Received-SPF: pass (google.com: domain of linux-crypto-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Nicolai Stange <nstange@suse.de>
To:     Herbert Xu <herbert@gondor.apana.org.au>,
        "David S. Miller" <davem@davemloft.net>
Cc:     =?UTF-8?q?Stephan=20M=C3=BCller?= <smueller@chronox.de>,
        Hannes Reinecke <hare@suse.de>, Torsten Duwe <duwe@suse.de>,
        David Howells <dhowells@redhat.com>,
        Jarkko Sakkinen <jarkko@kernel.org>,
        linux-crypto@vger.kernel.org, linux-kernel@vger.kernel.org,
        keyrings@vger.kernel.org, Nicolai Stange <nstange@suse.de>
Subject: [PATCH v3 05/15] crypto: dh - split out deserialization code from crypto_dh_decode()
Date:   Wed,  2 Feb 2022 11:40:02 +0100
Message-Id: <20220202104012.4193-6-nstange@suse.de>
In-Reply-To: <20220202104012.4193-1-nstange@suse.de>
References: <20220202104012.4193-1-nstange@suse.de>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

A subsequent commit will introduce "dh" wrapping templates of the form
"ffdhe2048(dh)", "ffdhe3072(dh)" and so on in order to provide built-in
support for the well-known safe-prime ffdhe group parameters specified in
RFC 7919.

Those templates' ->set_secret() will wrap the inner "dh" implementation's
->set_secret() and set the ->p and ->g group parameters as appropriate on
the way inwards. More specifically,
- A ffdheXYZ(dh) user would call crypto_dh_encode() on a struct dh instance
  having ->p == ->g == NULL as well as ->p_size == ->g_size == 0 and pass
  the resulting buffer to the outer ->set_secret().
- This outer ->set_secret() would then decode the struct dh via
  crypto_dh_decode_key(), set ->p, ->g, ->p_size as well as ->g_size as
  appropriate for the group in question and encode the struct dh again
  before passing it further down to the inner "dh"'s ->set_secret().

The problem is that crypto_dh_decode_key() implements some basic checks
which would reject parameter sets with ->p_size == 0 and thus, the ffdheXYZ
templates' ->set_secret() cannot use it as-is for decoding the passed
buffer. As the inner "dh"'s ->set_secret() will eventually conduct said
checks on the final parameter set anyway, the outer ->set_secret() really
only needs the decoding functionality.

Split out the pure struct dh decoding part from crypto_dh_decode_key() into
the new __crypto_dh_decode_key().

__crypto_dh_decode_key() gets defined in crypto/dh_helper.c, but will have
to get called from crypto/dh.c and thus, its declaration must be somehow
made available to the latter. Strictly speaking, __crypto_dh_decode_key()
is internal to the dh_generic module, yet it would be a bit over the top
to introduce a new header like e.g. include/crypto/internal/dh.h
containing just a single prototype. Add the __crypto_dh_decode_key()
declaration to include/crypto/dh.h instead.

Provide a proper kernel-doc annotation, even though
__crypto_dh_decode_key() is purposedly not on the function list specified
in Documentation/crypto/api-kpp.rst.

Signed-off-by: Nicolai Stange <nstange@suse.de>
---
 crypto/dh_helper.c  | 27 +++++++++++++++++++--------
 include/crypto/dh.h | 16 ++++++++++++++++
 2 files changed, 35 insertions(+), 8 deletions(-)

diff --git a/crypto/dh_helper.c b/crypto/dh_helper.c
index aabc91e4f63f..2d499879328b 100644
--- a/crypto/dh_helper.c
+++ b/crypto/dh_helper.c
@@ -63,7 +63,7 @@ int crypto_dh_encode_key(char *buf, unsigned int len, const struct dh *params)
 }
 EXPORT_SYMBOL_GPL(crypto_dh_encode_key);
 
-int crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params)
+int __crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params)
 {
 	const u8 *ptr = buf;
 	struct kpp_secret secret;
@@ -81,6 +81,24 @@ int crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params)
 	if (secret.len != crypto_dh_key_len(params))
 		return -EINVAL;
 
+	/* Don't allocate memory. Set pointers to data within
+	 * the given buffer
+	 */
+	params->key = (void *)ptr;
+	params->p = (void *)(ptr + params->key_size);
+	params->g = (void *)(ptr + params->key_size + params->p_size);
+
+	return 0;
+}
+
+int crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params)
+{
+	int err;
+
+	err = __crypto_dh_decode_key(buf, len, params);
+	if (err)
+		return err;
+
 	/*
 	 * Don't permit the buffer for 'key' or 'g' to be larger than 'p', since
 	 * some drivers assume otherwise.
@@ -89,13 +107,6 @@ int crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params)
 	    params->g_size > params->p_size)
 		return -EINVAL;
 
-	/* Don't allocate memory. Set pointers to data within
-	 * the given buffer
-	 */
-	params->key = (void *)ptr;
-	params->p = (void *)(ptr + params->key_size);
-	params->g = (void *)(ptr + params->key_size + params->p_size);
-
 	/*
 	 * Don't permit 'p' to be 0.  It's not a prime number, and it's subject
 	 * to corner cases such as 'mod 0' being undefined or
diff --git a/include/crypto/dh.h b/include/crypto/dh.h
index 67f3f6bca527..7b863e911cb4 100644
--- a/include/crypto/dh.h
+++ b/include/crypto/dh.h
@@ -79,4 +79,20 @@ int crypto_dh_encode_key(char *buf, unsigned int len, const struct dh *params);
  */
 int crypto_dh_decode_key(const char *buf, unsigned int len, struct dh *params);
 
+/**
+ * __crypto_dh_decode_key() - decode a private key without parameter checks
+ * @buf:	Buffer holding a packet key that should be decoded
+ * @len:	Length of the packet private key buffer
+ * @params:	Buffer allocated by the caller that is filled with the
+ *		unpacked DH private key.
+ *
+ * Internal function providing the same services as the exported
+ * crypto_dh_decode_key(), but without any of those basic parameter
+ * checks conducted by the latter.
+ *
+ * Return:	-EINVAL if buffer has insufficient size, 0 on success
+ */
+int __crypto_dh_decode_key(const char *buf, unsigned int len,
+			   struct dh *params);
+
 #endif
-- 
2.26.2