openssl/doc/man7/provider-keymgmt.pod

398 lines
15 KiB
Plaintext
Raw Normal View History

=pod
=head1 NAME
provider-keymgmt - The KEYMGMT library E<lt>-E<gt> provider functions
=head1 SYNOPSIS
#include <openssl/core_numbers.h>
/*
* None of these are actual functions, but are displayed like this for
* the function signatures for functions that are offered as function
* pointers in OSSL_DISPATCH arrays.
*/
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
/* Key object (keydata) creation and destruction */
void *OP_keymgmt_new(void *provctx);
void OP_keymgmt_free(void *keydata);
void *OP_keymgmt_gen_init(void *provctx, int selection);
int OP_keymgmt_gen_set_template(void *genctx, void *template);
int OP_keymgmt_gen_set_params(void *genctx, const OSSL_PARAM params[]);
const OSSL_PARAM *OP_keymgmt_gen_settable_params(void *provctx);
void *OP_keymgmt_gen(void *genctx, OSSL_CALLBACK *cb, void *cbarg);
void OP_keymgmt_gen_cleanup(void *genctx);
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
/* Key object information */
int OP_keymgmt_get_params(void *keydata, OSSL_PARAM params[]);
const OSSL_PARAM *OP_keymgmt_gettable_params(void);
int OP_keymgmt_set_params(void *keydata, const OSSL_PARAM params[]);
const OSSL_PARAM *OP_keymgmt_settable_params(void);
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
/* Key object content checks */
int OP_keymgmt_has(void *keydata, int selection);
int OP_keymgmt_match(const void *keydata1, const void *keydata2,
int selection);
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
/* Discovery of supported operations */
const char *OP_keymgmt_query_operation_name(int operation_id);
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
/* Key object import and export functions */
int OP_keymgmt_import(int selection, void *keydata, const OSSL_PARAM params[]);
const OSSL_PARAM *OP_keymgmt_import_types(int selection);
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
int OP_keymgmt_export(int selection, void *keydata,
OSSL_CALLBACK *param_cb, void *cbarg);
const OSSL_PARAM *OP_keymgmt_export_types(int selection);
/* Key object copy */
int OP_keymgmt_copy(void *keydata_to, const void *keydata_from, int selection);
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
/* Key object validation */
int OP_keymgmt_validate(void *keydata, int selection);
=head1 DESCRIPTION
The KEYMGMT operation doesn't have much public visibility in OpenSSL
libraries, it's rather an internal operation that's designed to work
in tandem with operations that use private/public key pairs.
Because the KEYMGMT operation shares knowledge with the operations it
works with in tandem, they must belong to the same provider.
The OpenSSL libraries will ensure that they do.
The primary responsibility of the KEYMGMT operation is to hold the
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
provider side key data for the OpenSSL library EVP_PKEY structure.
All "functions" mentioned here are passed as function pointers between
F<libcrypto> and the provider in B<OSSL_DISPATCH> arrays via
B<OSSL_ALGORITHM> arrays that are returned by the provider's
provider_query_operation() function
(see L<provider-base(7)/Provider Functions>).
All these "functions" have a corresponding function type definition
named B<OSSL_{name}_fn>, and a helper function to retrieve the
function pointer from a B<OSSL_DISPATCH> element named
B<OSSL_get_{name}>.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
For example, the "function" OP_keymgmt_new() has these:
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
typedef void *(OSSL_OP_keymgmt_new_fn)(void *provctx);
static ossl_inline OSSL_OP_keymgmt_new_fn
OSSL_get_OP_keymgmt_new(const OSSL_DISPATCH *opf);
B<OSSL_DISPATCH> arrays are indexed by numbers that are provided as
macros in L<openssl-core_numbers.h(7)>, as follows:
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
OP_keymgmt_new OSSL_FUNC_KEYMGMT_NEW
OP_keymgmt_free OSSL_FUNC_KEYMGMT_FREE
OP_keymgmt_gen_init OSSL_FUNC_KEYMGMT_GEN_INIT
OP_keymgmt_gen_set_template OSSL_FUNC_KEYMGMT_GEN_SET_TEMPLATE
OP_keymgmt_gen_set_params OSSL_FUNC_KEYMGMT_GEN_SET_PARAMS
OP_keymgmt_gen_settable_params OSSL_FUNC_KEYMGMT_GEN_SETTABLE_PARAMS
OP_keymgmt_gen OSSL_FUNC_KEYMGMT_GEN
OP_keymgmt_gen_cleanup OSSL_FUNC_KEYMGMT_GEN_CLEANUP
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
OP_keymgmt_get_params OSSL_FUNC_KEYMGMT_GET_PARAMS
OP_keymgmt_gettable_params OSSL_FUNC_KEYMGMT_GETTABLE_PARAMS
OP_keymgmt_set_params OSSL_FUNC_KEYMGMT_SET_PARAMS
OP_keymgmt_settable_params OSSL_FUNC_KEYMGMT_SETTABLE_PARAMS
OP_keymgmt_query_operation_name OSSL_FUNC_KEYMGMT_QUERY_OPERATION_NAME
