crypto-native: 256-bit AES CBC support

Used on intel client CPUs which suppport VAES instruction set without
AVX512

Type: improvement
Change-Id: I5f816a1ea9f89a8d298d2c0f38d8d7c06f414ba0
Signed-off-by: Damjan Marion <damarion@cisco.com>

1 files changed, 297 insertions, 60 deletions

diff --git a/src/plugins/crypto_native/aes_cbc.c b/src/plugins/crypto_native/aes_cbc.c
index 7896c8814b1..02d96b31c79 100644
--- a/src/plugins/crypto_native/aes_cbc.c
+++ b/src/plugins/crypto_native/aes_cbc.c
@@ -25,17 +25,35 @@
 #pragma GCC optimize ("O3")
 #endif
 
+#if defined(__VAES__) && defined(__AVX512F__)
+#define N		  16
+#define u8xN		  u8x64
+#define u32xN		  u32x16
+#define u32xN_min_scalar  u32x16_min_scalar
+#define u32xN_is_all_zero u32x16_is_all_zero
+#define u32xN_splat	  u32x16_splat
+#elif defined(__VAES__)
+#define N		  8
+#define u8xN		  u8x32
+#define u32xN		  u32x8
+#define u32xN_min_scalar  u32x8_min_scalar
+#define u32xN_is_all_zero u32x8_is_all_zero
+#define u32xN_splat	  u32x8_splat
+#else
+#define N		  4
+#define u8xN		  u8x16
+#define u32xN		  u32x4
+#define u32xN_min_scalar  u32x4_min_scalar
+#define u32xN_is_all_zero u32x4_is_all_zero
+#define u32xN_splat	  u32x4_splat
+#endif
+
 typedef struct
 {
   u8x16 encrypt_key[15];
-#if __VAES__
-  u8x64 decrypt_key[15];
-#else
-  u8x16 decrypt_key[15];
-#endif
+  u8xN decrypt_key[15];
 } aes_cbc_key_data_t;
 
-
 static_always_inline void __clib_unused
 aes_cbc_dec (u8x16 * k, u8x16u * src, u8x16u * dst, u8x16u * iv, int count,
 	     int rounds)
@@ -119,7 +137,7 @@ aes_cbc_dec (u8x16 * k, u8x16u * src, u8x16u * dst, u8x16u * iv, int count,
 }
 
 #if __x86_64__
-#ifdef __VAES__
+#if defined(__VAES__) && defined(__AVX512F__)
 
 static_always_inline u8x64
 aes_block_load_x4 (u8 * src[], int i)
@@ -142,14 +160,13 @@ aes_block_store_x4 (u8 * dst[], int i, u8x64 r)
 }
 
 static_always_inline u8x64
-aes_cbc_dec_permute (u8x64 a, u8x64 b)
+aes4_cbc_dec_permute (u8x64 a, u8x64 b)
 {
-  __m512i perm = { 6, 7, 8, 9, 10, 11, 12, 13 };
-  return (u8x64) _mm512_permutex2var_epi64 ((__m512i) a, perm, (__m512i) b);
+  return (u8x64) u64x8_shuffle2 (a, b, 6, 7, 8, 9, 10, 11, 12, 13);
 }
 
 static_always_inline void
-vaes_cbc_dec (u8x64 *k, u8x64u *src, u8x64u *dst, u8x16u *iv, int count,
+aes4_cbc_dec (u8x64 *k, u8x64u *src, u8x64u *dst, u8x16u *iv, int count,
 	      aes_key_size_t rounds)
 {
   u8x64 f, r[4], c[4] = { };
@@ -184,10 +201,10 @@ vaes_cbc_dec (u8x64 *k, u8x64u *src, u8x64u *dst, u8x16u *iv, int count,
       r[2] = aes_dec_last_round_x4 (r[2], k[i]);
       r[3] = aes_dec_last_round_x4 (r[3], k[i]);
 
