Avoid 'amdnop' when building with nasm

Other implementations of x86inc.asm have more comprehensive nasm
workarounds. This is the only thing that was changed for the previous
import to libvpx. See if we can still get away with it.

Change-Id: I3ef6fe9a4816461c89431a82b7e4a08b4b948d39

diff --git a/third_party/x86inc/README.libvpx b/third_party/x86inc/README.libvpx
index 629cecea7854f59cd57f50d30d74d19a6f0ec9fc..13d86adea9e1aad64225b4da02a028bf2d8c3156 100644 (file)
--- a/third_party/x86inc/README.libvpx
+++ b/third_party/x86inc/README.libvpx
@@ -14,3 +14,4 @@ Manage name mangling (prefixing with '_') manually because 'PREFIX' does not
   exist in libvpx.
 Expand PIC default to macho64 and respect CONFIG_PIC from libvpx
 Catch all elf formats for 'hidden' status and SECTION notes.
+Avoid 'amdnop' when building with nasm.

diff --git a/third_party/x86inc/x86inc.asm.new b/third_party/x86inc/x86inc.asm.new
index b38ae9ef0c8df50f01feab6f52997b66c9ada24c..c516a78773d5187c9117002426074044dc9770ec 100644 (file)
--- a/third_party/x86inc/x86inc.asm.new
+++ b/third_party/x86inc/x86inc.asm.new
@@ -806,7 +806,10 @@ SECTION .note.GNU-stack noalloc noexec nowrite progbits
         %endif
     %endif
 
-    %if ARCH_X86_64 || cpuflag(sse2)
+    %ifdef __NASM_VER__
+        %use smartalign
+        ALIGNMODE k7
+    %elif ARCH_X86_64 || cpuflag(sse2)
         CPU amdnop
     %else
         CPU basicnop