From patchwork Thu Feb 29 00:32:02 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Stephen Hemminger X-Patchwork-Id: 137458 X-Patchwork-Delegate: thomas@monjalon.net Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 2847B43C20; Thu, 29 Feb 2024 01:33:22 +0100 (CET) Received: from mails.dpdk.org (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id A0BC6402B4; Thu, 29 Feb 2024 01:33:21 +0100 (CET) Received: from mail-pf1-f182.google.com (mail-pf1-f182.google.com [209.85.210.182]) by mails.dpdk.org (Postfix) with ESMTP id 8E7EA402AF for ; Thu, 29 Feb 2024 01:33:20 +0100 (CET) Received: by mail-pf1-f182.google.com with SMTP id d2e1a72fcca58-6da4a923b1bso258854b3a.2 for ; Wed, 28 Feb 2024 16:33:20 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=networkplumber-org.20230601.gappssmtp.com; s=20230601; t=1709166799; x=1709771599; darn=dpdk.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to; bh=jb4taKT9Bu6mP1Vwvx0msDDKpinRZuv2mV3iUPOQhFc=; b=YEgOYdPOv5DtEUPo+yOtV6vk/bEb0HPiXEKooeIJXVv75awVYO8iZBhyH83hOindW/ siLSsPjucYAyo88gOvNzfNjLqIlIpckpTAyEHwI7dNF43RYLlo7PkV0nZaDdXS4wXfjT KcVvuNAxDaMOn6KU9EcFC/+nKBeTRAHha2te+vcBBdlhlzDckvd+P9G3OmwDENSvVhDz k/pA+uu7+iCAE8XB3ClqPguegv4LFGMqTmLAw+iI9MO39qIaFhw5bncOyhdrSMjcjQNr BNs/wX6LFCW0LAscrH8c+QmsGacVeXphGwgxzNkhyfszF5Yvb9y63ZiWFy9xBfe40+Sv ymkA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1709166799; x=1709771599; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=jb4taKT9Bu6mP1Vwvx0msDDKpinRZuv2mV3iUPOQhFc=; b=W7wbGZ2wE5WrN1txmPgHcjQkz6yYzgwUZnBc9TI0G84XoT4SeLXKp0fwanESkc8wxx K0wsjeP3NOGvFAmSa2TzpHsXRWLocbme0SpUvAtfsM35jfc0P6sQf7Q7tgRNk51dmJ9d R2dW4ASMG2nh2/iqkM1XNg4oVybJ7IpuzeEcXtOV5cNog590bGdYkVNwzQn6OGRY4k7v WXYcTTjKUfthLbCED6qqjQ8VSBMNWL47VmwBk40hRrvD1LUynVP/yS8QQl/sX06t5hBT 8P2vQZPnMXc0XPxbrrGZxs2olWW0UibSrDyOlfIDw6zCfByZbd8F+A4FR2PXsBK7xjEj QR6Q== X-Gm-Message-State: AOJu0YzoLLqUAF+AaN2JUdCJU1xjKhaRl2WiLFItqCBEzbKN05eDWLjX Wi7Zycv7KiOWMctX6DY4kj0uM0pu6vCJVjrTveTbHzst01Yx3evG6KM6tQhlI3qqVGwYMdinVS8 K X-Google-Smtp-Source: AGHT+IFX0kHMlJg9hAo5sQ5UQylJXwVS719GIs+P+QfBTIYqykI4e9RKZTRk06RrOfaRbQ4Llk6WOg== X-Received: by 2002:a05:6a20:a892:b0:1a1:3b8:827b with SMTP id ca18-20020a056a20a89200b001a103b8827bmr779098pzb.52.1709166799456; Wed, 28 Feb 2024 16:33:19 -0800 (PST) Received: from hermes.local (204-195-123-141.wavecable.com. [204.195.123.141]) by smtp.gmail.com with ESMTPSA id s13-20020a170902ea0d00b001dcc0c84721sm61902plg.99.2024.02.28.16.33.18 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 28 Feb 2024 16:33:19 -0800 (PST) From: Stephen Hemminger To: dev@dpdk.org Cc: Stephen Hemminger , Yipeng Wang , Sameh Gobriel , Bruce Richardson , Vladimir Medvedkin Subject: [PATCH v3] lib/hash: add siphash Date: Wed, 28 Feb 2024 16:32:02 -0800 Message-ID: <20240229003201.186997-2-stephen@networkplumber.org> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20240227174012.343004-1-stephen@networkplumber.org> References: <20240227174012.343004-1-stephen@networkplumber.org> MIME-Version: 1.0 X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Add SipHash which is a fast and cryptographicly sound hash created by Jean-Philippe Aumasson and Daniel J. Bernstein. Siphash is widely used by Linux, FreeBSD, OpenBSD and other projects because it is fast and resistant to DoS attacks. This version is designed to be useful as alternative hash with cuckoo hash in DPDK. Implementation is based of the public domain and Creative Common license reference version as well as some optimizations from the GPL-2 or BSD-3 licensed version in Linux. Signed-off-by: Stephen Hemminger --- v3 - unroll the rounds to get 3% speedup use memcpy to avoid having x86 only code app/test/test_hash_functions.c | 71 ++++++++++++- lib/hash/meson.build | 2 + lib/hash/rte_siphash.c | 176 +++++++++++++++++++++++++++++++++ lib/hash/rte_siphash.h | 86 ++++++++++++++++ lib/hash/version.map | 7 ++ 5 files changed, 338 insertions(+), 4 deletions(-) create mode 100644 lib/hash/rte_siphash.c create mode 100644 lib/hash/rte_siphash.h diff --git a/app/test/test_hash_functions.c b/app/test/test_hash_functions.c index 70820d1f1976..3bffbe4ed4ba 100644 --- a/app/test/test_hash_functions.c +++ b/app/test/test_hash_functions.c @@ -15,6 +15,7 @@ #include #include #include +#include #include "test.h" @@ -52,6 +53,42 @@ static uint32_t hash_values_crc[2][12] = {{ } }; +static uint32_t hash_values_hsiphash[2][12] = { + { + 0x8e01e473, 0xc41e3669, + 0x813e4dbd, 0x7ebe2eea, + 0x33a5c5b7, 0xc910629b, + 0xba50237f, 0xbbc870c6, + 0x95124362, 0x850f8e0d, + 0x192ff266, 0xb41d8206, + }, { + 0x66200aa0, 0x769e7201, + 0x0e934d03, 0x96d7c892, + 0xb4643534, 0xcb758913, + 0x498b66e9, 0x116b4082, + 0x603030dc, 0x644b608b, + 0x74b29c27, 0x513f3a9c, + } +}; + +static uint64_t hash_values_siphash[2][12] = { + { + 0x8b5a0baa49fbc58d, 0x62c3506f27376c25, + 0xef7bdf3ee24abec8, 0xc72b1c24fc2f7938, + 0xc902632ed88f897f, 0xab631e00063006f5, + 0x7b821577565ea3a4, 0x8b19265d1c12cdc7, + 0x610e7ab6ada60b22, 0x3c6f1970dd62f235, + 0x4b8db19fd6940031, 0xa43827a530b08989, + }, { + 0x86e1bbae2893aaf1, 0x9c05614a696cda03, + 0x9ee31847083019c3, 0x600e860c97264e31, + 0xda8721038e4972bc, 0xfeedeb1b8bfe5d1e, + 0xb2d6388922af426f, 0x7d05b8e82e38cf30, + 0x439caa6ecd0b628d, 0x6d4c4ba6f3f2aed5, + 0x5622bb0da20658ff, 0x409d76de4adcd475, + } +}; + /******************************************************************************* * Hash function performance test configuration section. Each performance test * will be performed HASHTEST_ITERATIONS times. @@ -61,9 +98,11 @@ static uint32_t hash_values_crc[2][12] = {{ */ #define HASHTEST_ITERATIONS 1000000 #define MAX_KEYSIZE 64 -static rte_hash_function hashtest_funcs[] = {rte_jhash, rte_hash_crc}; -static uint32_t hashtest_initvals[] = {0, 0xdeadbeef}; -static uint32_t hashtest_key_lens[] = { +static const rte_hash_function hashtest_funcs[] = {rte_jhash, rte_hash_crc, rte_hsiphash}; +static