[PATCH 0/3] crypto: [sha] Restructure x86 sha trasforms glue code to expose all available transform algorithms

[PATCH 0/3] crypto: [sha] Restructure x86 sha trasforms glue code to expose all available transform algorithms

[Tim Chen]

During the posting of patches of new x86 sha1 and sha256 algorithms based
on new x86 SHA extensions, I got requests to expose all the available
sha transform implementations that are based on SSSE3, AVX and AVX2 as
separate individual crypto driver.  The result is this series of patches.
I've also performed clean up of the glue code to make them easier to
maintain.

Thanks.

Tim

Pre-req patch series:
[1] https://lkml.org/lkml/2015/9/10/554

Tim Chen (3):
  crypto: [sha] Restructure x86 sha1 glue code to expose all the
    available sha1 transforms
  crypto: [sha] Restructure x86 sha256 glue code to expose all the
    available sha256 transforms
  crypto: [sha] Restructure x86 sha512 glue code to expose all the
    available sha512 transforms

 arch/x86/crypto/sha1_ssse3_glue.c   | 324 +++++++++++++++++++++++++++--------
 arch/x86/crypto/sha256_ssse3_glue.c | 331 ++++++++++++++++++++++++++++++------
 arch/x86/crypto/sha512_ssse3_glue.c | 249 ++++++++++++++++++++++-----
 3 files changed, 735 insertions(+), 169 deletions(-)

-- 
1.7.11.7

Re: [PATCH 0/3] crypto: [sha] Restructure x86 sha trasforms glue code to expose all available transform algorithms

[Herbert Xu]

On Wed, Sep 16, 2015 at 04:34:32PM -0700, Tim Chen wrote:
> 
> During the posting of patches of new x86 sha1 and sha256 algorithms based
> on new x86 SHA extensions, I got requests to expose all the available
> sha transform implementations that are based on SSSE3, AVX and AVX2 as
> separate individual crypto driver.  The result is this series of patches.
> I've also performed clean up of the glue code to make them easier to
> maintain.

All applied.  Thanks.
-- 
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

[PATCH 1/3] crypto: [sha] Restructure x86 sha1 glue code to expose all the available sha1 transforms

[Tim Chen]

Restructure the x86 sha1 glue code so we will expose sha1 transforms based
on SSSE3, AVX, AVX2 or SHA-NI extension as separate individual drivers
when cpu provides such support. This will make it easy for alternative
algorithms to be used if desired and makes the code cleaner and easier
to maintain.

Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
---
 arch/x86/crypto/sha1_ssse3_glue.c | 324 +++++++++++++++++++++++++++++---------
 1 file changed, 250 insertions(+), 74 deletions(-)

diff --git a/arch/x86/crypto/sha1_ssse3_glue.c b/arch/x86/crypto/sha1_ssse3_glue.c
index 98be8cc..c934197 100644
--- a/arch/x86/crypto/sha1_ssse3_glue.c
+++ b/arch/x86/crypto/sha1_ssse3_glue.c
@@ -31,28 +31,11 @@
 #include <crypto/sha1_base.h>
 #include <asm/fpu/api.h>
 
+typedef void (sha1_transform_fn)(u32 *digest, const char *data,
+				unsigned int rounds);
 
-asmlinkage void sha1_transform_ssse3(u32 *digest, const char *data,
-				     unsigned int rounds);
-#ifdef CONFIG_AS_AVX
-asmlinkage void sha1_transform_avx(u32 *digest, const char *data,
-				   unsigned int rounds);
-#endif
-#ifdef CONFIG_AS_AVX2
-#define SHA1_AVX2_BLOCK_OPTSIZE	4	/* optimal 4*64 bytes of SHA1 blocks */
-
-asmlinkage void sha1_transform_avx2(u32 *digest, const char *data,
-				    unsigned int rounds);
-#endif
-#ifdef CONFIG_AS_SHA1_NI
-asmlinkage void sha1_ni_transform(u32 *digest, const char *data,
-				   unsigned int rounds);
-#endif
-
-static void (*sha1_transform_asm)(u32 *, const char *, unsigned int);
-
-static int sha1_ssse3_update(struct shash_desc *desc, const u8 *data,
-			     unsigned int len)
+static int sha1_update(struct shash_desc *desc, const u8 *data,
+			     unsigned int len, sha1_transform_fn *sha1_xform)
 {
 	struct sha1_state *sctx = shash_desc_ctx(desc);
 
@@ -65,14 +48,14 @@ static int sha1_ssse3_update(struct shash_desc *desc, const u8 *data,
 
 	kernel_fpu_begin();
 	sha1_base_do_update(desc, data, len,
-			    (sha1_block_fn *)sha1_transform_asm);
+			    (sha1_block_fn *)sha1_xform);
 	kernel_fpu_end();
 
 	return 0;
 }
 
-static int sha1_ssse3_finup(struct shash_desc *desc, const u8 *data,
-			      unsigned int len, u8 *out)
+static int sha1_finup(struct shash_desc *desc, const u8 *data,
+		      unsigned int len, u8 *out, sha1_transform_fn *sha1_xform)
 {
 	if (!irq_fpu_usable())
 		return crypto_sha1_finup(desc, data, len, out);
@@ -80,32 +63,37 @@ static int sha1_ssse3_finup(struct shash_desc *desc, const u8 *data,
 	kernel_fpu_begin();
 	if (len)
 		sha1_base_do_update(desc, data, len,
-				    (sha1_block_fn *)sha1_transform_asm);
-	sha1_base_do_finalize(desc, (sha1_block_fn *)sha1_transform_asm);
+				    (sha1_block_fn *)sha1_xform);
+	sha1_base_do_finalize(desc, (sha1_block_fn *)sha1_xform);
 	kernel_fpu_end();
 
