MdePkg BaseMemoryLibMmx: Convert X64/SetMem.asm to NASM

The BaseTools/Scripts/ConvertMasmToNasm.py script was used to convert
X64/SetMem.asm to X64/SetMem.nasm

Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>

MdePkg/Library/BaseMemoryLibMmx/BaseMemoryLibMmx.inf

@@ -114,6 +114,7 @@
   X64/SetMem32.asm
   X64/SetMem16.nasm
   X64/SetMem16.asm
+  X64/SetMem.nasm
   X64/SetMem.asm
   X64/CopyMem.asm
   X64/ScanMem64.nasm
@@ -134,6 +135,7 @@
   X64/SetMem16.S
   X64/ZeroMem.nasm
   X64/ZeroMem.S
+  X64/SetMem.nasm
   X64/SetMem.S
   X64/CopyMem.S
 

MdePkg/Library/BaseMemoryLibMmx/X64/SetMem.nasm

@@ -0,0 +1,60 @@
+;------------------------------------------------------------------------------
+;
+; Copyright (c) 2006, Intel Corporation. All rights reserved.<BR>
+; This program and the accompanying materials
+; are licensed and made available under the terms and conditions of the BSD License
+; which accompanies this distribution.  The full text of the license may be found at
+; http://opensource.org/licenses/bsd-license.php.
+;
+; THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
+; WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
+;
+; Module Name:
+;
+;   SetMem.nasm
+;
+; Abstract:
+;
+;   SetMem function
+;
+; Notes:
+;
+;------------------------------------------------------------------------------
+
+    DEFAULT REL
+    SECTION .text
+
+;------------------------------------------------------------------------------
+; VOID *
+; EFIAPI
+; InternalMemSetMem (
+;   OUT     VOID                      *Buffer,
+;   IN      UINTN                     Length,
+;   IN      UINT8                     Value
+;   );
+;------------------------------------------------------------------------------
+global ASM_PFX(InternalMemSetMem)
+ASM_PFX(InternalMemSetMem):
+    push    rdi
+    mov     rax, r8
+    mov     ah, al
+    DB      0x48, 0xf, 0x6e, 0xc0         ; movd mm0, rax
+    mov     r8, rcx
+    mov     rdi, r8                     ; rdi <- Buffer
+    mov     rcx, rdx
+    and     edx, 7
+    shr     rcx, 3
+    jz      @SetBytes
+    DB      0xf, 0x70, 0xC0, 0x0         ; pshufw mm0, mm0, 0h
+.0:
+    DB      0xf, 0xe7, 0x7              ; movntq [rdi], mm0
+    add     rdi, 8
+    loop    .0
+    mfence
+@SetBytes:
+    mov     ecx, edx
+    rep     stosb
+    mov     rax, r8
+    pop     rdi
+    ret
+