Skip to content

GitLab

Commit f22f4ef5 authored by Sam Hocevar's avatar Sam Hocevar
Browse files
Options

* fixed several format string inconsistencies and deprecated C constructions.

parent f9e12228
Hide whitespace changes
Inline Side-by-side
Showing with 577 additions and 572 deletions
modules/access/ftp.c
View file @ f22f4ef5
......@@ -2,7 +2,7 @@
* ftp.c:
*****************************************************************************
* Copyright (C) 2001, 2002 VideoLAN
* $Id: ftp.c,v 1.2 2002/12/16 15:17:12 fenrir Exp $
* $Id: ftp.c,v 1.3 2002/12/18 14:17:09 sam Exp $
*
* Authors: Laurent Aimar <fenrir@via.ecp.fr>
*
......@@ -350,7 +350,7 @@ static int Open( vlc_object_t *p_this )
}
#endif
msg_Dbg( p_input, "file size:%d", p_access->i_filesize );
msg_Dbg( p_input, "file size: "I64Fd, p_access->i_filesize );
FREE( psz_arg );
if( ftp_StartStream( p_input, 0 ) < 0 )
......
modules/access/mms/mms.c
View file @ f22f4ef5
......@@ -2,7 +2,7 @@
* mms.c: MMS access plug-in
*****************************************************************************
* Copyright (C) 2001, 2002 VideoLAN
* $Id: mms.c,v 1.13 2002/12/12 15:10:58 gbazin Exp $
* $Id: mms.c,v 1.14 2002/12/18 14:17:09 sam Exp $
*
* Authors: Laurent Aimar <fenrir@via.ecp.fr>
*
......@@ -1559,11 +1559,15 @@ static int mms_ParseCommand( input_thread_t *p_input,
GET32( 0 ),
GET32( 4 ),
GET32( 8 ),
/* 12: protocol type "MMS " */
GET32( 16 ),
GET32( 20 ),
/* 24: unknown (0) */
/* 28: unknown (0) */
GET32( 32 ),
GET32( 36 ),
GET32( 40 ) );
GET32( 36 )
/* 40: switches */
/* 44: extra */ );
p_access->i_command = GET32( 36 ) & 0xffff;
......
modules/audio_output/oss.c
View file @ f22f4ef5
......@@ -2,7 +2,7 @@
* oss.c : OSS /dev/dsp module for vlc
*****************************************************************************
* Copyright (C) 2000-2002 VideoLAN
* $Id: oss.c,v 1.38 2002/12/10 18:22:01 gbazin Exp $
* $Id: oss.c,v 1.39 2002/12/18 14:17:09 sam Exp $
*
* Authors: Michel Kaempf <maxx@via.ecp.fr>
* Samuel Hocevar <sam@zoy.org>
......@@ -376,7 +376,7 @@ static int Open( vlc_object_t *p_this )
if( ioctl( p_sys->i_fd, SNDCTL_DSP_CHANNELS, &i_nb_channels ) < 0 ||
i_nb_channels != aout_FormatNbChannels( &p_aout->output.output ) )
{
msg_Err( p_aout, "cannot set number of audio channels (%x)",
msg_Err( p_aout, "cannot set number of audio channels (%s)",
aout_FormatPrintChannels( &p_aout->output.output) );
close( p_sys->i_fd );
free( p_sys );
......
modules/codec/adpcm.c
View file @ f22f4ef5
......@@ -2,7 +2,7 @@
* adpcm.c : adpcm variant audio decoder
*****************************************************************************
* Copyright (C) 2001, 2002 VideoLAN
* $Id: adpcm.c,v 1.1 2002/12/03 17:00:16 fenrir Exp $
* $Id: adpcm.c,v 1.2 2002/12/18 14:17:10 sam Exp $
*
* Authors: Laurent Aimar <fenrir@via.ecp.fr>
*
......@@ -378,6 +378,7 @@ static void DecodeThread( adec_thread_t *p_adec )
DecodeAdpcmMs( p_adec, p_aout_buffer );
break;
default:
break;
}
......
