Added general IDCT library. Added IEEE testbed.

Original commit message from CVS:
Added general IDCT library. Added IEEE testbed.
Update to the windows codec loader.
fixes in the riff parser/encoder.
cleanup in the getbits/putbits API.

libs/Makefile.am

@@ -1,3 +1,3 @@
-SUBDIRS = riff colorspace getbits putbits videoscale winloader
+SUBDIRS = riff colorspace getbits putbits videoscale winloader idct
 
-DIST_SUBDIRS = riff colorspace getbits putbits videoscale winloader
+DIST_SUBDIRS = riff colorspace getbits putbits videoscale winloader idct

libs/getbits/gstgetbits.c

@@ -200,9 +200,9 @@ void getbits_back_int(gst_getbits_t *gb,unsigned long bits) {
     }
   }
   gb->dword = bswap_32(*((unsigned long *)(gb->ptr)));
-  fprintf(stderr,"orignal new loaded word is %08x\n",gb->dword);
+  fprintf(stderr,"orignal new loaded word is %08lx\n",gb->dword);
   gb->dword <<= (32 - gb->bits);
-  fprintf(stderr,"shifted (by %lu) word is %08x\n",gb->bits,gb->dword);
+  fprintf(stderr,"shifted (by %lu) word is %08lx\n",gb->bits,gb->dword);
 }
 
 void getbits_byteback_int(gst_getbits_t *gb,unsigned long bytes) {

libs/getbits/gstgetbits_inl.h

@@ -277,7 +277,7 @@ extern unsigned long gst_getbits_nBitMask[];
 }
 
 
-#define gst_flushbits(gb, num)                                          \
+#define gst_flushbitsn(gb, num)                                         \
 {                                                                       \
   (gb)->bits += num;                                      		\
                                                                       	\

libs/idct/Makefile.am

@@ -0,0 +1,19 @@
+filterdir = $(libdir)/gst
+
+filter_LTLIBRARIES = libgstidct.la
+
+libgstidct_la_SOURCES = fastintidct.c floatidct.c gstidct.c intidct.c mmxidct.S mmx32idct.c
+
+bin_PROGRAMS = ieeetest
+
+ieeetest_SOURCES = ieeetest.c
+ieeetest_LDADD = libgstidct.la $(GLIB_LIBS) $(GTK_LIBS) $(top_srcdir)/gst/libgst.la
+ieeetest_CFLAGS = $(shell gnome-config --cflags gnomeui) -g -Wall
+ieeetest_LDFLAGS = $(shell gnome-config --libs gnomeui) 
+
+noinst_HEADERS = gstidct.h
+
+CFLAGS += -Wall -O2 -funroll-all-loops -finline-functions -ffast-math
+
+INCLUDES = $(GLIB_CFLAGS) $(GTK_CFLAGS) -I$(top_srcdir) -I$(top_srcdir)/include
+LDADD = $(GLIB_LIBS) $(GTK_LIBS) $(top_srcdir)/gst/libgst.la

libs/idct/README

@@ -0,0 +1,48 @@
+This archive contains a quick & dirty implementation of the IEEE Standard
+1180-1990 accuracy test for inverse DCT.  It is not guaranteed to be
+correct ... but if you find any bugs, please let me know (by email to
+tgl@cs.cmu.edu).
+
+The test harness consists of the C program ieeetest.c and shell script
+doieee.  For comparison purposes I have also supplied a copy of jrevdct.c,
+the inverse DCT routine from release 4 of the Independent JPEG Group's
+free JPEG software.  (jrevdct.c is slightly modified from the IJG release
+so that it will compile without the IJG include files.)  jrevdct.c passes
+the 1180 test --- or at least, this program thinks so.  jrevdct.out is
+the output from a test run.
+
+Note that numerical results may vary somewhat across machines.  This appears
+to be mostly due to differing results from the cosine function.
+
+
+INSTALLATION:
+  Check the Makefile, change CC and CFLAGS if needed.  Then say "make".
+If your C compiler is non-ANSI, you may need to change includes and/or
+function headers.
+
+  To test a different IDCT routine, link with that routine instead of
+jrevdct.o.  You will need to modify dct.h and/or ieeetest.c if your
+routine's calling convention is not in-place modification of an array
+of 64 "short"s.
+
+
+USAGE:
+  The standard test procedure is
+	doieee ieeetest >outputfile
+Expect it to take a while (almost 80 minutes on my old 68030 box).
+Each of the six passes will emit a row of 100 dots as it runs.
+
+You can grep the output for the word FAILS if you just want to know
+yea or nay.
+
+
+LEGAL MUMBO-JUMBO:
+  I hereby release the test harness to the public domain.
+			Thomas G. Lane, 22 Nov 1993
+
+IMPORTANT: jrevdct.c is NOT public domain, but is copyrighted free software
+(not the same thing at all).  It is subject to IJG's distribution terms, which
+primarily state that if you incorporate it into a program you must acknowledge
+IJG's contribution in your program documentation.  For more details and the
+complete IJG software, see the IJG FTP archive at ftp.uu.net, in directory
+/graphics/jpeg.

libs/idct/dct.h

@@ -0,0 +1,29 @@
+/* define DCT types */
+
+/*
+ * DCTSIZE      underlying (1d) transform size
+ * DCTSIZE2     DCTSIZE squared
+ */
+
+#define DCTSIZE      (8)
+#define DCTSIZE2     (DCTSIZE*DCTSIZE)
+
+#define EIGHT_BIT_SAMPLES	/* needed in jrevdct.c */
+
+typedef short DCTELEM;		/* must be at least 16 bits */
+
+typedef DCTELEM DCTBLOCK[DCTSIZE2];
+
+typedef long INT32;		/* must be at least 32 bits */
+
+extern void gst_idct_int_idct();
+
+extern void gst_idct_init_fast_int_idct (void);
+extern void gst_idct_fast_int_idct (short *block);
+
+extern void gst_idct_mmx_idct (short *block);
+extern void gst_idct_mmx32_idct (short *block);
+
+extern void gst_idct_init_float_idct(void);
+extern void gst_idct_float_idct (short *block);
+

libs/idct/doieee

@@ -0,0 +1,15 @@
+# perform IEEE 1180 test series
+# Typical usage:
+#    doieee  >outfile
+# where progname is ieeetest or a variant
+
+for i in 1 2 3 4 5; 
+do
+time ./ieeetest $i -256 255 1 10000
+time ./ieeetest $i -5 5     1 10000
+time ./ieeetest $i -300 300 1 10000
+
+time ./ieeetest $i -256 255 -1 10000
+time ./ieeetest $i -5 5     -1 10000
+time ./ieeetest $i -300 300 -1 10000
+done

libs/idct/fastintidct.c

@@ -0,0 +1,207 @@
+/* idct.c, inverse fast discrete cosine transform                           */
+
+/* Copyright (C) 1996, MPEG Software Simulation Group. All Rights Reserved. */
+
+/*
+ * Disclaimer of Warranty
+ *
+ * These software programs are available to the user without any license fee or
+ * royalty on an "as is" basis.  The MPEG Software Simulation Group disclaims
+ * any and all warranties, whether express, implied, or statuary, including any
+ * implied warranties or merchantability or of fitness for a particular
+ * purpose.  In no event shall the copyright-holder be liable for any
+ * incidental, punitive, or consequential damages of any kind whatsoever
+ * arising from the use of these programs.
+ *
+ * This disclaimer of warranty extends to the user of these programs and user's
+ * customers, employees, agents, transferees, successors, and assigns.
+ *
+ * The MPEG Software Simulation Group does not represent or warrant that the
+ * programs furnished hereunder are free of infringement of any third-party
+ * patents.
+ *
+ * Commercial implementations of MPEG-1 and MPEG-2 video, including shareware,
+ * are subject to royalty fees to patent holders.  Many of these patents are
+ * general enough such that they are unavoidable regardless of implementation
+ * design.
+ *
+ */
+
+/**********************************************************/
+/* inverse two dimensional DCT, Chen-Wang algorithm       */
+/* (cf. IEEE ASSP-32, pp. 803-816, Aug. 1984)             */
+/* 32-bit integer arithmetic (8 bit coefficients)         */
+/* 11 mults, 29 adds per DCT                              */
+/*                                      sE, 18.8.91       */
+/**********************************************************/
+/* coefficients extended to 12 bit for IEEE1180-1990      */
+/* compliance                           sE,  2.1.94       */
+/**********************************************************/
+
+/* this code assumes >> to be a two's-complement arithmetic */
+/* right shift: (-2)>>1 == -1 , (-3)>>1 == -2               */
+
+#define W1 2841 /* 2048*sqrt(2)*cos(1*pi/16) */
+#define W2 2676 /* 2048*sqrt(2)*cos(2*pi/16) */
+#define W3 2408 /* 2048*sqrt(2)*cos(3*pi/16) */
+#define W5 1609 /* 2048*sqrt(2)*cos(5*pi/16) */
+#define W6 1108 /* 2048*sqrt(2)*cos(6*pi/16) */
+#define W7 565  /* 2048*sqrt(2)*cos(7*pi/16) */
+
+#include "dct.h"
+
+/* private data */
+static short iclip[1024]; /* clipping table */
+static short *iclp;
+
+/* private prototypes */
+static void idctrow (short *blk);
+static void idctcol (short *blk);
+
+/* row (horizontal) IDCT
+ *
+ *           7                       pi         1
+ * dst[k] = sum c[l] * src[l] * cos( -- * ( k + - ) * l )
+ *          l=0                      8          2
+ *
+ * where: c[0]    = 128
+ *        c[1..7] = 128*sqrt(2)
+ */
+
+static void idctrow(blk)
+short *blk;
+{
+  int x0, x1, x2, x3, x4, x5, x6, x7, x8;
+
+  /* shortcut */
+  if (!((x1 = blk[4]<<11) | (x2 = blk[6]) | (x3 = blk[2]) |
+        (x4 = blk[1]) | (x5 = blk[7]) | (x6 = blk[5]) | (x7 = blk[3])))
+  {
+    blk[0]=blk[1]=blk[2]=blk[3]=blk[4]=blk[5]=blk[6]=blk[7]=blk[0]<<3;
+    return;
+  }
+
+  x0 = (blk[0]<<11) + 128; /* for proper rounding in the fourth stage */
+
+  /* first stage */
+  x8 = W7*(x4+x5);
+  x4 = x8 + (W1-W7)*x4;
+  x5 = x8 - (W1+W7)*x5;
+  x8 = W3*(x6+x7);
+  x6 = x8 - (W3-W5)*x6;
+  x7 = x8 - (W3+W5)*x7;
+  
+  /* second stage */
+  x8 = x0 + x1;
+  x0 -= x1;
+  x1 = W6*(x3+x2);
+  x2 = x1 - (W2+W6)*x2;
+  x3 = x1 + (W2-W6)*x3;
+  x1 = x4 + x6;
+  x4 -= x6;
+  x6 = x5 + x7;
+  x5 -= x7;
+  
+  /* third stage */
+  x7 = x8 + x3;
+  x8 -= x3;
+  x3 = x0 + x2;
+  x0 -= x2;
+  x2 = (181*(x4+x5)+128)>>8;
+  x4 = (181*(x4-x5)+128)>>8;
+  
+  /* fourth stage */
+  blk[0] = (x7+x1)>>8;
+  blk[1] = (x3+x2)>>8;
+  blk[2] = (x0+x4)>>8;
+  blk[3] = (x8+x6)>>8;
+  blk[4] = (x8-x6)>>8;
+  blk[5] = (x0-x4)>>8;
+  blk[6] = (x3-x2)>>8;
+  blk[7] = (x7-x1)>>8;
+}
+
+/* column (vertical) IDCT
+ *
+ *             7                         pi         1
+ * dst[8*k] = sum c[l] * src[8*l] * cos( -- * ( k + - ) * l )
+ *            l=0                        8          2
+ *
+ * where: c[0]    = 1/1024
+ *        c[1..7] = (1/1024)*sqrt(2)
+ */
+static void idctcol(blk)
+short *blk;
+{
+  int x0, x1, x2, x3, x4, x5, x6, x7, x8;
+
+  /* shortcut */
+  if (!((x1 = (blk[8*4]<<8)) | (x2 = blk[8*6]) | (x3 = blk[8*2]) |
+        (x4 = blk[8*1]) | (x5 = blk[8*7]) | (x6 = blk[8*5]) | (x7 = blk[8*3])))
+  {
+    blk[8*0]=blk[8*1]=blk[8*2]=blk[8*3]=blk[8*4]=blk[8*5]=blk[8*6]=blk[8*7]=
+      iclp[(blk[8*0]+32)>>6];
+    return;
+  }
+
+  x0 = (blk[8*0]<<8) + 8192;
+
+  /* first stage */
+  x8 = W7*(x4+x5) + 4;
+  x4 = (x8+(W1-W7)*x4)>>3;
+  x5 = (x8-(W1+W7)*x5)>>3;
+  x8 = W3*(x6+x7) + 4;
+  x6 = (x8-(W3-W5)*x6)>>3;
+  x7 = (x8-(W3+W5)*x7)>>3;
+  
+  /* second stage */
+  x8 = x0 + x1;
+  x0 -= x1;
+  x1 = W6*(x3+x2) + 4;
+  x2 = (x1-(W2+W6)*x2)>>3;
+  x3 = (x1+(W2-W6)*x3)>>3;
+  x1 = x4 + x6;
+  x4 -= x6;
+  x6 = x5 + x7;
+  x5 -= x7;
+  
+  /* third stage */
+  x7 = x8 + x3;
+  x8 -= x3;
+  x3 = x0 + x2;
+  x0 -= x2;
+  x2 = (181*(x4+x5)+128)>>8;
+  x4 = (181*(x4-x5)+128)>>8;
+  
+  /* fourth stage */
+  blk[8*0] = iclp[(x7+x1)>>14];
+  blk[8*1] = iclp[(x3+x2)>>14];
+  blk[8*2] = iclp[(x0+x4)>>14];
+  blk[8*3] = iclp[(x8+x6)>>14];
+  blk[8*4] = iclp[(x8-x6)>>14];
+  blk[8*5] = iclp[(x0-x4)>>14];
+  blk[8*6] = iclp[(x3-x2)>>14];
+  blk[8*7] = iclp[(x7-x1)>>14];
+}
+
+/* two dimensional inverse discrete cosine transform */
+void gst_idct_fast_int_idct(block)
+short *block;
+{
+  int i;
+
+  for (i=0; i<8; i++)
+    idctrow(block+8*i);
+
+  for (i=0; i<8; i++)
+    idctcol(block+i);
+}
+
+void gst_idct_init_fast_int_idct()
+{
+  int i;
+
+  iclp = iclip+512;
+  for (i= -512; i<512; i++)
+    iclp[i] = (i<-256) ? -256 : ((i>255) ? 255 : i);
+}

libs/idct/floatidct.c

@@ -0,0 +1,102 @@
+/* Reference_IDCT.c, Inverse Discrete Fourier Transform, double precision          */
+
+/* Copyright (C) 1996, MPEG Software Simulation Group. All Rights Reserved. */
+
+/*
+ * Disclaimer of Warranty
+ *
+ * These software programs are available to the user without any license fee or
+ * royalty on an "as is" basis.  The MPEG Software Simulation Group disclaims
+ * any and all warranties, whether express, implied, or statuary, including any
+ * implied warranties or merchantability or of fitness for a particular
+ * purpose.  In no event shall the copyright-holder be liable for any
+ * incidental, punitive, or consequential damages of any kind whatsoever
+ * arising from the use of these programs.
+ *
+ * This disclaimer of warranty extends to the user of these programs and user's
+ * customers, employees, agents, transferees, successors, and assigns.
+ *
+ * The MPEG Software Simulation Group does not represent or warrant that the
+ * programs furnished hereunder are free of infringement of any third-party
+ * patents.
+ *
+ * Commercial implementations of MPEG-1 and MPEG-2 video, including shareware,
+ * are subject to royalty fees to patent holders.  Many of these patents are
+ * general enough such that they are unavoidable regardless of implementation
+ * design.
+ *
+ */
+
+/*  Perform IEEE 1180 reference (64-bit floating point, separable 8x1
+ *  direct matrix multiply) Inverse Discrete Cosine Transform
+*/
+
+
+/* Here we use math.h to generate constants.  Compiler results may
+   vary a little */
+
+#include <math.h>
+
+#ifndef PI
+# ifdef M_PI
+#  define PI M_PI
+# else
+#  define PI 3.14159265358979323846
+# endif
+#endif
+
+/* private data */
+
+/* cosine transform matrix for 8x1 IDCT */
+static double gst_idct_float_c[8][8];
+
+/* initialize DCT coefficient matrix */
+
+void gst_idct_init_float_idct()
+{
+  int freq, time;
+  double scale;
+
+  for (freq=0; freq < 8; freq++)
+  {
+    scale = (freq == 0) ? sqrt(0.125) : 0.5;
+    for (time=0; time<8; time++)
+      gst_idct_float_c[freq][time] = scale*cos((PI/8.0)*freq*(time + 0.5));
+  }
+}
+
+/* perform IDCT matrix multiply for 8x8 coefficient block */
+
+void gst_idct_float_idct(block)
+short *block;
+{
+  int i, j, k, v;
+  double partial_product;
+  double tmp[64];
+
+  for (i=0; i<8; i++)
+    for (j=0; j<8; j++)
+    {
+      partial_product = 0.0;
+
+      for (k=0; k<8; k++)
+        partial_product+= gst_idct_float_c[k][j]*block[8*i+k];
+
+      tmp[8*i+j] = partial_product;
+    }
+
+  /* Transpose operation is integrated into address mapping by switching 
+     loop order of i and j */
+
+  for (j=0; j<8; j++)
+    for (i=0; i<8; i++)
+    {
+      partial_product = 0.0;
+
+      for (k=0; k<8; k++)
+        partial_product+= gst_idct_float_c[k][i]*tmp[8*k+j];
+
+      v = (int) floor(partial_product+0.5);
+      block[8*i+j] = (v<-256) ? -256 : ((v>255) ? 255 : v);
+    }
+}

libs/idct/gstidct.c

@@ -0,0 +1,111 @@
+/* Gnome-Streamer
+ * Copyright (C) <1999> Erik Walthinsen <omega@cse.ogi.edu>
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library 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 GNU
+ * Library General Public License for more details.
+ *
+ * You should have received a copy of the GNU Library General Public
+ * License along with this library; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+
+#include <config.h>
+
+#include "gstidct.h"
+#include "dct.h"
+
+static void gst_idct_int_sparse_idct(short *data);
+
+GstIDCT *gst_idct_new(GstIDCTMethod method) 
+{
+  GstIDCT *new = g_malloc(sizeof(GstIDCT));
+
+  new->need_transpose = FALSE;
+
+  if (method == GST_IDCT_DEFAULT) {
+#ifdef HAVE_LIBMMX
+	 method = GST_IDCT_MMX32;
+#else
+	 method = GST_IDCT_FAST_INT;
+#endif
+  }
+
+  new->convert_sparse = gst_idct_int_sparse_idct;
+
+  switch (method) {
+	 case GST_IDCT_FAST_INT:
+		g_print("GstIDCT: using fast_int_idct\n");
+	   gst_idct_init_fast_int_idct();
+		new->convert = gst_idct_fast_int_idct;
+		break;
+	 case GST_IDCT_INT:
+		g_print("GstIDCT: using int_idct\n");
+		new->convert = gst_idct_int_idct;
+		break;
+	 case GST_IDCT_FLOAT:
+		g_print("GstIDCT: using float_idct\n");
+		gst_idct_init_float_idct();
+		new->convert = gst_idct_float_idct;
+		break;
+	 case GST_IDCT_MMX:
+		g_print("GstIDCT: using MMX_idct\n");
+		new->convert = gst_idct_mmx_idct;
+		new->need_transpose = TRUE;
+		break;
+	 case GST_IDCT_MMX32:
+		g_print("GstIDCT: using MMX32_idct\n");
+		new->convert = gst_idct_mmx32_idct;
+		new->need_transpose = TRUE;
+		break;
+	 default:
+		g_print("GstIDCT: method not supported\n");
+		g_free(new);
+		return NULL;
+  }
+  return new;
+}
+
+static void gst_idct_int_sparse_idct(short *data) 
+{
+  short val;
+  gint32 v, *dp = (guint32 *)data;
+
+  v = *data;
+
+  if (v < 0) {
+    val = -v;
+    val += (8 >> 1);
+    val /= 8;
+    val = -val;
+  }
+  else {
+    val = (v + (8 >> 1)) / 8;
+  }
+  v = (( val & 0xffff) | (val << 16));
+
+  dp[0] = v;  dp[1] = v;  dp[2] = v;  dp[3] = v;
+  dp[4] = v;  dp[5] = v;  dp[6] = v;  dp[7] = v;
+  dp[8] = v;  dp[9] = v;  dp[10] = v; dp[11] = v;
+  dp[12] = v; dp[13] = v; dp[14] = v; dp[15] = v;
+  dp[16] = v; dp[17] = v; dp[18] = v; dp[19] = v;
+  dp[20] = v; dp[21] = v; dp[22] = v; dp[23] = v;
+  dp[24] = v; dp[25] = v; dp[26] = v; dp[27] = v;
+  dp[28] = v; dp[29] = v; dp[30] = v; dp[31] = v;
+}
+
+void gst_idct_destroy(GstIDCT *idct) 
+{
+  g_return_if_fail(idct != NULL);
+
+  g_free(idct);
+}
+

libs/idct/gstidct.h

@@ -0,0 +1,53 @@
+/* Gnome-Streamer
+ * Copyright (C) <1999> Erik Walthinsen <omega@cse.ogi.edu>
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library 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 GNU
+ * Library General Public License for more details.
+ *
+ * You should have received a copy of the GNU Library General Public
+ * License along with this library; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+
+#ifndef __GST_IDCT_H__
+#define __GST_IDCT_H__
+
+#include <glib.h>
+
+typedef enum {
+  GST_IDCT_DEFAULT,         // default 
+  GST_IDCT_INT,				 // integer IDCT
+  GST_IDCT_FAST_INT,        // fastest integer
+  GST_IDCT_FLOAT,	          // accurate float version
+  GST_IDCT_MMX,			    // fast MMX (not accurate)
+  GST_IDCT_MMX32,           // accurate MMX
+} GstIDCTMethod;
+
+typedef struct _GstIDCT GstIDCT;
+typedef void (*GstIDCTFunction) (gshort *block);
+
+#define GST_IDCT_TRANSPOSE(idct) ((idct)->need_transpose)
+
+struct _GstIDCT {
+  /* private */
+  GstIDCTFunction convert;
+  GstIDCTFunction convert_sparse;
+  gboolean need_transpose;
+};
+
+
+GstIDCT *gst_idct_new(GstIDCTMethod method);
+#define gst_idct_convert(idct, blocks) (idct)->convert((blocks))
+#define gst_idct_convert_sparse(idct, blocks) (idct)->convert_sparse((blocks))
+void gst_idct_destroy(GstIDCT *idct);
+
+#endif /* __GST_IDCT_H__ */

libs/idct/ieeetest.c

@@ -0,0 +1,339 @@
+/*
+ * ieeetest.c --- test IDCT code against the IEEE Std 1180-1990 spec
+ *
+ * Note that this does only one pass of the test.
+ * Six invocations of ieeetest are needed to complete the entire spec.
+ * The shell script "doieee" performs the complete test.
+ *
+ * Written by Tom Lane (tgl@cs.cmu.edu).
+ * Released to public domain 11/22/93.
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <math.h>
+
+#include <gst/gst.h>
+#include "gstidct.h"
+#include "dct.h"
+
+
+/* prototypes */
+
+void usage (char *msg);
+long ieeerand (long L, long H);
+void dct_init(void);
+void ref_fdct(DCTELEM block[8][8]);
+void ref_idct(DCTELEM block[8][8]);
+
+/* error stat accumulators -- assume initialized to 0 */
+
+long sumerrs[DCTSIZE2];
+long sumsqerrs[DCTSIZE2];
+int maxerr[DCTSIZE2];
+
+
+char * meets (double val, double limit)
+{
+  return ((fabs(val) <= limit) ? "meets" : "FAILS");
+}
+
+__inline__ void read_tsc(guint64 *dst) {
+  __asm__ __volatile__
+    ("rdtsc"
+     : "=a" (*(guint32 *)dst), "=d" (*(((guint32 *)dst) + 1))
+     :
+     : "eax", "edx");
+}
+
+
+int
+main(int argc, char **argv)
+{
+  long minpix, maxpix, sign;
+  long curiter, niters;
+  int i, j;
+  double max, total;
+  int method;
+  DCTELEM   block[DCTSIZE2];	/* random source data */
+  DCTELEM   refcoefs[DCTSIZE2]; /* coefs from reference FDCT */
+  DCTELEM   refout[DCTSIZE2];	/* output from reference IDCT */
+  DCTELEM   testout[DCTSIZE2]; /* output from test IDCT */
+  GstIDCT   *idct;
+  guint64  tscstart, tscmin = ~0, tscmax = 0;
+  guint64  tscstop;
+
+  /* Argument parsing --- not very bulletproof at all */
+
+  if (argc != 6) usage(NULL);
+
+  method = atoi(argv[1]);
+  minpix = atoi(argv[2]);
+  maxpix = atoi(argv[3]);
+  sign   = atoi(argv[4]);
+  niters = atol(argv[5]);
+
+  gst_library_load("gstidct");
+
+  idct = gst_idct_new(method);
+
+  dct_init();
+
+  /* Loop once per generated random-data block */
+
+  for (curiter = 0; curiter < niters; curiter++) {
+
+    /* generate a pseudo-random block of data */
+    for (i = 0; i < DCTSIZE2; i++)
+      block[i] = (DCTELEM) (ieeerand(-minpix,maxpix) * sign);
+
+    /* perform reference FDCT */
+    memcpy(refcoefs, block, sizeof(DCTELEM)*DCTSIZE2);
+    ref_fdct(refcoefs);
+    /* clip */
+    for (i = 0; i < DCTSIZE2; i++) {
+      if (refcoefs[i] < -2048) refcoefs[i] = -2048;
+      else if (refcoefs[i] > 2047) refcoefs[i] = 2047;
+    }
+
+    /* perform reference IDCT */
+    memcpy(refout, refcoefs, sizeof(DCTELEM)*DCTSIZE2);
+    ref_idct(refout);
+    /* clip */
+    for (i = 0; i < DCTSIZE2; i++) {
+      if (refout[i] < -256) refout[i] = -256;
+      else if (refout[i] > 255) refout[i] = 255;
+    }
+
+    /* perform test IDCT */
+	 if (GST_IDCT_TRANSPOSE(idct)) {
+      for (j = 0; j < DCTSIZE; j++) {
+        for (i = 0; i < DCTSIZE; i++) {
+		    testout[i*DCTSIZE+j] = refcoefs[j*DCTSIZE+i];
+		  }
+	   }
+	 }
+	 else {
+      memcpy(testout, refcoefs, sizeof(DCTELEM)*DCTSIZE2);
+	 }
+
+	 read_tsc(&tscstart);
+    gst_idct_convert(idct, testout);
+	 read_tsc(&tscstop);
+	 //printf("time %llu, %llu %lld\n", tscstart, tscstop, tscstop-tscstart);
+	 if (tscstop - tscstart < tscmin) tscmin = tscstop-tscstart;
+	 if (tscstop - tscstart > tscmax) tscmax = tscstop-tscstart;
+
+    /* clip */
+    for (i = 0; i < DCTSIZE2; i++) {
+      if (testout[i] < -256) testout[i] = -256;
+      else if (testout[i] > 255) testout[i] = 255;
+    }
+
+    /* accumulate error stats */
+    for (i = 0; i < DCTSIZE2; i++) {
+      register int err = testout[i] - refout[i];
+      sumerrs[i] += err;
+      sumsqerrs[i] += err * err;
+      if (err < 0) err = -err;
+      if (maxerr[i] < err) maxerr[i] = err;
+    }
+
+    if (curiter % 100 == 99) {
+      fprintf(stderr, ".");
+      fflush(stderr);
+    }
+  }
+  fprintf(stderr, "\n");
+
+  /* print results */
+
+  printf("IEEE test conditions: -L = %ld, +H = %ld, sign = %ld, #iters = %ld\n",
+	 minpix, maxpix, sign, niters);
+
+  printf("Speed, min time %lld, max %lld\n", tscmin, tscmax);
+
+  printf("Peak absolute values of errors:\n");
+  for (i = 0, j = 0; i < DCTSIZE2; i++) {
+    if (j < maxerr[i]) j = maxerr[i];
+    printf("%4d", maxerr[i]);
+    if ((i%DCTSIZE) == DCTSIZE-1) printf("\n");
+  }
+  printf("Worst peak error = %d  (%s spec limit 1)\n\n", j,
+	 meets((double) j, 1.0));
+
+  printf("Mean square errors:\n");
+  max = total = 0.0;
+  for (i = 0; i < DCTSIZE2; i++) {
+    double err = (double) sumsqerrs[i]  / ((double) niters);
+    total += (double) sumsqerrs[i];
+    if (max < err) max = err;
+    printf(" %8.4f", err);
+    if ((i%DCTSIZE) == DCTSIZE-1) printf("\n");
+  }
+  printf("Worst pmse = %.6f  (%s spec limit 0.06)\n", max, meets(max, 0.06));
+  total /= (double) (64*niters);
+  printf("Overall mse = %.6f  (%s spec limit 0.02)\n\n", total,
+	 meets(total, 0.02));
+
+  printf("Mean errors:\n");
+  max = total = 0.0;
+  for (i = 0; i < DCTSIZE2; i++) {
+    double err = (double) sumerrs[i]  / ((double) niters);
+    total += (double) sumerrs[i];
+    printf(" %8.4f", err);
+    if (err < 0.0) err = -err;
+    if (max < err) max = err;
+    if ((i%DCTSIZE) == DCTSIZE-1) printf("\n");
+  }
+  printf("Worst mean error = %.6f  (%s spec limit 0.015)\n", max,
+	 meets(max, 0.015));
+  total /= (double) (64*niters);
+  printf("Overall mean error = %.6f  (%s spec limit 0.0015)\n\n", total,
+	 meets(total, 0.0015));
+
+  /* test for 0 input giving 0 output */
+  memset(testout, 0, sizeof(DCTELEM)*DCTSIZE2);
+  gst_idct_convert(idct, testout);
+  for (i = 0, j=0; i < DCTSIZE2; i++) {
+    if (testout[i]) {
+      printf("Position %d of IDCT(0) = %d (FAILS)\n", i, testout[i]);
+      j++;
+    }
+  }
+  printf("%d elements of IDCT(0) were not zero\n\n\n", j);
+
+  exit(0);
+  return 0;
+}
+
+
+void usage (char *msg)
+{
+  if (msg != NULL)
+    fprintf(stderr, "\nerror: %s\n", msg);
+
+  fprintf(stderr, "\n");
+  fprintf(stderr, "usage: ieeetest minpix maxpix sign niters\n");
+  fprintf(stderr, "\n");
+  fprintf(stderr, "  test = 1 - 5\n");
+  fprintf(stderr, "  minpix = -L value per IEEE spec\n");
+  fprintf(stderr, "  maxpix =  H value per IEEE spec\n");
+  fprintf(stderr, "  sign = +1 for normal, -1 to run negated test\n");
+  fprintf(stderr, "  niters = # iterations (10000 for full test)\n");
+  fprintf(stderr, "\n");
+
+  exit(1);
+}
+
+
+/* Pseudo-random generator specified by IEEE 1180 */
+
+long ieeerand (long L, long H)
+{
+  static long randx = 1;
+  static double z = (double) 0x7fffffff;
+
+  long i,j;
+  double x;
+
+  randx = (randx * 1103515245) + 12345;
+  i = randx & 0x7ffffffe;
+  x = ((double) i) / z;
+  x *= (L+H+1);
+  j = x;
+  return j-L;
+}
+
+
+/* Reference double-precision FDCT and IDCT */
+
+
+/* The cosine lookup table */
+/* coslu[a][b] = C(b)/2 * cos[(2a+1)b*pi/16] */
+double coslu[8][8];
+
+
+/* Routine to initialise the cosine lookup table */
+void dct_init(void)
+{
+  int a,b;
+  double tmp;
+
+  for(a=0;a<8;a++)
+    for(b=0;b<8;b++) {
+      tmp = cos((double)((a+a+1)*b) * (3.14159265358979323846 / 16.0));
+      if(b==0)
+	tmp /= sqrt(2.0);
+      coslu[a][b] = tmp * 0.5;
+    }
+}
+
+
+void ref_fdct (DCTELEM block[8][8])
+{
+  int x,y,u,v;
+  double tmp, tmp2;
+  double res[8][8];
+
+  for (v=0; v<8; v++) {
+    for (u=0; u<8; u++) {
+      tmp = 0.0;
+      for (y=0; y<8; y++) {
+	tmp2 = 0.0;
+	for (x=0; x<8; x++) {
+	  tmp2 += (double) block[y][x] * coslu[x][u];
+	}
+	tmp += coslu[y][v] * tmp2;
+      }
+      res[v][u] = tmp;
+    }
+  }
+
+  for (v=0; v<8; v++) {
+    for (u=0; u<8; u++) {
+      tmp = res[v][u];
+      if (tmp < 0.0) {
+	x = - ((int) (0.5 - tmp));
+      } else {
+	x = (int) (tmp + 0.5);
+      }
+      block[v][u] = (DCTELEM) x;
+    }
+  }
+}
+
+
+void ref_idct (DCTELEM block[8][8])
+{
+  int x,y,u,v;
+  double tmp, tmp2;
+  double res[8][8];
+
+  for (y=0; y<8; y++) {
+    for (x=0; x<8; x++) {
+      tmp = 0.0;
+      for (v=0; v<8; v++) {
+	tmp2 = 0.0;
+	for (u=0; u<8; u++) {
+	  tmp2 += (double) block[v][u] * coslu[x][u];
+	}
+	tmp += coslu[y][v] * tmp2;
+      }
+      res[y][x] = tmp;
+    }
+  }
+
+  for (v=0; v<8; v++) {
+    for (u=0; u<8; u++) {
+      tmp = res[v][u];
+      if (tmp < 0.0) {
+	x = - ((int) (0.5 - tmp));
+      } else {
+	x = (int) (tmp + 0.5);
+      }
+      block[v][u] = (DCTELEM) x;
+    }
+  }
+}

