
issues, cast "kk" to a signed type before subtraction to ensure
safe arithmetic and assignment.
Fixed: CID 473874
Signed-off-by: Lidong Chen <lidong.chen@oracle.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Trying to start grub-emu with a module that calls grub_dl_set_persistent()
and grub_dl_is_persistent() will crash because grub-emu fakes modules and
passes NULL to the module init function.
Provide an empty function for the emu case.
Fixes: ee7808e2197c (dl: Add support for persistent modules)
Signed-off-by: Daniel Axtens <dja@axtens.net>
Signed-off-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Avnish Chouhan <avnish@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In function protect_tpm2_export_tpm2key(), the "err" variable
is uninitialized in the normal (error free) path, so ensure this
defaults to GRUB_ERR_NONE.
This causes the GRUB build to fail with clang (observed with clang-14).
Fixes: 5934bf51c (util/grub-protect: Support NV index mode)
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This patch resolved a minor issue spotted by Coverity:
a983d36bd917 (gnulib/regexec: Resolve unused variable)
But, it was removed by the Gnulib update:
2b7902459803 (Update gnulib version and drop most gnulib patches)
It caused Coverity to continue to flag the issue. Daniel Kiper
suggested to bring back the patch a983d36bd917 (gnulib/regexec: Resolve
unused variable).
Fixes: CID 292459
Signed-off-by: Lidong Chen <lidong.chen@oracle.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Pull in Gnulib fix to allow lib/base64.c to compile using GCC 15 or newer.
Pulled from Gnulib commit 25df6dc425 (Silence some
-Wunterminated-string-initialization warnings.)
GCC 15 adds a new compiler warning "-Wunterminated-string-initialization"
that will trigger what is considered a false-positive in lib/base64.c as
this array is not treated as a string but an array of characters so the
lack of NUL string terminator is expected.
GCC team has added ability to flag such instances of arrays that the
compiler may think are strings as "nonstring" arrays to avoid this
warning: __attribute__((nonstring)).
Fixes: https://savannah.gnu.org/bugs/?66470
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In the function merge_state_with_log(), memory is allocated for the variable
next_nodes when creating a union of the variables table_nodes and log_nodes.
However, if next_state->entrance_nodes is NULL, then table_nodes becomes NULL
and we still allocate memory to copy the content of log_nodes. This can cause
a resource leak since we only free the memory for next_nodes if table_nodes
isn't NULL. To prevent this, we need to check that next_state->entrance_nodes
isn't NULL before allocating memory for the union.
This issue has been fixed in the latest version of gnulib and I've backported
this change to maintain consistency.
This issue was found by a Coverity scan of GRUB2 under the CID 473887.
Fixes: CID 473887
Signed-off-by: Alec Brown <alec.r.brown@oracle.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In the functions create_initial_state() and calc_eclosure_iter(), memory
is allocated for the elems member of a re_node_set structure but that
memory isn't freed on error. Before returning an error, a call to
re_node_set_free() should be made to prevent the resource leak.
This issue has been fixed in the latest version of gnulib and I've
backported this change to maintain consistency.
This issue was found by a Coverity scan of GRUB2 under the following
CIDs: 473869, 473888.
Fixes: CID 473869
Fixes: CID 473888
Signed-off-by: Alec Brown <alec.r.brown@oracle.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Add a few more tests to seal and unseal the key with the SHA-384 PCR
bank instead of the default SHA-256 PCR bank.
Signed-off-by: Gary Lin <glin@suse.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
TPM 2.0 Key File format stores the PCR selection in the parameters
for TPM2_PolicyPCR and it already contains the selected PCR bank.
Currently, tpm2_key_protector dumped the PCR bank specified by the
--bank option, and it may not be the PCR bank for key unsealing.
To dump the real PCR bank for key unsealing, this commit records the PCR
bank used by TPM2_PolicyPCR and dumps PCR values from that bank when
necessary.
Signed-off-by: Gary Lin <glin@suse.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
For tpm2_key_protector and grub-protect, SHA-256 is chosen as the hash
algorithm for the TPM session. However, grub-protect mistakenly used the
hash algorithm of the PCR bank to calculate PCR digest. If the user
chose a PCR bank other than SHA-256, grub-protect created a non-SHA-256
PCR digest to seal the key. But, tpm2_key_protector expects a SHA-256
PCR digest to the TPM unsealing session, so it would fail due to digest
mismatch.
This commit fixes the hash algorithm of PCR digest in grub-protect to
avoid the potential unsealing failure.