modules/codec/ffmpeg/audio.c
View file @ f22f4ef5
......@@ -2,7 +2,7 @@
* audio.c: audio decoder using ffmpeg library
*****************************************************************************
* Copyright (C) 1999-2001 VideoLAN
* $Id: audio.c,v 1.7 2002/12/06 14:22:55 fenrir Exp $
* $Id: audio.c,v 1.8 2002/12/18 14:17:10 sam Exp $
*
* Authors: Laurent Aimar <fenrir@via.ecp.fr>
*
......@@ -238,7 +238,7 @@ void E_( DecodeThread_Audio )( adec_thread_t *p_adec )
p_adec->p_context->channels > 5 )
{
msg_Warn( p_adec->p_fifo,
"invalid channels count",
"invalid channels count %d",
p_adec->p_context->channels );
}
......
modules/codec/ffmpeg/postprocessing/postprocessing_mmx.c
View file @ f22f4ef5
......@@ -2,15 +2,15 @@
* postprocessing_mmx.c: Post Processing library in MMX
*****************************************************************************
* Copyright (C) 2001 VideoLAN
* $Id: postprocessing_mmx.c,v 1.2 2002/12/06 16:34:05 sam Exp $
* $Id: postprocessing_mmx.c,v 1.3 2002/12/18 14:17:10 sam Exp $
*
* Authors: Laurent Aimar <fenrir@via.ecp.fr>
*
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
......@@ -21,7 +21,7 @@
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA.
*****************************************************************************/
#include <vlc/vlc.h> /* only use u8, u32 .... */
#include <vlc/vlc.h> /* only use uint8_t, uint32_t .... */
#include "postprocessing.h"
#include "postprocessing_common.h"
......@@ -42,8 +42,8 @@
/* Use same things as in idct but how it work ? */
#define UNUSED_LONGLONG( foo ) \
static const unsigned long long foo __asm__ (#foo) __attribute__((unused))
/* to calculate isDC_mode for mmx */
/* to calculate isDC_mode for mmx */
UNUSED_LONGLONG( mmx_127_thr1 ) = ( ( 127ULL - PP_THR1 ) << 56 )|
( ( 127ULL - PP_THR1 ) << 48 )|
( ( 127ULL - PP_THR1 ) << 40 )|
......@@ -67,38 +67,38 @@ UNUSED_LONGLONG( mmx_m2_5_m5_2 ) = 0xfffe0005fffb0002ULL;
#if 0
/* find min bytes from r ans set it in r, t is destroyed */
#define MMXEXT_GET_PMIN( r, t ) \
"movq " #r ", " #t " \n\
psrlq $8, " #t " \n\
pminub " #t ", " #r " \n\
pshufw $0xf5, " #r ", " #t " #instead of shift with tmp reg \n\
pminub " #t ", " #r " \n\
pshufw $0xfe, " #r ", " #t " \n\
pminub " #t ", " #r " \n"
"movq " #r ", " #t " \n" \
"psrlq $8, " #t " \n" \
"pminub " #t ", " #r " \n" \
"pshufw $0xf5, " #r ", " #t " #instead of shift with tmp reg \n" \
"pminub " #t ", " #r " \n" \
"pshufw $0xfe, " #r ", " #t " \n" \
"pminub " #t ", " #r " \n"
/* find mzx bytes from r ans set it in r, t is destroyed */
#define MMXEXT_GET_PMAX( r, t ) \
"movq " #r ", " #t " \n\
psrlq $8, " #t " \n\
pmaxub " #t ", " #r " \n\
pshufw $0xf5, " #r ", " #t " \n\
pmaxub " #t ", " #r " \n\
pshufw $0xfe, " #r ", " #t " \n\
pmaxub " #t ", " #r " \n"
"movq " #r ", " #t " \n" \
"psrlq $8, " #t " \n" \
"pmaxub " #t ", " #r " \n" \
"pshufw $0xf5, " #r ", " #t " \n" \
"pmaxub " #t ", " #r " \n" \
"pshufw $0xfe, " #r ", " #t " \n" \
"pmaxub " #t ", " #r " \n"
#define MMXEXT_GET_LMINMAX( s, m, M, t ) \
"movq " #s ", " #t " \n\
pminub " #t ", " #m " \n\
pmaxub " #t ", " #M " \n"
"movq " #s ", " #t " \n" \
"pminub " #t ", " #m " \n" \
"pmaxub " #t ", " #M " \n"
/* Some tips for MMX
* |a-b| :
d1 = a - b with unsigned saturate
d2 = b - a with ...
