File:  [gforth] / gforth / doc / vmgen.texi
Revision 1.25: download - view: text, annotated - select for diffs
Sun Mar 9 15:17:02 2003 UTC (20 years, 6 months ago) by anton
Branches: MAIN
CVS tags: v0-6-1, v0-6-0, HEAD
updated copyright years

\input texinfo    @c -*-texinfo-*-
@comment %**start of header
@include version.texi
@settitle Vmgen (Gforth @value{VERSION})
@c @syncodeindex pg cp
@comment %**end of header
This manual is for Vmgen
(version @value{VERSION}, @value{UPDATED}),
the virtual machine interpreter generator

Copyright @copyright{} 2002, 03,2003 Free Software Foundation, Inc.

Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.1 or
any later version published by the Free Software Foundation; with no
Invariant Sections, with the Front-Cover texts being ``A GNU Manual,''
and with the Back-Cover Texts as in (a) below.  A copy of the
license is included in the section entitled ``GNU Free Documentation

(a) The FSF's Back-Cover Text is: ``You have freedom to copy and modify
this GNU Manual, like GNU software.  Copies published by the Free
Software Foundation raise funds for GNU development.''
@end quotation
@end copying

@dircategory Software development
* Vmgen: (vmgen).               Virtual machine interpreter generator
@end direntry

@title Vmgen
@subtitle for Gforth version @value{VERSION}, @value{UPDATED}
@author M. Anton Ertl (@email{})
@vskip 0pt plus 1filll
@end titlepage


@node Top, Introduction, (dir), (dir)
@top Vmgen

@end ifnottex

* Introduction::                What can Vmgen do for you?
* Why interpreters?::           Advantages and disadvantages
* Concepts::                    VM interpreter background
* Invoking Vmgen::              
* Example::                     
* Input File Format::           
* Error messages::              reported by Vmgen
* Using the generated code::    
* Hints::                       VM archictecture, efficiency
* The future::                  
* Changes::                     from earlier versions
* Contact::                     Bug reporting etc.
* Copying This Manual::         Manual License
* Index::                       

 --- The Detailed Node Listing ---


* Front end and VM interpreter::  Modularizing an interpretive system
* Data handling::               Stacks, registers, immediate arguments
* Dispatch::                    From one VM instruction to the next


* Example overview::            
* Using profiling to create superinstructions::  

Input File Format

* Input File Grammar::          
* Simple instructions::         
* Superinstructions::           
* Store Optimization::          
* Register Machines::           How to define register VM instructions

Input File Grammar

* Eval escapes::                what follows \E

Simple instructions

* C Code Macros::               Macros recognized by Vmgen
* C Code restrictions::         Vmgen makes assumptions about C code
* Stack growth direction::      is configurable per stack

Using the generated code

* VM engine::                   Executing VM code
* VM instruction table::        
* VM code generation::          Creating VM code (in the front-end)
* Peephole optimization::       Creating VM superinstructions
* VM disassembler::             for debugging the front end
* VM profiler::                 for finding worthwhile superinstructions


* Floating point::              and stacks

Copying This Manual

* GNU Free Documentation License::  License for copying this manual.

@end detailmenu
@end menu

@c @ifnottex
@c This file documents Vmgen (Gforth @value{VERSION}).

@c ************************************************************
@node Introduction, Why interpreters?, Top, Top
@chapter Introduction

Vmgen is a tool for writing efficient interpreters.  It takes a simple
virtual machine description and generates efficient C code for dealing
with the virtual machine code in various ways (in particular, executing
it).  The run-time efficiency of the resulting interpreters is usually
within a factor of 10 of machine code produced by an optimizing

The interpreter design strategy supported by Vmgen is to divide the
interpreter into two parts:

@itemize @bullet

@item The @emph{front end} takes the source code of the language to be
implemented, and translates it into virtual machine code.  This is
similar to an ordinary compiler front end; typically an interpreter
front-end performs no optimization, so it is relatively simple to
implement and runs fast.

@item The @emph{virtual machine interpreter} executes the virtual
machine code.

@end itemize

Such a division is usually used in interpreters, for modularity as well
as for efficiency.  The virtual machine code is typically passed between
front end and virtual machine interpreter in memory, like in a
load-and-go compiler; this avoids the complexity and time cost of
writing the code to a file and reading it again.

