openssl/providers/implementations/rands/drbg.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1042 lines
33 KiB
C
Raw Normal View History

/*
* Copyright 2011-2025 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
* https://www.openssl.org/source/license.html
*/
#include <string.h>
#include <openssl/crypto.h>
#include <openssl/err.h>
#include <openssl/rand.h>
#include <openssl/evp.h>
#include "crypto/rand.h"
#include <openssl/proverr.h>
#include "drbg_local.h"
#include "internal/thread_once.h"
#include "crypto/cryptlib.h"
#include "prov/seeding.h"
#include "crypto/rand_pool.h"
#include "prov/provider_ctx.h"
#include "prov/providercommon.h"
#include "crypto/context.h"
/*
* Support framework for NIST SP 800-90A DRBG
*
* See manual page PROV_DRBG(7) for a general overview.
*
* The OpenSSL model is to have new and free functions, and that new
* does all initialization. That is not the NIST model, which has
* instantiation and un-instantiate, and reuse within a new/free
* lifecycle. (No doubt this comes from the desire to support hardware
* DRBG, where allocation of resources on something like an HSM is
* a much bigger deal than just re-setting an allocated resource.)
*/
/* NIST SP 800-90A DRBG recommends the use of a personalization string. */
static const char ossl_pers_string[] = DRBG_DEFAULT_PERS_STRING;
static const OSSL_DISPATCH *find_call(const OSSL_DISPATCH *dispatch,
int function);
static int rand_drbg_restart(PROV_DRBG *drbg);
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
/*
* We interpret a call to this function as a hint only and ignore it. This
* occurs when the EVP layer thinks we should do some locking. In practice
* however we manage for ourselves when we take a lock or not on the basis
* of whether drbg->lock is present or not.
*/
int ossl_drbg_lock(void *vctx)
{
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
return 1;
}
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
/* Interpreted as a hint only and ignored as for ossl_drbg_lock() */
void ossl_drbg_unlock(void *vctx)
{
}
static int ossl_drbg_lock_parent(PROV_DRBG *drbg)
{
void *parent = drbg->parent;
if (parent != NULL
&& drbg->parent_lock != NULL
&& !drbg->parent_lock(parent)) {
ERR_raise(ERR_LIB_PROV, PROV_R_PARENT_LOCKING_NOT_ENABLED);
return 0;
}
return 1;
}
static void ossl_drbg_unlock_parent(PROV_DRBG *drbg)
{
void *parent = drbg->parent;
if (parent != NULL && drbg->parent_unlock != NULL)
drbg->parent_unlock(parent);
}
static int get_parent_strength(PROV_DRBG *drbg, unsigned int *str)
{
OSSL_PARAM params[2] = { OSSL_PARAM_END, OSSL_PARAM_END };
void *parent = drbg->parent;
int res;
if (drbg->parent_get_ctx_params == NULL) {
ERR_raise(ERR_LIB_PROV, PROV_R_UNABLE_TO_GET_PARENT_STRENGTH);
return 0;
}
*params = OSSL_PARAM_construct_uint(OSSL_RAND_PARAM_STRENGTH, str);
if (!ossl_drbg_lock_parent(drbg)) {
ERR_raise(ERR_LIB_PROV, PROV_R_UNABLE_TO_LOCK_PARENT);
return 0;
}
res = drbg->parent_get_ctx_params(parent, params);
ossl_drbg_unlock_parent(drbg);
if (!res) {
ERR_raise(ERR_LIB_PROV, PROV_R_UNABLE_TO_GET_PARENT_STRENGTH);
return 0;
}
return 1;
}
static unsigned int get_parent_reseed_count(PROV_DRBG *drbg)
{
OSSL_PARAM params[2] = { OSSL_PARAM_END, OSSL_PARAM_END };
void *parent = drbg->parent;
unsigned int r = 0;
*params = OSSL_PARAM_construct_uint(OSSL_DRBG_PARAM_RESEED_COUNTER, &r);
if (!ossl_drbg_lock_parent(drbg)) {
ERR_raise(ERR_LIB_PROV, PROV_R_UNABLE_TO_LOCK_PARENT);
goto err;
}
if (!drbg->parent_get_ctx_params(parent, params))
r = 0;
ossl_drbg_unlock_parent(drbg);
return r;
err:
r = tsan_load(&drbg->reseed_counter) - 2;
if (r == 0)
r = UINT_MAX;
return r;
}
/*
* Implements the get_entropy() callback
*
* If the DRBG has a parent, then the required amount of entropy input
* is fetched using the parent's ossl_prov_drbg_generate().
*
* Otherwise, the entropy is polled from the system entropy sources
* using ossl_pool_acquire_entropy().
*
* If a random pool has been added to the DRBG using RAND_add(), then
* its entropy will be used up first.