|a-b| = d1 | d2
d2 = b - a with ...
|a-b| = d1 | d2
*/
......@@ -112,45 +112,45 @@ UNUSED_LONGLONG( mmx_m2_5_m5_2 ) = 0xfffe0005fffb0002ULL;
* so need to be fast ...
*
****************************************************************************/
static inline int pp_deblock_isDC_mode( u8 *p_v )
static inline int pp_deblock_isDC_mode( uint8_t *p_v )
{
unsigned int i_eq_cnt;
/* algo :
x = v[i] - v[i+1] without signed saturation
( XXX see if there is'nt problem, but can't be with signed
sat because pixel will be saturate :(
x = v[i] - v[i+1] without signed saturation
( XXX see if there is'nt problem, but can't be with signed
sat because pixel will be saturate :(
so x within [-128, 127] and we have to test if it fit in [-M, M]
we add 127-M with wrap around -> good value fit in [ 127-2*M, 127]
and if x >= 127 - 2 * M ie x > 127 -2*M - 1 value is good
*/
__asm__ __volatile__ (" \n\
#* Do (v0-v1) to (v7-v8) \n\
movq (%1), %%mm1 # load v0->v7 \n\
movq 1(%1), %%mm2 # load v1->v8 \n\
psubb %%mm2, %%mm1 # v[i]-v[i+1] \n\
paddb mmx_127_thr1, %%mm1 # + 127-THR1 with wrap \n\
pcmpgtb mmx_127_2xthr1_1, %%mm1 # > 127 -2*thr1 - 1 \n"
"movq %%mm1, %%mm0 # \n\
psrlw $8, %%mm1 # \n\
paddb %%mm1, %%mm0 # \n\
# \n\
movq %%mm0, %%mm1 # Now sum to create eq_cnt \n\
psrld $16, %%mm0 # \n\
paddb %%mm0, %%mm1 # \n\
# \n\
movq %%mm1, %%mm0 # \n\
psrlq $32, %%mm1 # \n\
paddb %%mm1, %%mm0 \n"
"movd %%mm0, %0 # \n\
negl %0 \n\
andl $255, %0"
__asm__ __volatile__ (
" #* Do (v0-v1) to (v7-v8) \n"
"movq (%1), %%mm1 # load v0->v7 \n"
"movq 1(%1), %%mm2 # load v1->v8 \n"
"psubb %%mm2, %%mm1 # v[i]-v[i+1] \n"
"paddb mmx_127_thr1, %%mm1 # + 127-THR1 with wrap \n"
"pcmpgtb mmx_127_2xthr1_1, %%mm1 # > 127 -2*thr1 - 1 \n"
"movq %%mm1, %%mm0 # \n"
"psrlw $8, %%mm1 # \n"
"paddb %%mm1, %%mm0 # \n"
" # \n"
"movq %%mm0, %%mm1 # Now sum to create eq_cnt \n"
"psrld $16, %%mm0 # \n"
"paddb %%mm0, %%mm1 # \n"
" # \n"
"movq %%mm1, %%mm0 # \n"
"psrlq $32, %%mm1 # \n"
"paddb %%mm1, %%mm0 \n"
"movd %%mm0, %0 # \n"
"negl %0 \n"
"andl $255, %0"
: "=r"(i_eq_cnt) : "r" (p_v) );
/* last test, hey, 9 don't fit in MMX */
