At 13:32 3/11/99 -0800, you wrote:
>I believe I've just found a really nasty hardware bug in the 16F876.  I
>figure I would tell you guys first then figure out who at MChip to give a
>piece of my mind.
>
>Here's the scenario : I'm using pin 0 of port A to wakeup an external
>device.  This involves toggling the pin.  If only pin 0 is written to, a 0
>never comes out on the port.  To get it to write something another pin on
>port A must also be written to.
>
>My bare bones software test cases are :
>1: xor pin 0 with 1
>2: delay
>
>In this case pin 0 stays a 1, constantly.  If I change it to :
>
>1: xor pin 0 with 1
>2: delay
>3: write a 1 to pin 1,2,3 or 5;
>4: write a 0 to pin 1,2,3, or 5;
>
>Then it works.  Otherwise nix.  It operates in exactly the same manner with
>bsf and bcf on that particular pin.  I didn't see anything in the errata on
>this.  I'm more than a bit irritated.  If someone can show me how this is a
>bug on my part I'd be grateful (as well as surprised).
>
>Here's the source for working :
>
>m015    MOVLW .1
>00F0 1283  0529         BCF   0x03,RP0
>00F1 1303  0530         BCF   0x03,RP1
>00F2 0685  0531         XORWF PORTA,1
>00F3 0000  0533         NOP
>00F4 0000  0535         NOP
>00F5 0000  0537         NOP
>00F6 0000  0539         NOP
>00F7 0000  0541         NOP
>00F8 1685  0544         BSF   PORTA,5
>00F9 1285  0546         BCF   PORTA,5
>00FA 28EF  0549         GOTO  m015
>
>And for not working :
>m015    MOVLW .1
>00F0 1283  0529         BCF   0x03,RP0
>00F1 1303  0530         BCF   0x03,RP1
>00F2 0685  0531         XORWF PORTA,1
>00F3 0000  0533         NOP
>00F4 0000  0535         NOP
>00F5 0000  0537         NOP
>00F6 0000  0539         NOP
>00F7 0000  0541         NOP
>00F8 28EF  0547         GOTO  m015
>
>


Is this not a read write modify type instruction?
Where to XORWF with portA it must read PORTA, in this case it is an output
so portA has an "undefined" input value on 0.



Dennis