This commit addresses multiple side-channel vulnerabilities present
during RSA key validation.
Private key parameters are re-computed using variable-time functions.
This issue was discovered and reported by the NISEC group at TAU Finland.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9779)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9639)
(cherry picked from commit c70e2ec33943d3bd46d3d9950f774307feda832b)
This will never be the case for 1.1.1 so removed.
Fixes: comment 1 of #9757
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9762)
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/9734)
(cherry picked from commit 46a9cc9451213039fd53f62733b2ccd04e853bb2)
This commit addresses a side-channel vulnerability present when
PVK and MSBLOB key formats are loaded into OpenSSL.
The public key was not computed using a constant-time exponentiation
function.
This issue was discovered and reported by the NISEC group at TAU Finland.
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9587)
(cherry picked from commit 724339ff44)
There is a problem in the rand_unix.c code when the random seed fd is greater
than or equal to FD_SETSIZE and the FDSET overruns its limit and walks the
stack.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/9686)
(cherry picked from commit e1f8584d47)
Improve handling of low entropy at start up from /dev/urandom by waiting for
a read(2) call on /dev/random to succeed. Once one such call has succeeded,
a shared memory segment is created and persisted as an indicator to other
processes that /dev/urandom is properly seeded.
This does not fully prevent against attacks weakening the entropy source.
An attacker who has control of the machine early in its boot sequence
could create the shared memory segment preventing detection of low entropy
conditions. However, this is no worse than the current situation.
An attacker would also be capable of removing the shared memory segment
and causing seeding to reoccur resulting in a denial of service attack.
This is partially mitigated by keeping the shared memory alive for the
duration of the process's existence. Thus, an attacker would not only need
to have called call shmctl(2) with the IPC_RMID command but the system
must subsequently enter a state where no instances of libcrypto exist in
any process. Even one long running process will prevent this attack.
The System V shared memory calls used here go back at least as far as
Linux kernel 2.0. Linux kernels 4.8 and later, don't have a reliable way
to detect that /dev/urandom has been properly seeded, so a failure is raised
for this case (i.e. the getentropy(2) call has already failed).
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/9595)
[manual merge]
Requesting zero bytes from shake previously led to out-of-bounds write
on some platforms.
Signed-off-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9433)
(cherry picked from commit a890ef833d)
When OpenSSL is configured with 'no-stdio', TEST_ENG_OPENSSL_RC4_P_INIT
shouldn't be defined, as that test uses stdio.
Fixes#9597
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9598)
(cherry picked from commit 9f643f5423)
Fix a few places where calling ossl_isdigit does the wrong thing on
EBCDIC based systems.
Replaced with ascii_isdigit.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9556)
(cherry picked from commit 48102247ff)
We should not retry on EAI_MEMORY as that error is most probably
fatal and not depending on AI_ADDRCONFIG hint.
Also report the error from the first call if the second call fails
as that one would be most probably the more interesting one.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9535)
(cherry picked from commit 91cb81d40a)
Do not try to discern the error return value on
getaddrinfo() failure but when retrying set the AI_NUMERICHOST
to avoid DNS lookups.
Fixes: #9053
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9535)
(cherry picked from commit 7f616a00e9)
A default digest of SHA256 was being returned for RSA PSS even if the
PSS parameters indicated a different digest must be used. We change this
so that the correct default digest is returned and additionally mark this
as mandatory for PSS.
This bug had an impact on sig alg selection in libssl. Due to this issue
an incorrect sig alg might be selected in the event that a server is
configured with an RSA-PSS cert with parameter restrictions.
Fixes#9545
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9553)
(cherry picked from commit 9bcc9f973b)
Note a flag needed to be added since some ssl tests fail if they output any error
(even if the error is ignored). Only ciphers that handle the GET_IV_LEN control set this flag.
Fixes#8330
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9499)
CLA: trivial
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/9295)
Fix: crypto\whrlpool\wp_block.c(90) : warning C4164: '_rotl64' : intrinsic function not declared.
