drop support for x86_32 PIC.

Author: pengvado

common/x86/cabac-a.asm

@@ -63,20 +63,13 @@ struc cb
 endstruc
 
 %macro LOAD_GLOBAL 4
-%ifdef PIC64
+%ifdef PIC
     ; this would be faster if the arrays were declared in asm, so that I didn't have to duplicate the lea
     lea   r11, [%2 GLOBAL]
     %ifnidn %3, 0
     add   r11, %3
     %endif
     movzx %1, byte [r11+%4]
-%elifdef PIC32
-    %ifnidn %3, 0
-    lea   %1, [%3+%4]
-    movzx %1, byte [%2+%1 GLOBAL]
-    %else
-    movzx %1, byte [%2+%3+%4 GLOBAL]
-    %endif
 %else
     movzx %1, byte [%2+%3+%4]
 %endif
@@ -85,7 +78,6 @@ endstruc
 cglobal x264_cabac_encode_decision_asm, 0,7
     movifnidn t0d, r0m
     movifnidn t1d, r1m
-    picgetgot t2
     mov   t5d, [r0+cb.range]
     movzx t3d, byte [r0+cb.state+t1]
     mov   t4d, t5d
@@ -95,22 +87,13 @@ cglobal x264_cabac_encode_decision_asm, 0,7
     sub   t4d, t5d
     mov   t6d, t3d
     shr   t6d, 6
-%ifdef PIC32
-    cmp   t6d, r2m
-%else
     movifnidn t2d, r2m
     cmp   t6d, t2d
-%endif
     mov   t6d, [r0+cb.low]
     lea   t7,  [t6+t4]
     cmovne t4d, t5d
     cmovne t6d, t7d
-%ifdef PIC32
-    mov   t1,  r2m
-    LOAD_GLOBAL t3d, x264_cabac_transition, t1, t3*2
-%else
     LOAD_GLOBAL t3d, x264_cabac_transition, t2, t3*2
-%endif
     movifnidn t1d, r1m
     mov   [r0+cb.state+t1], t3b
 .renorm:

common/x86/dct-32.asm

@@ -378,7 +378,6 @@ global x264_add8x8_idct8_sse2 %+ .skip_prologue
     IDCT8_1D   0,1,2,3,4,5,6,7,r1
     SPILL r1, 6
     TRANSPOSE8x8W 0,1,2,3,4,5,6,7,[r1+0x60],[r1+0x40],1
-    picgetgot  edx
     paddw      m0, [pw_32 GLOBAL]
     SPILL r1, 0
     IDCT8_1D   0,1,2,3,4,5,6,7,r1

common/x86/dct-a.asm

@@ -41,7 +41,7 @@ SECTION .text
 ;-----------------------------------------------------------------------------
 ; void x264_dct4x4dc_mmx( int16_t d[4][4] )
 ;-----------------------------------------------------------------------------
-cglobal x264_dct4x4dc_mmx, 1,1,1
+cglobal x264_dct4x4dc_mmx, 1,1
     movq   m0, [r0+ 0]
     movq   m1, [r0+ 8]
     movq   m2, [r0+16]
@@ -119,7 +119,7 @@ cglobal x264_sub4x4_dct_mmx, 3,3
 ;-----------------------------------------------------------------------------
 ; void x264_add4x4_idct_mmx( uint8_t *p_dst, int16_t dct[4][4] )
 ;-----------------------------------------------------------------------------
-cglobal x264_add4x4_idct_mmx, 2,2,1
+cglobal x264_add4x4_idct_mmx, 2,2
 .skip_prologue:
     movq  m0, [r1+ 0]
     movq  m1, [r1+ 8]
@@ -155,7 +155,7 @@ cglobal x264_sub8x8_dct_sse2, 3,3
     movhps [r0+56], m3
     ret
 
-cglobal x264_add8x8_idct_sse2, 2,2,1
+cglobal x264_add8x8_idct_sse2, 2,2
 .skip_prologue:
     call .8x4
     add  r1, 64
@@ -197,7 +197,7 @@ cglobal %1, 3,3
 ; void x264_add8x8_idct_mmx( uint8_t *pix, int16_t dct[4][4][4] )
 ;-----------------------------------------------------------------------------
 %macro ADD_NxN_IDCT 6
-cglobal %1, 2,2,1
+cglobal %1, 2,2
 .skip_prologue:
     call %2
     add  r0, %4-%5-%6*FDEC_STRIDE
@@ -268,7 +268,6 @@ cglobal x264_zigzag_sub_4x4_frame_ssse3, 3,3
     movd      [r2+1*FDEC_STRIDE], xmm1
     movd      [r2+2*FDEC_STRIDE], xmm2
     movd      [r2+3*FDEC_STRIDE], xmm3
-    picgetgot r1
     punpckldq xmm0, xmm1
     punpckldq xmm2, xmm3
     punpckldq xmm4, xmm5