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
OP_keymgmt_has OSSL_FUNC_KEYMGMT_HAS
OP_keymgmt_validate OSSL_FUNC_KEYMGMT_VALIDATE
OP_keymgmt_match OSSL_FUNC_KEYMGMT_MATCH
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
OP_keymgmt_import OSSL_FUNC_KEYMGMT_IMPORT
OP_keymgmt_import_types OSSL_FUNC_KEYMGMT_IMPORT_TYPES
OP_keymgmt_export OSSL_FUNC_KEYMGMT_EXPORT
OP_keymgmt_export_types OSSL_FUNC_KEYMGMT_EXPORT_TYPES
OP_keymgmt_copy OSSL_FUNC_KEYMGMT_COPY
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=head2 Key Objects
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
A key object is a collection of data for an asymmetric key, and is
represented as I<keydata> in this manual.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
The exact contents of a key object are defined by the provider, and it
is assumed that different operations in one and the same provider use
the exact same structure to represent this collection of data, so that
for example, a key object that has been created using the KEYMGMT
interface that we document here can be passed as is to other provider
operations, such as OP_signature_sign_init() (see
L<provider-signature(7)>).
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
With some of the KEYMGMT functions, it's possible to select a specific
subset of data to handle, governed by the bits in a I<selection>
indicator. The bits are:
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=over 4
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=item B<OSSL_KEYMGMT_SELECT_PRIVATE_KEY>
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
Indicating that the private key data in a key object should be
considered.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=item B<OSSL_KEYMGMT_SELECT_PUBLIC_KEY>
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
Indicating that the public key data in a key object should be
considered.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=item B<OSSL_KEYMGMT_SELECT_DOMAIN_PARAMETERS>
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
Indicating that the domain parameters in a key object should be
considered.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=item B<OSSL_KEYMGMT_SELECT_OTHER_PARAMETERS>
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
Indicating that other parameters in a key object should be
considered.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
Other parameters are key parameters that don't fit any other
classification. In other words, this particular selector bit works as
a last resort bit bucket selector.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=back
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
Some selector bits have also been combined for easier use:
=over 4
=item B<OSSL_KEYMGMT_SELECT_ALL_PARAMETERS>
Indicating that all key object parameters should be considered,
regardless of their more granular classification.
=for comment This should used by EVP functions such as
EVP_PKEY_copy_parameters() and EVP_PKEY_cmp_parameters()
This is a combination of B<OSSL_KEYMGMT_SELECT_DOMAIN_PARAMETERS> and
B<OSSL_KEYMGMT_SELECT_OTHER_PARAMETERS>.
=for comment If more parameter categories are added, they should be
mentioned here too.
=item B<OSSL_KEYMGMT_SELECT_KEYPAIR>
Indicating that both the whole key pair in a key object should be
considered, i.e. the combination of public and private key.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
This is a combination of B<OSSL_KEYMGMT_SELECT_PRIVATE_KEY> and
B<OSSL_KEYMGMT_SELECT_PUBLIC_KEY>.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=item B<OSSL_KEYMGMT_SELECT_ALL>
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
Indicating that everything in a key object should be considered.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=back
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
The exact interpretation of those bits or how they combine is left to
each function where you can specify a selector.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=for comment One might think that a combination of bits means that all
the selected data subsets must be considered, but then you have to
consider that when comparing key objects (future function), an
implementation might opt to not compare the private key if it has
compared the public key, since a match of one half implies a match of
the other half.
=head2 Constructing and Destructing Functions
OP_keymgmt_new() should create a provider side key object. The
provider context I<provctx> is passed and may be incorporated in the
key object, but that is not mandatory.
OP_keymgmt_free() should free the passed I<keydata>.