Fixes#9487
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9488)
(cherry picked from commit 0c789f59f1)
The rand pool support allocates maximal sized buffers -- this is typically
12288 bytes in size. These pools are allocated in secure memory which is a
scarse resource. They are also allocated per DRBG of which there are up to two
per thread.
This change allocates 64 byte pools and grows them dynamically if required.
64 is chosen to be sufficiently large so that pools do not normally need to
grow.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/9428)
(cherry picked from commit a6a66e4511)
The additional data allocates 12K per DRBG instance in the
secure memory, which is not necessary. Also nonces are not
considered secret.
[extended tests]
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9424)
The check is redundant, because <openssl/x509v3.h> is included.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/9365)
This include guard inside an object file comes as a surprise and
serves no purpose anymore. It seems like this object file was
included by crypto/threads/mttest.c at some time, but the include
directive was removed in commit bb8abd6.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/9365)
These weren't available in Cygwin at the time our DSO code was
written, but things have changed since.
Fixes#9385
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9402)
(cherry picked from commit 38f6f99cdf)
Cosmetic changes to use the X509_STORE_lock/unlock functions.
Renamed some ctx variables to store.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9366)
(cherry picked from commit 7a9abccde7)
x509 store's objects cache can get corrupted when using dir lookup
method in multithreaded application. Claim x509 store's lock when
accessing objects cache.
CLA: trivial
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9326)
(cherry picked from commit a161738a70)
Modified rev to rev64, because rev only takes integer registers.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90827
Otherwise, the following error will occur.
Error: operand 1 must be an integer register -- `rev v31.16b,v31.16b'
CLA: trivial
Signed-off-by: Lei Maohui <leimaohui@cn.fujitsu.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9151)
(cherry picked from commit 7b0fceed21)
Happens when trying to generate 4 or 5 bit safe primes.
[extended tests]
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9311)
(cherry picked from commit 291f616ced)
BOOLEAN does not have valid data in the value.ptr member,
thus don't use it here.
Fixes#9276
[extended tests]
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9278)
(cherry picked from commit 6335f837cf)
The maximum key length for rc5 is 2040 bits so we should not attempt to
use keys longer than this.
Issue found by OSS-Fuzz and Guido Vranken.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8834)
(cherry picked from commit 792cb4ee8d)
CLA: trivial
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/9275)
This feature is enabled by default outside of FIPS builds
which ban such actions completely.
Encryption is always disallowed and will generate an error.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9112)
(cherry picked from commit 2c840201e5)
This is a bit annoying, if for instance "openssl genrsa -aes128"
tries to read a 4+ character size password, but CTRL-C does no longer
work after a RETURN key, since the flag UI_FLAG_REDOABLE is set by
UI_set_result_ex, together with the error "You must type in 4 to 1023 characters".
Thus remove the REDOABLE flag to allow CTRL-C to work.
[extended tests]
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9170)
(cherry picked from commit f8922b5107)
CLA: trivial
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9149)
The BIO_FLAGS_NONCLEAR_RST flag behavior was not properly documented
and it also caused the length to be incorrectly set after the reset
operation.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9179)
(cherry picked from commit 8b7b32921e)
When bufsize == 0, openssl_strerror_r should return 0 (if _GNU_SOURCE is defined),
to be consistent with non-_GNU_SOURCE variants, which exhibit the same behavior.
Fix a few cases, where the return value of openssl_strerror_r was ignored.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9163)
(cherry picked from commit e7a4682d0b)
This avoids the case where a UEFI build on FreeBSD tries to call the system
issetugid function instead of returning 0 as it should do.
CLA: trivial
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9158)
When compiling with --strict-warnings using gcc 7.4.0 the compiler
complains that a case falls through, even though there is an explicit
comment stating this. Moving the comment outside of the conditional
compilation section resolves this.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9131)
(cherry picked from commit a2e520447e)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9123)
(cherry picked from commit e98e586b31)
The lookup for ::1 with getaddrinfo() might return error even if
the ::1 would work if AI_ADDRCONFIG flag is used.