*/
size_t ossl_drbg_get_seed(void *vdrbg, unsigned char **pout,
int entropy, size_t min_len,
size_t max_len, int prediction_resistance,
const unsigned char *adin, size_t adin_len)
{
PROV_DRBG *drbg = (PROV_DRBG *)vdrbg;
size_t bytes_needed;
unsigned char *buffer;
/* Figure out how many bytes we need */
bytes_needed = entropy >= 0 ? (entropy + 7) / 8 : 0;
if (bytes_needed < min_len)
bytes_needed = min_len;
if (bytes_needed > max_len)
bytes_needed = max_len;
/* Allocate storage */
buffer = OPENSSL_secure_malloc(bytes_needed);
if (buffer == NULL)
return 0;
/*
* Get random data. Include our DRBG address as
* additional input, in order to provide a distinction between
* different DRBG child instances.
*
* Note: using the sizeof() operator on a pointer triggers
* a warning in some static code analyzers, but it's
* intentional and correct here.
*/
if (!ossl_prov_drbg_generate(drbg, buffer, bytes_needed,
drbg->strength, prediction_resistance,
(unsigned char *)&drbg, sizeof(drbg))) {
OPENSSL_secure_clear_free(buffer, bytes_needed);
ERR_raise(ERR_LIB_PROV, PROV_R_GENERATE_ERROR);
return 0;
}
*pout = buffer;
return bytes_needed;
}
/* Implements the cleanup_entropy() callback */
void ossl_drbg_clear_seed(ossl_unused void *vdrbg,
unsigned char *out, size_t outlen)
{
OPENSSL_secure_clear_free(out, outlen);
}
static size_t get_entropy(PROV_DRBG *drbg, unsigned char **pout, int entropy,
size_t min_len, size_t max_len,
int prediction_resistance)
{
size_t bytes;
unsigned int p_str;
if (drbg->parent == NULL)
/*
* In normal use (i.e. OpenSSL's own uses), this is never called.
* This remains purely for legacy reasons.
*/
return ossl_prov_get_entropy(drbg->provctx, pout, entropy, min_len,
max_len);
if (drbg->parent_get_seed == NULL) {
ERR_raise(ERR_LIB_PROV, PROV_R_PARENT_CANNOT_SUPPLY_ENTROPY_SEED);
return 0;
}
if (!get_parent_strength(drbg, &p_str))
return 0;
if (drbg->strength > p_str) {
/*
* We currently don't support the algorithm from NIST SP 800-90C
* 10.1.2 to use a weaker DRBG as source
*/
ERR_raise(ERR_LIB_PROV, PROV_R_PARENT_STRENGTH_TOO_WEAK);
return 0;
}
/*
* Our lock is already held, but we need to lock our parent before
* generating bits from it. Note: taking the lock will be a no-op
* if locking is not required (while drbg->parent->lock == NULL).
*/
if (!ossl_drbg_lock_parent(drbg))
return 0;
/*
* Get random data from parent. Include our DRBG address as
* additional input, in order to provide a distinction between
* different DRBG child instances.
*
* Note: using the sizeof() operator on a pointer triggers
* a warning in some static code analyzers, but it's
* intentional and correct here.
*/
bytes = drbg->parent_get_seed(drbg->parent, pout,
entropy > 0 ? entropy : (int) drbg->strength,
min_len, max_len, prediction_resistance,
(unsigned char *)&drbg, sizeof(drbg));
ossl_drbg_unlock_parent(drbg);
return bytes;
}
static void cleanup_entropy(PROV_DRBG *drbg, unsigned char *out, size_t outlen)
{
if (drbg->parent == NULL) {
ossl_prov_cleanup_entropy(drbg->provctx, out, outlen);
} else if (drbg->parent_clear_seed != NULL) {
if (!ossl_drbg_lock_parent(drbg))
return;
drbg->parent_clear_seed(drbg->parent, out, outlen);
ossl_drbg_unlock_parent(drbg);
}
}
#ifndef PROV_RAND_GET_RANDOM_NONCE
typedef struct prov_drbg_nonce_global_st {
CRYPTO_RWLOCK *rand_nonce_lock;
int rand_nonce_count;
} PROV_DRBG_NONCE_GLOBAL;
/*
* drbg_ossl_ctx_new() calls drgb_setup() which calls rand_drbg_get_nonce()
* which needs to get the rand_nonce_lock out of the OSSL_LIB_CTX...but since
* drbg_ossl_ctx_new() hasn't finished running yet we need the rand_nonce_lock
* to be in a different global data object. Otherwise we will go into an
* infinite recursion loop.
