libvpx: Pull from upstream

source/libvpx/vp8/common/x86/recon_sse2.asm

 ;
 ;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
 ;
 ;  Use of this source code is governed by a BSD-style license
 ;  that can be found in the LICENSE file in the root of the source
 ;  tree. An additional intellectual property rights grant can be found
 ;  in the file PATENTS.  All contributing project authors may
 ;  be found in the AUTHORS file in the root of the source tree.
 ;
 
@@ skipping 872 matching lines @@
 ;    unsigned char *above,
 ;    unsigned char *left,
 ;    int left_stride
 ;    )
 %macro vp8_intra_pred_y_tm 1
 global sym(vp8_intra_pred_y_tm_%1) PRIVATE
 sym(vp8_intra_pred_y_tm_%1):
     push        rbp
     mov         rbp, rsp
     SHADOW_ARGS_TO_STACK 5
+    SAVE_XMM 7
     push        rsi
     push        rdi
     GET_GOT     rbx
     ; end prolog
 
     ; read top row
     mov         edx,        8
     mov         rsi,        arg(2) ;above
     movsxd      rax,        dword ptr arg(4) ;left_stride;
     pxor        xmm0,       xmm0
@@ skipping 47 matching lines @@
     movdqa [rdi+rcx],       xmm5
     lea         rsi,        [rsi+rax*2]
     lea         rdi,        [rdi+rcx*2]
     dec         edx
     jnz vp8_intra_pred_y_tm_%1_loop
 
     ; begin epilog
     RESTORE_GOT
     pop         rdi
     pop         rsi
+    RESTORE_XMM
     UNSHADOW_ARGS
     pop         rbp
     ret
 %endmacro
 
 vp8_intra_pred_y_tm sse2
 vp8_intra_pred_y_tm ssse3
 
 ;void vp8_intra_pred_y_ve_sse2(
 ;    unsigned char *dst,
@@ skipping 101 matching lines @@
 align 16
 dc_8:
     times 8 dw 8
 align 16
 dc_1024:
     times 8 dw 0x400
 align 16
 dc_00001111:
     times 8 db 0
     times 8 db 1