view libvo/vo_xvmc.c @ 34410:bb440bfcade9

configure: Simplify Theora check. The Theora check does some very elaborate tests to account for issues in old libtheora versions. Nowadays libtheora is stable and easy to test for, so there is no longer a need for the extra complexity. Also drop the Theora check that linked against tremor/bitwise.c. Its purpose is now lost in the mists of time.
author diego
date Wed, 04 Jan 2012 10:39:38 +0000
parents 229982126641
children 33e13b4ddfcb
line wrap: on
line source

/*
 * This file is part of MPlayer.
 *
 * MPlayer is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * MPlayer 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 MPlayer; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
 */

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>

#include "config.h"
#include "mp_msg.h"
#include "video_out.h"
#include "video_out_internal.h"
#include "osdep/timer.h"
#include "libmpcodecs/vf.h"

#include <X11/Xlib.h>
#include <X11/Xutil.h>
#include <X11/Xatom.h>

#ifdef HAVE_SHM
#include <sys/ipc.h>
#include <sys/shm.h>
#include <X11/extensions/XShm.h>
#endif

#include <X11/extensions/Xv.h>
#include <X11/extensions/Xvlib.h>
#include <X11/extensions/XvMClib.h>

#include "x11_common.h"
#include "libavcodec/avcodec.h"
#include "libavcodec/xvmc.h"

#include "sub/sub.h"
#include "aspect.h"

#include "subopt-helper.h"

#include "libavutil/common.h"

//no chance for xinerama to be supported in the near future
#undef CONFIG_XINERAMA

#undef NDEBUG
#include <assert.h>


static int benchmark;
static int use_sleep;
static int first_frame;//draw colorkey on first frame
static int use_queue;
static int xv_port_request = 0;
static int xv_adaptor = -1;
static int bob_deinterlace;
static int top_field_first;

static int image_width,image_height;
static int image_format;

#define NO_SUBPICTURE      0
#define OVERLAY_SUBPICTURE 1
#define BLEND_SUBPICTURE   2
#define BACKEND_SUBPICTURE 3

static int subpicture_mode;
static int subpicture_alloc;
static XvMCSubpicture subpicture;
static XvImageFormatValues subpicture_info;
static int subpicture_clear_color;//transparent color for the subpicture or color key for overlay

static XvMCSurfaceInfo surface_info;
static XvMCContext ctx;
static XvMCBlockArray data_blocks;
static XvMCMacroBlockArray mv_blocks;

#define MAX_SURFACES 8
static int number_of_surfaces=0;
static XvMCSurface surface_array[MAX_SURFACES];
struct xvmc_render{
    struct xvmc_pix_fmt shared;
#define STATE_DISPLAY_PENDING 1  /**  the surface should be shown, the video driver manipulates this */
#define STATE_PREDICTION      2  /**  the surface is needed for prediction, the codec manipulates this */
#define STATE_OSD_SOURCE      4  /**  the surface is needed for subpicture rendering */
    int state;
    void *p_osd_target_surface_render;
    mp_image_t *mpi;
};
static struct xvmc_render *surface_render;

static struct xvmc_render *p_render_surface_to_show = NULL;
static struct xvmc_render *p_render_surface_visible = NULL;

//display queue, kinda render ahead
static struct xvmc_render *show_queue[MAX_SURFACES];
static int free_element;


static void (*draw_osd_fnc)(int x0,int y0, int w,int h, unsigned char* src, unsigned char *srca, int stride);
static void (*clear_osd_fnc)(int x0,int y0, int w,int h);
static void (*init_osd_fnc)(void);

static void   draw_osd_AI44(int x0,int y0, int w,int h, unsigned char* src, unsigned char *srca, int stride);
static void   draw_osd_IA44(int x0,int y0, int w,int h, unsigned char* src, unsigned char *srca, int stride);
static void   clear_osd_subpic(int x0,int y0, int w,int h);
static void   init_osd_yuv_pal(void);


static const struct{
    int id;//id as xvimages or as mplayer RGB|{8,15,16,24,32}
    void (* init_func_ptr)(void);
    void (* draw_func_ptr)(int, int, int, int, unsigned char *, unsigned char *, int);
    void (* clear_func_ptr)(int, int, int, int);
    } osd_render[]={
                        {0x34344149,init_osd_yuv_pal,draw_osd_AI44,clear_osd_subpic},
                        {0x34344941,init_osd_yuv_pal,draw_osd_IA44,clear_osd_subpic},
                        {0,NULL,NULL,NULL}
                    };

static void xvmc_free(void);
static void xvmc_clean_surfaces(void);
static int count_free_surfaces(void);
static struct xvmc_render *find_free_surface(void);

static const vo_info_t info = {
    "XVideo Motion Compensation",
    "xvmc",
    "Ivan Kalvachev <iive@users.sf.net>",
    ""
};

const LIBVO_EXTERN(xvmc);

//shm stuff from vo_xv
#ifdef HAVE_SHM
static XShmSegmentInfo Shminfo;
static int Shmem_Flag;
#endif
static XvImage * xvimage;

static void add_state(struct xvmc_render *cur_render, int flags)
{
    int bit;
    if (cur_render->mpi) {
        for (bit = 1; flags >= bit; bit <<= 1) {
            if (flags & ~cur_render->state & bit)
                cur_render->mpi->usage_count++;
        }
    }
    cur_render->state |= flags;
}

static void remove_state(struct xvmc_render *cur_render, int flags)
{
    int bit;
    if (cur_render->mpi) {
        for (bit = 1; flags >= bit; bit <<= 1) {
            if (flags & cur_render->state & bit)
                cur_render->mpi->usage_count--;
        }
    }
    cur_render->state &= ~flags;
}

static int in_use(struct xvmc_render *cur_render)
{
    return cur_render->state || (cur_render->mpi && cur_render->mpi->usage_count);
}

static void allocate_xvimage(int xvimage_width,int xvimage_height,int xv_format)
{
 /*
  * allocate XvImages.  FIXME: no error checking, without
  * mit-shm this will bomb... trzing to fix ::atmos
  */
#ifdef HAVE_SHM
    if ( mLocalDisplay && XShmQueryExtension( mDisplay ) ) Shmem_Flag = 1;
    else
    {
        Shmem_Flag = 0;
        mp_msg(MSGT_VO,MSGL_WARN, "Shared memory not supported\nReverting to normal Xv\n" );
    }
    if ( Shmem_Flag )
    {
        xvimage = (XvImage *) XvShmCreateImage(mDisplay, xv_port, xv_format,
                             NULL, xvimage_width, xvimage_height, &Shminfo);