Fixes: #9053
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9108)
(cherry picked from commit 3f91ede9ae)
The DEVRANDOM_WAIT feature added a select() call to wait for the
`/dev/random` device to become readable before reading from the
`/dev/urandom` device. It was introduced in commit 38023b87f0
in order to mitigate the fact that the `/dev/urandom` device
does not block until the initial seeding of the kernel CSPRNG
has completed, contrary to the behaviour of the `getrandom()`
system call.
It turned out that this change had negative side effects on
performance which were not acceptable. After some discussion it
was decided to revert this feature and leave it up to the OS
resp. the platform maintainer to ensure a proper initialization
during early boot time.
Fixes#9078
This partially reverts commit 38023b87f0.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(cherry picked from commit a08714e181)
(Merged from https://github.com/openssl/openssl/pull/9118)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/9101)
(cherry picked from commit bab6046146)
The 4 kB SPACE_SYS_STR_REASONS in crypto/err/err.c isn't enough for some locales.
The Russian locales consume 6856 bytes, Ukrainian even 7000.
build_SYS_str_reasons() contains an overflow check:
if (cnt > sizeof(strerror_pool))
cnt = sizeof(strerror_pool);
But since commit 9f15e5b911 it no longer
works as cnt is incremented once more after the condition.
cnt greater than sizeof(strerror_pool) results in an unbounded
OPENSSL_strlcpy() in openssl_strerror_r(), eventually causing a crash.
When the first received error string was empty or contained only
spaces, cur would move in front of the start of the strerror_pool.
Also don't call openssl_strerror_r when the pool is full.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/8966)
(cherry picked from commit fac9200a88)
Fixes#8923
Found using the openssl cms -resign option.
This uses an alternate path to do the signing which was not adding the required signed attribute
content type. The content type attribute should always exist since it is required is there are
any signed attributes.
As the signing time attribute is always added in code, the content type attribute is also required.
The CMS_si_check_attributes() method adds validity checks for signed and unsigned attributes
e.g. The message digest attribute is a signed attribute that must exist if any signed attributes
exist, it cannot be an unsigned attribute and there must only be one instance containing a single
value.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8944)
(cherry picked from commit 19e512a824)
openssl_config_int() returns the uninitialized variable `ret`
when compiled with OPENSSL_SYS_UEFI.
Fixes#9026
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9029)
(cherry picked from commit f4a96507fb)
The #7408 implemented mandatory digest checking in TLS.
However this broke compatibility of DSS support with GnuTLS
which supports only SHA1 with DSS.
There is no reason why SHA256 would be a mandatory digest
for DSA as other digests in SHA family can be used as well.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9015)
(cherry picked from commit cd4c83b524)
Add a few coverage test case.
Fixes#8949
[extended tests]
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8959)
(cherry picked from commit 5b3accde60)
67c81ec311 forgot about s390x
Signed-off-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8971)
(cherry picked from commit 887e22dd8b)
This change allows to pass the authentication tag after specifying
the AAD in CCM mode. This is already true for the other two supported
AEAD modes (GCM and OCB) and it seems appropriate to match the
behavior.
GCM and OCB also support to set the tag at any point before the call
to `EVP_*Final`, but this won't work for CCM due to a restriction
imposed by section 2.6 of RFC3610: The tag must be set before
actually decrypting data.
This commit also adds a test case for setting the tag after supplying
plaintext length and AAD.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7243)
(cherry picked from commit 67c81ec311)
Return error if the output tag buffer size doesn't match
the tag size exactly. This prevents the caller from
using that portion of the tag buffer that remains
uninitialized after an otherwise succesfull call to
CRYPTO_ccm128_tag.
Bug found by OSS-Fuzz.
Fix suggested by Kurt Roeckx.
Signed-off-by: Guido Vranken <guidovranken@gmail.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8810)
(cherry picked from commit 514c9da48b)
This happens in ec_key_simple_check_key and EC_GROUP_check.
Since the the group order is not a secret scalar, it is
unnecessary to use coordinate blinding.
