From 37d63769683a5d79570620da54eb79d9eed0aa64 Mon Sep 17 00:00:00 2001
From: Morph <39850852+Morph1984@users.noreply.github.com>
Date: Mon, 5 Sep 2022 17:00:02 -0400
Subject: [PATCH] aesni: Unroll aes rounds

---
 library/aesni.c | 80 +++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 78 insertions(+), 2 deletions(-)

diff --git a/library/aesni.c b/library/aesni.c
index 5af0e4c11..e54ab0c33 100644
--- a/library/aesni.c
+++ b/library/aesni.c
@@ -117,14 +117,90 @@ int mbedtls_aesni_crypt_ecb( mbedtls_aes_context *ctx,
 
     if (mode == MBEDTLS_AES_ENCRYPT)
     {
-        for (i = ctx->nr - 1; i; --i)
+        if (ctx->nr == 10) {
             a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+        } else if (ctx->nr == 12) {
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+        } else if (ctx->nr == 14) {
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+        } else {
+            for (i = ctx->nr - 1; i; --i)
+                a = _mm_aesenc_si128( a, _mm_loadu_si128( rk++ ) );
+        }
         a = _mm_aesenclast_si128( a, _mm_loadu_si128( rk ) );
     }
     else
     {
-        for (i = ctx->nr - 1; i; --i)
+        if (ctx->nr == 10) {
             a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+        } else if (ctx->nr == 12) {
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+        } else if (ctx->nr == 14) {
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+            a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+        } else {
+            for (i = ctx->nr - 1; i; --i)
+                a = _mm_aesdec_si128( a, _mm_loadu_si128( rk++ ) );
+        }
         a = _mm_aesdeclast_si128( a, _mm_loadu_si128( rk ) );
     }