        Shminfo.shmid    = shmget(IPC_PRIVATE, xvimage->data_size, IPC_CREAT | 0777);
        Shminfo.shmaddr  = (char *) shmat(Shminfo.shmid, 0, 0);
        Shminfo.readOnly = False;

        xvimage->data = Shminfo.shmaddr;
        XShmAttach(mDisplay, &Shminfo);
        XSync(mDisplay, False);
        shmctl(Shminfo.shmid, IPC_RMID, 0);
    }
    else
#endif
    {
        xvimage = (XvImage *) XvCreateImage(mDisplay, xv_port, xv_format, NULL, xvimage_width, xvimage_height);
        xvimage->data = malloc(xvimage->data_size);
        XSync(mDisplay,False);
    }
// memset(xvimage->data,128,xvimage->data_size);
    return;
}

static void deallocate_xvimage(void)
{
#ifdef HAVE_SHM
    if ( Shmem_Flag )
    {
        XShmDetach( mDisplay,&Shminfo );
        shmdt( Shminfo.shmaddr );
    }
    else
#endif
    {
        free(xvimage->data);
    }
    XFree(xvimage);

    XSync(mDisplay, False);
    return;
}
//end of vo_xv shm/xvimage code

static int xvmc_check_surface_format(uint32_t format, XvMCSurfaceInfo * surf_info){
    if ( format == IMGFMT_XVMC_IDCT_MPEG2 ){
        if( surf_info->mc_type != (XVMC_IDCT|XVMC_MPEG_2) ) return -1;
        if( surf_info->chroma_format != XVMC_CHROMA_FORMAT_420 ) return -1;
        return 0;
    }
    if ( format == IMGFMT_XVMC_MOCO_MPEG2 ){
        if(surf_info->mc_type != XVMC_MPEG_2) return -1;
        if(surf_info->chroma_format != XVMC_CHROMA_FORMAT_420) return -1;
        return 0;
    }
return -1;//fail
}

//print all info needed to add new format
static void print_xvimage_format_values(XvImageFormatValues *xifv){
    int i;
    const int msgl=MSGL_DBG4;

    mp_msg(MSGT_VO,msgl,"Format_ID = 0x%X\n",xifv->id);

    mp_msg(MSGT_VO,msgl,"  type = ");
    if(xifv->type == XvRGB) mp_msg(MSGT_VO,msgl,"RGB\n");
    else if(xifv->type == XvYUV) mp_msg(MSGT_VO,msgl,"YUV\n");
    else mp_msg(MSGT_VO,msgl,"Unknown\n");

    mp_msg(MSGT_VO,msgl,"  byte_order = ");
    if(xifv->byte_order == LSBFirst) mp_msg(MSGT_VO,msgl,"LSB First\n");
    else if(xifv->type == MSBFirst) mp_msg(MSGT_VO,msgl,"MSB First\n");
    else mp_msg(MSGT_VO,msgl,"Unknown\n");//yes Linux support other types too

    mp_msg(MSGT_VO,msgl,"  guid = ");
    for(i=0;i<16;i++)
        mp_msg(MSGT_VO,msgl,"%02X ",(unsigned char)xifv->guid[i]);
    mp_msg(MSGT_VO,msgl,"\n");

    mp_msg(MSGT_VO,msgl,"  bits_per_pixel = %d\n",xifv->bits_per_pixel);

    mp_msg(MSGT_VO,msgl,"  format = ");
    if(xifv->format == XvPacked) mp_msg(MSGT_VO,msgl,"XvPacked\n");
    else if(xifv->format == XvPlanar) mp_msg(MSGT_VO,msgl,"XvPlanar\n");
    else mp_msg(MSGT_VO,msgl,"Unknown\n");

    mp_msg(MSGT_VO,msgl,"  num_planes = %d\n",xifv->num_planes);

    if(xifv->type == XvRGB){
        mp_msg(MSGT_VO,msgl,"  red_mask = %0X\n",  xifv->red_mask);
        mp_msg(MSGT_VO,msgl,"  green_mask = %0X\n",xifv->green_mask);
        mp_msg(MSGT_VO,msgl,"  blue_mask = %0X\n", xifv->blue_mask);
    }
    if(xifv->type == XvYUV){
        mp_msg(MSGT_VO,msgl,"  y_sample_bits = %d\n  u_sample_bits = %d\n  v_sample_bits = %d\n",
                xifv->y_sample_bits,xifv->u_sample_bits,xifv->v_sample_bits);
        mp_msg(MSGT_VO,msgl,"  horz_y_period = %d\n  horz_u_period = %d\n  horz_v_period = %d\n",
                xifv->horz_y_period,xifv->horz_u_period,xifv->horz_v_period);
        mp_msg(MSGT_VO,msgl,"  vert_y_period = %d\n  vert_u_period = %d\n  vert_v_period = %d\n",
                xifv->vert_y_period,xifv->vert_u_period,xifv->vert_v_period);

        mp_msg(MSGT_VO,msgl,"  component_order = ");
        for(i=0;i<32;i++)
            if(xifv->component_order[i]>=32)
                mp_msg(MSGT_VO,msgl,"%c",xifv->component_order[i]);
        mp_msg(MSGT_VO,msgl,"\n");

        mp_msg(MSGT_VO,msgl,"  scanline = ");
        if(xifv->scanline_order == XvTopToBottom) mp_msg(MSGT_VO,msgl,"XvTopToBottom\n");
        else if(xifv->scanline_order == XvBottomToTop) mp_msg(MSGT_VO,msgl,"XvBottomToTop\n");
        else mp_msg(MSGT_VO,msgl,"Unknown\n");
    }
    mp_msg(MSGT_VO,msgl,"\n");
}