Fixes: #8731
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8734)
(cherry picked from commit 3051bf2afa)
Even with custome ciphers, the combination in == NULL && inl == 0
should not be passed down to the backend cipher function. The reason
is that these are the values passed by EVP_*Final, and some of the
backend cipher functions do check for these to see if a "final" call
is made.
Fixes#8675
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8676)
(cherry picked from commit dcb982d792)
'no-dso' is meaningless, as it doesn't get any macro defined.
Therefore, we remove all checks of OPENSSL_NO_DSO. However, there may
be some odd platforms with no DSO scheme. For those, we generate the
internal macro DSO_NONE aand use it.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8622)
If using a custom X509_LOOKUP_METHOD then calls to
X509_STORE_CTX_get_by_subject may crash due to an incorrectly initialised
X509_OBJECT being passed to the callback get_by_subject function.
Fixes#8673
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8698)
(cherry picked from commit b926f9deb3)
It was assumed that the config functionality returned a boolean.
However, it may return a negative number on error, so we need to take
that into account.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8679)
(cherry picked from commit e3af453bac)
This prevents failure of openssl s_server socket binding to wildcard
address on hosts with disabled IPv6.
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8550)
(cherry picked from commit b8472b4e67)
I turns out that this made crypto/rand/rand_win.c to never build with
BCrypt support unless the user sets _WIN32_WINNT. That wasn't the
intent.
This reverts commit cc8926ec8f.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8641)
(cherry picked from commit 705a27f7e0)
Revert win32_pathbyaddr() which is used in DSO_dsobyaddr().
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8596)
(cherry picked from commit 9c98aa354d)
Replace it with InitializeCriticalSection()
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8596)
(cherry picked from commit 09305a7d0a)
DH_check is used to test the validity of Diffie-Hellman parameter sets (p, q, g). Among the tests performed are primality tests on p and q, for this BN_is_prime_ex is called with the rounds of Miller-Rabin set as default. This will therefore use the average case error estimates derived from the function BN_prime_checks_for_size based on the bit size of the number tested.
However, these bounds are only accurate on testing random input. Within this testing scenario, where we are checking the validity of a DH parameter set, we can not assert that these parameters are randomly generated. Thus we must treat them as if they are adversarial in nature and increase the rounds of Miller-Rabin performed.
Generally, each round of Miller-Rabin can declare a composite number prime with probability at most (1/4), thus 64 rounds is sufficient in thwarting known generation techniques (even in safe prime settings - see https://eprint.iacr.org/2019/032 for full analysis). The choice of 64 rounds is also consistent with SRP_NUMBER_ITERATIONS_FOR_PRIME 64 as used in srp_Verify_N_and_g in openssl/apps/s_client.c.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8593)
(cherry picked from commit 2500c093aa)
We treat that as automatic success. Other EVP_*Update functions already do
this (e.g. EVP_EncryptUpdate, EVP_DecryptUpdate etc). EVP_EncodeUpdate is
a bit of an anomoly. That treats 0 byte input length as an error.
Fixes#8576
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8587)
(cherry picked from commit a8274ea351)
constant time with a memory access pattern that does not depend
on secret information.
[extended tests]
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8543)
(cherry picked from commit 9c0cf214e7)
Great effort has been made to make initialization more configurable.
However, the behavior of OPENSSL_config() was lost in the process,
having it suddenly generate errors it didn't previously, which is not
how it's documented to behave.
A simple setting of default flags fixes this problem.
Fixes#8528
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8533)
(cherry picked from commit 905c9a72a7)
There are some compiling errors for mips32r6 and mips64r6:
crypto/bn/bn-mips.S:56: Error: opcode not supported on this processor: mips2 (mips2) `mulu $1,$12,$7'
crypto/mips_arch.h: Assembler messages:
crypto/mips_arch.h:15: Error: junk at end of line, first unrecognized character is `&'
Signed-off-by: Hua Zhang <hua.zhang1974@hotmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8464)
(cherry picked from commit 1b9c5f2e2f)
The secret point R can be recovered from S using the equation R = S - P.