libs/idct/intidct.c

@@ -0,0 +1,391 @@
+/*
+ * jrevdct.c
+ *
+ * Copyright (C) 1991, 1992, Thomas G. Lane.
+ * This file is part of the Independent JPEG Group's software.
+ * For conditions of distribution and use, see the accompanying README file.
+ *
+ * This file contains the basic inverse-DCT transformation subroutine.
+ *
+ * This implementation is based on an algorithm described in
+ *   C. Loeffler, A. Ligtenberg and G. Moschytz, "Practical Fast 1-D DCT
+ *   Algorithms with 11 Multiplications", Proc. Int'l. Conf. on Acoustics,
+ *   Speech, and Signal Processing 1989 (ICASSP '89), pp. 988-991.
+ * The primary algorithm described there uses 11 multiplies and 29 adds.
+ * We use their alternate method with 12 multiplies and 32 adds.
+ * The advantage of this method is that no data path contains more than one
+ * multiplication; this allows a very simple and accurate implementation in
+ * scaled fixed-point arithmetic, with a minimal number of shifts.
+ */
+
+#include "dct.h"
+
+/* We assume that right shift corresponds to signed division by 2 with
+ * rounding towards minus infinity.  This is correct for typical "arithmetic
+ * shift" instructions that shift in copies of the sign bit.  But some
+ * C compilers implement >> with an unsigned shift.  For these machines you
+ * must define RIGHT_SHIFT_IS_UNSIGNED.
+ * RIGHT_SHIFT provides a proper signed right shift of an INT32 quantity.
+ * It is only applied with constant shift counts.  SHIFT_TEMPS must be
+ * included in the variables of any routine using RIGHT_SHIFT.
+ */
+
+#ifdef RIGHT_SHIFT_IS_UNSIGNED
+#define SHIFT_TEMPS	INT32 shift_temp;
+#define RIGHT_SHIFT(x,shft)  \
+	((shift_temp = (x)) < 0 ? \
+	 (shift_temp >> (shft)) | ((~((INT32) 0)) << (32-(shft))) : \
+	 (shift_temp >> (shft)))
+#else
+#define SHIFT_TEMPS
+#define RIGHT_SHIFT(x,shft)	((x) >> (shft))
+#endif
+
+
+/*
+ * This routine is specialized to the case DCTSIZE = 8.
+ */
+
+#if DCTSIZE != 8
+  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
+#endif
+
+
+/*
+ * A 2-D IDCT can be done by 1-D IDCT on each row followed by 1-D IDCT
+ * on each column.  Direct algorithms are also available, but they are
+ * much more complex and seem not to be any faster when reduced to code.
+ *
+ * The poop on this scaling stuff is as follows:
+ *
+ * Each 1-D IDCT step produces outputs which are a factor of sqrt(N)
+ * larger than the true IDCT outputs.  The final outputs are therefore
+ * a factor of N larger than desired; since N=8 this can be cured by
+ * a simple right shift at the end of the algorithm.  The advantage of
+ * this arrangement is that we save two multiplications per 1-D IDCT,
+ * because the y0 and y4 inputs need not be divided by sqrt(N).
+ *
+ * We have to do addition and subtraction of the integer inputs, which
+ * is no problem, and multiplication by fractional constants, which is
+ * a problem to do in integer arithmetic.  We multiply all the constants
+ * by CONST_SCALE and convert them to integer constants (thus retaining
+ * CONST_BITS bits of precision in the constants).  After doing a
+ * multiplication we have to divide the product by CONST_SCALE, with proper
+ * rounding, to produce the correct output.  This division can be done
+ * cheaply as a right shift of CONST_BITS bits.  We postpone shifting
+ * as long as possible so that partial sums can be added together with
+ * full fractional precision.
+ *
+ * The outputs of the first pass are scaled up by PASS1_BITS bits so that
+ * they are represented to better-than-integral precision.  These outputs
+ * require BITS_IN_JSAMPLE + PASS1_BITS + 3 bits; this fits in a 16-bit word
+ * with the recommended scaling.  (To scale up 12-bit sample data further, an
+ * intermediate INT32 array would be needed.)
+ *
+ * To avoid overflow of the 32-bit intermediate results in pass 2, we must
+ * have BITS_IN_JSAMPLE + CONST_BITS + PASS1_BITS <= 26.  Error analysis
+ * shows that the values given below are the most effective.
+ */
+
+#ifdef EIGHT_BIT_SAMPLES
+#define CONST_BITS  13
+#define PASS1_BITS  2
+#else
+#define CONST_BITS  13
+#define PASS1_BITS  1		/* lose a little precision to avoid overflow */
+#endif
+
+#define ONE	((INT32) 1)
+
+#define CONST_SCALE (ONE << CONST_BITS)
+
+/* Convert a positive real constant to an integer scaled by CONST_SCALE. */
+
+#define FIX(x)	((INT32) ((x) * CONST_SCALE + 0.5))
+
+/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
+ * causing a lot of useless floating-point operations at run time.
+ * To get around this we use the following pre-calculated constants.
+ * If you change CONST_BITS you may want to add appropriate values.
+ * (With a reasonable C compiler, you can just rely on the FIX() macro...)
+ */
+
+#if CONST_BITS == 13
+#define FIX_0_298631336  ((INT32)  2446)	/* FIX(0.298631336) */
+#define FIX_0_390180644  ((INT32)  3196)	/* FIX(0.390180644) */
+#define FIX_0_541196100  ((INT32)  4433)	/* FIX(0.541196100) */
+#define FIX_0_765366865  ((INT32)  6270)	/* FIX(0.765366865) */
+#define FIX_0_899976223  ((INT32)  7373)	/* FIX(0.899976223) */
+#define FIX_1_175875602  ((INT32)  9633)	/* FIX(1.175875602) */
+#define FIX_1_501321110  ((INT32)  12299)	/* FIX(1.501321110) */
+#define FIX_1_847759065  ((INT32)  15137)	/* FIX(1.847759065) */
+#define FIX_1_961570560  ((INT32)  16069)	/* FIX(1.961570560) */
+#define FIX_2_053119869  ((INT32)  16819)	/* FIX(2.053119869) */
+#define FIX_2_562915447  ((INT32)  20995)	/* FIX(2.562915447) */
+#define FIX_3_072711026  ((INT32)  25172)	/* FIX(3.072711026) */
+#else
+#define FIX_0_298631336  FIX(0.298631336)
+#define FIX_0_390180644  FIX(0.390180644)
+#define FIX_0_541196100  FIX(0.541196100)
+#define FIX_0_765366865  FIX(0.765366865)
+#define FIX_0_899976223  FIX(0.899976223)
+#define FIX_1_175875602  FIX(1.175875602)
+#define FIX_1_501321110  FIX(1.501321110)
+#define FIX_1_847759065  FIX(1.847759065)
+#define FIX_1_961570560  FIX(1.961570560)
+#define FIX_2_053119869  FIX(2.053119869)
+#define FIX_2_562915447  FIX(2.562915447)
+#define FIX_3_072711026  FIX(3.072711026)
+#endif
+
+
+/* Descale and correctly round an INT32 value that's scaled by N bits.
+ * We assume RIGHT_SHIFT rounds towards minus infinity, so adding
+ * the fudge factor is correct for either sign of X.
+ */
+
+#define DESCALE(x,n)  RIGHT_SHIFT((x) + (ONE << ((n)-1)), n)
+
+/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
+ * For 8-bit samples with the recommended scaling, all the variable
+ * and constant values involved are no more than 16 bits wide, so a
+ * 16x16->32 bit multiply can be used instead of a full 32x32 multiply;
+ * this provides a useful speedup on many machines.
+ * There is no way to specify a 16x16->32 multiply in portable C, but
+ * some C compilers will do the right thing if you provide the correct
+ * combination of casts.
+ * NB: for 12-bit samples, a full 32-bit multiplication will be needed.
+ */
+
+#ifdef EIGHT_BIT_SAMPLES
+#ifdef SHORTxSHORT_32		/* may work if 'int' is 32 bits */
+#define MULTIPLY(var,const)  (((INT16) (var)) * ((INT16) (const)))
+#endif
+#ifdef SHORTxLCONST_32		/* known to work with Microsoft C 6.0 */
+#define MULTIPLY(var,const)  (((INT16) (var)) * ((INT32) (const)))
+#endif
+#endif
+
+#ifndef MULTIPLY		/* default definition */
+#define MULTIPLY(var,const)  ((var) * (const))
+#endif
+
+
+/*
+ * Perform the inverse DCT on one block of coefficients.
+ */
+
+void
+gst_idct_int_idct (DCTBLOCK data)
+{
+  INT32 tmp0, tmp1, tmp2, tmp3;
+  INT32 tmp10, tmp11, tmp12, tmp13;
+  INT32 z1, z2, z3, z4, z5;
+  register DCTELEM *dataptr;
+  int rowctr;
+  SHIFT_TEMPS
+
+  /* Pass 1: process rows. */
+  /* Note results are scaled up by sqrt(8) compared to a true IDCT; */
+  /* furthermore, we scale the results by 2**PASS1_BITS. */
+
+  dataptr = data;
+  for (rowctr = DCTSIZE-1; rowctr >= 0; rowctr--) {
+    /* Due to quantization, we will usually find that many of the input
+     * coefficients are zero, especially the AC terms.  We can exploit this
+     * by short-circuiting the IDCT calculation for any row in which all
+     * the AC terms are zero.  In that case each output is equal to the
+     * DC coefficient (with scale factor as needed).
+     * With typical images and quantization tables, half or more of the
+     * row DCT calculations can be simplified this way.
+     */
+
+    if ((dataptr[1] | dataptr[2] | dataptr[3] | dataptr[4] |
+	 dataptr[5] | dataptr[6] | dataptr[7]) == 0) {
+      /* AC terms all zero */
+      DCTELEM dcval = (DCTELEM) (dataptr[0] << PASS1_BITS);
+      
+      dataptr[0] = dcval;
+      dataptr[1] = dcval;
+      dataptr[2] = dcval;
+      dataptr[3] = dcval;
+      dataptr[4] = dcval;
+      dataptr[5] = dcval;
+      dataptr[6] = dcval;
+      dataptr[7] = dcval;
+      
+      dataptr += DCTSIZE;	/* advance pointer to next row */
+      continue;
+    }
+
+    /* Even part: reverse the even part of the forward DCT. */
+    /* The rotator is sqrt(2)*c(-6). */
+
+    z2 = (INT32) dataptr[2];
+    z3 = (INT32) dataptr[6];
+
+    z1 = MULTIPLY(z2 + z3, FIX_0_541196100);
+    tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065);
+    tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865);
+
+    tmp0 = ((INT32) dataptr[0] + (INT32) dataptr[4]) << CONST_BITS;
+    tmp1 = ((INT32) dataptr[0] - (INT32) dataptr[4]) << CONST_BITS;
+
+    tmp10 = tmp0 + tmp3;
+    tmp13 = tmp0 - tmp3;
+    tmp11 = tmp1 + tmp2;
+    tmp12 = tmp1 - tmp2;
+    
+    /* Odd part per figure 8; the matrix is unitary and hence its
+     * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
+     */
+
+    tmp0 = (INT32) dataptr[7];
+    tmp1 = (INT32) dataptr[5];
+    tmp2 = (INT32) dataptr[3];
+    tmp3 = (INT32) dataptr[1];
+
+    z1 = tmp0 + tmp3;
+    z2 = tmp1 + tmp2;
+    z3 = tmp0 + tmp2;
+    z4 = tmp1 + tmp3;
+    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
+    
+    tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
+    tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
+    tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
+    tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
+    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
+    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
+    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
+    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
+    
+    z3 += z5;
+    z4 += z5;
+    
+    tmp0 += z1 + z3;
+    tmp1 += z2 + z4;
+    tmp2 += z2 + z3;
+    tmp3 += z1 + z4;
+
+    /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
+
+    dataptr[0] = (DCTELEM) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
+    dataptr[7] = (DCTELEM) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
+    dataptr[1] = (DCTELEM) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
+    dataptr[6] = (DCTELEM) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
+    dataptr[2] = (DCTELEM) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
+    dataptr[5] = (DCTELEM) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
+    dataptr[3] = (DCTELEM) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
+    dataptr[4] = (DCTELEM) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS);
+
+    dataptr += DCTSIZE;		/* advance pointer to next row */
+  }
+
+  /* Pass 2: process columns. */
+  /* Note that we must descale the results by a factor of 8 == 2**3, */
+  /* and also undo the PASS1_BITS scaling. */
+
+  dataptr = data;
+  for (rowctr = DCTSIZE-1; rowctr >= 0; rowctr--) {
+    /* Columns of zeroes can be exploited in the same way as we did with rows.
+     * However, the row calculation has created many nonzero AC terms, so the
+     * simplification applies less often (typically 5% to 10% of the time).
+     * On machines with very fast multiplication, it's possible that the
+     * test takes more time than it's worth.  In that case this section
+     * may be commented out.
+     */
+
+#ifndef NO_ZERO_COLUMN_TEST
+    if ((dataptr[DCTSIZE*1] | dataptr[DCTSIZE*2] | dataptr[DCTSIZE*3] |
+	 dataptr[DCTSIZE*4] | dataptr[DCTSIZE*5] | dataptr[DCTSIZE*6] |
+	 dataptr[DCTSIZE*7]) == 0) {
+      /* AC terms all zero */
+      DCTELEM dcval = (DCTELEM) DESCALE((INT32) dataptr[0], PASS1_BITS+3);
+      
+      dataptr[DCTSIZE*0] = dcval;
+      dataptr[DCTSIZE*1] = dcval;
+      dataptr[DCTSIZE*2] = dcval;
+      dataptr[DCTSIZE*3] = dcval;
+      dataptr[DCTSIZE*4] = dcval;
+      dataptr[DCTSIZE*5] = dcval;
+      dataptr[DCTSIZE*6] = dcval;
+      dataptr[DCTSIZE*7] = dcval;
+      
+      dataptr++;		/* advance pointer to next column */
+      continue;
+    }
+#endif
+
+    /* Even part: reverse the even part of the forward DCT. */
+    /* The rotator is sqrt(2)*c(-6). */
+
+    z2 = (INT32) dataptr[DCTSIZE*2];
+    z3 = (INT32) dataptr[DCTSIZE*6];
+
+    z1 = MULTIPLY(z2 + z3, FIX_0_541196100);
+    tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065);
+    tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865);
+
+    tmp0 = ((INT32) dataptr[DCTSIZE*0] + (INT32) dataptr[DCTSIZE*4]) << CONST_BITS;
+    tmp1 = ((INT32) dataptr[DCTSIZE*0] - (INT32) dataptr[DCTSIZE*4]) << CONST_BITS;
+
+    tmp10 = tmp0 + tmp3;
+    tmp13 = tmp0 - tmp3;
+    tmp11 = tmp1 + tmp2;
+    tmp12 = tmp1 - tmp2;
+    
+    /* Odd part per figure 8; the matrix is unitary and hence its
+     * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
+     */
+
+    tmp0 = (INT32) dataptr[DCTSIZE*7];
+    tmp1 = (INT32) dataptr[DCTSIZE*5];
+    tmp2 = (INT32) dataptr[DCTSIZE*3];
+    tmp3 = (INT32) dataptr[DCTSIZE*1];
+
+    z1 = tmp0 + tmp3;
+    z2 = tmp1 + tmp2;
+    z3 = tmp0 + tmp2;
+    z4 = tmp1 + tmp3;
+    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
+    
+    tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
+    tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
+    tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
+    tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
+    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
+    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
+    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
+    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
+    
+    z3 += z5;
+    z4 += z5;
+    
+    tmp0 += z1 + z3;
+    tmp1 += z2 + z4;
+    tmp2 += z2 + z3;
+    tmp3 += z1 + z4;
+
+    /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
+
+    dataptr[DCTSIZE*0] = (DCTELEM) DESCALE(tmp10 + tmp3,
+					   CONST_BITS+PASS1_BITS+3);
+    dataptr[DCTSIZE*7] = (DCTELEM) DESCALE(tmp10 - tmp3,
+					   CONST_BITS+PASS1_BITS+3);
+    dataptr[DCTSIZE*1] = (DCTELEM) DESCALE(tmp11 + tmp2,
+					   CONST_BITS+PASS1_BITS+3);
+    dataptr[DCTSIZE*6] = (DCTELEM) DESCALE(tmp11 - tmp2,
+					   CONST_BITS+PASS1_BITS+3);
+    dataptr[DCTSIZE*2] = (DCTELEM) DESCALE(tmp12 + tmp1,
+					   CONST_BITS+PASS1_BITS+3);
+    dataptr[DCTSIZE*5] = (DCTELEM) DESCALE(tmp12 - tmp1,
+					   CONST_BITS+PASS1_BITS+3);
+    dataptr[DCTSIZE*3] = (DCTELEM) DESCALE(tmp13 + tmp0,
+					   CONST_BITS+PASS1_BITS+3);
+    dataptr[DCTSIZE*4] = (DCTELEM) DESCALE(tmp13 - tmp0,
+					   CONST_BITS+PASS1_BITS+3);
+    
+    dataptr++;			/* advance pointer to next column */
+  }
+}