// WARNING This function may changes xv_port and surface_info!
static int xvmc_find_surface_by_format(int format,int width,int height,
                                        XvMCSurfaceInfo * surf_info,int query){
    int rez;
    XvAdaptorInfo * ai;
    int num_adaptors,i;
    unsigned long p;
    int s,mc_surf_num;
    XvMCSurfaceInfo * mc_surf_list;

    rez = XvQueryAdaptors(mDisplay,DefaultRootWindow(mDisplay),&num_adaptors,&ai);
    if( rez != Success ) return -1;
    mp_msg(MSGT_VO,MSGL_DBG3,"vo_xvmc: Querying %d adaptors\n",num_adaptors);
    for(i=0; i<num_adaptors; i++)
    {
        /* check if adaptor number has been specified */
        if (xv_adaptor != -1 && xv_adaptor != i)
            continue;
        mp_msg(MSGT_VO,MSGL_DBG3,"vo_xvmc: Querying adaptor #%d\n",i);
        if( ai[i].type == 0 ) continue;// we need at least dummy type!
//probing ports
        for(p=ai[i].base_id; p<ai[i].base_id+ai[i].num_ports; p++)
        {
            mp_msg(MSGT_VO,MSGL_DBG3,"vo_xvmc: probing port #%ld\n",p);
            mc_surf_list = XvMCListSurfaceTypes(mDisplay,p,&mc_surf_num);
            if( mc_surf_list == NULL || mc_surf_num == 0){
                mp_msg(MSGT_VO,MSGL_DBG3,"vo_xvmc: No XvMC supported. \n");
                continue;
            }
            mp_msg(MSGT_VO,MSGL_DBG3,"vo_xvmc: XvMC list have %d surfaces\n",mc_surf_num);
//we have XvMC list!
            for(s=0; s<mc_surf_num; s++)
            {
                if( width > mc_surf_list[s].max_width ) continue;
                if( height > mc_surf_list[s].max_height ) continue;
                if( xvmc_check_surface_format(format,&mc_surf_list[s])<0 ) continue;
//we have match!
                /* respect the users wish */
                if ( xv_port_request != 0 && xv_port_request != p )
                {
                    continue;
                }

                if(!query){
                    rez = XvGrabPort(mDisplay,p,CurrentTime);
                    if(rez != Success){
                        mp_msg(MSGT_VO,MSGL_DBG3,"vo_xvmc: Fail to grab port %ld\n",p);
                        continue;
                    }
                    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: Using Xv Adaptor #%d (%s)\n", i, ai[i].name);
                    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: Port %ld grabed\n",p);
                    xv_port = p;
                }
                goto surface_found;
            }//for mc surf
            XFree(mc_surf_list);//if mc_surf_num==0 is list==NULL ?
        }//for ports
    }//for adaptors
    XvFreeAdaptorInfo(ai);

    if(!query) mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: Could not find free matching surface. Sorry.\n");
    return 0;

// somebody know cleaner way to escape from 3 internal loops?
surface_found:
    XvFreeAdaptorInfo(ai);

    memcpy(surf_info,&mc_surf_list[s],sizeof(XvMCSurfaceInfo));
    mp_msg(MSGT_VO, query?MSGL_INFO:MSGL_DBG3,
                "vo_xvmc: Found matching surface with id=%X on %ld port at %d adapter\n",
                mc_surf_list[s].surface_type_id,p,i);
    return mc_surf_list[s].surface_type_id;
}

static uint32_t xvmc_draw_image(mp_image_t *mpi){
    struct xvmc_render *rndr;

    assert(mpi!=NULL);
    assert(mpi->flags &MP_IMGFLAG_DIRECT);
//   assert(mpi->flags &MP_IMGFLAGS_DRAWBACK);

    rndr = (struct xvmc_render*)mpi->priv; //there is copy in plane[2]
    assert( rndr != NULL );
    assert( rndr->shared.xvmc_id == AV_XVMC_ID );
    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: draw_image(show rndr=%p)\n",rndr);
// the surface have passed vf system without been skiped, it will be displayed
    add_state(rndr, STATE_DISPLAY_PENDING);
    p_render_surface_to_show = rndr;
    top_field_first = mpi->fields & MP_IMGFIELD_TOP_FIRST;
    return VO_TRUE;
}

static int preinit(const char *arg){
    int xv_version,xv_release,xv_request_base,xv_event_base,xv_error_base;
    int mc_eventBase,mc_errorBase;
    int mc_ver,mc_rev;
    strarg_t ck_src_arg = { 0, NULL };
    strarg_t ck_method_arg = { 0, NULL };
    const opt_t subopts [] =
    {
        /* name         arg type      arg var           test */
        {  "port",      OPT_ARG_INT,  &xv_port_request, int_pos },
        {  "adaptor",   OPT_ARG_INT,  &xv_adaptor,      int_non_neg },
        {  "ck",        OPT_ARG_STR,  &ck_src_arg,      xv_test_ck },
        {  "ck-method", OPT_ARG_STR,  &ck_method_arg,   xv_test_ckm },
        {  "benchmark", OPT_ARG_BOOL, &benchmark,       NULL },
        {  "sleep",     OPT_ARG_BOOL, &use_sleep,       NULL },
        {  "queue",     OPT_ARG_BOOL, &use_queue,       NULL },
        {  "bobdeint",  OPT_ARG_BOOL, &bob_deinterlace, NULL },
        {  NULL }
    };

    //Obtain display handler
    if (!vo_init()) return -1;//vo_xv

    //XvMC is subdivision of XVideo
    if (Success != XvQueryExtension(mDisplay,&xv_version,&xv_release,&xv_request_base,
                                    &xv_event_base,&xv_error_base) ){
        mp_msg(MSGT_VO,MSGL_ERR,"Sorry, Xv(MC) not supported by this X11 version/driver\n");
        mp_msg(MSGT_VO,MSGL_ERR,"********** Try with  -vo x11  or  -vo sdl  ***********\n");
        return -1;
    }
    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: X-Video extension %d.%d\n",xv_version,xv_release);

    if( True != XvMCQueryExtension(mDisplay,&mc_eventBase,&mc_errorBase) ){
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: No X-Video MotionCompensation Extension on %s\n",
                XDisplayName(NULL));
        return -1;
    }

    if(Success == XvMCQueryVersion(mDisplay, &mc_ver, &mc_rev) ){
        mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: X-Video MotionCompensation Extension version %i.%i\n",
                mc_ver,mc_rev);
    }
    else{
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: Error querying version info!\n");
        return -1;
    }
    surface_render = NULL;
    xv_port = 0;
    number_of_surfaces = 0;
    subpicture_alloc = 0;

    benchmark = 0; //disable PutImageto allow faster display than screen refresh
    use_sleep = 0;
    use_queue = 0;
    bob_deinterlace = 0;

    /* parse suboptions */
    if ( subopt_parse( arg, subopts ) != 0 )
    {
        return -1;
    }

    xv_setup_colorkeyhandling( ck_method_arg.str, ck_src_arg.str );

    return 0;
}

static int config(uint32_t width, uint32_t height,
                       uint32_t d_width, uint32_t d_height,
                       uint32_t flags, char *title, uint32_t format){
    int i,mode_id,rez;
    int numblocks,blocks_per_macroblock;//bpmb we have 6,8,12