A @emph{virtual machine} (VM) represents the program as a sequence of
@emph{VM instructions}, following each other in memory, similar to real
machine code.  Control flow occurs through VM branch instructions, like
in a real machine.

@cindex functionality features overview
In this setup, Vmgen can generate most of the code dealing with virtual
machine instructions from a simple description of the virtual machine
instructions (@pxref{Input File Format}), in particular:

@table @strong

@item VM instruction execution

@item VM code generation
Useful in the front end.

@item VM code decompiler
Useful for debugging the front end.

@item VM code tracing
Useful for debugging the front end and the VM interpreter.  You will
typically provide other means for debugging the user's programs at the
source level.

@item VM code profiling
Useful for optimizing the VM interpreter with superinstructions
(@pxref{VM profiler}).

@end table

To create parts of the interpretive system that do not deal with VM
instructions, you have to use other tools (e.g., @command{bison}) and/or
hand-code them.

@cindex efficiency features overview
Vmgen supports efficient interpreters though various optimizations, in

@itemize @bullet

@item Threaded code

@item Caching the top-of-stack in a register

@item Combining VM instructions into superinstructions

Replicating VM (super)instructions for better BTB prediction accuracy
(not yet in vmgen-ex, but already in Gforth).

@end itemize

@cindex speed for JVM
As a result, Vmgen-based interpreters are only about an order of
magnitude slower than native code from an optimizing C compiler on small
benchmarks; on large benchmarks, which spend more time in the run-time
system, the slowdown is often less (e.g., the slowdown of a
Vmgen-generated JVM interpreter over the best JVM JIT compiler we
measured is only a factor of 2-3 for large benchmarks; some other JITs
and all other interpreters we looked at were slower than our

VMs are usually designed as stack machines (passing data between VM
instructions on a stack), and Vmgen supports such designs especially
well; however, you can also use Vmgen for implementing a register VM
(@pxref{Register Machines}) and still benefit from most of the advantages
offered by Vmgen.

There are many potential uses of the instruction descriptions that are
not implemented at the moment, but we are open for feature requests, and
we will consider new features if someone asks for them; so the feature
list above is not exhaustive.

@c *********************************************************************
@node Why interpreters?, Concepts, Introduction, Top
@chapter Why interpreters?
@cindex interpreters, advantages
@cindex advantages of interpreters
@cindex advantages of vmgen

Interpreters are a popular language implementation technique because
they combine all three of the following advantages:

@itemize @bullet

@item Ease of implementation

@item Portability

@item Fast edit-compile-run cycle

@end itemize

Vmgen makes it even easier to implement interpreters.

@cindex speed of interpreters
The main disadvantage of interpreters is their run-time speed.  However,
there are huge differences between different interpreters in this area:
the slowdown over optimized C code on programs consisting of simple
operations is typically a factor of 10 for the more efficient
interpreters, and a factor of 1000 for the less efficient ones (the
slowdown for programs executing complex operations is less, because the
time spent in libraries for executing complex operations is the same in
all implementation strategies).

Vmgen supports techniques for building efficient interpreters.

@c ********************************************************************
@node Concepts, Invoking Vmgen, Why interpreters?, Top
@chapter Concepts

* Front end and VM interpreter::  Modularizing an interpretive system
* Data handling::               Stacks, registers, immediate arguments
* Dispatch::                    From one VM instruction to the next
@end menu

@c --------------------------------------------------------------------
@node Front end and VM interpreter, Data handling, Concepts, Concepts
@section Front end and VM interpreter
@cindex modularization of interpreters

@cindex front-end
Interpretive systems are typically divided into a @emph{front end} that
parses the input language and produces an intermediate representation
for the program, and an interpreter that executes the intermediate
representation of the program.

@cindex virtual machine
@cindex VM
@cindex VM instruction
@cindex instruction, VM
@cindex VM branch instruction
@cindex branch instruction, VM
@cindex VM register
@cindex register, VM
@cindex opcode, VM instruction
@cindex immediate argument, VM instruction
For efficient interpreters the intermediate representation of choice is
virtual machine code (rather than, e.g., an abstract syntax tree).
@emph{Virtual machine} (VM) code consists of VM instructions arranged
sequentially in memory; they are executed in sequence by the VM
interpreter, but VM branch instructions can change the control flow and
are used for implementing control structures.  The conceptual similarity
to real machine code results in the name @emph{virtual machine}.
Various terms similar to terms for real machines are used; e.g., there
are @emph{VM registers} (like the instruction pointer and stack
pointer(s)), and the VM instruction consists of an @emph{opcode} and
@emph{immediate arguments}.

