Fix yasm warnings in libjpeg-turbo.

simd/jdmrgss2.asm

 ;
 ; jdmrgss2.asm - merged upsampling/color conversion (SSE2)
 ;
 ; Copyright 2009 Pierre Ossman <ossman@cendio.se> for Cendio AB
 ;
 ; Based on
 ; x86 SIMD extension for IJG JPEG library
 ; Copyright (C) 1999-2006, MIYASAKA Masaru.
 ; For conditions of distribution and use, see copyright notice in jsimdext.inc
 ;
 ; This file should be assembled with NASM (Netwide Assembler),
 ; can *not* be assembled with Microsoft's MASM or any compatible
 ; assembler (including Borland's Turbo Assembler).
 ; NASM is available from http://nasm.sourceforge.net/ or
 ; http://sourceforge.net/project/showfiles.php?group_id=6208
 ;
 ; [TAB8]
 
 %include "jcolsamp.inc"
                                 
 ; --------------------------------------------------------------------------
+%ifndef NEED_SECTION
+%define NEED_SECTION
         SECTION SEG_TEXT
         BITS    32
+%endif
+
 ;
 ; Upsample and color convert for the case of 2:1 horizontal and 1:1 vertical.
 ;
 ; GLOBAL(void)
 ; jsimd_h2v1_merged_upsample_sse2 (JDIMENSION output_width,
 ;                                  JSAMPIMAGE input_buf,
 ;                                  JDIMENSION in_row_group_ctr,
 ;                                  JSAMPARRAY output_buf);
 ;
 
@@ skipping 437 matching lines @@
         add     edi, byte SIZEOF_XMMWORD        ; outptr
         movdqa  xmmA,xmmD
         sub     ecx, byte SIZEOF_XMMWORD/4
 .column_st15:
 %ifdef STRICT_MEMORY_ACCESS
         ; Store two pixels (8 bytes) of xmmA to the output when it has enough
         ; space.
         cmp     ecx, byte SIZEOF_XMMWORD/8
         jb      short .column_st7
         movq    MMWORD [edi], xmmA
-        add     edi, byte SIZEOF_XMMWORD/2
+        add     edi, byte SIZEOF_XMMWORD/8*4
         sub     ecx, byte SIZEOF_XMMWORD/8
-        psrldq  xmmA, 64
+        psrldq  xmmA, SIZEOF_XMMWORD/8*4
 .column_st7:
         ; Store one pixel (4 bytes) of xmmA to the output when it has enough
         ; space.
         test    ecx, ecx
         jz      short .endcolumn
         movd    DWORD [edi], xmmA
 %else
         cmp     ecx, byte SIZEOF_XMMWORD/16
         jb      short .endcolumn
         mov     eax,ecx
@@ skipping 114 matching lines @@
         pop     esi
 ;       pop     edx             ; need not be preserved
 ;       pop     ecx             ; need not be preserved
         pop     ebx
         pop     ebp
         ret
 
 ; For some reason, the OS X linker does not honor the request to align the
 ; segment unless we do this.
         align   16