-      dst[0] = r[0] ^= aes_cbc_dec_permute (f, c[0]);
-      dst[1] = r[1] ^= aes_cbc_dec_permute (c[0], c[1]);
-      dst[2] = r[2] ^= aes_cbc_dec_permute (c[1], c[2]);
-      dst[3] = r[3] ^= aes_cbc_dec_permute (c[2], c[3]);
+      dst[0] = r[0] ^= aes4_cbc_dec_permute (f, c[0]);
+      dst[1] = r[1] ^= aes4_cbc_dec_permute (c[0], c[1]);
+      dst[2] = r[2] ^= aes4_cbc_dec_permute (c[1], c[2]);
+      dst[3] = r[3] ^= aes4_cbc_dec_permute (c[2], c[3]);
       f = c[3];
 
       n_blocks -= 16;
@@ -195,40 +212,248 @@ vaes_cbc_dec (u8x64 *k, u8x64u *src, u8x64u *dst, u8x16u *iv, int count,
       dst += 4;
     }
 
-  while (n_blocks > 0)
+  if (n_blocks >= 12)
+    {
+      c[0] = src[0];
+      c[1] = src[1];
+      c[2] = src[2];
+
+      r[0] = c[0] ^ k[0];
+      r[1] = c[1] ^ k[0];
+      r[2] = c[2] ^ k[0];
+
+      for (i = 1; i < rounds; i++)
+	{
+	  r[0] = aes_dec_round_x4 (r[0], k[i]);
+	  r[1] = aes_dec_round_x4 (r[1], k[i]);
+	  r[2] = aes_dec_round_x4 (r[2], k[i]);
+	}
+
+      r[0] = aes_dec_last_round_x4 (r[0], k[i]);
+      r[1] = aes_dec_last_round_x4 (r[1], k[i]);
+      r[2] = aes_dec_last_round_x4 (r[2], k[i]);
+
+      dst[0] = r[0] ^= aes4_cbc_dec_permute (f, c[0]);
+      dst[1] = r[1] ^= aes4_cbc_dec_permute (c[0], c[1]);
+      dst[2] = r[2] ^= aes4_cbc_dec_permute (c[1], c[2]);
+      f = c[2];
+
+      n_blocks -= 12;
+      src += 3;
+      dst += 3;
+    }
+  else if (n_blocks >= 8)
+    {
+      c[0] = src[0];
+      c[1] = src[1];
+
+      r[0] = c[0] ^ k[0];
+      r[1] = c[1] ^ k[0];
+
+      for (i = 1; i < rounds; i++)
+	{
+	  r[0] = aes_dec_round_x4 (r[0], k[i]);
+	  r[1] = aes_dec_round_x4 (r[1], k[i]);
+	}
+
+      r[0] = aes_dec_last_round_x4 (r[0], k[i]);
+      r[1] = aes_dec_last_round_x4 (r[1], k[i]);
+
+      dst[0] = r[0] ^= aes4_cbc_dec_permute (f, c[0]);
+      dst[1] = r[1] ^= aes4_cbc_dec_permute (c[0], c[1]);
+      f = c[1];
+
+      n_blocks -= 8;
+      src += 2;
+      dst += 2;
+    }
+  else if (n_blocks >= 4)
+    {
+      c[0] = src[0];
+
+      r[0] = c[0] ^ k[0];
+
+      for (i = 1; i < rounds; i++)
+	{
+	  r[0] = aes_dec_round_x4 (r[0], k[i]);
+	}
+
+      r[0] = aes_dec_last_round_x4 (r[0], k[i]);
+
+      dst[0] = r[0] ^= aes4_cbc_dec_permute (f, c[0]);
+      f = c[0];
+
+      n_blocks -= 4;
+      src += 1;
+      dst += 1;
+    }
+
+  if (n_blocks > 0)
     {
       m = (1 << (n_blocks * 2)) - 1;
       c[0] = (u8x64) _mm512_mask_loadu_epi64 ((__m512i) c[0], m,
 					      (__m512i *) src);
-      f = aes_cbc_dec_permute (f, c[0]);
+      f = aes4_cbc_dec_permute (f, c[0]);
       r[0] = c[0] ^ k[0];
       for (i = 1; i < rounds; i++)
 	r[0] = aes_dec_round_x4 (r[0], k[i]);
       r[0] = aes_dec_last_round_x4 (r[0], k[i]);
       _mm512_mask_storeu_epi64 ((__m512i *) dst, m, (__m512i) (r[0] ^ f));
-      f = c[0];
+    }
+}
+#elif defined(__VAES__)
+
+static_always_inline u8x32
+aes_block_load_x2 (u8 *src[], int i)
+{
+  u8x32 r = {};
+  r = u8x32_insert_lo (r, aes_block_load (src[0] + i));
+  r = u8x32_insert_hi (r, aes_block_load (src[1] + i));
+  return r;
+}
+
+static_always_inline void
+aes_block_store_x2 (u8 *dst[], int i, u8x32 r)
+{
+  aes_block_store (dst[0] + i, u8x32_extract_lo (r));
+  aes_block_store (dst[1] + i, u8x32_extract_hi (r));
+}
+
+static_always_inline u8x32
