crypto: x86/sha - glue code for Intel SHA extensions optimized SHA1 & SHA256

This patch adds the glue code to detect and utilize the Intel SHA
extensions optimized SHA1 and SHA256 update transforms when available.

This code has been tested on Broxton for functionality.

Originally-by: Chandramouli Narayanan <mouli_7982@yahoo.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

2 files changed, 34 insertions, 16 deletions

diff --git a/arch/x86/crypto/sha1_ssse3_glue.c b/arch/x86/crypto/sha1_ssse3_glue.c
index 7c48e8b20848..98be8cc17ca2 100644
--- a/arch/x86/crypto/sha1_ssse3_glue.c
+++ b/arch/x86/crypto/sha1_ssse3_glue.c
@@ -44,6 +44,10 @@ asmlinkage void sha1_transform_avx(u32 *digest, const char *data,
 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);
 
@@ -166,12 +170,18 @@ static int __init sha1_ssse3_mod_init(void)
 #endif
 	}
 #endif
+#ifdef CONFIG_AS_SHA1_NI
+	if (boot_cpu_has(X86_FEATURE_SHA_NI)) {
+		sha1_transform_asm = sha1_ni_transform;
+		algo_name = "SHA-NI";
+	}
+#endif
 
 	if (sha1_transform_asm) {
 		pr_info("Using %s optimized SHA-1 implementation\n", algo_name);
 		return crypto_register_shash(&alg);
 	}
-	pr_info("Neither AVX nor AVX2 nor SSSE3 is available/usable.\n");
+	pr_info("Neither AVX nor AVX2 nor SSSE3/SHA-NI is available/usable.\n");
 
 	return -ENODEV;
 }
diff --git a/arch/x86/crypto/sha256_ssse3_glue.c b/arch/x86/crypto/sha256_ssse3_glue.c
index f8097fc0d1d1..9c7b22c489f6 100644
--- a/arch/x86/crypto/sha256_ssse3_glue.c
+++ b/arch/x86/crypto/sha256_ssse3_glue.c
@@ -50,6 +50,10 @@ asmlinkage void sha256_transform_avx(u32 *digest, const char *data,
 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
 
 static void (*sha256_transform_asm)(u32 *, const char *, u64);
 
@@ -142,36 +146,40 @@ static bool __init avx_usable(void)
 
 static int __init sha256_ssse3_mod_init(void)
 {
+	char *algo;
+
 	/* test for SSSE3 first */
-	if (cpu_has_ssse3)
+	if (cpu_has_ssse3) {
 		sha256_transform_asm = sha256_transform_ssse3;
+		algo = "SSSE3";
+	}
 
 #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))
+		if (boot_cpu_has(X86_FEATURE_AVX2) &&
+		    boot_cpu_has(X86_FEATURE_BMI2)) {
 			sha256_transform_asm = sha256_transform_rorx;
-		else
+			algo = "AVX2";
+		}
+#endif
+	}
 #endif
-			sha256_transform_asm = sha256_transform_avx;
+#ifdef CONFIG_AS_SHA256_NI
+	if (boot_cpu_has(X86_FEATURE_SHA_NI)) {
+		sha256_transform_asm = sha256_ni_transform;
+		algo = "SHA-256-NI";
 	}
 #endif
 
 	if (sha256_transform_asm) {
-#ifdef CONFIG_AS_AVX
-		if (sha256_transform_asm == sha256_transform_avx)
-			pr_info("Using AVX optimized SHA-256 implementation\n");
-#ifdef CONFIG_AS_AVX2
-		else if (sha256_transform_asm == sha256_transform_rorx)
-			pr_info("Using AVX2 optimized SHA-256 implementation\n");
-#endif
-		else
-#endif
-			pr_info("Using SSSE3 optimized SHA-256 implementation\n");
+		pr_info("Using %s optimized SHA-256 implementation\n", algo);
 		return crypto_register_shashes(algs, ARRAY_SIZE(algs));
 	}
-	pr_info("Neither AVX nor SSSE3 is available/usable.\n");
+	pr_info("Neither AVX nor SSSE3/SHA-NI is available/usable.\n");
 
 	return -ENODEV;
 }