postprocess.c

        PPContext *c= av_malloc(sizeof(PPContext));
        int stride= (width+15)&(~15);    //assumed / will realloc if needed
        int qpStride= (width+15)/16 + 2; //assumed / will realloc if needed

        global_init();

        memset(c, 0, sizeof(PPContext));
        c->cpuCaps= cpuCaps;
        if(cpuCaps&PP_FORMAT){
                c->hChromaSubSample= cpuCaps&0x3;
                c->vChromaSubSample= (cpuCaps>>4)&0x3;
        }else{
                c->hChromaSubSample= 1;
                c->vChromaSubSample= 1;
        }

        reallocBuffers(c, width, height, stride, qpStride);

        c->frameNum=-1;

        return c;
}

void pp_free_context(void *vc){
        PPContext *c = (PPContext*)vc;
        int i;

        for(i=0; i<3; i++) av_free(c->tempBlured[i]);
        for(i=0; i<3; i++) av_free(c->tempBluredPast[i]);

        av_free(c->tempBlocks);
        av_free(c->yHistogram);
        av_free(c->tempDst);
        av_free(c->tempSrc);
        av_free(c->deintTemp);
        av_free(c->stdQPTable);
        av_free(c->nonBQPTable);
        av_free(c->forcedQPTable);

        memset(c, 0, sizeof(PPContext));

        av_free(c);
}

void  pp_postprocess(uint8_t * src[3], int srcStride[3],
                 uint8_t * dst[3], int dstStride[3],
                 int width, int height,
                 QP_STORE_T *QP_store,  int QPStride,
                 pp_mode_t *vm,  void *vc, int pict_type)
{
        int mbWidth = (width+15)>>4;
        int mbHeight= (height+15)>>4;
        PPMode *mode = (PPMode*)vm;
        PPContext *c = (PPContext*)vc;
        int minStride= MAX(ABS(srcStride[0]), ABS(dstStride[0]));
        int absQPStride = ABS(QPStride);

        // c->stride and c->QPStride are always positive
        if(c->stride < minStride || c->qpStride < absQPStride)
                reallocBuffers(c, width, height,
                                MAX(minStride, c->stride),
                                MAX(c->qpStride, absQPStride));

        if(QP_store==NULL || (mode->lumMode & FORCE_QUANT))
        {
                int i;
                QP_store= c->forcedQPTable;
                absQPStride = QPStride = 0;
                if(mode->lumMode & FORCE_QUANT)
                        for(i=0; i<mbWidth; i++) QP_store[i]= mode->forcedQuant;
                else
                        for(i=0; i<mbWidth; i++) QP_store[i]= 1;
        }
//printf("pict_type:%d\n", pict_type);

        if(pict_type & PP_PICT_TYPE_QP2){
                int i;
                const int count= mbHeight * absQPStride;
                for(i=0; i<(count>>2); i++){
                        ((uint32_t*)c->stdQPTable)[i] = (((uint32_t*)QP_store)[i]>>1) & 0x7F7F7F7F;
                }
                for(i<<=2; i<count; i++){
                        c->stdQPTable[i] = QP_store[i]>>1;
                }
                QP_store= c->stdQPTable;
                QPStride= absQPStride;
        }

if(0){
int x,y;
for(y=0; y<mbHeight; y++){
        for(x=0; x<mbWidth; x++){
                printf("%2d ", QP_store[x + y*QPStride]);
        }
        printf("\n");
}
        printf("\n");
}

        if((pict_type&7)!=3)
        {
                if (QPStride >= 0) {
                        int i;
                        const int count= mbHeight * QPStride;
                        for(i=0; i<(count>>2); i++){
                                ((uint32_t*)c->nonBQPTable)[i] = ((uint32_t*)QP_store)[i] & 0x3F3F3F3F;
                        }
                        for(i<<=2; i<count; i++){
                                c->nonBQPTable[i] = QP_store[i] & 0x3F;
                        }
                } else {
                        int i,j;
                        for(i=0; i<mbHeight; i++) {
                                    for(j=0; j<absQPStride; j++) {
                                        c->nonBQPTable[i*absQPStride+j] = QP_store[i*QPStride+j] & 0x3F;
                                }
                        }
                }
        }

        if(verbose>2)
        {
                printf("using npp filters 0x%X/0x%X\n", mode->lumMode, mode->chromMode);
        }

        postProcess(src[0], srcStride[0], dst[0], dstStride[0],
                width, height, QP_store, QPStride, 0, mode, c);

        width  = (width )>>c->hChromaSubSample;
        height = (height)>>c->vChromaSubSample;

        if(mode->chromMode)
        {
                postProcess(src[1], srcStride[1], dst[1], dstStride[1],
                        width, height, QP_store, QPStride, 1, mode, c);
                postProcess(src[2], srcStride[2], dst[2], dstStride[2],
                        width, height, QP_store, QPStride, 2, mode, c);
        }
        else if(srcStride[1] == dstStride[1] && srcStride[2] == dstStride[2])
        {
                linecpy(dst[1], src[1], height, srcStride[1]);
                linecpy(dst[2], src[2], height, srcStride[2]);
        }
        else
        {
                int y;
                for(y=0; y<height; y++)
                {
                        memcpy(&(dst[1][y*dstStride[1]]), &(src[1][y*srcStride[1]]), width);
                        memcpy(&(dst[2][y*dstStride[2]]), &(src[2][y*srcStride[2]]), width);
                }
        }
}