+aes2_cbc_dec_permute (u8x32 a, u8x32 b)
+{
+  return (u8x32) u64x4_shuffle2 ((u64x4) a, (u64x4) b, 2, 3, 4, 5);
+}
+
+static_always_inline void
+aes2_cbc_dec (u8x32 *k, u8x32u *src, u8x32u *dst, u8x16u *iv, int count,
+	      aes_key_size_t rounds)
+{
+  u8x32 f = {}, r[4], c[4] = {};
+  int i, n_blocks = count >> 4;
+
+  f = u8x32_insert_hi (f, *iv);
+
+  while (n_blocks >= 8)
+    {
+      c[0] = src[0];
+      c[1] = src[1];
+      c[2] = src[2];
+      c[3] = src[3];
+
+      r[0] = c[0] ^ k[0];
+      r[1] = c[1] ^ k[0];
+      r[2] = c[2] ^ k[0];
+      r[3] = c[3] ^ k[0];
+
+      for (i = 1; i < rounds; i++)
+	{
+	  r[0] = aes_dec_round_x2 (r[0], k[i]);
+	  r[1] = aes_dec_round_x2 (r[1], k[i]);
+	  r[2] = aes_dec_round_x2 (r[2], k[i]);
+	  r[3] = aes_dec_round_x2 (r[3], k[i]);
+	}
+
+      r[0] = aes_dec_last_round_x2 (r[0], k[i]);
+      r[1] = aes_dec_last_round_x2 (r[1], k[i]);
+      r[2] = aes_dec_last_round_x2 (r[2], k[i]);
+      r[3] = aes_dec_last_round_x2 (r[3], k[i]);
+
+      dst[0] = r[0] ^= aes2_cbc_dec_permute (f, c[0]);
+      dst[1] = r[1] ^= aes2_cbc_dec_permute (c[0], c[1]);
+      dst[2] = r[2] ^= aes2_cbc_dec_permute (c[1], c[2]);
+      dst[3] = r[3] ^= aes2_cbc_dec_permute (c[2], c[3]);
+      f = c[3];
+
+      n_blocks -= 8;
+      src += 4;
+      dst += 4;
+    }
+
+  if (n_blocks >= 6)
+    {
+      c[0] = src[0];
+      c[1] = src[1];
+      c[2] = src[2];
+
+      r[0] = c[0] ^ k[0];
+      r[1] = c[1] ^ k[0];
+      r[2] = c[2] ^ k[0];
+
+      for (i = 1; i < rounds; i++)
+	{
+	  r[0] = aes_dec_round_x2 (r[0], k[i]);
+	  r[1] = aes_dec_round_x2 (r[1], k[i]);
+	  r[2] = aes_dec_round_x2 (r[2], k[i]);
+	}
+
+      r[0] = aes_dec_last_round_x2 (r[0], k[i]);
+      r[1] = aes_dec_last_round_x2 (r[1], k[i]);
+      r[2] = aes_dec_last_round_x2 (r[2], k[i]);
+
+      dst[0] = r[0] ^= aes2_cbc_dec_permute (f, c[0]);
+      dst[1] = r[1] ^= aes2_cbc_dec_permute (c[0], c[1]);
+      dst[2] = r[2] ^= aes2_cbc_dec_permute (c[1], c[2]);
+      f = c[2];
+
+      n_blocks -= 6;
+      src += 3;
+      dst += 3;
+    }
+  else if (n_blocks >= 4)
+    {
+      c[0] = src[0];
+      c[1] = src[1];
+
+      r[0] = c[0] ^ k[0];
+      r[1] = c[1] ^ k[0];
+
+      for (i = 1; i < rounds; i++)
+	{
+	  r[0] = aes_dec_round_x2 (r[0], k[i]);
+	  r[1] = aes_dec_round_x2 (r[1], k[i]);
+	}
+
+      r[0] = aes_dec_last_round_x2 (r[0], k[i]);
+      r[1] = aes_dec_last_round_x2 (r[1], k[i]);
+
+      dst[0] = r[0] ^= aes2_cbc_dec_permute (f, c[0]);
+      dst[1] = r[1] ^= aes2_cbc_dec_permute (c[0], c[1]);
+      f = c[1];
+
       n_blocks -= 4;
+      src += 2;
+      dst += 2;
+    }
+  else if (n_blocks >= 2)
+    {
+      c[0] = src[0];
+      r[0] = c[0] ^ k[0];
+
+      for (i = 1; i < rounds; i++)
+	r[0] = aes_dec_round_x2 (r[0], k[i]);
+
+      r[0] = aes_dec_last_round_x2 (r[0], k[i]);
+      dst[0] = r[0] ^= aes2_cbc_dec_permute (f, c[0]);
+      f = c[0];
+
+      n_blocks -= 2;
       src += 1;
       dst += 1;
     }
+
+  if (n_blocks > 0)
+    {
+      u8x16 rl = *(u8x16u *) src ^ u8x32_extract_lo (k[0]);
+      for (i = 1; i < rounds; i++)
+	rl = aes_dec_round (rl, u8x32_extract_lo (k[i]));
+      rl = aes_dec_last_round (rl, u8x32_extract_lo (k[i]));
+      *(u8x16 *) dst = rl ^ u8x32_extract_hi (f);
+    }
 }
 #endif
 #endif
 
