Fix memory leak if legacy test is skipped.
Using EVP_KDF_CTX_get_params() to get OSSL_KDF_PARAM_SIZE will now
return 0 if the returned size is 0.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15977)
Passing the return value from gmtime() directly to mktime() was producing
incorrect results under windows (but not under wine) when built with mingw
32-bit (but not VC-WIN32). We implement a workaround for this.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15939)
This gives better diagnostic output
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15939)
There were 4 classes of failure:
- line ending problems;
- unicode problems;
- file path munging problems; and
- a "hang" in test_cmp_http.
The unicode problems appear to be somewhere between wine or msys - they
don't actually appear to be a problem with the built binaries. We just skip
those tests for now.
Fixes#13558
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15939)
When creating a signed S/MIME message using SMIME_write_CMS()
if the reading from the bio fails, the state is therefore
still ASN1_STATE_START when BIO_flush() is called by i2d_ASN1_bio_stream().
This results in calling asn1_bio_flush_ex cleanup but will only
reset retry flags as the state is not ASN1_STATE_POST_COPY.
Therefore 48 bytes (Linux x86_64) leaked since the
ndef_prefix_free / ndef_suffix_free callbacks are not executed
and the ndef_aux structure is not freed.
By always calling free function callback in asn1_bio_free() the
memory leak is fixed.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14844)
Fixes#15963
INSTALL.md uses these exact options as an example so it should work.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15965)
The engine is modifying memory without the sanitiser realising. By pre-
initialising this memory, the sanitiser now thinks that read accesses are okay.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/15952)
48f1739600 did not convert the RSA OAEP
tests correctly. The corrupted ciphertext and truncation tests were
really decrypting uninitialized memory, rather than the sample
ciphertext. This results in an error in tools like MSan.
The test is somewhat roundabout. In the original version, before the
conversion, ctext_ex was an OAEP test vector from key1(), etc.,
functions. The test would:
1. Encrypt ptext_ex as ctext.
2. Decrypt ctext and check it gives ptext_ex.
3. Decrypt ctext_ex and check it gives ptext_ex.
4. Try corrupted and truncated versions of ctext.
48f1739600 then moved steps 1 and 2 into
test_rsa_simple, which meant ctext is no longer available for step 4. It
then mistakenly left the variable around, but uninitialized, so the test
wasn't testing anything. (Confusingly, test_rsa_simple outputs ctext_ex
to the caller, but doesn't do anything with it. The ctext_ex output is
also only usable for OAEP, not PKCS#1 v1.5.)
It doesn't really matter whether we use ctext or ctext_ex for step 4, so
this PR fixes it by using ctext_ex instead.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15950)
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15790)
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15790)
However, when passing it through the OPENSSL_CONFIG environment
variable, we still need the quotes, just to make sure.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15823)
Some tests are designed to test UTF-8 on the command line.
We simply disable those on VMS.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15823)
Occasionally, we get an error code on VMS that doesn't translate
into POSIX, and the error string reflects that
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15823)
The providers have to call up_ref to keep the cbio pointer, just like
the internal bio_prov.c does.
OSSL_STORE_attach passes a cbio pointer to the provider and then calls
ossl_core_bio_free(cbio). If up_ref is not called, the cbio gets
freed way too early.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15163)
Previously, when locks were held while calling a provider init function,
then RAND_bytes_ex() would fail if called from the init function and
used in conjunction with a child lib ctx. We add an explicit test of that.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15854)
These 2 functions have become so close to each other that they may as well
be just one function.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15854)
Update use_fallbacks to zero when we add a provider to the store rather
than when we activate it. Its only at the point that we add it to the store
that it is actually usable and visible to other threads.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15854)
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15874)
Using File::Temp::tempfile() is admirable, but isn't necessary for the
sort of thing we use it for.
Furthermore, since tempfile() returns an opened file handle for
reading for the file in question, it may have effect that the file
becomes unwritable. This is the default on VMS, and since tempfile()
doesn't seem to have any option to affect this, it means that
test/shlibloadtest.c can't write the magic line to that file.
Also, if we consider forensics, to be able to see what a test produced
to determine what went wrong, it's better to use specific and known
file names.
Therefore, this test is modified to use well known file names, and to
open them for reading after the shlibloadtest program has been run
instead of before.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15872)
OpenSSL::Test::cmd() should be used with caution, as it is for special
cases only.
It's preferable to use OpenSSL::Test::app() or OpenSSL::Test::test().
Fixes#15833
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15846)
app_malloc() terminates execution if the allocation fails. The tests implement
their own app_malloc() in an attempt to reduce the amount of code pulled in.
This version also needs to terminate on failed allocation. The alternative
would be adding failed allocation checks pervasively throughout the apps's
commands.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15836)
This is a request from the lab that changes the AES_GCM test back to perform both a encrypt and
decrypt. (This makes no logical sense since this is not an inverse cipher).
I have left the AES_ECB decrypt test in (although it may not be needed)
since it is actually testing the inverse cipher case.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15844)
- Use SSL_set_ciphersuites for TLS 1.3 tests instead of using
SSL_set_cipher_list.
- Don't bother passing a sequence number size to KTLS test functions.
These functions always test TLS (and not DTLS) for which the
sequence size is always the same. In addition, even for DTLS the
check in question (verifying that the sequence number fields in SSL
do not change) should still pass when doing a before/after
comparison of the field.
- Define a helper structure to hold the TLS version and cipher name
for a single KTLS test.
- Define an array of such structures with valid KTLS ciphers and move
#ifdef's for TLS versions and supported ciphers out of test
functions and instead use them to define the valid members of this
array. This also permits using TLS 1.3 cipher suite names for
TLS 1.3 tests.
- Use separate tests per cipher for test_ktls to give more
fine-grained pass/fail results as is already done for
test_ktls_sendfile.
- While here, rename test_ktls_sendfile to execute_test_ktls_sendfile
and test_ktls_sendfile_anytls to test_ktls_sendfile. This is more
consistent with the naming used for test_ktls as well as other tests
in this file.
- Close the file descriptors used for temporary sockets in ktls tests.
- Don't assume that KTLS is supported for all compile-time supported
cipher suites at runtime. If the kernel fails to offload a given
cipher suite, skip the test rather than failing it. FreeBSD kernels
may not offload all of the cipher suites supported by its KTLS if a
suitable driver or KTLS backend is not present.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15814)