libs/idct/mmx32idct.c

@@ -0,0 +1,783 @@
+/* 
+ *  idctmmx32.cpp 
+ *
+ *	Copyright (C) Alberto Vigata - January 2000 - ultraflask@yahoo.com
+ *
+ *  This file is part of FlasKMPEG, a free MPEG to MPEG/AVI converter
+ *	
+ *  FlasKMPEG 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, or (at your option)
+ *  any later version.
+ *   
+ *  FlasKMPEG 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
+ *  GNU General Public License for more details.
+ *   
+ *  You should have received a copy of the GNU General Public License
+ *  along with GNU Make; see the file COPYING.  If not, write to
+ *  the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. 
+ *
+ */
+
+
+// MMX32 iDCT algorithm  (IEEE-1180 compliant) :: idct_mmx32()
+//
+// MPEG2AVI
+// --------
+//  v0.16B33 initial release
+//
+// This was one of the harder pieces of work to code.
+// Intel's app-note focuses on the numerical issues of the algorithm, but
+// assumes the programmer is familiar with IDCT mathematics, leaving the
+// form of the complete function up to the programmer's imagination.
+//
+//  ALGORITHM OVERVIEW
+//  ------------------
+// I played around with the code for quite a few hours.  I came up
+// with *A* working IDCT algorithm, however I'm not sure whether my routine
+// is "the correct one."  But rest assured, my code passes all six IEEE 
+// accuracy tests with plenty of margin.
+//
+//   My IDCT algorithm consists of 4 steps:
+//
+//   1) IDCT-row transformation (using the IDCT-row function) on all 8 rows
+//      This yields an intermediate 8x8 matrix.
+//
+//   2) intermediate matrix transpose (mandatory)
+//
+//   3) IDCT-row transformation (2nd time) on all 8 rows of the intermediate
+//      matrix.  The output is the final-result, in transposed form.
+//
+//   4) post-transformation matrix transpose 
+//      (not necessary if the input-data is already transposed, this could
+//       be done during the MPEG "zig-zag" scan, but since my algorithm
+//       requires at least one transpose operation, why not re-use the
+//       transpose-code.)
+//
+//   Although the (1st) and (3rd) steps use the SAME row-transform operation,
+//   the (3rd) step uses different shift&round constants (explained later.)
+//
+//   Also note that the intermediate transpose (2) would not be neccessary,
+//   if the subsequent operation were a iDCT-column transformation.  Since
+//   we only have the iDCT-row transform, we transpose the intermediate
+//   matrix and use the iDCT-row transform a 2nd time.
+//
+//   I had to change some constants/variables for my method to work :
+//
+//      As given by Intel, the #defines for SHIFT_INV_COL and RND_INV_COL are
+//      wrong.  Not surprising since I'm not using a true column-transform 
+//      operation, but the row-transform operation (as mentioned earlier.)
+//      round_inv_col[], which is given as "4 short" values, should have the
+//      same dimensions as round_inv_row[].  The corrected variables are 
+//      shown.
+//
+//      Intel's code defines a different table for each each row operation.
+//      The tables given are 0/4, 1/7, 2/6, and 5/3.  My code only uses row#0.
+//      Using the other rows messes up the overall transform.
+//
+//   IMPLEMENTATION DETAILs
+//   ----------------------
+// 
+//   I divided the algorithm's work into two subroutines,
+//    1) idct_mmx32_rows() - transforms 8 rows, then transpose
+//    2) idct_mmx32_cols() - transforms 8 rows, then transpose
+//       yields final result ("drop-in" direct replacement for INT32 IDCT)
+//
+//   The 2nd function is a clone of the 1st, with changes made only to the
+//   shift&rounding instructions.
+//
+//      In the 1st function (rows), the shift & round instructions use 
+//       SHIFT_INV_ROW & round_inv_row[] (renamed to r_inv_row[])
+//
+//      In the 2nd function (cols)-> r_inv_col[], and
+//       SHIFT_INV_COL & round_inv_col[] (renamed to r_inv_col[])
+//
+//   Each function contains an integrated transpose-operator, which comes
+//   AFTER the primary transformation operation.  In the future, I'll optimize
+//   the code to do more of the transpose-work "in-place".  Right now, I've
+//   left the code as two subroutines and a main calling function, so other
+//   people can read the code more easily.
+//
+//   liaor@umcc.ais.org  http://members.tripod.com/~liaor
+//  
+
+
+//;=============================================================================
+//;
+//;  AP-922   http://developer.intel.com/vtune/cbts/strmsimd
+//; These examples contain code fragments for first stage iDCT 8x8
+//; (for rows) and first stage DCT 8x8 (for columns)
+//;
+//;=============================================================================
+/*
+mword typedef qword
+qword ptr equ mword ptr */
+
+#include <mmx.h>
+
+#define BITS_INV_ACC	4	//; 4 or 5 for IEEE
+	// 5 yields higher accuracy, but lessens dynamic range on the input matrix
+#define SHIFT_INV_ROW	(16 - BITS_INV_ACC)
+#define SHIFT_INV_COL	(1 + BITS_INV_ACC +14 )  // changed from Intel's val)
+//#define SHIFT_INV_COL	(1 + BITS_INV_ACC )
+
+#define RND_INV_ROW		(1 << (SHIFT_INV_ROW-1))
+#define RND_INV_COL		(1 << (SHIFT_INV_COL-1)) 
+#define RND_INV_CORR	(RND_INV_COL - 1)		//; correction -1.0 and round
+//#define RND_INV_ROW		(1024 * (6 - BITS_INV_ACC)) //; 1 << (SHIFT_INV_ROW-1)
+//#define RND_INV_COL		(16 * (BITS_INV_ACC - 3)) //; 1 << (SHIFT_INV_COL-1)
+
+
+//.data
+//Align 16
+const static long r_inv_row[2] = { RND_INV_ROW, RND_INV_ROW};
+const static long r_inv_col[2] = {RND_INV_COL, RND_INV_COL};
+const static long r_inv_corr[2] = {RND_INV_CORR, RND_INV_CORR };
+
+//const static short r_inv_col[4] = 
+//	{RND_INV_COL, RND_INV_COL, RND_INV_COL, RND_INV_COL};
+//const static short r_inv_corr[4] =
+//	{RND_INV_CORR, RND_INV_CORR, RND_INV_CORR, RND_INV_CORR};
+
+/* constants for the forward DCT
+
+//#define BITS_FRW_ACC	3 //; 2 or 3 for accuracy
+//#define SHIFT_FRW_COL	BITS_FRW_ACC
+//#define SHIFT_FRW_ROW	(BITS_FRW_ACC + 17)
+//#define RND_FRW_ROW		(262144 * (BITS_FRW_ACC - 1)) //; 1 << (SHIFT_FRW_ROW-1)
+
+const static __int64 one_corr = 0x0001000100010001;
+const static long r_frw_row[2] = {RND_FRW_ROW, RND_FRW_ROW };
+
+//const static short tg_1_16[4] = {13036, 13036, 13036, 13036 }; //tg * (2<<16) + 0.5
+//const static short tg_2_16[4] = {27146, 27146, 27146, 27146 }; //tg * (2<<16) + 0.5
+//const static short tg_3_16[4] = {-21746, -21746, -21746, -21746 }; //tg * (2<<16) + 0.5
+//const static short cos_4_16[4] = {-19195, -19195, -19195, -19195 }; //cos * (2<<16) + 0.5
+//const static short ocos_4_16[4] = {23170, 23170, 23170, 23170 }; //cos * (2<<15) + 0.5
+
+//concatenated table, for forward DCT transformation
+const static short tg_all_16[] = {
+	13036, 13036, 13036, 13036,		// tg * (2<<16) + 0.5
+	27146, 27146, 27146, 27146,		//tg * (2<<16) + 0.5
+	-21746, -21746, -21746, -21746,	// tg * (2<<16) + 0.5
+	-19195, -19195, -19195, -19195,	//cos * (2<<16) + 0.5
+	23170, 23170, 23170, 23170 };	//cos * (2<<15) + 0.5
+
+#define tg_1_16 (tg_all_16 + 0)
+#define tg_2_16 (tg_all_16 + 8)
+#define tg_3_16 (tg_all_16 + 16)
+#define cos_4_16 (tg_all_16 + 24)
+#define ocos_4_16 (tg_all_16 + 32)
+*/
+/*
+;=============================================================================
+;
+; The first stage iDCT 8x8 - inverse DCTs of rows
+;
+;-----------------------------------------------------------------------------
+; The 8-point inverse DCT direct algorithm
+;-----------------------------------------------------------------------------
+;
+; static const short w[32] = {
+; FIX(cos_4_16), FIX(cos_2_16), FIX(cos_4_16), FIX(cos_6_16),
+; FIX(cos_4_16), FIX(cos_6_16), -FIX(cos_4_16), -FIX(cos_2_16),
+; FIX(cos_4_16), -FIX(cos_6_16), -FIX(cos_4_16), FIX(cos_2_16),
+; FIX(cos_4_16), -FIX(cos_2_16), FIX(cos_4_16), -FIX(cos_6_16),
+; FIX(cos_1_16), FIX(cos_3_16), FIX(cos_5_16), FIX(cos_7_16),
+; FIX(cos_3_16), -FIX(cos_7_16), -FIX(cos_1_16), -FIX(cos_5_16),
+; FIX(cos_5_16), -FIX(cos_1_16), FIX(cos_7_16), FIX(cos_3_16),
+; FIX(cos_7_16), -FIX(cos_5_16), FIX(cos_3_16), -FIX(cos_1_16) };
+;
+; #define DCT_8_INV_ROW(x, y)
+
+;{
+; int a0, a1, a2, a3, b0, b1, b2, b3;
+;
+; a0 =x[0]*w[0]+x[2]*w[1]+x[4]*w[2]+x[6]*w[3];
+; a1 =x[0]*w[4]+x[2]*w[5]+x[4]*w[6]+x[6]*w[7];
+; a2 = x[0] * w[ 8] + x[2] * w[ 9] + x[4] * w[10] + x[6] * w[11];
+; a3 = x[0] * w[12] + x[2] * w[13] + x[4] * w[14] + x[6] * w[15];
+; b0 = x[1] * w[16] + x[3] * w[17] + x[5] * w[18] + x[7] * w[19];
+; b1 = x[1] * w[20] + x[3] * w[21] + x[5] * w[22] + x[7] * w[23];
+; b2 = x[1] * w[24] + x[3] * w[25] + x[5] * w[26] + x[7] * w[27];
+; b3 = x[1] * w[28] + x[3] * w[29] + x[5] * w[30] + x[7] * w[31];
+;
+; y[0] = SHIFT_ROUND ( a0 + b0 );
+; y[1] = SHIFT_ROUND ( a1 + b1 );
+; y[2] = SHIFT_ROUND ( a2 + b2 );
+; y[3] = SHIFT_ROUND ( a3 + b3 );
+; y[4] = SHIFT_ROUND ( a3 - b3 );
+; y[5] = SHIFT_ROUND ( a2 - b2 );
+; y[6] = SHIFT_ROUND ( a1 - b1 );
+; y[7] = SHIFT_ROUND ( a0 - b0 );
+;}
+;
+;-----------------------------------------------------------------------------
+;
+; In this implementation the outputs of the iDCT-1D are multiplied
+; for rows 0,4 - by cos_4_16,
+; for rows 1,7 - by cos_1_16,
+; for rows 2,6 - by cos_2_16,
+; for rows 3,5 - by cos_3_16
+; and are shifted to the left for better accuracy
+;
+; For the constants used,
+; FIX(float_const) = (short) (float_const * (1<<15) + 0.5)
+;
+;=============================================================================
+;=============================================================================
+IF _MMX ; MMX code
+;=============================================================================
+
+//; Table for rows 0,4 - constants are multiplied by cos_4_16
+const short tab_i_04[] = {
+	16384, 16384, 16384, -16384,	// ; movq-> w06 w04 w02 w00
+	21407, 8867, 8867, -21407,		// w07 w05 w03 w01
+	16384, -16384, 16384, 16384,	//; w14 w12 w10 w08
+	-8867, 21407, -21407, -8867,	//; w15 w13 w11 w09
+	22725, 12873, 19266, -22725,	//; w22 w20 w18 w16
+	19266, 4520, -4520, -12873,		//; w23 w21 w19 w17
+	12873, 4520, 4520, 19266,		//; w30 w28 w26 w24
+	-22725, 19266, -12873, -22725 };//w31 w29 w27 w25
+
+//; Table for rows 1,7 - constants are multiplied by cos_1_16
+const short tab_i_17[] = {
+	22725, 22725, 22725, -22725,	// ; movq-> w06 w04 w02 w00
+	29692, 12299, 12299, -29692,	//	; w07 w05 w03 w01
+	22725, -22725, 22725, 22725,	//; w14 w12 w10 w08
+	-12299, 29692, -29692, -12299,	//; w15 w13 w11 w09
+	31521, 17855, 26722, -31521,	//; w22 w20 w18 w16
+	26722, 6270, -6270, -17855,		//; w23 w21 w19 w17
+	17855, 6270, 6270, 26722,		//; w30 w28 w26 w24
+	-31521, 26722, -17855, -31521};	// w31 w29 w27 w25
+
+//; Table for rows 2,6 - constants are multiplied by cos_2_16
+const short tab_i_26[] = {
+	21407, 21407, 21407, -21407,	// ; movq-> w06 w04 w02 w00
+	27969, 11585, 11585, -27969,	// ; w07 w05 w03 w01
+	21407, -21407, 21407, 21407,	// ; w14 w12 w10 w08
+	-11585, 27969, -27969, -11585,	//  ;w15 w13 w11 w09
+	29692, 16819, 25172, -29692, 	// ;w22 w20 w18 w16
+	25172, 5906, -5906, -16819, 	// ;w23 w21 w19 w17
+	16819, 5906, 5906, 25172, 		// ;w30 w28 w26 w24
+	-29692, 25172, -16819, -29692};	//  ;w31 w29 w27 w25
+
+
+//; Table for rows 3,5 - constants are multiplied by cos_3_16
+const short tab_i_35[] = {
+	19266, 19266, 19266, -19266,	//; movq-> w06 w04 w02 w00
+	25172, 10426, 10426, -25172,	//; w07 w05 w03 w01
+	19266, -19266, 19266, 19266,	//; w14 w12 w10 w08
+	-10426, 25172, -25172, -10426,	//; w15 w13 w11 w09
+	26722, 15137, 22654, -26722,	//; w22 w20 w18 w16
+	22654, 5315, -5315, -15137,		//; w23 w21 w19 w17
+	15137, 5315, 5315, 22654,		//; w30 w28 w26 w24
+	-26722, 22654, -15137, -26722};	//; w31 w29 w27 w25
+*/
+
+// CONCATENATED TABLE, rows 0,1,2,3,4,5,6,7 (in order )
+//
+// In our implementation, however, we only use row0 !
+//
+static const short tab_i_01234567[] = {
+	//row0, this row is required
+	16384, 16384, 16384, -16384,	// ; movq-> w06 w04 w02 w00
+	21407, 8867, 8867, -21407,		// w07 w05 w03 w01
+	16384, -16384, 16384, 16384,	//; w14 w12 w10 w08
+	-8867, 21407, -21407, -8867,	//; w15 w13 w11 w09
+	22725, 12873, 19266, -22725,	//; w22 w20 w18 w16
+	19266, 4520, -4520, -12873,		//; w23 w21 w19 w17
+	12873, 4520, 4520, 19266,		//; w30 w28 w26 w24
+	-22725, 19266, -12873, -22725,  //w31 w29 w27 w25
+
+	// the rest of these rows (1-7), aren't used !
+
+	//row1
+	22725, 22725, 22725, -22725,	// ; movq-> w06 w04 w02 w00
+	29692, 12299, 12299, -29692,	//	; w07 w05 w03 w01
+	22725, -22725, 22725, 22725,	//; w14 w12 w10 w08
+	-12299, 29692, -29692, -12299,	//; w15 w13 w11 w09
+	31521, 17855, 26722, -31521,	//; w22 w20 w18 w16
+	26722, 6270, -6270, -17855,		//; w23 w21 w19 w17
+	17855, 6270, 6270, 26722,		//; w30 w28 w26 w24
+	-31521, 26722, -17855, -31521,	// w31 w29 w27 w25
+
+	//row2
+	21407, 21407, 21407, -21407,	// ; movq-> w06 w04 w02 w00
+	27969, 11585, 11585, -27969,	// ; w07 w05 w03 w01
+	21407, -21407, 21407, 21407,	// ; w14 w12 w10 w08
+	-11585, 27969, -27969, -11585,	//  ;w15 w13 w11 w09
+	29692, 16819, 25172, -29692, 	// ;w22 w20 w18 w16
+	25172, 5906, -5906, -16819, 	// ;w23 w21 w19 w17
+	16819, 5906, 5906, 25172, 		// ;w30 w28 w26 w24
+	-29692, 25172, -16819, -29692,	//  ;w31 w29 w27 w25
+
+	//row3
+	19266, 19266, 19266, -19266,	//; movq-> w06 w04 w02 w00
+	25172, 10426, 10426, -25172,	//; w07 w05 w03 w01
+	19266, -19266, 19266, 19266,	//; w14 w12 w10 w08
+	-10426, 25172, -25172, -10426,	//; w15 w13 w11 w09
+	26722, 15137, 22654, -26722,	//; w22 w20 w18 w16
+	22654, 5315, -5315, -15137,		//; w23 w21 w19 w17
+	15137, 5315, 5315, 22654,		//; w30 w28 w26 w24
+	-26722, 22654, -15137, -26722,	//; w31 w29 w27 w25
+
+	//row4
+	16384, 16384, 16384, -16384,	// ; movq-> w06 w04 w02 w00
+	21407, 8867, 8867, -21407,		// w07 w05 w03 w01
+	16384, -16384, 16384, 16384,	//; w14 w12 w10 w08
+	-8867, 21407, -21407, -8867,	//; w15 w13 w11 w09
+	22725, 12873, 19266, -22725,	//; w22 w20 w18 w16
+	19266, 4520, -4520, -12873,		//; w23 w21 w19 w17
+	12873, 4520, 4520, 19266,		//; w30 w28 w26 w24
+	-22725, 19266, -12873, -22725,  //w31 w29 w27 w25
+
+	//row5
+	19266, 19266, 19266, -19266,	//; movq-> w06 w04 w02 w00
+	25172, 10426, 10426, -25172,	//; w07 w05 w03 w01
+	19266, -19266, 19266, 19266,	//; w14 w12 w10 w08
+	-10426, 25172, -25172, -10426,	//; w15 w13 w11 w09
+	26722, 15137, 22654, -26722,	//; w22 w20 w18 w16
+	22654, 5315, -5315, -15137,		//; w23 w21 w19 w17
+	15137, 5315, 5315, 22654,		//; w30 w28 w26 w24
+	-26722, 22654, -15137, -26722,	//; w31 w29 w27 w25
+
+	//row6
+	21407, 21407, 21407, -21407,	// ; movq-> w06 w04 w02 w00
+	27969, 11585, 11585, -27969,	// ; w07 w05 w03 w01
+	21407, -21407, 21407, 21407,	// ; w14 w12 w10 w08
+	-11585, 27969, -27969, -11585,	//  ;w15 w13 w11 w09
+	29692, 16819, 25172, -29692, 	// ;w22 w20 w18 w16
+	25172, 5906, -5906, -16819, 	// ;w23 w21 w19 w17
+	16819, 5906, 5906, 25172, 		// ;w30 w28 w26 w24
+	-29692, 25172, -16819, -29692,	//  ;w31 w29 w27 w25
+
+	//row7
+	22725, 22725, 22725, -22725,	// ; movq-> w06 w04 w02 w00
+	29692, 12299, 12299, -29692,	//	; w07 w05 w03 w01
+	22725, -22725, 22725, 22725,	//; w14 w12 w10 w08
+	-12299, 29692, -29692, -12299,	//; w15 w13 w11 w09
+	31521, 17855, 26722, -31521,	//; w22 w20 w18 w16
+	26722, 6270, -6270, -17855,		//; w23 w21 w19 w17
+	17855, 6270, 6270, 26722,		//; w30 w28 w26 w24
+	-31521, 26722, -17855, -31521};	// w31 w29 w27 w25
+
+
+#define INP eax		// pointer to (short *blk)
+#define OUT ecx		// pointer to output (temporary store space qwTemp[])
+#define TABLE ebx	// pointer to tab_i_01234567[]
+#define round_inv_row edx
+#define round_inv_col edx
+
+#define ROW_STRIDE 8 // for 8x8 matrix transposer
+
+// private variables and functions
+
+//temporary storage space, 8x8 of shorts
+
+__inline static void idct_mmx32_rows( short *blk ); // transform rows
+__inline static void idct_mmx32_cols( short *blk ); // transform "columns"
+	// the "column" transform actually transforms rows, it is
+	// identical to the row-transform except for the ROUNDING
+	// and SHIFTING coefficients.
+
+ 
+static void 
+idct_mmx32_rows( short *blk )	// transform all 8 rows of 8x8 iDCT block
+{
+  int x;
+  short qwTemp[64];
+  short *out = &qwTemp[0];
+  short *inptr = blk;
+  // this subroutine performs two operations
+  // 1) iDCT row transform
+  //		for( i = 0; i < 8; ++ i)
+  //			DCT_8_INV_ROW_1( blk[i*8], qwTemp[i] );
+  //
+  // 2) transpose the matrix (which was stored in qwTemp[])
+  //        qwTemp[] -> [8x8 matrix transpose] -> blk[]
+
+  for (x=0; x<8; x++) {  // transform one row per iteration
+	 movq_m2r(*(inptr), mm0);		// 0 ; x3 x2 x1 x0
+
+	 movq_m2r(*(inptr+4), mm1);	// 1 ; x7 x6 x5 x4
+	 movq_r2r(mm0, mm2);				// 2 ; x3 x2 x1 x0
+
+	 movq_m2r(*(tab_i_01234567), mm3);	// 3 ; w06 w04 w02 w00
+	 punpcklwd_r2r(mm1, mm0);			// x5 x1 x4 x0
+
+    // ----------
+	 movq_r2r(mm0, mm5);					// 5 ; x5 x1 x4 x0
+	 punpckldq_r2r(mm0, mm0);			// x4 x0 x4 x0
+
+	 movq_m2r(*(tab_i_01234567+4), mm4);	// 4 ; w07 w05 w03 w01
+	 punpckhwd_r2r(mm1, mm2);			// 1 ; x7 x3 x6 x2
+
+	 pmaddwd_r2r(mm0, mm3);				// x4*w06+x0*w04 x4*w02+x0*w00
+	 movq_r2r(mm2, mm6);				// 6 ; x7 x3 x6 x2
+
+	 movq_m2r(*(tab_i_01234567+16), mm1);// 1 ; w22 w20 w18 w16
+	 punpckldq_r2r(mm2, mm2);			// x6 x2 x6 x2
+
+	 pmaddwd_r2r(mm2, mm4);				// x6*w07+x2*w05 x6*w03+x2*w01
+	 punpckhdq_r2r(mm5, mm5);			// x5 x1 x5 x1
+
+	 pmaddwd_m2r(*(tab_i_01234567+8), mm0);// x4*w14+x0*w12 x4*w10+x0*w08
+	 punpckhdq_r2r(mm6, mm6);			// x7 x3 x7 x3
+
+	 movq_m2r(*(tab_i_01234567+20), mm7);// 7 ; w23 w21 w19 w17
+	 pmaddwd_r2r(mm5, mm1);				// x5*w22+x1*w20 x5*w18+x1*w16
+
+	 paddd_m2r(*(r_inv_row), mm3);// +rounder
+	 pmaddwd_r2r(mm6, mm7);				// x7*w23+x3*w21 x7*w19+x3*w17
+
+	 pmaddwd_m2r(*(tab_i_01234567+12), mm2);// x6*w15+x2*w13 x6*w11+x2*w09
+	 paddd_r2r(mm4, mm3);				// 4 ; a1=sum(even1) a0=sum(even0)
+
+	 pmaddwd_m2r(*(tab_i_01234567+24), mm5);// x5*w30+x1*w28 x5*w26+x1*w24
+	 movq_r2r(mm3, mm4);				// 4 ; a1 a0
+
+	 pmaddwd_m2r(*(tab_i_01234567+28), mm6);// x7*w31+x3*w29 x7*w27+x3*w25
+	 paddd_r2r(mm7, mm1);				// 7 ; b1=sum(odd1) b0=sum(odd0)
+
+	 paddd_m2r(*(r_inv_row), mm0);// +rounder
+	 psubd_r2r(mm1, mm3);				// a1-b1 a0-b0
+
+	 psrad_i2r(SHIFT_INV_ROW, mm3);		// y6=a1-b1 y7=a0-b0
+	 paddd_r2r(mm4, mm1);				// 4 ; a1+b1 a0+b0
+
+	 paddd_r2r(mm2, mm0);				// 2 ; a3=sum(even3) a2=sum(even2)
+	 psrad_i2r(SHIFT_INV_ROW, mm1);		// y1=a1+b1 y0=a0+b0
+
+	 paddd_r2r(mm6, mm5);				// 6 ; b3=sum(odd3) b2=sum(odd2)
+	 movq_r2r(mm0, mm4);				// 4 ; a3 a2
+
+	 paddd_r2r(mm5, mm0);				// a3+b3 a2+b2
+	 psubd_r2r(mm5, mm4);				// 5 ; a3-b3 a2-b2
+
+	 psrad_i2r(SHIFT_INV_ROW, mm4);		// y4=a3-b3 y5=a2-b2
+	 psrad_i2r(SHIFT_INV_ROW, mm0);		// y3=a3+b3 y2=a2+b2
+
+	 packssdw_r2r(mm3, mm4);				// 3 ; y6 y7 y4 y5
+
+	 packssdw_r2r(mm0, mm1);				// 0 ; y3 y2 y1 y0
+	 movq_r2r(mm4, mm7);				// 7 ; y6 y7 y4 y5
+
+	 psrld_i2r(16, mm4);					// 0 y6 0 y4
+
+	 movq_r2m(mm1, *(out));	// 1 ; save y3 y2 y1 y0
+	 pslld_i2r(16, mm7);					// y7 0 y5 0
+
+	 por_r2r(mm4, mm7);					// 4 ; y7 y6 y5 y4
+
+   // begin processing row 1
+	 movq_r2m(mm7, *(out+4));	// 7 ; save y7 y6 y5 y4
+
+	 inptr += 8;
+	 out += 8;
+  }
+	 
+
+	// done with the iDCT row-transformation
+
+	// now we have to transpose the output 8x8 matrix
+	// 8x8 (OUT) -> 8x8't' (IN)
+	// the transposition is implemented as 4 sub-operations.
+	// 1) transpose upper-left quad
+	// 2) transpose lower-right quad
+	// 3) transpose lower-left quad
+	// 4) transpose upper-right quad
+
+ 
+	// mm0 = 1st row [ A B C D ] row1
+	// mm1 = 2nd row [ E F G H ] 2
+	// mm2 = 3rd row [ I J K L ] 3
+	// mm3 = 4th row [ M N O P ] 4
+
+	// 1) transpose upper-left quad
+  out = &qwTemp[0];
+
+  movq_m2r(*(out + ROW_STRIDE * 0), mm0);
+
+  movq_m2r(*(out + ROW_STRIDE * 1), mm1);
+  movq_r2r(mm0, mm4);	// mm4 = copy of row1[A B C D]
+	
+  movq_m2r(*(out + ROW_STRIDE * 2), mm2);
+  punpcklwd_r2r(mm1, mm0); // mm0 = [ 0 4 1 5]
+	
+  movq_m2r(*(out + ROW_STRIDE * 3), mm3);
+  punpckhwd_r2r(mm1, mm4); // mm4 = [ 2 6 3 7]
+
+  movq_r2r(mm2, mm6);
+  punpcklwd_r2r(mm3, mm2);	// mm2 = [ 8 12 9 13]
+
+  punpckhwd_r2r(mm3, mm6);	// mm6 = 10 14 11 15]
+  movq_r2r(mm0, mm1);	// mm1 = [ 0 4 1 5]
+
+  inptr = blk;
+
+  punpckldq_r2r(mm2, mm0);	// final result mm0 = row1 [0 4 8 12]
+
+  movq_r2r(mm4, mm3);	// mm3 = [ 2 6 3 7]
+  punpckhdq_r2r(mm2, mm1); // mm1 = final result mm1 = row2 [1 5 9 13]
+
+  movq_r2m(mm0, *(inptr + ROW_STRIDE * 0)); // store row 1
+  punpckldq_r2r(mm6, mm4); // final result mm4 = row3 [2 6 10 14]
+
+// begin reading next quadrant (lower-right)
+  movq_m2r(*(out + ROW_STRIDE*4 + 4), mm0); 
+  punpckhdq_r2r(mm6, mm3); // final result mm3 = row4 [3 7 11 15]
+
+  movq_r2m(mm4, *(inptr + ROW_STRIDE * 2)); // store row 3
+  movq_r2r(mm0, mm4);	// mm4 = copy of row1[A B C D]
+
+  movq_r2m(mm1, *(inptr + ROW_STRIDE * 1)); // store row 2
+
+  movq_m2r(*(out + ROW_STRIDE*5 + 4), mm1);
+
+  movq_r2m(mm3, *(inptr + ROW_STRIDE * 3)); // store row 4
+  punpcklwd_r2r(mm1, mm0); // mm0 = [ 0 4 1 5]
+
+	// 2) transpose lower-right quadrant
+
+//	movq mm0, qword ptr [OUT + ROW_STRIDE*4 + 8]
+
+//	movq mm1, qword ptr [OUT + ROW_STRIDE*5 + 8]
+//	 movq mm4, mm0;	// mm4 = copy of row1[A B C D]
+	
+  movq_m2r(*(out + ROW_STRIDE*6 + 4), mm2);
+//	 punpcklwd mm0, mm1; // mm0 = [ 0 4 1 5]
+  punpckhwd_r2r(mm1, mm4); // mm4 = [ 2 6 3 7]
+	
+  movq_m2r(*(out + ROW_STRIDE*7 + 4), mm3);
+  movq_r2r(mm2, mm6);
+
+  punpcklwd_r2r(mm3, mm2);	// mm2 = [ 8 12 9 13]
+  movq_r2r(mm0, mm1);	// mm1 = [ 0 4 1 5]
+
+  punpckhwd_r2r(mm3, mm6);	// mm6 = 10 14 11 15]
+  movq_r2r(mm4, mm3);	// mm3 = [ 2 6 3 7]
+
+  punpckldq_r2r(mm2, mm0);	// final result mm0 = row1 [0 4 8 12]
+
+  punpckhdq_r2r(mm2, mm1); // mm1 = final result mm1 = row2 [1 5 9 13]
+  ; // slot
+
+  movq_r2m(mm0, *(inptr + ROW_STRIDE*4 + 4)); // store row 1
+  punpckldq_r2r(mm6, mm4); // final result mm4 = row3 [2 6 10 14]
+
+  movq_m2r(*(out + ROW_STRIDE * 4 ), mm0);
+  punpckhdq_r2r(mm6, mm3); // final result mm3 = row4 [3 7 11 15]
+  
+  movq_r2m(mm4, *(inptr + ROW_STRIDE*6 + 4)); // store row 3
+  movq_r2r(mm0, mm4);	// mm4 = copy of row1[A B C D]
+
+  movq_r2m(mm1, *(inptr + ROW_STRIDE*5 + 4)); // store row 2
+  ; // slot
+  
+  movq_m2r(*(out + ROW_STRIDE * 5 ), mm1);
+  ; // slot
+
+  movq_r2m(mm3, *(inptr + ROW_STRIDE*7 + 4)); // store row 4
+  punpcklwd_r2r(mm1, mm0); // mm0 = [ 0 4 1 5]
+
+  // 3) transpose lower-left
+//	movq mm0, qword ptr [OUT + ROW_STRIDE * 4 ]
+
+//	movq mm1, qword ptr [OUT + ROW_STRIDE * 5 ]
+//	 movq mm4, mm0;	// mm4 = copy of row1[A B C D]
+	
+  movq_m2r(*(out + ROW_STRIDE * 6 ), mm2);
+//	 punpcklwd mm0, mm1; // mm0 = [ 0 4 1 5]
+  punpckhwd_r2r(mm1, mm4); // mm4 = [ 2 6 3 7]
+	
+  movq_m2r(*(out + ROW_STRIDE * 7 ), mm3);
+  movq_r2r(mm2, mm6);
+
+  punpcklwd_r2r(mm3, mm2);	// mm2 = [ 8 12 9 13]
+  movq_r2r(mm0, mm1);	// mm1 = [ 0 4 1 5]
+
+  punpckhwd_r2r(mm3, mm6);	// mm6 = 10 14 11 15]
+  movq_r2r(mm4, mm3);	// mm3 = [ 2 6 3 7]
+
+  punpckldq_r2r(mm2, mm0);	// final result mm0 = row1 [0 4 8 12]
+
+  punpckhdq_r2r(mm2, mm1); // mm1 = final result mm1 = row2 [1 5 9 13]
+  ;//slot
+
+  movq_r2m(mm0, *(inptr + ROW_STRIDE * 0 + 4 )); // store row 1
+  punpckldq_r2r(mm6, mm4); // final result mm4 = row3 [2 6 10 14]
+
+// begin reading next quadrant (upper-right)
+  movq_m2r(*(out + ROW_STRIDE*0 + 4), mm0);
+  punpckhdq_r2r(mm6, mm3); // final result mm3 = row4 [3 7 11 15]
+
+  movq_r2m(mm4, *(inptr + ROW_STRIDE * 2 + 4)); // store row 3
+  movq_r2r(mm0, mm4);	// mm4 = copy of row1[A B C D]
+
+  movq_r2m(mm1, *(inptr + ROW_STRIDE * 1 + 4)); // store row 2
+  movq_m2r(*(out + ROW_STRIDE*1 + 4), mm1);
+
+  movq_r2m(mm3, *(inptr + ROW_STRIDE * 3 + 4)); // store row 4
+  punpcklwd_r2r(mm1, mm0); // mm0 = [ 0 4 1 5]
+
+
+	// 2) transpose lower-right quadrant
+
+//	movq mm0, qword ptr [OUT + ROW_STRIDE*4 + 8]
+
+//	movq mm1, qword ptr [OUT + ROW_STRIDE*5 + 8]
+//	 movq mm4, mm0;	// mm4 = copy of row1[A B C D]
+	
+  movq_m2r(*(out + ROW_STRIDE*2 + 4), mm2);
+//	 punpcklwd mm0, mm1; // mm0 = [ 0 4 1 5]
+  punpckhwd_r2r(mm1, mm4); // mm4 = [ 2 6 3 7]
+	
+  movq_m2r(*(out + ROW_STRIDE*3 + 4), mm3);
+  movq_r2r(mm2, mm6);
+
+  punpcklwd_r2r(mm3, mm2);	// mm2 = [ 8 12 9 13]
+  movq_r2r(mm0, mm1);	// mm1 = [ 0 4 1 5]
+
+  punpckhwd_r2r(mm3, mm6);	// mm6 = 10 14 11 15]
+  movq_r2r(mm4, mm3);	// mm3 = [ 2 6 3 7]
+
+  punpckldq_r2r(mm2, mm0);	// final result mm0 = row1 [0 4 8 12]
+
+  punpckhdq_r2r(mm2, mm1); // mm1 = final result mm1 = row2 [1 5 9 13]
+  ; // slot
+
+  movq_r2m(mm0, *(inptr + ROW_STRIDE*4)); // store row 1
+  punpckldq_r2r(mm6, mm4); // final result mm4 = row3 [2 6 10 14]
+
+  movq_r2m(mm1, *(inptr + ROW_STRIDE*5)); // store row 2
+  punpckhdq_r2r(mm6, mm3); // final result mm3 = row4 [3 7 11 15]
+
+  movq_r2m(mm4, *(inptr + ROW_STRIDE*6)); // store row 3
+  ; // slot
+
+  movq_r2m(mm3, *(inptr + ROW_STRIDE*7)); // store row 4
+  ; // slot
+ 
+}
+
+
+static void 
+idct_mmx32_cols( short *blk )	// transform all 8 cols of 8x8 iDCT block
+{
+  int x;
+  short *inptr = blk;
+
+	// Despite the function's name, the matrix is transformed
+	// row by row.  This function is identical to idct_mmx32_rows(),
+	// except for the SHIFT amount and ROUND_INV amount.
+
+	// this subroutine performs two operations
+	// 1) iDCT row transform
+	//		for( i = 0; i < 8; ++ i)
+	//			DCT_8_INV_ROW_1( blk[i*8], qwTemp[i] );
+	//
+	// 2) transpose the matrix (which was stored in qwTemp[])
+	//        qwTemp[] -> [8x8 matrix transpose] -> blk[]
+
+
+  for (x=0; x<8; x++) {  // transform one row per iteration
+
+    movq_m2r(*(inptr), mm0);		// 0 ; x3 x2 x1 x0
+
+    movq_m2r(*(inptr+4), mm1);	// 1 ; x7 x6 x5 x4
+	 movq_r2r(mm0, mm2);				// 2 ; x3 x2 x1 x0
+
+	 movq_m2r(*(tab_i_01234567), mm3);	// 3 ; w06 w04 w02 w00
+	 punpcklwd_r2r(mm1, mm0);			// x5 x1 x4 x0
+
+// ----------
+	 movq_r2r(mm0, mm5);					// 5 ; x5 x1 x4 x0
+	 punpckldq_r2r(mm0, mm0);			// x4 x0 x4 x0
+
+	 movq_m2r(*(tab_i_01234567+4), mm4);	// 4 ; w07 w05 w03 w01
+	 punpckhwd_r2r(mm1, mm2);			// 1 ; x7 x3 x6 x2
+
+	 pmaddwd_r2r(mm0, mm3);				// x4*w06+x0*w04 x4*w02+x0*w00
+	 movq_r2r(mm2, mm6);				// 6 ; x7 x3 x6 x2
+
+	 movq_m2r(*(tab_i_01234567+16), mm1);// 1 ; w22 w20 w18 w16
+	 punpckldq_r2r(mm2, mm2);			// x6 x2 x6 x2
+
+	 pmaddwd_r2r(mm2, mm4);				// x6*w07+x2*w05 x6*w03+x2*w01
+	 punpckhdq_r2r(mm5, mm5);			// x5 x1 x5 x1
+
+	 pmaddwd_m2r(*(tab_i_01234567+8), mm0);// x4*w14+x0*w12 x4*w10+x0*w08
+	 punpckhdq_r2r(mm6, mm6);			// x7 x3 x7 x3
+
+	 movq_m2r(*(tab_i_01234567+20), mm7);// 7 ; w23 w21 w19 w17
+	 pmaddwd_r2r(mm5, mm1);				// x5*w22+x1*w20 x5*w18+x1*w16
+
+	 paddd_m2r(*(r_inv_col), mm3);// +rounder
+	 pmaddwd_r2r(mm6, mm7);				// x7*w23+x3*w21 x7*w19+x3*w17
+
+	 pmaddwd_m2r(*(tab_i_01234567+12), mm2);// x6*w15+x2*w13 x6*w11+x2*w09
+	 paddd_r2r(mm4, mm3);				// 4 ; a1=sum(even1) a0=sum(even0)
+
+	 pmaddwd_m2r(*(tab_i_01234567+24), mm5);// x5*w30+x1*w28 x5*w26+x1*w24
+	 movq_r2r(mm3, mm4);				// 4 ; a1 a0
+
+	 pmaddwd_m2r(*(tab_i_01234567+28), mm6);// x7*w31+x3*w29 x7*w27+x3*w25
+	 paddd_r2r(mm7, mm1);				// 7 ; b1=sum(odd1) b0=sum(odd0)
+
+	 paddd_m2r(*(r_inv_col), mm0);// +rounder
+	 psubd_r2r(mm1, mm3);				// a1-b1 a0-b0
+
+	 psrad_i2r(SHIFT_INV_COL, mm3);		// y6=a1-b1 y7=a0-b0
+	 paddd_r2r(mm4, mm1);				// 4 ; a1+b1 a0+b0
+
+	 paddd_r2r(mm2, mm0);				// 2 ; a3=sum(even3) a2=sum(even2)
+	 psrad_i2r(SHIFT_INV_COL, mm1);		// y1=a1+b1 y0=a0+b0
+
+	 paddd_r2r(mm6, mm5);				// 6 ; b3=sum(odd3) b2=sum(odd2)
+	 movq_r2r(mm0, mm4);				// 4 ; a3 a2
+
+	 paddd_r2r(mm5, mm0);				// a3+b3 a2+b2
+	 psubd_r2r(mm5, mm4);				// 5 ; a3-b3 a2-b2
+
+
+	 psrad_i2r(SHIFT_INV_COL, mm4);		// y4=a3-b3 y5=a2-b2
+	 psrad_i2r(SHIFT_INV_COL, mm0);		// y3=a3+b3 y2=a2+b2
+
+	 packssdw_r2r(mm3, mm4);				// 3 ; y6 y7 y4 y5
+
+	 packssdw_r2r(mm0, mm1);				// 0 ; y3 y2 y1 y0
+	 movq_r2r(mm4, mm7);				// 7 ; y6 y7 y4 y5
+
+	 psrld_i2r(16, mm4);					// 0 y6 0 y4
+
+	 movq_r2m(mm1, *(inptr));	// 1 ; save y3 y2 y1 y0
+	 pslld_i2r(16, mm7);					// y7 0 y5 0
+
+	 por_r2r(mm4, mm7);					// 4 ; y7 y6 y5 y4
+
+   // begin processing row 1
+	 movq_r2m(mm7, *(inptr+4));	// 7 ; save y7 y6 y5 y4
+
+	 inptr += 8;
+  }
+  // done with the iDCT column-transformation
+}
+
+//	
+// public interface to MMX32 IDCT 8x8 operation
+//
+void
+gst_idct_mmx32_idct( short *blk )
+{
+	// 1) iDCT row transformation
+	idct_mmx32_rows( blk ); // 1) transform iDCT row, and transpose
+
+	// 2) iDCT column transformation
+	idct_mmx32_cols( blk ); // 2) transform iDCT row, and transpose
+
+	emms();  // restore processor state
+	// all done
+}

