target/linux/generic/backport-5.4/080-wireguard-0003-crypto-x86-chacha-depend-on-generic-chacha-library-i.patch

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Ard Biesheuvel <ardb@kernel.org>
Date: Fri, 8 Nov 2019 13:22:09 +0100
Subject: [PATCH] crypto: x86/chacha - depend on generic chacha library instead
 of crypto driver

commit 28e8d89b1ce8d2e7badfb5f69971dd635acb8863 upstream.

In preparation of extending the x86 ChaCha driver to also expose the ChaCha
library interface, drop the dependency on the chacha_generic crypto driver
as a non-SIMD fallback, and depend on the generic ChaCha library directly.
This way, we only pull in the code we actually need, without registering
a set of ChaCha skciphers that we will never use.

Since turning the FPU on and off is cheap these days, simplify the SIMD
routine by dropping the per-page yield, which makes for a cleaner switch
to the library API as well. This also allows use to invoke the skcipher
walk routines in non-atomic mode.

Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
---
 arch/x86/crypto/chacha_glue.c | 90 ++++++++++++++---------------------
 crypto/Kconfig                |  2 +-
 2 files changed, 36 insertions(+), 56 deletions(-)

--- a/arch/x86/crypto/chacha_glue.c
+++ b/arch/x86/crypto/chacha_glue.c
@@ -123,37 +123,38 @@ static void chacha_dosimd(u32 *state, u8
 	}
 }
 
-static int chacha_simd_stream_xor(struct skcipher_walk *walk,
+static int chacha_simd_stream_xor(struct skcipher_request *req,
 				  const struct chacha_ctx *ctx, const u8 *iv)
 {
 	u32 *state, state_buf[16 + 2] __aligned(8);
-	int next_yield = 4096; /* bytes until next FPU yield */
-	int err = 0;
+	struct skcipher_walk walk;
+	int err;
+
+	err = skcipher_walk_virt(&walk, req, false);
 
 	BUILD_BUG_ON(CHACHA_STATE_ALIGN != 16);
 	state = PTR_ALIGN(state_buf + 0, CHACHA_STATE_ALIGN);
 
-	crypto_chacha_init(state, ctx, iv);
+	chacha_init_generic(state, ctx->key, iv);
 
-	while (walk->nbytes > 0) {
-		unsigned int nbytes = walk->nbytes;
+	while (walk.nbytes > 0) {
+		unsigned int nbytes = walk.nbytes;
 
-		if (nbytes < walk->total) {
-			nbytes = round_down(nbytes, walk->stride);
-			next_yield -= nbytes;
-		}
-
-		chacha_dosimd(state, walk->dst.virt.addr, walk->src.virt.addr,
-			      nbytes, ctx->nrounds);
+		if (nbytes < walk.total)
+			nbytes = round_down(nbytes, walk.stride);
 
-		if (next_yield <= 0) {
-			/* temporarily allow preemption */
-			kernel_fpu_end();
+		if (!crypto_simd_usable()) {
+			chacha_crypt_generic(state, walk.dst.virt.addr,
+					     walk.src.virt.addr, nbytes,
+					     ctx->nrounds);
+		} else {
 			kernel_fpu_begin();
-			next_yield = 4096;
+			chacha_dosimd(state, walk.dst.virt.addr,
+				      walk.src.virt.addr, nbytes,
+				      ctx->nrounds);
+			kernel_fpu_end();
 		}
-
-		err = skcipher_walk_done(walk, walk->nbytes - nbytes);
+		err = skcipher_walk_done(&walk, walk.nbytes - nbytes);
 	}
 
 	return err;
@@ -163,55 +164,34 @@ static int chacha_simd(struct skcipher_r
 {
 	struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
 	struct chacha_ctx *ctx = crypto_skcipher_ctx(tfm);
-	struct skcipher_walk walk;
-	int err;
-
-	if (req->cryptlen <= CHACHA_BLOCK_SIZE || !crypto_simd_usable())
-		return crypto_chacha_crypt(req);
 
