MSP430F5438A: Instruction cycle differences with jmp $+2

Jürgen Wissenwasser

Other Parts Discussed in Thread: MSP430F5438A

Hi!

With the MSP430F5438A I have some strange behaviour that I'd like to mention with the following code snippet:

For oscilloscope measurement I simply did the following:

                bis.b   #BIT2,&P8OUT
                jmp     $+2
                nop
                bic.b   #BIT2,&P8OUT
                jmp     $+2
                nop
                bis.b   #BIT2,&P8OUT
                jmp     $+2
                nop
                bic.b   #BIT2,&P8OUT
                jmp     $+2
                nop

followed by

                bis.b   #BIT2,&P8OUT
                nop
                jmp     $+2
                bic.b   #BIT2,&P8OUT
                nop
                jmp     $+2
                bis.b   #BIT2,&P8OUT
                nop
                jmp     $+2
                bic.b   #BIT2,&P8OUT
                nop
                jmp     $+2

The first sequence takes 8 clock cylces for keeping the pin high or low (1 too much) and the second version is correct with 7 cycles.

What's the problem with the jmp instruction?

over 9 years ago

0 Jürgen Wissenwasser over 9 years ago

Intellectual 490 points

Edit: This problem does ONLY occur in the case of bis- and bic-operations, also if I utlize mapping to the timer, i.e. bis/bic #BIT2,&TA0CCTL2.
It does NOT occur in the case of MOV operations, i.e. mov #0,&TA0CCTL2, mov #BIT2,&TA0CCTL2 or mov.b R15,&P8OUT, mov.b R14,&P8OUT (where R15 and R14 hold properly set values with respect to other port pin settings).

0 Katie Pier over 9 years ago in reply to Jürgen Wissenwasser

TI__Guru 70306 points

Hi wissenwasserj,

I think this other post might be helpful for you: e2e.ti.com/.../356887

Regards,
Katie

**Attention** This is a public forum

MSP low-power microcontrollers

MSP low-power microcontroller forum

MSP430F5438A: Instruction cycle differences with jmp $+2