View Full Version : Confused about DMA buffers [C]

March 13th, 2009, 12:45 PM
Hi all,
First off thanks for taking the time to help me out ;)

I'm writing a program that should read from /dev/dsp in C.

Some of the code:

/* open /dev/dsp */

#define CHANNELS 2 /* 2 = stereo */
#define SIZE 8 /* sample size: 8 bits */

int sc_open() {
int ffd = 0;
int rate = 44100;
int arg;
int status;

ffd = open("/dev/dsp", O_RDONLY);
if (ffd < 0) {
perror("open of /dev/dsp failed");

/* set sampling parameters */
arg = SIZE; /* sample size */
status = ioctl(ffd, SOUND_PCM_WRITE_BITS, &arg);
if (status == -1)
perror("SOUND_PCM_WRITE_BITS ioctl failed");
if (arg != SIZE)
perror("unable to set sample size");

arg = CHANNELS; /* mono or stereo */
status = ioctl(ffd, SOUND_PCM_WRITE_CHANNELS, &arg);
if (status == -1)
perror("SOUND_PCM_WRITE_CHANNELS ioctl failed");
if (arg != CHANNELS)
perror("unable to set number of channels");

arg = rate; /* sampling rate */
status = ioctl(ffd, SOUND_PCM_WRITE_RATE, &rate);
if (status == -1)
perror("SOUND_PCM_WRITE_WRITE ioctl failed");

return ffd;

/* create stream */
FILE* sc_open_stream ( int ffd ){
FILE *fd;
fd = fdopen(ffd, "r");
if (fd < 0) {
perror("Failed to connect stream to /dev/dsp");
return fd;

And that's where I get confused. I start reading from fd using fgetc and get a byte at a time. But between one reading and another I need to do some operations (like applying a simple software low pass filter) until I got 1000 samples. I know I could just read all samples and then do the post processing, but I would like all to be as close to real time as possible.

I guess I could do some multithreading, one thread could read /dev/dsp and store the samples in a buffer while the another thread could do the processing. But it doesn't sound like a good solution, because if the CPU will get busy the first thread might miss some samples and mess up my data.

I thought DMA buffers are there exactly for the purpose of buffering data, so that an application can read the data at it's own speed without worrying about missing sample, but I have no clue how to set this up and use it.

Many thanks to everyone.


March 13th, 2009, 04:12 PM
Do you really need to manually handle /dev/dsp?
Why not use alsa-lib?

If it's for some educational purpose or something that alsa-lib doesn't provide, then I have nothing more to say. But, I doubt there is anything that alsa-lib doesn't provide. You can use it as a fairly low level API and it's the standard Linux audio subsystem.

For more details:


Couple of tutorial links:

Feel free to disagree :)

March 13th, 2009, 06:16 PM
Thanks, I'll look into it. It might just be what I'm looking for.

One more question:
Can an application that uses libalsa be ported to windows with cygwin?

Many thanks,

March 13th, 2009, 06:48 PM
One more question:
Can an application that uses libalsa be ported to windows with cygwin?

Unfortunately, no.
ALSA stands for Advanced Linux Sound Architecture - it's specific to Linux.

However, if you want to write portable code, you might want to take a look at SDL:

It's a cross-platform multimedia programming library - includes various subsystems like video, audio, OpenGL etc.

On linux, it will use whatever is available to talk to the audio h/w - it's completely customizable.
On windows, it will use the native Win32 API (or DirectX - I'm not sure) for audio - so, you don't need an emulation like cygwin for it.

Here are some links for getting SDL working on Windows:

There is also JACK:

which is cross-platform. But, I believe, it requires POSIX API (cygwin) :(.

HTH ;)