lib/crypto: nh: Add NH library

Add support for the NH "almost-universal hash function" to lib/crypto/,
specifically the variant of NH used in Adiantum.

This will replace the need for the "nhpoly1305" crypto_shash algorithm.
All the implementations of "nhpoly1305" use architecture-optimized code
only for the NH stage; they just use the generic C Poly1305 code for the
Poly1305 stage.  We can achieve the same result in a simpler way using
an (architecture-optimized) nh() function combined with code in
crypto/adiantum.c that passes the results to the Poly1305 library.

This commit begins this cleanup by adding the nh() function.  The code
is derived from crypto/nhpoly1305.c and include/crypto/nhpoly1305.h.

Link: https://lore.kernel.org/r/20251211011846.8179-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@kernel.org>

include/crypto/nh.h

@@ -0,0 +1,52 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * NH hash function for Adiantum
+ */
+
+#ifndef _CRYPTO_NH_H
+#define _CRYPTO_NH_H
+
+#include <linux/types.h>
+
+/* NH parameterization: */
+
+/* Endianness: little */
+/* Word size: 32 bits (works well on NEON, SSE2, AVX2) */
+
+/* Stride: 2 words (optimal on ARM32 NEON; works okay on other CPUs too) */
+#define NH_PAIR_STRIDE		2
+#define NH_MESSAGE_UNIT		(NH_PAIR_STRIDE * 2 * sizeof(u32))
+
+/* Num passes (Toeplitz iteration count): 4, to give ε = 2^{-128} */
+#define NH_NUM_PASSES		4
+#define NH_HASH_BYTES		(NH_NUM_PASSES * sizeof(u64))
+
+/* Max message size: 1024 bytes (32x compression factor) */
+#define NH_NUM_STRIDES		64
+#define NH_MESSAGE_WORDS	(NH_PAIR_STRIDE * 2 * NH_NUM_STRIDES)
+#define NH_MESSAGE_BYTES	(NH_MESSAGE_WORDS * sizeof(u32))
+#define NH_KEY_WORDS		(NH_MESSAGE_WORDS + \
+				 NH_PAIR_STRIDE * 2 * (NH_NUM_PASSES - 1))
+#define NH_KEY_BYTES		(NH_KEY_WORDS * sizeof(u32))
+
+/**
+ * nh() - NH hash function for Adiantum
+ * @key: The key.  @message_len + 48 bytes of it are used.  This is NH_KEY_BYTES
+ *	 if @message_len has its maximum length of NH_MESSAGE_BYTES.
+ * @message: The message
+ * @message_len: The message length in bytes.  Must be a multiple of 16
+ *		 (NH_MESSAGE_UNIT) and at most 1024 (NH_MESSAGE_BYTES).
+ * @hash: (output) The resulting hash value
+ *
+ * Note: the pseudocode for NH in the Adiantum paper iterates over 1024-byte
+ * segments of the message, computes a 32-byte hash for each, and returns all
+ * the hashes concatenated together.  In contrast, this function just hashes one
+ * segment and returns one hash.  It's the caller's responsibility to call this
+ * function for each 1024-byte segment and collect all the hashes.
+ *
+ * Context: Any context.
+ */
+void nh(const u32 *key, const u8 *message, size_t message_len,
+	__le64 hash[NH_NUM_PASSES]);
+
+#endif /* _CRYPTO_NH_H */

lib/crypto/Kconfig

@@ -108,6 +108,16 @@ config CRYPTO_LIB_MLDSA
 	  The ML-DSA library functions.  Select this if your module uses any of
 	  the functions from <crypto/mldsa.h>.
 