*/
void *ossl_prov_drbg_nonce_ctx_new(OSSL_LIB_CTX *libctx)
{
PROV_DRBG_NONCE_GLOBAL *dngbl = OPENSSL_zalloc(sizeof(*dngbl));
if (dngbl == NULL)
return NULL;
dngbl->rand_nonce_lock = CRYPTO_THREAD_lock_new();
if (dngbl->rand_nonce_lock == NULL) {
OPENSSL_free(dngbl);
return NULL;
}
return dngbl;
}
void ossl_prov_drbg_nonce_ctx_free(void *vdngbl)
{
PROV_DRBG_NONCE_GLOBAL *dngbl = vdngbl;
if (dngbl == NULL)
return;
CRYPTO_THREAD_lock_free(dngbl->rand_nonce_lock);
OPENSSL_free(dngbl);
}
/* Get a nonce from the operating system */
static size_t prov_drbg_get_nonce(PROV_DRBG *drbg, unsigned char **pout,
size_t min_len, size_t max_len)
{
size_t ret = 0, n;
unsigned char *buf = NULL;
OSSL_LIB_CTX *libctx = ossl_prov_ctx_get0_libctx(drbg->provctx);
PROV_DRBG_NONCE_GLOBAL *dngbl
= ossl_lib_ctx_get_data(libctx, OSSL_LIB_CTX_DRBG_NONCE_INDEX);
struct {
void *drbg;
int count;
} data;
if (dngbl == NULL)
return 0;
if (drbg->parent != NULL && drbg->parent_nonce != NULL) {
n = drbg->parent_nonce(drbg->parent, NULL, 0, drbg->min_noncelen,
drbg->max_noncelen);
if (n > 0 && (buf = OPENSSL_malloc(n)) != NULL) {
ret = drbg->parent_nonce(drbg->parent, buf, 0,
drbg->min_noncelen, drbg->max_noncelen);
if (ret == n) {
*pout = buf;
return ret;
}
OPENSSL_free(buf);
}
}
/* Use the built in nonce source plus some of our specifics */
memset(&data, 0, sizeof(data));
data.drbg = drbg;
if (!CRYPTO_atomic_add(&dngbl->rand_nonce_count, 1, &data.count,
dngbl->rand_nonce_lock))
return 0;
return ossl_prov_get_nonce(drbg->provctx, pout, min_len, max_len,
&data, sizeof(data));
}
#endif /* PROV_RAND_GET_RANDOM_NONCE */
/*
* Instantiate |drbg|, after it has been initialized. Use |pers| and
* |perslen| as prediction-resistance input.
*
* Requires that drbg->lock is already locked for write, if non-null.
*
* Returns 1 on success, 0 on failure.
*/
int ossl_prov_drbg_instantiate(PROV_DRBG *drbg, unsigned int strength,
int prediction_resistance,
const unsigned char *pers, size_t perslen)
{
unsigned char *nonce = NULL, *entropy = NULL;
size_t noncelen = 0, entropylen = 0;
unsigned int min_entropy;
size_t min_entropylen, max_entropylen;
if (strength > drbg->strength) {
ERR_raise(ERR_LIB_PROV, PROV_R_INSUFFICIENT_DRBG_STRENGTH);
goto end;
}
min_entropy = drbg->strength;
min_entropylen = drbg->min_entropylen;
max_entropylen = drbg->max_entropylen;
if (pers == NULL) {
pers = (const unsigned char *)ossl_pers_string;
perslen = sizeof(ossl_pers_string);
}
if (perslen > drbg->max_perslen) {
ERR_raise(ERR_LIB_PROV, PROV_R_PERSONALISATION_STRING_TOO_LONG);
goto end;
}
if (drbg->state != EVP_RAND_STATE_UNINITIALISED) {
if (drbg->state == EVP_RAND_STATE_ERROR)
ERR_raise(ERR_LIB_PROV, PROV_R_IN_ERROR_STATE);
else
ERR_raise(ERR_LIB_PROV, PROV_R_ALREADY_INSTANTIATED);
goto end;
}
drbg->state = EVP_RAND_STATE_ERROR;
if (drbg->min_noncelen > 0) {
if (drbg->parent_nonce != NULL) {
noncelen = drbg->parent_nonce(drbg->parent, NULL, drbg->strength,
drbg->min_noncelen,
drbg->max_noncelen);
if (noncelen == 0) {
ERR_raise(ERR_LIB_PROV, PROV_R_ERROR_RETRIEVING_NONCE);
goto end;
}
nonce = OPENSSL_malloc(noncelen);
if (nonce == NULL) {
ERR_raise(ERR_LIB_PROV, PROV_R_ERROR_RETRIEVING_NONCE);
goto end;
}
if (noncelen != drbg->parent_nonce(drbg->parent, nonce,
drbg->strength,
drbg->min_noncelen,
drbg->max_noncelen)) {
ERR_raise(ERR_LIB_PROV, PROV_R_ERROR_RETRIEVING_NONCE);
goto end;
}
#ifndef PROV_RAND_GET_RANDOM_NONCE
} else if (drbg->parent != NULL) {
#endif
/*
* NIST SP800-90Ar1 section 9.1 says you can combine getting
* the entropy and nonce in 1 call by increasing the entropy
* with 50% and increasing the minimum length to accommodate
* the length of the nonce. We do this in case a nonce is
* required and there is no parental nonce capability.