In this framework, Vmgen supports building the VM interpreter and any
other component dealing with VM instructions.  It does not have any
support for the front end, apart from VM code generation support.  The
front end can be implemented with classical compiler front-end
techniques, supported by tools like @command{flex} and @command{bison}.

The intermediate representation is usually just internal to the
interpreter, but some systems also support saving it to a file, either
as an image file, or in a full-blown linkable file format (e.g., JVM).
Vmgen currently has no special support for such features, but the
information in the instruction descriptions can be helpful, and we are
open to feature requests and suggestions.

@c --------------------------------------------------------------------
@node Data handling, Dispatch, Front end and VM interpreter, Concepts
@section Data handling

@cindex stack machine
@cindex register machine
Most VMs use one or more stacks for passing temporary data between VM
instructions.  Another option is to use a register machine architecture
for the virtual machine; we believe that using a stack architecture is
usually both simpler and faster.

however, this option is slower or
significantly more complex to implement than a stack machine architecture.

Vmgen has special support and optimizations for stack VMs, making their
implementation easy and efficient.

You can also implement a register VM with Vmgen (@pxref{Register
Machines}), and you will still profit from most Vmgen features.

@cindex stack item size
@cindex size, stack items
Stack items all have the same size, so they typically will be as wide as
an integer, pointer, or floating-point value.  Vmgen supports treating
two consecutive stack items as a single value, but anything larger is
best kept in some other memory area (e.g., the heap), with pointers to
the data on the stack.

@cindex instruction stream
@cindex immediate arguments
Another source of data is immediate arguments VM instructions (in the VM
instruction stream).  The VM instruction stream is handled similar to a
stack in Vmgen.

@cindex garbage collection
@cindex reference counting
Vmgen has no built-in support for, nor restrictions against
@emph{garbage collection}.  If you need garbage collection, you need to
provide it in your run-time libraries.  Using @emph{reference counting}
is probably harder, but might be possible (contact us if you are
@c reference counting might be possible by including counting code in 
@c the conversion macros.

@c --------------------------------------------------------------------
@node Dispatch,  , Data handling, Concepts
@section Dispatch
@cindex Dispatch of VM instructions
@cindex main interpreter loop

Understanding this section is probably not necessary for using Vmgen,
but it may help.  You may want to skip it now, and read it if you find statements about dispatch methods confusing.

After executing one VM instruction, the VM interpreter has to dispatch
the next VM instruction (Vmgen calls the dispatch routine @samp{NEXT}).
Vmgen supports two methods of dispatch:

@table @strong

@item switch dispatch
@cindex switch dispatch
In this method the VM interpreter contains a giant @code{switch}
statement, with one @code{case} for each VM instruction.  The VM
instruction opcodes are represented by integers (e.g., produced by an
@code{enum}) in the VM code, and dispatch occurs by loading the next
opcode, @code{switch}ing on it, and continuing at the appropriate
@code{case}; after executing the VM instruction, the VM interpreter
jumps back to the dispatch code.

@item threaded code
@cindex threaded code
This method represents a VM instruction opcode by the address of the
start of the machine code fragment for executing the VM instruction.
Dispatch consists of loading this address, jumping to it, and
incrementing the VM instruction pointer.  Typically the threaded-code
dispatch code is appended directly to the code for executing the VM
instruction.  Threaded code cannot be implemented in ANSI C, but it can
be implemented using GNU C's labels-as-values extension (@pxref{Labels
as Values, , Labels as Values,, GNU C Manual}).

@c call threading
@end table

Threaded code can be twice as fast as switch dispatch, depending on the
interpreter, the benchmark, and the machine.

