Check that there are sufficient bytes in the buffer before reading the
cipherSize from it. Also, reduce the bufferSize variable by the number
of bytes that make up the cipherSize to avoid reading and writing bytes
beyond the buffer in subsequent steps that do in-place decryption.
This fixes CVE-2023-1017 & CVE-2023-1018.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Openssl 3.0 did return an error if EVP_PKEY_CTX_set0_rsa_oaep_label was called
with label size 0. The function should only be called if the size of the label
is greater 0.
With this fix TPM2_RSA_Encrypt/Decrypt did work with OpenSSL 1.1 and 3.0
for encryption without label.
Signed-off-by: Juergen Repp <juergen.repp@sit.fraunhofer.de>
When decryption fails then return TPM_RC_VALUE rather than TPM_RC_FAILURE.
The old error code could indicate to an application or driver that
something is wrong with the TPM (has possibly gone into failure mode) even
though only the decryption failed, possibly due to a wrong key.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
The JSON returned by TPM2_GetInfo contains a leading zero in the level.
$> swtpm_ioctl --tcp :10000 --info 1
{"TPMSpecification":{"family":"2.0","level":00,"revision":162+0}}
This patch fixes this to:
$> swtpm_ioctl --tcp :10000 --info 1
{"TPMSpecification":{"family":"2.0","level":0,"revision":162+0}}
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Fedora Rawhide and CentOS 10 do not support OpenSSL engine anymore.
Therefore, replace include of engine.h with err.h since the engine is not
needed anyway but we only need the prototype of ERR_get_error_line_data.
Add missing includes for evp.h and rsa.h.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Do not set the RSA_FLAG_NO_BLINDING on RSA keys anymore. The BoringSSL
documentation states the following:
"RSA_FLAG_NO_BLINDING disables blinding of private operations, which is
a dangerous thing to do. It is deprecated and should not be used. It will
be ignored whenever possible."
So, follow this documentation. OpenSSL does not seem to say much about it.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Check the secret size against the size of the buffer, not the size
member that has not been set yet.
Reported by Coverity.
Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
To avoid timeouts on short-running commands, such as TPM2_PCR_Extend,
avoid triggering the writing of the permanent state of the TPM 2
if only the clock was updated. So the clock by itself will not cause
the permanent state to be written out anymore but there have to be
other reasons as well.
The state will still be written out upon a TPM2_Shutdown, which is
supposed to be the last command to be sent to the TPM when shutting
down the VM/vTPM. Also, the permanent state will still carry the
latest clock value if it is retrieved via control channel for
VM/VTPM suspend.
The case that may be affected, but is of lesser importance, is the one
where swtpm's volatile state is written to storage using 'swtpm_ioctl -v'
and then swtpm is terminated and restarted (similar to suspend/resume)
and the permanent state file is read from storage but does not contain
the latest clock value. In this case the go.clock will be updated when
the first command after resume is executed.
This fixes the swtpm issue https://github.com/stefanberger/swtpm/issues/597.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
OpenSSL 3.0 has changed the signature of EVP_PKEY_get0_RSA() from
struct rsa_st *EVP_PKEY_get0_RSA(EVP_PKEY *pkey);
to
const struct rsa_st *EVP_PKEY_get0_RSA(const EVP_PKEY *pkey);
We now have to use EVP_PKEY_get1_RSA with this signature so that we can
access the RSA key. The signature of that function hasn't changed between
OpenSSL 1.1.0 and 3.0.0.
struct rsa_st *EVP_PKEY_get1_RSA(EVP_PKEY *pkey);
Free the additional reference held on the RSA key.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Event sequence objects were never properly marshalled and when their state
was saved and later restored their state may have been corrupted. Fix this
now by also marshalling the state of event sequence objects.
Bump up the version of the HASH_OBJECT's header to '3' so that previously
written state can be resumed if an event sequence object is encountered
and we only unmarshal an event sequence object when the version is at least
'3'.
Fixes issue #259.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
The NVRAM entries in s_indexOrderlyRam array do not need to contain a
0-sized terminating node. Instead, the entries may fill up this 512
byte array so that no NV_RAM_HEADER structure fits anymore. The fact
that no more NV_RAM_HEADER structure fits is also an indicator for the
last entry. We need to account for this in the code marshalling and
unmarshalling the entries so that we stop marshalling the entries
then and similarly stop unmarshalling.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Initialize a whole OBJECT before using it. This is necessary since
an OBJECT may also be used as a HASH_OBJECT via the ANY_OBJECT
union and that HASH_OBJECT can leave bad size inidicators in TPM2B
buffer in the OBJECT. To get rid of this problem we reset the whole
OBJECT to 0 before using it. This is as if the memory for the
OBJECT was just initialized.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Restore the original value of the memory location where data from
a stream was unmarshalled and the unmarshalled value was found to
be illegal. The goal is to not keep illegal values in memory.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Add maxSize parameter to TPM2B_Marshal and assert on it checking
the size of the data intended to be marshaled versus the maximum
buffer size.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Reset the buffer size indicator in a TPM2B type of buffer after it failed
the test for the maximum buffer size it allows. This prevents having bad
buffer sizes in memory that can come to haunt us when writing the volatile
state for example.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Do not call BLOCK_SKIP_READ once rc has been set to any error value.
