Post by thread:[PIC]: HiTech PICC and Assembler variables

I am trying to implement Dmitry's bit reversal algorithm under HiTech PICC, but am having a problem with the variables. The function I have is: char reverse_bits(char source)// reverses bits 0 to 6 - source = 0ABCDEFG { char result; #asm swapf _source,w ;W= DEFG0ABC btfsc _source,3 ; If D = 1, xorlw 0x88 ;convert now sure W= 0EFGDABC btfsc _source,6 ;Test bit A xorlw 0x05 ;Invert bits A and C btfsc _source,4 ;Test bit C xorlw 0x05 ;Invert bits A and C ;now W = 0EFGDCBA btfsc _source,2 ;Do the same with E and G xorlw 0x50 btfsc _source,0 xorlw 0x50 movwf _result ;so now W = 0GFEDCBA (done) #endasm return(result); } But, on building, the linker says that _source and _result are undefined variables. I thought that, to use a C variable in a piece of assembler code, you simply prepend an underscore to it. Anyone know what I'm doing wrong? Graham -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

I am not an expert on this, but I would tend to believe that since PICC keeps hands-off anything in an asm block, and since you don't use the variables source and result other than in the assembler block, that it just figures it doesn't need those two variables, so they don't exist for the assembler block to use. (W is used for passed values). As a test, perhaps add some do-nothing code that uses those variables within the function, or use static variables. Cheers, Bob {Quote hidden}> char reverse_bits(char source)// reverses bits 0 to 6 - source = 0ABCDEFG > { > char result; > #asm > swapf _source,w ;W= DEFG0ABC > btfsc _source,3 ; If D = 1, > xorlw 0x88 ;convert now sure W= 0EFGDABC > > btfsc _source,6 ;Test bit A > xorlw 0x05 ;Invert bits A and C > btfsc _source,4 ;Test bit C > xorlw 0x05 ;Invert bits A and C > ;now W = 0EFGDCBA > btfsc _source,2 ;Do the same with E and G > xorlw 0x50 > btfsc _source,0 > xorlw 0x50 > movwf _result ;so now W = 0GFEDCBA (done) > #endasm > return(result); > } > > But, on building, the linker says that _source and _result are undefined variables. I thought that, to use a C variable in a piece of assembler code, you simply prepend an underscore to it. Anyone know what I'm doing wrong? > -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

On Fri, 30 Mar 2001 16:55:07 +0000, you wrote: {Quote hidden}>I am not an expert on this, but I would tend to believe that since PICC >keeps hands-off anything in an asm block, and since you don't use the >variables source and result other than in the assembler block, that it >just figures it doesn't need those two variables, so they don't exist for >the assembler block to use. (W is used for passed values). > >As a test, perhaps add some do-nothing code that uses those variables >within the function, or use static variables. > >Cheers, > >Bob > >> char reverse_bits(char source)// reverses bits 0 to 6 - source = 0ABCDEFG >> { >> char result; >> #asm >> swapf _source,w ;W= DEFG0ABC >> btfsc _source,3 ; If D = 1, >> xorlw 0x88 ;convert now sure W= 0EFGDABC >> >> btfsc _source,6 ;Test bit A >> xorlw 0x05 ;Invert bits A and C >> btfsc _source,4 ;Test bit C >> xorlw 0x05 ;Invert bits A and C >> ;now W = 0EFGDCBA >> btfsc _source,2 ;Do the same with E and G >> xorlw 0x50 >> btfsc _source,0 >> xorlw 0x50 >> movwf _result ;so now W = 0GFEDCBA (done) >> #endasm >> return(result); >> } >> >> But, on building, the linker says that _source and _result are undefined variables. I thought that, to use a C variable in a piece of assembler code, you simply prepend an underscore to it. Anyone know what I'm doing wrong? >> I found that Hi-Tech had a nasty habit of optimising out vars that were only used inside ASM blocks. I can't remember If I found a solution to this situation for return vars. I think doing something like result=result; outside the #asm can fix it, but can also generate unnecessary code. The other problems is that with full optimisation on. it will 'helpfully' optimise out all your assembler code! I found it to be much less hassle to put asm stuff in a seperate .as file. Under MPLAB this is relatively painless to do. -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

> From: Mike Harrison [spam_OUTmikeTakeThisOuTwhitewing.co.uk] ... {Quote hidden}>>> char reverse_bits(char source)// reverses bits 0 to 6 - >>> source = 0ABCDEFG >>> { >>> char result; >>> #asm >>> swapf _source,w ;W= DEFG0ABC >>> btfsc _source,3 ; If D = 1, >>> xorlw 0x88 ;convert now sure W= 0EFGDABC >>> >>> btfsc _source,6 ;Test bit A >>> xorlw 0x05 ;Invert bits A and C >>> btfsc _source,4 ;Test bit C >>> xorlw 0x05 ;Invert bits A and C >>> ;now W = 0EFGDCBA >>> btfsc _source,2 ;Do the same with E and G >>> xorlw 0x50 >>> btfsc _source,0 >>> xorlw 0x50 >>> movwf _result ;so now W = 0GFEDCBA (done) >>> #endasm >>> return(result); >>> } >>> >>> But, on building, the linker says that _source and _result >>> are undefined variables. I thought that, to use a C variable >>> in a piece of assembler code, you simply prepend an >>> underscore to it. Anyone know what I'm doing wrong? Global variables usually use underscores local variables are of the format ?a_procedurename + offset > I found that Hi-Tech had a nasty habit of optimising out vars that > were only used inside ASM blocks. > I can't remember If I found a solution to this situation for return > vars. > I think doing something like > result=result; > outside the #asm can fix it, but can also generate unnecessary code. In this case, since result is returned, it didn't get optimized out. source was the one that was lost, so you would have to add a line like result = source; However, you can also cheat a bit. Result is only used to return the end result. There's no reason that source can't be used for that. This way no extra code is added. In assembly language this kind of cheating could effect some variable outside of the function, but in C, source is just the value of the variable in the calling function, it stops being a meaningful reference once this function exits anyway, so it's okay to stomp on. The code at the bottom is what I came up with. If you really want to use result, you can start the function with the line: source = (source << 4) | (source >> 4) that will generate: swapf ?a_reverse_bits start the asm with movf ?a_reverse_bits, w change all the bit tests to handle the swap, and end the code with movwf ?a_reverse_bits+1 /* Converts 0ABCDEFG to 0GFEDCBA */ char reverse_bits(char source) { #asm ;source is ?a_reverse_bits swapf ?a_reverse_bits, w ;W = DEFG0ABC btfsc ?a_reverse_bits, 3 ;If D = 1, xorlw 0x88 ;Now we're sure W = 0EFGDABC btfsc ?a_reverse_bits, 6 ;Test bit A xorlw 0x05 ;Invert bits A and C btfsc ?a_reverse_bits, 4 ;Test bit C xorlw 0x05 ;Invert bits A and C ;now W = 0EFGDCBA btfsc ?a_reverse_bits, 2 ;Do the same with E and G xorlw 0x50 btfsc ?a_reverse_bits, 0 xorlw 0x50 movwf ?a_reverse_bits ;so now W = 0GFEDCBA, copy to source #endasm return(source); } -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.