diff -urp libdv-0.104-old/libdv/dct_block_mmx.S libdv-0.104/libdv/dct_block_mmx.S
--- libdv-0.104-old/libdv/dct_block_mmx.S	2005-10-23 19:40:58.000000000 +0200
+++ libdv-0.104/libdv/dct_block_mmx.S	2005-10-24 00:11:39.000000000 +0200
@@ -53,6 +53,17 @@ scratch2:       .quad 0
 scratch3:       .quad 0
 scratch4:       .quad 0
 
+#ifdef __PIC__
+# undef __i686 /* gcc define gets in our way */
+	.section .gnu.linkonce.t.__i686.get_pc_thunk.bp,"ax",@progbits
+.globl __i686.get_pc_thunk.bp
+	.hidden  __i686.get_pc_thunk.bp
+	.type    __i686.get_pc_thunk.bp,@function
+__i686.get_pc_thunk.bp:
+	movl (%esp), %ebp
+	ret
+#endif
+
 .text
 
 .align 8	
@@ -60,10 +71,14 @@ scratch4:       .quad 0
 _dv_dct_88_block_mmx:
 
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 
-	movl    8(%ebp), %esi          # source
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    12(%ebp), %esi          # source
 
 # column 0
 	movq 16*0(%esi), %mm0          # v0
@@ -86,22 +101,45 @@ _dv_dct_88_block_mmx:
 
 	movq 16*3(%esi), %mm5          # v3
 	movq 16*4(%esi), %mm7          # v4
+#ifdef __PIC__
+	movq  %mm7, scratch1@GOTOFF(%ebp)           # scratch1: v4   ; 
+#else
 	movq  %mm7, scratch1           # scratch1: v4   ; 
+#endif
 	movq  %mm5, %mm7               # duplicate v3 
+#ifdef __PIC__
+	paddw scratch1@GOTOFF(%ebp), %mm5           # v03: v3+v4  
+	psubw scratch1@GOTOFF(%ebp), %mm7           # v04: v3-v4  
+	movq  %mm5, scratch2@GOTOFF(%ebp)           # scratch2: v03
+#else
 	paddw scratch1, %mm5           # v03: v3+v4  
 	psubw scratch1, %mm7           # v04: v3-v4  
 	movq  %mm5, scratch2           # scratch2: v03
+#endif
 	movq  %mm0, %mm5               # mm5: v00
 
+#ifdef __PIC__
+	paddw scratch2@GOTOFF(%ebp), %mm0           # v10: v00+v03   
+	psubw scratch2@GOTOFF(%ebp), %mm5           # v13: v00-v03   
+	movq  %mm3, scratch3@GOTOFF(%ebp)           # scratch3: v02
+#else
 	paddw scratch2, %mm0           # v10: v00+v03   
 	psubw scratch2, %mm5           # v13: v00-v03   
 	movq  %mm3, scratch3           # scratch3: v02
+#endif
 	movq  %mm1, %mm3               # duplicate v01
 
+#ifdef __PIC__
+	paddw scratch3@GOTOFF(%ebp), %mm1          # v11: v01+v02
+	psubw scratch3@GOTOFF(%ebp), %mm3          # v12: v01-v02
+
+	movq  %mm6, scratch4@GOTOFF(%ebp)           # scratch4: v05
+#else
 	paddw scratch3, %mm1          # v11: v01+v02
 	psubw scratch3, %mm3          # v12: v01-v02
 
 	movq  %mm6, scratch4           # scratch4: v05
+#endif
 	movq  %mm0, %mm6               # duplicate v10
 
 	paddw %mm1, %mm0              # v10+v11
@@ -111,10 +149,18 @@ _dv_dct_88_block_mmx:
 	movq  %mm6, 16*4(%esi)         # out4: v10-v11 
 
 	movq  %mm4, %mm0               # mm0: v06
+#ifdef __PIC__
+	paddw scratch4@GOTOFF(%ebp), %mm4          # v15: v05+v06 
+#else
 	paddw scratch4, %mm4          # v15: v05+v06 
+#endif
 	paddw  %mm2, %mm0             # v16: v07+v06
 
+#ifdef __PIC__
+	pmulhw WA3@GOTOFF(%ebp), %mm4               # v35~: WA3*v15
+#else
 	pmulhw WA3, %mm4               # v35~: WA3*v15
+#endif
 	psllw  $1, %mm4                # v35: compensate the coeefient scale
 
 	movq   %mm4, %mm6              # duplicate v35
@@ -123,7 +169,11 @@ _dv_dct_88_block_mmx:
 
 	paddw  %mm5, %mm3             # v22: v12+v13
 
+#ifdef __PIC__
+	pmulhw WA1@GOTOFF(%ebp), %mm3               # v32~: WA1*v22
+#else
 	pmulhw WA1, %mm3               # v32~: WA1*v22
+#endif
 	psllw  $16-NSHIFT, %mm3        # v32: compensate the coeefient scale
 	movq   %mm5, %mm6              # duplicate v13
 
@@ -134,13 +184,23 @@ _dv_dct_88_block_mmx:
 	movq  %mm6, 16*6(%esi)         # out6: v13-v32 
 
 
+#ifdef __PIC__
+	paddw  scratch4@GOTOFF(%ebp), %mm7         # v14n: v04+v05
+#else
 	paddw  scratch4, %mm7         # v14n: v04+v05
+#endif
 	movq   %mm0, %mm5              # duplicate v16
 
 	psubw  %mm7, %mm0             # va1: v16-v14n
+#ifdef __PIC__
+	pmulhw WA5@GOTOFF(%ebp), %mm0               # va0~:  va1*WA5
+	pmulhw WA4@GOTOFF(%ebp), %mm5               # v36~~: v16*WA4
+	pmulhw WA2@GOTOFF(%ebp), %mm7               # v34~~: v14n*WA2
+#else
 	pmulhw WA5, %mm0               # va0~:  va1*WA5
 	pmulhw WA4, %mm5               # v36~~: v16*WA4
 	pmulhw WA2, %mm7               # v34~~: v14n*WA2
+#endif
 	psllw  $16-WA4_SHIFT, %mm5     # v36: compensate the coeefient scale 
 	psllw  $16-NSHIFT, %mm7        # v34: compensate the coeefient scale
 
@@ -188,22 +248,45 @@ _dv_dct_88_block_mmx:
 
 	movq 16*3(%esi), %mm5              # v3
 	movq 16*4(%esi), %mm7              # v4
+#ifdef __PIC__
+	movq  %mm7, scratch1@GOTOFF(%ebp)                    # scratch1: v4   ; 
+#else
 	movq  %mm7, scratch1                    # scratch1: v4   ; 
+#endif
 	movq  %mm5, %mm7               # duplicate v3 
+#ifdef __PIC__
+	paddw scratch1@GOTOFF(%ebp), %mm5           # v03: v3+v4  
+	psubw scratch1@GOTOFF(%ebp), %mm7           # v04: v3-v4  
+	movq  %mm5, scratch2@GOTOFF(%ebp)        # scratch2: v03
+#else
 	paddw scratch1, %mm5           # v03: v3+v4  
 	psubw scratch1, %mm7           # v04: v3-v4  
 	movq  %mm5, scratch2        # scratch2: v03
+#endif
 	movq  %mm0, %mm5               # mm5: v00
 
+#ifdef __PIC__
+	paddw scratch2@GOTOFF(%ebp), %mm0           # v10: v00+v03   
+	psubw scratch2@GOTOFF(%ebp), %mm5           # v13: v00-v03   
+	movq  %mm3, scratch3@GOTOFF(%ebp)         # scratc3: v02
+#else
 	paddw scratch2, %mm0           # v10: v00+v03   
 	psubw scratch2, %mm5           # v13: v00-v03   
 	movq  %mm3, scratch3         # scratc3: v02
+#endif
 	movq  %mm1, %mm3               # duplicate v01
 
+#ifdef __PIC__
+	paddw scratch3@GOTOFF(%ebp), %mm1           # v11: v01+v02
+	psubw scratch3@GOTOFF(%ebp), %mm3           # v12: v01-v02
+
+	movq  %mm6, scratch4@GOTOFF(%ebp)         # scratc4: v05
+#else
 	paddw scratch3, %mm1           # v11: v01+v02
 	psubw scratch3, %mm3           # v12: v01-v02
 
 	movq  %mm6, scratch4         # scratc4: v05
+#endif
 	movq  %mm0, %mm6               # duplicate v10
 
 	paddw %mm1, %mm0                            # v10+v11
@@ -213,10 +296,18 @@ _dv_dct_88_block_mmx:
 	movq  %mm6, 16*4(%esi)          # out4: v10-v11 
 
 	movq  %mm4, %mm0             # mm0: v06
+#ifdef __PIC__
+	paddw scratch4@GOTOFF(%ebp), %mm4         # v15: v05+v06 
+#else
 	paddw scratch4, %mm4         # v15: v05+v06 
+#endif
 	paddw  %mm2, %mm0                       # v16: v07+v06
 
+#ifdef __PIC__
+	pmulhw WA3@GOTOFF(%ebp), %mm4           # v35~: WA3*v15
+#else
 	pmulhw WA3, %mm4           # v35~: WA3*v15
+#endif
 	psllw  $16-NSHIFT, %mm4       # v35: compensate the coeefient scale
 
 	movq   %mm4, %mm6            # duplicate v35
@@ -225,7 +316,11 @@ _dv_dct_88_block_mmx:
 
 	paddw  %mm5, %mm3            # v22: v12+v13
 
+#ifdef __PIC__
+	pmulhw WA1@GOTOFF(%ebp), %mm3           # v32~: WA3*v15
+#else
 	pmulhw WA1, %mm3           # v32~: WA3*v15
+#endif
 	psllw  $16-NSHIFT, %mm3       # v32: compensate the coeefient scale
 	movq   %mm5, %mm6            # duplicate v13
 
@@ -235,13 +330,23 @@ _dv_dct_88_block_mmx:
 	movq  %mm5, 16*2(%esi)          # out2: v13+v32 
 	movq  %mm6, 16*6(%esi)          # out6: v13-v32 
 
