Floppy keeps spinning after fread

[Csoft]

Hello all,

I'm developing a game in C, in my code I load assets from floppy with fread/fclose functions, it works, but the floppy keeps spinning even after the load is completed.

How can I solve it?

Thank you in advance,

Lorenzo

[Galahad/FLT]

Originally Posted by Csoft

Hello all,

I'm developing a game in C, in my code I load assets from floppy with fread/fclose functions, it works, but the floppy keeps spinning even after the load is completed.

How can I solve it?

Thank you in advance,

Lorenzo

Are you shutting down the system after loading files or stopping multitasking with Forbid?

[Csoft]

Originally Posted by Galahad/FLT

Are you shutting down the system after loading files or stopping multitasking with Forbid?

No, I'm not shutting down the system at all. I'm trying to keep everything OS friendly.

[alkis]

what's the flow?
fopen/fread/fclose? (c library)
Open/Read/Close? (OS)

[phx]

Originally Posted by Csoft

No, I'm not shutting down the system at all. I'm trying to keep everything OS friendly.

Then there is no reason it could ever happen. And it should be easy to isolate the problem with some test code doing fopen/fread/fclose.

Is it some strange file system? Or standard FFS?
Note, that a running motor for a few seconds after the last access is normal.

[Csoft]

Originally Posted by phx

Then there is no reason it could ever happen. And it should be easy to isolate the problem with some test code doing fopen/fread/fclose.

Is it some strange file system? Or standard FFS?
Note, that a running motor for a few seconds after the last access is normal.

Simply this, with standard Amiga FFS, it keeps spinning forever:

FILE *fp = fopen(filename, "rb");
    if (fp) {
        fread(dest, sizeof(unsigned char), size, fp);
    } else {
        printf("error: file '%s' not found\n", filename);
    }
    fclose(fp);

[Thomas Richter]

Please post a complete source. For example, you could run into a buffer overrun here, overwriting some Os variables if "dest" is smaller than "size" bytes. Thus, what is "dest", and "size", and how are they allocated? A standard "fread" (or an underlying Read()) will not cause this.

[Asman]

Originally Posted by Csoft

Simply this, with standard Amiga FFS, it keeps spinning forever:

FILE *fp = fopen(filename, "rb");
    if (fp) {
        fread(dest, sizeof(unsigned char), size, fp);
    } else {
        printf("error: file '%s' not found\n", filename);
    }
    fclose(fp);

What about such code, it still keeps spinning forever ?

FILE *fp = fopen(filename, "rb");

if (fp)
{
    fread(dest, sizeof(unsigned char), size, fp);
    fclose(fp);
}
else
{
    printf("error: file '%s' not found\n", filename);
}

Furthermore you can also print value from fclose, should be 0.

[Olaf Barthel]

Originally Posted by Csoft

Hello all,

I'm developing a game in C, in my code I load assets from floppy with fread/fclose functions, it works, but the floppy keeps spinning even after the load is completed.

Which operating system version are you developing and testing your game on? Are you using timer.device I/O requests in your game? Any chance you might be using a vertical blanking interrupt handler in your game?

It's the job of the file system to tell the floppy disk drive to spin down and give it a rest. This should happen within 3-4 seconds of no file system activity taking place, such as the last read or write operation. Because the file system uses timer.device to schedule periodic tests for activity, accidentally knocking out timer.device processing may cause the spindown never to take place (trackdisk.device does not spin down on its own accord).

Question is whether your game code might be responsible for that.

[Csoft]

Originally Posted by Olaf Barthel

Which operating system version are you developing and testing your game on? Are you using timer.device I/O requests in your game? Any chance you might be using a vertical blanking interrupt handler in your game?

It's the job of the file system to tell the floppy disk drive to spin down and give it a rest. This should happen within 3-4 seconds of no file system activity taking place, such as the last read or write operation. Because the file system uses timer.device to schedule periodic tests for activity, accidentally knocking out timer.device processing may cause the spindown never to take place (trackdisk.device does not spin down on its own accord).

Question is whether your game code might be responsible for that.

I'm working on 3.0. No use of timer.device, but I DO use a vblank interrupt, yes. Could be this the source of the problem?

[Thomas Richter]

If the system timer is not run, yes. What is the return code of your code? Note, in particular, that your code shall return with Z flag set as otherwise the exec interrupt handler assumes that your code handled the interrupt completely, and will not call the remaining system interrupt handlers, including the one of the timer.device.

[Csoft]

Originally Posted by Thomas Richter

If the system timer is not run, yes. What is the return code of your code? Note, in particular, that your code shall return with Z flag set as otherwise the exec interrupt handler assumes that your code handled the interrupt completely, and will not call the remaining system interrupt handlers, including the one of the timer.device.

