Re: wanted: good references for [45]86 code generation

dgay@barnowl.CS.Berkeley.EDU (David Gay)
1 Feb 1998 14:29:15 -0500

          From comp.compilers

Related articles
wanted: good references for [45]86 code generation kers@cdollin.hpl.hp.com (Chris Dollin) (1998-01-21)
Re: wanted: good references for [45]86 code generation sergey@solyanik.com (Sergey Solyanik) (1998-01-23)
Re: wanted: good references for [45]86 code generation ast@halcyon.com (1998-01-24)
Re: wanted: good references for [45]86 code generation sergey@solyanik.com (Sergey Solyanik) (1998-01-30)
Re: wanted: good references for [45]86 code generation dgay@barnowl.CS.Berkeley.EDU (1998-02-01)
Re: wanted: good references for [45]86 code generation mike.hobbs@ccmail.fingerhut.com (Michael Hobbs) (1998-02-01)
Re: wanted: good references for [45]86 code generation jacob@jacob.remcomp.fr (1998-02-01)
Re: wanted: good references for [45]86 code generation mzraly@world.std.com (1998-02-03)
Re: wanted: good references for [45]86 code generation bs@hplb.hpl.hp.com (1998-02-07)
| List of all articles for this month |

From: dgay@barnowl.CS.Berkeley.EDU (David Gay)
Newsgroups: comp.compilers
Date: 1 Feb 1998 14:29:15 -0500
Organization: University of California, Berkeley
References: 98-01-083 98-01-094 98-01-101 98-01-117
Keywords: 386, code

Andrew Tucker <ast@halcyon.com> wrote ...
      > Fraser and Hanson's lcc book describes their code generator for x86
      > as well as MIPS and SPARC.


"Sergey Solyanik" <sergey@solyanik.com> writes:
      Well... I guess (hope) lcc-style code generator was not what author of
      original article had in mind. Their code generator is very-very basic.


      F&H book does not even describe x86-specific code generation for per se.
      Rather, it provides scripts for lburg - an automated generator generator.


This is something of a mischaracterisation. The F&H book is the
'webbed' source of the lcc compiler, i.e. the documentation and source
code are written as a whole, and appropriate generate the book and
compilable source code. The section on x86 code generation thus
contains the lburg specification, interspersed with comments
explaining and justifying its design.


Given that the emphasis of lcc is not efficient code, I doubt the x86
code is spectacularly fast, but I suspect this is more due to the
general lack of optimisation in lcc and its simple register
allocation, rather than the actual x86 code generator.
--
David Gay - Yet Another Starving Grad Student
dgay@cs.berkeley.edu
--


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.