The X and Z coordinates should be sufficient for that.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8504)
(cherry picked from commit 8a74bb5c7b)
There are two copy-paste errors in handling CTR mode. When dealing
with a 2 or 3 block tail, the code branches to the CBC decryption exit
path, rather than to the CTR exit path.
This can lead to data corruption: in the Linux kernel we have a copy
of this file, and the bug leads to corruption of the IV, which leads
to data corruption when we call the encryption function again later to
encrypt subsequent blocks.
Originally reported to the Linux kernel by Ondrej Mosnáček <omosnacek@gmail.com>
CLA: trivial
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8510)
(cherry picked from commit f643deac41)
The function felem_diff_128_64 in ecp_nistp521.c substracts the number |in|
from |out| mod p. In order to avoid underflow it first adds 32p mod p
(which is equivalent to 0 mod p) to |out|. The comments and variable naming
suggest that the original author intended to add 64p mod p. In fact it
has been shown that with certain unusual co-ordinates it is possible to
cause an underflow in this function when only adding 32p mod p while
performing a point double operation. By changing this to 64p mod p the
underflow is avoided.
It turns out to be quite difficult to construct points that satisfy the
underflow criteria although this has been done and the underflow
demonstrated. However none of these points are actually on the curve.
Finding points that satisfy the underflow criteria and are also *on* the
curve is considered significantly more difficult. For this reason we do
not believe that this issue is currently practically exploitable and
therefore no CVE has been assigned.
This only impacts builds using the enable-ec_nistp_64_gcc_128 Configure
option.
With thanks to Bo-Yin Yang, Billy Brumley and Dr Liu for their significant
help in investigating this issue.
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/8405)
(cherry picked from commit 13fbce17fc)
ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input for
every encryption operation. RFC 7539 specifies that the nonce value (IV)
should be 96 bits (12 bytes). OpenSSL allows a variable nonce length and
front pads the nonce with 0 bytes if it is less than 12 bytes. However it
also incorrectly allows a nonce to be set of up to 16 bytes. In this case
only the last 12 bytes are significant and any additional leading bytes are
ignored.
It is a requirement of using this cipher that nonce values are unique.
Messages encrypted using a reused nonce value are susceptible to serious
confidentiality and integrity attacks. If an application changes the
default nonce length to be longer than 12 bytes and then makes a change to
the leading bytes of the nonce expecting the new value to be a new unique
nonce then such an application could inadvertently encrypt messages with a
reused nonce.
Additionally the ignored bytes in a long nonce are not covered by the
integrity guarantee of this cipher. Any application that relies on the
integrity of these ignored leading bytes of a long nonce may be further
affected.
Any OpenSSL internal use of this cipher, including in SSL/TLS, is safe
because no such use sets such a long nonce value. However user
applications that use this cipher directly and set a non-default nonce
length to be longer than 12 bytes may be vulnerable.
CVE-2019-1543
Fixes#8345
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8406)
(cherry picked from commit 2a3d0ee9d5)
GNU strerror_r may return either a pointer to a string that the function
stores in buf, or a pointer to some (immutable) static string in which case
buf is unused.
In such a case we need to set buf manually.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8371)
(cherry picked from commit e3b35d2b29)
Use select to wait for /dev/random in readable state,
but do not actually read anything from /dev/random,
use /dev/urandom first.
Use linux define __NR_getrandom instead of the
glibc define SYS_getrandom, in case the kernel headers
are more current than the glibc headers.
Fixes#8215
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/8251)
(cherry picked from commit 38023b87f0)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8281)
(cherry picked from commit 54d00677f3)
Currently SM2 shares the ameth with EC, so the current default digest
algorithm returned is SHA256. This fixes the default digest algorithm of
SM2 to SM3, which is the only valid digest algorithm for SM2 signature.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8186)
(cherry picked from commit e766f4a053)
This restores the behavior of previous versions of the /dev/crypto
engine, in alignment with the default implementation.
Reported-by: Gerard Looije <lglooije@hotmail.com>
Signed-off-by: Eneas U de Queiroz <cote2004-github@yahoo.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/8306)
cipher_init may be called on an already initialized context, without a
necessary cleanup. This separates cleanup from initialization, closing
an eventual open session before creating a new one.