OP_keymgmt_gen_init(), OP_keymgmt_gen_set_template(),
OP_keymgmt_gen_set_params(), OP_keymgmt_gen_settable_params(),
OP_keymgmt_gen() and OP_keymgmt_gen_cleanup() work together as a more
elaborate context based key object constructor.
OP_keymgmt_gen_init() should create the key object generation context
and initialize it with I<selections>, which will determine what kind
of contents the key object to be generated should get.
OP_keymgmt_gen_set_template() should add I<template> to the context
I<genctx>. The I<template> is assumed to be a key object constructed
with the same KEYMGMT, and from which content that the implementation
chooses can be used as a template for the key object to be generated.
Typically, the generation of a DSA or DH key would get the domain
parameters from this I<template>.
OP_keymgmt_gen_set_params() should set additional parameters from
I<params> in the key object generation context I<genctx>.
OP_keymgmt_gen_settable_params() should return a constant array of
descriptor B<OSSL_PARAM>, for parameters that OP_keymgmt_gen_set_params()
can handle.
OP_keymgmt_gen() should perform the key object generation itself, and
return the result. The callback I<cb> should be called at regular
intervals with indications on how the key object generation
progresses.
OP_keymgmt_gen_cleanup() should clean up and free the key object
generation context I<genctx>
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
At least one of OP_keymgmt_new() and OP_keymgmt_gen() are mandatory,
as well as OP_keymgmt_free(). Additionally, if OP_keymgmt_gen() is
present, OP_keymgmt_gen_init() and OP_keymgmt_gen_cleanup() must be
present as well.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=head2 Key Object Information Functions
OP_keymgmt_get_params() should extract information data associated
with the given I<keydata>, see L</Common Information Parameters>.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
OP_keymgmt_gettable_params() should return a constant array of
descriptor B<OSSL_PARAM>, for parameters that OP_keymgmt_get_params()
can handle.
If OP_keymgmt_gettable_params() is present, OP_keymgmt_get_params()
must also be present, and vice versa.
OP_keymgmt_set_params() should update information data associated
with the given I<keydata>, see L</Common Information Parameters>.
OP_keymgmt_settable_params() should return a constant array of
descriptor B<OSSL_PARAM>, for parameters that OP_keymgmt_set_params()
can handle.
If OP_keymgmt_settable_params() is present, OP_keymgmt_set_params()
must also be present, and vice versa.
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
=head2 Key Object Checking Functions
OP_keymgmt_query_operation_name() should return the name of the
supported algorithm for the operation I<operation_id>. This is
similar to provider_query_operation() (see L<provider-base(7)>),
but only works as an advisory. If this function is not present, or
returns NULL, the caller is free to assume that there's an algorithm
from the same provider, of the same name as the one used to fetch the
keymgmt and try to use that.
OP_keymgmt_has() should check whether the given I<keydata> contains the subsets
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
of data indicated by the I<selector>. A combination of several
selector bits must consider all those subsets, not just one. An
implementation is, however, free to consider an empty subset of data
to still be a valid subset.
OP_keymgmt_validate() should check if the I<keydata> contains valid
data subsets indicated by I<selection>. Some combined selections of
data subsets may cause validation of the combined data.
For example, the combination of B<OSSL_KEYMGMT_SELECT_PRIVATE_KEY> and
B<OSSL_KEYMGMT_SELECT_PUBLIC_KEY> (or B<OSSL_KEYMGMT_SELECT_KEYPAIR>
for short) is expected to check that the pairwise consistency of
I<keydata> is valid.
OP_keymgmt_match() should check if the data subset indicated by
I<selection> in I<keydata1> and I<keydata2> match. It is assumed that
the caller has ensured that I<keydata1> and I<keydata2> are both owned
by the implementation of this function.
=head2 Key Object Import, Export and Copy Functions
Redesign the KEYMGMT libcrypto <-> provider interface - the basics The KEYMGMT libcrypto <-> provider interface currently makes a few assumptions: 1. provider side domain parameters and key data isn't mutable. In other words, as soon as a key has been created in any (loaded, imported data, ...), it's set in stone. 2. provider side domain parameters can be strictly separated from the key data. This does work for the most part, but there are places where that's a bit too rigid for the functionality that the EVP_PKEY API delivers. Key data needs to be mutable to allow the flexibility that functions like EVP_PKEY_copy_parameters promise, as well as to provide the combinations of data that an EVP_PKEY is generally assumed to be able to hold: - domain parameters only - public key only - public key + private key - domain parameters + public key - domain parameters + public key + private key To remedy all this, we: 1. let go of the distinction between domain parameters and key material proper in the libcrypto <-> provider interface. As a consequence, functions that still need it gain a selection argument, which is a set of bits that indicate what parts of the key object are to be considered in a specific call. This allows a reduction of very similar functions into one. 2. Rework the libcrypto <-> provider interface so provider side key objects are created and destructed with a separate function, and get their data filled and extracted in through import and export. (future work will see other key object constructors and other functions to fill them with data) Fixes #10979 squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics Remedy 1 needs a rewrite: Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Shane Lontis <shane.lontis@oracle.com> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/11006)
2020-02-03 01:56:07 +08:00
OP_keymgmt_import() should import data indicated by I<selection> into
I<keydata> with values taken from the B<OSSL_PARAM> array I<params>.
