A provider could be linked against a different version of libcrypto than
the version of libcrypto that loaded the provider. Different versions of
libcrypto could define opaque types differently. It must never occur that
a type created in one libcrypto is used directly by the other libcrypto.
This will cause crashes.
We can "cheat" for "built-in" providers that are part of libcrypto itself,
because we know that the two libcrypto versions are the same - but not for
other providers.
To ensure this does not occur we use different types names for the handful
of opaque types that are passed between the core and providers.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11758)
The provider context structure is made to include the following information:
- The core provider handle (first argument to the provider init
function). This handle is meant to be used in all upcalls that need
it.
- A library context, used for any libcrypto calls that need it, done in
the provider itself.
Regarding the library context, that's generally only needed if the
provider makes any libcrypto calls, i.e. is linked with libcrypto. That
happens to be the case for all OpenSSL providers, but is applicable for
other providers that use libcrypto internally as well.
The normal thing to do for a provider init function is to create its own
library context. For a provider that's meant to become a dynamically
loadable module, this is what MUST be done.
However, we do not do that in the default provider; it uses the library
context associated with the core provider handle instead. This is
permissible, although generally discouraged, as long as the provider in
question is guaranteed to be built-in, into libcrypto or into the
application that uses it.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11803)