common/x86/deblock-a.asm

@@ -373,7 +373,7 @@ cglobal x264_deblock_h_luma_sse2
 ;-----------------------------------------------------------------------------
 ; void x264_deblock_v8_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-cglobal x264_deblock_%2_luma_%1, 5,5,1
+cglobal x264_deblock_%2_luma_%1, 5,5
     lea     r4, [r1*3]
     dec     r2     ; alpha-1
     neg     r4
@@ -609,7 +609,7 @@ DEBLOCK_LUMA sse2, v, 16
 ;-----------------------------------------------------------------------------
 ; void x264_deblock_v_luma_intra_sse2( uint8_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal x264_deblock_%2_luma_intra_%1, 4,6,1
+cglobal x264_deblock_%2_luma_intra_%1, 4,6
 %ifndef ARCH_X86_64
     sub     esp, 0x60
 %endif
@@ -813,7 +813,6 @@ chroma_inter_body_mmxext:
     movd       m6, [r4] ; tc0
     punpcklbw  m6, m6
     pand       m7, m6
-    picgetgot r4
     DEBLOCK_P0_Q0
     ret
 
@@ -862,7 +861,6 @@ chroma_intra_body_mmxext:
     LOAD_MASK r2d, r3d
     movq   m5, m1
     movq   m6, m2
-    picgetgot r2
     CHROMA_INTRA_P0  m1, m0, m3
     CHROMA_INTRA_P0  m2, m3, m0
     psubb  m1, m5

common/x86/mc-a.asm

@@ -284,17 +284,9 @@ cglobal x264_pixel_avg2_w20_sse2, 6,7
 %macro INIT_SHIFT 2
     and    eax, 7
     shl    eax, 3
-%ifdef PIC32
-    ; both versions work, but picgetgot is slower than gpr->mmx is slower than mem->mmx
-    mov    r2, 64
-    sub    r2, eax
-    movd   %2, eax
-    movd   %1, r2
-%else
     movd   %1, [sw_64 GLOBAL]
     movd   %2, eax
     psubw  %1, %2
-%endif
 %endmacro
 
 %macro AVG_CACHELINE_CHECK 3 ; width, cacheline, instruction set
@@ -316,7 +308,7 @@ cglobal x264_pixel_avg2_w%1_cache%2_%3, 0,0
     INIT_SHIFT mm6, mm7
     mov    eax, r4m
     INIT_SHIFT mm4, mm5
-    PROLOGUE 6,6,0
+    PROLOGUE 6,6
     and    r2, ~7
     and    r4, ~7
     sub    r4, r2
@@ -510,7 +502,6 @@ COPY_W16_SSE2 x264_mc_copy_w16_aligned_sse2, movdqa
 %else
     SPLATW  m4, r4m
 %endif
-    picgetgot r4
     mova    m5, [pw_64 GLOBAL]
     psubw   m5, m4      ; weight_src
     mova    m6, [pw_32 GLOBAL] ; rounding
@@ -530,7 +521,7 @@ INIT_MMX
 ;-----------------------------------------------------------------------------
 ; int x264_pixel_avg_weight_w16_mmxext( uint8_t *dst, int, uint8_t *src, int, int i_weight, int )
 ;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_weight_4x4_mmxext, 4,4,1
+cglobal x264_pixel_avg_weight_4x4_mmxext, 4,4
     BIWEIGHT_START 0
     BIWEIGHT  [r0     ], [r2     ]
     BIWEIGHT  [r0+r1  ], [r2+r3  ]
@@ -671,7 +662,7 @@ cglobal x264_prefetch_ref_mmxext, 3,3
 ;                             int width, int height )
 ;-----------------------------------------------------------------------------
 %macro MC_CHROMA 1
-cglobal x264_mc_chroma_%1, 0,6,1
+cglobal x264_mc_chroma_%1, 0,6
 %if mmsize == 16
     cmp dword r6m, 4
     jle x264_mc_chroma_mmxext %+ .skip_prologue
@@ -833,7 +824,7 @@ INIT_XMM
 MC_CHROMA sse2
 