 	return sha1_base_finish(desc, out);
 }
 
-/* Add padding and return the message digest. */
-static int sha1_ssse3_final(struct shash_desc *desc, u8 *out)
+asmlinkage void sha1_transform_ssse3(u32 *digest, const char *data,
+				     unsigned int rounds);
+
+static int sha1_ssse3_update(struct shash_desc *desc, const u8 *data,
+			     unsigned int len)
 {
-	return sha1_ssse3_finup(desc, NULL, 0, out);
+	return sha1_update(desc, data, len,
+			(sha1_transform_fn *) sha1_transform_ssse3);
 }
 
-#ifdef CONFIG_AS_AVX2
-static void sha1_apply_transform_avx2(u32 *digest, const char *data,
-				unsigned int rounds)
+static int sha1_ssse3_finup(struct shash_desc *desc, const u8 *data,
+			      unsigned int len, u8 *out)
 {
-	/* Select the optimal transform based on data block size */
-	if (rounds >= SHA1_AVX2_BLOCK_OPTSIZE)
-		sha1_transform_avx2(digest, data, rounds);
-	else
-		sha1_transform_avx(digest, data, rounds);
+	return sha1_finup(desc, data, len, out,
+			(sha1_transform_fn *) sha1_transform_ssse3);
+}
+
+/* Add padding and return the message digest. */
+static int sha1_ssse3_final(struct shash_desc *desc, u8 *out)
+{
+	return sha1_ssse3_finup(desc, NULL, 0, out);
 }
-#endif
 
-static struct shash_alg alg = {
+static struct shash_alg sha1_ssse3_alg = {
 	.digestsize	=	SHA1_DIGEST_SIZE,
 	.init		=	sha1_base_init,
 	.update		=	sha1_ssse3_update,
@@ -114,7 +102,7 @@ static struct shash_alg alg = {
 	.descsize	=	sizeof(struct sha1_state),
 	.base		=	{
 		.cra_name	=	"sha1",
-		.cra_driver_name=	"sha1-ssse3",
+		.cra_driver_name =	"sha1-ssse3",
 		.cra_priority	=	150,
 		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
 		.cra_blocksize	=	SHA1_BLOCK_SIZE,
@@ -122,8 +110,60 @@ static struct shash_alg alg = {
 	}
 };
 
+static int register_sha1_ssse3(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SSSE3))
+		return crypto_register_shash(&sha1_ssse3_alg);
+	return 0;
+}
+
+static void unregister_sha1_ssse3(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SSSE3))
+		crypto_unregister_shash(&sha1_ssse3_alg);
+}
+
 #ifdef CONFIG_AS_AVX
-static bool __init avx_usable(void)
+asmlinkage void sha1_transform_avx(u32 *digest, const char *data,
+				   unsigned int rounds);
+
+static int sha1_avx_update(struct shash_desc *desc, const u8 *data,
+			     unsigned int len)
+{
+	return sha1_update(desc, data, len,
+			(sha1_transform_fn *) sha1_transform_avx);
+}
+
+static int sha1_avx_finup(struct shash_desc *desc, const u8 *data,
+			      unsigned int len, u8 *out)
+{
+	return sha1_finup(desc, data, len, out,
+			(sha1_transform_fn *) sha1_transform_avx);
+}
+
+static int sha1_avx_final(struct shash_desc *desc, u8 *out)
+{
+	return sha1_avx_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha1_avx_alg = {
+	.digestsize	=	SHA1_DIGEST_SIZE,
+	.init		=	sha1_base_init,
+	.update		=	sha1_avx_update,
+	.final		=	sha1_avx_final,
+	.finup		=	sha1_avx_finup,
+	.descsize	=	sizeof(struct sha1_state),
+	.base		=	{
+		.cra_name	=	"sha1",
+		.cra_driver_name =	"sha1-avx",
+		.cra_priority	=	160,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA1_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+};
+
+static bool avx_usable(void)
 {
 	if (!cpu_has_xfeatures(XSTATE_SSE | XSTATE_YMM, NULL)) {
 		if (cpu_has_avx)
@@ -134,61 +174,197 @@ static bool __init avx_usable(void)
 	return true;
 }
 
-#ifdef CONFIG_AS_AVX2
-static bool __init avx2_usable(void)
+static int register_sha1_avx(void)
+{
+	if (avx_usable())
+		return crypto_register_shash(&sha1_avx_alg);
+	return 0;
+}
+
+static void unregister_sha1_avx(void)
+{
+	if (avx_usable())
+		crypto_unregister_shash(&sha1_avx_alg);
+}
+
+#else  /* CONFIG_AS_AVX */
+static inline int register_sha1_avx(void) { return 0; }
+static inline void unregister_sha1_avx(void) { }
+#endif /* CONFIG_AS_AVX */
+
+
+#if defined(CONFIG_AS_AVX2) && (CONFIG_AS_AVX)
+#define SHA1_AVX2_BLOCK_OPTSIZE	4	/* optimal 4*64 bytes of SHA1 blocks */
+
+asmlinkage void sha1_transform_avx2(u32 *digest, const char *data,
+				    unsigned int rounds);
+
+static bool avx2_usable(void)
 {
-	if (avx_usable() && cpu_has_avx2 && boot_cpu_has(X86_FEATURE_BMI1) &&
-	    boot_cpu_has(X86_FEATURE_BMI2))
+	if (avx_usable() && boot_cpu_has(X86_FEATURE_AVX2)
+		&& boot_cpu_has(X86_FEATURE_BMI1)
+		&& boot_cpu_has(X86_FEATURE_BMI2))
 		return true;
 
 	return false;
 }
-#endif
-#endif
 
-static int __init sha1_ssse3_mod_init(void)
+static void sha1_apply_transform_avx2(u32 *digest, const char *data,
+				unsigned int rounds)
 {
-	char *algo_name;
+	/* Select the optimal transform based on data block size */
+	if (rounds >= SHA1_AVX2_BLOCK_OPTSIZE)
+		sha1_transform_avx2(digest, data, rounds);
+	else
+		sha1_transform_avx(digest, data, rounds);
+}
 