-#ifdef __VAES__
-#define N 16
-#define u32xN u32x16
-#define u32xN_min_scalar u32x16_min_scalar
-#define u32xN_is_all_zero u32x16_is_all_zero
-#define u32xN_splat u32x16_splat
-#else
-#define N 4
-#define u32xN u32x4
-#define u32xN_min_scalar u32x4_min_scalar
-#define u32xN_is_all_zero u32x4_is_all_zero
-#define u32xN_splat u32x4_splat
-#endif
-
 static_always_inline u32
 aes_ops_enc_aes_cbc (vlib_main_t * vm, vnet_crypto_op_t * ops[],
 		     u32 n_ops, aes_key_size_t ks)
@@ -242,14 +467,8 @@ aes_ops_enc_aes_cbc (vlib_main_t * vm, vnet_crypto_op_t * ops[],
   vnet_crypto_key_index_t key_index[N];
   u8 *src[N] = { };
   u8 *dst[N] = { };
-#if __VAES__
-  u8x64 r[N / 4] = { };
-  u8x64 k[15][N / 4] = { };
-  u8x16 *kq, *rq = (u8x16 *) r;
-#else
-  u8x16 r[N] = { };
-  u8x16 k[15][N] = { };
-#endif
+  u8xN r[4] = {};
+  u8xN k[15][4] = {};
 
   for (i = 0; i < N; i++)
     key_index[i] = ~0;
@@ -268,11 +487,7 @@ more:
 	else
 	  {
 	    u8x16 t = aes_block_load (ops[0]->iv);
-#if __VAES__
-	    rq[i] = t;
-#else
-	    r[i] = t;
-#endif
+	    ((u8x16 *) r)[i] = t;
 
 	    src[i] = ops[0]->src;
 	    dst[i] = ops[0]->dst;
@@ -284,14 +499,7 @@ more:
 		key_index[i] = ops[0]->key_index;
 		kd = (aes_cbc_key_data_t *) cm->key_data[key_index[i]];
 		for (j = 0; j < rounds + 1; j++)
-		  {
-#if __VAES__
-		    kq = (u8x16 *) k[j];
-		    kq[i] = kd->encrypt_key[j];
-#else
-		    k[j][i] = kd->encrypt_key[j];
-#endif
-		  }
+		  ((u8x16 *) k[j])[i] = kd->encrypt_key[j];
 	      }
 	    ops[0]->status = VNET_CRYPTO_OP_STATUS_COMPLETED;
 	    n_left--;
@@ -305,7 +513,7 @@ more:
 
