Instead of emitting the per TPM PCR bank "boot_aggregate" values one
at a time, store them in a buffer and emit them all at once.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Run `make check' to execute the tests.
This commit only adds ima_hash test.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
The IMA measurement list boot_aggregate is the link between the preboot
event log and the IMA measurement list. Read and calculate all the
possible per TPM bank boot_aggregate digests based on PCRs 0 - 7.
Reading the TPM PCRs requires root permission, unless access to the
device (/dev/tpm0 or /dev/tpmrm0) has been granted.
Prior to calculating the boot_aggregate, the TPM PCRs themselves should
be validated by walking the TPM event log and re-calculating the PCRs.
(Such a test should be included as part of the TSS regression testsuites.)
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
This patch makes it possible to use the Intel TSS2 for getting
PCR values from the SHA1/SHA256 banks on a TPM2.
It is somewhat naive as it doesn't use the multi-PCR selection
that TSS2 is capable of, that is for a future patch.
Signed-off-by: Patrick Uiterwijk <patrick@puiterwijk.org>
[zohar@linux.ibm.com: added missing "stdint.h" in pcr_tsspcrread.c]
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Now that read_tpm_banks() reads the TPM 1.2 PCRs, remove the TPM 1.2
specific code for reading and verifying the SHA1 PCRs.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Extend read_tpm_banks() to support TPM 1.2, by reading TPM 1.2 SHA1 PCRs
into the first bank and mark the other banks as disabled.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
After walking the measurement list, re-calculating and extending the TPM
PCRs with the appropriate template digest for each bank, compare the
re-calculated PCR values for each TPM bank with the actual TPM values.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
tpm2_read_pcrs() reads the sha1 PCRs in order to verify the measurmeent
list. This patch adds support for reading other TPM banks.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
IMA currently extends the different TPM banks by padding/truncating the
SHA1 template digest. Although the IMA measurement list only includes
the SHA1 template digest, the template digest could be re-calculated
properly for each bank.
This patch adds support for properly calculating the template hash for
multiple TPM banks - "sha1" and "sha256".
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Opening a file for write when it is already opened for read, results in
a time of measure, time of use (ToMToU) error. Similarly, when opening
a file for read, when it is already opened for write, results in a file
measurement error. These violations are flagged by including 0x00's as
the template digest in the measurement list, but extending the TPM with
0xFF's.
In preparation of extending the TPM banks with bank specific digest
values, increase the "zero" and "fox" variable sizes.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
The TPM spec differentiates between an unknown bank and an unallocated
bank. In terms of re-calculating the PCR, treat them as equivalent.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Change main() return code from -1 to 125 as -1 is not really valid exit
code. 125 is choosen because exit codes for signals start from 126.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Previously for EVM verify you should specify `--hashalgo' option while
for IMA ima_verify you didn't.
Allow EVM verify to determine hash algo from signature.
Also, this makes two previously static functions to become exportable
and renamed:
get_hash_algo_from_sig -> imaevm_hash_algo_from_sig
get_hash_algo_by_id -> imaevm_hash_algo_by_id
This is needed because EVM hash is calculated (in calc_evm_hash) outside
of library.
imaevm_hash_algo_by_id() will now return NULL if algo is not found.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
If user wants to verify v1 signature and specify RSA public key in `-k'
option, this key will be attempted to be loaded as x509 certificate and
this process will output errors.
Do not load a key as a x509 cert if user pass `--rsa'.
This is not perfect solution. As now it's possible to specify `-k' and
`--rsa' and v2 signatures will not verify, because of no keys.
This improvement is not added into ima_measurement().
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
If any tested file results in failure produce failure exit code.
Previously exit code affected only by the last file tested.
Fixes: "Allow multiple files in ima_verify"
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
OPENSSL_init_crypto() was introduced in version 1.1. When using an
older version of openssl, don't call OPENSSL_init_crypto. Partially
revert commit 782224f33cd7 ("ima-evm-utils: Rework openssl init").
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Two chars were left in a AC_DEFINE() in configure.ac, leading to an error
message during ./configure call:
checking for tsspcrread... yes
./configure: line 9894: ],: command not found
Signed-off-by: Bruno E. O. Meneguele <bmeneg@redhat.com>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Updated both the release and library (ABI change) versions. See the
"Changelog" for a short list of the new features, bug fixes, and code
cleanup.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Make sign_hash_v1(), sign_hash_v2(), get_hash_algo_by_id, and
pkey_hash_algo[] and pkey_hash_algo_kern[] static.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Prefix `dump', `do_dump', and `params' with `imaevm_' to avoid colliding
with other global symbols.
Also, rename `libevm_' to `libimaevm_`, only used with `params'.
Additionally, rename `dump' into `hexdump'.
Finally, rename `get_hash_algo' to `imaevm_get_hash_algo' as suggested by
Mimi Zohar.
Lines that became too long are split, indent corrected. No code changes.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Reviewed-by: Bruno E. O. Meneguele <bmeneg@redhat.com>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Falling back and permissiveness could have security implications.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
The kernel does not expose the crypto agile TPM 2.0 PCR banks to
userspace like it exposes PCRs for TPM 1.2. As a result, a userspace
application is required to read PCRs.
This patch adds tsspcrread support for reading the TPM 2.0 PCRs.
tsspcrread is one application included in the ibmtss package.