*/
min_entropy += drbg->strength / 2;
min_entropylen += drbg->min_noncelen;
max_entropylen += drbg->max_noncelen;
}
#ifndef PROV_RAND_GET_RANDOM_NONCE
else { /* parent == NULL */
noncelen = prov_drbg_get_nonce(drbg, &nonce, drbg->min_noncelen,
drbg->max_noncelen);
if (noncelen < drbg->min_noncelen
|| noncelen > drbg->max_noncelen) {
ERR_raise(ERR_LIB_PROV, PROV_R_ERROR_RETRIEVING_NONCE);
goto end;
}
}
#endif
}
drbg->reseed_next_counter = tsan_load(&drbg->reseed_counter);
if (drbg->reseed_next_counter) {
drbg->reseed_next_counter++;
if (!drbg->reseed_next_counter)
drbg->reseed_next_counter = 1;
}
entropylen = get_entropy(drbg, &entropy, min_entropy,
min_entropylen, max_entropylen,
prediction_resistance);
if (entropylen < min_entropylen
|| entropylen > max_entropylen) {
ERR_raise(ERR_LIB_PROV, PROV_R_ERROR_RETRIEVING_ENTROPY);
goto end;
}
if (!drbg->instantiate(drbg, entropy, entropylen, nonce, noncelen,
pers, perslen)) {
cleanup_entropy(drbg, entropy, entropylen);
ERR_raise(ERR_LIB_PROV, PROV_R_ERROR_INSTANTIATING_DRBG);
goto end;
}
cleanup_entropy(drbg, entropy, entropylen);
drbg->state = EVP_RAND_STATE_READY;
drbg->generate_counter = 1;
drbg->reseed_time = time(NULL);
tsan_store(&drbg->reseed_counter, drbg->reseed_next_counter);
end:
if (nonce != NULL)
ossl_prov_cleanup_nonce(drbg->provctx, nonce, noncelen);
if (drbg->state == EVP_RAND_STATE_READY)
return 1;
return 0;
}
/*
* Uninstantiate |drbg|. Must be instantiated before it can be used.
*
* Requires that drbg->lock is already locked for write, if non-null.
*
* Returns 1 on success, 0 on failure.
*/
int ossl_prov_drbg_uninstantiate(PROV_DRBG *drbg)
{
drbg->state = EVP_RAND_STATE_UNINITIALISED;
return 1;
}
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
static int ossl_prov_drbg_reseed_unlocked(PROV_DRBG *drbg,
int prediction_resistance,
const unsigned char *ent,
size_t ent_len,
const unsigned char *adin,
size_t adinlen)
{
unsigned char *entropy = NULL;
size_t entropylen = 0;
if (!ossl_prov_is_running())
return 0;
if (drbg->state != EVP_RAND_STATE_READY) {
/* try to recover from previous errors */
rand_drbg_restart(drbg);
if (drbg->state == EVP_RAND_STATE_ERROR) {
ERR_raise(ERR_LIB_PROV, PROV_R_IN_ERROR_STATE);
return 0;
}
if (drbg->state == EVP_RAND_STATE_UNINITIALISED) {
ERR_raise(ERR_LIB_PROV, PROV_R_NOT_INSTANTIATED);
return 0;
}
}
if (ent != NULL) {
if (ent_len < drbg->min_entropylen) {
ERR_raise(ERR_LIB_RAND, RAND_R_ENTROPY_OUT_OF_RANGE);
drbg->state = EVP_RAND_STATE_ERROR;
return 0;
}
if (ent_len > drbg->max_entropylen) {
ERR_raise(ERR_LIB_RAND, RAND_R_ENTROPY_INPUT_TOO_LONG);
drbg->state = EVP_RAND_STATE_ERROR;
return 0;
}
}
if (adin == NULL) {
adinlen = 0;
} else if (adinlen > drbg->max_adinlen) {
ERR_raise(ERR_LIB_PROV, PROV_R_ADDITIONAL_INPUT_TOO_LONG);
return 0;
}
drbg->state = EVP_RAND_STATE_ERROR;
drbg->reseed_next_counter = tsan_load(&drbg->reseed_counter);
if (drbg->reseed_next_counter) {
drbg->reseed_next_counter++;
if (!drbg->reseed_next_counter)
drbg->reseed_next_counter = 1;
}
if (ent != NULL) {
#ifdef FIPS_MODULE
/*
* NIST SP-800-90A mandates that entropy *shall not* be provided
* by the consuming application. Instead the data is added as additional
* input.