+#ifdef __PIC__
+	paddw  scratch4@GOTOFF(%ebp), %mm7                           # v14n: v04+v05
+#else
 	paddw  scratch4, %mm7                           # v14n: v04+v05
+#endif
 	movq   %mm0, %mm5                               # duplicate v16
 
 	psubw  %mm7, %mm0                               # va1: v16-v14n
+#ifdef __PIC__
+	pmulhw WA2@GOTOFF(%ebp), %mm7                # v34~~: v14n*WA2
+	pmulhw WA5@GOTOFF(%ebp), %mm0                # va0~:  va1*WA5
+	pmulhw WA4@GOTOFF(%ebp), %mm5                        # v36~~: v16*WA4
+#else
 	pmulhw WA2, %mm7                # v34~~: v14n*WA2
 	pmulhw WA5, %mm0                # va0~:  va1*WA5
 	pmulhw WA4, %mm5                        # v36~~: v16*WA4
+#endif
 	psllw  $16-NSHIFT, %mm7
 	psllw  $16-WA4_SHIFT, %mm5      # v36: compensate the coeffient 
 		# scale note that WA4 is shifted 1 bit less than the others
@@ -751,11 +856,15 @@ _dv_dct_block_mmx_postscale_88:
 _dv_dct_248_block_mmx:
 
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl   %edi
 
-	movl    8(%ebp), %esi          # source
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    16(%ebp), %esi          # source
 
 # column 0
 
@@ -779,7 +888,11 @@ _dv_dct_248_block_mmx:
 	paddw %mm1, %mm0	       # v20: v10+v11
 	psubw %mm1, %mm3	       # v21: v10-v11
 
+#ifdef __PIC__
+	pmulhw WA1@GOTOFF(%ebp), %mm5               # v32~: WA1*v22
+#else
 	pmulhw WA1, %mm5               # v32~: WA1*v22
+#endif
 	movq  %mm4, %mm2	
 	psllw  $16-NSHIFT, %mm5        # v32: compensate the coeffient scale
 
@@ -818,7 +931,11 @@ _dv_dct_248_block_mmx:
 	paddw %mm1, %mm0	       # v20: v10+v11
 	psubw %mm1, %mm3	       # v21: v10-v11
 
+#ifdef __PIC__
+	pmulhw WA1@GOTOFF(%ebp), %mm5               # v32~: WA1*v22
+#else
 	pmulhw WA1, %mm5               # v32~: WA1*v22
+#endif
 	movq  %mm4, %mm2	
 	psllw  $16-NSHIFT, %mm5        # v32: compensate the coeffient scale
 
@@ -855,7 +972,11 @@ _dv_dct_248_block_mmx:
 	paddw %mm1, %mm0	       # v20: v10+v11
 	psubw %mm1, %mm3	       # v21: v10-v11
 
+#ifdef __PIC__
+	pmulhw WA1@GOTOFF(%ebp), %mm5               # v32~: WA1*v22
+#else
 	pmulhw WA1, %mm5               # v32~: WA1*v22
+#endif
 	movq  %mm4, %mm2	
 	psllw  $16-NSHIFT, %mm5        # v32: compensate the coeffient scale
 
@@ -892,7 +1013,11 @@ _dv_dct_248_block_mmx:
 	paddw %mm1, %mm0	       # v20: v10+v11
 	psubw %mm1, %mm3	       # v21: v10-v11
 
+#ifdef __PIC__
+	pmulhw WA1@GOTOFF(%ebp), %mm5               # v32~: WA1*v22
+#else
 	pmulhw WA1, %mm5               # v32~: WA1*v22
+#endif
 	movq  %mm4, %mm2	
 	psllw  $16-NSHIFT, %mm5        # v32: compensate the coeffient scale
 
diff -urp libdv-0.104-old/libdv/dv.c libdv-0.104/libdv/dv.c
--- libdv-0.104-old/libdv/dv.c	2004-10-20 05:49:24.000000000 +0200
+++ libdv-0.104/libdv/dv.c	2005-10-24 00:59:57.000000000 +0200
@@ -205,6 +205,9 @@ dv_reconfigure(int clamp_luma, int clamp
 } /* dv_reconfigure */
 
 
+extern uint8_t dv_quant_offset[4];
+extern uint8_t dv_quant_shifts[22][4];
+
 static inline void 
 dv_decode_macroblock(dv_decoder_t *dv, dv_macroblock_t *mb, unsigned int quality) {
   int i;
@@ -218,7 +221,7 @@ dv_decode_macroblock(dv_decoder_t *dv, d
       dv_idct_248 (co248, mb->b[i].coeffs);
     } else {
 #if ARCH_X86
-      _dv_quant_88_inverse_x86(mb->b[i].coeffs,mb->qno,mb->b[i].class_no);
+      _dv_quant_88_inverse_x86(mb->b[i].coeffs,mb->qno,mb->b[i].class_no,dv_quant_offset,dv_quant_shifts);
       _dv_idct_88(mb->b[i].coeffs);
 #elif ARCH_X86_64
       _dv_quant_88_inverse_x86_64(mb->b[i].coeffs,mb->qno,mb->b[i].class_no);
@@ -250,7 +253,7 @@ dv_decode_video_segment(dv_decoder_t *dv
 	dv_idct_248 (co248, mb->b[b].coeffs);
       } else {
 #if ARCH_X86
-	_dv_quant_88_inverse_x86(bl->coeffs,mb->qno,bl->class_no);
+	_dv_quant_88_inverse_x86(bl->coeffs,mb->qno,bl->class_no,dv_quant_offset,dv_quant_shifts);
 	_dv_weight_88_inverse(bl->coeffs);
 	_dv_idct_88(bl->coeffs);
 #elif ARCH_X86_64
diff -urp libdv-0.104-old/libdv/encode.c libdv-0.104/libdv/encode.c
--- libdv-0.104-old/libdv/encode.c	2004-11-17 04:36:30.000000000 +0100
+++ libdv-0.104/libdv/encode.c	2005-10-24 01:17:41.000000000 +0200
@@ -521,7 +521,8 @@ static void reorder_block(dv_block_t *bl
 }
 
 extern unsigned long _dv_vlc_encode_block_mmx(dv_coeff_t* coeffs,
-					  dv_vlc_entry_t ** out);
+					  dv_vlc_entry_t ** out,
+					  dv_vlc_entry_t * lookup);
 
 extern unsigned long _dv_vlc_encode_block_mmx_x86_64(dv_coeff_t* coeffs,
 					  dv_vlc_entry_t ** out);
@@ -558,7 +559,7 @@ static unsigned long vlc_encode_block(dv
 #elif ARCH_X86
 	int num_bits;
 
-	num_bits = _dv_vlc_encode_block_mmx(coeffs, &o);
+	num_bits = _dv_vlc_encode_block_mmx(coeffs, &o, vlc_encode_lookup);
 	emms();
 #else
 	int num_bits;
@@ -574,7 +575,7 @@ static unsigned long vlc_encode_block(dv
 	return num_bits;
 }
 
-extern unsigned long _dv_vlc_num_bits_block_x86(dv_coeff_t* coeffs);
+extern unsigned long _dv_vlc_num_bits_block_x86(dv_coeff_t* coeffs, unsigned char* lookup);
 extern unsigned long _dv_vlc_num_bits_block_x86_64(dv_coeff_t* coeffs);
 
 extern unsigned long _dv_vlc_num_bits_block(dv_coeff_t* coeffs)
@@ -600,7 +601,7 @@ extern unsigned long _dv_vlc_num_bits_bl
 #elif ARCH_X86_64
 	return _dv_vlc_num_bits_block_x86_64(coeffs);
 #else
-	return _dv_vlc_num_bits_block_x86(coeffs);
+	return _dv_vlc_num_bits_block_x86(coeffs, vlc_num_bits_lookup);
 #endif
 }
 
diff -urp libdv-0.104-old/libdv/encode_x86.S libdv-0.104/libdv/encode_x86.S
--- libdv-0.104-old/libdv/encode_x86.S	2005-10-23 19:40:58.000000000 +0200
+++ libdv-0.104/libdv/encode_x86.S	2005-10-24 01:18:32.000000000 +0200
@@ -23,10 +23,6 @@
  *  The libdv homepage is http://libdv.sourceforge.net/.  
  */
 
-.data
-ALLONE:		.word 1,1,1,1
-VLCADDMASK:	.byte 255,0,0,0,255,0,0,0
-		
 .text
 
 .global _dv_vlc_encode_block_mmx
@@ -45,11 +41,14 @@ _dv_vlc_encode_block_mmx:	
 
 	movl	$63, %ecx
 
-	movl	vlc_encode_lookup, %esi
+	movl	4+4*4+8(%esp), %esi              # vlc_encode_lookup
 
 	pxor	%mm0, %mm0
 	pxor	%mm2, %mm2
-	movq	VLCADDMASK, %mm1
+	pushl	$0x000000FF
+	pushl	$0x000000FF
+	movq	(%esp), %mm1
+	addl	$8, %esp
 	xorl	%ebp, %ebp
 	subl	$8, %edx
 vlc_encode_block_mmx_loop:
@@ -121,7 +120,7 @@ _dv_vlc_num_bits_block_x86:	
 	addl	$2, %edi
 
 	movl	$63, %ecx
-	movl	vlc_num_bits_lookup, %esi
+	movl	4+4*4+4(%esp), %esi              # vlc_num_bits_lookup
 	
 vlc_num_bits_block_x86_loop:
 	movw	(%edi), %ax
@@ -579,8 +578,11 @@ _dv_need_dct_248_mmx_rows:
 	paddw	%mm5, %mm1
 
 	paddw	%mm1, %mm0
-	
-	pmaddwd	ALLONE, %mm0	
+
+	pushl	$0x00010001
+	pushl	$0x00010001
+	pmaddwd	(%esp), %mm0
+	addl	$8, %esp
 	movq	%mm0, %mm1
 	psrlq	$32, %mm1
 	paddd	%mm1, %mm0
diff -urp libdv-0.104-old/libdv/idct_block_mmx.S libdv-0.104/libdv/idct_block_mmx.S
--- libdv-0.104-old/libdv/idct_block_mmx.S	2005-10-23 19:40:58.000000000 +0200
+++ libdv-0.104/libdv/idct_block_mmx.S	2005-10-24 01:12:12.000000000 +0200
@@ -8,16 +8,37 @@
 
 
 
+#ifdef __PIC__
+# undef __i686 /* gcc define gets in our way */
+	.section .gnu.linkonce.t.__i686.get_pc_thunk.bp,"ax",@progbits
+.globl __i686.get_pc_thunk.bp
+	.hidden  __i686.get_pc_thunk.bp
+	.type    __i686.get_pc_thunk.bp,@function
+__i686.get_pc_thunk.bp:
+	movl (%esp), %ebp
+	ret
+#endif
+
 .text
+
 	.align 4
 .globl _dv_idct_block_mmx 
 	.type	 _dv_idct_block_mmx,@function
 _dv_idct_block_mmx:
 	pushl	 %ebp
-	movl	 %esp,%ebp
 	pushl	 %esi
+
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+#ifdef __PIC__
+	leal	 preSC@GOTOFF(%ebp), %ecx
+#else
 	leal	 preSC, %ecx
-	movl	 8(%ebp),%esi		/* source matrix */
+#endif
+	movl	 12(%esp),%esi		/* source matrix */
 
