Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!zephyr.ens.tek.com!uw-beaver!uw-june!pardo
From: pardo@cs.washington.edu (David Keppel)
Newsgroups: comp.sys.mips
Subject: 8-byte floating-point alignment restrictions
Message-ID: <11441@june.cs.washington.edu>
Date: 13 Apr 90 23:14:37 GMT
Reply-To: pardo@june.cs.washington.edu (David Keppel)
Organization: University of Washington, Computer Science, Seattle
Lines: 21

The R2000 hardware disallows references to 4-byte quantities (e.g.,
ints and floats) that are less than 4-byte aligned.  Some DECstation
3100 routines (e.g., /usr/include/varargs.h) assume 8-byte
floating-point numbers are 8-byte aligned, tho' I can find no
hardware requirement that forces more than 4-byte alignment.

Some questions:
* Why this convention?
* How widespread is it?

And a warning:
* If you get totally bizarre 8-byte floating-point numbers, check
  the stack alignment where *printf routines are expected to print
  floating numbers.

Please e-mail answers to the questions; I will post a brief summary.

	;-D on  ( Brief summary: thin underwear )  Pardo
-- 
		    pardo@cs.washington.edu
    {rutgers,cornell,ucsd,ubc-cs,tektronix}!uw-beaver!june!pardo