*
* (NIST SP-800-90Ar1, Sections 9.1 and 9.2)
*/
if (!drbg->reseed(drbg, NULL, 0, ent, ent_len)) {
ERR_raise(ERR_LIB_PROV, PROV_R_UNABLE_TO_RESEED);
return 0;
}
#else
if (!drbg->reseed(drbg, ent, ent_len, adin, adinlen)) {
ERR_raise(ERR_LIB_PROV, PROV_R_UNABLE_TO_RESEED);
return 0;
}
/* There isn't much point adding the same additional input twice */
adin = NULL;
adinlen = 0;
#endif
}
/* Reseed using our sources in addition */
entropylen = get_entropy(drbg, &entropy, drbg->strength,
drbg->min_entropylen, drbg->max_entropylen,
prediction_resistance);
if (entropylen < drbg->min_entropylen
|| entropylen > drbg->max_entropylen) {
ERR_raise(ERR_LIB_PROV, PROV_R_ERROR_RETRIEVING_ENTROPY);
goto end;
}
if (!drbg->reseed(drbg, entropy, entropylen, adin, adinlen))
goto end;
drbg->state = EVP_RAND_STATE_READY;
drbg->generate_counter = 1;
drbg->reseed_time = time(NULL);
tsan_store(&drbg->reseed_counter, drbg->reseed_next_counter);
if (drbg->parent != NULL)
drbg->parent_reseed_counter = get_parent_reseed_count(drbg);
end:
cleanup_entropy(drbg, entropy, entropylen);
if (drbg->state == EVP_RAND_STATE_READY)
return 1;
return 0;
}
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
/*
* Reseed |drbg|, mixing in the specified data
*
* Acquires the drbg->lock for writing, if non-null.
*
* Returns 1 on success, 0 on failure.
*/
int ossl_prov_drbg_reseed(PROV_DRBG *drbg, int prediction_resistance,
const unsigned char *ent, size_t ent_len,
const unsigned char *adin, size_t adinlen)
{
int ret;
if (drbg->lock != NULL && !CRYPTO_THREAD_write_lock(drbg->lock))
return 0;
ret = ossl_prov_drbg_reseed_unlocked(drbg, prediction_resistance, ent,
ent_len, adin, adinlen);
if (drbg->lock != NULL)
CRYPTO_THREAD_unlock(drbg->lock);
return ret;
}
/*
* Generate |outlen| bytes into the buffer at |out|. Reseed if we need
* to or if |prediction_resistance| is set. Additional input can be
* sent in |adin| and |adinlen|.
*
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
* Acquires the drbg->lock for writing if available
*
* Returns 1 on success, 0 on failure.
*
*/
int ossl_prov_drbg_generate(PROV_DRBG *drbg, unsigned char *out, size_t outlen,
unsigned int strength, int prediction_resistance,
const unsigned char *adin, size_t adinlen)
{
int fork_id;
int reseed_required = 1;
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
int ret = 0;
if (!ossl_prov_is_running())
return 0;
fork_id = openssl_get_fork_id();
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
if (drbg->lock != NULL && !CRYPTO_THREAD_read_lock(drbg->lock))
return 0;
if (strength > drbg->strength) {
ERR_raise(ERR_LIB_PROV, PROV_R_INSUFFICIENT_DRBG_STRENGTH);
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
goto err;
}
if (outlen > drbg->max_request) {
ERR_raise(ERR_LIB_PROV, PROV_R_REQUEST_TOO_LARGE_FOR_DRBG);
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
goto err;
}
if (adinlen > drbg->max_adinlen) {
ERR_raise(ERR_LIB_PROV, PROV_R_ADDITIONAL_INPUT_TOO_LONG);
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
goto err;
}
if (drbg->state != EVP_RAND_STATE_READY)
goto reseed;
if (prediction_resistance)
goto reseed;
if (drbg->fork_id != fork_id)
goto reseed;
if (drbg->reseed_interval > 0
&& drbg->generate_counter >= drbg->reseed_interval)
goto reseed;
if (drbg->reseed_time_interval > 0) {
time_t now = time(NULL);
if (now < drbg->reseed_time
|| now - drbg->reseed_time >= drbg->reseed_time_interval)
goto reseed;
}
if (drbg->parent != NULL
&& get_parent_reseed_count(drbg) != drbg->parent_reseed_counter)
goto reseed;
reseed_required = 0;
/*
* From here to the end of the function, if we are doing locking, we do so
* under the write lock
*/
reseed:
if (drbg->lock != NULL) {
/* switch to write lock */
CRYPTO_THREAD_unlock(drbg->lock);
if (!CRYPTO_THREAD_write_lock(drbg->lock))
return 0;
}
if (reseed_required) {
drbg->fork_id = fork_id;
/*
* Note that ossl_prov_drbg_reseed_unlocked also handles
* the need for restart for us
*/
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
if (!ossl_prov_drbg_reseed_unlocked(drbg, prediction_resistance, NULL,
0, adin, adinlen)) {
ERR_raise(ERR_LIB_PROV, PROV_R_RESEED_ERROR);
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
goto err;
}
adin = NULL;
adinlen = 0;
}
if (!drbg->generate(drbg, out, outlen, adin, adinlen)) {
drbg->state = EVP_RAND_STATE_ERROR;
ERR_raise(ERR_LIB_PROV, PROV_R_GENERATE_ERROR);
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
goto err;
}
drbg->generate_counter++;
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
ret = 1;
err:
if (drbg->lock != NULL)
CRYPTO_THREAD_unlock(drbg->lock);
return ret;
}
/*
* Restart |drbg|, using the specified entropy or additional input
*
* Tries its best to get the drbg instantiated by all means,
* regardless of its current state.