if(( ( p_v[8] - p_v[9] + PP_THR1 )&0xffff )<= PP_2xTHR1 )
{
if(( ( p_v[8] - p_v[9] + PP_THR1 )&0xffff )<= PP_2xTHR1 )
{
i_eq_cnt++;
}
......@@ -160,8 +160,8 @@ static inline int pp_deblock_isDC_mode( u8 *p_v )
for( i =0; i < 9; i++ )
{
if(( ( p_v[i] - p_v[i+1] + PP_THR1 )&0xffff )<= PP_2xTHR1 )
{
if(( ( p_v[i] - p_v[i+1] + PP_THR1 )&0xffff )<= PP_2xTHR1 )
{
i_eq_cnt++;
}
}
......@@ -170,25 +170,25 @@ static inline int pp_deblock_isDC_mode( u8 *p_v )
return( (i_eq_cnt >= PP_THR2 ) ? 1 : 0 );
}
static inline int pp_deblock_isMinMaxOk( u8 *p_v, int i_QP )
static inline int pp_deblock_isMinMaxOk( uint8_t *p_v, int i_QP )
{
int i_range;
#if 0
__asm__ __volatile__ (
__asm__ __volatile__ (
"movq 1(%1), %%mm0 # 8 bytes \n"
"movq %%mm0, %%mm1 \n"
MMXEXT_GET_PMIN( %%mm0, %%mm7 )
MMXEXT_GET_PMAX( %%mm1, %%mm7 )
"psubd %%mm0, %%mm1 # max - min \n\
movd %%mm1, %0 \n\
andl $255, %0" : "=r"(i_range) : "r"(p_v) );
"psubd %%mm0, %%mm1 # max - min \n"
"movd %%mm1, %0 \n"
"andl $255, %0" : "=r"(i_range) : "r"(p_v) );
#endif
int i_max, i_min;
int i;
i_min = i_max = p_v[1];
i_min = i_max = p_v[1];
for( i = 2; i < 9; i++ )
{
if( i_max < p_v[i] ) i_max = p_v[i];
......@@ -200,7 +200,7 @@ static inline int pp_deblock_isMinMaxOk( u8 *p_v, int i_QP )
}
static inline void pp_deblock_DefaultMode( u8 i_v[10], int i_stride,
static inline void pp_deblock_DefaultMode( uint8_t i_v[10], int i_stride,
int i_QP )
{
int d, i_delta;
......@@ -210,16 +210,16 @@ static inline void pp_deblock_DefaultMode( u8 i_v[10], int i_stride,
/* d = CLIP( 5(a3x0' - a3x0)//8, 0, (v4-v5)/2 ).d( abs(a3x0) < QP ) */
/* First calculate a3x0 */
__asm__ __volatile__ ( " \n\
pxor %%mm7, %%mm7 # mm7 = 0 \n\
movq mmx_m2_5_m5_2, %%mm6 # mm6 =(2,-5,5,-2) \n\
movd 3(%1), %%mm0 \n\
punpcklbw %%mm7,%%mm0 \n\
pmaddwd %%mm6, %%mm0 \n"
"movq %%mm0, %%mm1 \n\
psrlq $32, %%mm1 \n"
"paddd %%mm1, %%mm0 \n\
movd %%mm0, %0" : "=r"(a3x0) :"r"(i_v) );
__asm__ __volatile__ (
"pxor %%mm7, %%mm7 # mm7 = 0 \n"
"movq mmx_m2_5_m5_2, %%mm6 # mm6 =(2,-5,5,-2) \n"
"movd 3(%1), %%mm0 \n"
"punpcklbw %%mm7,%%mm0 \n"