 /* 
  *	column 0: even part
@@ -35,7 +56,11 @@ _dv_idct_block_mmx:
 	movq %mm1, %mm2			/* added 11/1/96 */
 	pmulhw 8*8(%esi),%mm5		/* V8 */
 	psubsw %mm0, %mm1		/* V16 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp), %mm1	/* 23170 ->V18 */
+#else
 	pmulhw x5a825a825a825a82, %mm1	/* 23170 ->V18 */
+#endif
 	paddsw %mm0, %mm2		/* V17 */
 	movq %mm2, %mm0			/* duplicate V17 */
 	psraw $1, %mm2			/* t75=t82 */
@@ -76,7 +101,11 @@ _dv_idct_block_mmx:
 	paddsw %mm0, %mm3		/* V29 ; free mm0 */
 	movq %mm7, %mm1			/* duplicate V26 */
 	psraw $1, %mm3			/* t91=t94 */
+#ifdef __PIC__
+	pmulhw x539f539f539f539f@GOTOFF(%ebp),%mm7	/* V33 */
+#else
 	pmulhw x539f539f539f539f,%mm7	/* V33 */
+#endif
 	psraw $1, %mm1			/* t96 */
 	movq %mm5, %mm0			/* duplicate V2 */
 	psraw $2, %mm4			/* t85=t87 */
@@ -84,15 +113,27 @@ _dv_idct_block_mmx:
 	psubsw %mm4, %mm0		/* V28 ; free mm4 */
 	movq %mm0, %mm2			/* duplicate V28 */
 	psraw $1, %mm5			/* t90=t93 */
+#ifdef __PIC__
+	pmulhw x4546454645464546@GOTOFF(%ebp),%mm0	/* V35 */
+#else
 	pmulhw x4546454645464546,%mm0	/* V35 */
+#endif
 	psraw $1, %mm2			/* t97 */
 	movq %mm5, %mm4			/* duplicate t90=t93 */
 	psubsw %mm2, %mm1		/* V32 ; free mm2 */
+#ifdef __PIC__
+	pmulhw x61f861f861f861f8@GOTOFF(%ebp),%mm1	/* V36 */
+#else
 	pmulhw x61f861f861f861f8,%mm1	/* V36 */
+#endif
 	psllw $1, %mm7			/* t107 */
 	paddsw %mm3, %mm5		/* V31 */
 	psubsw %mm3, %mm4		/* V30 ; free mm3 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp),%mm4	/* V34 */
+#else
 	pmulhw x5a825a825a825a82,%mm4	/* V34 */
+#endif
 	nop
 	psubsw %mm1, %mm0		/* V38 */
 	psubsw %mm7, %mm1		/* V37 ; free mm7 */
@@ -159,7 +200,11 @@ _dv_idct_block_mmx:
 	psubsw %mm7, %mm1		/* V50 */
 	pmulhw 8*9(%esi), %mm5		/* V9 */
 	paddsw %mm7, %mm2		/* V51 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp), %mm1	/* 23170 ->V52 */
+#else
 	pmulhw x5a825a825a825a82, %mm1	/* 23170 ->V52 */
+#endif
 	movq %mm2, %mm6			/* duplicate V51 */
 	psraw $1, %mm2			/* t138=t144 */
 	movq %mm3, %mm4			/* duplicate V1 */
@@ -200,11 +245,19 @@ _dv_idct_block_mmx:
  * even more by doing the correction step in a later stage when the number
  * is actually multiplied by 16
  */
+#ifdef __PIC__
+	paddw x0005000200010001@GOTOFF(%ebp), %mm4
+#else
 	paddw x0005000200010001, %mm4
+#endif
 	psubsw %mm6, %mm3		/* V60 ; free mm6 */
 	psraw $1, %mm0			/* t154=t156 */
 	movq %mm3, %mm1			/* duplicate V60 */
+#ifdef __PIC__
+	pmulhw x539f539f539f539f@GOTOFF(%ebp), %mm1	/* V67 */
+#else
 	pmulhw x539f539f539f539f, %mm1	/* V67 */
+#endif
 	movq %mm5, %mm6			/* duplicate V3 */
 	psraw $2, %mm4			/* t148=t150 */
 	paddsw %mm4, %mm5		/* V61 */
@@ -213,13 +266,25 @@ _dv_idct_block_mmx:
 	psllw $1, %mm1			/* t169 */
 	paddsw %mm0, %mm5		/* V65 -> result */
 	psubsw %mm0, %mm4		/* V64 ; free mm0 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp), %mm4	/* V68 */
+#else
 	pmulhw x5a825a825a825a82, %mm4	/* V68 */
+#endif
 	psraw $1, %mm3			/* t158 */
 	psubsw %mm6, %mm3		/* V66 */
 	movq %mm5, %mm2			/* duplicate V65 */
+#ifdef __PIC__
+	pmulhw x61f861f861f861f8@GOTOFF(%ebp), %mm3	/* V70 */
+#else
 	pmulhw x61f861f861f861f8, %mm3	/* V70 */
+#endif
 	psllw $1, %mm6			/* t165 */
+#ifdef __PIC__
+	pmulhw x4546454645464546@GOTOFF(%ebp), %mm6	/* V69 */
+#else
 	pmulhw x4546454645464546, %mm6	/* V69 */
+#endif
 	psraw $1, %mm2			/* t172 */
 /* moved from next block */
 	movq 8*5(%esi), %mm0		/* V56 */
@@ -344,7 +409,11 @@ _dv_idct_block_mmx:
 *	movq 8*13(%esi), %mm4		tmt13
 */
 	psubsw %mm4, %mm3		/* V134 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp), %mm3	/* 23170 ->V136 */
+#else
 	pmulhw x5a825a825a825a82, %mm3	/* 23170 ->V136 */
+#endif
 	movq 8*9(%esi), %mm6		/* tmt9 */
 	paddsw %mm4, %mm5		/* V135 ; mm4 free */
 	movq %mm0, %mm4			/* duplicate tmt1 */
@@ -373,17 +442,33 @@ _dv_idct_block_mmx:
 	psubsw %mm7, %mm0		/* V144 */
 	movq %mm0, %mm3			/* duplicate V144 */
 	paddsw %mm7, %mm2		/* V147 ; free mm7 */
+#ifdef __PIC__
+	pmulhw x539f539f539f539f@GOTOFF(%ebp), %mm0	/* 21407-> V151 */
+#else
 	pmulhw x539f539f539f539f, %mm0	/* 21407-> V151 */
+#endif
 	movq %mm1, %mm7			/* duplicate tmt3 */
 	paddsw %mm5, %mm7		/* V145 */
 	psubsw %mm5, %mm1		/* V146 ; free mm5 */
 	psubsw %mm1, %mm3		/* V150 */
 	movq %mm7, %mm5			/* duplicate V145 */
+#ifdef __PIC__
+	pmulhw x4546454645464546@GOTOFF(%ebp), %mm1	/* 17734-> V153 */
+#else
 	pmulhw x4546454645464546, %mm1	/* 17734-> V153 */
+#endif
 	psubsw %mm2, %mm5		/* V148 */
+#ifdef __PIC__
+	pmulhw x61f861f861f861f8@GOTOFF(%ebp), %mm3	/* 25080-> V154 */
+#else
 	pmulhw x61f861f861f861f8, %mm3	/* 25080-> V154 */
+#endif
 	psllw $2, %mm0			/* t311 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp), %mm5	/* 23170-> V152 */
+#else
 	pmulhw x5a825a825a825a82, %mm5	/* 23170-> V152 */
+#endif
 	paddsw %mm2, %mm7		/* V149 ; free mm2 */
 	psllw $1, %mm1			/* t313 */
 	nop	/* without the nop - freeze here for one clock */
@@ -409,7 +494,11 @@ _dv_idct_block_mmx:
 	paddsw %mm3, %mm6		/* V164 ; free mm3 */
 	movq %mm4, %mm3			/* duplicate V142 */
 	psubsw %mm5, %mm4		/* V165 ; free mm5 */
+#ifdef __PIC__
+	movq %mm2, scratch7@GOTOFF(%ebp)		/* out7 */
+#else
 	movq %mm2, scratch7		/* out7 */
+#endif
 	psraw $4, %mm6
 	psraw $4, %mm4
 	paddsw %mm5, %mm3		/* V162 */
@@ -420,11 +509,19 @@ _dv_idct_block_mmx:
  */
 	movq %mm6, 8*9(%esi)		/* out9 */
 	paddsw %mm1, %mm0		/* V161 */
+#ifdef __PIC__
+	movq %mm3, scratch5@GOTOFF(%ebp)		/* out5 */
+#else
 	movq %mm3, scratch5		/* out5 */
+#endif
 	psubsw %mm1, %mm5		/* V166 ; free mm1 */
 	movq %mm4, 8*11(%esi)		/* out11 */
 	psraw $4, %mm5
+#ifdef __PIC__
+	movq %mm0, scratch3@GOTOFF(%ebp)		/* out3 */
+#else
 	movq %mm0, scratch3		/* out3 */
+#endif
 	movq %mm2, %mm4			/* duplicate V140 */
 	movq %mm5, 8*13(%esi)		/* out13 */
 	paddsw %mm7, %mm2		/* V160 */
@@ -434,7 +531,11 @@ _dv_idct_block_mmx:
 /* moved from the next block */
 	movq 8*3(%esi), %mm7
 	psraw $4, %mm4
+#ifdef __PIC__
+	movq %mm2, scratch1@GOTOFF(%ebp)		/* out1 */
+#else
 	movq %mm2, scratch1		/* out1 */
+#endif
 /* moved from the next block */
 	movq %mm0, %mm1
 	movq %mm4, 8*15(%esi)		/* out15 */
@@ -491,15 +592,31 @@ _dv_idct_block_mmx:
 	paddsw %mm4, %mm3		/* V113 ; free mm4 */
 	movq %mm0, %mm4			/* duplicate V110 */
 	paddsw %mm1, %mm2		/* V111 */
+#ifdef __PIC__
+	pmulhw x539f539f539f539f@GOTOFF(%ebp), %mm0	/* 21407-> V117 */
+#else
 	pmulhw x539f539f539f539f, %mm0	/* 21407-> V117 */
+#endif
 	psubsw %mm1, %mm5		/* V112 ; free mm1 */
 	psubsw %mm5, %mm4		/* V116 */
 	movq %mm2, %mm1			/* duplicate V111 */
+#ifdef __PIC__
+	pmulhw x4546454645464546@GOTOFF(%ebp), %mm5	/* 17734-> V119 */
+#else
 	pmulhw x4546454645464546, %mm5	/* 17734-> V119 */
+#endif
 	psubsw %mm3, %mm2		/* V114 */
+#ifdef __PIC__
+	pmulhw x61f861f861f861f8@GOTOFF(%ebp), %mm4	/* 25080-> V120 */
+#else
 	pmulhw x61f861f861f861f8, %mm4	/* 25080-> V120 */
+#endif
 	paddsw %mm3, %mm1		/* V115 ; free mm3 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp), %mm2	/* 23170-> V118 */
+#else
 	pmulhw x5a825a825a825a82, %mm2	/* 23170-> V118 */
+#endif
 	psllw $2, %mm0			/* t266 */
 	movq %mm1, (%esi)		/* save V115 */
 	psllw $1, %mm5			/* t268 */
@@ -517,7 +634,11 @@ _dv_idct_block_mmx:
 	movq %mm6, %mm3			/* duplicate tmt4 */
 	psubsw %mm0, %mm6		/* V100 */
 	paddsw %mm0, %mm3		/* V101 ; free mm0 */
+#ifdef __PIC__
+	pmulhw x5a825a825a825a82@GOTOFF(%ebp), %mm6	/* 23170 ->V102 */
+#else
 	pmulhw x5a825a825a825a82, %mm6	/* 23170 ->V102 */
+#endif
 	movq %mm7, %mm5			/* duplicate tmt0 */
 	movq 8*8(%esi), %mm1		/* tmt8 */
 	paddsw %mm1, %mm7		/* V103 */
@@ -551,10 +672,18 @@ _dv_idct_block_mmx:
 	movq 8*2(%esi), %mm3		/* V123 */
 	paddsw %mm4, %mm7		/* out0 */
 /* moved up from next block */
+#ifdef __PIC__
+	movq scratch3@GOTOFF(%ebp), %mm0
+#else
 	movq scratch3, %mm0
+#endif
 	psraw $4, %mm7
 /* moved up from next block */
+#ifdef __PIC__
+	movq scratch5@GOTOFF(%ebp), %mm6 
+#else
 	movq scratch5, %mm6 
+#endif
 	psubsw %mm4, %mm1		/* out14 ; free mm4 */
 	paddsw %mm3, %mm5		/* out2 */
 	psraw $4, %mm1
@@ -565,7 +694,11 @@ _dv_idct_block_mmx:
 	movq %mm5, 8*2(%esi)		/* out2 ; free mm5 */
 	psraw $4, %mm2
 /* moved up to the prev block */
+#ifdef __PIC__
+	movq scratch7@GOTOFF(%ebp), %mm4
+#else
 	movq scratch7, %mm4
+#endif
 /* moved up to the prev block */
 	psraw $4, %mm0
 	movq %mm2, 8*12(%esi)		/* out12 ; free mm2 */
@@ -579,7 +712,11 @@ _dv_idct_block_mmx:
  *	psraw $4, %mm0
  *	psraw $4, %mm6
 */
+#ifdef __PIC__
+	movq scratch1@GOTOFF(%ebp), %mm1
+#else
 	movq scratch1, %mm1
+#endif
 	psraw $4, %mm4
 	movq %mm0, 8*3(%esi)		/* out3 */
 	psraw $4, %mm1
diff -urp libdv-0.104-old/libdv/quant.c libdv-0.104/libdv/quant.c
--- libdv-0.104-old/libdv/quant.c	2004-10-20 05:49:24.000000000 +0200
+++ libdv-0.104/libdv/quant.c	2005-10-24 01:06:24.000000000 +0200
@@ -144,7 +144,7 @@ uint8_t  dv_quant_offset[4] = { 6,3,0,1 
 uint32_t	dv_quant_248_mul_tab [2] [22] [64];
 uint32_t dv_quant_88_mul_tab [2] [22] [64];
 