*
* Optionally, a |buffer| of |len| random bytes can be passed,
* which is assumed to contain at least |entropy| bits of entropy.
*
* If |entropy| > 0, the buffer content is used as entropy input.
*
* If |entropy| == 0, the buffer content is used as additional input
*
* Returns 1 on success, 0 on failure.
*
* This function is used internally only.
*/
static int rand_drbg_restart(PROV_DRBG *drbg)
{
/* repair error state */
if (drbg->state == EVP_RAND_STATE_ERROR)
drbg->uninstantiate(drbg);
/* repair uninitialized state */
if (drbg->state == EVP_RAND_STATE_UNINITIALISED)
/* reinstantiate drbg */
ossl_prov_drbg_instantiate(drbg, drbg->strength, 0, NULL, 0);
return drbg->state == EVP_RAND_STATE_READY;
}
/* Provider support from here down */
static const OSSL_DISPATCH *find_call(const OSSL_DISPATCH *dispatch,
int function)
{
if (dispatch != NULL)
while (dispatch->function_id != 0) {
if (dispatch->function_id == function)
return dispatch;
dispatch++;
}
return NULL;
}
int ossl_drbg_enable_locking(void *vctx)
{
PROV_DRBG *drbg = vctx;
if (drbg != NULL && drbg->lock == NULL) {
if (drbg->parent_enable_locking != NULL)
if (!drbg->parent_enable_locking(drbg->parent)) {
ERR_raise(ERR_LIB_PROV, PROV_R_PARENT_LOCKING_NOT_ENABLED);
return 0;
}
drbg->lock = CRYPTO_THREAD_lock_new();
if (drbg->lock == NULL) {
ERR_raise(ERR_LIB_PROV, PROV_R_FAILED_TO_CREATE_LOCK);
return 0;
}
}
return 1;
}
/*
* Allocate memory and initialize a new DRBG. The DRBG is allocated on
* the secure heap if |secure| is nonzero and the secure heap is enabled.
* The |parent|, if not NULL, will be used as random source for reseeding.
* This also requires the parent's provider context and the parent's lock.
*
* Returns a pointer to the new DRBG instance on success, NULL on failure.