+config CRYPTO_LIB_NH
+	tristate
+	help
+	  Implementation of the NH almost-universal hash function, specifically
+	  the variant of NH used in Adiantum.
+
+config CRYPTO_LIB_NH_ARCH
+	bool
+	depends on CRYPTO_LIB_NH && !UML
+
 config CRYPTO_LIB_POLY1305
 	tristate
 	help

lib/crypto/Makefile

@@ -131,6 +131,14 @@ libmldsa-y := mldsa.o
 
 ################################################################################
 
+obj-$(CONFIG_CRYPTO_LIB_NH) += libnh.o
+libnh-y := nh.o
+ifeq ($(CONFIG_CRYPTO_LIB_NH_ARCH),y)
+CFLAGS_nh.o += -I$(src)/$(SRCARCH)
+endif
+
+################################################################################
+
 obj-$(CONFIG_CRYPTO_LIB_POLY1305) += libpoly1305.o
 libpoly1305-y := poly1305.o
 ifeq ($(CONFIG_ARCH_SUPPORTS_INT128),y)

lib/crypto/nh.c

@@ -0,0 +1,82 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2018 Google LLC
+ */
+
+/*
+ * Implementation of the NH almost-universal hash function, specifically the
+ * variant of NH used in Adiantum.  This is *not* a cryptographic hash function.
+ *
+ * Reference: section 6.3 of "Adiantum: length-preserving encryption for
+ * entry-level processors" (https://eprint.iacr.org/2018/720.pdf).
+ */
+
+#include <crypto/nh.h>
+#include <linux/export.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/unaligned.h>
+
+#ifdef CONFIG_CRYPTO_LIB_NH_ARCH
+#include "nh.h" /* $(SRCARCH)/nh.h */
+#else
+static bool nh_arch(const u32 *key, const u8 *message, size_t message_len,
+		    __le64 hash[NH_NUM_PASSES])
+{
+	return false;
+}
+#endif
+
+void nh(const u32 *key, const u8 *message, size_t message_len,
+	__le64 hash[NH_NUM_PASSES])
+{
+	u64 sums[4] = { 0, 0, 0, 0 };
+
+	if (nh_arch(key, message, message_len, hash))
+		return;
+
+	static_assert(NH_PAIR_STRIDE == 2);
+	static_assert(NH_NUM_PASSES == 4);
+
+	while (message_len) {
+		u32 m0 = get_unaligned_le32(message + 0);
+		u32 m1 = get_unaligned_le32(message + 4);
+		u32 m2 = get_unaligned_le32(message + 8);
+		u32 m3 = get_unaligned_le32(message + 12);
+
+		sums[0] += (u64)(u32)(m0 + key[0]) * (u32)(m2 + key[2]);
+		sums[1] += (u64)(u32)(m0 + key[4]) * (u32)(m2 + key[6]);
+		sums[2] += (u64)(u32)(m0 + key[8]) * (u32)(m2 + key[10]);
+		sums[3] += (u64)(u32)(m0 + key[12]) * (u32)(m2 + key[14]);
+		sums[0] += (u64)(u32)(m1 + key[1]) * (u32)(m3 + key[3]);
+		sums[1] += (u64)(u32)(m1 + key[5]) * (u32)(m3 + key[7]);
+		sums[2] += (u64)(u32)(m1 + key[9]) * (u32)(m3 + key[11]);
+		sums[3] += (u64)(u32)(m1 + key[13]) * (u32)(m3 + key[15]);
+		key += NH_MESSAGE_UNIT / sizeof(key[0]);
+		message += NH_MESSAGE_UNIT;
+		message_len -= NH_MESSAGE_UNIT;
+	}
+
+	hash[0] = cpu_to_le64(sums[0]);
+	hash[1] = cpu_to_le64(sums[1]);
+	hash[2] = cpu_to_le64(sums[2]);
+	hash[3] = cpu_to_le64(sums[3]);
+}
+EXPORT_SYMBOL_GPL(nh);
+
+#ifdef nh_mod_init_arch
+static int __init nh_mod_init(void)
+{
+	nh_mod_init_arch();
+	return 0;
+}
+subsys_initcall(nh_mod_init);
+
+static void __exit nh_mod_exit(void)
+{
+}
+module_exit(nh_mod_exit);
+#endif
+
+MODULE_DESCRIPTION("NH almost-universal hash function");
+MODULE_LICENSE("GPL");