"pmaddwd %%mm6, %%mm0 \n"
"movq %%mm0, %%mm1 \n"
"psrlq $32, %%mm1 \n"
"paddd %%mm1, %%mm0 \n"
"movd %%mm0, %0" : "=r"(a3x0) :"r"(i_v) );
#if 0
a3x0 = 2 * ( i_v[3] - i_v[6] ) + 5 *( i_v[5] - i_v[4] );
#endif
......@@ -228,7 +228,7 @@ static inline void pp_deblock_DefaultMode( u8 i_v[10], int i_stride,
{
b_neg = 1;
a3x0 = -a3x0;
}
}
else
{
b_neg = 0;
......@@ -237,24 +237,24 @@ static inline void pp_deblock_DefaultMode( u8 i_v[10], int i_stride,
if( ( a3x0 < 8 * i_QP )&&( a3x0 != 0 ) ) /* |a3x0| < 8*i_QP */
{
/* calculate a3x1 et a3x2 */
__asm__ __volatile__ ( " \n\
# mm7 = 0 \n\
# mm6 = ( 2, -5, 5, -2 ) \n\
movd 1(%2), %%mm0 \n\
movd 5(%2), %%mm2 \n\
punpcklbw %%mm7,%%mm0 \n\
punpcklbw %%mm7,%%mm2 \n\
pmaddwd %%mm6, %%mm0 \n\
pmaddwd %%mm6, %%mm2 \n"
"movq %%mm0, %%mm1 \n\
psrlq $32, %%mm1 \n"
"paddd %%mm1, %%mm0 # mm0 = a3x1 \n\
movd %%mm0, %0 \n"
"movq %%mm2, %%mm1 \n\
psrlq $32, %%mm1 \n"
"paddd %%mm1, %%mm2 # mm2 = a3x2 \n\
movd %%mm2, %1 \n\
" : "=r"(a3x1), "=r"(a3x2) : "r"(i_v) );
__asm__ __volatile__ (
" # mm7 = 0 \n"
" # mm6 = ( 2, -5, 5, -2 ) \n"
"movd 1(%2), %%mm0 \n"
"movd 5(%2), %%mm2 \n"
"punpcklbw %%mm7,%%mm0 \n"
"punpcklbw %%mm7,%%mm2 \n"
"pmaddwd %%mm6, %%mm0 \n"
"pmaddwd %%mm6, %%mm2 \n"
"movq %%mm0, %%mm1 \n"
"psrlq $32, %%mm1 \n"
"paddd %%mm1, %%mm0 # mm0 = a3x1 \n"
"movd %%mm0, %0 \n"
"movq %%mm2, %%mm1 \n"
"psrlq $32, %%mm1 \n"
"paddd %%mm1, %%mm2 # mm2 = a3x2 \n"
"movd %%mm2, %1 \n"
: "=r"(a3x1), "=r"(a3x2) : "r"(i_v) );
#if 0
a3x1 = 2 * ( i_v[1] - i_v[4] ) + 5 * ( i_v[3] - i_v[2] );
a3x2 = 2 * ( i_v[5] - i_v[8] ) + 5 * ( i_v[7] - i_v[6] );
......@@ -264,7 +264,7 @@ static inline void pp_deblock_DefaultMode( u8 i_v[10], int i_stride,
if( a3x2 < 0) a3x2 = -a3x2; /* abs( a3x2 ) */
a3x0_ = PP_MIN3( a3x0, a3x1, a3x2 );
d = 5 *( a3x0 - a3x0_ ) / 8; /* always > 0 */
i_delta = ( i_v[4] - i_v[5] ) / 2;
......@@ -293,7 +293,7 @@ static inline void pp_deblock_DefaultMode( u8 i_v[10], int i_stride,
static inline void pp_deblock_DCMode( u8 *p_v, /* = int i_v[10] */
static inline void pp_deblock_DCMode( uint8_t *p_v, /* = int i_v[10] */
int i_QP )