libs/idct/mmxidct.S

@@ -0,0 +1,725 @@
+/*
+ * the input data is tranposed and each 16 bit element in the 8x8 matrix
+ * is left aligned:
+ * for example in 11...1110000 format
+ * If the iDCT is of I macroblock then 0.5 needs to be added to the;DC Component
+ * (element[0][0] of the matrix)
+ */
+
+/* extrn re_matrix */
+
+.data
+	.align 16
+	.type	 preSC,@object
+preSC:  .short  16384,22725,21407,19266,16384,12873,8867,4520
+        .short  22725,31521,29692,26722,22725,17855,12299,6270
+        .short  21407,29692,27969,25172,21407,16819,11585,5906
+        .short  19266,26722,25172,22654,19266,15137,10426,5315
+        .short  16384,22725,21407,19266,16384,12873,8867,4520
+        .short  12873,17855,16819,15137,25746,20228,13933,7103
+        .short  17734,24598,23170,20853,17734,13933,9597,4892
+        .short  18081,25080,23624,21261,18081,14206,9785,4988
+	.size	 preSC,128
+	.align 8
+	.type	x0005000200010001,@object
+	.size	x0005000200010001,8
+x0005000200010001:
+	.long	0x00010001,0x00050002
+	.align 8
+	.type	x0040000000000000,@object
+	.size	x0040000000000000,8
+x0040000000000000:
+	.long	0, 0x00400000
+	.align 8
+	.type	x5a825a825a825a82,@object
+	.size	x5a825a825a825a82,8
+x5a825a825a825a82:
+	.long	0x5a825a82, 0x5a825a82
+	.align 8
+	.type	x539f539f539f539f,@object
+	.size	x539f539f539f539f,8
+x539f539f539f539f:
+	.long	0x539f539f,0x539f539f
+	.align 8
+	.type	x4546454645464546,@object
+	.size	x4546454645464546,8
+x4546454645464546:
+	.long	0x45464546,0x45464546
+	.align 8
+	.type	x61f861f861f861f8,@object
+	.size	x61f861f861f861f8,8
+x61f861f861f861f8:
+	.long	0x61f861f8,0x61f861f8
+	.type	x0004000000000000,@object
+	.size	x0004000000000000,8
+x0004000000000000:
+	.long	0x00000000,0x00040000
+	.type	x0000000000000004,@object
+	.size	x0000000000000004,8
+x0000000000000004:
+	.long	0x00000004,0x00000000
+	.align 8
+	.type	 scratch1,@object
+	.size	 scratch1,8
+scratch1:
+	.long 0,0
+	.align 8
+	.type	 scratch3,@object
+	.size	 scratch3,8
+scratch3:
+	.long 0,0
+	.align 8
+	.type	 scratch5,@object
+	.size	 scratch5,8
+scratch5:
+	.long 0,0
+	.align 8
+	.type	 scratch7,@object
+	.size	 scratch7,8
+scratch7:
+	.long 0,0
+	.type	 x0,@object
+	.size	 x0,8
+x0:
+	.long 0,0
+	.align 8
+.text
+	.align 4
+.globl gst_idct_mmx_idct
+	.type	 gst_idct_mmx_idct,@function
+gst_idct_mmx_idct:
+	pushl %ebp
+	movl %esp,%ebp
+	pushl %ebx
+	pushl %ecx
+	pushl %edx
+	pushl %esi
+	pushl %edi
+	movl 8(%ebp),%esi		/* source matrix */
+	movq (%esi), %mm0
+	paddw x0000000000000004, %mm0
+	movq 8(%esi), %mm1
+	psllw $4, %mm0
+	movq 16(%esi), %mm2
+	psllw $4, %mm1
+	movq 24(%esi), %mm3
+	psllw $4, %mm2
+	movq 32(%esi), %mm4
+	psllw $4, %mm3
+	movq 40(%esi), %mm5
+	psllw $4, %mm4
+	movq 48(%esi), %mm6
+	psllw $4, %mm5
+	movq 56(%esi), %mm7
+	psllw $4, %mm6
+	psllw $4, %mm7
+	movq %mm0,  (%esi)
+	movq %mm1, 8(%esi)
+	movq %mm2,16(%esi)
+	movq %mm3,24(%esi)
+	movq %mm4,32(%esi)
+	movq %mm5,40(%esi)
+	movq %mm6,48(%esi)
+	movq %mm7,56(%esi)
+	movq 64(%esi), %mm0
+	movq 72(%esi), %mm1
+	psllw $4, %mm0
+	movq 80(%esi), %mm2
+	psllw $4, %mm1
+	movq 88(%esi), %mm3
+	psllw $4, %mm2
+	movq 96(%esi), %mm4
+	psllw $4, %mm3
+	movq 104(%esi), %mm5
+	psllw $4, %mm4
+	movq 112(%esi), %mm6
+	psllw $4, %mm5
+	movq 120(%esi), %mm7
+	psllw $4, %mm6
+	psllw $4, %mm7
+	movq %mm0,64(%esi)
+	movq %mm1,72(%esi)
+	movq %mm2,80(%esi)
+	movq %mm3,88(%esi)
+	movq %mm4,96(%esi)
+	movq %mm5,104(%esi)
+	movq %mm6,112(%esi)
+	movq %mm7,120(%esi)
+	leal preSC, %ecx
+/* column 0: even part
+ * use V4, V12, V0, V8 to produce V22..V25
+ */
+	movq 8*12(%ecx), %mm0	/* maybe the first mul can be done together */
+				/* with the dequantization in iHuff module */
+	pmulhw 8*12(%esi), %mm0		/* V12 */
+	movq 8*4(%ecx), %mm1
+	pmulhw 8*4(%esi), %mm1		/* V4 */
+	movq (%ecx), %mm3
+	psraw $1, %mm0			/* t64=t66 */
+	pmulhw (%esi), %mm3		/* V0 */
+	movq 8*8(%ecx), %mm5		/* duplicate V4 */
+	movq %mm1, %mm2			/* added 11/1/96 */
+	pmulhw 8*8(%esi),%mm5		/* V8 */
+	psubsw %mm0, %mm1		/* V16 */
+	pmulhw x5a825a825a825a82, %mm1	/* 23170 ->V18 */
+	paddsw %mm0, %mm2		/* V17 */
+	movq %mm2, %mm0			/* duplicate V17 */
+	psraw $1, %mm2			/* t75=t82 */
+	psraw $2, %mm0			/* t72 */
+	movq %mm3, %mm4			/* duplicate V0 */
+	paddsw %mm5, %mm3		/* V19 */
+	psubsw %mm5, %mm4		/* V20 ;mm5 free */
+/* moved from the block below */
+	movq 8*10(%ecx), %mm7
+	psraw $1, %mm3			/* t74=t81 */
+	movq %mm3, %mm6			/* duplicate t74=t81 */
+	psraw $2, %mm4			/* t77=t79 */
+	psubsw %mm0, %mm1		/* V21 ; mm0 free */
+	paddsw %mm2, %mm3		/* V22 */
+	movq %mm1, %mm5			/* duplicate V21 */
+	paddsw %mm4, %mm1		/* V23 */
+	movq %mm3, 8*4(%esi)		/* V22 */
+	psubsw %mm5, %mm4		/* V24; mm5 free */
+	movq %mm1, 8*12(%esi)		/* V23 */
+	psubsw %mm2, %mm6		/* V25; mm2 free */
+	movq %mm4, (%esi)		/* V24 */
+/* keep mm6 alive all along the next block */
+	/* movq %mm6, 8*8(%esi) 	V25 */
+/* column 0: odd part
+ * use V2, V6, V10, V14 to produce V31, V39, V40, V41
+ */
+/* moved above: movq 8*10(%ecx), %mm7 */
+
+	pmulhw 8*10(%esi), %mm7		/* V10 */
+	movq 8*6(%ecx), %mm0
+	pmulhw 8*6(%esi), %mm0		/* V6 */
+	movq 8*2(%ecx), %mm5
+	movq %mm7, %mm3			/* duplicate V10 */
+	pmulhw 8*2(%esi), %mm5		/* V2 */
+	movq 8*14(%ecx), %mm4
+	psubsw %mm0, %mm7		/* V26 */
+	pmulhw 8*14(%esi), %mm4		/* V14 */
+	paddsw %mm0, %mm3		/* V29 ; free mm0 */
+	movq %mm7, %mm1			/* duplicate V26 */
+	psraw $1, %mm3			/* t91=t94 */
+	pmulhw x539f539f539f539f,%mm7	/* V33 */
+	psraw $1, %mm1			/* t96 */
+	movq %mm5, %mm0			/* duplicate V2 */
+	psraw $2, %mm4			/* t85=t87 */
+	paddsw %mm4,%mm5		/* V27 */
+	psubsw %mm4, %mm0		/* V28 ; free mm4 */
+	movq %mm0, %mm2			/* duplicate V28 */
+	psraw $1, %mm5			/* t90=t93 */
+	pmulhw x4546454645464546,%mm0	/* V35 */
+	psraw $1, %mm2			/* t97 */
+	movq %mm5, %mm4			/* duplicate t90=t93 */
+	psubsw %mm2, %mm1		/* V32 ; free mm2 */
+	pmulhw x61f861f861f861f8,%mm1	/* V36 */
+	psllw $1, %mm7			/* t107 */
+	paddsw %mm3, %mm5		/* V31 */
+	psubsw %mm3, %mm4		/* V30 ; free mm3 */
+	pmulhw x5a825a825a825a82,%mm4	/* V34 */
+	nop
+	psubsw %mm1, %mm0		/* V38 */
+	psubsw %mm7, %mm1		/* V37 ; free mm7 */
+	psllw $1, %mm1			/* t114 */
+/* move from the next block */
+	movq %mm6, %mm3			/* duplicate V25 */
+/* move from the next block */
+	movq 8*4(%esi), %mm7		/* V22 */
+	psllw $1, %mm0			/* t110 */
+	psubsw %mm5, %mm0		/* V39 (mm5 needed for next block) */
+	psllw $2, %mm4			/* t112 */
+/* moved from the next block */
+	movq 8*12(%esi), %mm2		/* V23 */
+	psubsw %mm0, %mm4		/* V40 */
+	paddsw %mm4, %mm1		/* V41; free mm0 */
+/* moved from the next block */
+	psllw $1, %mm2			/* t117=t125 */
+/* column 0: output butterfly */
+/* moved above:
+ * movq %mm6, %mm3			duplicate V25
+ * movq 8*4(%esi), %mm7			V22
+ * movq 8*12(%esi), %mm2		V23
+ * psllw $1, %mm2			t117=t125
+ */
+	psubsw %mm1, %mm6		/* tm6 */
+	paddsw %mm1, %mm3		/* tm8; free mm1 */
+	movq %mm7, %mm1			/* duplicate V22 */
+	paddsw %mm5, %mm7		/* tm0 */
+	movq %mm3, 8*8(%esi)		/* tm8; free mm3 */
+	psubsw %mm5, %mm1		/* tm14; free mm5 */
+	movq %mm6, 8*6(%esi)		/* tm6; free mm6 */
+	movq %mm2, %mm3			/* duplicate t117=t125 */
+	movq (%esi), %mm6		/* V24 */
+	paddsw %mm0, %mm2		/* tm2 */
+	movq %mm7, (%esi)		/* tm0; free mm7 */
+	psubsw %mm0, %mm3		/* tm12; free mm0 */
+	movq %mm1, 8*14(%esi)		/* tm14; free mm1 */
+	psllw $1, %mm6			/* t119=t123 */
+	movq %mm2, 8*2(%esi)		/* tm2; free mm2 */
+	movq %mm6, %mm0			/* duplicate t119=t123 */
+	movq %mm3, 8*12(%esi)		/* tm12; free mm3 */
+	paddsw %mm4, %mm6		/* tm4 */
+/* moved from next block */
+	movq 8*5(%ecx), %mm1
+	psubsw %mm4, %mm0		/* tm10; free mm4 */
+/* moved from next block */
+	pmulhw 8*5(%esi), %mm1		/* V5 */
+	movq %mm6, 8*4(%esi)		/* tm4; free mm6 */
+	movq %mm0, 8*10(%esi)		/* tm10; free mm0 */
+/* column 1: even part
+ * use V5, V13, V1, V9 to produce V56..V59
+ */
+/* moved to prev block:
+ *	movq 8*5(%ecx), %mm1
+ *	pmulhw 8*5(%esi), %mm1		 V5
+ */
+	movq 8*13(%ecx), %mm7
+	psllw $1, %mm1			/* t128=t130 */
+	pmulhw 8*13(%esi), %mm7		/* V13 */
+	movq %mm1, %mm2			/* duplicate t128=t130 */
+	movq 8(%ecx), %mm3
+	pmulhw 8(%esi), %mm3		/* V1 */
+	movq 8*9(%ecx), %mm5
+	psubsw %mm7, %mm1		/* V50 */
+	pmulhw 8*9(%esi), %mm5		/* V9 */
+	paddsw %mm7, %mm2		/* V51 */
+	pmulhw x5a825a825a825a82, %mm1	/* 23170 ->V52 */
+	movq %mm2, %mm6			/* duplicate V51 */
+	psraw $1, %mm2			/* t138=t144 */
+	movq %mm3, %mm4			/* duplicate V1 */
+	psraw $2, %mm6			/* t136 */
+	paddsw %mm5, %mm3		/* V53 */
+	psubsw %mm5, %mm4		/* V54 ;mm5 free */
+	movq %mm3, %mm7			/* duplicate V53 */
+/* moved from next block */
+	movq 8*11(%ecx), %mm0
+	psraw $1, %mm4			/* t140=t142 */
+	psubsw %mm6, %mm1		/* V55 ; mm6 free */
+	paddsw %mm2, %mm3		/* V56 */
+	movq %mm4, %mm5			/* duplicate t140=t142 */
+	paddsw %mm1, %mm4		/* V57 */
+	movq %mm3, 8*5(%esi)		/* V56 */
+	psubsw %mm1, %mm5		/* V58; mm1 free */
+	movq %mm4, 8*13(%esi)		/* V57 */
+	psubsw %mm2, %mm7		/* V59; mm2 free */
+	movq %mm5, 8*9(%esi)		/* V58 */
+/* keep mm7 alive all along the next block
+ *	movq %mm7, 8(%esi)		V59
+ * moved above
+ *	movq 8*11(%ecx), %mm0
+ */
+	pmulhw 8*11(%esi), %mm0		/* V11 */
+	movq 8*7(%ecx), %mm6
+	pmulhw 8*7(%esi), %mm6		/* V7 */
+	movq 8*15(%ecx), %mm4
+	movq %mm0, %mm3			/* duplicate V11 */
+	pmulhw 8*15(%esi), %mm4		/* V15 */
+	movq 8*3(%ecx), %mm5
+	psllw $1, %mm6			/* t146=t152 */
+	pmulhw 8*3(%esi), %mm5		/* V3 */
+	paddsw %mm6, %mm0		/* V63 */
+/* note that V15 computation has a correction step: 
+ * this is a 'magic' constant that rebiases the results to be closer to the
+ * expected result.  this magic constant can be refined to reduce the error
+ * even more by doing the correction step in a later stage when the number
+ * is actually multiplied by 16
+ */
+	paddw x0005000200010001, %mm4
+	psubsw %mm6, %mm3		/* V60 ; free mm6 */
+	psraw $1, %mm0			/* t154=t156 */
+	movq %mm3, %mm1			/* duplicate V60 */
+	pmulhw x539f539f539f539f, %mm1	/* V67 */
+	movq %mm5, %mm6			/* duplicate V3 */
+	psraw $2, %mm4			/* t148=t150 */
+	paddsw %mm4, %mm5		/* V61 */
+	psubsw %mm4, %mm6		/* V62 ; free mm4 */
+	movq %mm5, %mm4			/* duplicate V61 */
+	psllw $1, %mm1			/* t169 */
+	paddsw %mm0, %mm5		/* V65 -> result */
+	psubsw %mm0, %mm4		/* V64 ; free mm0 */
+	pmulhw x5a825a825a825a82, %mm4	/* V68 */
+	psraw $1, %mm3			/* t158 */
+	psubsw %mm6, %mm3		/* V66 */
+	movq %mm5, %mm2			/* duplicate V65 */
+	pmulhw x61f861f861f861f8, %mm3	/* V70 */
+	psllw $1, %mm6			/* t165 */
+	pmulhw x4546454645464546, %mm6	/* V69 */
+	psraw $1, %mm2			/* t172 */
+/* moved from next block */
+	movq 8*5(%esi), %mm0		/* V56 */
+	psllw $1, %mm4			/* t174 */
+/* moved from next block */
+	psraw $1, %mm0			/* t177=t188 */
+	nop
+	psubsw %mm3, %mm6		/* V72 */
+	psubsw %mm1, %mm3		/* V71 ; free mm1 */
+	psubsw %mm2, %mm6		/* V73 ; free mm2 */
+/* moved from next block */
+	psraw $1, %mm5			/* t178=t189 */
+	psubsw %mm6, %mm4		/* V74 */
+/* moved from next block */
+	movq %mm0, %mm1			/* duplicate t177=t188 */
+	paddsw %mm4, %mm3		/* V75 */
+/* moved from next block */
+	paddsw %mm5, %mm0		/* tm1 */
+/* location
+ *  5 - V56
+ * 13 - V57
+ *  9 - V58
+ *  X - V59, mm7
+ *  X - V65, mm5
+ *  X - V73, mm6
+ *  X - V74, mm4
+ *  X - V75, mm3
+ * free mm0, mm1 & mm2
+ * moved above
+ *	movq 8*5(%esi), %mm0		V56
+ *	psllw $1, %mm0			t177=t188 ! new !!
+ *	psllw $1, %mm5			t178=t189 ! new !!
+ *	movq %mm0, %mm1			duplicate t177=t188
+ *	paddsw %mm5, %mm0		tm1
+ */
+	movq 8*13(%esi), %mm2		/* V57 */
+	psubsw %mm5, %mm1		/* tm15; free mm5 */
+	movq %mm0, 8(%esi)		/* tm1; free mm0 */
+	psraw $1, %mm7			/* t182=t184 ! new !! */
+/* save the store as used directly in the transpose
+ *	movq %mm1, 120(%esi)		tm15; free mm1
+ */
+	movq %mm7, %mm5			/* duplicate t182=t184 */
+	psubsw %mm3, %mm7		/* tm7 */
+	paddsw %mm3, %mm5		/* tm9; free mm3 */
+	movq 8*9(%esi), %mm0		/* V58 */
+	movq %mm2, %mm3			/* duplicate V57 */
+	movq %mm7, 8*7(%esi)		/* tm7; free mm7 */
+	psubsw %mm6, %mm3		/* tm13 */
+	paddsw %mm6, %mm2		/* tm3 ; free mm6 */
+/* moved up from the transpose */
+	movq %mm3, %mm7
+/* moved up from the transpose */
+	punpcklwd %mm1, %mm3
+	movq %mm0, %mm6			/* duplicate V58 */
+	movq %mm2, 8*3(%esi)		/* tm3; free mm2 */
+	paddsw %mm4, %mm0		/* tm5 */
+	psubsw %mm4, %mm6		/* tm11; free mm4 */
+/* moved up from the transpose */
+	punpckhwd %mm1, %mm7
+	movq %mm0, 8*5(%esi)		/* tm5; free mm0 */
+/* moved up from the transpose */
+	movq %mm5, %mm2
+/* transpose - M4 part
+ *  ---------       ---------
+ * | M1 | M2 |     | M1'| M3'|
+ *  ---------  -->  ---------
+ * | M3 | M4 |     | M2'| M4'|
+ *  ---------       ---------
+ * Two alternatives: use full mmword approach so the following code can be
+ * scheduled before the transpose is done without stores, or use the faster
+ * half mmword stores (when possible)
+ */
+	movd %mm3, 8*9+4(%esi)		/* MS part of tmt9 */
+	punpcklwd %mm6, %mm5
+	movd %mm7, 8*13+4(%esi)		/* MS part of tmt13 */
+	punpckhwd %mm6, %mm2
+	movd %mm5, 8*9(%esi)		/* LS part of tmt9 */
+	punpckhdq %mm3, %mm5		/* free mm3 */
+	movd %mm2, 8*13(%esi)		/* LS part of tmt13 */
+	punpckhdq %mm7, %mm2		/* free mm7 */
+/* moved up from the M3 transpose */
+	movq 8*8(%esi), %mm0
+/* moved up from the M3 transpose */
+	movq 8*10(%esi), %mm1
+/* moved up from the M3 transpose */
+	movq %mm0, %mm3
+/* shuffle the rest of the data, and write it with 2 mmword writes */
+	movq %mm5, 8*11(%esi)		/* tmt11 */
+/* moved up from the M3 transpose */
+	punpcklwd %mm1, %mm0
+	movq %mm2, 8*15(%esi)		/* tmt15 */
+/* moved up from the M3 transpose */
+	punpckhwd %mm1, %mm3
+/* transpose - M3 part
+ * moved up to previous code section
+ *	movq 8*8(%esi), %mm0
+ *	movq 8*10(%esi), %mm1
+ *	movq %mm0, %mm3
+ *	punpcklwd %mm1, %mm0
+ *	punpckhwd %mm1, %mm3
+ */
+	movq 8*12(%esi), %mm6
+	movq 8*14(%esi), %mm4
+	movq %mm6, %mm2
+/* shuffle the data and write the lower parts of the transposed in 4 dwords */
+	punpcklwd %mm4, %mm6
+	movq %mm0, %mm1
+	punpckhdq %mm6, %mm1
+	movq %mm3, %mm7
+	punpckhwd %mm4, %mm2		/* free mm4 */
+	punpckldq %mm6, %mm0		/* free mm6 */
+/* moved from next block */
+	movq 8*13(%esi), %mm4		/* tmt13 */
+	punpckldq %mm2, %mm3
+	punpckhdq %mm2, %mm7		/* free mm2 */
+/* moved from next block */
+	movq %mm3, %mm5			/* duplicate tmt5 */
+/* column 1: even part (after transpose)
+* moved above
+*	movq %mm3, %mm5			duplicate tmt5
+*	movq 8*13(%esi), %mm4		tmt13
+*/
+	psubsw %mm4, %mm3		/* V134 */
+	pmulhw x5a825a825a825a82, %mm3	/* 23170 ->V136 */
+	movq 8*9(%esi), %mm6		/* tmt9 */
+	paddsw %mm4, %mm5		/* V135 ; mm4 free */
+	movq %mm0, %mm4			/* duplicate tmt1 */
+	paddsw %mm6, %mm0		/* V137 */
+	psubsw %mm6, %mm4		/* V138 ; mm6 free */
+	psllw $2, %mm3			/* t290 */
+	psubsw %mm5, %mm3		/* V139 */
+	movq %mm0, %mm6			/* duplicate V137 */
+	paddsw %mm5, %mm0		/* V140 */
+	movq %mm4, %mm2			/* duplicate V138 */
+	paddsw %mm3, %mm2		/* V141 */
+	psubsw %mm3, %mm4		/* V142 ; mm3 free */
+	movq %mm0, 8*9(%esi)		/* V140 */
+	psubsw %mm5, %mm6		/* V143 ; mm5 free */
+/* moved from next block */
+	movq 8*11(%esi), %mm0		/* tmt11 */
+	movq %mm2, 8*13(%esi)		/* V141 */
+/* moved from next block */
+	movq %mm0, %mm2			/* duplicate tmt11 */
+/* column 1: odd part (after transpose) */
+/* moved up to the prev block
+ *	movq 8*11(%esi), %mm0		tmt11
+ *	movq %mm0, %mm2			duplicate tmt11
+ */
+	movq 8*15(%esi), %mm5		/* tmt15 */
+	psubsw %mm7, %mm0		/* V144 */
+	movq %mm0, %mm3			/* duplicate V144 */
+	paddsw %mm7, %mm2		/* V147 ; free mm7 */
+	pmulhw x539f539f539f539f, %mm0	/* 21407-> V151 */
+	movq %mm1, %mm7			/* duplicate tmt3 */
+	paddsw %mm5, %mm7		/* V145 */
+	psubsw %mm5, %mm1		/* V146 ; free mm5 */
+	psubsw %mm1, %mm3		/* V150 */
+	movq %mm7, %mm5			/* duplicate V145 */
+	pmulhw x4546454645464546, %mm1	/* 17734-> V153 */
+	psubsw %mm2, %mm5		/* V148 */
+	pmulhw x61f861f861f861f8, %mm3	/* 25080-> V154 */
+	psllw $2, %mm0			/* t311 */
+	pmulhw x5a825a825a825a82, %mm5	/* 23170-> V152 */
+	paddsw %mm2, %mm7		/* V149 ; free mm2 */
+	psllw $1, %mm1			/* t313 */
+	nop	/* without the nop - freeze here for one clock */
+	movq %mm3, %mm2			/* duplicate V154 */
+	psubsw %mm0, %mm3		/* V155 ; free mm0 */
+	psubsw %mm2, %mm1		/* V156 ; free mm2 */
+/* moved from the next block */
+	movq %mm6, %mm2			/* duplicate V143 */
+/* moved from the next block */
+	movq 8*13(%esi), %mm0		/* V141 */
+	psllw $1, %mm1			/* t315 */
+	psubsw %mm7, %mm1		/* V157 (keep V149) */
+	psllw $2, %mm5			/* t317 */
+	psubsw %mm1, %mm5		/* V158 */
+	psllw $1, %mm3			/* t319 */
+	paddsw %mm5, %mm3		/* V159 */
+/* column 1: output butterfly (after transform)
+ * moved to the prev block
+ *	movq %mm6, %mm2			duplicate V143
+ *	movq 8*13(%esi), %mm0		V141
+ */
+	psubsw %mm3, %mm2		/* V163 */
+	paddsw %mm3, %mm6		/* V164 ; free mm3 */
+	movq %mm4, %mm3			/* duplicate V142 */
+	psubsw %mm5, %mm4		/* V165 ; free mm5 */
+	movq %mm2, scratch7		/* out7 */
+	psraw $4, %mm6
+	psraw $4, %mm4
+	paddsw %mm5, %mm3		/* V162 */
+	movq 8*9(%esi), %mm2		/* V140 */
+	movq %mm0, %mm5			/* duplicate V141 */
+/* in order not to perculate this line up,
+ * we read 72(%esi) very near to this location
+ */
+	movq %mm6, 8*9(%esi)		/* out9 */
+	paddsw %mm1, %mm0		/* V161 */
+	movq %mm3, scratch5		/* out5 */
+	psubsw %mm1, %mm5		/* V166 ; free mm1 */
+	movq %mm4, 8*11(%esi)		/* out11 */
+	psraw $4, %mm5
+	movq %mm0, scratch3		/* out3 */
+	movq %mm2, %mm4			/* duplicate V140 */
+	movq %mm5, 8*13(%esi)		/* out13 */
+	paddsw %mm7, %mm2		/* V160 */
+/* moved from the next block */
+	movq 8(%esi), %mm0
+	psubsw %mm7, %mm4		/* V167 ; free mm7 */
+/* moved from the next block */
+	movq 8*3(%esi), %mm7
+	psraw $4, %mm4
+	movq %mm2, scratch1		/* out1 */
+/* moved from the next block */
+	movq %mm0, %mm1
+	movq %mm4, 8*15(%esi)		/* out15 */
+/* moved from the next block */
+	punpcklwd %mm7, %mm0
+/* transpose - M2 parts
+ * moved up to the prev block
+ *	movq 8(%esi), %mm0
+ *	movq 8*3(%esi), %mm7
+ *	movq %mm0, %mm1
+ *	punpcklwd %mm7, %mm0
+ */
+	movq 8*5(%esi), %mm5
+	punpckhwd %mm7, %mm1
+	movq 8*7(%esi), %mm4
+	movq %mm5, %mm3
+/* shuffle the data and write the lower parts of the trasposed in 4 dwords */
+	movd %mm0, 8*8(%esi)		/* LS part of tmt8 */
+	punpcklwd %mm4, %mm5
+	movd %mm1, 8*12(%esi)		/* LS part of tmt12 */
+	punpckhwd %mm4, %mm3
+	movd %mm5, 8*8+4(%esi)		/* MS part of tmt8 */
+	punpckhdq %mm5, %mm0		/* tmt10 */
+	movd %mm3, 8*12+4(%esi)		/* MS part of tmt12 */
+	punpckhdq %mm3, %mm1		/* tmt14 */
+/* transpose - M1 parts */
+	movq (%esi), %mm7
+	movq 8*2(%esi), %mm2
+	movq %mm7, %mm6
+	movq 8*4(%esi), %mm5
+	punpcklwd %mm2, %mm7
+	movq 8*6(%esi), %mm4
+	punpckhwd %mm2, %mm6		/* free mm2 */
+	movq %mm5, %mm3
+	punpcklwd %mm4, %mm5
+	punpckhwd %mm4, %mm3		/* free mm4 */
+	movq %mm7, %mm2
+	movq %mm6, %mm4
+	punpckldq %mm5, %mm7		/* tmt0 */
+	punpckhdq %mm5, %mm2		/* tmt2 ; free mm5 */
+/* shuffle the rest of the data, and write it with 2 mmword writes */
+	punpckldq %mm3, %mm6		/* tmt4 */
+/* moved from next block */
+	movq %mm2, %mm5			/* duplicate tmt2 */
+	punpckhdq %mm3, %mm4		/* tmt6 ; free mm3 */
+/* moved from next block */
+	movq %mm0, %mm3			/* duplicate tmt10 */
+/* column 0: odd part (after transpose)
+ *moved up to prev block
+ *	movq %mm0, %mm3			duplicate tmt10
+ *	movq %mm2, %mm5			duplicate tmt2
+ */
+	psubsw %mm4, %mm0		/* V110 */
+	paddsw %mm4, %mm3		/* V113 ; free mm4 */
+	movq %mm0, %mm4			/* duplicate V110 */
+	paddsw %mm1, %mm2		/* V111 */
+	pmulhw x539f539f539f539f, %mm0	/* 21407-> V117 */
+	psubsw %mm1, %mm5		/* V112 ; free mm1 */
+	psubsw %mm5, %mm4		/* V116 */
+	movq %mm2, %mm1			/* duplicate V111 */
+	pmulhw x4546454645464546, %mm5	/* 17734-> V119 */
+	psubsw %mm3, %mm2		/* V114 */
+	pmulhw x61f861f861f861f8, %mm4	/* 25080-> V120 */
+	paddsw %mm3, %mm1		/* V115 ; free mm3 */
+	pmulhw x5a825a825a825a82, %mm2	/* 23170-> V118 */
+	psllw $2, %mm0			/* t266 */
+	movq %mm1, (%esi)		/* save V115 */
+	psllw $1, %mm5			/* t268 */
+	psubsw %mm4, %mm5		/* V122 */
+	psubsw %mm0, %mm4		/* V121 ; free mm0 */
+	psllw $1, %mm5			/* t270 */
+	psubsw %mm1, %mm5		/* V123 ; free mm1 */
+	psllw $2, %mm2			/* t272 */
+	psubsw %mm5, %mm2		/* V124 (keep V123) */
+	psllw $1, %mm4			/* t274 */
+	movq %mm5, 8*2(%esi)		/* save V123 ; free mm5 */
+	paddsw %mm2, %mm4		/* V125 (keep V124) */
+/* column 0: even part (after transpose) */
+	movq 8*12(%esi), %mm0		/* tmt12 */
+	movq %mm6, %mm3			/* duplicate tmt4 */
+	psubsw %mm0, %mm6		/* V100 */
+	paddsw %mm0, %mm3		/* V101 ; free mm0 */
+	pmulhw x5a825a825a825a82, %mm6	/* 23170 ->V102 */
+	movq %mm7, %mm5			/* duplicate tmt0 */
+	movq 8*8(%esi), %mm1		/* tmt8 */
+	paddsw %mm1, %mm7		/* V103 */
+	psubsw %mm1, %mm5		/* V104 ; free mm1 */
+	movq %mm7, %mm0			/* duplicate V103 */
+	psllw $2, %mm6			/* t245 */
+	paddsw %mm3, %mm7		/* V106 */
+	movq %mm5, %mm1			/* duplicate V104 */
+	psubsw %mm3, %mm6		/* V105 */
+	psubsw %mm3, %mm0		/* V109; free mm3 */
+	paddsw %mm6, %mm5		/* V107 */
+	psubsw %mm6, %mm1		/* V108 ; free mm6 */
+/* column 0: output butterfly (after transform) */
+	movq %mm1, %mm3			/* duplicate V108 */
+	paddsw %mm2, %mm1		/* out4 */
+	psraw $4, %mm1
+	psubsw %mm2, %mm3		/* out10 ; free mm2 */
+	psraw $4, %mm3
+	movq %mm0, %mm6			/* duplicate V109 */
+	movq %mm1, 8*4(%esi)		/* out4 ; free mm1 */
+	psubsw %mm4, %mm0		/* out6 */
+	movq %mm3, 8*10(%esi)		/* out10 ; free mm3 */
+	psraw $4, %mm0
+	paddsw %mm4, %mm6		/* out8 ; free mm4 */
+	movq %mm7, %mm1			/* duplicate V106 */
+	movq %mm0, 8*6(%esi)		/* out6 ; free mm0 */
+	psraw $4, %mm6
+	movq (%esi), %mm4		/* V115 */
+	movq %mm6, 8*8(%esi)		/* out8 ; free mm6 */
+	movq %mm5, %mm2			/* duplicate V107 */
+	movq 8*2(%esi), %mm3		/* V123 */
+	paddsw %mm4, %mm7		/* out0 */
+/* moved up from next block */
+	movq scratch3, %mm0
+	psraw $4, %mm7
+/* moved up from next block */
+	movq scratch5, %mm6 
+	psubsw %mm4, %mm1		/* out14 ; free mm4 */
+	paddsw %mm3, %mm5		/* out2 */
+	psraw $4, %mm1
+	movq %mm7, (%esi)		/* out0 ; free mm7 */
+	psraw $4, %mm5
+	movq %mm1, 8*14(%esi)		/* out14 ; free mm1 */
+	psubsw %mm3, %mm2		/* out12 ; free mm3 */
+	movq %mm5, 8*2(%esi)		/* out2 ; free mm5 */
+	psraw $4, %mm2
+/* moved up to the prev block */
+	movq scratch7, %mm4
+/* moved up to the prev block */
+	psraw $4, %mm0
+	movq %mm2, 8*12(%esi)		/* out12 ; free mm2 */
+/* moved up to the prev block */
+	psraw $4, %mm6
+/* move back the data to its correct place
+* moved up to the prev block
+ *	movq scratch3, %mm0
+ *	movq scratch5, %mm6
+ *	movq scratch7, %mm4
+ *	psraw $4, %mm0
+ *	psraw $4, %mm6
+*/
+	movq scratch1, %mm1
+	psraw $4, %mm4
+	movq %mm0, 8*3(%esi)		/* out3 */
+	psraw $4, %mm1
+	movq %mm6, 8*5(%esi)		/* out5 */
+	movq %mm4, 8*7(%esi)		/* out7 */
+	movq %mm1, 8(%esi)		/* out1 */
+	emms
+	popl %edi
+	popl %esi
+	popl %edx
+	popl %ecx
+	popl %ebx
+	movl %ebp,%esp
+	popl %ebp
+	ret
+.Lfe1:
+	.size	 gst_idct_mmx_idct,.Lfe1-gst_idct_mmx_idct

libs/putbits/gstputbits.c

@@ -63,7 +63,7 @@ void gst_putbits(gst_putbits_t *pb, int val, int n)
   int i;
   unsigned int mask;
 