-	err = skcipher_walk_virt(&walk, req, true);
-	if (err)
-		return err;
-
-	kernel_fpu_begin();
-	err = chacha_simd_stream_xor(&walk, ctx, req->iv);
-	kernel_fpu_end();
-	return err;
+	return chacha_simd_stream_xor(req, ctx, req->iv);
 }
 
 static int xchacha_simd(struct skcipher_request *req)
 {
 	struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
 	struct chacha_ctx *ctx = crypto_skcipher_ctx(tfm);
-	struct skcipher_walk walk;
-	struct chacha_ctx subctx;
 	u32 *state, state_buf[16 + 2] __aligned(8);
+	struct chacha_ctx subctx;
 	u8 real_iv[16];
-	int err;
-
-	if (req->cryptlen <= CHACHA_BLOCK_SIZE || !crypto_simd_usable())
-		return crypto_xchacha_crypt(req);
-
-	err = skcipher_walk_virt(&walk, req, true);
-	if (err)
-		return err;
 
 	BUILD_BUG_ON(CHACHA_STATE_ALIGN != 16);
 	state = PTR_ALIGN(state_buf + 0, CHACHA_STATE_ALIGN);
-	crypto_chacha_init(state, ctx, req->iv);
+	chacha_init_generic(state, ctx->key, req->iv);
 
-	kernel_fpu_begin();
-
-	hchacha_block_ssse3(state, subctx.key, ctx->nrounds);
+	if (req->cryptlen > CHACHA_BLOCK_SIZE && crypto_simd_usable()) {
+		kernel_fpu_begin();
+		hchacha_block_ssse3(state, subctx.key, ctx->nrounds);
+		kernel_fpu_end();
+	} else {
+		hchacha_block_generic(state, subctx.key, ctx->nrounds);
+	}
 	subctx.nrounds = ctx->nrounds;
 
 	memcpy(&real_iv[0], req->iv + 24, 8);
 	memcpy(&real_iv[8], req->iv + 16, 8);
-	err = chacha_simd_stream_xor(&walk, &subctx, real_iv);
-
-	kernel_fpu_end();
-
-	return err;
+	return chacha_simd_stream_xor(req, &subctx, real_iv);
 }
 
 static struct skcipher_alg algs[] = {
@@ -227,7 +207,7 @@ static struct skcipher_alg algs[] = {
 		.max_keysize		= CHACHA_KEY_SIZE,
 		.ivsize			= CHACHA_IV_SIZE,
 		.chunksize		= CHACHA_BLOCK_SIZE,
-		.setkey			= crypto_chacha20_setkey,
+		.setkey			= chacha20_setkey,
 		.encrypt		= chacha_simd,
 		.decrypt		= chacha_simd,
 	}, {
@@ -242,7 +222,7 @@ static struct skcipher_alg algs[] = {
 		.max_keysize		= CHACHA_KEY_SIZE,
 		.ivsize			= XCHACHA_IV_SIZE,
 		.chunksize		= CHACHA_BLOCK_SIZE,
-		.setkey			= crypto_chacha20_setkey,
+		.setkey			= chacha20_setkey,
 		.encrypt		= xchacha_simd,
 		.decrypt		= xchacha_simd,
 	}, {
@@ -257,7 +237,7 @@ static struct skcipher_alg algs[] = {
 		.max_keysize		= CHACHA_KEY_SIZE,
 		.ivsize			= XCHACHA_IV_SIZE,
 		.chunksize		= CHACHA_BLOCK_SIZE,
-		.setkey			= crypto_chacha12_setkey,
+		.setkey			= chacha12_setkey,
 		.encrypt		= xchacha_simd,
 		.decrypt		= xchacha_simd,
 	},
--- a/crypto/Kconfig
+++ b/crypto/Kconfig
@@ -1417,7 +1417,7 @@ config CRYPTO_CHACHA20_X86_64
 	tristate "ChaCha stream cipher algorithms (x86_64/SSSE3/AVX2/AVX-512VL)"
 	depends on X86 && 64BIT
 	select CRYPTO_BLKCIPHER
-	select CRYPTO_CHACHA20
+	select CRYPTO_LIB_CHACHA_GENERIC
 	help
 	  SSSE3, AVX2, and AVX-512VL optimized implementations of the ChaCha20,
 	  XChaCha20, and XChaCha12 stream ciphers.