Fixes: https://github.com/lcp/grub2/issues/4
Signed-off-by: Gary Lin <glin@suse.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Several new header files have been added to GRUB which need
to be manually added to the dist archive. This allows building
from the tar archive created by "make dist".
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This file is auto-generated based on the selected platform and should
not be included in the source tarball.
Fixes: 6744840b (build: Track explicit module dependencies in Makefile.core.def)
Signed-off-by: Mike Gilbert <floppym@gentoo.org>
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In LzmaEnc_CodeOneBlock(), both GetOptimumFast() and GetOptimum()
returns a value of greater or equal to 1, which is assigned to
"len". But since LZMA_MATCH_LEN_MIN == 2, "len" should be validated
before performing "len - LZMA_MATCH_LEN_MIN" to avoid underflow
when "len" equals to 1.
Fixes: CID 51508
Signed-off-by: Lidong Chen <lidong.chen@oracle.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Reviewed-by: Sudhakar Kuppusamy <sudhakar@linux.ibm.com>
The potential overflow issue arises at "size += ret;" because "size"
is of type ssize_t (signed) while "len" is size_t (unsigned). Repeatedly
adding read sizes, "ret", to "size" can potentially exceed the maximum
value of ssize_t, causing it to overflow into a negative or incorrect value.
The fix is to ensure "len" is within the range of SSIZE_MAX.
Fixes: CID 473850
Fixes: CID 473863
Signed-off-by: Lidong Chen <lidong.chen@oracle.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Similarly to Intel IMSM, there are BIOS and UEFI implementations that
support DDF containers natively.
DDF and IMSM are very similar in handling, especially these should not
be considered as RAID abstraction. This fixes the requirement of having
a device map when probing DDF containers.
Fixes: https://issues.redhat.com/browse/RHEL-44336
Signed-off-by: Renaud Métrich <rmetrich@redhat.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Avoid attempting to defererence a NULL pointer to call read_symlink() when
the given filesystem does not provide a read_symlink() function. This could
be triggered if the calling filesystem had a file marked as a symlink.
This appears possible for HFS and was observed during fuzzing of NTFS.
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Correct several infinite loops/hangs found during fuzzing. The issues
fixed here could occur if certain specific malformed NTFS file systems
were presented to GRUB. Currently, GRUB does not allow NTFS file system
access when lockdown mode is enforced, so these should be of minimal
impact.
The changes made in this commit generally correct issues such as attempting
to iterate through a buffer using a length read from the NTFS file system
without confirming the length is larger than 0.
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Correct several memory access violations found during fuzzing.
The issues fixed here could occur if certain specific malformed NTFS
file systems were presented to GRUB. Currently, GRUB does not allow NTFS
file system access when lockdown mode is enforced, so these should be of
minimal impact.
The changes made in this commit generally correct issues where pointers
into data buffers were being calculated using lengths read from the
NTFS file system without sufficient bounds/sanity checking; or
attempting to access elements of a structure to free them, when the
structure pointer is NULL.
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Correct ntfs_test test failures around attempting to validate attribute
list entries as attributes. The NTFS code uses common logic in some
places to parse both attributes and attribute_lists which complicates
validation. Attribute lists contain different headers including a
different size of the length field (2 bytes) at offset 4 instead of the
4 byte length field used in attributes at offset 4. There are other
differences as well, but attempting to validate attribute list types
using attribute header validation was causing failure of the NTFS test
suite. This change restores some of the validation logic which may be
shared between attributes and attribute lists to be closer to the
original logic prior to fixes for previous CVEs. A following commit will
address some of the implications of removing this validation logic by
correcting some fuzzer failures (some which are exposed by removing the
validation in some of the cases).
Fixes: 067b6d225 (fs/ntfs: Implement attribute verification)
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Correct ntfs_test test failures around attempting to validate attribute
run list values. The calculation was incorrect for the "curr" variable.
With previous calculation, some file systems would fail validation
despite being well-formed and valid. This was caused by incrementing
"curr" by min_size which included both the (already accounted for)
min_size as well as the size of the run list. Correct by making a new
variable "run_size" to denote the current run list size to increment
both "curr" and "min_size" separately.
Fixes: 067b6d225 (fs/ntfs: Implement attribute verification)
Signed-off-by: Andrew Hamilton <adhamilt@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Environment files may contain empty lines, which should be ignored
during parsing. Currently, these lines are not skipped and resulting in
incorrect behavior. This patch adds a check to skip empty lines along
with those starting with "#".