*/
PROV_DRBG *ossl_rand_drbg_new
(void *provctx, void *parent, const OSSL_DISPATCH *p_dispatch,
int (*dnew)(PROV_DRBG *ctx),
void (*dfree)(void *vctx),
int (*instantiate)(PROV_DRBG *drbg,
const unsigned char *entropy, size_t entropylen,
const unsigned char *nonce, size_t noncelen,
const unsigned char *pers, size_t perslen),
int (*uninstantiate)(PROV_DRBG *ctx),
int (*reseed)(PROV_DRBG *drbg, const unsigned char *ent, size_t ent_len,
const unsigned char *adin, size_t adin_len),
int (*generate)(PROV_DRBG *, unsigned char *out, size_t outlen,
const unsigned char *adin, size_t adin_len))
{
PROV_DRBG *drbg;
unsigned int p_str;
const OSSL_DISPATCH *pfunc;
if (!ossl_prov_is_running())
return NULL;
drbg = OPENSSL_zalloc(sizeof(*drbg));
if (drbg == NULL)
return NULL;
drbg->provctx = provctx;
drbg->instantiate = instantiate;
drbg->uninstantiate = uninstantiate;
drbg->reseed = reseed;
drbg->generate = generate;
drbg->fork_id = openssl_get_fork_id();
/* Extract parent's functions */
drbg->parent = parent;
if ((pfunc = find_call(p_dispatch, OSSL_FUNC_RAND_ENABLE_LOCKING)) != NULL)
drbg->parent_enable_locking = OSSL_FUNC_rand_enable_locking(pfunc);
if ((pfunc = find_call(p_dispatch, OSSL_FUNC_RAND_LOCK)) != NULL)
drbg->parent_lock = OSSL_FUNC_rand_lock(pfunc);
if ((pfunc = find_call(p_dispatch, OSSL_FUNC_RAND_UNLOCK)) != NULL)
drbg->parent_unlock = OSSL_FUNC_rand_unlock(pfunc);
if ((pfunc = find_call(p_dispatch, OSSL_FUNC_RAND_GET_CTX_PARAMS)) != NULL)
drbg->parent_get_ctx_params = OSSL_FUNC_rand_get_ctx_params(pfunc);
if ((pfunc = find_call(p_dispatch, OSSL_FUNC_RAND_NONCE)) != NULL)
drbg->parent_nonce = OSSL_FUNC_rand_nonce(pfunc);
if ((pfunc = find_call(p_dispatch, OSSL_FUNC_RAND_GET_SEED)) != NULL)
drbg->parent_get_seed = OSSL_FUNC_rand_get_seed(pfunc);
if ((pfunc = find_call(p_dispatch, OSSL_FUNC_RAND_CLEAR_SEED)) != NULL)
drbg->parent_clear_seed = OSSL_FUNC_rand_clear_seed(pfunc);
/* Set some default maximums up */
drbg->max_entropylen = DRBG_MAX_LENGTH;
drbg->max_noncelen = DRBG_MAX_LENGTH;
drbg->max_perslen = DRBG_MAX_LENGTH;
drbg->max_adinlen = DRBG_MAX_LENGTH;
drbg->generate_counter = 1;
drbg->reseed_counter = 1;
drbg->reseed_interval = RESEED_INTERVAL;
drbg->reseed_time_interval = TIME_INTERVAL;
if (!dnew(drbg))
goto err;
if (parent != NULL) {
if (!get_parent_strength(drbg, &p_str))
goto err;
if (drbg->strength > p_str) {
/*
* We currently don't support the algorithm from NIST SP 800-90C
* 10.1.2 to use a weaker DRBG as source
*/
ERR_raise(ERR_LIB_PROV, PROV_R_PARENT_STRENGTH_TOO_WEAK);
goto err;
}
}
#ifdef TSAN_REQUIRES_LOCKING
if (!ossl_drbg_enable_locking(drbg))
goto err;
#endif
return drbg;
err:
dfree(drbg);
return NULL;
}
void ossl_rand_drbg_free(PROV_DRBG *drbg)
{
if (drbg == NULL)
return;
CRYPTO_THREAD_lock_free(drbg->lock);
OPENSSL_free(drbg);
}
Refactor the DRBG implementations to manage locking themselves Previously the EVP layer would call lock and unlock functions on the underlying DRBG implementation to say when a lock should be acquired and released. This gives the DRBG implementation no say as to what kind of lock should obtained (e.g. read/write) or even whether a lock is actually needed or not. In reality we know whether a DRBG is supposed to be in locking mode or not because the "enable_locking()" function will have been called if locks should be used. Therefore we re-interpret the lock and unlock functions as "hints" from the EVP layer which we ignore. Instead we acquire locks only when we need them. By knowing the context we can obtain either a read or a write lock as appropriate. This may mean that in some rare cases we acquire the locks more than once for a single EVP call, if the EVP call makes several calls to the underlying DRBG. But in practice almost all EVP calls only make one such call. EVP_RAND_generate() is an example of a call where multiple DRBG calls may be made. One of these gets the "max_request" parameter (which is constant for all of our own DRBGs) and it may make several calls to the DRBG generate call - but only if the requested size is very large which will rarely be the case. Partially fixes #20286 Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Todd Short <todd.short@me.com> Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/20970)