-extern void             _dv_quant_x86(dv_coeff_t *block,int qno,int klass);
+extern void             _dv_quant_x86(dv_coeff_t *block,int qno,int klass,uint8_t  dv_quant_offset[],uint8_t dv_quant_shifts[][]);
 extern void             _dv_quant_x86_64(dv_coeff_t *block,int qno,int klass);
 static void quant_248_inverse_std(dv_coeff_t *block,int qno,int klass,dv_248_coeff_t *co);
 static void quant_248_inverse_mmx(dv_coeff_t *block,int qno,int klass,dv_248_coeff_t *co);
@@ -210,7 +210,7 @@ void _dv_quant(dv_coeff_t *block,int qno
 		_dv_quant_x86_64(block, qno, klass);
 		emms();
 #else
-		_dv_quant_x86(block, qno, klass);
+		_dv_quant_x86(block, qno, klass, dv_quant_offset, dv_quant_shifts);
 		emms();
 #endif
 	}
diff -urp libdv-0.104-old/libdv/quant.h libdv-0.104/libdv/quant.h
--- libdv-0.104-old/libdv/quant.h	2004-10-20 05:49:24.000000000 +0200
+++ libdv-0.104/libdv/quant.h	2005-10-24 00:57:43.000000000 +0200
@@ -27,7 +27,7 @@ extern void _dv_quant(dv_coeff_t *block,
 extern void _dv_quant_88_inverse(dv_coeff_t *block,int qno,int klass);
 extern void (*_dv_quant_248_inverse) (dv_coeff_t *block,int qno,int klass,
                                   dv_248_coeff_t *co);
-extern void _dv_quant_88_inverse_x86(dv_coeff_t *block,int qno,int klass);
+extern void _dv_quant_88_inverse_x86(dv_coeff_t *block,int qno,int klass, uint8_t offset[], uint8_t shifts[][]);
 extern void _dv_quant_88_inverse_x86_64(dv_coeff_t *block,int qno,int klass);
 extern void dv_quant_init (void);
 #ifdef __cplusplus
diff -urp libdv-0.104-old/libdv/quant_x86.S libdv-0.104/libdv/quant_x86.S
--- libdv-0.104-old/libdv/quant_x86.S	2005-10-23 19:40:58.000000000 +0200
+++ libdv-0.104/libdv/quant_x86.S	2005-10-24 01:10:21.000000000 +0200
@@ -71,10 +71,13 @@ _dv_quant_88_inverse_x86:	
 	
 	/*  pq = dv_quant_shifts[qno + dv_quant_offset[class]]; */
 	movl	ARGn(1),%eax	/* qno */
+	movl	ARGn(3),%ebx	/* dv_quant_offset */
+	addl	ARGn(2),%ebx	/* class */
+	movzbl	(%ebx),%ecx
 	movl	ARGn(2),%ebx	/* class */
-	movzbl	dv_quant_offset(%ebx),%ecx
 	addl	%ecx,%eax
-	leal	dv_quant_shifts(,%eax,4),%edx	/* edx is pq */
+	movl	ARGn(4),%edx	/* dv_quant_shifts */
+	leal	(%edx,%eax,4),%edx	/* edx is pq */
 
 	/* extra = (class == 3); */
 				/*  0   1   2   3 */
@@ -212,11 +215,13 @@ _dv_quant_x86:	
 	
 	/*  pq = dv_quant_shifts[qno + dv_quant_offset[class]]; */
 	movl	ARGn(1),%eax	/* qno */
+	movl	ARGn(3),%ebx	/* offset */
+	addl	ARGn(2),%ebx	/* class */
+	movzbl	(%ebx),%ecx
 	movl	ARGn(2),%ebx	/* class */
-
-	movzbl	dv_quant_offset(%ebx),%ecx
+	movl	ARGn(4),%edx	/* shifts */
 	addl	%ecx,%eax
-	leal	dv_quant_shifts(,%eax,4),%edx	/* edx is pq */
+	leal	(%edx,%eax,4),%edx	/* edx is pq */
 
 	/* extra = (class == 3); */
 				/*  0   1   2   3 */
diff -urp libdv-0.104-old/libdv/rgbtoyuv.S libdv-0.104/libdv/rgbtoyuv.S
--- libdv-0.104-old/libdv/rgbtoyuv.S	2005-10-23 19:40:58.000000000 +0200
+++ libdv-0.104/libdv/rgbtoyuv.S	2005-10-24 00:46:34.000000000 +0200
@@ -110,20 +110,30 @@ VR0GR:  .long   0,0
 VBG0B:  .long   0,0
 	
 #endif	
-	
+
+#ifdef __PIC__
+# undef __i686 /* gcc define gets in our way */
+	.section .gnu.linkonce.t.__i686.get_pc_thunk.bp,"ax",@progbits
+.globl __i686.get_pc_thunk.bp
+	.hidden  __i686.get_pc_thunk.bp
+	.type    __i686.get_pc_thunk.bp,@function
+__i686.get_pc_thunk.bp:
+	movl (%esp), %ebp
+	ret
+#endif
+
 .text
 
-#define _inPtr     8
-#define _rows      12
-#define _columns   16
-#define _outyPtr   20
-#define _outuPtr   24
-#define _outvPtr   28
+#define _inPtr     24+8
+#define _rows      24+12
+#define _columns   24+16
+#define _outyPtr   24+20
+#define _outuPtr   24+24
+#define _outvPtr   24+28
 
 _dv_rgbtoycb_mmx:
 