Sample error messages:
Failed to read PCRs: (tsspcrread failed: No such file or directory)
Failed to read PCRs: (TSS_Dev_Open: Error opening /dev/tpmrm0)
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Reviewed-by: Petr Vorel <pvorel@suse.cz>
Reviewed-by: Bruno E. O. Meneguele <bmeneg@redhat.com>
ERR_load_crypto_strings() is already called in other place.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Each tima a new unknown key is encountered, emit a message of the format
"key #: <keyid> (unknown keyid)". The individual files using unknown
keys are then only logged in verbose mode. Also update the message
emitted to be consistent with other "verification failed" messages.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Changlog:
- Incorporated Vitaly's fix to prevent a null dereference in `tail->next`
Previously OpenSSL errors was delayed until evmctl exit (sometimes not).
Since we try to make libimaevm more robust, there could be many errors
accumulated, so it's useful to output OpenSSL errors as they happen.
This will also make output more understandable as you can see which
openssl error correspond to which libimaevm error.
Additionally, change spelling of read_pub_pkey and read_priv_pkey to
include key file name.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Since we now always call verify_hash_v2() with NULL keyfile (assuming
all keys are already loaded into public_keys list), remove keyfile
argument and its handling from verify_hash_v2().
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
[zohar@linux.ibm.com: make verify_hash_v1() and verify_hash_v2() static.]
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
This is more human understandable and also will improve handling of
the sources by cscope.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
When public keys are specified on the boot command line (--key "<public
key file>,[<public key file>,...]"), the appropriate public key is used
to verify EVM or file signatures. If no keys are specified, the default
x509_evm.der or pubkey_evm.pem file is used to verify the DIGSIG_VERSION_2
or DIGSIG_VERSION_1 signatures respectively, without first checking the
keyids. Instead of emitting a "verification failed: 0 (invalid
padding)" message, an "unknown keyid" message would be clearer.
To address this problem, when no public keys are specified, this patch
loads the x509_evm.der default public key onto the "public_keys" list,
while the pubkey_evm.pem continues to be passed to verify_hash_v1()
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Check return value of fstat(2) in add_file_hash() and remove
now unused get_fdsize().
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
`selinux_str',`caps_str', and `ima_str' are passed from the command line
but copied into the fixed-size buffer.
Yes, length of `selinux_str' is calculated differently than of `caps_str'.
Fixes: CID 229895.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
file2bin() may return NULL, which is set to tmp, which is passed to
memcpy. Add explicit check for it.
Fixes: CID 229904.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Other than the "boot-aggregate" measurement entry in the IMA
measuremeent list, all other measurements are of file data. Kernel
support was recently added to support measuring the kexec boot command
line buffer, which is stored in a new template field named 'buf'.
This patch adds support for a new template named "ima-buf", defined as
"d-ng|n-ng|buf".
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
New, unknown template formats containing unknown fields are not
processed, resulting in "remain unprocessed data" messages. Processing
these unknown fields is unnecessary for walking the measurement list to
re-calculate the PCRs.
The "remain unproccessed data" may also be emitted for malformed, known
template records.
This patch limits the number of messages emitted to once per template
format and includes the template name in the message.
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Struct signature_v2_hdr is packed so clang complains that taking address
of packed member may result in an unaligned pointer value:
libimaevm.c:481:21: warning: taking address of packed member 'keyid' of class or structure 'signature_v2_hdr' may result in an unaligned pointer value
[-Waddress-of-packed-member]
__be32_to_cpup(&hdr->keyid));
^~~~~~~~~~
libimaevm.c:905:17: warning: taking address of packed member 'keyid' of class or structure 'signature_v2_hdr' may result in an unaligned pointer value
[-Waddress-of-packed-member]
calc_keyid_v2(&hdr->keyid, name, pkey);
^~~~~~~~~~
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
This allows testing multiple verify in a row, similar to ima_measurement.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
When pctx is allocated using EVP_MD_CTX_new() it should be freed.
Found with ASan.
Fixes: 81010f0 ("ima-evm-utils: Add backward compatible support for openssl 1.1")
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
It's useful to know not just a hash value but also which algorithm is
used.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Move sign_hash()/ima_calc_hash()/calc_evm_hmac()/calc_evm_hash() status
checking before assert()'ing of their return values, so it can be passed
to the upper level callers. Especially useful for showing errors.
Fixes: 1d9c279279 ("Define hash and sig buffer sizes and add asserts")
Fixes: 9643544701 ("Fix hash buffer overflow in verify_evm and hmac_evm")
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
ima-evm-utils: Fix assert after ima_calc_hash
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
RSA_ASN1_templates[] are not needed anymore, because we switched to the
generic EVP_PKEY OpenSSL API to generate v2 signatures instead of
constructing PKCS1 ourselves.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Acked-by: Dmitry Kasatkin <dmitry.kasatkin@gmail.com>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Finish conversion of calc keyid v2 to EVP_PKEY API. After sign_hash_v2()
is switched to EVP_PKEY API (in previous commit), older RSA-specific
calc_keyid_v2() does not needed anymore and can be replaced with
calc_pkeyid_v2().
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Acked-by: Dmitry Kasatkin <dmitry.kasatkin@gmail.com>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Convert sign_hash_v2() to use more generic EVP_PKEY API instead of RSA
API. This enables generation of more signatures out of the box, such as
EC-RDSA (GOST) and any other that OpenSSL supports. This conversion also
fixes generation of MD4 signatures, because it didn't have proper
RSA_ASN1_template.
Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Acked-by: Dmitry Kasatkin <dmitry.kasatkin@gmail.com>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