//from vo_xv
    XVisualInfo vinfo;
    XSetWindowAttributes xswa;
    XWindowAttributes attribs;
    unsigned long xswamask;
    int depth;
#ifdef CONFIG_XF86VM
    int vm = flags & VOFLAG_MODESWITCHING;
#endif
//end of vo_xv

    if( !IMGFMT_IS_XVMC(format) )
    {
        assert(0);//should never happen, abort on debug or
        return 1;//return error on relese
    }

// Find free port that supports MC, by querying adaptors
    if( xv_port != 0 || number_of_surfaces != 0 ){
        if( height==image_height && width==image_width && image_format==format){
            xvmc_clean_surfaces();
            goto skip_surface_allocation;
        }
        xvmc_free();
    };
    numblocks=((width+15)/16)*((height+15)/16);
// Find Supported Surface Type
    mode_id = xvmc_find_surface_by_format(format,width,height,&surface_info,0);//false=1 to grab port, not query
    if ( mode_id == 0 )
    {
        return -1;
    }

    rez = XvMCCreateContext(mDisplay, xv_port,mode_id,width,height,XVMC_DIRECT,&ctx);
    if( rez != Success ){
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: XvMCCreateContext failed with error %d\n",rez);
        return -1;
    }
    if( ctx.flags & XVMC_DIRECT ){
        mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: Allocated Direct Context\n");
    }else{
        mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: Allocated Indirect Context!\n");
    }


    blocks_per_macroblock = 6;
    if(surface_info.chroma_format == XVMC_CHROMA_FORMAT_422)
        blocks_per_macroblock = 8;
    if(surface_info.chroma_format == XVMC_CHROMA_FORMAT_444)
        blocks_per_macroblock = 12;

    rez = XvMCCreateBlocks(mDisplay,&ctx,numblocks*blocks_per_macroblock,&data_blocks);
    if( rez != Success ){
        XvMCDestroyContext(mDisplay,&ctx);
        return -1;
    }
    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: data_blocks allocated\n");

    rez = XvMCCreateMacroBlocks(mDisplay,&ctx,numblocks,&mv_blocks);
    if( rez != Success ){
        XvMCDestroyBlocks(mDisplay,&data_blocks);
        XvMCDestroyContext(mDisplay,&ctx);
        return -1;
    }
    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: mv_blocks allocated\n");

    if(surface_render==NULL)
        surface_render = malloc(MAX_SURFACES * sizeof(struct xvmc_render)); //easy mem debug
    memset(surface_render, 0, MAX_SURFACES * sizeof(struct xvmc_render));

    for(i=0; i<MAX_SURFACES; i++){
        rez=XvMCCreateSurface(mDisplay,&ctx,&surface_array[i]);
        if( rez != Success )
            break;
        surface_render[i].shared.xvmc_id = AV_XVMC_ID;
        surface_render[i].shared.data_blocks = data_blocks.blocks;
        surface_render[i].shared.mv_blocks = mv_blocks.macro_blocks;
        surface_render[i].shared.allocated_mv_blocks = numblocks;
        surface_render[i].shared.allocated_data_blocks = numblocks*blocks_per_macroblock;
        surface_render[i].shared.idct = (surface_info.mc_type & XVMC_IDCT) == XVMC_IDCT;
        surface_render[i].shared.unsigned_intra = (surface_info.flags & XVMC_INTRA_UNSIGNED) == XVMC_INTRA_UNSIGNED;
        surface_render[i].shared.p_surface = &surface_array[i];
        mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: surface[%d] = %p .rndr=%p\n",
                i,&surface_array[i], &surface_render[i]);
    }
    number_of_surfaces = i;
    if( number_of_surfaces < 4 ){// +2 I or P and +2 for B (to avoid visible motion drawing)
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: Unable to allocate at least 4 Surfaces\n");
        uninit();
        return -1;
    }
    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: Motion Compensation context allocated - %d surfaces\n",
            number_of_surfaces);

    //debug
    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: idct=%d unsigned_intra=%d\n",
            (surface_info.mc_type & XVMC_IDCT) == XVMC_IDCT,
            (surface_info.flags & XVMC_INTRA_UNSIGNED) == XVMC_INTRA_UNSIGNED);

// Find way to display OSD & subtitle
    mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: looking for OSD support\n");
    subpicture_mode = NO_SUBPICTURE;
    if(surface_info.flags & XVMC_OVERLAID_SURFACE)
        subpicture_mode = OVERLAY_SUBPICTURE;

    if(surface_info.subpicture_max_width  != 0 &&
       surface_info.subpicture_max_height != 0  ){
        int s,k,num_subpic;

        XvImageFormatValues * xvfmv;
        xvfmv = XvMCListSubpictureTypes(mDisplay, xv_port,
                        surface_info.surface_type_id, &num_subpic);

        if(num_subpic != 0 && xvfmv != NULL){
            if( mp_msg_test(MSGT_VO,MSGL_DBG4) ){//Print all subpicture types for debug
                for(s=0;s<num_subpic;s++)
                    print_xvimage_format_values(&xvfmv[s]);
            }

            for(s=0;s<num_subpic;s++){
                for(k=0;osd_render[k].draw_func_ptr!=NULL;k++){
                    if(xvfmv[s].id == osd_render[k].id)
                    {
                        init_osd_fnc  = osd_render[k].init_func_ptr;
                        draw_osd_fnc  = osd_render[k].draw_func_ptr;
                        clear_osd_fnc = osd_render[k].clear_func_ptr;

                        subpicture_mode = BLEND_SUBPICTURE;
                        subpicture_info = xvfmv[s];
                        mp_msg(MSGT_VO,MSGL_INFO,"    Subpicture id 0x%08X\n",subpicture_info.id);
                        goto found_subpic;
                    }
                }
            }
found_subpic:
            XFree(xvfmv);
        }
        //Blend2 supicture is always possible, blend1 only at backend
        if( (subpicture_mode == BLEND_SUBPICTURE) &&
            (surface_info.flags & XVMC_BACKEND_SUBPICTURE) )
        {
            subpicture_mode = BACKEND_SUBPICTURE;
        }

    }

    switch(subpicture_mode){
        case NO_SUBPICTURE:
            mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: No OSD support for this mode\n");
            break;
        case OVERLAY_SUBPICTURE:
            mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: OSD support via color key tricks\n");
            mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: not yet implemented:(\n");
            break;
        case BLEND_SUBPICTURE:
            mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: OSD support by additional frontend rendering\n");
            break;
        case BACKEND_SUBPICTURE:
            mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: OSD support by backend rendering (fast)\n");
            mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: Please send feedback to confirm that it works,otherwise send bugreport!\n");
            break;
    }

