14a32e5dcSEric Biggers // SPDX-License-Identifier: GPL-2.0
24a32e5dcSEric Biggers /*
34a32e5dcSEric Biggers * OpenSSL/Cryptogams accelerated Poly1305 transform for ARM
44a32e5dcSEric Biggers *
54a32e5dcSEric Biggers * Copyright (C) 2019 Linaro Ltd. <ard.biesheuvel@linaro.org>
64a32e5dcSEric Biggers */
74a32e5dcSEric Biggers
84a32e5dcSEric Biggers #include <asm/hwcap.h>
94a32e5dcSEric Biggers #include <asm/neon.h>
10*52c3e242SEric Biggers #include <asm/simd.h>
114a32e5dcSEric Biggers #include <crypto/internal/poly1305.h>
124a32e5dcSEric Biggers #include <linux/cpufeature.h>
134a32e5dcSEric Biggers #include <linux/jump_label.h>
144a32e5dcSEric Biggers #include <linux/kernel.h>
154a32e5dcSEric Biggers #include <linux/module.h>
164a32e5dcSEric Biggers #include <linux/unaligned.h>
174a32e5dcSEric Biggers
184a32e5dcSEric Biggers asmlinkage void poly1305_block_init_arch(
194a32e5dcSEric Biggers struct poly1305_block_state *state,
204a32e5dcSEric Biggers const u8 raw_key[POLY1305_BLOCK_SIZE]);
214a32e5dcSEric Biggers EXPORT_SYMBOL_GPL(poly1305_block_init_arch);
224a32e5dcSEric Biggers asmlinkage void poly1305_blocks_arm(struct poly1305_block_state *state,
234a32e5dcSEric Biggers const u8 *src, u32 len, u32 hibit);
244a32e5dcSEric Biggers asmlinkage void poly1305_blocks_neon(struct poly1305_block_state *state,
254a32e5dcSEric Biggers const u8 *src, u32 len, u32 hibit);
264a32e5dcSEric Biggers asmlinkage void poly1305_emit_arch(const struct poly1305_state *state,
274a32e5dcSEric Biggers u8 digest[POLY1305_DIGEST_SIZE],
284a32e5dcSEric Biggers const u32 nonce[4]);
294a32e5dcSEric Biggers EXPORT_SYMBOL_GPL(poly1305_emit_arch);
304a32e5dcSEric Biggers
314a32e5dcSEric Biggers static __ro_after_init DEFINE_STATIC_KEY_FALSE(have_neon);
324a32e5dcSEric Biggers
poly1305_blocks_arch(struct poly1305_block_state * state,const u8 * src,unsigned int len,u32 padbit)334a32e5dcSEric Biggers void poly1305_blocks_arch(struct poly1305_block_state *state, const u8 *src,
344a32e5dcSEric Biggers unsigned int len, u32 padbit)
354a32e5dcSEric Biggers {
364a32e5dcSEric Biggers len = round_down(len, POLY1305_BLOCK_SIZE);
374a32e5dcSEric Biggers if (IS_ENABLED(CONFIG_KERNEL_MODE_NEON) &&
38*52c3e242SEric Biggers static_branch_likely(&have_neon) && likely(may_use_simd())) {
394a32e5dcSEric Biggers do {
404a32e5dcSEric Biggers unsigned int todo = min_t(unsigned int, len, SZ_4K);
414a32e5dcSEric Biggers
424a32e5dcSEric Biggers kernel_neon_begin();
434a32e5dcSEric Biggers poly1305_blocks_neon(state, src, todo, padbit);
444a32e5dcSEric Biggers kernel_neon_end();
454a32e5dcSEric Biggers
464a32e5dcSEric Biggers len -= todo;
474a32e5dcSEric Biggers src += todo;
484a32e5dcSEric Biggers } while (len);
494a32e5dcSEric Biggers } else
504a32e5dcSEric Biggers poly1305_blocks_arm(state, src, len, padbit);
514a32e5dcSEric Biggers }
524a32e5dcSEric Biggers EXPORT_SYMBOL_GPL(poly1305_blocks_arch);
534a32e5dcSEric Biggers
poly1305_is_arch_optimized(void)544a32e5dcSEric Biggers bool poly1305_is_arch_optimized(void)
554a32e5dcSEric Biggers {
564a32e5dcSEric Biggers /* We always can use at least the ARM scalar implementation. */
574a32e5dcSEric Biggers return true;
584a32e5dcSEric Biggers }
594a32e5dcSEric Biggers EXPORT_SYMBOL(poly1305_is_arch_optimized);
604a32e5dcSEric Biggers
arm_poly1305_mod_init(void)614a32e5dcSEric Biggers static int __init arm_poly1305_mod_init(void)
624a32e5dcSEric Biggers {
634a32e5dcSEric Biggers if (IS_ENABLED(CONFIG_KERNEL_MODE_NEON) &&
644a32e5dcSEric Biggers (elf_hwcap & HWCAP_NEON))
654a32e5dcSEric Biggers static_branch_enable(&have_neon);
664a32e5dcSEric Biggers return 0;
674a32e5dcSEric Biggers }
684a32e5dcSEric Biggers subsys_initcall(arm_poly1305_mod_init);
694a32e5dcSEric Biggers
arm_poly1305_mod_exit(void)704a32e5dcSEric Biggers static void __exit arm_poly1305_mod_exit(void)
714a32e5dcSEric Biggers {
724a32e5dcSEric Biggers }
734a32e5dcSEric Biggers module_exit(arm_poly1305_mod_exit);
744a32e5dcSEric Biggers
754a32e5dcSEric Biggers MODULE_DESCRIPTION("Accelerated Poly1305 transform for ARM");
764a32e5dcSEric Biggers MODULE_LICENSE("GPL v2");
77