 INIT_MMX
-cglobal x264_mc_chroma_ssse3, 0,6,1
+cglobal x264_mc_chroma_ssse3, 0,6
     MC_CHROMA_START
     and       r4d, 7
     and       r5d, 7

common/x86/mc-a2.asm

@@ -111,7 +111,7 @@ INIT_MMX
 ;-----------------------------------------------------------------------------
 ; void x264_hpel_filter_v_mmxext( uint8_t *dst, uint8_t *src, int16_t *buf, int stride, int width );
 ;-----------------------------------------------------------------------------
-cglobal x264_hpel_filter_v_%1, 5,6,1
+cglobal x264_hpel_filter_v_%1, 5,6
     lea r5, [r1+r3]
     sub r1, r3
     sub r1, r3
@@ -148,7 +148,7 @@ HPEL_V mmxext
 ;-----------------------------------------------------------------------------
 ; void x264_hpel_filter_c_mmxext( uint8_t *dst, int16_t *buf, int width );
 ;-----------------------------------------------------------------------------
-cglobal x264_hpel_filter_c_mmxext, 3,3,1
+cglobal x264_hpel_filter_c_mmxext, 3,3
     add r0, r2
     lea r1, [r1+r2*2]
     neg r2
@@ -177,7 +177,7 @@ cglobal x264_hpel_filter_c_mmxext, 3,3,1
 ;-----------------------------------------------------------------------------
 ; void x264_hpel_filter_h_mmxext( uint8_t *dst, uint8_t *src, int width );
 ;-----------------------------------------------------------------------------
-cglobal x264_hpel_filter_h_mmxext, 3,3,1
+cglobal x264_hpel_filter_h_mmxext, 3,3
     add r0, r2
     add r1, r2
     neg r2
@@ -224,7 +224,7 @@ INIT_XMM
 ;-----------------------------------------------------------------------------
 ; void x264_hpel_filter_c_sse2( uint8_t *dst, int16_t *buf, int width );
 ;-----------------------------------------------------------------------------
-cglobal x264_hpel_filter_c_%1, 3,3,1
+cglobal x264_hpel_filter_c_%1, 3,3
     add r0, r2
     lea r1, [r1+r2*2]
     neg r2
@@ -267,7 +267,7 @@ cglobal x264_hpel_filter_c_%1, 3,3,1
 ;-----------------------------------------------------------------------------
 ; void x264_hpel_filter_h_sse2( uint8_t *dst, uint8_t *src, int width );
 ;-----------------------------------------------------------------------------
-cglobal x264_hpel_filter_h_sse2, 3,3,1
+cglobal x264_hpel_filter_h_sse2, 3,3
     add r0, r2
     add r1, r2
     neg r2

common/x86/pixel-a.asm

@@ -481,7 +481,6 @@ SATD_W4 mmxext
 %endmacro
 
 %macro SATD_END_SSE2 0
-    picgetgot ebx
     psrlw   m6, 1
     HADDW   m6, m7
     movd   eax, m6
@@ -643,7 +642,6 @@ cglobal x264_pixel_sa8d_8x8_%1, 4,7
     paddusw m0, m7
     paddusw m5, m4
     pavgw   m0, m5
-    picgetgot ebx
     HADDW   m0, m7
     movd eax, m0
     mov  ecx, eax ; preserve rounding for 16x16
@@ -1220,7 +1218,6 @@ cglobal x264_pixel_ssim_4x4x2_core_sse2, 4,4
 %endrep
     ; PHADDW m1, m2
     ; PHADDD m3, m4
-    picgetgot eax
     movdqa    m7, [pw_1 GLOBAL]
     pshufd    m5, m3, 0xb1
     pmaddwd   m1, m7
@@ -1267,7 +1264,6 @@ cglobal x264_pixel_ssim_end4_sse2, 3,3
     paddd     m1, m2
     paddd     m2, m3
     paddd     m3, m4
-    picgetgot r1
     movdqa    m5, [ssim_c1 GLOBAL]
     movdqa    m6, [ssim_c2 GLOBAL]
     TRANSPOSE4x4D  0, 1, 2, 3, 4
@@ -1298,8 +1294,8 @@ cglobal x264_pixel_ssim_end4_sse2, 3,3
     cmp       r2d, 4
     je .skip ; faster only if this is the common case; remove branch if we use ssim on a macroblock level
     neg       r2
-%ifdef PIC64
-    lea       r3,   [mask_ff + 16 GLOBAL]
+%ifdef PIC
+    lea       r3, [mask_ff + 16 GLOBAL]
     movdqu    m1, [r3 + r2*4]
 %else
     movdqu    m1, [mask_ff + r2*4 + 16 GLOBAL]