{
int i_p0, i_p9;
......@@ -302,20 +302,20 @@ static inline void pp_deblock_DCMode( u8 *p_v, /* = int i_v[10] */
i_p9 = PP_ABS( p_v[8] - p_v[9] ) < i_QP ? p_v[9] : p_v[8];
/* mm0 = 8 pix unmodified
-We will process first 4 pixel
-We will process first 4 pixel
mm0 = 8 pix unmodified
mm1 = for the first part of the 4 first pix
(v1) -> (p0) -> ... ( word )
(v2) (v1)
(v3) (v2)
(v4) (v3)
= for the commoin part between first and last pix
(v2) -> (v3) -> ... ( word )
(v3) (v4)
(v4) (v5)
(v5) (v6)
= for the last part of the 4 last pix
(v5) -> (v6) -> ... ( word )
(v6) (v7)
......@@ -323,110 +323,110 @@ static inline void pp_deblock_DCMode( u8 *p_v, /* = int i_v[10] */
(v8) (p9)
mm2 = acu for first new pix
mm3 = acu for last pix
mm3 = acu for last pix
mm4 = unused
mm5 = p0
mm5 = p0
mm6 = p9 << 48
mm7 = 0 */
__asm__ __volatile__ (
"pxor %%mm7, %%mm7 \n\
movq 1(%0), %%mm0 # get 8 pix \n\
# unpack into mm1 \n\
movq %%mm0, %%mm1 \n\
punpcklbw %%mm7, %%mm1 \n\
# get p_0 and i_p9 \n\
movd %1, %%mm5 \n\
movd %2, %%mm6 \n\
psllq $48, %%mm6 \n
\n\
movq %%mm1, %%mm3 # p_v[5-8] = v[1-4] !! \n\
movq %%mm1, %%mm2 \n\
psllw $2, %%mm2 # p_v[1-4] = 4*v[1-4] \n\
\n\
psllq $16, %%mm1 \n\
por %%mm5, %%mm1 # mm1 =( p0, v1, v2 ,v3)\n\
\n\
paddw %%mm1, %%mm2 \n\
paddw %%mm1, %%mm2 \n\
\n\
psllq $16, %%mm1 \n\
por %%mm5, %%mm1 # mm1 =( p0, p0, v1, v2)\n\
\n\
paddw %%mm1, %%mm2 \n\
paddw %%mm1, %%mm2 \n\
\n\
psllq $16, %%mm1 \n\
por %%mm5, %%mm1 # mm1 =( p0, p0, p0, v1)\n\
\n\
paddw %%mm1, %%mm2 \n\
\n\
psllq $16, %%mm1 \n\
por %%mm5, %%mm1 # mm1 =( p0, p0, p0, p0)\n\
\n\
paddw %%mm1, %%mm2 \n\
# Now last part a little borring\n\
# last part for mm2, beginig for mm3
movq %%mm0, %%mm1 \n\
psrlq $8, %%mm1 \n\
punpcklbw %%mm7, %%mm1 # mm1 =( v2, v3, v4, v5 )\n\
paddw %%mm1, %%mm2 \n\
paddw %%mm1, %%mm2 \n\
paddw %%mm1, %%mm3 \n\
\n\
movq %%mm0, %%mm1 \n\
psrlq $16, %%mm1 \n\
punpcklbw %%mm7, %%mm1 # mm1 =( v3, v4, v5, v6 )\n\
psllw $1, %%mm1 \n\
paddw %%mm1, %%mm2 \n\
paddw %%mm1, %%mm3 \n\
\n\
movq %%mm0, %%mm1 \n\
psrlq $24, %%mm1 \n\
punpcklbw %%mm7, %%mm1 # mm1 =( v4, v5, v6, v7) \n\
