diff --git a/wolfcrypt/src/aes.c b/wolfcrypt/src/aes.c
index 5ced9b2f7..4b940cfc5 100644
--- a/wolfcrypt/src/aes.c
+++ b/wolfcrypt/src/aes.c
@@ -6987,10 +6987,14 @@ static int wc_AesGcmEncrypt_STM32(Aes* aes, byte* out, const byte* in, word32 sz
     /* for cases where hardware cannot be used for authTag calculate it */
     /* if IV is not 12 calculate GHASH using software */
     if (ivSz != GCM_NONCE_MID_SZ
-#ifndef STM32_AESGCM_PARTIAL
+    #ifndef CRYP_HEADERWIDTHUNIT_BYTE
+        /* or harware that does not support partial block */
+        || sz == 0 || partial != 0
+    #endif
+    #ifndef STM32_AESGCM_PARTIAL
         /* or authIn is not a multiple of 4  */
-        || authPadSz != authInSz || sz == 0 || partial != 0
-#endif
+        || authPadSz != authInSz
+    #endif
     ) {
         useSwGhash = 1;
     }
@@ -7008,17 +7012,42 @@ static int wc_AesGcmEncrypt_STM32(Aes* aes, byte* out, const byte* in, word32 sz
 
 #if defined(STM32_HAL_V2)
     hcryp.Init.Algorithm = CRYP_AES_GCM;
+    #ifdef CRYP_HEADERWIDTHUNIT_BYTE
+    /* V2 with CRYP_HEADERWIDTHUNIT_BYTE uses byte size for header */
+    hcryp.Init.HeaderSize = authPadSz;
+    #else
     hcryp.Init.HeaderSize = authPadSz/sizeof(word32);
+    #endif
     #ifdef STM32_AESGCM_PARTIAL
     hcryp.Init.HeaderPadSize = authPadSz - authInSz;
     #endif
-    ByteReverseWords(partialBlock, ctr, AES_BLOCK_SIZE);
-    hcryp.Init.pInitVect = (STM_CRYPT_TYPE*)partialBlock;
+    #ifdef CRYP_KEYIVCONFIG_ONCE
+    /* allows repeated calls to HAL_CRYP_Encrypt */
+    hcryp.Init.KeyIVConfigSkip = CRYP_KEYIVCONFIG_ONCE;
+    #endif
+    ByteReverseWords(ctr, ctr, AES_BLOCK_SIZE);
+    hcryp.Init.pInitVect = (STM_CRYPT_TYPE*)ctr;
     HAL_CRYP_Init(&hcryp);
 
+    #ifndef CRYP_KEYIVCONFIG_ONCE
     /* GCM payload phase - can handle partial blocks */
     status = HAL_CRYP_Encrypt(&hcryp, (uint32_t*)in,
         (blocks * AES_BLOCK_SIZE) + partial, (uint32_t*)out, STM32_HAL_TIMEOUT);
+    #else
+    /* GCM payload phase - blocks */
+    if (blocks) {
+        status = HAL_CRYP_Encrypt(&hcryp, (uint32_t*)in,
+            (blocks * AES_BLOCK_SIZE), (uint32_t*)out, STM32_HAL_TIMEOUT);
+    }
+    /* GCM payload phase - partial remainder */
+    if (status == HAL_OK && (partial != 0 || blocks == 0)) {
+        XMEMSET(partialBlock, 0, sizeof(partialBlock));
+        XMEMCPY(partialBlock, in + (blocks * AES_BLOCK_SIZE), partial);
+        status = HAL_CRYP_Encrypt(&hcryp, (uint32_t*)partialBlock, partial,
+            (uint32_t*)partialBlock, STM32_HAL_TIMEOUT);
+        XMEMCPY(out + (blocks * AES_BLOCK_SIZE), partialBlock, partial);
+    }
+    #endif
     if (status == HAL_OK && !useSwGhash) {
         /* Compute the authTag */
         status = HAL_CRYPEx_AESGCM_GenerateAuthTAG(&hcryp, (uint32_t*)tag,
@@ -7447,16 +7476,20 @@ static int wc_AesGcmDecrypt_STM32(Aes* aes, byte* out,
 
     /* for cases where hardware cannot be used for authTag calculate it */
     /* if IV is not 12 calculate GHASH using software */
-    if (ivSz != GCM_NONCE_MID_SZ || sz == 0 || partial != 0
-#ifndef STM32_AESGCM_PARTIAL
+    if (ivSz != GCM_NONCE_MID_SZ 
+    #ifndef CRYP_HEADERWIDTHUNIT_BYTE
+        /* or harware that does not support partial block */
+        || sz == 0 || partial != 0
+    #endif
+    #ifndef STM32_AESGCM_PARTIAL
         /* or authIn is not a multiple of 4  */
         || authPadSz != authInSz
-#endif
+    #endif
     ) {
-		GHASH(aes, authIn, authInSz, in, sz, (byte*)tag, sizeof(tag));
-		wc_AesEncrypt(aes, (byte*)ctr, (byte*)partialBlock);
-		xorbuf(tag, partialBlock, sizeof(tag));
-		tagComputed = 1;
+        GHASH(aes, authIn, authInSz, in, sz, (byte*)tag, sizeof(tag));
+        wc_AesEncrypt(aes, (byte*)ctr, (byte*)partialBlock);
+        xorbuf(tag, partialBlock, sizeof(tag));
+        tagComputed = 1;
     }
 
