Post by thread:Someone tell me if I'm blind

Hello All, I am trying to debug a simple routine that adds two 32 bit numbers and I can't figure out why it keeps setting bits in the hi order bytes inappropriately. If you will take a look at this code and give some suggestions they would be much appreciated. The two arguments are in eight locations. Arg. 1 is, from least to most significant, in registers da1_l da1_m da1_h and da1_x. Arg. 2 is in da2_l da2_m da2_h and da2_x. On exit the sum is supposed to be in da1_l da1_m da1_h and da1_x. Here it is: movf da2_l, W addwf da1_l, SAME btfsc CARRY incf da2_m, SAME btfsc ZERO incf da2_h, SAME btfsc ZERO incf da2_x, SAME movf da2_m, W addwf da1_m, SAME btfsc CARRY incf da2_h, SAME btfsc ZERO incf da2_x, SAME movf da2_h, W addwf da1_h, SAME btfsc CARRY incf da2_x, SAME movf da2_x, W addwf da1_x, SAME If you look at this code Many Thanks Keith Causey

Keith Causey wrote: >Hello All, > I am trying to debug a simple routine that adds two 32 bit numbers and I >can't figure out why it keeps setting bits in the hi order bytes >inappropriately. If you will take a look at this code and give some >suggestions they would be much appreciated. <snipped down to 1st byte section> > movf da2_l, W > addwf da1_l, SAME > btfsc CARRY > incf da2_m, SAME > btfsc ZERO > incf da2_h, SAME > btfsc ZERO > incf da2_x, SAME First I will assume SAME is defined as 1 and CARRY and ZERO are defined to correctly reference the status bits. Look at the case of an input of all 0. First addwf will results in clear carry, set zero. Skips the first incf, but then the "btfsc ZERO" will not skip, incrementing the da2_h value. A much faster and simpler approach is: movf da2_l,w ; add first two bytes. addwf da1_1,f ; add second two bytes with carry in and carry out movf da2_m,w skpnc ; this is a predefined macro in MPASM, skip if no carry. incfsz da2_m,w ; note use of "sz" version of incf, carry, zero not changed. ; if above hits 0 then next add is skipped, carry is preserved! addwf da1_m,f ; do next to lowest add. ; add third two bytes with carry in and carry out. movf da2_h,w skpnc incfsz da2_h,w addwf da1_h,f ; repeat again for fourth section. movf da2_x,w skpnc incfsz da2_x,w addwf da1_x,f Chip Weller

Hi, Your add routine may not work if the result of da2_l+da_1 has a carry and da2_m is equal to 255 since it will become zero and the next addition will be wrong. Use the standard routine: add32: movf da2_l,w addwf da1_l,f movf da2_m,w skpnc incfsz da2_m,w addwf da1_m,f movf da2_h,w skpnc incfsz da2_h,w addwf da1_h,f regards, Reggie Keith Causey wrote: {Quote hidden}> > Hello All, > I am trying to debug a simple routine that adds two 32 bit numbers and I > can't figure out why it keeps setting bits in the hi order bytes > inappropriately. If you will take a look at this code and give some > suggestions they would be much appreciated. The two arguments are in eight > locations. Arg. 1 is, from least to most significant, in registers da1_l > da1_m da1_h and da1_x. Arg. 2 is in da2_l da2_m da2_h and da2_x. On exit the > sum is supposed to be in da1_l da1_m da1_h and da1_x. > Here it is: > movf da2_l, W > addwf da1_l, SAME > btfsc CARRY > incf da2_m, SAME > btfsc ZERO > incf da2_h, SAME > btfsc ZERO > incf da2_x, SAME > movf da2_m, W > addwf da1_m, SAME > btfsc CARRY > incf da2_h, SAME > btfsc ZERO > incf da2_x, SAME > movf da2_h, W > addwf da1_h, SAME > btfsc CARRY > incf da2_x, SAME > movf da2_x, W > addwf da1_x, SAME > If you look at this code > Many Thanks > Keith Causey

Thanks Chip and Regulus, that worked. Now another question. How did you know that MPLAB contained that macro? Is it possible to view the source for tant macro? Are there other macros of the same utility in that same location? Thank You Keith Causey {Original Message removed}