-  //printf("putbits: %d %d %ld\n", val, n, pb->outcnt);
+  //printf("putbits: %d %d %ld %ld\n", val, n, pb->outcnt, pb->newlen);
   mask = 1 << (n-1); /* selects first (leftmost) bit */
 
   for (i=0; i<n; i++)

libs/riff/gstriff.h

@@ -71,6 +71,7 @@
 #define GST_RIFF_FCC_auds MAKE_FOUR_CC('a','u','d','s')
 #define GST_RIFF_FCC_pads MAKE_FOUR_CC('p','a','d','s')
 #define GST_RIFF_FCC_txts MAKE_FOUR_CC('t','x','t','s')
+#define GST_RIFF_FCC_vidc MAKE_FOUR_CC('v','i','d','c')
 /* fcc handlers */
 #define GST_RIFF_FCCH_RLE  MAKE_FOUR_CC('R','L','E',' ')
 #define GST_RIFF_FCCH_msvc MAKE_FOUR_CC('m','s','v','c')
@@ -184,6 +185,8 @@
 #define GST_RIFF_vyuy MAKE_FOUR_CC( 'v', 'y', 'u', 'y')
 #define GST_RIFF_VYUY MAKE_FOUR_CC( 'V', 'Y', 'U', 'Y')
 
+#define GST_RIFF_DIV3 MAKE_FOUR_CC( 'D', 'I', 'V', '3')
+
 #define GST_RIFF_rpza MAKE_FOUR_CC( 'r', 'p', 'z', 'a')
 /* And this here's the mistakes that need to be supported */
 #define GST_RIFF_azpr MAKE_FOUR_CC( 'a', 'z', 'p', 'r')  /* recognize Apple's rpza mangled? */
@@ -273,10 +276,14 @@ struct _gst_riff_strf_auds {       /* == WaveHeader (?) */
 #define GST_RIFF_WAVE_FORMAT_YAMAHA_ADPCM   (0x0020)
 #define GST_RIFF_WAVE_FORMAT_DSP_TRUESPEECH (0x0022)
 #define GST_RIFF_WAVE_FORMAT_GSM610         (0x0031)
-#define GST_RIFF_WAVE_FORMAT_MPEG           (0x0055)
+#define GST_RIFF_WAVE_FORMAT_MSN            (0x0032)
+#define GST_RIFF_WAVE_FORMAT_MPEGL12        (0x0050)
+#define GST_RIFF_WAVE_FORMAT_MPEGL3         (0x0055)
 #define GST_RIFF_IBM_FORMAT_MULAW           (0x0101)
 #define GST_RIFF_IBM_FORMAT_ALAW            (0x0102)
 #define GST_RIFF_IBM_FORMAT_ADPCM           (0x0103)
+#define GST_RIFF_WAVE_FORMAT_DIVX           (0x0160)
+#define GST_RIFF_WAVE_FORMAT_divx           (0x0161)
   guint16 channels;
   guint32 rate;
   guint32 av_bps;
@@ -360,6 +367,7 @@ gint gst_riff_encoder_chunk(GstRiff *riff, guint32 chunk_type, void *chunk, gulo
 
 GstBuffer *gst_riff_encoder_get_buffer(GstRiff *riff);
 GstBuffer *gst_riff_encoder_get_and_reset_buffer(GstRiff *riff);
+
 /* from gstriffutil.c */
 gulong gst_riff_fourcc_to_id(gchar *fourcc);
 gchar *gst_riff_id_to_fourcc(gulong id);

libs/riff/gstriffencode.c

@@ -18,20 +18,19 @@
  */
 
 
+//#define DEBUG_ENABLED
+#include <gst/gst.h>
 #include <gstriff.h>
 
 #define GST_RIFF_ENCODER_BUF_SIZE    1024
 
-//#define debug(format,args...) g_print(format,##args)
-#define debug(format,args...)
-
 #define ADD_CHUNK(riffenc, chunkid, chunksize) \
 { \
   gst_riff_chunk *chunk;\
   chunk = (gst_riff_chunk *)(riffenc->dataleft + riffenc->nextlikely);\
   chunk->id = chunkid; \
   chunk->size = chunksize; \
-  riffenc->nextlikely += sizeof(gst_riff_chunk); \
+  riffenc->nextlikely += sizeof(gst_riff_chunk) + (chunksize&1); \
 }
 
 #define ADD_LIST(riffenc, listsize, listtype) \
@@ -49,7 +48,7 @@ GstRiff *gst_riff_encoder_new(guint32 type) {
   GstRiff *riff;
   gst_riff_list *list;
 
-  debug("gst_riff_encoder: making %4.4s encoder\n", (char *)&type);
+  DEBUG("gst_riff_encoder: making %4.4s encoder\n", (char *)&type);
   riff = (GstRiff *)g_malloc(sizeof(GstRiff));
   g_return_val_if_fail(riff != NULL, NULL);
 
@@ -77,7 +76,7 @@ gint gst_riff_encoder_avih(GstRiff *riff, gst_riff_avih *head, gulong size) {
 
   g_return_val_if_fail(riff->state == GST_RIFF_STATE_INITIAL, GST_RIFF_EINVAL);
 
-  debug("gst_riff_encoder: add avih\n");
+  DEBUG("gst_riff_encoder: add avih\n");
 
   ADD_LIST(riff, 0xB8, GST_RIFF_LIST_hdrl);
 
@@ -97,7 +96,7 @@ gint gst_riff_encoder_strh(GstRiff *riff, guint32 fcc_type, gst_riff_strh *head,
   g_return_val_if_fail(riff->state == GST_RIFF_STATE_HASAVIH ||
 		       riff->state == GST_RIFF_STATE_HASSTRF, GST_RIFF_EINVAL);
 
-  debug("gst_riff_encoder: add strh type %08x (%4.4s)\n", fcc_type, (char *)&fcc_type);
+  DEBUG("gst_riff_encoder: add strh type %08x (%4.4s)\n", fcc_type, (char *)&fcc_type);
 
   ADD_LIST(riff, 108, GST_RIFF_LIST_strl);
 
@@ -118,7 +117,7 @@ gint gst_riff_encoder_strf(GstRiff *riff, void *format, gulong size) {
 
   g_return_val_if_fail(riff->state == GST_RIFF_STATE_HASSTRH, GST_RIFF_EINVAL);
 
-  debug("gst_riff_encoder: add strf\n");
+  DEBUG("gst_riff_encoder: add strf\n");
 
   ADD_CHUNK(riff, GST_RIFF_TAG_strf, size);
 
@@ -141,14 +140,14 @@ gint gst_riff_encoder_chunk(GstRiff *riff, guint32 chunk_type, void *chunkdata,
     riff->state = GST_RIFF_STATE_MOVI;
   }
 
-  debug("gst_riff_encoder: add chunk type %08x (%4.4s)\n", chunk_type, (char *)&chunk_type);
+  DEBUG("gst_riff_encoder: add chunk type %08x (%4.4s)\n", chunk_type, (char *)&chunk_type);
   
   ADD_CHUNK(riff, chunk_type, size);
 
   if (chunkdata != NULL) {
     chunk = (gst_riff_chunk *)(riff->dataleft + riff->nextlikely);
     memcpy(chunk, chunkdata, size);
-    riff->nextlikely += size;
+    riff->nextlikely += size + (size&1);
   }
 
   return GST_RIFF_OK;

libs/winloader/Makefile.am

@@ -2,7 +2,7 @@ filterdir = $(libdir)/gst
 
 filter_LTLIBRARIES = libwinloader.la
 
-libwinloader_la_SOURCES = driver.c elfdll.c ext.c externals.c module.c pe_image.c pe_resource.c registry.c resource.c stubs.s vfl.c
+libwinloader_la_SOURCES = driver.c elfdll.c ext.c externals.c module.c pe_image.c pe_resource.c registry.c resource.c stubs.s vfl.c afl.c
 
 libwinloaderincludedir = $(includedir)/gst/libs/winloader.h
 libwinloaderinclude_HEADERS = 

libs/winloader/afl.c

@@ -0,0 +1,758 @@
+/**************************************************************************
+
+
+  This file will contain an interface to ACM drivers.
+  Its content will be based mainly on wine/dlls/msacm32
+  actually, for audio decompression only the following functions
+  are needed:
+  
+  acmStreamOpen ( takes formats of src and dest, returns stream handle )
+  acmStreamPrepareHeader ( takes stream handler and info on data )
+  acmStreamConvert ( the same as PrepareHeader )
+  acmStreamUnprepareHeader
+  acmStreamClose
+  acmStreamSize
+  maybe acmStreamReset
+  
+  In future I'll also add functions for format enumeration, 
+  but not right now.
+
+  
+***************************************************************************/
+#include <stdio.h>
+#include <string.h>
+
+#include <wine/winbase.h>
+#include <wine/windef.h>
+#include <wine/winuser.h>
+#include <wine/vfw.h>
+#include <wine/winestring.h>
+#include <wine/driver.h>
+#include <wine/winerror.h>
+#include <wine/msacm.h>
+#include <wine/msacmdrv.h>
+#include "wineacm.h"
+#pragma pack(1)
+#define OpenDriverA DrvOpen
+extern HDRVR VFWAPI DrvOpen(long);
+#define CloseDriver DrvClose
+extern HDRVR VFWAPI DrvClose(long);
+
+static PWINE_ACMSTREAM	ACM_GetStream(HACMSTREAM has)
+{
+    return (PWINE_ACMSTREAM)has;
+}
+
+/***********************************************************************
+ *           acmDriverAddA (MSACM32.2)
+ */
+MMRESULT WINAPI acmDriverAddA(PHACMDRIVERID phadid, HINSTANCE hinstModule,
+			      LPARAM lParam, DWORD dwPriority, DWORD fdwAdd)
+{
+    if (!phadid)
+	return MMSYSERR_INVALPARAM;
+    
+    /* Check if any unknown flags */
+    if (fdwAdd & 
+	~(ACM_DRIVERADDF_FUNCTION|ACM_DRIVERADDF_NOTIFYHWND|
+	  ACM_DRIVERADDF_GLOBAL))
+	return MMSYSERR_INVALFLAG;
+    
+    /* Check if any incompatible flags */
+    if ((fdwAdd & ACM_DRIVERADDF_FUNCTION) && 
+	(fdwAdd & ACM_DRIVERADDF_NOTIFYHWND))
+	return MMSYSERR_INVALFLAG;
+    
+    /* FIXME: in fact, should GetModuleFileName(hinstModule) and do a 
+     * LoadDriver on it, to be sure we can call SendDriverMessage on the
+     * hDrvr handle.
+     */
+    *phadid = (HACMDRIVERID) MSACM_RegisterDriver(NULL, NULL, hinstModule);
+    
+    /* FIXME: lParam, dwPriority and fdwAdd ignored */
+    
+    return MMSYSERR_NOERROR;
+}
+
+/***********************************************************************
+ *           acmDriverClose (MSACM32.4)
+ */
+MMRESULT WINAPI acmDriverClose(HACMDRIVER had, DWORD fdwClose)
+{
+    PWINE_ACMDRIVER  p;
+    PWINE_ACMDRIVER* tp;
+    
+    if (fdwClose)
+	return MMSYSERR_INVALFLAG;
+    
+    p = MSACM_GetDriver(had);
+    if (!p)
+	return MMSYSERR_INVALHANDLE;
+
+    for (tp = &(p->obj.pACMDriverID->pACMDriverList); *tp; *tp = (*tp)->pNextACMDriver) {
+	if (*tp == p) {
+	    *tp = (*tp)->pNextACMDriver;
+	    break;
+	}
+    }
+    
+    if (p->hDrvr && !p->obj.pACMDriverID->pACMDriverList)
+	CloseDriver(p->hDrvr);
+    
+    HeapFree(MSACM_hHeap, 0, p);
+    
+    return MMSYSERR_NOERROR;
+}
+
+/***********************************************************************
+ *           acmDriverEnum (MSACM32.7)
+ */
+MMRESULT WINAPI acmDriverEnum(ACMDRIVERENUMCB fnCallback, DWORD dwInstance, DWORD fdwEnum)
+{
+    PWINE_ACMDRIVERID	p;
+    DWORD		fdwSupport;
+
+    if (!fnCallback) {
+	return MMSYSERR_INVALPARAM;
+    }
+    
+    if (fdwEnum && ~(ACM_DRIVERENUMF_NOLOCAL|ACM_DRIVERENUMF_DISABLED)) {
+	return MMSYSERR_INVALFLAG;
+    }
+    
+    for (p = MSACM_pFirstACMDriverID; p; p = p->pNextACMDriverID) {
+	fdwSupport = ACMDRIVERDETAILS_SUPPORTF_CODEC;
+	if (!p->bEnabled) {
+	    if (fdwEnum & ACM_DRIVERENUMF_DISABLED)
+		fdwSupport |= ACMDRIVERDETAILS_SUPPORTF_DISABLED;
+	    else
+		continue;
+	}
+	(*fnCallback)((HACMDRIVERID) p, dwInstance, fdwSupport);
+    }
+    
+    return MMSYSERR_NOERROR;
+}
+
+/***********************************************************************
+ *           acmDriverID (MSACM32.8)
+ */
+MMRESULT WINAPI acmDriverID(HACMOBJ hao, PHACMDRIVERID phadid, DWORD fdwDriverID)
+{
+    PWINE_ACMOBJ pao;
+    
+    pao = MSACM_GetObj(hao);
+    if (!pao)
+	return MMSYSERR_INVALHANDLE;
+    
+    if (!phadid)
+	return MMSYSERR_INVALPARAM;
+    
+    if (fdwDriverID)
+	return MMSYSERR_INVALFLAG;
+    
+    *phadid = (HACMDRIVERID) pao->pACMDriverID;
+    
+    return MMSYSERR_NOERROR;
+}
+
+/***********************************************************************
+ *           acmDriverMessage (MSACM32.9)
+ * FIXME
+ *   Not implemented
+ */
+LRESULT WINAPI acmDriverMessage(HACMDRIVER had, UINT uMsg, LPARAM lParam1, LPARAM lParam2)
+{
+    PWINE_ACMDRIVER pad = MSACM_GetDriver(had);
+    if (!pad)
+	return MMSYSERR_INVALPARAM;
+    
+    /* FIXME: Check if uMsg legal */
+    
+    if (!SendDriverMessage(pad->hDrvr, uMsg, lParam1, lParam2))
+	return MMSYSERR_NOTSUPPORTED;
+    
+    return MMSYSERR_NOERROR;
+}
+
+
+/***********************************************************************
+ *           acmDriverOpen (MSACM32.10)
+ */
+MMRESULT WINAPI acmDriverOpen(PHACMDRIVER phad, HACMDRIVERID hadid, DWORD fdwOpen)
+{
+    PWINE_ACMDRIVERID	padid;
+    PWINE_ACMDRIVER	pad;
+    ICOPEN		icopen;
+    HDRVR		hdrv;
+
+
+
+    TRACE("(%p, %x, %08lu)\n", phad, hadid, fdwOpen);
+
+    if (!phad)
+	return MMSYSERR_INVALPARAM;
+    
+    padid = MSACM_GetDriverID(hadid); 
+    if (!padid)
+	return MMSYSERR_INVALHANDLE;
+    
+    if (fdwOpen)
+	return MMSYSERR_INVALFLAG;
+    
+    pad = HeapAlloc(MSACM_hHeap, 0, sizeof(WINE_ACMDRIVER));
+    if (!pad) return MMSYSERR_NOMEM;
+
+    pad->obj.pACMDriverID = padid;
+    icopen.fccType		= mmioFOURCC('a', 'u', 'd', 'c');
+    icopen.fccHandler		= (long)padid->pszFileName;
+    icopen.dwSize		= sizeof(ICOPEN);
+    icopen.dwFlags		= 0;
+
+   if (!padid->hInstModule)
+	pad->hDrvr = OpenDriverA((long)&icopen);
+    else
+	pad->hDrvr = padid->hInstModule;
+    
+    if (!pad->hDrvr) {
+	HeapFree(MSACM_hHeap, 0, pad);
+	return MMSYSERR_ERROR;
+    }
+    
+    pad->pfnDriverProc = GetProcAddress(pad->hDrvr, "DriverProc");
+
+    /* insert new pad at beg of list */
+    pad->pNextACMDriver = padid->pACMDriverList;
+    padid->pACMDriverList = pad;
+
+    /* FIXME: Create a WINE_ACMDRIVER32 */
+    *phad = (HACMDRIVER)pad;
+    
+    return MMSYSERR_NOERROR;
+}
+
+/***********************************************************************
+ *           acmDriverRemove (MSACM32.12)
+ */
+MMRESULT WINAPI acmDriverRemove(HACMDRIVERID hadid, DWORD fdwRemove)
+{
+    PWINE_ACMDRIVERID padid;
+    
+    padid = MSACM_GetDriverID(hadid);
+    if (!padid)
+	return MMSYSERR_INVALHANDLE;
+    
+    if (fdwRemove)
+	return MMSYSERR_INVALFLAG;
+    
+    MSACM_UnregisterDriver(padid);
+    
+    return MMSYSERR_NOERROR;
+}
+
+
+
+/**********************************************************************/
+
+HANDLE MSACM_hHeap = (HANDLE) NULL;
+PWINE_ACMDRIVERID MSACM_pFirstACMDriverID = NULL;
+PWINE_ACMDRIVERID MSACM_pLastACMDriverID = NULL;
+
+/***********************************************************************
+ *           MSACM_RegisterDriver32() 
+ */
+PWINE_ACMDRIVERID MSACM_RegisterDriver(LPSTR pszDriverAlias, LPSTR pszFileName,
+				       HINSTANCE hinstModule)
+//
+// File names are stored in driver.c. I reuse this variable to store driver ID
+// in it. If it's <0x10000, it is primary codec for corresponding format.
+//				       
+{ 
+    PWINE_ACMDRIVERID padid;
+
+    TRACE("('%s', '%x', 0x%08x)\n", pszDriverAlias, pszFileName, hinstModule);
+
+    padid = (PWINE_ACMDRIVERID) HeapAlloc(MSACM_hHeap, 0, sizeof(WINE_ACMDRIVERID));
+    padid->pszDriverAlias = (char*)malloc(strlen(pszDriverAlias)+1);
+    strcpy(padid->pszDriverAlias, pszDriverAlias);
+//    1~strdup(pszDriverAlias);
+    padid->pszFileName = pszFileName;
+    padid->hInstModule = hinstModule;
+    padid->bEnabled = TRUE;
+    padid->pACMDriverList = NULL;
+    padid->pNextACMDriverID = NULL;
+    padid->pPrevACMDriverID = MSACM_pLastACMDriverID;
+    if (MSACM_pLastACMDriverID)
+	MSACM_pLastACMDriverID->pNextACMDriverID = padid;
+    MSACM_pLastACMDriverID = padid;
+    if (!MSACM_pFirstACMDriverID)
+	MSACM_pFirstACMDriverID = padid;
+    
+    return padid;
+}
+
+/***********************************************************************
+ *           MSACM_RegisterAllDrivers32() 
+ */
+void MSACM_RegisterAllDrivers(void)
+{
+    LPSTR pszBuffer;
+    DWORD dwBufferLength;
+    
+    if (MSACM_pFirstACMDriverID)
+	return;
+    
+    MSACM_RegisterDriver("divxa32", (LPSTR)0x161, 0);
+    MSACM_RegisterDriver("msadp32", (LPSTR)0x2, 0);
+}
+
+/***********************************************************************
+ *           MSACM_UnregisterDriver32()
+ */
+PWINE_ACMDRIVERID MSACM_UnregisterDriver(PWINE_ACMDRIVERID p)
+{
+    PWINE_ACMDRIVERID pNextACMDriverID;
+    
+    while (p->pACMDriverList)
+	acmDriverClose((HACMDRIVER) p->pACMDriverList, 0);
+    
+    if (p->pszDriverAlias)
+	HeapFree(MSACM_hHeap, 0, p->pszDriverAlias);
+//    if (p->pszFileName)
+//	HeapFree(MSACM_hHeap, 0, p->pszFileName);
+    
+    if (p == MSACM_pFirstACMDriverID)
+	MSACM_pFirstACMDriverID = p->pNextACMDriverID;
+    if (p == MSACM_pLastACMDriverID)
+	MSACM_pLastACMDriverID = p->pPrevACMDriverID;
+
+    if (p->pPrevACMDriverID)
+	p->pPrevACMDriverID->pNextACMDriverID = p->pNextACMDriverID;
+    if (p->pNextACMDriverID)
+	p->pNextACMDriverID->pPrevACMDriverID = p->pPrevACMDriverID;
+    
+    pNextACMDriverID = p->pNextACMDriverID;
+    
+    HeapFree(MSACM_hHeap, 0, p);
+    
+    return pNextACMDriverID;
+}
+
+/***********************************************************************
+ *           MSACM_UnregisterAllDrivers32()
+ * FIXME
+ *   Where should this function be called?
+ */
+void MSACM_UnregisterAllDrivers(void)
+{
+    PWINE_ACMDRIVERID p;
+
+    for (p = MSACM_pFirstACMDriverID; p; p = MSACM_UnregisterDriver(p));
+}
+
+/***********************************************************************
+ *           MSACM_GetDriverID32() 
+ */
+PWINE_ACMDRIVERID MSACM_GetDriverID(HACMDRIVERID hDriverID)
+{
+    return (PWINE_ACMDRIVERID)hDriverID;
+}
+
+/***********************************************************************
+ *           MSACM_GetDriver32()
+ */
+PWINE_ACMDRIVER MSACM_GetDriver(HACMDRIVER hDriver)
+{
+    return (PWINE_ACMDRIVER)hDriver;
+}
+
+/***********************************************************************
+ *           MSACM_GetObj32()
+ */
+PWINE_ACMOBJ MSACM_GetObj(HACMOBJ hObj)
+{
+    return (PWINE_ACMOBJ)hObj;
+}
+
+
+
+/***********************************************************************
+ *           acmStreamOpen (MSACM32.40)
+ */
+MMRESULT WINAPI acmStreamOpen(PHACMSTREAM phas, HACMDRIVER had, PWAVEFORMATEX pwfxSrc,
+			      PWAVEFORMATEX pwfxDst, PWAVEFILTER pwfltr, DWORD dwCallback,
+			      DWORD dwInstance, DWORD fdwOpen)
+{
+    PWINE_ACMSTREAM	was;
+    PWINE_ACMDRIVER	wad;
+    MMRESULT		ret;
+    int			wfxSrcSize;
+    int			wfxDstSize;
+    
+    TRACE("(%p, 0x%08x, %p, %p, %p, %ld, %ld, %ld)\n",
+	  phas, had, pwfxSrc, pwfxDst, pwfltr, dwCallback, dwInstance, fdwOpen);
+
+    TRACE("src [wFormatTag=%u, nChannels=%u, nSamplesPerSec=%lu, nAvgBytesPerSec=%lu, nBlockAlign=%u, wBitsPerSample=%u, cbSize=%u]\n", 
+	  pwfxSrc->wFormatTag, pwfxSrc->nChannels, pwfxSrc->nSamplesPerSec, pwfxSrc->nAvgBytesPerSec, 
+	  pwfxSrc->nBlockAlign, pwfxSrc->wBitsPerSample, pwfxSrc->cbSize);
+
+    TRACE("dst [wFormatTag=%u, nChannels=%u, nSamplesPerSec=%lu, nAvgBytesPerSec=%lu, nBlockAlign=%u, wBitsPerSample=%u, cbSize=%u]\n", 
+	  pwfxDst->wFormatTag, pwfxDst->nChannels, pwfxDst->nSamplesPerSec, pwfxDst->nAvgBytesPerSec, 
+	  pwfxDst->nBlockAlign, pwfxDst->wBitsPerSample, pwfxDst->cbSize);
+
+#define SIZEOF_WFX(wfx) (sizeof(WAVEFORMATEX) + ((wfx->wFormatTag == WAVE_FORMAT_PCM) ? 0 : wfx->cbSize))
+    wfxSrcSize = SIZEOF_WFX(pwfxSrc);
+    wfxDstSize = SIZEOF_WFX(pwfxDst);
+#undef SIZEOF_WFX
+
+    was = HeapAlloc(MSACM_hHeap, 0, sizeof(*was) + wfxSrcSize + wfxDstSize + ((pwfltr) ? sizeof(WAVEFILTER) : 0));
+    if (was == NULL)
+	return MMSYSERR_NOMEM;
+    
+    was->drvInst.cbStruct = sizeof(was->drvInst);
+    was->drvInst.pwfxSrc = (PWAVEFORMATEX)((LPSTR)was + sizeof(*was));
+    memcpy(was->drvInst.pwfxSrc, pwfxSrc, wfxSrcSize);
+    was->drvInst.pwfxDst = (PWAVEFORMATEX)((LPSTR)was + sizeof(*was) + wfxSrcSize);
+    memcpy(was->drvInst.pwfxDst, pwfxDst, wfxDstSize);
+    if (pwfltr) {
+	was->drvInst.pwfltr = (PWAVEFILTER)((LPSTR)was + sizeof(*was) + wfxSrcSize + wfxDstSize);
+	memcpy(was->drvInst.pwfltr, pwfltr, sizeof(WAVEFILTER));
+    } else {
+	was->drvInst.pwfltr = NULL;
+    }
+    was->drvInst.dwCallback = dwCallback;    
+    was->drvInst.dwInstance = dwInstance;
+    was->drvInst.fdwOpen = fdwOpen;
+    was->drvInst.fdwDriver = 0L;  
+    was->drvInst.dwDriver = 0L;     
+    was->drvInst.has = (HACMSTREAM)was;
+    
+    if (had) {
+	if (!(wad = MSACM_GetDriver(had))) {
+	    ret = MMSYSERR_INVALPARAM;
+	    goto errCleanUp;
+	}
+	
+	was->obj.pACMDriverID = wad->obj.pACMDriverID;
+	was->pDrv = wad;
+	was->hAcmDriver = 0; /* not to close it in acmStreamClose */
+
+	ret = SendDriverMessage(wad->hDrvr, ACMDM_STREAM_OPEN, (DWORD)&was->drvInst, 0L);
+	if (ret != MMSYSERR_NOERROR)
+	    goto errCleanUp;
+    } else {
+	PWINE_ACMDRIVERID wadi;
+	short drv_tag;
+	ret = ACMERR_NOTPOSSIBLE;
+/*	if(pwfxSrc->wFormatTag==1)//compression
+	    drv_tag=pwfxDst->wFormatTag;
+	    else
+	    if(pwfxDst->wFormatTag==1)//decompression
+		drv_tag=pwfxSrc->wFormatTag;
+		else
+		goto errCleanUp;
+
+	    ret=acmDriverOpen2(drv_tag); 
+	    if (ret == MMSYSERR_NOERROR) {
+		if ((wad = MSACM_GetDriver(had)) != 0) {
+		    was->obj.pACMDriverID = wad->obj.pACMDriverID;
+		    was->pDrv = wad;
+		    was->hAcmDriver = had;
+		    
+		    ret = SendDriverMessage(wad->hDrvr, ACMDM_STREAM_OPEN, (DWORD)&was->drvInst, 0L);
+		    if (ret == MMSYSERR_NOERROR) {
+			if (fdwOpen & ACM_STREAMOPENF_QUERY) {
+			    acmDriverClose(had, 0L);
+			}
+			break;
+		    }
+		}
+		acmDriverClose(had, 0L);*/
+	if(MSACM_pFirstACMDriverID==NULL)
+	    MSACM_RegisterAllDrivers();	
+	
+	for (wadi = MSACM_pFirstACMDriverID; wadi; wadi = wadi->pNextACMDriverID) {
+	    ret = acmDriverOpen(&had, (HACMDRIVERID)wadi, 0L);
+	    if (ret == MMSYSERR_NOERROR) {
+		if ((wad = MSACM_GetDriver(had)) != 0) {
+		    was->obj.pACMDriverID = wad->obj.pACMDriverID;
+		    was->pDrv = wad;
+		    was->hAcmDriver = had;
+		    
+		    ret = SendDriverMessage(wad->hDrvr, ACMDM_STREAM_OPEN, (DWORD)&was->drvInst, 0L);
+		    if (ret == MMSYSERR_NOERROR) {
+			if (fdwOpen & ACM_STREAMOPENF_QUERY) {
+			    acmDriverClose(had, 0L);
+			}
+			break;
+		    }
+		}
+		// no match, close this acm driver and try next one 
+		acmDriverClose(had, 0L);
+	    }
+	}
+	if (ret != MMSYSERR_NOERROR) {
+	    ret = ACMERR_NOTPOSSIBLE;
+	    goto errCleanUp;
+	}
+    }
+    ret = MMSYSERR_NOERROR;
+    if (!(fdwOpen & ACM_STREAMOPENF_QUERY)) {
+	if (phas)
+	    *phas = (HACMSTREAM)was;
+	TRACE("=> (%d)\n", ret);
+	return ret;
+    }
+errCleanUp:		
+    if (phas)
+	*phas = (HACMSTREAM)0;
+    HeapFree(MSACM_hHeap, 0, was);
+    TRACE("=> (%d)\n", ret);
+    return ret;
+}
+
+
+MMRESULT WINAPI acmStreamClose(HACMSTREAM has, DWORD fdwClose)
+{
+    PWINE_ACMSTREAM	was;
+    MMRESULT		ret;
+		
+    TRACE("(0x%08x, %ld)\n", has, fdwClose);
+    
+    if ((was = ACM_GetStream(has)) == NULL) {
+	return MMSYSERR_INVALHANDLE;
+    }
+    ret = SendDriverMessage(was->pDrv->hDrvr, ACMDM_STREAM_CLOSE, (DWORD)&was->drvInst, 0);
+    if (ret == MMSYSERR_NOERROR) {
+	if (was->hAcmDriver)
+	    acmDriverClose(was->hAcmDriver, 0L);	
+	HeapFree(MSACM_hHeap, 0, was);
+    }
+    TRACE("=> (%d)\n", ret);
+    return ret;
+}
+
+/***********************************************************************
+ *           acmStreamConvert (MSACM32.38)
+ */
+MMRESULT WINAPI acmStreamConvert(HACMSTREAM has, PACMSTREAMHEADER pash, 
+				 DWORD fdwConvert)
+{
+    PWINE_ACMSTREAM	was;
+    MMRESULT		ret = MMSYSERR_NOERROR;
+    PACMDRVSTREAMHEADER	padsh;
+
+    TRACE("(0x%08x, %p, %ld)\n", has, pash, fdwConvert);
+    
+    if ((was = ACM_GetStream(has)) == NULL)
+	return MMSYSERR_INVALHANDLE;
+    if (!pash || pash->cbStruct < sizeof(ACMSTREAMHEADER))
+	return MMSYSERR_INVALPARAM;
+
+    if (!(pash->fdwStatus & ACMSTREAMHEADER_STATUSF_PREPARED))
+	return ACMERR_UNPREPARED;
+
+    /* Note: the ACMSTREAMHEADER and ACMDRVSTREAMHEADER structs are of same
+     * size. some fields are private to msacm internals, and are exposed
+     * in ACMSTREAMHEADER in the dwReservedDriver array
+     */
+    padsh = (PACMDRVSTREAMHEADER)pash;
+
+    /* check that pointers have not been modified */
+    if (padsh->pbPreparedSrc != padsh->pbSrc ||
+	padsh->cbPreparedSrcLength < padsh->cbSrcLength ||
+	padsh->pbPreparedDst != padsh->pbDst ||
+	padsh->cbPreparedDstLength < padsh->cbDstLength) {
+	return MMSYSERR_INVALPARAM;
+    }	
+
+    padsh->fdwConvert = fdwConvert;
+
+    ret = SendDriverMessage(was->pDrv->hDrvr, ACMDM_STREAM_CONVERT, (DWORD)&was->drvInst, (DWORD)padsh);
+    if (ret == MMSYSERR_NOERROR) {
+	padsh->fdwStatus |= ACMSTREAMHEADER_STATUSF_DONE;
+    }
+    TRACE("=> (%d)\n", ret);
+    return ret;
+}
+
+
+/***********************************************************************
+ *           acmStreamPrepareHeader (MSACM32.41)
+ */
+MMRESULT WINAPI acmStreamPrepareHeader(HACMSTREAM has, PACMSTREAMHEADER pash, 
+				       DWORD fdwPrepare)
+{
+    PWINE_ACMSTREAM	was;
+    MMRESULT		ret = MMSYSERR_NOERROR;
+    PACMDRVSTREAMHEADER	padsh;
+
+    TRACE("(0x%08x, %p, %ld)\n", has, pash, fdwPrepare);
+    
+    if ((was = ACM_GetStream(has)) == NULL)
+	return MMSYSERR_INVALHANDLE;
+    if (!pash || pash->cbStruct < sizeof(ACMSTREAMHEADER))
+	return MMSYSERR_INVALPARAM;
+    if (fdwPrepare)
+	ret = MMSYSERR_INVALFLAG;
+
+    if (pash->fdwStatus & ACMSTREAMHEADER_STATUSF_DONE)
+	return MMSYSERR_NOERROR;
+
+    /* Note: the ACMSTREAMHEADER and ACMDRVSTREAMHEADER structs are of same
+     * size. some fields are private to msacm internals, and are exposed
+     * in ACMSTREAMHEADER in the dwReservedDriver array
+     */
+    padsh = (PACMDRVSTREAMHEADER)pash;
+
+    padsh->fdwConvert = fdwPrepare;
+    padsh->padshNext = NULL;
+    padsh->fdwDriver = padsh->dwDriver = 0L;
+
+    padsh->fdwPrepared = 0;
+    padsh->dwPrepared = 0;
+    padsh->pbPreparedSrc = 0;
+    padsh->cbPreparedSrcLength = 0;
+    padsh->pbPreparedDst = 0;
+    padsh->cbPreparedDstLength = 0;
+
+    ret = SendDriverMessage(was->pDrv->hDrvr, ACMDM_STREAM_PREPARE, (DWORD)&was->drvInst, (DWORD)padsh);
+    if (ret == MMSYSERR_NOERROR || ret == MMSYSERR_NOTSUPPORTED) {
+	ret = MMSYSERR_NOERROR;
+	padsh->fdwStatus &= ~(ACMSTREAMHEADER_STATUSF_DONE|ACMSTREAMHEADER_STATUSF_INQUEUE);
+	padsh->fdwStatus |= ACMSTREAMHEADER_STATUSF_PREPARED;
+	padsh->fdwPrepared = padsh->fdwStatus;
+	padsh->dwPrepared = 0;
+	padsh->pbPreparedSrc = padsh->pbSrc;
+	padsh->cbPreparedSrcLength = padsh->cbSrcLength;
+	padsh->pbPreparedDst = padsh->pbDst;
+	padsh->cbPreparedDstLength = padsh->cbDstLength;
+    } else {
+	padsh->fdwPrepared = 0;
+	padsh->dwPrepared = 0;
+	padsh->pbPreparedSrc = 0;
+	padsh->cbPreparedSrcLength = 0;
+	padsh->pbPreparedDst = 0;
+	padsh->cbPreparedDstLength = 0;
+    }
+    TRACE("=> (%d)\n", ret);
+    return ret;
+}
+
+/***********************************************************************
+ *           acmStreamReset (MSACM32.42)
+ */
+MMRESULT WINAPI acmStreamReset(HACMSTREAM has, DWORD fdwReset)
+{
+    PWINE_ACMSTREAM	was;
+    MMRESULT		ret = MMSYSERR_NOERROR;
+
+    TRACE("(0x%08x, %ld)\n", has, fdwReset);
+
+    if (fdwReset) {
+	ret = MMSYSERR_INVALFLAG;
+    } else if ((was = ACM_GetStream(has)) == NULL) {
+	return MMSYSERR_INVALHANDLE;
+    } else if (was->drvInst.fdwOpen & ACM_STREAMOPENF_ASYNC) {
+	ret = SendDriverMessage(was->pDrv->hDrvr, ACMDM_STREAM_RESET, (DWORD)&was->drvInst, 0);
+    }
+    TRACE("=> (%d)\n", ret);
+    return ret;
+}
+
+/***********************************************************************
+ *           acmStreamSize (MSACM32.43)
+ */
+MMRESULT WINAPI acmStreamSize(HACMSTREAM has, DWORD cbInput, 
+			      LPDWORD pdwOutputBytes, DWORD fdwSize)
+{
+    PWINE_ACMSTREAM	was;
+    ACMDRVSTREAMSIZE	adss;
+    MMRESULT		ret;
+    
+    TRACE("(0x%08x, %ld, %p, %ld)\n", has, cbInput, pdwOutputBytes, fdwSize);
+    
+    if ((was = ACM_GetStream(has)) == NULL) {
+	return MMSYSERR_INVALHANDLE;
+    }
+    if ((fdwSize & ~ACM_STREAMSIZEF_QUERYMASK) != 0) {
+	return MMSYSERR_INVALFLAG;
+    }
+
+    *pdwOutputBytes = 0L;
+    
+    switch (fdwSize & ACM_STREAMSIZEF_QUERYMASK) {
+    case ACM_STREAMSIZEF_DESTINATION:
+	adss.cbDstLength = cbInput;
+	adss.cbSrcLength = 0;
+	break;
+    case ACM_STREAMSIZEF_SOURCE:
+	adss.cbSrcLength = cbInput;
+	adss.cbDstLength = 0;
+	break;
+    default:	
+	return MMSYSERR_INVALFLAG;
+    }
+    
+    adss.cbStruct = sizeof(adss);
+    adss.fdwSize = fdwSize;
+    ret = SendDriverMessage(was->pDrv->hDrvr, ACMDM_STREAM_SIZE, 
+			    (DWORD)&was->drvInst, (DWORD)&adss);
+    if (ret == MMSYSERR_NOERROR) {
+	switch (fdwSize & ACM_STREAMSIZEF_QUERYMASK) {
+	case ACM_STREAMSIZEF_DESTINATION:
+	    *pdwOutputBytes = adss.cbSrcLength;
+	    break;
+	case ACM_STREAMSIZEF_SOURCE:
+	    *pdwOutputBytes = adss.cbDstLength;
+	    break;
+	}
+    }
+    TRACE("=> (%d) [%lu]\n", ret, *pdwOutputBytes);
+    return ret;
+}
+
+/***********************************************************************
+ *           acmStreamUnprepareHeader (MSACM32.44)
+ */
+MMRESULT WINAPI acmStreamUnprepareHeader(HACMSTREAM has, PACMSTREAMHEADER pash, 
+					 DWORD fdwUnprepare)
+{
+    PWINE_ACMSTREAM	was;
+    MMRESULT		ret = MMSYSERR_NOERROR;
+    PACMDRVSTREAMHEADER	padsh;
+
+    TRACE("(0x%08x, %p, %ld)\n", has, pash, fdwUnprepare);
+    
+    if ((was = ACM_GetStream(has)) == NULL)
+	return MMSYSERR_INVALHANDLE;
+    if (!pash || pash->cbStruct < sizeof(ACMSTREAMHEADER))
+	return MMSYSERR_INVALPARAM;
+
+    if (!(pash->fdwStatus & ACMSTREAMHEADER_STATUSF_PREPARED))
+	return ACMERR_UNPREPARED;
+
+    /* Note: the ACMSTREAMHEADER and ACMDRVSTREAMHEADER structs are of same
+     * size. some fields are private to msacm internals, and are exposed
+     * in ACMSTREAMHEADER in the dwReservedDriver array
+     */
+    padsh = (PACMDRVSTREAMHEADER)pash;
+
+    /* check that pointers have not been modified */
+    if (padsh->pbPreparedSrc != padsh->pbSrc ||
+	padsh->cbPreparedSrcLength < padsh->cbSrcLength ||
+	padsh->pbPreparedDst != padsh->pbDst ||
+	padsh->cbPreparedDstLength < padsh->cbDstLength) {
+	return MMSYSERR_INVALPARAM;
+    }	
+
+    padsh->fdwConvert = fdwUnprepare;
+
+    ret = SendDriverMessage(was->pDrv->hDrvr, ACMDM_STREAM_UNPREPARE, (DWORD)&was->drvInst, (DWORD)padsh);
+    if (ret == MMSYSERR_NOERROR || ret == MMSYSERR_NOTSUPPORTED) {
+	ret = MMSYSERR_NOERROR;
+	padsh->fdwStatus &= ~(ACMSTREAMHEADER_STATUSF_DONE|ACMSTREAMHEADER_STATUSF_INQUEUE|ACMSTREAMHEADER_STATUSF_PREPARED);
+    }
+    TRACE("=> (%d)\n", ret);
+    return ret;
+}