     /* if using hardware for authentication tag make sure its aligned and zero padded */
@@ -7492,18 +7525,42 @@ static int wc_AesGcmDecrypt_STM32(Aes* aes, byte* out,
 
 #if defined(STM32_HAL_V2)
     hcryp.Init.Algorithm = CRYP_AES_GCM;
+    #ifdef CRYP_HEADERWIDTHUNIT_BYTE
+    /* V2 with CRYP_HEADERWIDTHUNIT_BYTE uses byte size for header */
+    hcryp.Init.HeaderSize = authPadSz;
+    #else
     hcryp.Init.HeaderSize = authPadSz/sizeof(word32);
+    #endif
     #ifdef STM32_AESGCM_PARTIAL
     hcryp.Init.HeaderPadSize = authPadSz - authInSz;
     #endif
-    ByteReverseWords(partialBlock, ctr, AES_BLOCK_SIZE);
-    hcryp.Init.pInitVect = (STM_CRYPT_TYPE*)partialBlock;
+    #ifdef CRYP_KEYIVCONFIG_ONCE
+    /* allows repeated calls to HAL_CRYP_Decrypt */
+    hcryp.Init.KeyIVConfigSkip = CRYP_KEYIVCONFIG_ONCE;
+    #endif
+    ByteReverseWords(ctr, ctr, AES_BLOCK_SIZE);
+    hcryp.Init.pInitVect = (STM_CRYPT_TYPE*)ctr;
     HAL_CRYP_Init(&hcryp);
 
-    /* GCM payload phase - can handle partial blocks */
+    #ifndef CRYP_KEYIVCONFIG_ONCE
     status = HAL_CRYP_Decrypt(&hcryp, (uint32_t*)in,
         (blocks * AES_BLOCK_SIZE) + partial, (uint32_t*)out, STM32_HAL_TIMEOUT);
-    if (status == HAL_OK && tagComputed == 0) {
+    #else
+    /* GCM payload phase - blocks */
+    if (blocks) {
+        status = HAL_CRYP_Decrypt(&hcryp, (uint32_t*)in,
+            (blocks * AES_BLOCK_SIZE), (uint32_t*)out, STM32_HAL_TIMEOUT);
+    }
+    /* GCM payload phase - partial remainder */
+    if (status == HAL_OK && (partial != 0 || blocks == 0)) {
+        XMEMSET(partialBlock, 0, sizeof(partialBlock));
+        XMEMCPY(partialBlock, in + (blocks * AES_BLOCK_SIZE), partial);
+        status = HAL_CRYP_Decrypt(&hcryp, (uint32_t*)partialBlock, partial,
+(           uint32_t*)partialBlock, STM32_HAL_TIMEOUT);
+        XMEMCPY(out + (blocks * AES_BLOCK_SIZE), partialBlock, partial);
+    }
+    #endif
+    if (status == HAL_OK && !tagComputed) {
         /* Compute the authTag */
         status = HAL_CRYPEx_AESGCM_GenerateAuthTAG(&hcryp, (uint32_t*)tag,
             STM32_HAL_TIMEOUT);
@@ -7588,7 +7645,7 @@ static int wc_AesGcmDecrypt_STM32(Aes* aes, byte* out,
     if (status != SUCCESS)
         ret = AES_GCM_AUTH_E;
     if (tagComputed == 0)
-    	XMEMCPY(tag, partialBlock, authTagSz);
+        XMEMCPY(tag, partialBlock, authTagSz);
 #endif /* WOLFSSL_STM32_CUBEMX */
     wolfSSL_CryptHwMutexUnLock();
 
diff --git a/wolfcrypt/src/port/st/stm32.c b/wolfcrypt/src/port/st/stm32.c
index dd19797a8..390fa23f4 100644
--- a/wolfcrypt/src/port/st/stm32.c
+++ b/wolfcrypt/src/port/st/stm32.c
@@ -294,6 +294,9 @@ int wc_Stm32_Aes_Init(Aes* aes, CRYP_HandleTypeDef* hcryp)
     hcryp->Init.pKey = (STM_CRYPT_TYPE*)aes->key;
 #ifdef STM32_HAL_V2
     hcryp->Init.DataWidthUnit = CRYP_DATAWIDTHUNIT_BYTE;
+    #ifdef CRYP_HEADERWIDTHUNIT_BYTE
+    hcryp->Init.HeaderWidthUnit = CRYP_HEADERWIDTHUNIT_BYTE;
+    #endif
 #endif
 
     return 0;