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %eax
 	pushl   %ebx
 	pushl   %ecx
@@ -131,46 +141,103 @@ _dv_rgbtoycb_mmx:
 	pushl   %esi
 	pushl   %edi
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+#ifdef __PIC__
+	leal    ZEROSX@GOTOFF(%ebp), %eax    #This section gets around a bug
+#else
 	leal    ZEROSX, %eax    #This section gets around a bug
+#endif
 	movq    (%eax), %mm0    #unlikely to persist
+#ifdef __PIC__
+	movq    %mm0, ZEROS@GOTOFF(%ebp)
+	leal    OFFSETDX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, ZEROS
 	leal    OFFSETDX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, OFFSETD@GOTOFF(%ebp)
+	leal    OFFSETWX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, OFFSETD
 	leal    OFFSETWX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, OFFSETW@GOTOFF(%ebp)
+	leal    OFFSETBX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, OFFSETW
 	leal    OFFSETBX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, OFFSETB@GOTOFF(%ebp)
+	leal    YR0GRX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, OFFSETB
 	leal    YR0GRX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, YR0GR@GOTOFF(%ebp)
+	leal    YBG0BX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, YR0GR
 	leal    YBG0BX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, YBG0B@GOTOFF(%ebp)
+	leal    UR0GRX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, YBG0B
 	leal    UR0GRX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, UR0GR@GOTOFF(%ebp)
+	leal    UBG0BX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, UR0GR
 	leal    UBG0BX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, UBG0B@GOTOFF(%ebp)
+	leal    VR0GRX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, UBG0B
 	leal    VR0GRX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, VR0GR@GOTOFF(%ebp)
+	leal    VBG0BX@GOTOFF(%ebp), %eax
+#else
 	movq    %mm0, VR0GR
 	leal    VBG0BX, %eax
+#endif
 	movq    (%eax), %mm0
+#ifdef __PIC__
+	movq    %mm0, VBG0B@GOTOFF(%ebp)
+#else
 	movq    %mm0, VBG0B
-	
-	movl    _rows(%ebp), %eax
-	movl    _columns(%ebp), %ebx
+#endif
+	movl    _rows(%esp), %eax
+	movl    _columns(%esp), %ebx
 	mull    %ebx            #number pixels
 	shrl    $3, %eax        #number of loops
 	movl    %eax, %edi      #loop counter in edi
-	movl    _inPtr(%ebp), %eax
-	movl    _outyPtr(%ebp), %ebx
-	movl    _outuPtr(%ebp), %ecx
-	movl    _outvPtr(%ebp), %edx
+	movl    _inPtr(%esp), %eax
+	movl    _outyPtr(%esp), %ebx
+	movl    _outuPtr(%esp), %ecx
+	movl    _outvPtr(%esp), %edx
 rgbtoycb_mmx_loop: 
 	movq    (%eax), %mm1    #load G2R2B1G1R1B0G0R0
 	pxor    %mm6, %mm6      #0 -> mm6
@@ -184,29 +251,57 @@ rgbtoycb_mmx_loop: 
 	punpcklbw %mm6, %mm1     #B1G1R1B0 -> mm1
 	movq    %mm0, %mm2      #R1B0G0R0 -> mm2
 
+#ifdef __PIC__
+	pmaddwd YR0GR@GOTOFF(%ebp), %mm0     #yrR1,ygG0+yrR0 -> mm0
+#else
 	pmaddwd YR0GR, %mm0     #yrR1,ygG0+yrR0 -> mm0
+#endif
 	movq    %mm1, %mm3      #B1G1R1B0 -> mm3
 
+#ifdef __PIC__
+	pmaddwd YBG0B@GOTOFF(%ebp), %mm1     #ybB1+ygG1,ybB0 -> mm1
+#else
 	pmaddwd YBG0B, %mm1     #ybB1+ygG1,ybB0 -> mm1
+#endif
 	movq    %mm2, %mm4      #R1B0G0R0 -> mm4
 
+#ifdef __PIC__
+	pmaddwd UR0GR@GOTOFF(%ebp), %mm2     #urR1,ugG0+urR0 -> mm2
+#else
 	pmaddwd UR0GR, %mm2     #urR1,ugG0+urR0 -> mm2
+#endif
 	movq    %mm3, %mm5      #B1G1R1B0 -> mm5
 
+#ifdef __PIC__
+	pmaddwd UBG0B@GOTOFF(%ebp), %mm3     #ubB1+ugG1,ubB0 -> mm3
+#else
 	pmaddwd UBG0B, %mm3     #ubB1+ugG1,ubB0 -> mm3
+#endif
 	punpckhbw       %mm6, %mm7 #    00G2R2 -> mm7
 
+#ifdef __PIC__
+	pmaddwd VR0GR@GOTOFF(%ebp), %mm4     #vrR1,vgG0+vrR0 -> mm4
+#else
 	pmaddwd VR0GR, %mm4     #vrR1,vgG0+vrR0 -> mm4
+#endif
 	paddd   %mm1, %mm0      #Y1Y0 -> mm0
 
+#ifdef __PIC__
+	pmaddwd VBG0B@GOTOFF(%ebp), %mm5     #vbB1+vgG1,vbB0 -> mm5
+#else
 	pmaddwd VBG0B, %mm5     #vbB1+vgG1,vbB0 -> mm5
+#endif
 
 	movq    8(%eax), %mm1   #R5B4G4R4B3G3R3B2 -> mm1
 	paddd   %mm3, %mm2      #U1U0 -> mm2
 
 	movq    %mm1, %mm6      #R5B4G4R4B3G3R3B2 -> mm6
 
+#ifdef __PIC__
+	punpcklbw       ZEROS@GOTOFF(%ebp), %mm1     #B3G3R3B2 -> mm1
+#else
 	punpcklbw       ZEROS, %mm1     #B3G3R3B2 -> mm1
+#endif
 	paddd   %mm5, %mm4      #V1V0 -> mm4
 
 	movq    %mm1, %mm5      #B3G3R3B2 -> mm5
@@ -214,29 +309,61 @@ rgbtoycb_mmx_loop: 
 
 	paddd   %mm7, %mm1      #R3B200+00G2R2=R3B2G2R2->mm1
 
+#ifdef __PIC__
+	punpckhbw       ZEROS@GOTOFF(%ebp), %mm6     #R5B4G4R3 -> mm6
+#else
 	punpckhbw       ZEROS, %mm6     #R5B4G4R3 -> mm6
+#endif
 	movq    %mm1, %mm3      #R3B2G2R2 -> mm3
 
+#ifdef __PIC__
+	pmaddwd YR0GR@GOTOFF(%ebp), %mm1     #yrR3,ygG2+yrR2 -> mm1
+#else
 	pmaddwd YR0GR, %mm1     #yrR3,ygG2+yrR2 -> mm1
+#endif
 	movq    %mm5, %mm7      #B3G3R3B2 -> mm7
 
+#ifdef __PIC__
+	pmaddwd YBG0B@GOTOFF(%ebp), %mm5     #ybB3+ygG3,ybB2 -> mm5
+#else
 	pmaddwd YBG0B, %mm5     #ybB3+ygG3,ybB2 -> mm5
+#endif
 	psrad   $FIXPSHIFT, %mm0       #32-bit scaled Y1Y0 -> mm0
 
+#ifdef __PIC__
+	movq    %mm6, TEMP0@GOTOFF(%ebp)     #R5B4G4R4 -> TEMP0
+#else
 	movq    %mm6, TEMP0     #R5B4G4R4 -> TEMP0
+#endif
 	movq    %mm3, %mm6      #R3B2G2R2 -> mm6
+#ifdef __PIC__
+	pmaddwd UR0GR@GOTOFF(%ebp), %mm6     #urR3,ugG2+urR2 -> mm6
+#else
 	pmaddwd UR0GR, %mm6     #urR3,ugG2+urR2 -> mm6
+#endif
 	psrad   $FIXPSHIFT, %mm2       #32-bit scaled U1U0 -> mm2
 
 	paddd   %mm5, %mm1      #Y3Y2 -> mm1
 	movq    %mm7, %mm5      #B3G3R3B2 -> mm5
+#ifdef __PIC__
+	pmaddwd UBG0B@GOTOFF(%ebp), %mm7     #ubB3+ugG3,ubB2
+#else
 	pmaddwd UBG0B, %mm7     #ubB3+ugG3,ubB2
+#endif
 	psrad   $FIXPSHIFT, %mm1 #32-bit scaled Y3Y2 -> mm1
 
+#ifdef __PIC__
+	pmaddwd VR0GR@GOTOFF(%ebp), %mm3     #vrR3,vgG2+vgR2
+#else
 	pmaddwd VR0GR, %mm3     #vrR3,vgG2+vgR2
+#endif
 	packssdw        %mm1, %mm0      #Y3Y2Y1Y0 -> mm0
 
+#ifdef __PIC__
+	pmaddwd VBG0B@GOTOFF(%ebp), %mm5     #vbB3+vgG3,vbB2 -> mm5
+#else
 	pmaddwd VBG0B, %mm5     #vbB3+vgG3,vbB2 -> mm5
+#endif
 	psrad   $FIXPSHIFT, %mm4       #32-bit scaled V1V0 -> mm4
 
 	movq    16(%eax), %mm1  #B7G7R7B6G6R6B5G5 -> mm7
@@ -251,58 +378,114 @@ rgbtoycb_mmx_loop: 
 	movq    %mm7, %mm5      #R7B6G6R6B5G500 -> mm5
 	psrad   $FIXPSHIFT, %mm3       #32-bit scaled V3V2 -> mm3
 
+#ifdef __PIC__
+	paddw	OFFSETY@GOTOFF(%ebp), %mm0
+#else
 	paddw	OFFSETY, %mm0
+#endif
 	movq    %mm0, (%ebx)     #store Y3Y2Y1Y0 
 	packssdw %mm6, %mm2      #32-bit scaled U3U2U1U0 -> mm2
 
+#ifdef __PIC__
+	movq    TEMP0@GOTOFF(%ebp), %mm0     #R5B4G4R4 -> mm0
+#else
 	movq    TEMP0, %mm0     #R5B4G4R4 -> mm0
+#endif
 	addl	$8, %ebx
-	
+
+#ifdef __PIC__
+	punpcklbw       ZEROS@GOTOFF(%ebp), %mm7     #B5G500 -> mm7
+#else
 	punpcklbw       ZEROS, %mm7     #B5G500 -> mm7
+#endif
 	movq    %mm0, %mm6      #R5B4G4R4 -> mm6
 
+#ifdef __PIC__
+	movq    %mm2, TEMPU@GOTOFF(%ebp)     #32-bit scaled U3U2U1U0 -> TEMPU
+#else
 	movq    %mm2, TEMPU     #32-bit scaled U3U2U1U0 -> TEMPU
+#endif
 	psrlq   $32, %mm0       #00R5B4 -> mm0
 
 	paddw   %mm0, %mm7      #B5G5R5B4 -> mm7
 	movq    %mm6, %mm2      #B5B4G4R4 -> mm2
 
+#ifdef __PIC__
+	pmaddwd YR0GR@GOTOFF(%ebp), %mm2     #yrR5,ygG4+yrR4 -> mm2
+#else
 	pmaddwd YR0GR, %mm2     #yrR5,ygG4+yrR4 -> mm2
+#endif
 	movq    %mm7, %mm0      #B5G5R5B4 -> mm0
 
