summaryrefslogtreecommitdiff
path: root/arch/x86/crypto/aria_gfni_avx512_glue.c
diff options
context:
space:
mode:
authorTaehee Yoo <ap420073@gmail.com>2023-01-01 12:12:52 +0300
committerHerbert Xu <herbert@gondor.apana.org.au>2023-01-06 12:15:47 +0300
commitc970d42001f2dffd587cc38458fea4130796be43 (patch)
tree20c1189e2f1e6f7958c132d6d63206f1550851fd /arch/x86/crypto/aria_gfni_avx512_glue.c
parent37d8d3ae7a58cb16fa3f4f1992d2ee36bc621438 (diff)
downloadlinux-c970d42001f2dffd587cc38458fea4130796be43.tar.xz
crypto: x86/aria - implement aria-avx512
aria-avx512 implementation uses AVX512 and GFNI. It supports 64way parallel processing. So, byteslicing code is changed to support 64way parallel. And it exports some aria-avx2 functions such as encrypt() and decrypt(). AVX and AVX2 have 16 registers. They should use memory to store/load state because of lack of registers. But AVX512 supports 32 registers. So, it doesn't require store/load in the s-box layer. It means that it can reduce overhead of store/load in the s-box layer. Also code become much simpler. Benchmark with modprobe tcrypt mode=610 num_mb=8192, i3-12100: ARIA-AVX512(128bit and 256bit) testing speed of multibuffer ecb(aria) (ecb-aria-avx512) encryption tcrypt: 1 operation in 1504 cycles (1024 bytes) tcrypt: 1 operation in 4595 cycles (4096 bytes) tcrypt: 1 operation in 1763 cycles (1024 bytes) tcrypt: 1 operation in 5540 cycles (4096 bytes) testing speed of multibuffer ecb(aria) (ecb-aria-avx512) decryption tcrypt: 1 operation in 1502 cycles (1024 bytes) tcrypt: 1 operation in 4615 cycles (4096 bytes) tcrypt: 1 operation in 1759 cycles (1024 bytes) tcrypt: 1 operation in 5554 cycles (4096 bytes) ARIA-AVX2 with GFNI(128bit and 256bit) testing speed of multibuffer ecb(aria) (ecb-aria-avx2) encryption tcrypt: 1 operation in 2003 cycles (1024 bytes) tcrypt: 1 operation in 5867 cycles (4096 bytes) tcrypt: 1 operation in 2358 cycles (1024 bytes) tcrypt: 1 operation in 7295 cycles (4096 bytes) testing speed of multibuffer ecb(aria) (ecb-aria-avx2) decryption tcrypt: 1 operation in 2004 cycles (1024 bytes) tcrypt: 1 operation in 5956 cycles (4096 bytes) tcrypt: 1 operation in 2409 cycles (1024 bytes) tcrypt: 1 operation in 7564 cycles (4096 bytes) Signed-off-by: Taehee Yoo <ap420073@gmail.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Diffstat (limited to 'arch/x86/crypto/aria_gfni_avx512_glue.c')
-rw-r--r--arch/x86/crypto/aria_gfni_avx512_glue.c250
1 files changed, 250 insertions, 0 deletions
diff --git a/arch/x86/crypto/aria_gfni_avx512_glue.c b/arch/x86/crypto/aria_gfni_avx512_glue.c