const uint32_t hashtest_initvals[] = {0, 0xdeadbeef}; +static const uint64_t hashtest_initkeys[] = {0, UINT64_C(0xfeedc0dedeadbeef)}; + +static const uint32_t hashtest_key_lens[] = { 1, 2, /* Unusual key sizes */ 4, 8, 16, 32, 48, 64, /* standard key sizes */ 9, /* IPv4 SRC + DST + protocol, unpadded */ @@ -85,6 +124,9 @@ get_hash_name(rte_hash_function f) if (f == rte_hash_crc) return "rte_hash_crc"; + if (f == rte_hsiphash) + return "hsiphash"; + return "UnknownHash"; } @@ -126,6 +168,7 @@ run_hash_func_perf_tests(void) printf(" Number of iterations for each test = %d\n", HASHTEST_ITERATIONS); printf("Hash Func. , Key Length (bytes), Initial value, Ticks/Op.\n"); + fflush(stdout); for (i = 0; i < RTE_DIM(hashtest_initvals); i++) { for (j = 0; j < RTE_DIM(hashtest_key_lens); j++) { @@ -147,13 +190,15 @@ verify_precalculated_hash_func_tests(void) { unsigned i, j; uint8_t key[64]; - uint32_t hash; for (i = 0; i < 64; i++) key[i] = (uint8_t) i; for (i = 0; i < RTE_DIM(hashtest_key_lens); i++) { for (j = 0; j < RTE_DIM(hashtest_initvals); j++) { + uint32_t hash; + uint64_t shash; + hash = rte_jhash(key, hashtest_key_lens[i], hashtest_initvals[j]); if (hash != hash_values_jhash[j][i]) { @@ -173,6 +218,24 @@ verify_precalculated_hash_func_tests(void) hash_values_crc[j][i], hash); return -1; } + + hash = rte_hsiphash(key, hashtest_key_lens[i], hashtest_initvals[j]); + if (hash != hash_values_hsiphash[j][i]) { + printf("Hsiphash for %u bytes with initial value 0x%x." + "Expected 0x%x, but got 0x%x\n", + hashtest_key_lens[i], hashtest_initvals[j], + hash_values_hsiphash[j][i], hash); + return -1; + } + + shash = rte_siphash(key, hashtest_key_lens[i], hashtest_initkeys[j]); + if (shash != hash_values_siphash[j][i]) { + printf("siphash for %u bytes with initial %#"PRIx64"." + "Expected %#"PRIx64", but got %#"PRIx64"\n", + hashtest_key_lens[i], hashtest_initkeys[j], + hash_values_siphash[j][i], shash); + return -1; + } } } diff --git a/lib/hash/meson.build b/lib/hash/meson.build index 277eb9fa9366..913855124d2b 100644 --- a/lib/hash/meson.build +++ b/lib/hash/meson.build @@ -6,6 +6,7 @@ headers = files( 'rte_hash_crc.h', 'rte_hash.h', 'rte_jhash.h', + 'rte_siphash.h', 'rte_thash.h', 'rte_thash_gfni.h', ) @@ -21,6 +22,7 @@ sources = files( 'rte_cuckoo_hash.c', 'rte_hash_crc.c', 'rte_fbk_hash.c', + 'rte_siphash.c', 'rte_thash.c', 'rte_thash_gfni.c', ) diff --git a/lib/hash/rte_siphash.c b/lib/hash/rte_siphash.c new file mode 100644 index 000000000000..85f0c54c280e --- /dev/null +++ b/lib/hash/rte_siphash.c @@ -0,0 +1,176 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * + * Based on code reference code licensed as CC0 and MIT. + * Copyright (c) 2012-2022 Jean-Philippe Aumasson + * Copyright (c) 2012-2014 Daniel J. Bernstein + */ + +#include +#include + +#include +#include + +#include "rte_siphash.h" + +static __rte_always_inline uint64_t +rol64(uint64_t x, unsigned int b) +{ + return (x << b) | (x >> (64 - b)); +} + +static __rte_always_inline uint64_t u8to64_le(const uint8_t *p) +{ + uint64_t w; + + memcpy(&w, p, sizeof(w)); + return rte_cpu_to_le_64(w); +} + +static __rte_always_inline uint32_t u8to32_le(const uint8_t *p) +{ + uint32_t w; + + memcpy(&w, p, sizeof(w)); + return rte_cpu_to_le_32(w); +} + +static __rte_always_inline uint16_t u8to16_le(const uint8_t *p) +{ + uint16_t w; + + memcpy(&w, p, sizeof(w)); + return rte_cpu_to_le_16(w); +} + + +/* + * Use a 64bit version of SipHash for both full and + * half versions, The difference is the number of rounds. + */ +#define PREAMBLE(len) \ + uint64_t v0 = UINT64_C(0x736f6d6570736575); \ + uint64_t v1 = UINT64_C(0x646f72616e646f6d); \ + uint64_t v2 = UINT64_C(0x6c7967656e657261); \ + uint64_t v3 = UINT64_C(0x7465646279746573); \ + uint64_t b = (uint64_t)len << 56; \ + v2 ^= init_val; \ + v0 ^= init_val + +#define SIPROUND do { \ + v0 += v1; \ + v1 = rol64(v1, 13); \ + v1 ^= v0; \ + v0 = rol64(v0, 32); \ + v2 += v3; \ + v3 = rol64(v3, 16); \ + v3 ^= v2; \ + v0 += v3; \ + v3 = rol64(v3, 21); \ + v3 ^= v0; \ + v2 += v1; \ + v1 = rol64(v1, 17); \ + v1 ^= v2; \ + v2 = rol64(v2, 32); \ +} while (0) + +uint64_t +rte_siphash(const void *key, uint32_t len, uint64_t init_val) +{ + const uint8_t *data = key; + const uint32_t left = len & 7; + const uint8_t *end = data + (len - left); + + PREAMBLE(len); + + for (; data != end; data += 8) { + uint64_t m = u8to64_le(data); + v3 ^= m; + SIPROUND; + SIPROUND; + v0 ^= m; + } + + switch (left) { + case 7: + b |= ((uint64_t)end[6]) << 48; /* fallthrough */ + case 6: + b |= ((uint64_t)end[5]) << 40; /* fallthrough */ + case 5: + b |= ((uint64_t)end[4]) << 32; /* fallthrough */ + case 4: + b |= u8to32_le(data); + break; + case 3: + b |= ((uint64_t)end[2]) << 16; /* fallthrough */ + case 2: + b |= u8to16_le(data); + break; + case 1: + b |= data[0]; + } + + v3 ^= b; + SIPROUND; + SIPROUND; + + v0 ^= b; + v2 ^= 0xff; + + SIPROUND; + SIPROUND; + SIPROUND; + SIPROUND; + + return (v0 ^ v1) ^ (v2 ^ v3); +} + +uint32_t +rte_hsiphash(const void *key, uint32_t len, uint32_t init_val) +{ + const uint8_t *data = key; + const uint32_t left = len & 7; + const uint8_t *end = data + (len - left); + + PREAMBLE(len); + + for (; data != end; data += 8) { + uint64_t m = u8to64_le(data); + v3 ^= m; + SIPROUND; + v0 ^= m; + } + + + + switch (left) { + case 7: + b |= ((uint64_t)end[6]) << 48; /* fallthrough */ + case 6: + b |= ((uint64_t)end[5]) << 40; /* fallthrough */ + case 5: + b |= ((uint64_t)end[4]) << 32; /* fallthrough */ + case 4: + b |= u8to32_le(data); + break; + case 3: + b |= ((uint64_t)end[2]) << 16; /* fallthrough */ + case 2: + b |= u8to16_le(data); + break; + case 1: + b |= data[0]; + } + + v3 ^= b; + SIPROUND; + + v0 ^= b; + v2 ^= 0xff; + + SIPROUND; + SIPROUND; + SIPROUND; + + return (v0 ^ v1) ^ (v2 ^ v3); +} diff --git a/lib/hash/rte_siphash.h b/lib/hash/rte_siphash.h new file mode 100644 index 000000000000..d9fbbe2e9275 --- /dev/null +++ b/lib/hash/rte_siphash.h @@ -0,0 +1,86 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * + * Based on code reference code licensed as CC0 and MIT. + * Copyright (c) 2012-2022 Jean-Philippe Aumasson + * Copyright (c) 2012-2014 Daniel J. Bernstein + */ + +#ifndef _RTE_SIPHASH_H +#define _RTE_SIPHASH_H + +/** + * @file + * + * SipHash is a family of pseudorandom functions (PRFs) optimized for speed on short messages. + * + * SipHash was designed in 2012 by Jean-Philippe Aumasson and Daniel J. Bernstein as a defense + * against hash-flooding DoS attacks. + * + * SipHash is simpler and faster on short messages than previous cryptographic algorithms, + * such as MACs based on universal hashing. + * Competitive in performance with insecure non-cryptographic algorithms. + * + * Cryptographically secure, with no sign of weakness despite multiple cryptanalysis projects + * by leading cryptographers. + * + * Battle-tested, with successful integration in OSs (Linux kernel, OpenBSD, FreeBSD, FreeRTOS), + * languages (Perl, Python, Ruby, etc.), libraries (OpenSSL libcrypto, Sodium, etc.) + * and applications (Wireguard, Redis, etc.). + * + * As a secure pseudorandom function (a.k.a. keyed hash function), SipHash can also be used as + * a secure message authentication code (MAC). But SipHash is not a hash in the sense of + * general-purpose key-less hash function such as BLAKE3 or SHA-3. + * SipHash should therefore always be used with a secret key in order to be secure. + * siphash functions. + * + */ + +#ifdef __cplusplus +extern "C" { +#endif + +#include + +#include + +/** + * Compute a SipHash-2-4. + * This version is the original version described in the reference version. + * It uses a 64 bit key; does 2 compression rounds, and 4 finalization rounds. + * + * @param data + * Data to perform hash on. + * @param len + * How many bytes to use to calculate hash value. + * @param init_val + * Value to initialize hash generator. + * @return + * 64bit calculated hash value. + */ +__rte_experimental +uint64_t +rte_siphash(const void *data, uint32_t len, uint64_t init_val); + +/** + * Compute Siphash-1-3. + * This is the faster version which is used by Linux and other OS's. + * It uses a 32 bit key; does 1 compression round, and 3 finalization rounds. + * + * @param data + * Data to perform hash on. + * @param len + * How many bytes to use to calculate hash value. + * @param init_val + * Value to initialize hash generator. + * @return + * 32bit calculated hash value. + */ +__rte_experimental +uint32_t +rte_hsiphash(const void *data, uint32_t len, uint32_t init_val); + +#ifdef __cplusplus +} +#endif + +#endif /* _RTE_SIPHASH_H */ diff --git a/lib/hash/version.map b/lib/hash/version.map index 6b2afebf6b46..16221fc1d8c7 100644 --- a/lib/hash/version.map +++ b/lib/hash/version.map @@ -48,3 +48,10 @@ DPDK_24 { local: *; }; + +EXPERIMENTAL { + global: + + rte_siphash; + rte_hsiphash; +};