libs/winloader/driver.c

@@ -6,8 +6,6 @@
 #include <wine/vfw.h>
 #include <registry.h>
 
-#include <config.h>
-
 #define STORE_ALL \
     __asm__ ( \
     "push %%ebx\n\t" \
@@ -39,6 +37,26 @@ typedef DRVR  *LPDRVR;
     
 static DWORD dwDrvID = 0;
 
+LRESULT WINAPI SendDriverMessage( HDRVR hDriver, UINT message,
+                                    LPARAM lParam1, LPARAM lParam2 )
+{
+    DRVR* module=(DRVR*)hDriver;
+    int result;
+#ifdef DETAILED_OUT    
+    printf("SendDriverMessage: driver %X, message %X, arg1 %X, arg2 %X\n", hDriver, message, lParam1, lParam2);
+#endif
+    if(module==0)return -1;
+    if(module->hDriverModule==0)return -1;
+    if(module->DriverProc==0)return -1;
+    STORE_ALL;
+    result=module->DriverProc(module->dwDriverID,1,message,lParam1,lParam2);
+    REST_ALL;
+#ifdef DETAILED_OUT    
+    printf("\t\tResult: %X\n", result);
+#endif    
+    return result;
+}				    
+
 static NPDRVR DrvAlloc(HDRVR*lpDriver, LPUINT lpDrvResult)
 {
     NPDRVR npDriver;
@@ -66,28 +84,44 @@ typedef struct
     int usage;
 }codec_t;
 
-static codec_t codecs[3]={
+#define Win32Path "/usr/lib/win32/"
- {0, PLUGINS_SRCDIR "/win32/divxc32.dll", 0},
+static codec_t avi_codecs[]={
- {0, PLUGINS_SRCDIR "/win32/ir50_32.dll", 0},
+ {0, Win32Path"divxc32.dll", 0},	//0
-// {0, "./mpg4c32.dll", 0},
+ {0, Win32Path"ir50_32.dll", 0},
- {0, PLUGINS_SRCDIR "/win32/libvideodll.so", 0},
+ {0, Win32Path"ir41_32.dll", 0},
+ {0, Win32Path"ir32_32.dll", 0},    
+ {0, Win32Path"mpg4c32.dll", 0},
+ {0, Win32Path"iccvid.dll", 0},		//5
+ {0, Win32Path"libvideodll.so", 0},
+ {0, Win32Path"divxa32.acm", 0},	//7
+ {0, Win32Path"msadp32.acm", 0},
 };
 
                                                                                                                     
 static void DrvFree(HDRVR hDriver)
 {
     int i;
-    FreeLibrary(((DRVR*)hDriver)->hDriverModule);
     if(hDriver)
-    for(i=0; i<sizeof(codecs)/sizeof(codecs[0]); i++)
+    	if(((DRVR*)hDriver)->hDriverModule)
-	if(codecs[i].handle==((DRVR*)hDriver)->hDriverModule)
+    	if(((DRVR*)hDriver)->DriverProc)
+	(((DRVR*)hDriver)->DriverProc)(((DRVR*)hDriver)->dwDriverID, hDriver, DRV_CLOSE, 0, 0);
+    if(hDriver)
+    for(i=0; i<sizeof(avi_codecs)/sizeof(avi_codecs[0]); i++)
+	if(avi_codecs[i].handle==((DRVR*)hDriver)->hDriverModule)
 	{
-	    codecs[i].handle=0;	    
+	    avi_codecs[i].usage--;	
-	    codecs[i].usage--;	 
+	    if(avi_codecs[i].usage==0)
+	    { 
+    		avi_codecs[i].handle=0;	    
+                 if(((DRVR*)hDriver)->hDriverModule)
+    		if(((DRVR*)hDriver)->DriverProc)
+			(((DRVR*)hDriver)->DriverProc)(0, hDriver, DRV_FREE, 0, 0);
+		FreeLibrary(((DRVR*)hDriver)->hDriverModule);
     		if (hDriver)
         	    		free((NPDRVR)hDriver);
 		return;	
 	    }	
+	}      
 }       
 
 void DrvClose(HDRVR hdrvr)
@@ -110,21 +144,46 @@ DrvOpen(LPARAM lParam2)
     int regs[10];
 
     int fccHandler=*((int*)lParam2+2);
+    int fccType=*((int*)lParam2+1);
+    if(fccType==0x63646976)//vidc
 	switch(fccHandler)
 	{
 	case mmioFOURCC('D', 'I', 'V', '3'):
 	case mmioFOURCC('D', 'I', 'V', '4'):
 	case mmioFOURCC('d', 'i', 'v', '3'):
         case mmioFOURCC('d', 'i', 'v', '4'):
+	    printf("Video in DivX ;-) format\n");
 	    drv_id=0;
 	    break;
 	case mmioFOURCC('I', 'V', '5', '0'):	    
 	case mmioFOURCC('i', 'v', '5', '0'):	 
+	    printf("Video in Indeo Video 5 format\n");   
 	    drv_id=1;
 	    break;
+	case mmioFOURCC('I', 'V', '4', '1'):	    
+	case mmioFOURCC('i', 'v', '4', '1'):	    
+	    printf("Video in Indeo Video 4.1 format\n");   
+	    drv_id=2;
+	    break;
+	case mmioFOURCC('I', 'V', '3', '2'):	    
+	case mmioFOURCC('i', 'v', '3', '2'):
+	    printf("Video in Indeo Video 3.2 format\n");   
+	    drv_id=3;
+	    break;
+	
+	case mmioFOURCC('m', 'p', '4', '1'):
+	case mmioFOURCC('m', 'p', '4', '2'):
         case mmioFOURCC('m', 'p', '4', '3'):
 	case mmioFOURCC('M', 'P', 'G', '4'):
-	drv_id=2;
+	case mmioFOURCC('M', 'P', '4', '1'):
+	case mmioFOURCC('M', 'P', '4', '2'):
+	case mmioFOURCC('M', 'P', '4', '3'):
+	    printf("Video in Microsoft MPEG-4 format\n");   
+	    drv_id=4;
+	    break;
+	case mmioFOURCC('c', 'v', 'i', 'd'):
+	    printf("Video in Cinepak format\n");   
+	    drv_id=5;
 	    break;	
 	default:
 	    printf("Unknown codec %X='%c%c%c%c'\n", fccHandler, 
@@ -132,24 +191,38 @@ DrvOpen(LPARAM lParam2)
     	    (fccHandler&0xFF0000)>>16, (fccHandler&0xFF000000)>>24);
     	    return (HDRVR)0;	
         }
-    
+    else
+	switch(fccHandler)
+	{
+    	case 0x160://DivX audio
+    	case 0x161://DivX audio
+	    drv_id=7; 
+	    break;
+	case 0x2://MS ADPCM
+	    drv_id=8;
+	    break;
+	default:
+	    printf("Unknown ACM codec 0x%X\n", fccHandler);
+	    return (HDRVR)0;
+	}
 	
     if (!(npDriver = DrvAlloc(&hDriver, &uDrvResult)))
 	return ((HDRVR) 0);
 
-    if(codecs[drv_id].handle==0)
+    if(avi_codecs[drv_id].handle==0)
     {
-     if (!(codecs[drv_id].handle=npDriver->hDriverModule = LoadLibraryA(codecs[drv_id].name)))
+     if (!(avi_codecs[drv_id].handle=npDriver->hDriverModule = LoadLibraryA(avi_codecs[drv_id].name)))
      {
+     	printf("Can't open library %s\n", avi_codecs[drv_id].name);
         DrvFree(hDriver);
         return ((HDRVR) 0);
      }
-     else codecs[drv_id].usage=1;
+     else avi_codecs[drv_id].usage=1;
     }
     else
     {
-	npDriver->hDriverModule=codecs[drv_id].handle;
+	npDriver->hDriverModule=avi_codecs[drv_id].handle;
-	codecs[drv_id].usage++;
+	avi_codecs[drv_id].usage++;
     }	
     
 //    14c0 
@@ -177,24 +250,25 @@ DrvOpen(LPARAM lParam2)
      if (!(npDriver->DriverProc = (DRIVERPROC)
              GetProcAddress(npDriver->hDriverModule, "DriverProc")))
          {
+         printf("Library %s is not a valid codec\n", avi_codecs[drv_id].name);
          FreeLibrary(npDriver->hDriverModule);
          DrvFree(hDriver);
          return ((HDRVR) 0);
      }
      
-     //printf("DriverProc == %X\n", npDriver->DriverProc);
+    TRACE("DriverProc == %X\n", npDriver->DriverProc);
      npDriver->dwDriverID = ++dwDrvID;
 
-     if (codecs[drv_id].usage==1)
+     if (avi_codecs[drv_id].usage==1)
      {
 	STORE_ALL;
         (npDriver->DriverProc)(0, hDriver, DRV_LOAD, 0, 0);
 	REST_ALL;
-	//printf("DRV_LOAD Ok!\n");
+	TRACE("DRV_LOAD Ok!\n");
 	STORE_ALL;
 	(npDriver->DriverProc)(0, hDriver, DRV_ENABLE, 0, 0);
 	REST_ALL;
-	//printf("DRV_ENABLE Ok!\n");
+	TRACE("DRV_ENABLE Ok!\n");
      }
 
      // open driver 
@@ -203,7 +277,7 @@ DrvOpen(LPARAM lParam2)
          (LPARAM) (LPSTR) unknown, lParam2);
     REST_ALL;
 
-    //printf("DRV_OPEN Ok!(%X)\n", npDriver->dwDriverID);
+    TRACE("DRV_OPEN Ok!(%X)\n", npDriver->dwDriverID);
 
     if (uDrvResult)
     {

libs/winloader/elfdll.c

@@ -3,6 +3,9 @@
  *
  * Copyright 1999 Bertho A. Stultiens
  */
+#include <config.h>
+
+#ifdef HAVE_LIBDL
 
 #include <string.h>
 #include <ctype.h>
@@ -42,7 +45,7 @@ extern DWORD fixup_imports(WINE_MODREF *wm);
 extern void dump_exports(HMODULE hModule);
 /*---------------- END HACKS ---------------*/
 
-char *extra_ld_library_path = NULL;	/* The extra search-path set in wine.conf */
+char *extra_ld_library_path = "/usr/lib/win32";
 
 struct elfdll_image
 {
@@ -201,7 +204,8 @@ static WINE_MODREF *ELFDLL_CreateModref(HMODULE hModule, LPCSTR path)
 //		wm->binfmt.pe.pe_resource = (PIMAGE_RESOURCE_DIRECTORY)RVA(hModule, dir->VirtualAddress);
 
 
-	wm->filename = strdup( path );
+	wm->filename = malloc(strlen(path)+1);
+	strcpy(wm->filename, path);
 	wm->modname = strrchr( wm->filename, '\\' );
 	if (!wm->modname) wm->modname = wm->filename;
 	else wm->modname++;
@@ -275,7 +279,7 @@ WINE_MODREF *ELFDLL_LoadLibraryExA(LPCSTR path, DWORD flags)
 	}
 
 */
-	wm = ELFDLL_CreateModref(dlhandle, path);
+	wm = ELFDLL_CreateModref((int)dlhandle, path);
 	if(!wm)
 	{
 		ERR("Could not create WINE_MODREF for %s\n", path);
@@ -297,4 +301,4 @@ void ELFDLL_UnloadLibrary(WINE_MODREF *wm)
 {
 }
 
-
+#endif /*HAVE_LIBDL*/

libs/winloader/ext.c

@@ -11,8 +11,9 @@
 #include <errno.h>
 #include <fcntl.h>
 #include <string.h>
-
+#include <stdarg.h>
 #include <wine/windef.h>
+//#include <wine/winbase.h>
 int dbg_header_err( const char *dbg_channel, const char *func )
 {
     return 0; 
@@ -35,6 +36,12 @@ int dbg_vprintf( const char *format, ... )
 }
 int __vprintf( const char *format, ... )
 {
+#ifdef DETAILED_OUT
+    va_list va;
+    va_start(va, format);
+    vprintf(format, va);
+    va_end(va);
+#endif
     return 0; 
 }
     
@@ -56,17 +63,7 @@ int HeapFree(int heap, int flags, void* mem)
     free(mem);
     return 1;
 }     	
-/*
-void EnterCriticalSection(void* q)
-{
-    return;
-}
 
-void LeaveCriticalSection(void* q)
-{
-    return;
-}   
-*/
 static int last_error;
 
 int GetLastError()
@@ -155,7 +152,10 @@ int IsBadReadPtr(void* data, int size)
 }   
 char* HEAP_strdupA(const char* string)
 {
-    return strdup(string);
+//    return strdup(string);
+    char* answ=malloc(strlen(string)+1);
+    strcpy(answ, string);
+    return answ;
 }
 short* HEAP_strdupAtoW(void* heap, void* hz, const char* string)
 {
@@ -337,8 +337,7 @@ DWORD flProtect, DWORD dwMaxHigh, DWORD dwMaxLow, const char* name)
 	lseek(hFile, 0, SEEK_SET);
     }
     else len=dwMaxLow;
-//    len=min(len, dwMaxLow);
+
-#warning fixme - should analyze flProtect
     if(flProtect & PAGE_READONLY)
 	mmap_access |=PROT_READ;
     else
@@ -363,7 +362,10 @@ DWORD flProtect, DWORD dwMaxHigh, DWORD dwMaxLow, const char* name)
 	fm->next=NULL;    
 	fm->handle=answer;
 	if(name)
-	    fm->name=strdup(name);
+	{
+	    fm->name=malloc(strlen(name)+1);
+	    strcpy(fm->name, name);
+	}    
 	else
 	    fm->name=NULL;
 	fm->mapping_size=len;
@@ -397,15 +399,60 @@ int UnmapViewOfFile(HANDLE handle)
     }
     return 0;	
 }    
-static int va_size=0;
+//static int va_size=0;
+struct virt_alloc_s;
+typedef struct virt_alloc_s
+{
+    int mapping_size;
+    char* address;
+    struct virt_alloc_s* next;
+    struct virt_alloc_s* prev;
+    int state;
+}virt_alloc;
+static virt_alloc* vm=0;
+#define MEM_COMMIT              0x00001000
+#define MEM_RESERVE             0x00002000
+
 void* VirtualAlloc(void* address, DWORD size, DWORD type,  DWORD protection)
 {
     void* answer;
     int fd=open("/dev/zero", O_RDWR);
     size=(size+0xffff)&(~0xffff);
 //    printf("VirtualAlloc(0x%08X, %d)\n", address
+    if(address!=0)
+    {
+    //check whether we can allow to allocate this 
+        virt_alloc* str=vm;
+        while(str)
+        {
+	    if((unsigned)address>=(unsigned)str->address+str->mapping_size)
+	    {
+		str=str->prev;
+		continue;
+	    }	
+	    if((unsigned)address+size<(unsigned)str->address)
+	    {
+		str=str->prev;
+		continue;
+	    }
+	    if(str->state==0)
+	    {
+#warning FIXME
+		if(((unsigned)address+size<(unsigned)str->address+str->mapping_size) && (type & MEM_COMMIT))
+		{
+		    close(fd);
+		    return address; //returning previously reserved memory
+		}
+		return NULL;    
+	    }
+	    close(fd);
+	    return NULL;
+	}
+	answer=mmap(address, size, PROT_READ | PROT_WRITE | PROT_EXEC, 
+	    MAP_FIXED | MAP_PRIVATE, fd, 0);
+    }	    
+    else
     answer=mmap(address, size, PROT_READ | PROT_WRITE | PROT_EXEC, 
-//     ((address!=NULL) ? MAP_FIXED : MAP_SHARED), fd, 0);
 	 MAP_PRIVATE, fd, 0);
 //    answer=FILE_dommap(-1, address, 0, size, 0, 0,
 //	PROT_READ | PROT_WRITE | PROT_EXEC, MAP_PRIVATE);
@@ -418,31 +465,60 @@ void* VirtualAlloc(void* address, DWORD size, DWORD type,  DWORD protection)
     }
     else
     {
-	if(va_size!=0)
+	virt_alloc *new_vm=malloc(sizeof(virt_alloc));
-	    printf("Multiple VirtualAlloc!\n");	    
+	new_vm->mapping_size=size;
+	new_vm->address=answer;
+        new_vm->prev=vm;
+	if(type & MEM_RESERVE)
+	    new_vm->state=0;
+	else
+	    new_vm->state=1;
+	if(vm)
+	    vm->next=new_vm;
+    	vm=new_vm;
+	vm->next=0;
+//	if(va_size!=0)
+//	    printf("Multiple VirtualAlloc!\n");	    
 //	printf("answer=0x%08x\n", answer);
-	va_size=size;
         return answer;
     }	
 }    	
 int VirtualFree(void*  address, int t1, int t2)//not sure
 {
-    int answer=munmap(address, va_size);
+    virt_alloc* str=vm;
-    va_size=0;
+    int answer;
-    return answer;
+    while(str)
+    {
+    if(address!=str->address)
+    {
+	str=str->prev;
+	continue;
+    }	
+    answer=munmap(str->address, str->mapping_size);
+    if(str->next)str->next->prev=str->prev;
+    if(str->prev)str->prev->next=str->next;
+    if(vm==str)vm=0;
+    free(str);
+    return 0;
+    }
+    return -1;
 }
 
 int WideCharToMultiByte(unsigned int codepage, long flags, const short* src,
      int srclen,char* dest, int destlen, const char* defch, int* used_defch)
 {
-#warning FIXME
     int i;
-//    printf("WCh2MB: Src string ");
-//    for(i=0; i<=srclen; i++)printf(" %04X", src[i]);
     if(src==0)
 	return 0;
     if(dest==0)
-	return 0;
+  {
+    for(i=0; i<srclen; i++)
+    {
+	src++;
+	if(*src==0)
+	    return i+1;
+    }	    
+    }	
     if(used_defch)
 	*used_defch=0;	
     for(i=0; i<min(srclen, destlen); i++)

libs/winloader/externals.c

@@ -2,7 +2,9 @@
 #include <stdio.h>
 #include <pthread.h>
 #include <malloc.h>
+#include <time.h>
 #include <sys/types.h>
+#include <sys/time.h>
 #include <sys/timeb.h>
 
 #include <wine/winbase.h>
@@ -10,11 +12,23 @@
 #include <wine/winnt.h>
 #include <wine/winerror.h>
 #include <wine/debugtools.h>
+#include <wine/module.h>
 