//take keycolor value and choose method for handling it
    if ( !vo_xv_init_colorkey() )
    {
        return -1; // bail out, colorkey setup failed
    }

    vo_xv_enable_vsync();//it won't break anything

//taken from vo_xv
    image_height = height;
    image_width = width;

skip_surface_allocation:

    {
#ifdef CONFIG_XF86VM
        if ( vm )
        {
            vo_vm_switch();
        }
//        else
#endif
        XGetWindowAttributes(mDisplay, DefaultRootWindow(mDisplay), &attribs);
        depth=attribs.depth;
        if (depth != 15 && depth != 16 && depth != 24 && depth != 32) depth = 24;
        XMatchVisualInfo(mDisplay, mScreen, depth, TrueColor, &vinfo);

        xswa.border_pixel     = 0;
        xswamask = CWBorderPixel;
        if (xv_ck_info.method == CK_METHOD_BACKGROUND){
            xswa.background_pixel = xv_colorkey;
            xswamask |= CWBackPixel;
        }

        vo_x11_create_vo_window(&vinfo, vo_dx, vo_dy, d_width, d_height, flags,
                                CopyFromParent, "xvmc", title);
        XChangeWindowAttributes(mDisplay, vo_window, xswamask, &xswa);

#ifdef CONFIG_XF86VM
        if ( vm )
        {
            /* Grab the mouse pointer in our window */
            if(vo_grabpointer)
                XGrabPointer(mDisplay, vo_window, True, 0,
                             GrabModeAsync, GrabModeAsync,
                             vo_window, None, CurrentTime );
            XSetInputFocus(mDisplay, vo_window, RevertToNone, CurrentTime);
        }
#endif
    }

//end vo_xv

    /* store image dimesions for displaying */
    p_render_surface_visible = NULL;
    p_render_surface_to_show = NULL;

    free_element = 0;
    first_frame = 1;

    image_format=format;
    return 0;
}

static int draw_frame(uint8_t *srcp[]){
    assert(0);
}

static void init_osd_yuv_pal(void) {
    char * palette;
    int rez;
    int i,j;
    int snum,seb;
    int Y,U,V;

    subpicture_clear_color = 0;

    if(subpicture.num_palette_entries > 0){

        snum = subpicture.num_palette_entries;
        seb = subpicture.entry_bytes;
        palette = malloc(snum*seb);//check fail
        if(palette == NULL) return;
        for(i=0; i<snum; i++){
            // 0-black max-white the other are gradients
            Y = i*(1 << subpicture_info.y_sample_bits)/snum;//snum=2;->(0),(1*(1<<1)/2)
            U = 1 << (subpicture_info.u_sample_bits - 1);
            V = 1 << (subpicture_info.v_sample_bits - 1);
            for(j=0; j<seb; j++)
                switch(subpicture.component_order[j]){
                    case 'U': palette[i*seb+j] = U; break;
                    case 'V': palette[i*seb+j] = V; break;
                    case 'Y':
                    default:
                              palette[i*seb+j] = Y; break;
                }
        }
        rez = XvMCSetSubpicturePalette(mDisplay, &subpicture, palette);
        if(rez!=Success){
            mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: Setting palette failed.\n");
        }
        free(palette);
    }
}

static void clear_osd_subpic(int x0, int y0, int w, int h){
    int rez;

    rez=XvMCClearSubpicture(mDisplay, &subpicture,
                            x0, y0, w,h,
                            subpicture_clear_color);
    if(rez != Success)
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: XvMCClearSubpicture failed!\n");
}

static void OSD_init(void) {
    unsigned short osd_height, osd_width;
    int rez;

    if(subpicture_alloc){
        mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: destroying subpicture\n");
        XvMCDestroySubpicture(mDisplay,&subpicture);
        deallocate_xvimage();
        subpicture_alloc = 0;
    }

/*   if(surface_info.flags & XVMC_SUBPICTURE_INDEPENDENT_SCALING){
      osd_width = vo_dwidth;
      osd_height = vo_dheight;
   }else*/
    {
        osd_width = image_width;
        osd_height = image_height;
    }

    if(osd_width > surface_info.subpicture_max_width)
        osd_width = surface_info.subpicture_max_width;
    if(osd_height > surface_info.subpicture_max_height)
        osd_height = surface_info.subpicture_max_height;
    if(osd_width == 0 || osd_height == 0)
        return;//if called before window size is known

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: creating subpicture (%d,%d) format %X\n",
                osd_width,osd_height,subpicture_info.id);

    rez = XvMCCreateSubpicture(mDisplay,&ctx,&subpicture,
                                osd_width,osd_height,subpicture_info.id);
    if(rez != Success){
        subpicture_mode = NO_SUBPICTURE;
        mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: Create Subpicture failed, OSD disabled\n");
        return;
    }
    if( mp_msg_test(MSGT_VO,MSGL_DBG4) ){
        int i;
        mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: Created Subpicture:\n");
        mp_msg(MSGT_VO,MSGL_DBG4,"         xvimage_id=0x%X\n",subpicture.xvimage_id);
        mp_msg(MSGT_VO,MSGL_DBG4,"         width=%d\n",subpicture.width);
        mp_msg(MSGT_VO,MSGL_DBG4,"         height=%d\n",subpicture.height);
        mp_msg(MSGT_VO,MSGL_DBG4,"         num_palette_entries=0x%X\n",subpicture.num_palette_entries);
        mp_msg(MSGT_VO,MSGL_DBG4,"         entry_bytes=0x%X\n",subpicture.entry_bytes);

        mp_msg(MSGT_VO,MSGL_DBG4,"         component_order=\"");
        for(i=0; i<4; i++)
            if(subpicture.component_order[i] >= 32)
                mp_msg(MSGT_VO,MSGL_DBG4,"%c", subpicture.component_order[i]);
        mp_msg(MSGT_VO,MSGL_DBG4,"\"\n");
    }

    //call init for the surface type
    init_osd_fnc();//init palete,clear color etc ...
    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: clearing subpicture\n");
    clear_osd_fnc(0, 0, subpicture.width, subpicture.height);

    allocate_xvimage(subpicture.width, subpicture.height, subpicture_info.id);
    subpicture_alloc = 1;
}

static void draw_osd_IA44(int x0,int y0, int w,int h, unsigned char* src, unsigned char *srca, int stride){
    int ox,oy;
    int rez;