-	/* test for SSSE3 first */
-	if (cpu_has_ssse3) {
-		sha1_transform_asm = sha1_transform_ssse3;
-		algo_name = "SSSE3";
-	}
+static int sha1_avx2_update(struct shash_desc *desc, const u8 *data,
+			     unsigned int len)
+{
+	return sha1_update(desc, data, len,
+		(sha1_transform_fn *) sha1_apply_transform_avx2);
+}
 
-#ifdef CONFIG_AS_AVX
-	/* allow AVX to override SSSE3, it's a little faster */
-	if (avx_usable()) {
-		sha1_transform_asm = sha1_transform_avx;
-		algo_name = "AVX";
-#ifdef CONFIG_AS_AVX2
-		/* allow AVX2 to override AVX, it's a little faster */
-		if (avx2_usable()) {
-			sha1_transform_asm = sha1_apply_transform_avx2;
-			algo_name = "AVX2";
-		}
-#endif
+static int sha1_avx2_finup(struct shash_desc *desc, const u8 *data,
+			      unsigned int len, u8 *out)
+{
+	return sha1_finup(desc, data, len, out,
+		(sha1_transform_fn *) sha1_apply_transform_avx2);
+}
+
+static int sha1_avx2_final(struct shash_desc *desc, u8 *out)
+{
+	return sha1_avx2_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha1_avx2_alg = {
+	.digestsize	=	SHA1_DIGEST_SIZE,
+	.init		=	sha1_base_init,
+	.update		=	sha1_avx2_update,
+	.final		=	sha1_avx2_final,
+	.finup		=	sha1_avx2_finup,
+	.descsize	=	sizeof(struct sha1_state),
+	.base		=	{
+		.cra_name	=	"sha1",
+		.cra_driver_name =	"sha1-avx2",
+		.cra_priority	=	170,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA1_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
 	}
+};
+
+static int register_sha1_avx2(void)
+{
+	if (avx2_usable())
+		return crypto_register_shash(&sha1_avx2_alg);
+	return 0;
+}
+
+static void unregister_sha1_avx2(void)
+{
+	if (avx2_usable())
+		crypto_unregister_shash(&sha1_avx2_alg);
+}
+
+#else
+static inline int register_sha1_avx2(void) { return 0; }
+static inline void unregister_sha1_avx2(void) { }
 #endif
+
 #ifdef CONFIG_AS_SHA1_NI
-	if (boot_cpu_has(X86_FEATURE_SHA_NI)) {
-		sha1_transform_asm = sha1_ni_transform;
-		algo_name = "SHA-NI";
+asmlinkage void sha1_ni_transform(u32 *digest, const char *data,
+				   unsigned int rounds);
+
+static int sha1_ni_update(struct shash_desc *desc, const u8 *data,
+			     unsigned int len)
+{
+	return sha1_update(desc, data, len,
+		(sha1_transform_fn *) sha1_ni_transform);
+}
+
+static int sha1_ni_finup(struct shash_desc *desc, const u8 *data,
+			      unsigned int len, u8 *out)
+{
+	return sha1_finup(desc, data, len, out,
+		(sha1_transform_fn *) sha1_ni_transform);
+}
+
+static int sha1_ni_final(struct shash_desc *desc, u8 *out)
+{
+	return sha1_ni_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha1_ni_alg = {
+	.digestsize	=	SHA1_DIGEST_SIZE,
+	.init		=	sha1_base_init,
+	.update		=	sha1_ni_update,
+	.final		=	sha1_ni_final,
+	.finup		=	sha1_ni_finup,
+	.descsize	=	sizeof(struct sha1_state),
+	.base		=	{
+		.cra_name	=	"sha1",
+		.cra_driver_name =	"sha1-ni",
+		.cra_priority	=	250,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA1_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
 	}
+};
+
+static int register_sha1_ni(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SHA_NI))
+		return crypto_register_shash(&sha1_ni_alg);
+	return 0;
+}
+
+static void unregister_sha1_ni(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SHA_NI))
+		crypto_unregister_shash(&sha1_ni_alg);
+}
+
+#else
+static inline int register_sha1_ni(void) { return 0; }
+static inline void unregister_sha1_ni(void) { }
 #endif
 
-	if (sha1_transform_asm) {
-		pr_info("Using %s optimized SHA-1 implementation\n", algo_name);
-		return crypto_register_shash(&alg);
+static int __init sha1_ssse3_mod_init(void)
+{
+	if (register_sha1_ssse3())
+		goto fail;
+
+	if (register_sha1_avx()) {
+		unregister_sha1_ssse3();
+		goto fail;
 	}
-	pr_info("Neither AVX nor AVX2 nor SSSE3/SHA-NI is available/usable.\n");
 