Move the /dev/crypto session cleanup code to its own function.
Signed-off-by: Eneas U de Queiroz <cote2004-github@yahoo.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/8306)
The real cause for this change is that test/ec_internal_test.c
includes ec_lcl.h, and including curve448/curve448_lcl.h from there
doesn't work so well with compilers who always do inclusions relative
to the C file being compiled.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8334)
Thanks to David Benjamin who reported this, performed the analysis and
suggested the patch. I have incorporated some of his analysis in the
comments below.
This issue can cause an out-of-bounds read. It is believed that this was
not reachable until the recent "fixed top" changes. Analysis has so far
only identified one code path that can encounter this - although it is
possible that others may be found. The one code path only impacts 1.0.2 in
certain builds. The fuzzer found a path in RSA where iqmp is too large. If
the input is all zeros, the RSA CRT logic will multiply a padded zero by
iqmp. Two mitigating factors:
- Private keys which trip this are invalid (iqmp is not reduced mod p).
Only systems which take untrusted private keys care.
- In OpenSSL 1.1.x, there is a check which rejects the oversize iqmp,
so the bug is only reproducible in 1.0.2 so far.
Fortunately, the bug appears to be relatively harmless. The consequences of
bn_cmp_word's misbehavior are:
- OpenSSL may crash if the buffers are page-aligned and the previous page is
non-existent.
- OpenSSL will incorrectly treat two BN_ULONG buffers as not equal when they
are equal.
- Side channel concerns.
The first is indeed a concern and is a DoS bug. The second is fine in this
context. bn_cmp_word and bn_cmp_part_words are used to compute abs(a0 - a1)
in Karatsuba. If a0 = a1, it does not matter whether we use a0 - a1 or
a1 - a0. The third would be worth thinking about, but it is overshadowed
by the entire Karatsuba implementation not being constant time.
Due to the difficulty of tripping this and the low impact no CVE is felt
necessary for this issue.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8326)
(cherry picked from commit 576129cd72)
This commit adds a dedicated function in `EC_METHOD` to access a modular
field inversion implementation suitable for the specifics of the
implemented curve, featuring SCA countermeasures.
The new pointer is defined as:
`int (*field_inv)(const EC_GROUP*, BIGNUM *r, const BIGNUM *a, BN_CTX*)`
and computes the multiplicative inverse of `a` in the underlying field,
storing the result in `r`.
Three implementations are included, each including specific SCA
countermeasures:
- `ec_GFp_simple_field_inv()`, featuring SCA hardening through
blinding.
- `ec_GFp_mont_field_inv()`, featuring SCA hardening through Fermat's
Little Theorem (FLT) inversion.
- `ec_GF2m_simple_field_inv()`, that uses `BN_GF2m_mod_inv()` which
already features SCA hardening through blinding.
From a security point of view, this also helps addressing a leakage
previously affecting conversions from projective to affine coordinates.
This commit also adds a new error reason code (i.e.,
`EC_R_CANNOT_INVERT`) to improve consistency between the three
implementations as all of them could fail for the same reason but
through different code paths resulting in inconsistent error stack
states.
Co-authored-by: Nicola Tuveri <nic.tuv@gmail.com>
(cherry picked from commit e0033efc30)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/8262)
Cygwin binaries should not enforce text mode these days, just
use text mode if the underlying mount point requests it
Signed-off-by: Corinna Vinschen <vinschen@redhat.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8248)
(cherry picked from commit 9b57e4a1ef)
The add/double shortcut in ecp_nistz256-x86_64.pl left one instruction
point that did not unwind, and the "slow" path in AES_cbc_encrypt was
not annotated correctly. For the latter, add
.cfi_{remember,restore}_state support to perlasm.
Next, fill in a bunch of functions that are missing no-op .cfi_startproc
and .cfi_endproc blocks. libunwind cannot unwind those stack frames
otherwise.
