Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!cbatt!ucbvax!amdcad!tim
From: tim@amdcad.UUCP
Newsgroups: comp.sys.amiga
Subject: Re: PD Assembler;Label errors!
Message-ID: <15033@amdcad.UUCP>
Date: Mon, 2-Mar-87 11:32:45 EST
Article-I.D.: amdcad.15033
Posted: Mon Mar  2 11:32:45 1987
Date-Received: Tue, 3-Mar-87 20:51:43 EST
References: <859@crash.CTS.COM>
Reply-To: tim@amdcad.UUCP (Tim Olson)
Organization: Advanced Micro Devices, Sunnyvale, California
Lines: 36

In article <859@crash.CTS.COM> billk@pnet01.CTS.COM (Bill Kelly) writes:
>
>LabelOne:       movea.l         #4,a0
>                movea.l         (a0),a0
>                bra LabelTwo
>
>LabelTwo        move.l          a0,_ExecBase    ; colon on label name
>                beq NoExec                      ; purposely omitted
>
>      NoExec:   rts
>
>...this code is the kind of stuff I was writing to try to determine what in my
>LARGE file had caused the errors.   In a file as small as the one above, the
>same problems would occur.  Whenever I define *any* label, I get:
>
>***** 45  Label value changed between passes.

The MC68020 User's Manual states

	" Note: A short branch to the immediately following instruction
	  cannot be generated, because it would result in a zero offset,
	  which forces a word branch instruction definition."

The assembler assumes in pass 1 that the displacement of the branch
label can fit in an 8-bit field, and doesn't know the actual
displacement until the end of pass1 (since it is a forward reference). 
Then in pass2, it discovers that the branch matches the above condition,
and changes the displacement size to 16-bits.  This causes further
labels to now be off by 1 byte from pass1 to pass2, hence the error
message you see.  You will somehow have to force the branch to use the
16-bit displacement mode in pass1 (sorry, I don't know the syntax for
your assembler.)

	-- Tim Olson
	Advanced Micro Devices
	tim@amdcad