+	if (register_sha1_avx2()) {
+		unregister_sha1_avx();
+		unregister_sha1_ssse3();
+		goto fail;
+	}
+
+	if (register_sha1_ni()) {
+		unregister_sha1_avx2();
+		unregister_sha1_avx();
+		unregister_sha1_ssse3();
+		goto fail;
+	}
+
+	return 0;
+fail:
 	return -ENODEV;
 }
 
 static void __exit sha1_ssse3_mod_fini(void)
 {
-	crypto_unregister_shash(&alg);
+	unregister_sha1_ni();
+	unregister_sha1_avx2();
+	unregister_sha1_avx();
+	unregister_sha1_ssse3();
 }
 
 module_init(sha1_ssse3_mod_init);
-- 
1.7.11.7

[PATCH 2/3] crypto: [sha] Restructure x86 sha256 glue code to expose all the available sha256 transforms

[Tim Chen]

Restructure the x86 sha256 glue code so we will expose sha256 transforms
based on SSSE3, AVX, AVX2 or SHA-NI extension as separate individual
drivers when cpu provides such support. This will make it easy for
alternative algorithms to be used if desired and makes the code cleaner
and easier to maintain.

Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
---
 arch/x86/crypto/sha256_ssse3_glue.c | 331 ++++++++++++++++++++++++++++++------
 1 file changed, 281 insertions(+), 50 deletions(-)

diff --git a/arch/x86/crypto/sha256_ssse3_glue.c b/arch/x86/crypto/sha256_ssse3_glue.c
index 9c7b22c..863e2f6 100644
--- a/arch/x86/crypto/sha256_ssse3_glue.c
+++ b/arch/x86/crypto/sha256_ssse3_glue.c
@@ -42,23 +42,10 @@
 
 asmlinkage void sha256_transform_ssse3(u32 *digest, const char *data,
 				       u64 rounds);
-#ifdef CONFIG_AS_AVX
-asmlinkage void sha256_transform_avx(u32 *digest, const char *data,
-				     u64 rounds);
-#endif
-#ifdef CONFIG_AS_AVX2
-asmlinkage void sha256_transform_rorx(u32 *digest, const char *data,
-				      u64 rounds);
-#endif
-#ifdef CONFIG_AS_SHA256_NI
-asmlinkage void sha256_ni_transform(u32 *digest, const char *data,
-				   u64 rounds); /*unsigned int rounds);*/
-#endif
+typedef void (sha256_transform_fn)(u32 *digest, const char *data, u64 rounds);
 
-static void (*sha256_transform_asm)(u32 *, const char *, u64);
-
-static int sha256_ssse3_update(struct shash_desc *desc, const u8 *data,
-			     unsigned int len)
+static int sha256_update(struct shash_desc *desc, const u8 *data,
+			 unsigned int len, sha256_transform_fn *sha256_xform)
 {
 	struct sha256_state *sctx = shash_desc_ctx(desc);
 
@@ -71,14 +58,14 @@ static int sha256_ssse3_update(struct shash_desc *desc, const u8 *data,
 
 	kernel_fpu_begin();
 	sha256_base_do_update(desc, data, len,
-			      (sha256_block_fn *)sha256_transform_asm);
+			      (sha256_block_fn *)sha256_xform);
 	kernel_fpu_end();
 
 	return 0;
 }
 
-static int sha256_ssse3_finup(struct shash_desc *desc, const u8 *data,
-			      unsigned int len, u8 *out)
+static int sha256_finup(struct shash_desc *desc, const u8 *data,
+	      unsigned int len, u8 *out, sha256_transform_fn *sha256_xform)
 {
 	if (!irq_fpu_usable())
 		return crypto_sha256_finup(desc, data, len, out);
@@ -86,20 +73,32 @@ static int sha256_ssse3_finup(struct shash_desc *desc, const u8 *data,
 	kernel_fpu_begin();
 	if (len)
 		sha256_base_do_update(desc, data, len,
-				      (sha256_block_fn *)sha256_transform_asm);
-	sha256_base_do_finalize(desc, (sha256_block_fn *)sha256_transform_asm);
+				      (sha256_block_fn *)sha256_xform);
+	sha256_base_do_finalize(desc, (sha256_block_fn *)sha256_xform);
 	kernel_fpu_end();
 
 	return sha256_base_finish(desc, out);
 }
 
+static int sha256_ssse3_update(struct shash_desc *desc, const u8 *data,
+			 unsigned int len)
+{
+	return sha256_update(desc, data, len, sha256_transform_ssse3);
+}
+
+static int sha256_ssse3_finup(struct shash_desc *desc, const u8 *data,
+	      unsigned int len, u8 *out)
+{
+	return sha256_finup(desc, data, len, out, sha256_transform_ssse3);
+}
+
 /* Add padding and return the message digest. */
 static int sha256_ssse3_final(struct shash_desc *desc, u8 *out)
 {
 	return sha256_ssse3_finup(desc, NULL, 0, out);
 }
 
-static struct shash_alg algs[] = { {
+static struct shash_alg sha256_ssse3_algs[] = { {
 	.digestsize	=	SHA256_DIGEST_SIZE,
 	.init		=	sha256_base_init,
 	.update		=	sha256_ssse3_update,
@@ -131,8 +130,75 @@ static struct shash_alg algs[] = { {
 	}
 } };
 
