Failed to play multiple audio with ALSA DMIX

I am using alsa-lib-1.2.9 in my embedded Linux system, and now I want to play multiple audio from different threads, so I googled internet (I am NOT familiar with ALSA) and found I can use dmix. Here is the main part of my /etc/asound.conf in the board.

pcm.!default
{
        type asym
        playback.pcm "play_softvol"
        capture.pcm "cap_chn0"
}

pcm.play_softvol {
        type softvol
        slave {
                pcm play_chn0
        }

        control {
                name "Speaker Volume"
        }

        min_dB -60.0
        max_dB -10.0
        resolution 50
}

pcm.play_chn0 {
        type plug
        slave {
                pcm dmixer
        }
}

pcm.dmixer {
        type dmix
        ipc_key 77235
        ipc_key_add_uid true
        slave {
                pcm "hw:0,0"
                period_time 0
                period_size 320
                buffer_size 2560
                rate 32000
        }
}

ctl.dmixer {
        type hw
        card 0
}

pcm.cap_chn0 {
        type plug
        slave {
                pcm dsnooper
        }
}

pcm.tloop_cap {
        type plug
        slave.pcm "hw:Loopback,0,0"
}

pcm.dsnooper {
        type dsnoop
        ipc_key 77236
        ipc_key_add_uid true
        slave {
                pcm "hw:0,0"
                channels 4
                rate 16000
        }
        bindings {
                0 0
                1 1
                2 2
                3 3
        }
}

ctl.dsnooper {
        type hw
        card 0
}

I think the default device is using play_softvol->play_chn0->dmix. So I tried to play 2 PCM files as follows,

#include 
#include 
#include 
#include 

pthread_t tid, tid2;
static snd_pcm_t *playback_handle;
static int size;
static snd_pcm_uframes_t frames;

void *play_func(void *arg)
{
    char *buffer;
    int ret;
    FILE *fp = fopen(arg, "rb");
    if(fp == NULL)
        return 0;

    buffer = (char *) malloc(size);
    fprintf(stderr, "size = %d\n", size);

    while (1)
    {
        ret = fread(buffer, 1, size, fp);
        if(ret == 0)
        {
            fprintf(stderr, "end of file on input\n");
            break;
        }

        while(ret = snd_pcm_writei(playback_handle, buffer, frames) 2)
        ret = pthread_create(&tid2, NULL, play_func2, argv);

        pthread_join(tid, NULL);
        if (argc > 2)
        pthread_join(tid2, NULL);

    snd_pcm_close(playback_handle);

    return 0;
}

asound_threads 1.pcm 2.pcm
play song 1.pcm and 2.pcm
size = 320
size = 320

But it sounds messy, the two PCMs are played in an interleaved way, but the speed is wrong and flittered. So in my case, how can I play multiple audios (audio mixing)?