OP_keymgmt_export() should extract values indicated by I<selection>
from I<keydata>, create an B<OSSL_PARAM> array with them and call
I<param_cb> with that array as well as the given I<cbarg>.
OP_keymgmt_import_types() should return a constant array of descriptor
B<OSSL_PARAM> for data indicated by I<selection>, for parameters that
OP_keymgmt_import() can handle.
OP_keymgmt_export_types() should return a constant array of descriptor
B<OSSL_PARAM> for data indicated by I<selection>, that the
OP_keymgmt_export() callback can expect to receive.
OP_keymgmt_copy() should copy data subsets indicated by I<selection>
from I<keydata_from> to I<keydata_to>. It is assumed that the caller
has ensured that I<keydata_to> and I<keydata_from> are both owned by
the implementation of this function.
=head2 Common Information Parameters
See L<OSSL_PARAM(3)> for further details on the parameters structure.
Common information parameters currently recognised by all built-in
keymgmt algorithms are as follows:
=over 4
=item "bits" (B<OSSL_PKEY_PARAM_BITS>) <integer>
The value should be the cryptographic length of the cryptosystem to
which the key belongs, in bits. The definition of cryptographic
length is specific to the key cryptosystem.
=item "max-size" (B<OSSL_PKEY_PARAM_MAX_SIZE>) <integer>
The value should be the maximum size that a caller should allocate to
safely store a signature (called I<sig> in L<provider-signature(7)>),
the result of asymmmetric encryption / decryption (I<out> in
L<provider-asym_cipher(7)>, a derived secret (I<secret> in
L<provider-keyexch(7)>, and similar data).
Because an EVP_KEYMGMT method is always tightly bound to another method
(signature, asymmetric cipher, key exchange, ...) and must be of the
same provider, this number only needs to be synchronised with the
dimensions handled in the rest of the same provider.
=item "security-bits" (B<OSSL_PKEY_PARAM_SECURITY_BITS>) <integer>
The value should be the number of security bits of the given key.
Bits of security is defined in SP800-57.
=back
=head1 RETURN VALUES
OP_keymgmt_new() should return a valid reference to the newly created provider
side key object, or NULL on failure.
OP_keymgmt_import(), OP_keymgmt_export(), OP_keymgmt_get_params() and
OP_keymgmt_set_params() should return 1 for success or 0 on error.
OP_keymgmt_validate() should return 1 on successful validation, or 0 on
failure.
OP_keymgmt_has() should return 1 if all the selected data subsets are contained
in the given I<keydata> or 0 otherwise.
OP_keymgmt_query_operation_name() should return a pointer to a string matching
the requested operation, or NULL if the same name used to fetch the keymgmt
applies.
OP_keymgmt_gettable_params() and OP_keymgmt_settable_params()
OP_keymgmt_import_types(), OP_keymgmt_export_types()
should
always return a constant B<OSSL_PARAM> array.
=head1 SEE ALSO
L<provider(7)>,
L<EVP_PKEY-X25519(7)>, L<EVP_PKEY-X448(7)>, L<EVP_PKEY-ED25519(7)>,
L<EVP_PKEY-ED448(7)>, L<EVP_PKEY-EC(7)>, L<EVP_PKEY-RSA(7)>,
L<EVP_PKEY-DSA(7)>, L<EVP_PKEY-DH(7)>
=head1 HISTORY
The KEYMGMT interface was introduced in OpenSSL 3.0.
=head1 COPYRIGHT
Copyright 2019-2020 The OpenSSL Project Authors. All Rights Reserved.
Licensed under the Apache License 2.0 (the "License"). You may not use
this file except in compliance with the License. You can obtain a copy
in the file LICENSE in the source distribution or at
L<https://www.openssl.org/source/license.html>.
=cut