+static int register_sha256_ssse3(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SSSE3))
+		return crypto_register_shashes(sha256_ssse3_algs,
+				ARRAY_SIZE(sha256_ssse3_algs));
+	return 0;
+}
+
+static void unregister_sha256_ssse3(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SSSE3))
+		crypto_unregister_shashes(sha256_ssse3_algs,
+				ARRAY_SIZE(sha256_ssse3_algs));
+}
+
 #ifdef CONFIG_AS_AVX
-static bool __init avx_usable(void)
+asmlinkage void sha256_transform_avx(u32 *digest, const char *data,
+				     u64 rounds);
+
+static int sha256_avx_update(struct shash_desc *desc, const u8 *data,
+			 unsigned int len)
+{
+	return sha256_update(desc, data, len, sha256_transform_avx);
+}
+
+static int sha256_avx_finup(struct shash_desc *desc, const u8 *data,
+		      unsigned int len, u8 *out)
+{
+	return sha256_finup(desc, data, len, out, sha256_transform_avx);
+}
+
+static int sha256_avx_final(struct shash_desc *desc, u8 *out)
+{
+	return sha256_avx_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha256_avx_algs[] = { {
+	.digestsize	=	SHA256_DIGEST_SIZE,
+	.init		=	sha256_base_init,
+	.update		=	sha256_avx_update,
+	.final		=	sha256_avx_final,
+	.finup		=	sha256_avx_finup,
+	.descsize	=	sizeof(struct sha256_state),
+	.base		=	{
+		.cra_name	=	"sha256",
+		.cra_driver_name =	"sha256-avx",
+		.cra_priority	=	160,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA256_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+}, {
+	.digestsize	=	SHA224_DIGEST_SIZE,
+	.init		=	sha224_base_init,
+	.update		=	sha256_avx_update,
+	.final		=	sha256_avx_final,
+	.finup		=	sha256_avx_finup,
+	.descsize	=	sizeof(struct sha256_state),
+	.base		=	{
+		.cra_name	=	"sha224",
+		.cra_driver_name =	"sha224-avx",
+		.cra_priority	=	160,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA224_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+} };
+
+static bool avx_usable(void)
 {
 	if (!cpu_has_xfeatures(XSTATE_SSE | XSTATE_YMM, NULL)) {
 		if (cpu_has_avx)
@@ -142,51 +208,216 @@ static bool __init avx_usable(void)
 
 	return true;
 }
-#endif
 
-static int __init sha256_ssse3_mod_init(void)
+static int register_sha256_avx(void)
 {
-	char *algo;
+	if (avx_usable())
+		return crypto_register_shashes(sha256_avx_algs,
+				ARRAY_SIZE(sha256_avx_algs));
+	return 0;
+}
 
-	/* test for SSSE3 first */
-	if (cpu_has_ssse3) {
-		sha256_transform_asm = sha256_transform_ssse3;
-		algo = "SSSE3";
-	}
+static void unregister_sha256_avx(void)
+{
+	if (avx_usable())
+		crypto_unregister_shashes(sha256_avx_algs,
+				ARRAY_SIZE(sha256_avx_algs));
+}
 
