Could someone .. explain this bit of code in RF12demo ?

(2 posts) (2 voices)

Tags:

No tags yet.

raalst

I am reading the RF12demo code, and I think the code below might be imperfect. dfcommand issues a CLI code on the end. this means dfWriteCmd switches off interrupts and never switches them on again. perhaps the dfdeselect and last dfcommand should switch places in dfWriteCmd?

if this code is correct in it's current form, why does it not cause interrupt issues ?

     void dfcommand (byte cmd) {
      for (;;) {
        cli();
        dfenable();
        dfxfer(0x05); // Read Status Register
        byte status = dfxfer(0);
        dfdisable();
        sei();
        // don't wait for ready bit if there is clearly no dataflash connected
        if (status == 0xFF || (status & 1) == 0)
            break;
      }

     cli();
     dfenable();
     dfxfer(cmd);
    }

    static void dfwriteCmd (byte cmd) {
      dfcommand(0x06); // Write Enable
      dfdeselect();
      dfcommand(cmd);
    }

    static void dfdeselect () {
      dfdisable();
      sei();
    }


				Posted 11/07/2010 10:54:57


	
		
			
								
					jcw

				
			
			
				
You need to check all the places where dfWriteCmd() gets called. It does indeed return with interrupts disabled (and the DataFlash enabled) - but in each case the caller does a few things and then ends by calling dfdeselect().
I think all the code paths are correct in this sense. I have not seen errors with JeeLinks running for months on end. But this is tricky stuff, and should have been documented, I admit.
The code is sub-optimal, in that a page write to DataFlash will lock out the RFM12B too long to keep up when a packet comes in. A similar issue occurs with the EtherCard driver, which also uses SPI.



				Posted 12/07/2010 00:27:04