I'm puzzled. I tried to not stop the system with Permit(), I disabled the vblank interrupt but I have a mod playroutine that starts a CIA interrupt. I do the following:

Read everything I must read from disk.
Wait for 5 seconds.
The floppy stops to spin.
After 5 seconds I install the CIA interrupt and the floppy restarts to spin!!!

I'm using FS-Uae.

[hooverphonique]

Originally Posted by Csoft

Read everything I must read from disk.
Wait for 5 seconds.
The floppy stops to spin.
After 5 seconds I install the CIA interrupt and the floppy restarts to spin!!!

I'm using FS-Uae.

Assuming the emulator isn't faulty, I'd say your CIA ISR setup is (wrongly) messing with some CIA registers which then causes the floppy to turn on.

[Thomas Richter]

Originally Posted by Csoft

I'm puzzled. I tried to not stop the system with Permit(), I disabled the vblank interrupt but I have a mod playroutine that starts a CIA interrupt.

If you remove the system VBLank, then the timer.device will stop working, and by that the file system lost all its time information, including the timerequest it generated to stop the floppy motor.


In short: You cannot "partially" stop AmigaOs. Use the Os, dude!

[Csoft]

Originally Posted by Thomas Richter

If you remove the system VBLank, then the timer.device will stop working, and by that the file system lost all its time information, including the timerequest it generated to stop the floppy motor.


In short: You cannot "partially" stop AmigaOs. Use the Os, dude!

I didn't disable the system VBlank, I removed completely my VBlank interrupt installation by commenting it.

This problem persists even if I don't disable the OS at all.

Probabily is the ptplayer routine which do some mess with CIA, I'll check it.

[Csoft]

Originally Posted by Csoft

I didn't disable the system VBlank, I removed completely my VBlank interrupt installation by commenting it.

This problem persists even if I don't disable the OS at all.

Probabily is the ptplayer routine which do some mess with CIA, I'll check it.

No. It's not. It seems pure madness. I tried the following:

1) I do NOT call Forbid() anymore.
2) I do NOT install my VBlank
3) I do NOT install CIA interrupt (ptplayer)
3) After all disk read I wait 10 seconds and the spin stops.

I found the source of the problem:

4) The first time I play a sound the spin sound restarts!!!!!!!

This happens on both FS-Uae and Winuae, so I think it's not an emulator related issue.

[Thomas Richter]

Paula is unrelated to the floppy motor control, which is in the hands of CIA, resp. the trackdisk.device controlling CIA. Thus, if this is related to playing sounds, probably something is wrong with your player function.

[Csoft]

Originally Posted by Thomas Richter

Paula is unrelated to the floppy motor control, which is in the hands of CIA, resp. the trackdisk.device controlling CIA. Thus, if this is related to playing sounds, probably something is wrong with your player function.

I'm not doing anything special with my sound playing function, here it is

UWORD audioDmaBits[] = {
    DMAF_AUD0,
    DMAF_AUD1,
    DMAF_AUD2,
    DMAF_AUD3
};

/**
 * @brief Play the sound
 * 
 * Period = 1 / (samplerate * 2,81937 * 10^-7)
 *          
 * @param sound 
 * @param period 
 * @param volume 
 * @param audiochannel 
 */
void playsound(sound* sound,UWORD period,UWORD volume,USHORT audiochannel) {

    if (!mt_Enable) {
        
        // Stop audiochannel's DMA in case there is an already playing sound in the audiochannel
        custom.dmacon = audioDmaBits[audiochannel];    
        custom.aud[audiochannel].ac_ptr = sound->sampledata;
        custom.aud[audiochannel].ac_len = sound->size;
        custom.aud[audiochannel].ac_per = period;
        custom.aud[audiochannel].ac_vol = volume;
        custom.dmacon = DMAF_SETCLR | audioDmaBits[audiochannel];    // Start sound playback

        
    } 
    // The following is called if I'm using ptplayer, which have its own playback facilities
    // the problem IS present even in this case
    else {
        SfxStructure pt_sound;
        pt_sound.sfx_ptr = sound->sampledata;
        pt_sound.sfx_len = sound->size;
        pt_sound.sfx_per = period;
        pt_sound.sfx_vol = volume;
        pt_sound.sfx_cha = audiochannel;
        pt_sound.sfx_pri = 127;
        mt_playfx(&custom,&pt_sound);
    }
}

Cannot understand why the spin stops after n seconds and then restarts if I play a sound...

[Galahad/FLT]

When the floppy spin sound restarts, does the floppy light onscreen turn on with the sound or does it stay off?

[a/b]

audioChannel is 0-3 and audioDmaBits is { 1, 2, 4, 8 }?