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc:composite AI44 subpicture (%d,%d - %d,%d)\n",x0,y0,w,h);

    for(ox=0; ox<w; ox++){
        for(oy=0; oy<h; oy++){
            xvimage->data[oy*xvimage->width+ox] = (src[oy*stride+ox]>>4) | ((0-srca[oy*stride+ox])&0xf0);
        }
    }
    rez = XvMCCompositeSubpicture(mDisplay, &subpicture, xvimage, 0, 0,
                                    w,h,x0,y0);
    if(rez != Success){
        mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: composite subpicture failed\n");
        assert(0);
    }
}

static void draw_osd_AI44(int x0,int y0, int w,int h, unsigned char* src, unsigned char *srca, int stride){
    int ox,oy;
    int rez;

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc:composite AI44 subpicture (%d,%d - %d,%d)\n",x0,y0,w,h);

    for(ox=0; ox<w; ox++){
        for(oy=0; oy<h; oy++){
            xvimage->data[oy*xvimage->width+ox] = (src[oy*stride+ox]&0xf0) | (((0-srca[oy*stride+ox])>>4)&0xf);
        }
    }
    rez = XvMCCompositeSubpicture(mDisplay, &subpicture, xvimage, 0, 0,
                                    w,h,x0,y0);
    if(rez != Success){
        mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: composite subpicture failed\n");
        assert(0);
    }
}

static void draw_osd(void){
    struct xvmc_render *osd_rndr;
    int osd_has_changed;
    int have_osd_to_draw;
    int rez;

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: draw_osd ,OSD_mode=%d, surface_to_show=%p\n",
                subpicture_mode,p_render_surface_to_show);

    if(subpicture_mode == BLEND_SUBPICTURE ||
        subpicture_mode == BACKEND_SUBPICTURE ){

        if(!subpicture_alloc) //allocate subpicture when dimensions are known
            OSD_init();
        if(!subpicture_alloc)
            return;//dimensions still unknown.

        osd_has_changed = vo_update_osd(subpicture.width, subpicture.height);
        have_osd_to_draw = vo_osd_check_range_update(0, 0, subpicture.width,
                                                         subpicture.height);

        if(!have_osd_to_draw)
            return;//nothing to draw,no subpic, no blend

        if(osd_has_changed){
            //vo_remove_text(subpicture.width, subpicture.height,clear_osd_fnc)
            clear_osd_fnc(0,0,subpicture.width,subpicture.height);
            vo_draw_text(subpicture.width, subpicture.height, draw_osd_fnc);
        }
        XvMCSyncSubpicture(mDisplay,&subpicture);//todo usleeep wait!

        if(subpicture_mode == BLEND_SUBPICTURE){
            osd_rndr = find_free_surface();
            if(osd_rndr == NULL)
                return;// no free surface to draw OSD in

            rez = XvMCBlendSubpicture2(mDisplay,
                        p_render_surface_to_show->shared.p_surface, osd_rndr->shared.p_surface,
                        &subpicture,
                        0, 0, subpicture.width, subpicture.height,
                        0, 0, image_width, image_height);
            if(rez!=Success){
                mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: BlendSubpicture failed rez=%d\n",rez);
                assert(0);
                return;
            }
//       XvMCFlushSurface(mDisplay,osd_rndr->p_surface);//fixme- should I?

            //When replaceing the surface with osd one, save the flags too!
            osd_rndr->shared.picture_structure = p_render_surface_to_show->shared.picture_structure;
//add more if needed    osd_rndr-> = p_render_surface_to_show->;

            add_state(p_render_surface_to_show, STATE_OSD_SOURCE);
            remove_state(p_render_surface_to_show, STATE_DISPLAY_PENDING);
            p_render_surface_to_show->p_osd_target_surface_render = osd_rndr;

            p_render_surface_to_show = osd_rndr;
            add_state(p_render_surface_to_show, STATE_DISPLAY_PENDING);

            mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc:draw_osd: surface_to_show changed to %p\n",osd_rndr);
        }//endof if(BLEND)
        if(subpicture_mode == BACKEND_SUBPICTURE){
            rez = XvMCBlendSubpicture(mDisplay,
                        p_render_surface_to_show->shared.p_surface,
                        &subpicture,
                        0, 0, subpicture.width, subpicture.height,
                        0, 0, image_width, image_height);

        }

    }//if(BLEND||BACKEND)
}

static void xvmc_sync_surface(XvMCSurface * srf){
    int status,rez;

    rez = XvMCGetSurfaceStatus(mDisplay,srf,&status);
    assert(rez==Success);
    if((status & XVMC_RENDERING) == 0)
        return;//surface is already complete
    if(use_sleep){
        rez = XvMCFlushSurface(mDisplay, srf);
        assert(rez==Success);

        do{
            usec_sleep(1000);//1ms (may be 20ms on linux)
            XvMCGetSurfaceStatus(mDisplay,srf,&status);
        } while (status & XVMC_RENDERING);
        return;//done
    }

    XvMCSyncSurface(mDisplay, srf);
}

static void put_xvmc_image(struct xvmc_render *p_render_surface,
                           int draw_ck){
    int rez;
    struct vo_rect src_rect, dst_rect;
    int i;

    if(p_render_surface == NULL)
        return;

    calc_src_dst_rects(image_width, image_height, &src_rect, &dst_rect, NULL, NULL);

    if(draw_ck)
        vo_xv_draw_colorkey(dst_rect.left, dst_rect.top, dst_rect.width, dst_rect.height);

    if(benchmark)
        return;

    for (i = 1; i <= bob_deinterlace + 1; i++) {
        int field = top_field_first ? i : i ^ 3;
        rez = XvMCPutSurface(mDisplay, p_render_surface->shared.p_surface,
                            vo_window,
                            src_rect.left, src_rect.top, src_rect.width, src_rect.height,
                            dst_rect.left, dst_rect.top, dst_rect.width, dst_rect.height,
                            bob_deinterlace ? field : 3);
        if(rez != Success){
            mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: PutSurface failer, critical error %d!\n",rez);
            assert(0);
        }
    }
    XFlush(mDisplay);
}

static void flip_page(void){
    int i,cfs;

