x86: PMINUB: port to cpuflags

[ffmpeg] / libavutil / x86 / x86inc.asm

diff --git a/libavutil/x86/x86inc.asm b/libavutil/x86/x86inc.asm
index 1030f10a3277f0702d1eb8e5ec3183fedbd99a1b..b0df2b20f074f732bfe5fa5ce0070fee716dc2b2 100644 (file)
--- a/libavutil/x86/x86inc.asm
+++ b/libavutil/x86/x86inc.asm
@@ -34,7 +34,9 @@
 ; as this feature might be useful for others as well.  Send patches or ideas
 ; to x264-devel@videolan.org .
 
-%define program_name ff
+%ifndef program_name
+    %define program_name ff
+%endif
 
 %define WIN64  0
 %define UNIX64 0
@@ -91,8 +93,14 @@
     default rel
 %endif
 
+%macro CPUNOP 1
+    %if HAVE_CPUNOP
+        CPU %1
+    %endif
+%endmacro
+
 ; Always use long nops (reduces 0x90 spam in disassembly on x86_32)
-CPU amdnop
+CPUNOP amdnop
 
 ; Macros to eliminate most code duplication between x86_32 and x86_64:
 ; Currently this works only for leaf functions which load all their arguments
@@ -134,15 +142,16 @@ CPU amdnop
     %define r%1w %2w
     %define r%1b %2b
     %define r%1h %2h
+    %define %2q %2
     %if %0 == 2
         %define r%1m  %2d
         %define r%1mp %2
     %elif ARCH_X86_64 ; memory
         %define r%1m [rsp + stack_offset + %3]
-        %define r%1mp qword r %+ %1m
+        %define r%1mp qword r %+ %1 %+ m
     %else
         %define r%1m [esp + stack_offset + %3]
-        %define r%1mp dword r %+ %1m
+        %define r%1mp dword r %+ %1 %+ m
     %endif
     %define r%1  %2
 %endmacro
@@ -441,6 +450,9 @@ DECLARE_ARG 7, 8, 9, 10, 11, 12, 13, 14
 %macro PROLOGUE 2-4+ ; #args, #regs, #xmm_regs, arg_names...
     %assign num_args %1
     %assign regs_used %2
+    %if num_args > 7
+        %assign num_args 7
+    %endif
     %if regs_used > 7
         %assign regs_used 7
     %endif
@@ -496,12 +508,8 @@ DECLARE_ARG 7, 8, 9, 10, 11, 12, 13, 14
 ; Applies any symbol mangling needed for C linkage, and sets up a define such that
 ; subsequent uses of the function name automatically refer to the mangled version.
 ; Appends cpuflags to the function name if cpuflags has been specified.
-%macro cglobal 1-2+ ; name, [PROLOGUE args]
-%if %0 == 1
-    cglobal_internal %1 %+ SUFFIX
-%else
+%macro cglobal 1-2+ "" ; name, [PROLOGUE args]
     cglobal_internal %1 %+ SUFFIX, %2
-%endif
 %endmacro
 %macro cglobal_internal 1-2+
     %ifndef cglobaled_%1
@@ -519,7 +527,7 @@ DECLARE_ARG 7, 8, 9, 10, 11, 12, 13, 14
     %1:
     RESET_MM_PERMUTATION ; not really needed, but makes disassembly somewhat nicer
     %assign stack_offset 0
-    %if %0 > 1
+    %ifnidn %2, ""
         PROLOGUE %2
     %endif
 %endmacro
@@ -554,7 +562,7 @@ SECTION .note.GNU-stack noalloc noexec nowrite progbits
 %assign cpuflags_mmx      (1<<0)
 %assign cpuflags_mmx2     (1<<1) | cpuflags_mmx
 %assign cpuflags_3dnow    (1<<2) | cpuflags_mmx
-%assign cpuflags_3dnow2   (1<<3) | cpuflags_3dnow
+%assign cpuflags_3dnowext (1<<3) | cpuflags_3dnow
 %assign cpuflags_sse      (1<<4) | cpuflags_mmx2
 %assign cpuflags_sse2     (1<<5) | cpuflags_sse
 %assign cpuflags_sse2slow (1<<6) | cpuflags_sse2
@@ -586,7 +594,7 @@ SECTION .note.GNU-stack noalloc noexec nowrite progbits
 ; All subsequent functions (up to the next INIT_CPUFLAGS) is built for the specified cpu.
 ; You shouldn't need to invoke this macro directly, it's a subroutine for INIT_MMX &co.
 %macro INIT_CPUFLAGS 0-2
-    CPU amdnop
+    CPUNOP amdnop
     %if %0 >= 1
         %xdefine cpuname %1
         %assign cpuflags cpuflags_%1
@@ -609,7 +617,7 @@ SECTION .note.GNU-stack noalloc noexec nowrite progbits
             %define movu lddqu
         %endif
         %if notcpuflag(mmx2)
-            CPU basicnop
+            CPUNOP basicnop
         %endif
     %else
         %xdefine SUFFIX
@@ -672,14 +680,6 @@ SECTION .note.GNU-stack noalloc noexec nowrite progbits
     INIT_CPUFLAGS %1
 %endmacro
 
-; FIXME: INIT_AVX can be replaced by INIT_XMM avx
-%macro INIT_AVX 0
-    INIT_XMM
-    %assign avx_enabled 1
-    %define PALIGNR PALIGNR_SSSE3
-    %define RESET_MM_PERMUTATION INIT_AVX
-%endmacro
-
 %macro INIT_YMM 0-1+
     %assign avx_enabled 1
     %define RESET_MM_PERMUTATION INIT_YMM %1
@@ -786,13 +786,13 @@ INIT_XMM
 
 ; Append cpuflags to the callee's name iff the appended name is known and the plain name isn't
 %macro call 1
-    call_internal %1, %1 %+ SUFFIX
+    call_internal %1 %+ SUFFIX, %1
 %endmacro
 %macro call_internal 2
-    %xdefine %%i %1
-    %ifndef cglobaled_%1
-        %ifdef cglobaled_%2
-            %xdefine %%i %2
+    %xdefine %%i %2
+    %ifndef cglobaled_%2
+        %ifdef cglobaled_%1
+            %xdefine %%i %1
         %endif
     %endif
     call %%i