Therefore, surround all occurrences of BLOCK_SKIP_READ() with tests
of 'rc'.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
The TPM is supposed to provide the output IV in the ivInOut parameter in
CryptSymmetricEncrypt. In the case of using the openssl routines, the
output IV is missed, and the resulting output from the TPM is in the
input IV.
OpenSSL unfortunately does not export EVP_CIPHER_CTX_iv() until
tags/OpenSSL_1_1_0, so we have to fall back to the reference code for
previous OpenSSL versions.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
This patch fixes a VM suspend/resume issue when externally loaded public
keys are involved.
This patch addresses the bug reported in issue #195 where the saving of
an externally loaded public key's context doesn't work due to the usage of
ANY_CONTEXT_SAVE for saving key contexts. This patch fixes the issue by
creating local versions of TPM_SENSITIVE_Marshal/_Unmarshal that deals
with the case where sensitiveType is not a type of private key but a
public key instead that basically doesn't have much information in
TPM_SENSITIVE but is all zeros instead.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Use TPM_PrintFourLimit in those cases where less than 4 bytes of
valid data may exist. Hashes, nonces, encrypted data, and others
typically have more than 4 bytes but data read from NVRAM or
to be encrypted data may have less.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Sanitize some of the values read from the TPM state stream.
All Coverity discoveries seem to be false positives.
Coverity doesn't like to see array_size being used in the loop even
though it was compared against ARRAY_SIZE() before. We solve this by
using ARRAY_SIZE() as the loop limit now rather than array size.
Compare seed.b.size against PRIMARY_SEED_SIZE even though this is
already being done in TPM2B_Unmarshal().
The num_bytes parameter is sanitized via a comparison involving a
sum over a sum of values, but Coverity doesn't seem to detect this.
Then we have to use it as a loop limit. I don't see another way.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Prevent a potential buffer overrun by checking that EVP_DecryptUpdate()
has not overrun the buffer it was passed in, so this overrun should
never occurr unless EVP_DecryptUpdate() was wrong. Also the pAssert above
it should have taken care of it already.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Coverity complains that the *output* variable passed to
AES_set_encrypt_key contains uninitialized bytes, so we initialize
the variables now.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Coverity complains that nrh may not be initialize when copying nrh.size
from it into the buffer pointer to by nrhp. So resolve this by clearing
nrh at the beginning of the loop and checking 'rc' after the Unmarshal.
Previously we could have copied an uninitialized nrh.size but would have
propagated the rc error code from UINT32_Unmarshal(), so this fix doesn't
really change anything.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
This patch adds a description to the buggy RSAAdjustPrimeCandidate
implementation that can only be properly fixed after 0.7.x.
Reported-by: Nicolas Iooss <nicolas.iooss@ledger.fr>
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
This patch ensures that the leading zeros in the b parameter for NIST P521
are being kept so that HLK accepts the returned parameters from
TPM2_ECC_Parameters. Now 66 bytes are reported for 'b' rather than only 65.
Do the same for the 'a' parameter, though that one was properly reported
already because it didn't have any leading zeros.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
This patch addresses issue #177 by fixing some typos and error
reporting inconsistencies (how structures are spelled) in NVMarhsal.c.
Reported-by: Nicolas Iooss <nicolas.iooss@ledger.fr>
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
cppcheck has detected the following issues in 2 functions. However,
neither one of the out-of-bounds array access can happen with the
existing code (see comments in patch).