Finally, work around a bug in libunwind by not encoding rflags. (rflags
isn't a callee-saved register, so there's not much need to annotate it
anyway.)
These were found as part of ABI testing work in BoringSSL.
Reviewed-by: Richard Levitte <levitte@openssl.org>
GH: #8109
(cherry picked from commit c0e8e5007b)
CLA: trivial
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Yang <yang.yang@baishancloud.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8137)
(cherry picked from commit b754a8a159)
ARMv8.3 adds pointer authentication extension, which in this case allows
to ensure that, when offloaded to stack, return address is same at return
as at entry to the subroutine. The new instructions are nops on processors
that don't implement the extension, so that the vetification is backward
compatible.
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8205)
(cherry picked from commit 9a18aae5f2)
If the old openssl versions not supporting the .include directive
load a config file with it, they will bail out with error.
This change allows using the .include = <filename> syntax which
is interpreted as variable assignment by the old openssl
config file parser.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8141)
(cherry picked from commit 9d5560331d)
o2i_ECPublicKey() requires an EC_KEY structure filled with an EC_GROUP.
o2i_ECPublicKey() is called by d2i_PublicKey(). In order to fulfill the
o2i_ECPublicKey()'s requirement, d2i_PublicKey() needs to be called with
an EVP_PKEY with an EC_KEY containing an EC_GROUP.
However, the call to EVP_PKEY_set_type() frees any existing key structure
inside the EVP_PKEY, thus freeing the EC_KEY with the EC_GROUP that
o2i_ECPublicKey() needs.
This means you can't d2i_PublicKey() for an EC key...
The fix is to check to see if the type is already set appropriately, and
if so, not call EVP_PKEY_set_type().
Reviewed-by: Paul Yang <yang.yang@baishancloud.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8168)
(cherry picked from commit 2aa2beb06c)
Trim trailing whitespace. It doesn't match OpenSSL coding standards,
AFAICT, and it can cause problems with git tooling.
Trailing whitespace remains in test data and external source.
Backport-of: https://github.com/openssl/openssl/pull/8092
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8134)
If the call the ERR_set_error_data() in ERR_add_error_vdata() fails then
a mem leak can occur. This commit checks that we successfully added the
error data, and if not frees the buffer.
Fixes#8085
Reviewed-by: Paul Yang <yang.yang@baishancloud.com>
(Merged from https://github.com/openssl/openssl/pull/8105)
(cherry picked from commit fa6b1ee111)
When the ret parameter is NULL the generated prime
is in rnd variable and not in ret.
CLA: trivial
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8076)
(cherry picked from commit 3d43f9c809)
If this fails try out if mfspr268 works.
Use OPENSSL_ppccap=0x20 for enabling mftb,
OPENSSL_ppccap=0x40 for enabling mfspr268,
and OPENSSL_ppccap=0 for enabling neither.
Fixes#8012
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8043)
(cherry picked from commit c8f370485c)
If you use a BIO and set up your own buffer that is not freed, the
memory bio will leak the BIO_BUF_MEM object it allocates.
The trouble is that the BIO_BUF_MEM is allocated and kept around,
but it is not freed if BIO_NOCLOSE is set.
The freeing of BIO_BUF_MEM was fairly confusing, simplify things
so mem_buf_free only frees the memory buffer and free the BIO_BUF_MEM
in mem_free(), where it should be done.
Alse add a test for a leak in the memory bio
Setting a memory buffer caused a leak.
Signed-off-by: Corey Minyard <minyard@acm.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8051)
(cherry picked from commit c6048af23c)
Having a weak getauxval() and only depending on GNU C without looking
at the library we build against meant that it got picked up where not
really expected.
So we change this to check for the glibc version, and since we know it
exists from that version, there's no real need to make it weak.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/8028)
(cherry picked from commit 5f40dd158c)
It turns out that AT_SECURE may be defined through other means than
our inclusion of sys/auxv.h, so to be on the safe side, we define our
own guard and use that to determine if getauxval() should be used or
not.
Fixes#7932
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7933)
(cherry picked from commit aefb980c45)