new file mode 100644
index 000000000000..f4a2208d2638
--- /dev/null
+++ b/arch/x86/crypto/aria_gfni_avx512_glue.c
@@ -0,0 +1,250 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Glue Code for the AVX512/GFNI assembler implementation of the ARIA Cipher
+ *
+ * Copyright (c) 2022 Taehee Yoo <ap420073@gmail.com>
+ */
+
+#include <crypto/algapi.h>
+#include <crypto/internal/simd.h>
+#include <crypto/aria.h>
+#include <linux/crypto.h>
+#include <linux/err.h>
+#include <linux/module.h>
+#include <linux/types.h>
+
+#include "ecb_cbc_helpers.h"
+#include "aria-avx.h"
+
+asmlinkage void aria_gfni_avx512_encrypt_64way(const void *ctx, u8 *dst,
+ const u8 *src);
+asmlinkage void aria_gfni_avx512_decrypt_64way(const void *ctx, u8 *dst,
+ const u8 *src);
+asmlinkage void aria_gfni_avx512_ctr_crypt_64way(const void *ctx, u8 *dst,
+ const u8 *src,
+ u8 *keystream, u8 *iv);
+
+static struct aria_avx_ops aria_ops;
+
+struct aria_avx512_request_ctx {
+ u8 keystream[ARIA_GFNI_AVX512_PARALLEL_BLOCK_SIZE];
+};
+
+static int ecb_do_encrypt(struct skcipher_request *req, const u32 *rkey)
+{
+ ECB_WALK_START(req, ARIA_BLOCK_SIZE, ARIA_AESNI_PARALLEL_BLOCKS);
+ ECB_BLOCK(ARIA_GFNI_AVX512_PARALLEL_BLOCKS, aria_ops.aria_encrypt_64way);
+ ECB_BLOCK(ARIA_AESNI_AVX2_PARALLEL_BLOCKS, aria_ops.aria_encrypt_32way);
+ ECB_BLOCK(ARIA_AESNI_PARALLEL_BLOCKS, aria_ops.aria_encrypt_16way);
+ ECB_BLOCK(1, aria_encrypt);
+ ECB_WALK_END();
+}
+
+static int ecb_do_decrypt(struct skcipher_request *req, const u32 *rkey)
+{
+ ECB_WALK_START(req, ARIA_BLOCK_SIZE, ARIA_AESNI_PARALLEL_BLOCKS);
+ ECB_BLOCK(ARIA_GFNI_AVX512_PARALLEL_BLOCKS, aria_ops.aria_decrypt_64way);
+ ECB_BLOCK(ARIA_AESNI_AVX2_PARALLEL_BLOCKS, aria_ops.aria_decrypt_32way);
+ ECB_BLOCK(ARIA_AESNI_PARALLEL_BLOCKS, aria_ops.aria_decrypt_16way);
+ ECB_BLOCK(1, aria_decrypt);
+ ECB_WALK_END();
+}
+
+static int aria_avx512_ecb_encrypt(struct skcipher_request *req)
+{
+ struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
+ struct aria_ctx *ctx = crypto_skcipher_ctx(tfm);
+
+ return ecb_do_encrypt(req, ctx->enc_key[0]);
+}
+
+static int aria_avx512_ecb_decrypt(struct skcipher_request *req)
+{
+ struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
+ struct aria_ctx *ctx = crypto_skcipher_ctx(tfm);
+
+ return ecb_do_decrypt(req, ctx->dec_key[0]);
+}
+
+static int aria_avx512_set_key(struct crypto_skcipher *tfm, const u8 *key,
+ unsigned int keylen)
+{
+ return aria_set_key(&tfm->base, key, keylen);
+}
+
+static int aria_avx512_ctr_encrypt(struct skcipher_request *req)
+{
+ struct aria_avx512_request_ctx *req_ctx = skcipher_request_ctx(req);
+ struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
+ struct aria_ctx *ctx = crypto_skcipher_ctx(tfm);
+ struct skcipher_walk walk;
+ unsigned int nbytes;
+ int err;
+
+ err = skcipher_walk_virt(&walk, req, false);
+
+ while ((nbytes = walk.nbytes) > 0) {
+ const u8 *src = walk.src.virt.addr;
+ u8 *dst = walk.dst.virt.addr;
+
+ while (nbytes >= ARIA_GFNI_AVX512_PARALLEL_BLOCK_SIZE) {
+ kernel_fpu_begin();
+ aria_ops.aria_ctr_crypt_64way(ctx, dst, src,
+ &req_ctx->keystream[0],
+ walk.iv);
+ kernel_fpu_end();
+ dst += ARIA_GFNI_AVX512_PARALLEL_BLOCK_SIZE;
+ src += ARIA_GFNI_AVX512_PARALLEL_BLOCK_SIZE;
+ nbytes -= ARIA_GFNI_AVX512_PARALLEL_BLOCK_SIZE;
+ }
+
+ while (nbytes >= ARIA_AESNI_AVX2_PARALLEL_BLOCK_SIZE) {
+ kernel_fpu_begin();
+ aria_ops.aria_ctr_crypt_32way(ctx, dst, src,
+ &req_ctx->keystream[0],
+ walk.iv);
+ kernel_fpu_end();
+ dst += ARIA_AESNI_AVX2_PARALLEL_BLOCK_SIZE;
+ src += ARIA_AESNI_AVX2_PARALLEL_BLOCK_SIZE;
+ nbytes -= ARIA_AESNI_AVX2_PARALLEL_BLOCK_SIZE;
+ }
+
+ while (nbytes >= ARIA_AESNI_PARALLEL_BLOCK_SIZE) {
+ kernel_fpu_begin();
+ aria_ops.aria_ctr_crypt_16way(ctx, dst, src,
+ &req_ctx->keystream[0],
+ walk.iv);
+ kernel_fpu_end();
+ dst += ARIA_AESNI_PARALLEL_BLOCK_SIZE;
+ src += ARIA_AESNI_PARALLEL_BLOCK_SIZE;
+ nbytes -= ARIA_AESNI_PARALLEL_BLOCK_SIZE;
+ }
+
+ while (nbytes >= ARIA_BLOCK_SIZE) {
+ memcpy(&req_ctx->keystream[0], walk.iv,
+ ARIA_BLOCK_SIZE);
+ crypto_inc(walk.iv, ARIA_BLOCK_SIZE);
+
+ aria_encrypt(ctx, &req_ctx->keystream[0],
+ &req_ctx->keystream[0]);
+
+ crypto_xor_cpy(dst, src, &req_ctx->keystream[0],
+ ARIA_BLOCK_SIZE);
+ dst += ARIA_BLOCK_SIZE;
+ src += ARIA_BLOCK_SIZE;
+ nbytes -= ARIA_BLOCK_SIZE;
+ }
+
+ if (walk.nbytes == walk.total && nbytes > 0) {
+ memcpy(&req_ctx->keystream[0], walk.iv,
+ ARIA_BLOCK_SIZE);
+ crypto_inc(walk.iv, ARIA_BLOCK_SIZE);
+
+ aria_encrypt(ctx, &req_ctx->keystream[0],
+ &req_ctx->keystream[0]);
+
+ crypto_xor_cpy(dst, src, &req_ctx->keystream[0],
+ nbytes);
+ dst += nbytes;
+ src += nbytes;
+ nbytes = 0;
+ }
+ err = skcipher_walk_done(&walk, nbytes);
+ }
+
+ return err;
+}
+
+static int aria_avx512_init_tfm(struct crypto_skcipher *tfm)
+{
+ crypto_skcipher_set_reqsize(tfm,
+ sizeof(struct aria_avx512_request_ctx));
+
+ return 0;
+}
+
+static struct skcipher_alg aria_algs[] = {
+ {
+ .base.cra_name = "__ecb(aria)",
+ .base.cra_driver_name = "__ecb-aria-avx512",
+ .base.cra_priority = 600,
+ .base.cra_flags = CRYPTO_ALG_INTERNAL,
+ .base.cra_blocksize = ARIA_BLOCK_SIZE,
+ .base.cra_ctxsize = sizeof(struct aria_ctx),
