This is from early December, sorry about the delay:

> On Wed, 6 Dec 2000, Ignatios Souvatzis wrote:
> 
> > On Wed, Dec 06, 2000 at 07:04:52PM +0000, Steve Woodford wrote:
> > 
> > > NOP=0x4e75
> > 
> > is this something which will actually be executed a lot? In this
> > case the code for lea a5@,a5 will work better, as it won't synchronize
> 
> AFAICT, ld(1) doesn't insert this into the instruction stream of text
> sections, so it's not an issue.

Yes it does insert them, but only between object modules -- as padding.
The NOP opcode should be minimum-length, and IMHO a trap or breakpoint
instruction would be better. Just don't use 0x0000 because that's a two
word instruction on m68k. (ick.)

The performance of the instruction specified by NOP is only an issue if
the program jumps after the last return in a file, or if you are in the
habit of concatenating routines made up of multiple source files and let
alignment padding get in between them...

Todd Whitesel
toddpw @ best.com