+#ifdef __PIC__
+	pmaddwd YBG0B@GOTOFF(%ebp), %mm7     #ybB5+ygG5,ybB4 -> mm7
+#else
 	pmaddwd YBG0B, %mm7     #ybB5+ygG5,ybB4 -> mm7
+#endif
 	packssdw        %mm3, %mm4      #32-bit scaled V3V2V1V0 -> mm4
 
 	addl    $24, %eax       #increment RGB count
 
+#ifdef __PIC__
+	movq    %mm4, TEMPV@GOTOFF(%ebp)     #(V3V2V1V0)/256 -> mm4
+#else
 	movq    %mm4, TEMPV     #(V3V2V1V0)/256 -> mm4
+#endif
 	movq    %mm6, %mm4      #B5B4G4R4 -> mm4
 
+#ifdef __PIC__
+	pmaddwd UR0GR@GOTOFF(%ebp), %mm6     #urR5,ugG4+urR4
+#else
 	pmaddwd UR0GR, %mm6     #urR5,ugG4+urR4
+#endif
 	movq    %mm0, %mm3      #B5G5R5B4 -> mm0
 
+#ifdef __PIC__
+	pmaddwd UBG0B@GOTOFF(%ebp), %mm0     #ubB5+ugG5,ubB4
+#else
 	pmaddwd UBG0B, %mm0     #ubB5+ugG5,ubB4
+#endif
 	paddd   %mm7, %mm2      #Y5Y4 -> mm2
 
+#ifdef __PIC__
+	pmaddwd         VR0GR@GOTOFF(%ebp), %mm4     #vrR5,vgG4+vrR4 -> mm4
+#else
 	pmaddwd         VR0GR, %mm4     #vrR5,vgG4+vrR4 -> mm4
+#endif
 	pxor    %mm7, %mm7      #0 -> mm7
 
+#ifdef __PIC__
+	pmaddwd VBG0B@GOTOFF(%ebp), %mm3     #vbB5+vgG5,vbB4 -> mm3
+#else
 	pmaddwd VBG0B, %mm3     #vbB5+vgG5,vbB4 -> mm3
+#endif
 	punpckhbw       %mm7, %mm1      #B7G7R7B6 -> mm1
 
 	paddd   %mm6, %mm0      #U5U4 -> mm0
 	movq    %mm1, %mm6      #B7G7R7B6 -> mm6
 
+#ifdef __PIC__
+	pmaddwd YBG0B@GOTOFF(%ebp), %mm6     #ybB7+ygG7,ybB6 -> mm6
+#else
 	pmaddwd YBG0B, %mm6     #ybB7+ygG7,ybB6 -> mm6
+#endif
 	punpckhbw       %mm7, %mm5      #R7B6G6R6 -> mm5
 
 	movq    %mm5, %mm7      #R7B6G6R6 -> mm7
 	paddd   %mm4, %mm3      #V5V4 -> mm3
 
+#ifdef __PIC__
+	pmaddwd YR0GR@GOTOFF(%ebp), %mm5     #yrR7,ygG6+yrR6 -> mm5
+#else
 	pmaddwd YR0GR, %mm5     #yrR7,ygG6+yrR6 -> mm5
+#endif
 	movq    %mm1, %mm4      #B7G7R7B6 -> mm4
 
+#ifdef __PIC__
+	pmaddwd UBG0B@GOTOFF(%ebp), %mm4     #ubB7+ugG7,ubB6 -> mm4
+#else
 	pmaddwd UBG0B, %mm4     #ubB7+ugG7,ubB6 -> mm4
+#endif
 	psrad   $FIXPSHIFT, %mm0       #32-bit scaled U5U4 -> mm0
 
 	psrad   $FIXPSHIFT, %mm2       #32-bit scaled Y5Y4 -> mm2
@@ -310,25 +493,49 @@ rgbtoycb_mmx_loop: 
 	paddd   %mm5, %mm6      #Y7Y6 -> mm6
 	movq    %mm7, %mm5      #R7B6G6R6 -> mm5
 
+#ifdef __PIC__
+	pmaddwd UR0GR@GOTOFF(%ebp), %mm7     #urR7,ugG6+ugR6 -> mm7
+#else
 	pmaddwd UR0GR, %mm7     #urR7,ugG6+ugR6 -> mm7
+#endif
 	psrad   $FIXPSHIFT, %mm3       #32-bit scaled V5V4 -> mm3
 
+#ifdef __PIC__
+	pmaddwd VBG0B@GOTOFF(%ebp), %mm1     #vbB7+vgG7,vbB6 -> mm1
+#else
 	pmaddwd VBG0B, %mm1     #vbB7+vgG7,vbB6 -> mm1
+#endif
 	psrad   $FIXPSHIFT, %mm6 #32-bit scaled Y7Y6 -> mm6
 
 	packssdw %mm6, %mm2     #Y7Y6Y5Y4 -> mm2
 
+#ifdef __PIC__
+	pmaddwd VR0GR@GOTOFF(%ebp), %mm5     #vrR7,vgG6+vrR6 -> mm5
+#else
 	pmaddwd VR0GR, %mm5     #vrR7,vgG6+vrR6 -> mm5
+#endif
 	paddd   %mm4, %mm7      #U7U6 -> mm7    
 
 	psrad   $FIXPSHIFT, %mm7       #32-bit scaled U7U6 -> mm7
+#ifdef __PIC__
+	paddw	OFFSETY@GOTOFF(%ebp), %mm2
+#else
 	paddw	OFFSETY, %mm2
+#endif
 	movq	%mm2, (%ebx)    #store Y7Y6Y5Y4 
 
+#ifdef __PIC__
+	movq	ALLONE@GOTOFF(%ebp), %mm6
+#else
 	movq	ALLONE, %mm6
+#endif
 	packssdw %mm7, %mm0     #32-bit scaled U7U6U5U4 -> mm0
 
+#ifdef __PIC__
+	movq    TEMPU@GOTOFF(%ebp), %mm4     #32-bit scaled U3U2U1U0 -> mm4
+#else
 	movq    TEMPU, %mm4     #32-bit scaled U3U2U1U0 -> mm4
+#endif
 	pmaddwd	%mm6, %mm0      #U7U6U5U4 averaged -> (U7U6)(U5U4)=UU3 UU2->mm0
 	
 	pmaddwd	%mm6, %mm4      #U3U2U1U0 averaged -> (U3U2)(U1U0)=UU1 UU0->mm4
@@ -338,8 +545,12 @@ rgbtoycb_mmx_loop: 
 
 	psrad   $FIXPSHIFT, %mm1       #32-bit scaled V7V6 -> mm1
 	psraw	$1, %mm4 	#divide UU3 UU2 UU1 UU0 by 2 -> mm4
-		
+
+#ifdef __PIC__
+	movq    TEMPV@GOTOFF(%ebp), %mm5     #32-bit scaled V3V2V1V0 -> mm5
+#else
 	movq    TEMPV, %mm5     #32-bit scaled V3V2V1V0 -> mm5
+#endif
 
 	movq	%mm4, (%ecx)    # store U	
 
@@ -425,14 +636,22 @@ _dv_ppm_copy_y_block_mmx:
 _dv_pgm_copy_y_block_mmx:
 
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl	%edi
-	
-	movl    8(%ebp), %edi          # dest
-	movl    12(%ebp), %esi         # src
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    16(%esp), %edi          # dest
+	movl    20(%esp), %esi         # src
+
+#ifdef __PIC__
+	movq	OFFSETY@GOTOFF(%ebp), %mm7
+#else
 	movq	OFFSETY, %mm7
+#endif
 	pxor	%mm6, %mm6
 	
 	movq	(%esi), %mm0
@@ -567,14 +786,22 @@ _dv_pgm_copy_y_block_mmx:
 _dv_video_copy_y_block_mmx:
 
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl	%edi
-	
-	movl    8(%ebp), %edi          # dest
-	movl    12(%ebp), %esi         # src
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    16(%esp), %edi          # dest
+	movl    20(%esp), %esi         # src
+
+#ifdef __PIC__
+	movq	OFFSETBX@GOTOFF(%ebp), %mm7
+#else
 	movq	OFFSETBX, %mm7
+#endif
 	pxor	%mm6, %mm6
 	
 	movq	(%esi), %mm0
@@ -855,16 +1082,23 @@ _dv_ppm_copy_pal_c_block_mmx:
 _dv_pgm_copy_pal_c_block_mmx:
 				
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl	%edi
 	pushl	%ebx
-	
-	movl    8(%ebp), %edi          # dest
-	movl    12(%ebp), %esi         # src
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    20(%esp), %edi          # dest
+	movl    24(%esp), %esi         # src
 
+#ifdef __PIC__
+	movq	OFFSETBX@GOTOFF(%ebp), %mm7
+#else
 	movq	OFFSETBX, %mm7
+#endif
 	pxor	%mm6, %mm6
 
 	
@@ -1003,15 +1237,23 @@ _dv_pgm_copy_pal_c_block_mmx:
 _dv_video_copy_pal_c_block_mmx:
 				
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl	%edi
 	pushl	%ebx
-	
-	movl    8(%ebp), %edi          # dest
-	movl    12(%ebp), %esi         # src
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    20(%esp), %edi          # dest
+	movl    24(%esp), %esi         # src
+
+#ifdef __PIC__
+	movq	OFFSETBX@GOTOFF(%ebp), %mm7
+#else
 	movq	OFFSETBX, %mm7
+#endif
 	paddw	%mm7, %mm7
 	pxor	%mm6, %mm6
 
@@ -1098,18 +1340,25 @@ video_copy_pal_c_block_mmx_loop:	
 _dv_ppm_copy_ntsc_c_block_mmx:
 				
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl	%edi
 	pushl	%ebx
-	
-	movl    8(%ebp), %edi          # dest
-	movl    12(%ebp), %esi         # src
+
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    20(%esp), %edi          # dest
+	movl    24(%esp), %esi         # src
 
 	movl	$4, %ebx	
 
+#ifdef __PIC__
+	movq	ALLONE@GOTOFF(%ebp), %mm6
+#else
 	movq	ALLONE, %mm6
-	
+#endif
 ppm_copy_ntsc_c_block_mmx_loop:	
 	
 	movq	(%esi), %mm0
@@ -1171,14 +1420,22 @@ ppm_copy_ntsc_c_block_mmx_loop:	
 _dv_pgm_copy_ntsc_c_block_mmx:
 				