 #include <registry.h>
 
+static unsigned int localcount()
+{
+    int a;
+    __asm__ __volatile__("rdtsc\n\t"
+    :"=a"(a)
+    :
+    :"edx");
+    return a;
+}
+
 void dbgprintf(char* fmt, ...)
 {
+#ifdef DETAILED_OUT
     va_list va;
     FILE* f;
     va_start(va, fmt);
@@ -22,6 +36,7 @@ void dbgprintf(char* fmt, ...)
     vfprintf(f, fmt, va);
     fsync(f);
     fclose(f);
+#endif
 }    
 char export_names[500][30]={
 "name1",
@@ -64,6 +79,11 @@ void* my_mreq(int size, int to_zero)
 	heap=malloc(20000000);
 	memset(heap, 0xCC,20000000);
     }
+    if(heap==0)
+    {
+	printf("No enough memory\n");
+	return 0;
+    }	
     if(heap_counter+size>20000000)
     {
 	printf("No enough memory\n");
@@ -82,6 +102,11 @@ void* my_mreq(int size, int to_zero)
 int my_release(char* memory)
 {
     test_heap();
+    if(memory==NULL)
+    {
+	printf("ERROR: free(0)\n");
+	return 0;
+    }	
     if(*(int*)(memory-8)!=0x433476)
     {
 	printf("MEMORY CORRUPTION !!!!!!!!!!!!!!!!!!!\n");
@@ -118,6 +143,7 @@ int WINAPI ext_unknown()
 }    
 int WINAPI expIsBadWritePtr(void* ptr, unsigned int count)
 {
+    dbgprintf("IsBadWritePtr(%x, %x)\n", ptr, count);
     if(count==0)
 	return 0;
     if(ptr==0)
@@ -126,6 +152,7 @@ int WINAPI expIsBadWritePtr(void* ptr, unsigned int count)
 }
 int WINAPI expIsBadReadPtr(void* ptr, unsigned int count)
 {
+    dbgprintf("IsBadReadPtr(%x, %x)\n", ptr, count);
     if(count==0)
 	return 0;
     if(ptr==0)
@@ -136,11 +163,19 @@ void* CDECL expmalloc(int size)
 {
 //printf("malloc");
 //    return malloc(size);
-    return my_mreq(size,0);
+    void* result=my_mreq(size,0);
+    dbgprintf("malloc(%x)\n", size);
+    if(result==0)
+    {
+	dbgprintf("returns 0\n");
+	printf("WARNING: malloc() failed\n");
+    }	
+    return result;
 }
 void CDECL expfree(void* mem)
 {
 //    return free(mem);
+    dbgprintf("free(%x)\n", mem);
     my_release(mem);
 }
 void* CDECL expnew(int size)
@@ -149,15 +184,25 @@ void* CDECL expnew(int size)
 //    printf("%08x %08x %08x %08x\n",
 //    size, *(1+(int*)&size),
 //    *(2+(int*)&size),*(3+(int*)&size));
-    return malloc(size);
+    void* result=expmalloc(size);
+    dbgprintf("new(%x)\n", size);
+    if(result==0)
+    {
+	dbgprintf("returns 0\n");
+	printf("WARNING: malloc() failed\n");
+    }	
+    return result;
+
 }    
 int CDECL expdelete(void* memory)
 {
-    free(memory);
+    dbgprintf("delete(%x)\n", memory);
+    expfree(memory);
     return 0;
 }
 int WINAPI expDisableThreadLibraryCalls(int module)
 {
+    dbgprintf("DisableThreadLibraryCalls(%x)\n", module);
     return 0;
 }    
 int CDECL exp_initterm(int v1, int v2)
@@ -174,8 +219,18 @@ typedef struct {
 
 void* WINAPI expGetDriverModuleHandle(DRVR* pdrv)
 {
+    dbgprintf("GetDriverModuleHandle(%x)\n", pdrv);
     return pdrv->hDriverModule;
 }
+
+void* WINAPI expGetModuleHandleA(const char* name)
+{
+	WINE_MODREF* wm;
+        dbgprintf("GetModuleHandleA(%s)\n", name);
+        wm=MODULE_FindModule(name);
+        if(wm==0)return 0;
+        return (void*)(wm->module);
+}
 struct th_list_t;
 typedef struct th_list_t{
 int id;
@@ -233,7 +288,6 @@ void* WINAPI expCreateEventA(void* pSecAttr, char bManualReset,
 {
 #warning ManualReset
     pthread_mutex_t *pm;
-//    printf("CreateEvent:");
     dbgprintf("CreateEvent\n");
     if(mlist!=NULL)
     {
@@ -243,7 +297,7 @@ void* WINAPI expCreateEventA(void* pSecAttr, char bManualReset,
 	{
 	    if(strcmp(pp->name, name)==0)
 		return pp->pm;
-	}while(pp=pp->next);
+	}while(pp=pp->prev);
     }	
     pm=my_mreq(sizeof(pthread_mutex_t), 0);
     pthread_mutex_init(pm, NULL);
@@ -298,6 +352,7 @@ void WINAPI expGetSystemInfo(SYSTEM_INFO* si)
 	static int cache = 0;
 	static SYSTEM_INFO cachedsi;
 	HKEY	xhkey=0,hkey;
+        dbgprintf("GetSystemInfo()\n");
 
 	if (cache) {
 		memcpy(si,&cachedsi,sizeof(*si));
@@ -464,6 +519,14 @@ long WINAPI expHeapDestroy(void* heap)
     my_release(heap);
     return 1;
 }
+
+long WINAPI expHeapFree(int arg1, int arg2, void* ptr)
+{
+    dbgprintf("HeapFree(%X, %X, %X)\n", arg1, arg2, ptr);
+    my_release(ptr);
+    return 1;
+}    	
+
 void* WINAPI expVirtualAlloc(void* v1, long v2, long v3, long v4)
 {
     void* z;
@@ -478,28 +541,48 @@ int WINAPI expVirtualFree(void* v1, int v2, int v3)
     dbgprintf("VirtualFree(%X %X %X) \n",v1,v2,v3);
     return VirtualFree(v1,v2,v3);
 }    
+struct CRITSECT 
+{
+    pthread_t id;
+    pthread_mutex_t mutex;
+    int locked;
+};
 void WINAPI expInitializeCriticalSection(CRITICAL_SECTION* c)
 {
+    struct CRITSECT cs;
     dbgprintf("InitCriticalSection(%X) \n", c);
-    if(sizeof(pthread_mutex_t)>sizeof(CRITICAL_SECTION))
+/*    if(sizeof(pthread_mutex_t)>sizeof(CRITICAL_SECTION))
     {
 	printf(" ERROR:::: sizeof(pthread_mutex_t) is %d, expected <=%d!\n",
 	     sizeof(pthread_mutex_t), sizeof(CRITICAL_SECTION));
 	return;
-    }
+    }*/
-    pthread_mutex_init((pthread_mutex_t*)c, NULL);   
+/*    pthread_mutex_init((pthread_mutex_t*)c, NULL);   */
+    pthread_mutex_init(&cs.mutex, NULL);   
+    cs.locked=0;
+    *(void**)c=malloc(sizeof cs);
+    memcpy(*(void**)c, &cs, sizeof cs);
     return;
 }          
 void WINAPI expEnterCriticalSection(CRITICAL_SECTION* c)
 {
+    struct CRITSECT* cs=(struct CRITSECT*)c;
     dbgprintf("EnterCriticalSection(%X) \n",c);
-    pthread_mutex_lock((pthread_mutex_t*)c);
+//    cs.id=pthread_self();
+    if(cs->locked)
+	if(cs->id==pthread_self())
+	    return;
+    pthread_mutex_lock(&(cs->mutex));
+    cs->locked=1;
+    cs->id=pthread_self();
     return;
 }          
 void WINAPI expLeaveCriticalSection(CRITICAL_SECTION* c)
 {
+    struct CRITSECT* cs=(struct CRITSECT*)c;
     dbgprintf("LeaveCriticalSection(%X) \n",c);
-    pthread_mutex_unlock((pthread_mutex_t*)c);
+    cs->locked=0;
+    pthread_mutex_unlock(&(cs->mutex));
     return;
 }
 void WINAPI expDeleteCriticalSection(CRITICAL_SECTION *c)
@@ -613,9 +696,9 @@ int WINAPI expLoadStringA(long instance, long  id, void* buf, long size)
 
 long WINAPI expMultiByteToWideChar(long v1, long v2, char* s1, long siz1, char* s2, int siz2)
 {
-#warning fixme
+#warning FIXME
     dbgprintf("MB2WCh\n");
-//    printf("WARNING: Unsupported call: MBToWCh %s\n", s1);       
+    printf("WARNING: Unsupported call: MBToWCh %s\n", s1);       
     if(s2==0)
 	return 1;
     s2[0]=s2[1]=0;
@@ -636,12 +719,32 @@ long WINAPI expGetVersionExA(OSVERSIONINFOA* c)
     strcpy(c->szCSDVersion, "Win98");
     return 1;
 }        
+#include <sys/types.h>
+#include <sys/ipc.h>
+#include <sys/sem.h>
 HANDLE WINAPI expCreateSemaphoreA(char* v1, long init_count, long max_count, char* name)
 {
-#warning fixme
+#warning FIXME
+/*    struct sembuf buf[1];
+    int sem=semget(IPC_PRIVATE,1,IPC_CREAT);
+    if(sem==-1)
+    {
+	printf("semget() failed\n");
+	return (HANDLE)-1;
+    }	
+    buf[0].sem_num=0;
+    printf("%s\n", name);
+    printf("Init count %d, max count %d\n", init_count, max_count);
+    buf[0].sem_op=-max_count+init_count;
+    buf[0].sem_flg=0;
+    if(semop(sem, &buf, 1)<0)
+    {
+	printf("semop() failed\n");
+    }
+    return sem;	
+*/    
     void* z;
     dbgprintf("CreateSemaphoreA\n");
-//    printf("CreateSemaphore:");
     z=my_mreq(24, 0);
     pthread_mutex_init(z, NULL);
     return (HANDLE)z;
@@ -653,43 +756,62 @@ long WINAPI expReleaseSemaphore(long hsem, long increment, long* prev_count)
 // is greater than zero and nonsignaled when its count is equal to zero
 // Each time a waiting thread is released because of the semaphore's signaled
 // state, the count of the semaphore is decreased by one. 
+    struct sembuf buf[1];
     dbgprintf("ReleaseSemaphore\n");
     printf("WARNING: Unsupported call: ReleaseSemaphoreA\n");       
+/*    if(hsem==-1)return 0;
+    buf[0].sem_num=0;
+    buf[0].sem_op=-1;
+    buf[0].sem_flg=0;
+    if(semop(hsem, &buf, 1)<0)
+    {
+	printf("ReleaseSemaphore: semop() failed\n");
+    }*/
+
     return 1;//zero on error
 }
 
 
-long WINAPI expRegOpenKeyExA(long key, char* subkey, long reserved, long access, long* newkey)
+long WINAPI expRegOpenKeyExA(long key, const char* subkey, long reserved, long access, int* newkey)
 {
+    dbgprintf("RegOpenKeyExA(%d,%s)\n", key, subkey);
     return RegOpenKeyExA(key, subkey, reserved, access, newkey);
 }    
 long WINAPI expRegCloseKey(long key)
 {
+    dbgprintf("RegCloseKey()\n");
     return RegCloseKey(key);
 }         
-long WINAPI expRegQueryValueExA(long key, char* value, int* reserved, int* type, int* data, int* count)
+long WINAPI expRegQueryValueExA(long key, const char* value, int* reserved, int* type, int* data, int* count)
 {
+    dbgprintf("RegQueryValueExA()\n");
     return RegQueryValueExA(key, value, reserved, type, data, count);
 }  
-long WINAPI expRegCreateKeyExA(long key, char* name, long reserved,
+long WINAPI expRegCreateKeyExA(long key, const char* name, long reserved,
 							   void* classs, long options, long security,
 							   void* sec_attr, int* newkey, int* status) 
 {
+    dbgprintf("RegCreateKeyExA()\n");
     return RegCreateKeyExA(key, name, reserved, classs, options, security, sec_attr, newkey, status);
 }
-long WINAPI expRegSetValueExA(long key, char* name, long v1, long v2, void* data, long size)
+long WINAPI expRegSetValueExA(long key, const char* name, long v1, long v2, void* data, long size)
 {
+    dbgprintf("RegSetValueExA()\n");
     return RegSetValueExA(key, name, v1, v2, data, size);
 }        
 
-long WINAPI expRegOpenKeyA (
long hKey,
 LPCSTR lpSubKey,
 long* phkResult
)
+long WINAPI expRegOpenKeyA (
-{
+long hKey,
+ LPCSTR lpSubKey,
+ int* phkResult
+){
     return  RegOpenKeyExA(hKey, lpSubKey, 0, 0, phkResult);
 }
 
 long WINAPI expQueryPerformanceCounter(long long* z)
 {
-    __asm__(
+    dbgprintf("QueryPerformanceCounter()\n");
+    __asm__ __volatile__(
     "rdtsc\n\t"
     "movl %%eax, 0(%0)\n\t"
     "movl %%edx, 4(%0)\n\t"
@@ -697,17 +819,69 @@ long WINAPI expQueryPerformanceCounter(long long* z)
     return 1; 
 }
 
+static double old_freq()
+{
+    int i=time(NULL);
+    int x,y;
+    while(i==time(NULL));
+    x=localcount();
+    i++;
+    while(i==time(NULL));
+    y=localcount();
+    return (double)(y-x)/1000.;
+}
+static double CPU_Freq()
+{
+	FILE *f = fopen ("/proc/cpuinfo", "r");
+	char line[200];
+	char model[200]="unknown";
+	char flags[500]="";
+	char	*s,*value;
+	double freq=-1;
+	
+	if (!f)
+	{
+	    printf("Can't open /proc/cpuinfo for reading\n");
+	    return old_freq();
+	}    
+	while (fgets(line,200,f)!=NULL) 
+	{
+		/* NOTE: the ':' is the only character we can rely on */
+		if (!(value = strchr(line,':')))
+			continue;
+		/* terminate the valuename */
+		*value++ = '\0';
+		/* skip any leading spaces */
+		while (*value==' ') value++;
+		if ((s=strchr(value,'\n')))
+			*s='\0';
+
+		if (!strncasecmp(line, "cpu MHz",strlen("cpu MHz"))) 
+		{
+		    sscanf(value, "%lf", &freq);
+		    freq*=1000;
+		    break;
+		}
+		continue;
+		
+	}
+	fclose(f);
+	if(freq<0)return old_freq();
+	return freq;
+}
+
 long WINAPI expQueryPerformanceFrequency(long long* z)
 {
-#warning fixme
+    dbgprintf("QueryPerformanceFrequency()\n");
-    *z=(long long)550000000;
+    *z=(long long)CPU_Freq();
     return 1; 
 }
 long WINAPI exptimeGetTime()
 {
-    struct timeb t;
+    struct timeval t;
-    ftime(&t);
+    dbgprintf("timeGetTime()\n");
-    return 1000*t.time+t.millitm;
+    gettimeofday(&t, 0);
+    return 1000*t.tv_sec+t.tv_usec/1000;
 }
 void* WINAPI expLocalHandle(void* v)
 {
@@ -745,13 +919,11 @@ void* WINAPI expLocalFree(void* v)
     return 0;
 }        
 
-// HRSRC fun(HMODULE module, char* name, char* type)
 HRSRC WINAPI expFindResourceA(HMODULE module, char* name, char* type)
 {
     dbgprintf("FindResourceA\n");
     return FindResourceA(module, name, type);
 }
-//HGLOBAL fun(HMODULE module, HRSRC res)
 HGLOBAL WINAPI expLoadResource(HMODULE module, HRSRC res)
 {
     dbgprintf("LoadResource\n");
@@ -762,7 +934,7 @@ void* WINAPI expLockResource(long res)
     dbgprintf("LockResource\n");
     return LockResource(res);
 }    
-int /*bool*/ WINAPI expFreeResource(long res)
+int WINAPI expFreeResource(long res)
 {
     dbgprintf("FreeResource\n");
     return FreeResource(res);
@@ -778,11 +950,11 @@ int WINAPI expCloseHandle(long v1)
 const char* WINAPI expGetCommandLineA()
 {
     dbgprintf("GetCommandLine\n");
-    return "aviplay";
+    return "c:\\aviplay.exe";
 }
 LPWSTR WINAPI expGetEnvironmentStringsW()
 {
-    static short envs[]={0};
+    static short envs[]={'p', 'a', 't', 'h', ' ', 'c', ':', '\\', 0};
     dbgprintf("GetEnvStringsW\n");
     return envs;
 }
@@ -827,15 +999,27 @@ int WINAPI expGetACP()
     printf("WARNING: Unsupported call: GetACP\n");       
     return 0; 
 }
+extern WINE_MODREF *MODULE32_LookupHMODULE(HMODULE m);
 int WINAPI expGetModuleFileNameA(int module, char* s, int len)
 {
+    WINE_MODREF *mr;
     dbgprintf("GetModuleFileNameA\n");
     printf("File name of module %X requested\n", module);
     if(s==0)
     return 0;
-    if(len<10)
+    if(len<35)
     return 0;
-    strcpy(s, "aviplay");
+    strcpy(s, "c:\\windows\\system\\");
+    mr=MODULE32_LookupHMODULE(module);
+    if(mr==0)//oops
+    {
+        strcat(s, "aviplay.dll");
+	return 1;
+    }	
+    if(strrchr(mr->filename, '/')==NULL)
+	strcat(s, mr->filename);
+    else
+	strcat(s, strrchr(mr->filename, '/')+1);
     return 1;
 }    
     
@@ -856,7 +1040,6 @@ int WINAPI expFreeLibrary(int module)
 {
     dbgprintf("FreeLibrary\n");
     return FreeLibrary(module);
-//    return 0;
 }   
 void* WINAPI expGetProcAddress(HMODULE mod, char* name)
 {
@@ -873,10 +1056,8 @@ long WINAPI expCreateFileMappingA(int hFile, void* lpAttr,
 
 long WINAPI expOpenFileMappingA(long hFile, long hz, const char* name)
 {
-#warning fixme
+    dbgprintf("OpenFileMappingA\n");
-// dbgprintf("OpenFileMappingA\n");
     return OpenFileMappingA(hFile, hz, name);
-//  return 0;
 }
 
 void* WINAPI expMapViewOfFile(HANDLE file, DWORD mode, DWORD offHigh, DWORD offLow, DWORD size)
@@ -886,36 +1067,129 @@ void* WINAPI expMapViewOfFile(HANDLE file, DWORD mode, DWORD offHigh, DWORD offL
     return (char*)file+offLow;
 }
 
+void* WINAPI expUnmapViewOfFile(void* view)
+{
+    dbgprintf("UnmapViewOfFile()\n");
+    return 0;
+}
+
 void* WINAPI expSleep(int time)
 {
     dbgprintf("Sleep(%d)\n", time);
     usleep(time);
     return 0;
 }
+ // why does IV32 codec want to call this? I don't know ...
+void* WINAPI expCreateCompatibleDC(int hdc)
+{
+        dbgprintf("CreateCompatibleDC(%d)\n", hdc);
+        return (void*)129;
+}
 
+int WINAPI expGetDeviceCaps(int hdc, int unk)
+{
+        dbgprintf("GetDeviceCaps(%d, %d)\n", hdc, unk);
+        return 0;
+}
 
+WIN_BOOL WINAPI expDeleteDC(int hdc)
+{
+        dbgprintf("DeleteDC(%d)\n", hdc);
+        return 0;
+}
 
+int expwsprintfA(char* string, char* format, ...)
+{
+    va_list va;
+    va_start(va, format);
+    dbgprintf("wsprintfA\n");
+    return vsprintf(string, format, va);
+}
 
+int WINAPI expGetPrivateProfileIntA(char* appname, char* keyname, int default_value, char* filename)
+{
+    int size=4;
+    char* fullname;
+    dbgprintf("GetPrivateProfileIntA(%s, %s, %s)\n", appname, keyname, filename );
+    if(!(appname && keyname && filename) ) return default_value;
+    fullname=(char*)malloc(50+strlen(appname)+strlen(keyname)+strlen(filename));
+    strcpy(fullname, "Software\\IniFileMapping\\");
+    strcat(fullname, appname);
+    strcat(fullname, "\\");
+    strcat(fullname, keyname);
+    strcat(fullname, "\\");
+    strcat(fullname, filename);
+    RegQueryValueExA(HKEY_LOCAL_MACHINE, fullname, NULL, NULL, &default_value, &size);
+    free(fullname); 
+    return default_value;
+}
 
+int WINAPI expDefDriverProc(int _private, int id, int msg, int arg1, int arg2)
+{
+    printf("Called DefDriverProc(%X)\n", msg);
+    return 0;
+}    
 
+int WINAPI expSizeofResource(int v1, int v2)
+{
+    dbgprintf("SizeofResource()\n");
+    return SizeofResource(v1, v2);
+}    
 
+int WINAPI expGetLastError()
+{
+    dbgprintf("GetLastError()\n");
+    return GetLastError();
+}
 
+void WINAPI expSetLastError(int error)
+{
+    dbgprintf("SetLastError()\n");
+    SetLastError(error);
+}        
 
+char* expstrrchr(char* string, int value)
+{
+    return strrchr(string, value);
+}    
 
+char* expstrchr(char* string, int value)
+{
+    return strchr(string, value);
+}    
 
+int WINAPI expGetFileVersionInfoSizeA(const char* name, int* lpHandle)
+{
+    printf("GetFileVersionInfoSizeA(%s,0x%X)\n", name, lpHandle);
+    return 0;
+}    
 
+int WINAPI expIsBadStringPtrW(const short* string, int nchars)
+{
+    if(string==0)return 1;
+    return 0;
+}    
+extern long WINAPI InterlockedExchangeAdd( long* dest, long incr )
+{
+    long ret;
+    __asm__ __volatile__( "lock; xaddl %0,(%1)"
+                          : "=r" (ret) : "r" (dest), "0" (incr) : "memory" );
+    return ret;
+}
 
+extern long WINAPI expInterlockedIncrement( long* dest )
+{
+    return InterlockedExchangeAdd( dest, 1 ) + 1;
+}
+extern long WINAPI expInterlockedDecrement( long* dest )
+{
+    return InterlockedExchangeAdd( dest, -1 ) - 1;
+}
 
-
+extern void WINAPI expOutputDebugStringA( const char* string )
-
+{
-
+    fprintf(stderr, "DEBUG: %s\n", string);
-
+}    
-
-
-
-
-
-
 
 struct exports
 {
@@ -936,6 +1210,7 @@ struct libs
 struct exports exp_kernel32[]={
 FF(IsBadWritePtr, 357)
 FF(IsBadReadPtr, 354)
+FF(IsBadStringPtrW, -1)
 FF(DisableThreadLibraryCalls, -1)
 FF(CreateThread, -1)
 FF(CreateEventA, -1)
@@ -947,6 +1222,7 @@ FF(GetVersion, 332)
 FF(HeapCreate, 461)
 FF(HeapAlloc, -1)
 FF(HeapDestroy, -1)
+FF(HeapFree, -1)
 FF(VirtualAlloc, -1)
 FF(VirtualFree, -1)
 FF(InitializeCriticalSection, -1) 
@@ -979,6 +1255,7 @@ FF(ReleaseSemaphore, -1)
 FF(FindResourceA, -1)
 FF(LockResource, -1)
 FF(FreeResource, -1)
+FF(SizeofResource, -1)
 FF(CloseHandle, -1)
 FF(GetCommandLineA, -1)
 FF(GetEnvironmentStringsW, -1)
@@ -997,7 +1274,15 @@ FF(FreeLibrary, -1)
 FF(CreateFileMappingA, -1)
 FF(OpenFileMappingA, -1)
 FF(MapViewOfFile, -1)
+FF(UnmapViewOfFile, -1)
 FF(Sleep, -1)
+FF(GetModuleHandleA, -1)
+FF(GetPrivateProfileIntA, -1)
+FF(GetLastError, -1)
+FF(SetLastError, -1)
+FF(InterlockedIncrement, -1)
+FF(InterlockedDecrement, -1)
+FF(OutputDebugStringA, -1)
 };
 
 struct exports exp_msvcrt[]={
@@ -1006,13 +1291,17 @@ FF(_initterm, -1)
 FF(free, -1)
 {"??3@YAXPAX@Z", -1, expdelete},
 {"??2@YAPAXI@Z", -1, expnew},
+FF(strrchr, -1)
+FF(strchr, -1)
 };
 struct exports exp_winmm[]={
 FF(GetDriverModuleHandle, -1)
 FF(timeGetTime, -1)
+FF(DefDriverProc, -1)
 };
 struct exports exp_user32[]={
 FF(LoadStringA, -1)
+FF(wsprintfA, -1)
 };
 struct exports exp_advapi32[]={
 FF(RegOpenKeyA, -1)
@@ -1022,6 +1311,14 @@ FF(RegQueryValueExA, -1)
 FF(RegSetValueExA, -1)
 FF(RegCloseKey, -1)
 };
+struct exports exp_gdi32[]={
+FF(CreateCompatibleDC, -1)
+FF(GetDeviceCaps, -1)
+FF(DeleteDC, -1)
+};
+struct exports exp_version[]={
+FF(GetFileVersionInfoSizeA, -1)
+};
 #define LL(X) \
 {#X".dll", sizeof(exp_##X)/sizeof(struct exports), exp_##X},
 
@@ -1031,6 +1328,8 @@ LL(msvcrt)
 LL(winmm)
 LL(user32)
 LL(advapi32)
+LL(gdi32)
+LL(version)
 };
 
 void* LookupExternal(const char* library, int ordinal)
@@ -1070,6 +1369,7 @@ void* LookupExternalByName(const char* library, const char* name)
 {
     char* answ;
     int i,j;
+//   return (void*)ext_unknown;
     if(library==0)
     {
 	printf("ERROR: library=0\n");

libs/winloader/module.c

@@ -5,12 +5,42 @@
  */
 
 #include <assert.h>
+#include <errno.h>
 #include <fcntl.h>
-#include <stdlib.h>
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
-#include <sys/types.h>
 #include <unistd.h>
+
+#include <sys/mman.h>
+#include <sys/types.h>
+
+#ifdef __linux__
+#include <asm/unistd.h>
+#include <asm/ldt.h>
+#else
+#define LDT_ENTRIES     8192
+#define LDT_ENTRY_SIZE  8
+
+struct modify_ldt_ldt_s {
+        unsigned int  entry_number;
+        unsigned long base_addr;
+        unsigned int  limit;
+        unsigned int  seg_32bit:1;
+        unsigned int  contents:2;
+        unsigned int  read_exec_only:1;
+        unsigned int  limit_in_pages:1;
+        unsigned int  seg_not_present:1;
+        unsigned int  useable:1;
+};
+
+#define MODIFY_LDT_CONTENTS_DATA        0
+#define MODIFY_LDT_CONTENTS_STACK       1
+#define MODIFY_LDT_CONTENTS_CODE        2
+#define __NR_modify_ldt         123
+#endif
+
+
 #include <wine/windef.h>
 #include <wine/winerror.h>
 #include <wine/heap.h>
@@ -29,6 +59,132 @@ typedef struct modref_list_t
 modref_list;
 
 
+/***********************************************************************
+ *           LDT_EntryToBytes
+ *
+ * Convert an ldt_entry structure to the raw bytes of the descriptor.
+ */
+static void LDT_EntryToBytes( unsigned long *buffer, const struct modify_ldt_ldt_s *content )
+{
+    *buffer++ = ((content->base_addr & 0x0000ffff) << 16) |
+                 (content->limit & 0x0ffff);
+    *buffer = (content->base_addr & 0xff000000) |
+              ((content->base_addr & 0x00ff0000)>>16) |
+              (content->limit & 0xf0000) |
+              (content->contents << 10) |
+              ((content->read_exec_only == 0) << 9) |
+              ((content->seg_32bit != 0) << 22) |
+              ((content->limit_in_pages != 0) << 23) |
+              0xf000;
+}
+
+
+//
+// funcs:
+//
+// 0 read LDT
+// 1 write old mode
+// 0x11 write
+//
+static int modify_ldt( int func, struct modify_ldt_ldt_s *ptr,
+                                  unsigned long count )
+{
+    int res;
+#ifdef __PIC__
+    __asm__ __volatile__( "pushl %%ebx\n\t"
+                          "movl %2,%%ebx\n\t"
+                          "int $0x80\n\t"
+                          "popl %%ebx"
+                          : "=a" (res)
+                          : "0" (__NR_modify_ldt),
+                            "r" (func),
+                            "c" (ptr),
+                            "d" (sizeof(struct modify_ldt_ldt_s)*count) );
+#else
+    __asm__ __volatile__("int $0x80"
+                         : "=a" (res)
+                         : "0" (__NR_modify_ldt),
+                           "b" (func),
+                           "c" (ptr),
+                           "d" (sizeof(struct modify_ldt_ldt_s)*count) );
+#endif  /* __PIC__ */
+    if (res >= 0) return res;
+    errno = -res;
+    return -1;
+}
+static int fs_installed=0;
+static char* fs_seg=0;
+static int install_fs()
+{
+    struct modify_ldt_ldt_s array;
+    int fd;
+    int ret;
+    void* prev_struct;
+    
+    if(fs_installed)
+	return 0;
+
+    fd=open("/dev/zero", O_RDWR);
+    fs_seg=mmap((void*)0xbf000000, 0x30000, PROT_READ | PROT_WRITE, MAP_PRIVATE,
+	fd, 0);
+    if(fs_seg==0)
+    {
+	printf("ERROR: Couldn't allocate memory for fs segment\n");
+	return -1;
+    }	
+    array.base_addr=((int)fs_seg+0xffff) & 0xffff0000;
+    array.entry_number=0x1;
+    array.limit=array.base_addr+getpagesize()-1;
+    array.seg_32bit=1;
+    array.read_exec_only=0;
+    array.seg_not_present=0;
+    array.contents=MODIFY_LDT_CONTENTS_DATA;
+    array.limit_in_pages=0;
+#ifdef linux
+    ret=modify_ldt(0x1, &array, 1);
+    if(ret<0)
+    {
+	perror("install_fs");
+	MESSAGE("Couldn't install fs segment, expect segfault\n");
+    }	
+#endif /*linux*/
+
+#if defined(__NetBSD__) || defined(__FreeBSD__) || defined(__OpenBSD__)
+    {
+        long d[2];
+
+        LDT_EntryToBytes( d, &array );
+        ret = i386_set_ldt(0x1, (union descriptor *)d, 1);
+        if (ret < 0)
+        {
+            perror("install_fs");
+            MESSAGE("Did you reconfigure the kernel with \"options USER_LDT\"?\n");
+        }
+    }
+#endif  /* __NetBSD__ || __FreeBSD__ || __OpenBSD__ */
+    __asm__
+    (
+    "movl $0xf,%eax\n\t"
+    "pushw %ax\n\t"
+    "movw %ax, %fs\n\t"
+    );
+    prev_struct=malloc(8);
+    *(void**)array.base_addr=prev_struct;
+    printf("prev_struct: 0x%X\n", prev_struct);
+    close(fd);
+    
+    fs_installed=1;
+    return 0;
+};    	
+static int uninstall_fs()
+{
+    if(fs_seg==0)
+	return -1;
+    munmap(fs_seg, 0x30000);
+    return 0;
+}
+
+
 //WINE_MODREF *local_wm=NULL;
 modref_list* local_wm=NULL;
 
@@ -40,6 +196,7 @@ WINE_MODREF *MODULE_FindModule(LPCSTR m)
 	return NULL;
     while(strcmp(m, list->wm->filename))
     {
+	printf("%s: %x\n", list->wm->filename, list->wm->module);
 	list=list->prev;
 	if(list==NULL)
 	    return NULL;
@@ -59,6 +216,7 @@ void MODULE_RemoveFromList(WINE_MODREF *mod)
     {
 	free(list);
 	local_wm=NULL;
+	uninstall_fs();
 	return;
     }
     for(;list;list=list->prev)
@@ -241,330 +399,6 @@ void MODULE_DllProcessDetach( WINE_MODREF* wm, WIN_BOOL bForceDetach, LPVOID lpR
     MODULE_InitDll( wm, DLL_PROCESS_DETACH, lpReserved );
 }
 