Signed-off-by: Shreenidhi Shedi <shreenidhi.shedi@broadcom.com>
Reviewed-by: Alexey Makhalov <alexey.makhalov@broadcom.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Commit b66c6f918 (fs/zfs: Fix a number of memory leaks in ZFS code)
fixes many of the same leaks detected in bug #63846 except one, which
is fixed here.
Fixes: https://savannah.gnu.org/bugs/?63846
Fixes: b66c6f918 (fs/zfs: Fix a number of memory leaks in ZFS code)
Signed-off-by: Glenn Washburn <development@efficientek.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Those tests fail depending on the version of unifont. As we don't distribute
our own unifont it fails for most users. Disable them so that they don't mask
real failures. They can be reinstated once we solve unifont problem.
Signed-off-by: Glenn Washburn <development@efficientek.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
The option can be used to suppress output if we only want to test the
return value of the command.
Also, mention this option in the documentation.
Signed-off-by: Michael Chang <mchang@suse.com>
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Switching to another EFI boot application while there are secrets in
RAM is dangerous, because not all firmware is wiping memory on free.
To reduce the attack surface, wipe the passphrase acquired when
unlocking an encrypted volume.
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This commit adds the grub_cryptodisk_erasesecrets() function to wipe
master keys from all cryptodisks. This function is EFI-only.
Since there is no easy way to "force unmount" a given encrypted disk,
this function renders all mounted cryptodisks unusable. An attempt to
read them will return garbage.
This is why this function must be used in "no way back" conditions.
Currently, it is used when unloading the cryptodisk module and when
performing the "exit" command (it is often used to switch to the next
EFI application). This function is not called when performing the
"chainloader" command, because the callee may return to GRUB. For this
reason, users are encouraged to use "exit" instead of "chainloader" to
execute third-party boot applications.
This function does not guarantee that all secrets are wiped from RAM.
Console output, chunks from disk read requests and other may remain.
This function does not clear the IV prefix and rekey key for geli disks.
Also, this commit adds the relevant documentation improvements.
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Document the --cryptodisk-only argument. Also, document the
"cryptocheck" command invoked when that argument is processed.
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
When the --cryptodisk-only argument is given, also check the target
device using the "cryptocheck" command, if available.
This extends the checks to common layouts like LVM-on-LUKS, so the
--cryptodisk-only argument transparently handles such setups.
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This command examines a given diskfilter device, e.g., an LVM disk,
and checks if underlying disks, physical volumes, are cryptodisks,
e.g., LUKS disks, this layout is called "LVM-on-LUKS".
The return value is 0 when all underlying disks (of a given device)
are cryptodisks (1 if at least one disk is unencrypted or in an
unknown state).
Users are encouraged to include the relevant check before loading
anything from an LVM disk that is supposed to be encrypted.
This further supports the CLI authentication, blocking bypass
attempts when booting from an encrypted LVM disk.
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This allows users to restrict the "search" command's scope to
encrypted disks only.
Typically, this command is used to "rebase" $root and $prefix
before loading additional configuration files via "source" or
"configfile". Unfortunately, this leads to security problems,
like CVE-2023-4001, when an unexpected, attacker-controlled
device is chosen by the "search" command.
The --cryptodisk-only argument allows users to ensure that the
file system picked is encrypted.
This feature supports the CLI authentication, blocking bypass
attempts.
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This further mitigates potential misuse of the CLI after the
root device has been successfully unlocked via TPM.
Fixes: CVE-2025-4382
Signed-off-by: Maxim Suhanov <dfirblog@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
When large extent counter / NREXT64 support was added to GRUB, it missed
a couple of direct reads of nextents which need to be changed to the new
NREXT64-aware helper as well. Without this, we'll have mis-reads of some
directories with this feature enabled.
The large extent counter fix likely raced on merge with commit 07318ee7e
(fs/xfs: Fix XFS directory extent parsing) which added the new direct
nextents reads just prior, causing this issue.
Fixes: aa7c1322671e (fs/xfs: Add large extent counters incompat feature support)
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Anthony Iliopoulos <ailiop@suse.com>
Reviewed-by: Jon DeVree <nuxi@vault24.org>
Reviewed-by: Vladimir Serbinenko <phcoder@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
RAID 4 requires the same recovery module as RAID 5. Extend the condition to
cover both RAID levels.
Signed-off-by: Egor Ignatov <egori@altlinux.org>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