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl	%edi
-	
-	movl    8(%ebp), %edi          # dest
-	movl    12(%ebp), %esi         # src
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    16(%esp), %edi          # dest
+	movl    20(%esp), %esi         # src
+
+#ifdef __PIC__
+	movq	OFFSETBX@GOTOFF(%ebp), %mm7
+#else
 	movq	OFFSETBX, %mm7
+#endif
 	paddw	%mm7, %mm7
 	pxor	%mm6, %mm6
 
@@ -1328,15 +1585,23 @@ _dv_pgm_copy_ntsc_c_block_mmx:
 _dv_video_copy_ntsc_c_block_mmx:
 				
 	pushl   %ebp
-	movl    %esp, %ebp
 	pushl   %esi
 	pushl	%edi
 	pushl	%ebx
-	
-	movl    8(%ebp), %edi          # dest
-	movl    12(%ebp), %esi         # src
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
+
+	movl    20(%esp), %edi          # dest
+	movl    24(%esp), %esi         # src
+
+#ifdef __PIC__
+	movq	OFFSETBX@GOTOFF(%ebp), %mm7
+#else
 	movq	OFFSETBX, %mm7
+#endif
 	paddw	%mm7, %mm7
 	pxor	%mm6, %mm6
 
diff -urp libdv-0.104-old/libdv/vlc_x86.S libdv-0.104/libdv/vlc_x86.S
--- libdv-0.104-old/libdv/vlc_x86.S	2005-10-23 19:40:58.000000000 +0200
+++ libdv-0.104/libdv/vlc_x86.S	2005-10-25 01:47:14.000000000 +0200
@@ -1,29 +1,76 @@
 	#include "asmoff.h"
 .text
+
+#ifdef __PIC__
+# undef __i686 /* gcc define gets in our way */
+	.section .gnu.linkonce.t.__i686.get_pc_thunk.bp,"ax",@progbits
+.globl __i686.get_pc_thunk.bp
+	.hidden  __i686.get_pc_thunk.bp
+	.type    __i686.get_pc_thunk.bp,@function
+__i686.get_pc_thunk.bp:
+	movl (%esp), %ebp
+	ret
+
+	.section .gnu.linkonce.t.__i686.get_pc_thunk.si,"ax",@progbits
+.globl __i686.get_pc_thunk.si
+	.hidden  __i686.get_pc_thunk.si
+	.type    __i686.get_pc_thunk.si,@function
+__i686.get_pc_thunk.si:
+	movl (%esp), %esi
+	ret
+#endif
+
 	.align 4
 .globl dv_decode_vlc 
 	.type	 dv_decode_vlc,@function
 dv_decode_vlc:
 	pushl %ebx
+	pushl %ebp
+
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
 
-	/* Args are at 8(%esp). */
-	movl  8(%esp),%eax		/* %eax is bits */
-	movl  12(%esp),%ebx		/* %ebx is maxbits */
+	/* Args are at 12(%esp). */
+	movl  12(%esp),%eax		/* %eax is bits */
+	movl  16(%esp),%ebx		/* %ebx is maxbits */
 	andl  $0x3f,%ebx		/* limit index range STL*/
 
+#ifdef __ELF__
+	movl  dv_vlc_class_index_mask@GOTOFF(%ebp,%ebx,4),%edx
+#else
 	movl  dv_vlc_class_index_mask(,%ebx,4),%edx
+#endif
 	andl  %eax,%edx
+#ifdef __ELF__
+	movl  dv_vlc_class_index_rshift@GOTOFF(%ebp,%ebx,4),%ecx
+#else
 	movl  dv_vlc_class_index_rshift(,%ebx,4),%ecx
+#endif
 	sarl  %cl,%edx
+#ifdef __ELF__
+	movl  dv_vlc_classes@GOTOFF(%ebp,%ebx,4),%ecx
+#else
 	movl  dv_vlc_classes(,%ebx,4),%ecx
+#endif
 	movsbl  (%ecx,%edx,1),%edx	/* %edx is class */
-			
+
+#ifdef __ELF__
+	movl  dv_vlc_index_mask@GOTOFF(%ebp,%edx,4),%ebx
+	movl  dv_vlc_index_rshift@GOTOFF(%ebp,%edx,4),%ecx
+#else
 	movl  dv_vlc_index_mask(,%edx,4),%ebx
 	movl  dv_vlc_index_rshift(,%edx,4),%ecx
+#endif
 	andl  %eax,%ebx
 	sarl  %cl,%ebx
 
+#ifdef __ELF__
+	movl  dv_vlc_lookups@GOTOFF(%ebp,%edx,4),%edx
+#else
 	movl  dv_vlc_lookups(,%edx,4),%edx
+#endif
 	movl  (%edx,%ebx,4),%edx
 
 	/* Now %edx holds result, like this:
@@ -42,7 +89,11 @@ dv_decode_vlc:
 	movl  %edx,%ecx
 	sarl  $8,%ecx
 	andl  $0xff,%ecx
+#ifdef __ELF__
+	movl  sign_mask@GOTOFF(%ebp,%ecx,4),%ebx
+#else
 	movl  sign_mask(,%ecx,4),%ebx
+#endif
 	andl  %ebx,%eax
 	negl  %eax
 	sarl  $31,%eax
@@ -63,14 +114,14 @@ dv_decode_vlc:
 	    *result = broken;
 	Note that the 'broken' pattern is all ones (i.e. 0xffffffff)
 	*/
-	movl  12(%esp),%ebx		/* %ebx is maxbits */
+	movl  16(%esp),%ebx		/* %ebx is maxbits */
 	subl  %ecx,%ebx
 	sbbl  %ebx,%ebx
 	orl   %ebx,%edx
 
-	movl  16(%esp),%eax
+	movl  20(%esp),%eax
 	movl  %edx,(%eax)
-	
+	popl  %ebp
 	popl  %ebx
 	ret
 	
@@ -80,21 +131,38 @@ dv_decode_vlc:
 	.type	 __dv_decode_vlc,@function
 __dv_decode_vlc:
 	pushl %ebx
+	pushl %ebp
+
+#ifdef __PIC__
+	call __i686.get_pc_thunk.bp
+	addl $_GLOBAL_OFFSET_TABLE_, %ebp
+#endif
 
-	/* Args are at 8(%esp). */
-	movl  8(%esp),%eax		/* %eax is bits */
+	/* Args are at 12(%esp). */
+	movl  12(%esp),%eax		/* %eax is bits */
 	
 	movl  %eax,%edx			/* %edx is class */
 	andl  $0xfe00,%edx
 	sarl  $9,%edx
+#ifdef __PIC__
+	movsbl dv_vlc_class_lookup5@GOTOFF(%ebp,%edx),%edx
+
+	movl  dv_vlc_index_mask@GOTOFF(%ebp,%edx,4),%ebx
+	movl  dv_vlc_index_rshift@GOTOFF(%ebp,%edx,4),%ecx
+#else
 	movsbl dv_vlc_class_lookup5(%edx),%edx
-	
+
 	movl  dv_vlc_index_mask(,%edx,4),%ebx
 	movl  dv_vlc_index_rshift(,%edx,4),%ecx
+#endif
 	andl  %eax,%ebx
 	sarl  %cl,%ebx
 
+#ifdef __PIC__
+	movl  dv_vlc_lookups@GOTOFF(%ebp,%edx,4),%edx
+#else
 	movl  dv_vlc_lookups(,%edx,4),%edx
+#endif
 	movl  (%edx,%ebx,4),%edx
 
 	/* Now %edx holds result, like this:
@@ -112,7 +180,11 @@ __dv_decode_vlc:
 	movl  %edx,%ecx
 	sarl  $8,%ecx
 	andl  $0xff,%ecx
+#ifdef __PIC__
+	movl  sign_mask@GOTOFF(%ebp,%ecx,4),%ecx
+#else
 	movl  sign_mask(,%ecx,4),%ecx
+#endif
 	andl  %ecx,%eax
 	negl  %eax
 	sarl  $31,%eax
@@ -127,9 +199,9 @@ __dv_decode_vlc:
 	xorl  %eax,%edx
 	subl  %eax,%edx
 
-	movl  12(%esp),%eax
+	movl  16(%esp),%eax
 	movl  %edx,(%eax)
-	
+	popl  %ebp
 	popl  %ebx
 	ret
 
@@ -147,6 +219,11 @@ dv_parse_ac_coeffs_pass0:
 	pushl	%esi
 	pushl	%ebp
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.si
+	addl $_GLOBAL_OFFSET_TABLE_, %esi
+#endif
+
 #define ARGn(N)  (20+(4*(N)))(%esp)
 
 	/*
@@ -159,8 +236,10 @@ dv_parse_ac_coeffs_pass0:
 	ebp	bl
 	*/
 	movl    ARGn(2),%ebp
+#ifndef __PIC__
 	movl	ARGn(0),%esi
 	movl	bitstream_t_buf(%esi),%esi
+#endif
 	movl	dv_block_t_offset(%ebp),%edi
 	movl	dv_block_t_reorder(%ebp),%ebx
 
@@ -170,7 +249,11 @@ dv_parse_ac_coeffs_pass0:
 	
 	movq    dv_block_t_coeffs(%ebp),%mm1
 	pxor    %mm0,%mm0
+#ifdef __PIC__
+	pand    const_f_0_0_0@GOTOFF(%esi),%mm1
+#else
 	pand    const_f_0_0_0,%mm1
+#endif
 	movq    %mm1,dv_block_t_coeffs(%ebp)
 	movq    %mm0,(dv_block_t_coeffs + 8)(%ebp)
 	movq    %mm0,(dv_block_t_coeffs + 16)(%ebp)
@@ -191,9 +274,17 @@ dv_parse_ac_coeffs_pass0:
 readloop:
 	movl	%edi,%ecx
 	shrl	$3,%ecx
+#ifdef __PIC__
+	pushl	%esi
+	movl	ARGn(1),%esi
+	movl	bitstream_t_buf(%esi),%esi
+#endif
 	movzbl  (%esi,%ecx,1),%eax
 	movzbl  1(%esi,%ecx,1),%edx
 	movzbl  2(%esi,%ecx,1),%ecx
+#ifdef __PIC__
+	popl	%esi
+#endif
 	shll	$16,%eax
 	shll	$8,%edx
 	orl	%ecx,%eax
@@ -217,7 +308,11 @@ readloop:
 
 	/* Attempt to use the shortcut first.  If it hits, then
 	   this vlc term has been decoded. */
+#ifdef __PIC__
+	movl	dv_vlc_class1_shortcut@GOTOFF(%esi,%ecx,4),%edx
+#else
 	movl	dv_vlc_class1_shortcut(,%ecx,4),%edx
+#endif
 	test	$0x80,%edx
 	je	done_decode
 