   for (i = 0; i < count; i += 16)
     {
-#ifdef __VAES__
+#if defined(__VAES__) && defined(__AVX512F__)
       r[0] = u8x64_xor3 (r[0], aes_block_load_x4 (src, i), k[0][0]);
       r[1] = u8x64_xor3 (r[1], aes_block_load_x4 (src + 4, i), k[0][1]);
       r[2] = u8x64_xor3 (r[2], aes_block_load_x4 (src + 8, i), k[0][2]);
@@ -327,6 +535,28 @@ more:
       aes_block_store_x4 (dst + 4, i, r[1]);
       aes_block_store_x4 (dst + 8, i, r[2]);
       aes_block_store_x4 (dst + 12, i, r[3]);
+#elif defined(__VAES__)
+      r[0] = u8x32_xor3 (r[0], aes_block_load_x2 (src, i), k[0][0]);
+      r[1] = u8x32_xor3 (r[1], aes_block_load_x2 (src + 2, i), k[0][1]);
+      r[2] = u8x32_xor3 (r[2], aes_block_load_x2 (src + 4, i), k[0][2]);
+      r[3] = u8x32_xor3 (r[3], aes_block_load_x2 (src + 6, i), k[0][3]);
+
+      for (j = 1; j < rounds; j++)
+	{
+	  r[0] = aes_enc_round_x2 (r[0], k[j][0]);
+	  r[1] = aes_enc_round_x2 (r[1], k[j][1]);
+	  r[2] = aes_enc_round_x2 (r[2], k[j][2]);
+	  r[3] = aes_enc_round_x2 (r[3], k[j][3]);
+	}
+      r[0] = aes_enc_last_round_x2 (r[0], k[j][0]);
+      r[1] = aes_enc_last_round_x2 (r[1], k[j][1]);
+      r[2] = aes_enc_last_round_x2 (r[2], k[j][2]);
+      r[3] = aes_enc_last_round_x2 (r[3], k[j][3]);
+
+      aes_block_store_x2 (dst, i, r[0]);
+      aes_block_store_x2 (dst + 2, i, r[1]);
+      aes_block_store_x2 (dst + 4, i, r[2]);
+      aes_block_store_x2 (dst + 6, i, r[3]);
 #else
 #if __x86_64__
       r[0] = u8x16_xor3 (r[0], aes_block_load (src[0] + i), k[0][0]);
@@ -406,8 +636,11 @@ aes_ops_dec_aes_cbc (vlib_main_t * vm, vnet_crypto_op_t * ops[],
   ASSERT (n_ops >= 1);
 
 decrypt:
-#ifdef __VAES__
-  vaes_cbc_dec (kd->decrypt_key, (u8x64u *) op->src, (u8x64u *) op->dst,
+#if defined(__VAES__) && defined(__AVX512F__)
+  aes4_cbc_dec (kd->decrypt_key, (u8x64u *) op->src, (u8x64u *) op->dst,
+		(u8x16u *) op->iv, op->len, rounds);
+#elif defined(__VAES__)
+  aes2_cbc_dec (kd->decrypt_key, (u8x32u *) op->src, (u8x32u *) op->dst,
 		(u8x16u *) op->iv, op->len, rounds);
 #else
   aes_cbc_dec (kd->decrypt_key, (u8x16u *) op->src, (u8x16u *) op->dst,
@@ -435,8 +668,10 @@ aes_cbc_key_exp (vnet_crypto_key_t * key, aes_key_size_t ks)
   aes_key_enc_to_dec (e, d, ks);
   for (int i = 0; i < AES_KEY_ROUNDS (ks) + 1; i++)
     {
-#if __VAES__
-      kd->decrypt_key[i] = (u8x64) _mm512_broadcast_i64x2 ((__m128i) d[i]);
+#if defined(__VAES__) && defined(__AVX512F__)
+      kd->decrypt_key[i] = u8x64_splat_u8x16 (d[i]);
+#elif defined(__VAES__)
+      kd->decrypt_key[i] = u8x32_splat_u8x16 (d[i]);
 #else
       kd->decrypt_key[i] = d[i];
 #endif
@@ -463,8 +698,10 @@ foreach_aes_cbc_handler_type;
 #include <fcntl.h>
 
 clib_error_t *
-#ifdef __VAES__
-crypto_native_aes_cbc_init_icl (vlib_main_t * vm)
+#if defined(__VAES__) && defined(__AVX512F__)
+crypto_native_aes_cbc_init_icl (vlib_main_t *vm)
+#elif defined(__VAES__)
+crypto_native_aes_cbc_init_adl (vlib_main_t *vm)
 #elif __AVX512F__
 crypto_native_aes_cbc_init_skx (vlib_main_t * vm)
 #elif __aarch64__