2023-05-15 18:33:01 +08:00
/*
* Helper function called by internal DRBG implementations. Assumes that at
* least a read lock has been taken on drbg->lock
*/
int ossl_drbg_get_ctx_params(PROV_DRBG *drbg, OSSL_PARAM params[])
{
OSSL_PARAM *p;
p = OSSL_PARAM_locate(params, OSSL_RAND_PARAM_STATE);
if (p != NULL && !OSSL_PARAM_set_int(p, drbg->state))
return 0;
p = OSSL_PARAM_locate(params, OSSL_RAND_PARAM_STRENGTH);
if (p != NULL && !OSSL_PARAM_set_int(p, drbg->strength))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_MIN_ENTROPYLEN);
if (p != NULL && !OSSL_PARAM_set_size_t(p, drbg->min_entropylen))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_MAX_ENTROPYLEN);
if (p != NULL && !OSSL_PARAM_set_size_t(p, drbg->max_entropylen))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_MIN_NONCELEN);
if (p != NULL && !OSSL_PARAM_set_size_t(p, drbg->min_noncelen))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_MAX_NONCELEN);
if (p != NULL && !OSSL_PARAM_set_size_t(p, drbg->max_noncelen))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_MAX_PERSLEN);
if (p != NULL && !OSSL_PARAM_set_size_t(p, drbg->max_perslen))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_MAX_ADINLEN);
if (p != NULL && !OSSL_PARAM_set_size_t(p, drbg->max_adinlen))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_RESEED_REQUESTS);
if (p != NULL && !OSSL_PARAM_set_uint(p, drbg->reseed_interval))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_RESEED_TIME);
if (p != NULL && !OSSL_PARAM_set_time_t(p, drbg->reseed_time))
return 0;
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_RESEED_TIME_INTERVAL);
if (p != NULL && !OSSL_PARAM_set_time_t(p, drbg->reseed_time_interval))
return 0;
if (!OSSL_FIPS_IND_GET_CTX_PARAM(drbg, params))
return 0;
return 1;
}
/*
* Helper function to get certain params that require no lock to obtain. Sets
* *complete to 1 if all the params were processed, or 0 otherwise
*/
int ossl_drbg_get_ctx_params_no_lock(PROV_DRBG *drbg, OSSL_PARAM params[],
int *complete)
{
size_t cnt = 0;
OSSL_PARAM *p;
/* This value never changes once set */
p = OSSL_PARAM_locate(params, OSSL_RAND_PARAM_MAX_REQUEST);
if (p != NULL) {
if (!OSSL_PARAM_set_size_t(p, drbg->max_request))
return 0;
cnt++;
}
/*
* Can be changed by multiple threads, but we tolerate inaccuracies in this
* value.
*/
p = OSSL_PARAM_locate(params, OSSL_DRBG_PARAM_RESEED_COUNTER);
if (p != NULL) {
if (!OSSL_PARAM_set_uint(p, tsan_load(&drbg->reseed_counter)))
return 0;
cnt++;
}
if (params[cnt].key == NULL)
*complete = 1;
else
*complete = 0;
return 1;
}
int ossl_drbg_set_ctx_params(PROV_DRBG *drbg, const OSSL_PARAM params[])
{
const OSSL_PARAM *p;
if (ossl_param_is_empty(params))
return 1;
p = OSSL_PARAM_locate_const(params, OSSL_DRBG_PARAM_RESEED_REQUESTS);
if (p != NULL && !OSSL_PARAM_get_uint(p, &drbg->reseed_interval))
return 0;
p = OSSL_PARAM_locate_const(params, OSSL_DRBG_PARAM_RESEED_TIME_INTERVAL);
if (p != NULL && !OSSL_PARAM_get_time_t(p, &drbg->reseed_time_interval))
return 0;
return 1;
}
#ifdef FIPS_MODULE
static int digest_allowed(const EVP_MD *md)
{
/* FIPS 140-3 IG D.R limited DRBG digests to a specific set */
static const char *const allowed_digests[] = {
"SHA1", /* SHA 1 allowed */
"SHA2-256", "SHA2-512", /* non-truncated SHA2 allowed */
"SHA3-256", "SHA3-512", /* non-truncated SHA3 allowed */
};
size_t i;
for (i = 0; i < OSSL_NELEM(allowed_digests); i++) {
if (EVP_MD_is_a(md, allowed_digests[i]))
return 1;
}
return 0;
}
#endif
/* Confirm digest is allowed to be used with a DRBG */
int ossl_drbg_verify_digest(PROV_DRBG *drbg, OSSL_LIB_CTX *libctx,
const EVP_MD *md)
{
#ifdef FIPS_MODULE
int approved = digest_allowed(md);
if (!approved) {
if (!OSSL_FIPS_IND_ON_UNAPPROVED(drbg, OSSL_FIPS_IND_SETTABLE0,
libctx, "DRBG", "Digest",
ossl_fips_config_restricted_drbg_digests)) {
ERR_raise(ERR_LIB_PROV, PROV_R_DIGEST_NOT_ALLOWED);
return 0;
}
}
#else /* FIPS_MODULE */
/* Outside of FIPS, any digests that are not XOF are allowed */
if (EVP_MD_xof(md)) {
ERR_raise(ERR_LIB_PROV, PROV_R_XOF_DIGESTS_NOT_ALLOWED);
return 0;
}
#endif /* FIPS_MODULE */
return 1;
}