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: flip_page  show(rndr=%p)\n\n",p_render_surface_to_show);

    if(p_render_surface_to_show == NULL) return;
    assert( p_render_surface_to_show->shared.xvmc_id == AV_XVMC_ID );
//fixme   assert( p_render_surface_to_show != p_render_surface_visible);

    if(use_queue){
        // fill the queue until only n free surfaces remain
        // after that start displaying
        cfs = count_free_surfaces();
        show_queue[free_element++] = p_render_surface_to_show;
        if(cfs > 3){//well have 3 free surfaces after add queue
            if(free_element > 1)//a little voodoo magic
                xvmc_sync_surface(show_queue[0]->shared.p_surface);
            return;
        }
        p_render_surface_to_show=show_queue[0];
        mp_msg(MSGT_VO,MSGL_DBG5,"vo_xvmc: flip_queue free_element=%d\n",free_element);
        free_element--;
        for(i=0; i<free_element; i++){
            show_queue[i] = show_queue[i+1];
        }
        show_queue[free_element] = NULL;
    }

// make sure the rendering is done
    xvmc_sync_surface(p_render_surface_to_show->shared.p_surface);

//the visible surface won't be displayed anymore, mark it as free
    if(p_render_surface_visible != NULL)
        remove_state(p_render_surface_visible, STATE_DISPLAY_PENDING);

//!!fixme   assert(p_render_surface_to_show->state & STATE_DISPLAY_PENDING);

    //show it, displaying is always vsynced, so skip it for benchmark
    put_xvmc_image(p_render_surface_to_show,first_frame);
    first_frame=0;//make sure we won't draw it anymore

    p_render_surface_visible = p_render_surface_to_show;
    p_render_surface_to_show = NULL;
}

static void check_events(void){
    int e=vo_x11_check_events(mDisplay);

    if(e&VO_EVENT_RESIZE)
    {
        e |= VO_EVENT_EXPOSE;
    }
    if ( e & VO_EVENT_EXPOSE )
    {
        put_xvmc_image(p_render_surface_visible,1);
    }
}

static void xvmc_free(void){
    int i;

    if( subpicture_alloc ){

        XvMCDestroySubpicture(mDisplay,&subpicture);
        deallocate_xvimage();

        subpicture_alloc = 0;

        mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: subpicture destroyed\n");
    }

    if( number_of_surfaces ){

        XvMCDestroyMacroBlocks(mDisplay,&mv_blocks);
        XvMCDestroyBlocks(mDisplay,&data_blocks);

        for(i=0; i<number_of_surfaces; i++)
        {
            XvMCHideSurface(mDisplay,&surface_array[i]);//it doesn't hurt, I hope
            XvMCDestroySurface(mDisplay,&surface_array[i]);

            if( (surface_render[i].state != 0) &&
                (p_render_surface_visible != &surface_render[i]) )
                mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc::uninit surface_render[%d].status=%d\n",i,
                        surface_render[i].state);
        }

        memset(surface_render, 0, MAX_SURFACES * sizeof(struct xvmc_render)); //for debugging
        free(surface_render);surface_render=NULL;

        XvMCDestroyContext(mDisplay,&ctx);
        number_of_surfaces = 0;

        mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: Context sucessfuly freed\n");
    }


    if( xv_port !=0 ){
        XvUngrabPort(mDisplay,xv_port,CurrentTime);
        xv_port = 0;
        mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: xv_port sucessfuly ungrabed\n");
    }
}

static void uninit(void){
    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: uninit called\n");
    xvmc_free();
 //from vo_xv
#ifdef CONFIG_XF86VM
    vo_vm_close();
#endif
    vo_x11_uninit();
}

static int query_format(uint32_t format){
    uint32_t flags;
    XvMCSurfaceInfo qsurface_info;
    int mode_id;

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: query_format=%X\n",format);

    if(!IMGFMT_IS_XVMC(format)) return 0;// no caps supported
    mode_id = xvmc_find_surface_by_format(format, 16, 16, &qsurface_info, 1);//true=1 - querying

    if( mode_id == 0 ) return 0;

    flags = VFCAP_CSP_SUPPORTED |
            VFCAP_CSP_SUPPORTED_BY_HW |
            VFCAP_ACCEPT_STRIDE;

    if( (qsurface_info.subpicture_max_width  != 0) &&
        (qsurface_info.subpicture_max_height != 0) )
        flags|=VFCAP_OSD;
    return flags;
}


static int draw_slice(uint8_t *image[], int stride[],
                           int w, int h, int x, int y){
    struct xvmc_render *rndr;
    int rez;

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: draw_slice y=%d\n",y);

    rndr = (struct xvmc_render*)image[2]; //this is copy of priv-ate
    assert( rndr != NULL );
    assert( rndr->shared.xvmc_id == AV_XVMC_ID );

    rez = XvMCRenderSurface(mDisplay,&ctx,rndr->shared.picture_structure,
                            rndr->shared.p_surface,
                            rndr->shared.p_past_surface,
                            rndr->shared.p_future_surface,
                            rndr->shared.flags,
                            rndr->shared.filled_mv_blocks_num,rndr->shared.start_mv_blocks_num,
                            &mv_blocks,&data_blocks);
    if(rez != Success)
    {
    int i;
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc::slice: RenderSirface returned %d\n",rez);

        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc::slice: pict=%d,flags=%x,start_blocks=%d,num_blocks=%d\n",
                rndr->shared.picture_structure,rndr->shared.flags,rndr->shared.start_mv_blocks_num,
                rndr->shared.filled_mv_blocks_num);
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc::slice: this_surf=%p, past_surf=%p, future_surf=%p\n",
                rndr->shared.p_surface,rndr->shared.p_past_surface,rndr->shared.p_future_surface);

        for(i=0; i<rndr->shared.filled_mv_blocks_num; i++){
            XvMCMacroBlock* testblock;
            testblock = &mv_blocks.macro_blocks[i];

            mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc::slice: mv_block - x=%d,y=%d,mb_type=0x%x,mv_type=0x%x,mv_field_select=%d\n",
                    testblock->x,testblock->y,testblock->macroblock_type,
                    testblock->motion_type,testblock->motion_vertical_field_select);
            mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc::slice: dct_type=%d,data_index=0x%x,cbp=%d,pad0=%d\n",
                    testblock->dct_type,testblock->index,testblock->coded_block_pattern,
                    testblock->pad0);
            mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc::slice: PMV[0][0][0/1]=(%d,%d)\n",
                    testblock->PMV[0][0][0],testblock->PMV[0][0][1]);
        }
    }
    assert(rez==Success);
    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: flush surface\n");
    rez = XvMCFlushSurface(mDisplay, rndr->shared.p_surface);
    assert(rez==Success);