+ .base.cra_module = THIS_MODULE,
+ .min_keysize = ARIA_MIN_KEY_SIZE,
+ .max_keysize = ARIA_MAX_KEY_SIZE,
+ .setkey = aria_avx512_set_key,
+ .encrypt = aria_avx512_ecb_encrypt,
+ .decrypt = aria_avx512_ecb_decrypt,
+ }, {
+ .base.cra_name = "__ctr(aria)",
+ .base.cra_driver_name = "__ctr-aria-avx512",
+ .base.cra_priority = 600,
+ .base.cra_flags = CRYPTO_ALG_INTERNAL |
+ CRYPTO_ALG_SKCIPHER_REQSIZE_LARGE,
+ .base.cra_blocksize = 1,
+ .base.cra_ctxsize = sizeof(struct aria_ctx),
+ .base.cra_module = THIS_MODULE,
+ .min_keysize = ARIA_MIN_KEY_SIZE,
+ .max_keysize = ARIA_MAX_KEY_SIZE,
+ .ivsize = ARIA_BLOCK_SIZE,
+ .chunksize = ARIA_BLOCK_SIZE,
+ .setkey = aria_avx512_set_key,
+ .encrypt = aria_avx512_ctr_encrypt,
+ .decrypt = aria_avx512_ctr_encrypt,
+ .init = aria_avx512_init_tfm,
+ }
+};
+
+static struct simd_skcipher_alg *aria_simd_algs[ARRAY_SIZE(aria_algs)];
+
+static int __init aria_avx512_init(void)
+{
+ const char *feature_name;
+
+ if (!boot_cpu_has(X86_FEATURE_AVX) ||
+ !boot_cpu_has(X86_FEATURE_AVX2) ||
+ !boot_cpu_has(X86_FEATURE_AVX512F) ||
+ !boot_cpu_has(X86_FEATURE_AVX512VL) ||
+ !boot_cpu_has(X86_FEATURE_GFNI) ||
+ !boot_cpu_has(X86_FEATURE_OSXSAVE)) {
+ pr_info("AVX512/GFNI instructions are not detected.\n");
+ return -ENODEV;
+ }
+
+ if (!cpu_has_xfeatures(XFEATURE_MASK_SSE | XFEATURE_MASK_YMM |
+ XFEATURE_MASK_AVX512, &feature_name)) {
+ pr_info("CPU feature '%s' is not supported.\n", feature_name);
+ return -ENODEV;
+ }
+
+ aria_ops.aria_encrypt_16way = aria_aesni_avx_gfni_encrypt_16way;
+ aria_ops.aria_decrypt_16way = aria_aesni_avx_gfni_decrypt_16way;
+ aria_ops.aria_ctr_crypt_16way = aria_aesni_avx_gfni_ctr_crypt_16way;
+ aria_ops.aria_encrypt_32way = aria_aesni_avx2_gfni_encrypt_32way;
+ aria_ops.aria_decrypt_32way = aria_aesni_avx2_gfni_decrypt_32way;
+ aria_ops.aria_ctr_crypt_32way = aria_aesni_avx2_gfni_ctr_crypt_32way;
+ aria_ops.aria_encrypt_64way = aria_gfni_avx512_encrypt_64way;
+ aria_ops.aria_decrypt_64way = aria_gfni_avx512_decrypt_64way;
+ aria_ops.aria_ctr_crypt_64way = aria_gfni_avx512_ctr_crypt_64way;
+
+ return simd_register_skciphers_compat(aria_algs,
+ ARRAY_SIZE(aria_algs),
+ aria_simd_algs);
+}
+
+static void __exit aria_avx512_exit(void)
+{
+ simd_unregister_skciphers(aria_algs, ARRAY_SIZE(aria_algs),
+ aria_simd_algs);
+}
+
+module_init(aria_avx512_init);
+module_exit(aria_avx512_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Taehee Yoo <ap420073@gmail.com>");
+MODULE_DESCRIPTION("ARIA Cipher Algorithm, AVX512/GFNI optimized");
+MODULE_ALIAS_CRYPTO("aria");
+MODULE_ALIAS_CRYPTO("aria-gfni-avx512");