-/*************************************************************************
- *		MODULE_DllThreadAttach
- * 
- * Send DLL thread attach notifications. These are sent in the
- * reverse sequence of process detach notification.
- *
- */
- /*
-void MODULE_DllThreadAttach( LPVOID lpReserved )
-{
-    WINE_MODREF *wm;
-
-    MODULE_InitDll( wm, DLL_THREAD_ATTACH, lpReserved );
-}*/
-
-/*************************************************************************
- *		MODULE_DllThreadDetach
- * 
- * Send DLL thread detach notifications. These are sent in the
- * same sequence as process detach notification.
- *
- */
- /*
-void MODULE_DllThreadDetach( LPVOID lpReserved )
-{
-    WINE_MODREF *wm;
-    MODULE_InitDll( wm, DLL_THREAD_DETACH, lpReserved );
-}
-*/
-
-/***********************************************************************
- *           MODULE_CreateDummyModule
- *
- * Create a dummy NE module for Win32 or Winelib.
- */
-HMODULE MODULE_CreateDummyModule( LPCSTR filename, HMODULE module32 )
-{
-    printf("MODULE_CreateDummyModule:: Not implemented\n");
-    return 0;
-}
-/*    
-HMODULE MODULE_CreateDummyModule( LPCSTR filename, HMODULE module32 )
-{
-    HMODULE hModule;
-    NE_MODULE *pModule;
-    SEGTABLEENTRY *pSegment;
-    char *pStr,*s;
-    unsigned int len;
-    const char* basename;
-    OFSTRUCT *ofs;
-    int of_size, size;
-
-    // Extract base filename 
-    basename = strrchr(filename, '\\');
-    if (!basename) basename = filename;
-    else basename++;
-    len = strlen(basename);
-    if ((s = strchr(basename, '.'))) len = s - basename;
-
-    // Allocate module 
-    of_size = sizeof(OFSTRUCT) - sizeof(ofs->szPathName)
-                    + strlen(filename) + 1;
-    size = sizeof(NE_MODULE) +
-                // loaded file info 
-                 of_size +
-                // segment table: DS,CS 
-                 2 * sizeof(SEGTABLEENTRY) +
-                 // name table 
-                 len + 2 +
-                 // several empty tables 
-                 8;
-
-    hModule = GlobalAlloc16( GMEM_MOVEABLE | GMEM_ZEROINIT, size );
-    if (!hModule) return (HMODULE)11;  // invalid exe 
-
-    FarSetOwner16( hModule, hModule );
-    pModule = (NE_MODULE *)GlobalLock16( hModule );
-
-    // Set all used entries 
-    pModule->magic            = IMAGE_OS2_SIGNATURE;
-    pModule->count            = 1;
-    pModule->next             = 0;
-    pModule->flags            = 0;
-    pModule->dgroup           = 0;
-    pModule->ss               = 1;
-    pModule->cs               = 2;
-    pModule->heap_size        = 0;
-    pModule->stack_size       = 0;
-    pModule->seg_count        = 2;
-    pModule->modref_count     = 0;
-    pModule->nrname_size      = 0;
-    pModule->fileinfo         = sizeof(NE_MODULE);
-    pModule->os_flags         = NE_OSFLAGS_WINDOWS;
-    pModule->self             = hModule;
-    pModule->module32         = module32;
-
-    // Set version and flags 
-    if (module32)
-    {
-        pModule->expected_version =
-            ((PE_HEADER(module32)->OptionalHeader.MajorSubsystemVersion & 0xff) << 8 ) |
-             (PE_HEADER(module32)->OptionalHeader.MinorSubsystemVersion & 0xff);
-        pModule->flags |= NE_FFLAGS_WIN32;
-        if (PE_HEADER(module32)->FileHeader.Characteristics & IMAGE_FILE_DLL)
-            pModule->flags |= NE_FFLAGS_LIBMODULE | NE_FFLAGS_SINGLEDATA;
-    }
-
-    // Set loaded file information 
-    ofs = (OFSTRUCT *)(pModule + 1);
-    memset( ofs, 0, of_size );
-    ofs->cBytes = of_size < 256 ? of_size : 255;   // FIXME 
-    strcpy( ofs->szPathName, filename );
-
-    pSegment = (SEGTABLEENTRY*)((char*)(pModule + 1) + of_size);
-    pModule->seg_table = (int)pSegment - (int)pModule;
-    // Data segment
-    pSegment->size    = 0;
-    pSegment->flags   = NE_SEGFLAGS_DATA;
-    pSegment->minsize = 0x1000;
-    pSegment++;
-    // Code segment
-    pSegment->flags   = 0;
-    pSegment++;
-
-    // Module name 
-    pStr = (char *)pSegment;
-    pModule->name_table = (int)pStr - (int)pModule;
-    assert(len<256);
-    *pStr = len;
-    lstrcpynA( pStr+1, basename, len+1 );
-    pStr += len+2;
-
-    // All tables zero terminated 
-    pModule->res_table = pModule->import_table = pModule->entry_table =
-		(int)pStr - (int)pModule;
-
-    NE_RegisterModule( pModule );
-    return hModule;
-}
-
-*/
-
-/***********************************************************************
- *           MODULE_GetBinaryType
- *
- * The GetBinaryType function determines whether a file is executable
- * or not and if it is it returns what type of executable it is.
- * The type of executable is a property that determines in which
- * subsystem an executable file runs under.
- *
- * Binary types returned:
- * SCS_32BIT_BINARY: A Win32 based application
- * SCS_DOS_BINARY: An MS-Dos based application
- * SCS_WOW_BINARY: A Win16 based application
- * SCS_PIF_BINARY: A PIF file that executes an MS-Dos based app
- * SCS_POSIX_BINARY: A POSIX based application ( Not implemented )
- * SCS_OS216_BINARY: A 16bit OS/2 based application
- *
- * Returns TRUE if the file is an executable in which case
- * the value pointed by lpBinaryType is set.
- * Returns FALSE if the file is not an executable or if the function fails.
- *
- * To do so it opens the file and reads in the header information
- * if the extended header information is not present it will
- * assume that the file is a DOS executable.
- * If the extended header information is present it will
- * determine if the file is a 16 or 32 bit Windows executable
- * by check the flags in the header.
- *
- * Note that .COM and .PIF files are only recognized by their
- * file name extension; but Windows does it the same way ...
- */
- /*
-static WIN_BOOL MODULE_GetBinaryType( HANDLE hfile, LPCSTR filename,
-                                  LPDWORD lpBinaryType )
-{
-    IMAGE_DOS_HEADER mz_header;
-    char magic[4], *ptr;
-    DWORD len;
-
-    // Seek to the start of the file and read the DOS header information.
-    if (    SetFilePointer( hfile, 0, NULL, SEEK_SET ) != -1  
-         && ReadFile( hfile, &mz_header, sizeof(mz_header), &len, NULL )
-         && len == sizeof(mz_header) )
-    {
-        // Now that we have the header check the e_magic field
-         // to see if this is a dos image.
-         //
-        if ( mz_header.e_magic == IMAGE_DOS_SIGNATURE )
-        {
-            WIN_BOOL lfanewValid = FALSE;
-            // We do have a DOS image so we will now try to seek into
-            // the file by the amount indicated by the field
-            // "Offset to extended header" and read in the
-            // "magic" field information at that location.
-            // This will tell us if there is more header information
-            // to read or not.
-            //
-            // But before we do we will make sure that header
-             // structure encompasses the "Offset to extended header"
-             // field.
-             //
-            if ( (mz_header.e_cparhdr<<4) >= sizeof(IMAGE_DOS_HEADER) )
-                if ( ( mz_header.e_crlc == 0 ) ||
-                     ( mz_header.e_lfarlc >= sizeof(IMAGE_DOS_HEADER) ) )
-                    if (    mz_header.e_lfanew >= sizeof(IMAGE_DOS_HEADER)
-                         && SetFilePointer( hfile, mz_header.e_lfanew, NULL, SEEK_SET ) != -1  
-                         && ReadFile( hfile, magic, sizeof(magic), &len, NULL )
-                         && len == sizeof(magic) )
-                        lfanewValid = TRUE;
-
-            if ( !lfanewValid )
-            {
-                // If we cannot read this "extended header" we will
-                 // assume that we have a simple DOS executable.
-                 //
-                *lpBinaryType = SCS_DOS_BINARY;
-                return TRUE;
-            }
-            else
-            {
-                // Reading the magic field succeeded so
-                // we will try to determine what type it is.
-                 //
-                if ( *(DWORD*)magic      == IMAGE_NT_SIGNATURE )
-                {
-                    // This is an NT signature.
-                     //
-                    *lpBinaryType = SCS_32BIT_BINARY;
-                    return TRUE;
-                }
-                else if ( *(WORD*)magic == IMAGE_OS2_SIGNATURE )
-                {
-                    // The IMAGE_OS2_SIGNATURE indicates that the
-                    // "extended header is a Windows executable (NE)
-                    // header."  This can mean either a 16-bit OS/2
-                    // or a 16-bit Windows or even a DOS program 
-                    // (running under a DOS extender).  To decide
-                    // which, we'll have to read the NE header.
-                    ///
-
-                     IMAGE_OS2_HEADER ne;
-                     if (    SetFilePointer( hfile, mz_header.e_lfanew, NULL, SEEK_SET ) != -1  
-                          && ReadFile( hfile, &ne, sizeof(ne), &len, NULL )
-                          && len == sizeof(ne) )
-                     {
-                         switch ( ne.ne_exetyp )
-                         {
-                         case 2:  *lpBinaryType = SCS_WOW_BINARY;   return TRUE;
-                         case 5:  *lpBinaryType = SCS_DOS_BINARY;   return TRUE;
-                         default: *lpBinaryType = SCS_OS216_BINARY; return TRUE;
-                         }
-                     }
-                     // Couldn't read header, so abort. 
-                     return FALSE;
-                }
-                else
-                {
-                    // Unknown extended header, but this file is nonetheless
-		     //  DOS-executable.
-                     //
-                    *lpBinaryType = SCS_DOS_BINARY;
-	            return TRUE;
-                }
-            }
-        }
-    }
-
-    // If we get here, we don't even have a correct MZ header.
-    // Try to check the file extension for known types ...
-     //
-    ptr = strrchr( filename, '.' );
-    if ( ptr && !strchr( ptr, '\\' ) && !strchr( ptr, '/' ) )
-    {
-        if ( !lstrcmpiA( ptr, ".COM" ) )
-        {
-            *lpBinaryType = SCS_DOS_BINARY;
-            return TRUE;
-        }
-
-        if ( !lstrcmpiA( ptr, ".PIF" ) )
-        {
-            *lpBinaryType = SCS_PIF_BINARY;
-            return TRUE;
-        }
-    }
-
-    return FALSE;
-}
-*/
-/***********************************************************************
- *             GetBinaryTypeA                     [KERNEL32.280]
- */
-/*
-WIN_BOOL WINAPI GetBinaryTypeA( LPCSTR lpApplicationName, LPDWORD lpBinaryType )
-{
-    WIN_BOOL ret = FALSE;
-    HANDLE hfile;
-
-    TRACE_(win32)("%s\n", lpApplicationName );
-
-    // Sanity check.
-    
-    if ( lpApplicationName == NULL || lpBinaryType == NULL )
-        return FALSE;
-
-    // Open the file indicated by lpApplicationName for reading.
-     
-    hfile = CreateFileA( lpApplicationName, GENERIC_READ, 0,
-                         NULL, OPEN_EXISTING, 0, -1 );
-    if ( hfile == INVALID_HANDLE_VALUE )
-        return FALSE;
-
-    // Check binary type
-     
-    ret = MODULE_GetBinaryType( hfile, lpApplicationName, lpBinaryType );
-
-    // Close the file.
-     
-    CloseHandle( hfile );
-
-    return ret;
-}
-*/
 
 /***********************************************************************
  *           LoadLibraryExA   (KERNEL32)
@@ -578,6 +412,8 @@ HMODULE WINAPI LoadLibraryExA(LPCSTR libname, HANDLE hfile, DWORD flags)
 		SetLastError(ERROR_INVALID_PARAMETER);
 		return 0;
 	}
+	if(fs_installed==0)
+	    install_fs();
 	    
 
 	wm = MODULE_LoadLibraryExA( libname, hfile, flags );
@@ -619,11 +455,13 @@ WINE_MODREF *MODULE_LoadLibraryExA( LPCSTR libname, HFILE hfile, DWORD flags )
         SetLastError( ERROR_FILE_NOT_FOUND );
 	TRACE("Trying native dll '%s'\n", libname);
 	pwm = PE_LoadLibraryExA(libname, flags);
+#ifdef HAVE_LIBDL
 	if(!pwm)
 	{
     	    TRACE("Trying ELF dll '%s'\n", libname);
-	    pwm=ELFDLL_LoadLibraryExA(libname, flags);
+	    pwm=(WINE_MODREF*)ELFDLL_LoadLibraryExA(libname, flags);
 	}	
+#endif	
 //		printf("0x%08x\n", pwm);
 //		break;
 	if(pwm)
@@ -761,10 +599,12 @@ FARPROC MODULE_GetProcAddress(
      	retproc = PE_FindExportedFunction( wm, function, snoop );
 	if (!retproc) SetLastError(ERROR_PROC_NOT_FOUND);
 	return retproc;
+#ifdef HAVE_LIBDL	
     case MODULE32_ELF:
-	retproc = dlsym( wm->module, function);
+	retproc = (FARPROC) dlsym( wm->module, function);
 	if (!retproc) SetLastError(ERROR_PROC_NOT_FOUND);
 	return retproc;
+#endif
     default:
     	ERR("wine_modref type %d not handled.\n",wm->type);
     	SetLastError(ERROR_INVALID_HANDLE);

libs/winloader/pe_image.c

@@ -289,7 +289,7 @@ DWORD fixup_imports( WINE_MODREF *wm )
 		break;
 
 //#warning FIXME: here we should fill imports
-        printf("Loading imports for %s\n", name);
+        TRACE("Loading imports for %s.dll\n", name);
     
 	if (pe_imp->u.OriginalFirstThunk != 0) { 
 	    TRACE("Microsoft style imports used\n");
@@ -388,7 +388,7 @@ static void do_relocations( unsigned int load_addr, IMAGE_BASE_RELOCATION *r )
 		{
 			int offset = r->TypeOffset[i] & 0xFFF;
 			int type = r->TypeOffset[i] >> 12;
-			TRACE_(fixup)("patching %x type %x\n", offset, type);
+//			TRACE_(fixup)("patching %x type %x\n", offset, type);
 			switch(type)
 			{
 			case IMAGE_REL_BASED_ABSOLUTE: break;
@@ -797,20 +797,12 @@ WINE_MODREF *PE_CreateModule( HMODULE hModule,
     wm->binfmt.pe.pe_resource = pe_resource;
     wm->binfmt.pe.tlsindex = -1;
 
-    wm->filename = strdup( filename );
+    wm->filename = malloc(strlen(filename)+1);
+    strcpy(wm->filename, filename );
     wm->modname = strrchr( wm->filename, '\\' );
     if (!wm->modname) wm->modname = wm->filename;
     else wm->modname++;
 
-//    result = GetShortPathNameA( wm->filename, NULL, 0 );
-//    wm->short_filename = (char *)HeapAlloc( GetProcessHeap(), 0, result+1 );
-//    GetShortPathNameA( wm->filename, wm->short_filename, result+1 );
-//    wm->short_modname = strrchr( wm->short_filename, '\\' );
-//    if (!wm->short_modname) wm->short_modname = wm->short_filename;
-//    else wm->short_modname++;
-//    return NULL;  
-//    }
-
     if ( pe_export )
         dump_exports( hModule );
 
@@ -836,22 +828,15 @@ WINE_MODREF *PE_CreateModule( HMODULE hModule,
  */
 WINE_MODREF *PE_LoadLibraryExA (LPCSTR name, DWORD flags)
 {
-//        struct load_dll_request *req = get_req_buffer();
 	HMODULE		hModule32;
 	WINE_MODREF	*wm;
 	char        	filename[256];
-//	HANDLE		hFile;
 	int hFile;
 	WORD		version = 0;
 
 	
-//	if ( SearchPathA( NULL, name, ".DLL", 
-//	                  sizeof(filename), filename, NULL ) == 0 ) return NULL;
 	strncpy(filename, name, sizeof(filename));      
-//	hFile = CreateFileA( filename, GENERIC_READ, FILE_SHARE_READ,
-//                             NULL, OPEN_EXISTING, 0, -1 );
 	hFile=open(filename, O_RDONLY);
-//	if ( hFile == INVALID_HANDLE_VALUE ) return NULL;
 	if(hFile==-1)
 	    return NULL;
 	
@@ -863,34 +848,12 @@ WINE_MODREF *PE_LoadLibraryExA (LPCSTR name, DWORD flags)
 		return NULL;
 	}
 
-	// Create 16-bit dummy module 
-/*
-	if ((hModule16 = MODULE_CreateDummyModule( filename, hModule32 )) < 32)
-	{
-                CloseHandle( hFile );
-		SetLastError( (DWORD)hModule16 );	// This should give the correct error 
-		return NULL;
-	}
-*/
-	
 	if ( !(wm = PE_CreateModule( hModule32, filename, flags, FALSE )) )
 	{
 		ERR( "can't load %s\n", filename );
-	//	FreeLibrary16( hModule16 );
 		SetLastError( ERROR_OUTOFMEMORY );
 		return NULL;
 	}
-    /*
-	if (wm->binfmt.pe.pe_export)
-		SNOOP_RegisterDLL(wm->module,wm->modname,wm->binfmt.pe.pe_export->NumberOfFunctions);
-		
-        req->handle     = hFile;
-        req->base       = (void *)hModule32;
-        req->dbg_offset = 0;
-        req->dbg_size   = 0;
-        req->name       = &wm->modname;
-        server_call_noerr( REQ_LOAD_DLL );
-	*/
 	close(hFile);
 	return wm;
 }
@@ -944,13 +907,6 @@ WIN_BOOL PE_InitDLL( WINE_MODREF *wm, DWORD type, LPVOID lpReserved )
     return retv;
 }
 
-/************************************************************************
- *	PE_InitTls			(internal)
- *
- * If included, initialises the thread local storages of modules.
- * Pointers in those structs are not RVAs but real pointers which have been
- * relocated by do_relocations() already.
- */
 static LPVOID
 _fixup_address(PIMAGE_OPTIONAL_HEADER opt,int delta,LPVOID addr) {
 	if (	((DWORD)addr>opt->ImageBase) &&

libs/winloader/registry.c

@@ -1,8 +1,7 @@
 #include <stdio.h>
-#include <string.h>
 #include <fcntl.h>
-#include <errno.h>
+#include <pwd.h>
-#include <sys/param.h>
+#include <sys/types.h>
 
 #include <wine/winbase.h>
 #include <wine/winreg.h>
@@ -42,8 +41,8 @@ static void save_registry();
 
 
 
-static void create_registry()
+
-{
+static void create_registry(){
     if(regs)
     {
 	printf("Logic error: create_registry() called with existing registry\n");
@@ -52,8 +51,10 @@ static void create_registry()
     }	
     regs=(struct reg_value*)malloc(3*sizeof(struct reg_value));
     regs[0].type=regs[1].type=DIR;
-    regs[0].name=strdup("HKLM");
+    regs[0].name=(char*)malloc(5);
-    regs[1].name=strdup("HKCU");
+    strcpy(regs[0].name, "HKLM");
+    regs[1].name=(char*)malloc(5);
+    strcpy(regs[1].name, "HKCU");
     regs[0].value=regs[1].value=NULL;
     regs[0].len=regs[1].len=0;
     reg_size=2;
@@ -64,14 +65,19 @@ static void open_registry()
 	int fd;
 	int i;
 	int len;
-	char user_conf[PATH_MAX+1];
+         struct passwd* pwent;
+         char* pathname;
 	if(regs)
 	{
 		printf("Multiple open_registry(>\n");
 		return;
 	}
-	snprintf(user_conf, PATH_MAX, "%s/.gstreamer/win32/registry", getenv("HOME"));
+        pwent=getpwuid(getuid());
-	fd=open(user_conf, O_RDONLY);
+        pathname=(char*)malloc(strlen(pwent->pw_dir)+20);
+	strcpy(pathname, pwent->pw_dir);
+        strcat(pathname, "/.registry");
+	fd=open(pathname, O_RDONLY);
+        free(pathname);
 	if(fd==-1)
 	{
 	    printf("Creating new registry\n");
@@ -109,21 +115,15 @@ error:
 
 static void save_registry()
 {
-	int fd, i, len, res;
+	int fd, i, len;
-	char user_conf[PATH_MAX+1];
+         struct passwd* pwent;
-
+         char* pathname;
-	snprintf(user_conf, PATH_MAX, "%s/.gstreamer/", getenv("HOME"));
+        pwent=getpwuid(getuid());
-	res=mkdir(user_conf, 00777);
+        pathname=(char*)malloc(strlen(pwent->pw_dir)+20);
-	snprintf(user_conf, PATH_MAX, "%s/.gstreamer/win32/", getenv("HOME"));
+	strcpy(pathname, pwent->pw_dir);
-	res=mkdir(user_conf, 00777);
+        strcat(pathname, "/.registry");
-	if (res == -1 && errno != EEXIST) 
+	fd=open(pathname, O_WRONLY | O_CREAT, 00777);
-	{
+        free(pathname);
-		printf("Failed to create directory %s/.gstreamer/win32.\n", getenv("HOME"));
-		perror("mkdir");
-		return;
-	}
-	snprintf(user_conf, PATH_MAX, "%s/.gstreamer/win32/registry", getenv("HOME"));
-	fd=open(user_conf, O_WRONLY | O_CREAT, 00777);
 	if(fd==-1)
 	{
 		printf("Failed to open registry file for writing.\n");
@@ -196,7 +196,8 @@ static reg_handle_t* insert_handle(long handle, const char* name)
 		t->prev=head;
 	}
 	t->next=0;
-	t->name=strdup(name);
+	t->name=(char*)malloc(strlen(name)+1);
+	strcpy(t->name, name);
 	t->handle=handle;
 	head=t;
 	return t;
@@ -207,7 +208,7 @@ static char* build_keyname(long key, const char* subkey)
 	reg_handle_t* t;
  	if((t=find_handle(key))==0)
 	{
-		printf("Invalid key\n");
+		TRACE("Invalid key\n");
 		return NULL;
 	}
 	if(subkey==NULL)
@@ -225,7 +226,7 @@ struct reg_value* insert_reg_value(int handle, const char* name, int type, void*
 	char* fullname;
 	if((fullname=build_keyname(handle, name))==NULL)
 	{
-		printf("Invalid handle\n");
+		TRACE("Invalid handle\n");
 		return NULL;
 	}
 
@@ -248,7 +249,8 @@ struct reg_value* insert_reg_value(int handle, const char* name, int type, void*
 	v->len=len;
 	v->value=(char*)malloc(len);
 	memcpy(v->value, value, len);
-	v->name=strdup(fullname);
+	v->name=(char*)malloc(strlen(fullname)+1);
+	strcpy(v->name, fullname);
 	save_registry();
 	return v;
 }
@@ -260,13 +262,13 @@ static void init_registry()
 	insert_handle(HKEY_LOCAL_MACHINE, "HKLM");
 	insert_handle(HKEY_CURRENT_USER, "HKCU");
 }
-static reg_handle_t* find_handle_2(long key, char* subkey)
+static reg_handle_t* find_handle_2(long key, const char* subkey)
 {
 	char* full_name;
 	reg_handle_t* t;
  	if((t=find_handle(key))==0)
 	{
-		printf("Invalid key\n");
+		TRACE("Invalid key\n");
 		return (reg_handle_t*)-1;
 	}
 	if(subkey==NULL)
@@ -280,16 +282,16 @@ static reg_handle_t* find_handle_2(long key, char* subkey)
 	return t;
 }
 
-long RegOpenKeyExA(long key, char* subkey, long reserved, long access, int* newkey)
+long RegOpenKeyExA(long key, const char* subkey, long reserved, long access, int* newkey)
 {
     char* full_name;
     reg_handle_t* t;
     struct reg_value* v;
-    printf("Opening key %s\n", subkey);
+    TRACE("Opening key %s\n", subkey);
     
     if(!regs)
-        init_registry();
+        init_registry()
-	
+;	
 /*	t=find_handle_2(key, subkey);
 	
 	if(t==0)
@@ -297,8 +299,8 @@ long RegOpenKeyExA(long key, char* subkey, long reserved, long access, int* newk
 
 	if(t==(reg_handle_t*)-1)
 		return -1;
-*/
+
-    full_name=build_keyname(key, subkey);
+*/    full_name=build_keyname(key, subkey);
     if(!full_name)
         return -1;
     v=find_value_by_name(full_name);    
@@ -330,14 +332,14 @@ long RegCloseKey(long key)
     free(handle);
     return 1;
 }         
-long RegQueryValueExA(long key, char* value, int* reserved, int* type, int* data, int* count)
+long RegQueryValueExA(long key, const char* value, int* reserved, int* type, int* data, int* count)
 {
 	struct reg_value* t;
 	char* c;
-	printf("Querying value %s\n", value);
+	TRACE("Querying value %s\n", value);
 	if(!regs)
-	    init_registry();
+	    init_registry()
-	
+;	
     	c=build_keyname(key, value);
 	if(c==NULL)
 		return 1;
@@ -361,17 +363,17 @@ long RegQueryValueExA(long key, char* value, int* reserved, int* type, int* data
 	}
     return 0;
 }  
-long RegCreateKeyExA(long key, char* name, long reserved,
+long RegCreateKeyExA(long key, const char* name, long reserved,
 							   void* classs, long options, long security,
 							   void* sec_attr, int* newkey, int* status) 
 {
 	reg_handle_t* t;
 	char* fullname;
 	struct reg_value* v;
-        printf("Creating/Opening key %s\n", name);
+        TRACE("Creating/Opening key %s\n", name);
 	if(!regs)
-	    init_registry();
+	    init_registry()
-	
+;	
 	fullname=build_keyname(key, name);
 	if(fullname==NULL)
 		return 1;
@@ -391,14 +393,14 @@ long RegCreateKeyExA(long key, char* name, long reserved,
 	free(fullname);
 	return 0;
 }
-long RegSetValueExA(long key, char* name, long v1, long v2, void* data, long size)
+long RegSetValueExA(long key, const char* name, long v1, long v2, void* data, long size)
 {
     struct reg_value* t;
     char* c;
-    printf("Request to set value %s\n", name);
+    TRACE("Request to set value %s\n", name);
     if(!regs)
-        init_registry();
+        init_registry()
-	
+;	
     c=build_keyname(key, name);
     if(c==NULL)
 	return 1;

libs/winloader/vfl.c

@@ -31,6 +31,21 @@ HIC	VFWAPI ICLocate(long fccType, long fccHandler, LPBITMAPINFOHEADER lpbiIn, LP
 
 #define OpenDriverA DrvOpen
 extern HDRVR VFWAPI DrvOpen(long);
+#define STORE_ALL \
+    __asm__ ( \
+    "push %%ebx\n\t" \
+    "push %%ecx\n\t" \
+    "push %%edx\n\t" \
+    "push %%esi\n\t" \
+    "push %%edi\n\t"::)
+
+#define REST_ALL \
+    __asm__ ( \
+    "pop %%edi\n\t" \
+    "pop %%esi\n\t" \
+    "pop %%edx\n\t" \
+    "pop %%ecx\n\t" \
+    "pop %%ebx\n\t"::)
 
 
 typedef struct {
@@ -249,7 +264,7 @@ ICCompress(
 long VFWAPIV 
 ICDecompress(HIC hic,long dwFlags,LPBITMAPINFOHEADER lpbiFormat,void* lpData,LPBITMAPINFOHEADER  lpbi,void* lpBits) {
 	ICDECOMPRESS	icd;
-
+	int result;
 	icd.dwFlags	= dwFlags;
 	icd.lpbiInput	= lpbiFormat;
 	icd.lpInput	= lpData;
@@ -257,7 +272,10 @@ ICDecompress(HIC hic,long dwFlags,LPBITMAPINFOHEADER lpbiFormat,void* lpData,LPB
 	icd.lpbiOutput	= lpbi;
 	icd.lpOutput	= lpBits;
 	icd.ckid	= 0;
-	return ICSendMessage(hic,ICM_DECOMPRESS,(long)&icd,sizeof(icd));
+	STORE_ALL;
+	result=ICSendMessage(hic,ICM_DECOMPRESS,(long)&icd,sizeof(icd));
+	REST_ALL;
+	return result;
 }
 
 /***********************************************************************
@@ -267,7 +285,7 @@ LRESULT VFWAPI
 ICSendMessage(HIC hic,unsigned int msg,long lParam1,long lParam2) {
 	LRESULT		ret;
 	WINE_HIC	*whic = (WINE_HIC*)hic;
-
+	char qw[200];
 #define XX(x) case x: TRACE("(0x%08lx,"#x",0x%08lx,0x%08lx)\n",(long)hic,lParam1,lParam2);break;
 /*
 	switch (msg) {
@@ -317,25 +335,12 @@ ICSendMessage(HIC hic,unsigned int msg,long lParam1,long lParam2) {
 */
 //	if (whic->driverproc) {
 //		FIXME("(0x%08lx,0x%08lx,0x%08lx,0x%08lx), calling %p\n",(long)hic,(long)msg,lParam1,lParam2,whic->driverproc);
-#define STORE_ALL \
-    __asm__ ( \
-    "push %%ebx\n\t" \
-    "push %%ecx\n\t" \
-    "push %%edx\n\t" \
-    "push %%esi\n\t" \
-    "push %%edi\n\t"::)
-
-#define REST_ALL \
-    __asm__ ( \
-    "pop %%edi\n\t" \
-    "pop %%esi\n\t" \
-    "pop %%edx\n\t" \
-    "pop %%ecx\n\t" \
-    "pop %%ebx\n\t"::)
 //	printf("private=%x\n", whic->private);
+    __asm__ __volatile__ ("fsave (%0)\n\t": :"r"(&qw));    
     STORE_ALL;	
     	ret = whic->driverproc(whic->private,1,msg,lParam1,lParam2);
     REST_ALL;	
+    __asm__ __volatile__ ("frstor (%0)\n\t": :"r"(&qw));    
 //	} else
 
 //		ret = SendDriverMessage(whic->hdrv,msg,lParam1,lParam2);

libs/winloader/wineacm.h

@@ -0,0 +1,55 @@
+/* -*- tab-width: 8; c-basic-offset: 4 -*- */
+
+/***********************************************************************
+ * Wine specific - Win32
+ */
+typedef struct _WINE_ACMDRIVERID *PWINE_ACMDRIVERID;
+typedef struct _WINE_ACMDRIVER   *PWINE_ACMDRIVER;
+
+typedef struct _WINE_ACMOBJ
+{
+  PWINE_ACMDRIVERID	pACMDriverID;
+} WINE_ACMOBJ, *PWINE_ACMOBJ;
+
+typedef struct _WINE_ACMDRIVER
+{
+    WINE_ACMOBJ		obj;
+    HDRVR      		hDrvr;
+    DRIVERPROC		pfnDriverProc;
+    PWINE_ACMDRIVER	pNextACMDriver;
+} WINE_ACMDRIVER;
+
+typedef struct _WINE_ACMSTREAM
+{
+    WINE_ACMOBJ		obj;
+    PWINE_ACMDRIVER	pDrv;
+    ACMDRVSTREAMINSTANCE drvInst;
+    HACMDRIVER		hAcmDriver;
+} WINE_ACMSTREAM, *PWINE_ACMSTREAM;
+
+typedef struct _WINE_ACMDRIVERID
+{
+    LPSTR               pszDriverAlias;
+    LPSTR               pszFileName;
+    HINSTANCE		hInstModule;          /* NULL if global */
+    DWORD		dwProcessID;	      /* ID of process which installed a local driver */
+    WIN_BOOL                bEnabled;
+    PWINE_ACMDRIVER     pACMDriverList;
+    PWINE_ACMDRIVERID   pNextACMDriverID;
+    PWINE_ACMDRIVERID	pPrevACMDriverID;
+} WINE_ACMDRIVERID;
+
+/* From internal.c */
+extern HANDLE MSACM_hHeap;
+extern PWINE_ACMDRIVERID MSACM_pFirstACMDriverID;
+extern PWINE_ACMDRIVERID MSACM_pLastACMDriverID;
+PWINE_ACMDRIVERID MSACM_RegisterDriver(
+  LPSTR pszDriverAlias, LPSTR pszFileName,
+  HINSTANCE hinstModule);
+void MSACM_RegisterAllDrivers(void);
+PWINE_ACMDRIVERID MSACM_UnregisterDriver(PWINE_ACMDRIVERID p);
+void MSACM_UnregisterAllDrivers(void);
+PWINE_ACMDRIVERID MSACM_GetDriverID(HACMDRIVERID hDriverID);
+PWINE_ACMDRIVER MSACM_GetDriver(HACMDRIVER hDriver);
+PWINE_ACMOBJ MSACM_GetObj(HACMOBJ hObj);
+