Path: utzoo!attcan!uunet!ncrlnk!ncr-sd!hp-sdd!hplabs!decwrl!labrea!rutgers!cs.utexas.edu!oakhill!davet
From: davet@oakhill.UUCP (David Trissel)
Newsgroups: comp.sys.m68k
Subject: addq.w #n,sp and a pop quiz
Keywords: C asm addq
Message-ID: <1737@oakhill.UUCP>
Date: 11 Dec 88 14:16:41 GMT
References: <5005@bsu-cs.UUCP> <5460@cbmvax.UUCP> <1100@ncar.ucar.edu> <670@auspex.UUCP>
Reply-To: davet@oakhill.UUCP (David Trissel)
Distribution: na
Organization: Motorola Inc., Austin Tx.
Lines: 22


When I put the add.w into the compiler I wondered just how many people
would see it and think that it must be wrong. Craig Ruff had it
right in that the add.w requires 2 less bytes (and is faster) than add.l
for cases where the immediate will fit in a word and where addq cannot
be substituted by the assembler. Specifically the compiler generates this
for popping arguments and so there is never any danger of the immediate 
value being too large. 

Instructions depending on esoteric architectural features such as the above
separate the men from the boys. Here's another one:

There is only one place in the instruction set where a data register is
implicitly sign-extended. Where?

I'll post the answer in a few days (if nobody else does.)

  -- Dave Trissel  Motorola Semiconductor, Austin Texas
if anyone knows the answer to this without 
can answer this one without looking 
This type of architectural usage separate the men from the boys. Here's
another question on a subtlety which I doubt any of you would know without