-#ifdef CONFIG_AS_AVX
-	/* allow AVX to override SSSE3, it's a little faster */
-	if (avx_usable()) {
-		sha256_transform_asm = sha256_transform_avx;
-		algo = "AVX";
-#ifdef CONFIG_AS_AVX2
-		if (boot_cpu_has(X86_FEATURE_AVX2) &&
-		    boot_cpu_has(X86_FEATURE_BMI2)) {
-			sha256_transform_asm = sha256_transform_rorx;
-			algo = "AVX2";
-		}
+#else
+static inline int register_sha256_avx(void) { return 0; }
+static inline void unregister_sha256_avx(void) { }
 #endif
+
+#if defined(CONFIG_AS_AVX2) && defined(CONFIG_AS_AVX)
+asmlinkage void sha256_transform_rorx(u32 *digest, const char *data,
+				      u64 rounds);
+
+static int sha256_avx2_update(struct shash_desc *desc, const u8 *data,
+			 unsigned int len)
+{
+	return sha256_update(desc, data, len, sha256_transform_rorx);
+}
+
+static int sha256_avx2_finup(struct shash_desc *desc, const u8 *data,
+		      unsigned int len, u8 *out)
+{
+	return sha256_finup(desc, data, len, out, sha256_transform_rorx);
+}
+
+static int sha256_avx2_final(struct shash_desc *desc, u8 *out)
+{
+	return sha256_avx2_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha256_avx2_algs[] = { {
+	.digestsize	=	SHA256_DIGEST_SIZE,
+	.init		=	sha256_base_init,
+	.update		=	sha256_avx2_update,
+	.final		=	sha256_avx2_final,
+	.finup		=	sha256_avx2_finup,
+	.descsize	=	sizeof(struct sha256_state),
+	.base		=	{
+		.cra_name	=	"sha256",
+		.cra_driver_name =	"sha256-avx2",
+		.cra_priority	=	170,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA256_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
 	}
+}, {
+	.digestsize	=	SHA224_DIGEST_SIZE,
+	.init		=	sha224_base_init,
+	.update		=	sha256_avx2_update,
+	.final		=	sha256_avx2_final,
+	.finup		=	sha256_avx2_finup,
+	.descsize	=	sizeof(struct sha256_state),
+	.base		=	{
+		.cra_name	=	"sha224",
+		.cra_driver_name =	"sha224-avx2",
+		.cra_priority	=	170,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA224_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+} };
+
+static bool avx2_usable(void)
+{
+	if (avx_usable() && boot_cpu_has(X86_FEATURE_AVX2) &&
+		    boot_cpu_has(X86_FEATURE_BMI2))
+		return true;
+
+	return false;
+}
+
+static int register_sha256_avx2(void)
+{
+	if (avx2_usable())
+		return crypto_register_shashes(sha256_avx2_algs,
+				ARRAY_SIZE(sha256_avx2_algs));
+	return 0;
+}
+
+static void unregister_sha256_avx2(void)
+{
+	if (avx2_usable())
+		crypto_unregister_shashes(sha256_avx2_algs,
+				ARRAY_SIZE(sha256_avx2_algs));
+}
+
+#else
+static inline int register_sha256_avx2(void) { return 0; }
+static inline void unregister_sha256_avx2(void) { }
 #endif
+
 #ifdef CONFIG_AS_SHA256_NI
-	if (boot_cpu_has(X86_FEATURE_SHA_NI)) {
-		sha256_transform_asm = sha256_ni_transform;
-		algo = "SHA-256-NI";
+asmlinkage void sha256_ni_transform(u32 *digest, const char *data,
+				   u64 rounds); /*unsigned int rounds);*/
+
+static int sha256_ni_update(struct shash_desc *desc, const u8 *data,
+			 unsigned int len)
+{
+	return sha256_update(desc, data, len, sha256_ni_transform);
+}
+
+static int sha256_ni_finup(struct shash_desc *desc, const u8 *data,
+		      unsigned int len, u8 *out)
+{
+	return sha256_finup(desc, data, len, out, sha256_ni_transform);
+}
+
+static int sha256_ni_final(struct shash_desc *desc, u8 *out)
+{
+	return sha256_ni_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha256_ni_algs[] = { {
+	.digestsize	=	SHA256_DIGEST_SIZE,
+	.init		=	sha256_base_init,
+	.update		=	sha256_ni_update,
+	.final		=	sha256_ni_final,
+	.finup		=	sha256_ni_finup,
+	.descsize	=	sizeof(struct sha256_state),
+	.base		=	{
+		.cra_name	=	"sha256",
+		.cra_driver_name =	"sha256-ni",
+		.cra_priority	=	250,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA256_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+}, {
+	.digestsize	=	SHA224_DIGEST_SIZE,
+	.init		=	sha224_base_init,
+	.update		=	sha256_ni_update,
+	.final		=	sha256_ni_final,
+	.finup		=	sha256_ni_finup,
+	.descsize	=	sizeof(struct sha256_state),
+	.base		=	{
+		.cra_name	=	"sha224",
+		.cra_driver_name =	"sha224-ni",
+		.cra_priority	=	250,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA224_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
 	}
+} };
+
+static int register_sha256_ni(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SHA_NI))
+		return crypto_register_shashes(sha256_ni_algs,
+				ARRAY_SIZE(sha256_ni_algs));
+	return 0;
+}
+
+static void unregister_sha256_ni(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SHA_NI))
+		crypto_unregister_shashes(sha256_ni_algs,
+				ARRAY_SIZE(sha256_ni_algs));
+}
+
+#else
+static inline int register_sha256_ni(void) { return 0; }
+static inline void unregister_sha256_ni(void) { }
 #endif
 
-	if (sha256_transform_asm) {
-		pr_info("Using %s optimized SHA-256 implementation\n", algo);
-		return crypto_register_shashes(algs, ARRAY_SIZE(algs));
+static int __init sha256_ssse3_mod_init(void)
+{
+	if (register_sha256_ssse3())
+		goto fail;
+
+	if (register_sha256_avx()) {
+		unregister_sha256_ssse3();
+		goto fail;
 	}
-	pr_info("Neither AVX nor SSSE3/SHA-NI is available/usable.\n");
 
+	if (register_sha256_avx2()) {
+		unregister_sha256_avx();
+		unregister_sha256_ssse3();
+		goto fail;
+	}
+
+	if (register_sha256_ni()) {
+		unregister_sha256_avx2();
+		unregister_sha256_avx();
+		unregister_sha256_ssse3();
+		goto fail;
+	}
+
+	return 0;
+fail:
 	return -ENODEV;
 }
 
 static void __exit sha256_ssse3_mod_fini(void)
 {
-	crypto_unregister_shashes(algs, ARRAY_SIZE(algs));
+	unregister_sha256_ni();
+	unregister_sha256_avx2();
+	unregister_sha256_avx();
+	unregister_sha256_ssse3();
 }
 
 module_init(sha256_ssse3_mod_init);
-- 
1.7.11.7

[PATCH 3/3] crypto: [sha] Restructure x86 sha512 glue code to expose all the available sha512 transforms

[Tim Chen]

Restructure the x86 sha512 glue code so we will expose sha512 transforms
based on SSSE3, AVX or AVX2 as separate individual drivers when cpu
provides support. This will make it easy for alternative algorithms to
be used if desired and makes the code cleaner and easier to maintain.

Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
---
 arch/x86/crypto/sha512_ssse3_glue.c | 249 +++++++++++++++++++++++++++++-------
 1 file changed, 204 insertions(+), 45 deletions(-)

diff --git a/arch/x86/crypto/sha512_ssse3_glue.c b/arch/x86/crypto/sha512_ssse3_glue.c
index 2edad7b..0dfe9a2 100644
--- a/arch/x86/crypto/sha512_ssse3_glue.c
+++ b/arch/x86/crypto/sha512_ssse3_glue.c
@@ -41,19 +41,11 @@
 
 asmlinkage void sha512_transform_ssse3(u64 *digest, const char *data,
 				       u64 rounds);
-#ifdef CONFIG_AS_AVX
-asmlinkage void sha512_transform_avx(u64 *digest, const char *data,
-				     u64 rounds);
-#endif
-#ifdef CONFIG_AS_AVX2
-asmlinkage void sha512_transform_rorx(u64 *digest, const char *data,
-				      u64 rounds);
-#endif
 
-static void (*sha512_transform_asm)(u64 *, const char *, u64);
+typedef void (sha512_transform_fn)(u64 *digest, const char *data, u64 rounds);
 
-static int sha512_ssse3_update(struct shash_desc *desc, const u8 *data,
-			       unsigned int len)
+static int sha512_update(struct shash_desc *desc, const u8 *data,
+		       unsigned int len, sha512_transform_fn *sha512_xform)
 {
 	struct sha512_state *sctx = shash_desc_ctx(desc);
 
@@ -66,14 +58,14 @@ static int sha512_ssse3_update(struct shash_desc *desc, const u8 *data,
 
 	kernel_fpu_begin();
 	sha512_base_do_update(desc, data, len,
-			      (sha512_block_fn *)sha512_transform_asm);
+			      (sha512_block_fn *)sha512_xform);
 	kernel_fpu_end();
 
 	return 0;
 }
 
-static int sha512_ssse3_finup(struct shash_desc *desc, const u8 *data,
-			      unsigned int len, u8 *out)
+static int sha512_finup(struct shash_desc *desc, const u8 *data,
+	      unsigned int len, u8 *out, sha512_transform_fn *sha512_xform)
 {
 	if (!irq_fpu_usable())
 		return crypto_sha512_finup(desc, data, len, out);
@@ -81,20 +73,32 @@ static int sha512_ssse3_finup(struct shash_desc *desc, const u8 *data,
 	kernel_fpu_begin();
 	if (len)
 		sha512_base_do_update(desc, data, len,
-				      (sha512_block_fn *)sha512_transform_asm);
-	sha512_base_do_finalize(desc, (sha512_block_fn *)sha512_transform_asm);
+				      (sha512_block_fn *)sha512_xform);
+	sha512_base_do_finalize(desc, (sha512_block_fn *)sha512_xform);
 	kernel_fpu_end();
 
 	return sha512_base_finish(desc, out);
 }
 
+static int sha512_ssse3_update(struct shash_desc *desc, const u8 *data,
+		       unsigned int len)
+{
+	return sha512_update(desc, data, len, sha512_transform_ssse3);
+}
+
+static int sha512_ssse3_finup(struct shash_desc *desc, const u8 *data,
+	      unsigned int len, u8 *out)
+{
+	return sha512_finup(desc, data, len, out, sha512_transform_ssse3);
+}
+
 /* Add padding and return the message digest. */
 static int sha512_ssse3_final(struct shash_desc *desc, u8 *out)
 {
 	return sha512_ssse3_finup(desc, NULL, 0, out);
 }
 
-static struct shash_alg algs[] = { {
+static struct shash_alg sha512_ssse3_algs[] = { {
 	.digestsize	=	SHA512_DIGEST_SIZE,
 	.init		=	sha512_base_init,
 	.update		=	sha512_ssse3_update,
@@ -126,8 +130,25 @@ static struct shash_alg algs[] = { {
 	}
 } };
 
+static int register_sha512_ssse3(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SSSE3))
+		return crypto_register_shashes(sha512_ssse3_algs,
+			ARRAY_SIZE(sha512_ssse3_algs));
+	return 0;
+}
+
+static void unregister_sha512_ssse3(void)
+{
+	if (boot_cpu_has(X86_FEATURE_SSSE3))
+		crypto_unregister_shashes(sha512_ssse3_algs,
+			ARRAY_SIZE(sha512_ssse3_algs));
+}
+
 #ifdef CONFIG_AS_AVX
-static bool __init avx_usable(void)
+asmlinkage void sha512_transform_avx(u64 *digest, const char *data,
+				     u64 rounds);
+static bool avx_usable(void)
 {
 	if (!cpu_has_xfeatures(XSTATE_SSE | XSTATE_YMM, NULL)) {
 		if (cpu_has_avx)
@@ -137,47 +158,185 @@ static bool __init avx_usable(void)
 
 	return true;
 }
-#endif
 
-static int __init sha512_ssse3_mod_init(void)
+static int sha512_avx_update(struct shash_desc *desc, const u8 *data,
+		       unsigned int len)
 {
-	/* test for SSSE3 first */
-	if (cpu_has_ssse3)
-		sha512_transform_asm = sha512_transform_ssse3;
+	return sha512_update(desc, data, len, sha512_transform_avx);
+}
 
-#ifdef CONFIG_AS_AVX
-	/* allow AVX to override SSSE3, it's a little faster */
-	if (avx_usable()) {
-#ifdef CONFIG_AS_AVX2
-		if (boot_cpu_has(X86_FEATURE_AVX2))
-			sha512_transform_asm = sha512_transform_rorx;
-		else
-#endif
-			sha512_transform_asm = sha512_transform_avx;
+static int sha512_avx_finup(struct shash_desc *desc, const u8 *data,
+	      unsigned int len, u8 *out)
+{
+	return sha512_finup(desc, data, len, out, sha512_transform_avx);
+}
+
+/* Add padding and return the message digest. */
+static int sha512_avx_final(struct shash_desc *desc, u8 *out)
+{
+	return sha512_avx_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha512_avx_algs[] = { {
+	.digestsize	=	SHA512_DIGEST_SIZE,
+	.init		=	sha512_base_init,
+	.update		=	sha512_avx_update,
+	.final		=	sha512_avx_final,
+	.finup		=	sha512_avx_finup,
+	.descsize	=	sizeof(struct sha512_state),
+	.base		=	{
+		.cra_name	=	"sha512",
+		.cra_driver_name =	"sha512-avx",
+		.cra_priority	=	160,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA512_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
 	}
-#endif
+},  {
+	.digestsize	=	SHA384_DIGEST_SIZE,
+	.init		=	sha384_base_init,
+	.update		=	sha512_avx_update,
+	.final		=	sha512_avx_final,
+	.finup		=	sha512_avx_finup,
+	.descsize	=	sizeof(struct sha512_state),
+	.base		=	{
+		.cra_name	=	"sha384",
+		.cra_driver_name =	"sha384-avx",
+		.cra_priority	=	160,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA384_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+} };
 