//   rndr->start_mv_blocks_num += rndr->filled_mv_blocks_num;
    rndr->shared.start_mv_blocks_num = 0;
    rndr->shared.filled_mv_blocks_num = 0;

    rndr->shared.next_free_data_block_num = 0;

    return VO_TRUE;
}

//XvMCHide hides the surface on next retrace, so
//check if the surface is not still displaying
static void check_osd_source(struct xvmc_render *src_rndr) {
    struct xvmc_render *osd_rndr;
    int stat;

    //If this is source surface, check does the OSD rendering is compleate
    if(src_rndr->state & STATE_OSD_SOURCE){
        mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: OSD surface=%p querying\n",src_rndr);
        osd_rndr = src_rndr->p_osd_target_surface_render;
        XvMCGetSurfaceStatus(mDisplay, osd_rndr->shared.p_surface, &stat);
        if(!(stat & XVMC_RENDERING))
            remove_state(src_rndr, STATE_OSD_SOURCE);
    }
}
static int count_free_surfaces(void) {
    int i,num;

    num=0;
    for(i=0; i<number_of_surfaces; i++){
        check_osd_source(&surface_render[i]);
        if(!in_use(surface_render + i))
            num++;
    }
    return num;
}

static struct xvmc_render *find_free_surface(void) {
    int i,t;
    int stat;
    struct xvmc_render *visible_rndr;

    visible_rndr = NULL;
    for(i=0; i<number_of_surfaces; i++){

        check_osd_source(&surface_render[i]);
        if( !in_use(surface_render + i)){
            XvMCGetSurfaceStatus(mDisplay, surface_render[i].shared.p_surface,&stat);
            if( (stat & XVMC_DISPLAYING) == 0 )
                return &surface_render[i];
            visible_rndr = &surface_render[i];// remember it, use as last resort
        }
    }

    //all surfaces are busy, but there is one that will be free
    //on next monitor retrace, we just have to wait
    if(visible_rndr != NULL){
        mp_msg(MSGT_VO,MSGL_INFO,"vo_xvmc: waiting retrace\n");
        for(t=0;t<1000;t++){
            usec_sleep(1000);//1ms
            XvMCGetSurfaceStatus(mDisplay, visible_rndr->shared.p_surface,&stat);
            if( (stat & XVMC_DISPLAYING) == 0 )
                return visible_rndr;
        }
    }
//todo remove when stable
    mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: no free surfaces, this should not happen in g1\n");
    for(i=0;i<number_of_surfaces;i++)
        mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: surface[%d].state=%d\n",i,surface_render[i].state);
    return NULL;
}

static void xvmc_clean_surfaces(void){
    int i;

    for(i=0; i<number_of_surfaces; i++){

        remove_state(surface_render + i, STATE_DISPLAY_PENDING | STATE_OSD_SOURCE);
        surface_render[i].p_osd_target_surface_render=NULL;
        if(surface_render[i].state != 0){
            mp_msg(MSGT_VO,MSGL_WARN,"vo_xvmc: surface[%d].state=%d\n",
                                   i,surface_render[i].state);
        }
    }
    free_element=0;//clean up the queue
}

static uint32_t get_image(mp_image_t *mpi){
    struct xvmc_render *rndr;

    rndr = find_free_surface();

    if(rndr == NULL){
        mp_msg(MSGT_VO,MSGL_ERR,"vo_xvmc: get_image failed\n");
        return VO_FALSE;
    }

assert(rndr->shared.start_mv_blocks_num == 0);
assert(rndr->shared.filled_mv_blocks_num == 0);
assert(rndr->shared.next_free_data_block_num == 0);

    mpi->flags |= MP_IMGFLAG_DIRECT;
//keep strides 0 to avoid field manipulations
    mpi->stride[0] = 0;
    mpi->stride[1] = 0;
    mpi->stride[2] = 0;

// these are shared!! so watch out
// do call RenderSurface before overwriting
    mpi->planes[0] = (char*)data_blocks.blocks;
    mpi->planes[1] = (char*)mv_blocks.macro_blocks;
    mpi->priv =
    mpi->planes[2] = (char*)rndr;

    rndr->shared.picture_structure = 0;
    rndr->shared.flags = 0;
    rndr->state = 0;
    rndr->mpi = mpi;
    rndr->shared.start_mv_blocks_num = 0;
    rndr->shared.filled_mv_blocks_num = 0;
    rndr->shared.next_free_data_block_num = 0;

    mp_msg(MSGT_VO,MSGL_DBG4,"vo_xvmc: get_image: rndr=%p (surface=%p) \n",
                rndr,rndr->shared.p_surface);
return VO_TRUE;
}

static int control(uint32_t request, void *data)
{
    switch (request){
        case VOCTRL_GET_DEINTERLACE:
            *(int*)data = bob_deinterlace;
            return VO_TRUE;
        case VOCTRL_SET_DEINTERLACE:
            bob_deinterlace = *(int*)data;
            return VO_TRUE;
        case VOCTRL_QUERY_FORMAT:
            return query_format(*((uint32_t*)data));
        case VOCTRL_DRAW_IMAGE:
            return xvmc_draw_image((mp_image_t *)data);
        case VOCTRL_GET_IMAGE:
            return get_image((mp_image_t *)data);
        //vo_xv
        case VOCTRL_GUISUPPORT:
            return VO_TRUE;
        case VOCTRL_ONTOP:
            vo_x11_ontop();
            return VO_TRUE;
        case VOCTRL_FULLSCREEN:
            vo_x11_fullscreen();
        // indended, fallthrough to update panscan on fullscreen/windowed switch
        case VOCTRL_SET_PANSCAN:
            if ( ( vo_fs && ( vo_panscan != vo_panscan_amount ) ) || ( !vo_fs && vo_panscan_amount ) )
            {
                int old_y = vo_panscan_y;
                panscan_calc();

                if(old_y != vo_panscan_y)
                {
                    //this also draws the colorkey
                    put_xvmc_image(p_render_surface_visible,1);
                }
            }
            return VO_TRUE;
        case VOCTRL_GET_PANSCAN:
            if ( !vo_config_count || !vo_fs ) return VO_FALSE;
            return VO_TRUE;
        case VOCTRL_SET_EQUALIZER:
        {
            vf_equalizer_t *eq=data;
            return vo_xv_set_eq(xv_port, eq->item, eq->value);
        }

        case VOCTRL_GET_EQUALIZER:
        {
            vf_equalizer_t *eq=data;
            return vo_xv_get_eq(xv_port, eq->item, &eq->value);
        }
        case VOCTRL_UPDATE_SCREENINFO:
            update_xinerama_info();
            return VO_TRUE;
    }
    return VO_NOTIMPL;
}