Why do I get inconsistent clock rate for PRU 28-bit shift input mode?

[Yogi Salomo]

Hello everybody,

So I am trying to make a PRU program that reads Input to BBB in 28-bit shift mode. The following is my current code:

START:

//INITIALIZATION

    // Enable the OCP master port -- allows transfer of data to Linux userspace

LBCO    r0, C4, 4, 4     // * load SYSCFG reg into r0 (use c4 const addr)

CLR     r0, r0, 4        // * clear bit 4 (STANDBY_INIT)

SBCO    r0, C4, 4, 4     // * store the modified r0 back at the load addr

//SETUP GPI FOR BIT-SHIFT INPUT

LBCO r0, C4, 8, 4 // * load SYSCFG reg into r0 (use c4 const addr)

LDI r0.b0, 0xAA // 0-1: 0x2 (01), 2: 0x0 (0), 3-7: 0x15 (10101)

LDI r0.b1, 0x2 // 8-12: 0x2 (01)

SBCO r0, C4, 8, 4 // * store the modified r0 back to the CFG register

MOV r1, 0x00010000 // r1: Shared RAM Address

MOV r2, 1 // r2: High Value

MOV r4, 0 // r4: Low Value

MOV r3, 0 // r3: input value

SBBO r4, r1, 0 , 2

//WAIT UNTIL START BIT IS SET

WBS r31.t29

READDATA:

QBBC READDATA,r31.t28

CALL TOGGLEFLAG

SBBO r31.w0, r1, 1, 2

LBCO r0, C4, 8, 4 // * load SYSCFG reg into r0 (use c4 const addr)

SET r0.t13

SBCO r0, C4, 8, 4

QBA READDATA

END:

MOV r31.b0, PRU0_R31_VEC_VALID | PRU_EVTOUT_0

HALT                     // End of program -- below are the "procedures"

TOGGLEFLAG:

QBBC SETT2, r30.t2

CLR r30.t2

RET

SETT2:

SET r30.t2

RET

The idea behind this:

• READDATA is the main loop
  
• The program will wait until Cnt_16 flag is set and 16 bits are sampled
• write the read value to RAM and toggle flag (for debugging)
  
• Set Start Bit flag in GPCFG register to clear shift register

However, when I checked in oscilloscope, I found some weird behavior, that is every 4th iteration of this loop, the duration this program takes to toggle r30.t2 is different (e.q. 1.8626 microseconds, 1.8755 microseconds, 1.8705 microseconds, 1.9818 microseconds). Is this normal or is there any problem in my implementation?

Moreover, if you have any suggestion about best practice in implementing what I intended to do (or coding PRU program overall), please let me know. I am very new to PRU programming and there is nobody in my lab that has prior knowledge about this subject.

Thank you for your kind attention. Looking forward to hear from you all.

Regards,

Yogi Salomo