#2420 pCode optimization produces incorrect code for modulo operator

[Juho Salminen]

When the attached file is compiled with:

sdcc --use-non-free -mpic14 -p16f887 -c test.c

the code line

u8Digit = (uint8_t)((uint16_t)n16 % (uint16_t)10);

behaves as if n16 is always 0.

In the produced assembly, the MOVFs that should be copying the n16 operand to the stack are commented out. The result is that W and STK00 both contain 0x00.

_00108_DS_
;       .line   33; "test.c"    u8Digit = (uint8_t)((uint16_t)n16 % (uint16_t)10);
        MOVLW   0x0a
        MOVWF   STK02
        MOVLW   0x00
        MOVWF   STK01
;;102   MOVF    r0x100D,W
;;107   MOVF    r0x1008,W
        MOVWF   STK00
;;104   MOVF    r0x100E,W
;;109   MOVF    r0x1007,W
        PAGESEL __moduint
        CALL    __moduint

When compiling with

sdcc --no-pcode-opt --use-non-free -mpic14 -p16f887 -c test.c

the result is as intended.

sdcc -v:

SDCC : mcs51/z80/z180/r2k/r3ka/gbz80/tlcs90/ds390/pic16/pic14/TININative/ds400/hc08/s08/stm8 3.5.4 #9317 (Linux)