-	if (sha512_transform_asm) {
-#ifdef CONFIG_AS_AVX
-		if (sha512_transform_asm == sha512_transform_avx)
-			pr_info("Using AVX optimized SHA-512 implementation\n");
-#ifdef CONFIG_AS_AVX2
-		else if (sha512_transform_asm == sha512_transform_rorx)
-			pr_info("Using AVX2 optimized SHA-512 implementation\n");
+static int register_sha512_avx(void)
+{
+	if (avx_usable())
+		return crypto_register_shashes(sha512_avx_algs,
+			ARRAY_SIZE(sha512_avx_algs));
+	return 0;
+}
+
+static void unregister_sha512_avx(void)
+{
+	if (avx_usable())
+		crypto_unregister_shashes(sha512_avx_algs,
+			ARRAY_SIZE(sha512_avx_algs));
+}
+#else
+static inline int register_sha512_avx(void) { return 0; }
+static inline void unregister_sha512_avx(void) { }
 #endif
-		else
+
+#if defined(CONFIG_AS_AVX2) && defined(CONFIG_AS_AVX)
+asmlinkage void sha512_transform_rorx(u64 *digest, const char *data,
+				      u64 rounds);
+
+static int sha512_avx2_update(struct shash_desc *desc, const u8 *data,
+		       unsigned int len)
+{
+	return sha512_update(desc, data, len, sha512_transform_rorx);
+}
+
+static int sha512_avx2_finup(struct shash_desc *desc, const u8 *data,
+	      unsigned int len, u8 *out)
+{
+	return sha512_finup(desc, data, len, out, sha512_transform_rorx);
+}
+
+/* Add padding and return the message digest. */
+static int sha512_avx2_final(struct shash_desc *desc, u8 *out)
+{
+	return sha512_avx2_finup(desc, NULL, 0, out);
+}
+
+static struct shash_alg sha512_avx2_algs[] = { {
+	.digestsize	=	SHA512_DIGEST_SIZE,
+	.init		=	sha512_base_init,
+	.update		=	sha512_avx2_update,
+	.final		=	sha512_avx2_final,
+	.finup		=	sha512_avx2_finup,
+	.descsize	=	sizeof(struct sha512_state),
+	.base		=	{
+		.cra_name	=	"sha512",
+		.cra_driver_name =	"sha512-avx2",
+		.cra_priority	=	170,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA512_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+},  {
+	.digestsize	=	SHA384_DIGEST_SIZE,
+	.init		=	sha384_base_init,
+	.update		=	sha512_avx2_update,
+	.final		=	sha512_avx2_final,
+	.finup		=	sha512_avx2_finup,
+	.descsize	=	sizeof(struct sha512_state),
+	.base		=	{
+		.cra_name	=	"sha384",
+		.cra_driver_name =	"sha384-avx2",
+		.cra_priority	=	170,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA384_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+} };
+
+static bool avx2_usable(void)
+{
+	if (avx_usable() && boot_cpu_has(X86_FEATURE_AVX2) &&
+		    boot_cpu_has(X86_FEATURE_BMI2))
+		return true;
+
+	return false;
+}
+
+static int register_sha512_avx2(void)
+{
+	if (avx2_usable())
+		return crypto_register_shashes(sha512_avx2_algs,
+			ARRAY_SIZE(sha512_avx2_algs));
+	return 0;
+}
+
+static void unregister_sha512_avx2(void)
+{
+	if (avx2_usable())
+		crypto_unregister_shashes(sha512_avx2_algs,
+			ARRAY_SIZE(sha512_avx2_algs));
+}
+#else
+static inline int register_sha512_avx2(void) { return 0; }
+static inline void unregister_sha512_avx2(void) { }
 #endif
-			pr_info("Using SSSE3 optimized SHA-512 implementation\n");
-		return crypto_register_shashes(algs, ARRAY_SIZE(algs));
+
+static int __init sha512_ssse3_mod_init(void)
+{
+
+	if (register_sha512_ssse3())
+		goto fail;
+
+	if (register_sha512_avx()) {
+		unregister_sha512_ssse3();
+		goto fail;
 	}
-	pr_info("Neither AVX nor SSSE3 is available/usable.\n");
 
+	if (register_sha512_avx2()) {
+		unregister_sha512_avx();
+		unregister_sha512_ssse3();
+		goto fail;
+	}
+
+	return 0;
+fail:
 	return -ENODEV;
 }
 
 static void __exit sha512_ssse3_mod_fini(void)
 {
-	crypto_unregister_shashes(algs, ARRAY_SIZE(algs));
+	unregister_sha512_avx2();
+	unregister_sha512_avx();
+	unregister_sha512_ssse3();
 }
 
 module_init(sha512_ssse3_mod_init);
-- 
1.7.11.7