paddw %%mm1, %%mm2 \n\
paddw %%mm1, %%mm3 \n\
paddw %%mm1, %%mm3 \n\
\n\
movq %%mm0, %%mm1 \n\
psrlq $32, %%mm1 \n\
punpcklbw %%mm7, %%mm1 # mm1 =( v5, v6, v7, v8) \n\
paddw %%mm1, %%mm2 \n\
psllw $2, %%mm1
paddw %%mm1, %%mm3 \n\
# Now last part for last 4 pix \n\
# \n\
movq %%mm0, %%mm1 \n\
punpckhbw %%mm7, %%mm1 # mm1 = ( v5, v6, v7, v8) \n\
\n\
psrlq $16, %%mm1 \n\
por %%mm6, %%mm1 # mm1 =( v6, v7, v8, p9 )\n\
\n\
paddw %%mm1, %%mm3 \n\
paddw %%mm1, %%mm3 \n\
\n\
psrlq $16, %%mm1 \n\
por %%mm6, %%mm1 # mm1 =( v7, v8, p9, p9)\n\
\n\
paddw %%mm1, %%mm3 \n\
paddw %%mm1, %%mm3 \n\
\n\
psrlq $16, %%mm1 \n\
por %%mm6, %%mm1 # mm1 =( v8, p9, p9, p9 )\n\
\n\
paddw %%mm1, %%mm3 \n\
\n\
psrlq $16, %%mm1 \n\
por %%mm6, %%mm1 # mm1 =( p9, p9, p9, p9 )\n\
\n\
paddw %%mm1, %%mm3 \n\
psrlw $4, %%mm2 \n\
psrlw $4, %%mm3 \n\
packuswb %%mm3, %%mm2 \n\
movq %%mm2, 1(%0) \n\
": : "r"(p_v), "r"(i_p0), "r"(i_p9) : "memory" );
"pxor %%mm7, %%mm7 \n"
"movq 1(%0), %%mm0 # get 8 pix \n"
" # unpack into mm1 \n"
"movq %%mm0, %%mm1 \n"
"punpcklbw %%mm7, %%mm1 \n"
" # get p_0 and i_p9 \n"
"movd %1, %%mm5 \n"
"movd %2, %%mm6 \n"
"psllq $48, %%mm6 \n"
" \n"
"movq %%mm1, %%mm3 # p_v[5-8] = v[1-4] !! \n"
"movq %%mm1, %%mm2 \n"
"psllw $2, %%mm2 # p_v[1-4] = 4*v[1-4] \n"
" \n"
"psllq $16, %%mm1 \n"
"por %%mm5, %%mm1 # mm1 =( p0, v1, v2 ,v3)\n"
" \n"
"paddw %%mm1, %%mm2 \n"
"paddw %%mm1, %%mm2 \n"
" \n"
"psllq $16, %%mm1 \n"
"por %%mm5, %%mm1 # mm1 =( p0, p0, v1, v2)\n"
" \n"
"paddw %%mm1, %%mm2 \n"
"paddw %%mm1, %%mm2 \n"
" \n"
"psllq $16, %%mm1 \n"
"por %%mm5, %%mm1 # mm1 =( p0, p0, p0, v1)\n"
" \n"
"paddw %%mm1, %%mm2 \n"
" \n"
"psllq $16, %%mm1 \n"
"por %%mm5, %%mm1 # mm1 =( p0, p0, p0, p0)\n"
" \n"
"paddw %%mm1, %%mm2 \n"
" # Now last part a little borring\n"
" # last part for mm2, beginig for mm3\n"
"movq %%mm0, %%mm1 \n"
"psrlq $8, %%mm1 \n"
"punpcklbw %%mm7, %%mm1 # mm1 =( v2, v3, v4, v5 )\n"
"paddw %%mm1, %%mm2 \n"
"paddw %%mm1, %%mm2 \n"
"paddw %%mm1, %%mm3 \n"
" \n"
"movq %%mm0, %%mm1 \n"
"psrlq $16, %%mm1 \n"
"punpcklbw %%mm7, %%mm1 # mm1 =( v3, v4, v5, v6 )\n"