[M3commit] CVS Update: cm3

Jay Krell jkrell at elego.de
Tue May 13 09:44:03 CEST 2008

CVSROOT:	/usr/cvs
Changes by:	jkrell at birch.	08/05/13 09:44:03

Modified files:
	cm3/m3-sys/m3middle/src/: Target.i3 Target.m3 

Log message:
	let's consistently use bit sizes/offets/addresses
	and not byte sizes/offsets/addresses
	I don't really know what was intended here, but it
	seems that a mix of byte and bit addresses were used
	bit addresses seem better, they are the more general case,
	they can represent byte addresses, but byte addresses
	cannot represent bit addresses
	however that is not entirely true -- a bit address in
	an INTEGER has an eighth of the range of a byte address
	in an INTEGER; the multiplication/overflow was happening
	anyway, in NilChkExpr
	Look at this way -- the multiplications are optimized out.
	I remember thinking that multiplication by byte.size should
	omitted, since it is obviously 1, but it is not, it is bits
	not bytes
	verifying the correctness of all these numbers would be good
	setting the values lower than "correct" is ok, but less efficient

More information about the M3commit mailing list