@@ -228,12 +323,19 @@ readloop:
 	movl	%ebx,dv_block_t_reorder(%ebp)
 
 	/* %eax is bits */
-	
+#ifdef __PIC__
+	movsbl dv_vlc_class_lookup5@GOTOFF(%esi,%ecx),%ecx
+
+	movl  dv_vlc_index_mask@GOTOFF(%esi,%ecx,4),%ebx
+	movl  dv_vlc_lookups@GOTOFF(%esi,%ecx,4),%edx
+	movl  dv_vlc_index_rshift@GOTOFF(%esi,%ecx,4),%ecx
+#else
 	movsbl dv_vlc_class_lookup5(%ecx),%ecx
 
 	movl  dv_vlc_index_mask(,%ecx,4),%ebx
 	movl  dv_vlc_lookups(,%ecx,4),%edx
 	movl  dv_vlc_index_rshift(,%ecx,4),%ecx
+#endif
 	andl  %eax,%ebx
 	sarl  %cl,%ebx
 
@@ -256,7 +358,11 @@ readloop:
 	movl  %edx,%ecx
 	sarl  $8,%ecx
 	andl  $0xff,%ecx
+#ifdef __PIC__
+	movl  sign_mask@GOTOFF(%esi,%ecx,4),%ecx
+#else
 	movl  sign_mask(,%ecx,4),%ecx
+#endif
 	andl  %ecx,%eax
 	negl  %eax
 	sarl  $31,%eax
@@ -326,10 +432,20 @@ alldone:
 
 slowpath:
 	/* slow path:	 use dv_decode_vlc */;
+#ifdef __PIC__
+	pushl	%esi
+	leal	vlc@GOTOFF(%esi),%esi
+	xchgl	%esi,(%esp)	/* last parameter is &vlc */
+#else
 	pushl	$vlc		/* last parameter is &vlc */
+#endif
 	pushl	%edx		/* bits_left */
 	pushl	%eax		/* bits */
+#ifdef __PIC__
+	call	dv_decode_vlc@PLT
+#else
 	call	dv_decode_vlc
+#endif
 	addl	$12,%esp
 	test	$0x80,%edx	/* If (vlc.run < 0) break */
 	jne	escape
@@ -365,6 +481,11 @@ dv_parse_video_segment:
 	pushl	%esi
 	pushl	%ebp
 
+#ifdef __PIC__
+	call __i686.get_pc_thunk.si
+	addl $_GLOBAL_OFFSET_TABLE_, %esi
+#endif
+
 #define ARGn(N)  (20+(4*(N)))(%esp)
 
 	movl	ARGn(1),%eax			/* quality */
@@ -373,7 +494,11 @@ dv_parse_video_segment:
 	jz	its_mono
 	movl	$6,%ebx
 its_mono:
+#ifdef __PIC__
+	movl	%ebx,n_blocks@GOTOFF(%esi)
+#else
 	movl	%ebx,n_blocks
+#endif
 	
 	/*
 	 *	ebx	seg/b
@@ -384,15 +509,22 @@ its_mono:
 	 *	ebp	bl
 	 */
 	movl	ARGn(0),%ebx
+#ifndef __PIC__
 	movl	dv_videosegment_t_bs(%ebx),%esi
 	movl	bitstream_t_buf(%esi),%esi
+#endif
 	leal	dv_videosegment_t_mb(%ebx),%edi
 
 	movl	$0,%eax
 	movl	$0,%ecx
 macloop:
+#ifdef __PIC__
+	movl	%eax,m@GOTOFF(%esi)
+	movl	%ecx,mb_start@GOTOFF(%esi)
+#else
 	movl	%eax,m
 	movl	%ecx,mb_start
+#endif
 
 	movl	ARGn(0),%ebx
 	
@@ -400,7 +532,15 @@ macloop:
 	/* mb->qno = bitstream_get(bs,4); */
 	movl	%ecx,%edx
 	shr	$3,%edx
+#ifdef __PIC__
+	pushl	%esi
+	movl	dv_videosegment_t_bs(%ebx),%esi
+	movl	bitstream_t_buf(%esi),%esi
+#endif
 	movzbl	3(%esi,%edx,1),%edx
+#ifdef __PIC__
+	popl	%esi
+#endif
 	andl	$0xf,%edx
 	movl	%edx,dv_macroblock_t_qno(%edi)
 
@@ -411,7 +551,11 @@ macloop:
 	movl	%edx,dv_macroblock_t_eob_count(%edi)
 
 	/* mb->i = (seg->i + dv_super_map_vertical[m]) % (seg->isPAL?12:10); */
+#ifdef __PIC__
+	movl	dv_super_map_vertical@GOTOFF(%esi,%eax,4),%edx
+#else
 	movl	dv_super_map_vertical(,%eax,4),%edx
+#endif
 	movl	dv_videosegment_t_i(%ebx),%ecx
 	addl	%ecx,%edx
 
@@ -422,11 +566,20 @@ skarly:	
 	andl	$1,%ecx
 	shll	$5,%ecx		/* ecx = (isPAL ? 32 : 0) */
 
+#ifdef __PIC__
+	leal	mod_10@GOTOFF(%esi,%edx),%edx
+	movzbl	(%edx,%ecx,1),%edx	/* uses mod_12 for PAL */
+#else
 	movzbl	mod_10(%edx,%ecx,1),%edx	/* uses mod_12 for PAL */
+#endif
 	movl	%edx,dv_macroblock_t_i(%edi)
 
 	/*  mb->j = dv_super_map_horizontal[m]; */	
+#ifdef __PIC__
+	movl	dv_super_map_horizontal@GOTOFF(%esi,%eax,4),%edx
+#else
 	movl	dv_super_map_horizontal(,%eax,4),%edx
+#endif
 	movl	%edx,dv_macroblock_t_j(%edi)
 
 	/* mb->k = seg->k; */
@@ -445,12 +598,29 @@ blkloop:
 	        +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
 	*/
 	/* dc = bitstream_get(bs,9); */
+#ifdef __PIC__
+	movl	mb_start@GOTOFF(%esi),%ecx
+#else
 	movl	mb_start,%ecx
+#endif
 	shr	$3,%ecx
+#ifdef __PIC__
+	movzbl	blk_start@GOTOFF(%esi,%ebx),%edx
+#else
 	movzbl	blk_start(%ebx),%edx
+#endif
 	addl	%ecx,%edx
+#ifdef __PIC__
+	pushl	%esi
+	movl	ARGn(1),%esi
+	movl	dv_videosegment_t_bs(%esi),%esi
+	movl	bitstream_t_buf(%esi),%esi
+#endif
 	movzbl	(%esi,%edx,1),%eax	/* hi byte */
 	movzbl	1(%esi,%edx,1),%ecx	/* lo byte */
+#ifdef __PIC__
+	popl	%esi
+#endif
 	shll	$8,%eax
 	orl	%ecx,%eax
 
@@ -477,7 +647,11 @@ blkloop:
 
 	/* bl->reorder = &dv_reorder[bl->dct_mode][1]; */
 	shll	$6,%eax
+#ifdef __PIC__
+	addl	dv_reorder@GOTOFF+1(%esi),%eax
+#else
 	addl	$(dv_reorder+1),%eax
+#endif
 	movl	%eax,dv_block_t_reorder(%ebp)
 
 	/* bl->reorder_sentinel = bl->reorder + 63; */
@@ -485,13 +659,22 @@ blkloop:
 	movl	%eax,dv_block_t_reorder_sentinel(%ebp)
 
 	/* bl->offset= mb_start + dv_parse_bit_start[b]; */
+#ifdef __PIC__
+	movl	mb_start@GOTOFF(%esi),%ecx
+	movl	dv_parse_bit_start@GOTOFF(%esi,%ebx,4),%eax
+#else
 	movl	mb_start,%ecx
 	movl	dv_parse_bit_start(,%ebx,4),%eax
+#endif
 	addl	%ecx,%eax
 	movl	%eax,dv_block_t_offset(%ebp)
 
 	/* bl->end= mb_start + dv_parse_bit_end[b]; */
+#ifdef __PIC__
+	movl	dv_parse_bit_end@GOTOFF(%esi,%ebx,4),%eax
+#else
 	movl	dv_parse_bit_end(,%ebx,4),%eax
+#endif
 	addl	%ecx,%eax
 	movl	%eax,dv_block_t_end(%ebp)
 
@@ -503,7 +686,11 @@ blkloop:
 	/* no AC pass.  Just zero out the remaining coeffs */
 	movq    dv_block_t_coeffs(%ebp),%mm1
 	pxor    %mm0,%mm0
+#ifdef __PIC__
+	pand    const_f_0_0_0@GOTOFF(%esi),%mm1
+#else
 	pand    const_f_0_0_0,%mm1
+#endif
 	movq    %mm1,dv_block_t_coeffs(%ebp)
 	movq    %mm0,(dv_block_t_coeffs + 8)(%ebp)
 	movq    %mm0,(dv_block_t_coeffs + 16)(%ebp)
@@ -528,18 +715,31 @@ do_ac_pass:
 	pushl	%ebp
 	pushl	%edi
 	pushl	%eax
+#ifdef __PIC__
+	call	dv_parse_ac_coeffs_pass0@PLT
+#else
 	call	dv_parse_ac_coeffs_pass0
+#endif
 	addl	$12,%esp
 done_ac:
 
+#ifdef __PIC__
+	movl	n_blocks@GOTOFF(%esi),%eax
+#else
 	movl	n_blocks,%eax
+#endif
 	addl	$dv_block_t_size,%ebp
 	incl	%ebx
 	cmpl	%eax,%ebx
 	jnz	blkloop
 
+#ifdef __PIC__
+	movl	m@GOTOFF(%esi),%eax
+	movl	mb_start@GOTOFF(%esi),%ecx
+#else
 	movl	m,%eax
 	movl	mb_start,%ecx
+#endif
 	addl	$(8 * 80),%ecx
 	addl	$dv_macroblock_t_size,%edi
 	incl	%eax
@@ -557,7 +757,11 @@ done_ac:
 
 	andl	$DV_QUALITY_AC_MASK,%eax
 	cmpl	$DV_QUALITY_AC_2,%eax
+#ifdef __PIC__
+	jz	dv_parse_ac_coeffs@PLT
+#else
 	jz	dv_parse_ac_coeffs
+#endif
 	movl	$0,%eax
 	ret