src/tpm2/Session.c:399:5: note: After for loop, slotIndex has value 3
for(slotIndex = 0; slotIndex < MAX_LOADED_SESSIONS; slotIndex++)
^
src/tpm2/Session.c:414:15: note: Assuming condition is false
if(result != TPM_RC_SUCCESS)
^
src/tpm2/Session.c:419:15: note: Array index out of bounds
s_sessions[slotIndex].occupied = TRUE;
^
src/tpm2/Session.c:591:27: error: Array 's_sessions[3]' accessed at index 3, which is out of bounds. [arrayIndexOutOfBounds]
MemoryCopy(&s_sessions[slotIndex].session, session, sizeof(SESSION));
^
src/tpm2/Session.c:571:5: note: After for loop, slotIndex has value 3
for(slotIndex = 0; slotIndex < MAX_LOADED_SESSIONS; slotIndex++)
^
src/tpm2/Session.c:581:8: note: Assuming condition is false
&& contextIndex != s_oldestSavedSession)
^
src/tpm2/Session.c:591:27: note: Array index out of bounds
MemoryCopy(&s_sessions[slotIndex].session, session, sizeof(SESSION));
^
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
UBSAN detects possibly misaligned address when reading out of the
TPM 2's NVRAM and when writing back into it. The NV_RAM_HEADER may
be unaligned like this:
tests/test_tpm2_save_load_state_3.log:tpm2/Marshal.c:117:29: \
runtime error: load of misaligned address 0x7ffcb53b3bca for type 'UINT32', which requires 4 byte alignment
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Make sure that the value of bnK is not short so that the subsequent
BnEccModMult() runs in constant time. We take the same approach as with
the modifications to BnEccGenerateKeyPair() where we request bnK to have
all bytes set (no leading zeros that will be cut away) in case the order
of the curve is as byte boundary. In the other cases we add the order
to bnK, which creates bnK1, which we then use for BnEccModMult's scalar
parameter.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Suggested-by: Charanjit Jutla <csjutla@us.ibm.com>
Reviewed-by: Charanjit Jutla <csjutla@us.ibm.com>
Tested-by: Stefan Berger <stefanb@linux.ibm.com>
To avoid a potential side channel in the EcSchnorr signing algorithm,
enforce that the OpenSSL-generated bnD does not have leading zeros
that may then cause a timing side channel in the BnEccModMult() operation.
We modified BnEccGenerateKeyPair() so it calls BnEccModMult with a scalar
of constant number of bytes (for a particular curve):
In this version of BnEccGenerateKeyPair we take a dual approach to constant
time requirements: For curves whose order is at the byte boundary, e.g.
NIST P224/P256/P384, we make sure that bnD has all bytes set (no leading zeros)
so that OpenSSL BIGNUM code will not reduce the number of bytes and the
subsequent BnEccModMult() would run faster for a shoter value. For all other
curves whose order is not at the byte boundary, e.g. NIST P521, we simply
always add the order to bnD and call BnEccModMult() with the result bnD1,
which leads to the same result.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Suggested-by: Charanjit Jutla <csjutla@us.ibm.com>
Reviewed-by: Charanjit Jutla <csjutla@us.ibm.com>
Tested-by: Stefan Berger <stefanb@linux.ibm.com>
This patch fixes the following compilation error on Fedora 32 / s390x:
tpm2/Marshal.c: In function 'TPM2B_CREATION_DATA_Marshal':
tpm2/Marshal.c:95:19: error: 'sizePtr' may be used uninitialized in this function [-Werror=maybe-uninitialized]
95 | (*buffer)[0] = (BYTE)((*source >> 8) & 0xff);
| ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
tpm2/Marshal.c:2201:11: note: 'sizePtr' was declared here
2201 | BYTE *sizePtr;
| ^~~~~~~
The error is a false positive since sizePtr will have been initialized if
UINT16_Marshal() is called.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
tpm12/tpm_nvram.c: In function 'TPM_Process_NVWriteValue':
tpm12/tpm_nvram.c:2313:45: error: 'd1NvdataSensitive' may be used uninitialized in this function [-Werror=maybe-uninitialized]
2313 | if ((d1NvdataSensitive->pubInfo.permission.attributes & TPM_NV_PER_WRITEALL) &&
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~
This compiler error is a false positive since the above statement is inside
this if clause:
if ((returnCode == TPM_SUCCESS) && !done && !dir) {
However, if d1NvdataSensitive was not set then returnCode is
either != TPM_SUCCESS OR
- case index0 = FALSE : dir = TRUE per line 2106 OR
- case index0 = TRUE (nvIndex = 0): done = TRUE per line 2215.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Fix PCRBelongsTCBGroup by adjusting the set of PCRs that belong to the TCB
Group. The effect of this is that PCR changes to PCR 16 (for example) do
not change the pcrUpdateCounter anymore. The effect *should not* have any
negative side effects when using the TPM.
We also need to update the test cases that now show a different
pcrUpdateCounter in the responses. Also 'swtpm' test cases need
to be fixed to expect the changed result.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
In corner cases where the size of the salt and the size of the hash
to sign + 2 exceed the signature size we cannot use the salt length =
hash length but have to resort to using the maximum possible salt
length.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Call EVP_PKEY_CTX_set_rsa_pss_saltlen(ctx, -1) when creating an RSA
signature to set the PSS salt length to the digest length. Without
this call we previously set the salt length to the maximum
permissible value, but this is not how TPM 2 implements it.
Per interoperability testing between signatures created previously
with the max. permissible value and the new code, which does not
modify the signature verification code, old signatures still verify.
New signatures also verify.
This patch may solve interoperability with hardware TPMs that signatures
created following this patch now verify on hardware TPMs as well.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Always use a temporary buffer large enough to meet the requirements of the
EVP_DecryptUpdate() call.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
