Run more threads and load the legacy provider (which uses a child lib ctx)
in order to hit more possible thread failures.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
Not holding the flag lock when creating/removing child providers can
confuse the activation counts if the parent provider is loaded/unloaded
at the same time.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
A "find" operation on a stack can end up sorting the underlying stack. In
this case it is necessary to use a "write" lock to synchronise access to
the stack across multiple threads.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
If during OSSL_PROVIDER_try_load() we attempt to load a provider, but
adding to the store gives back a different provider, then we need to
ensure this different provider has its activation count increased.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
We should stop receiving child callbacks if we're about to free up
the child libctx. Otherwise we can get callbacks when the libctx is half
freed up.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
A provider may have been activated, but failed when being added to
the store. At this point we still need to deactivate it.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
We leave it to the caller to confirm that the provider does not exist
in the store. If it does exist then later adding it to the store will
fail.
It is possible that the provider could be added to the store in
between the caller checking, and the caller calling ossl_provider_new.
We leave it to the caller to properly handle the failure when it
attempts to add the provider to the store. This is simpler than
having ossl_provider_new try to handle it.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
This variable might have made sense at some point but it not longer does
so. It was being used to check whether we are still initing or not. If we
are still initing then the assumption was that we already hold the lock.
That assumption was untrue. We need to always take the lock.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
init_thread_stop() is called when a thread is stopping. It calls all
the callbacks that need to know about the demise of this thread. However,
the list of callbacks is also available globally and may be updated by
other threads so we need to make sure we use the right lock.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
If a provider doesn't have any child providers then there is no need
to attempt to remove them - so we should not do so. This removes some
potentialy thread races.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
These global variables were previously overwritten with the same value
every time we created a new OSSL_LIB_CTX. Instead we preinitialise them
with the correct values, and then confirm that settings for each
OSSL_LIB_CTX agree with the preinitialised values.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16980)
Negative return value indicates an error so we bail out.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16975)
Define a number of helper functions that ease the difficulty of detecting
integer overflows.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16930)
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/16342)
We currently load data byte by byte in order to byteswap it on big
endian. On little endian we can just do 8 byte loads.
A SHAKE128 benchmark runs 10% faster on POWER9 with this patch applied.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8455)
These are legacy of older versions of the code and are currently not used
anywhere.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16985)
In a recent upstream change
(43da9a14f0)
the parameter sigsize become a read/write input in
EVP_PKEY_sign(), and after signing, sigsize will be overwritten with
the actual size and used in the verify step. As the speed program
calls EVP_PKEY_sign() on the same context repeatedly, sigsize value is
no longer the initial available size, and may fail in later buffer
size checks.
This fix adds a new buflen member in struct loopargs (which is only
used within apps/speed.c), to save available buffer size and
to be used as sigsize input in EVP_PKEY_sign() calls.
Sigsize still contains the signature size for the verify step.
Signed-off-by: Xiaofei Bai <xiaofei.bai@arm.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16991)
The condition `userlen == -1` isn't possible because this is already checked
on line 159 above and the subsequent strlen(3) call guarantees that it's value
is positive.
Reviewed-by: Paul Yang <kaishen.yy@antfin.com>
(Merged from https://github.com/openssl/openssl/pull/16987)
The weekly build got lost when we stopped using Travis.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16988)
These DER encoder implementations are supposed to be aliases for the
"type-specific" output structure, but were made different in so far
that they would output a "type specific" public key, which turns out
to be garbage (it called i2o_ECPublicKey()). The "type-specific"
output structure doesn't support that, and shouldn't.
Fixes#16977
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16983)
(cherry picked from commit 2cb802e16f)
Signed-off-by: Tianjia Zhang <tianjia.zhang@linux.alibaba.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Yang <kaishen.yy@antfin.com>
(Merged from https://github.com/openssl/openssl/pull/16963)
The EVP_PKEY_fromdata man page has some code examples with various
errors in them. This fixes those errors.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16973)
The global variable `default_DSO_meth` was potentially set multiple times by
different threads. It turns out that it could only be set to a single value
so the race is harmless but still better avoided. The fix here simply removes
the global and accesses the value it was set to via the `DSO_METHOD_openssl()`
call.
Problem discovered via #16970, but this does not resolve that issue because
there are other concerns.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16972)