@c *************************************************************
@node Invoking Vmgen, Example, Concepts, Top
@chapter Invoking Vmgen
@cindex Invoking Vmgen

The usual way to invoke Vmgen is as follows:

vmgen @var{inputfile}
@end example

Here @var{inputfile} is the VM instruction description file, which
usually ends in @file{.vmg}.  The output filenames are made by taking
the basename of @file{inputfile} (i.e., the output files will be created
in the current working directory) and replacing @file{.vmg} with
@file{-vm.i}, @file{-disasm.i}, @file{-gen.i}, @file{-labels.i},
@file{-profile.i}, and @file{-peephole.i}.  E.g., @command{vmgen
hack/foo.vmg} will create @file{foo-vm.i}, @file{foo-disasm.i},
@file{foo-gen.i}, @file{foo-labels.i}, @file{foo-profile.i} and

The command-line options supported by Vmgen are

@table @option

@cindex -h, command-line option
@cindex --help, command-line option
@item --help
@itemx -h
Print a message about the command-line options

@cindex -v, command-line option
@cindex --version, command-line option
@item --version
@itemx -v
Print version and exit
@end table


@c ****************************************************************
@node Example, Input File Format, Invoking Vmgen, Top
@chapter Example
@cindex example of a Vmgen-based interpreter

* Example overview::            
* Using profiling to create superinstructions::  
@end menu

@c --------------------------------------------------------------------
@node Example overview, Using profiling to create superinstructions, Example, Example
@section Example overview
@cindex example overview
@cindex @file{vmgen-ex}
@cindex @file{vmgen-ex2}

There are two versions of the same example for using Vmgen:
@file{vmgen-ex} and @file{vmgen-ex2} (you can also see Gforth as
example, but it uses additional (undocumented) features, and also
differs in some other respects).  The example implements @emph{mini}, a
tiny Modula-2-like language with a small JavaVM-like virtual machine.

The difference between the examples is that @file{vmgen-ex} uses many
casts, and @file{vmgen-ex2} tries to avoids most casts and uses unions
instead.  In the rest of this manual we usually mention just files in
@file{vmgen-ex}; if you want to use unions, use the equivalent file in
@cindex unions example
@cindex casts example

The files provided with each example are:
@cindex example files

disasm.c           wrapper file
engine.c           wrapper file
peephole.c         wrapper file
profile.c          wrapper file
mini-inst.vmg      simple VM instructions
mini-super.vmg     superinstructions (empty at first)
mini.h             common declarations
mini.l             scanner
mini.y             front end (parser, VM code generator)
support.c          main() and other support functions           example mini program        example mini program          example mini program (tests everything)
test.out  output
stat.awk           script for aggregating profile information
peephole-blacklist list of instructions not allowed in superinstructions
seq2rule.awk       script for creating superinstructions
@end example

For your own interpreter, you would typically copy the following files
and change little, if anything:
@cindex wrapper files

disasm.c           wrapper file
engine.c           wrapper file
peephole.c         wrapper file
profile.c          wrapper file
stat.awk           script for aggregating profile information
seq2rule.awk       script for creating superinstructions
@end example

You would typically change much in or replace the following files:

mini-inst.vmg      simple VM instructions
mini.h             common declarations
mini.l             scanner
mini.y             front end (parser, VM code generator)
support.c          main() and other support functions
peephole-blacklist list of instructions not allowed in superinstructions
@end example

You can build the example by @code{cd}ing into the example's directory,
and then typing @code{make}; you can check that it works with @code{make
check}.  You can run run mini programs like this:

@end example

To learn about the options, type @code{./mini -h}.

@c --------------------------------------------------------------------
@node Using profiling to create superinstructions,  , Example overview, Example
@section Using profiling to create superinstructions
@cindex profiling example
@cindex superinstructions example

I have not added rules for this in the @file{Makefile} (there are many
options for selecting superinstructions, and I did not want to hardcode
one into the @file{Makefile}), but there are some supporting scripts, and
here's an example:

Suppose you want to use @file{} and @file{} as training
programs, you get the profiles like this:

make #takes a few seconds
@end example

You can aggregate these profiles with @file{stat.awk}:

awk -f stat.awk
@end example

The result contains lines like:

      2      16        36910041 loadlocal lit
@end example

This means that the sequence @code{loadlocal lit} statically occurs a
total of 16 times in 2 profiles, with a dynamic execution count of

The numbers can be used in various ways to select superinstructions.
E.g., if you just want to select all sequences with a dynamic
execution count exceeding 10000, you would use the following pipeline:

awk -f stat.awk|
awk '$3>=10000'|                #select sequences
fgrep -v -f peephole-blacklist| #eliminate wrong instructions
awk -f seq2rule.awk|  #transform sequences into superinstruction rules
sort -k 3 >mini-super.vmg       #sort sequences
@end example

The file @file{peephole-blacklist} contains all instructions that
directly access a stack or stack pointer (for mini: @code{call},
@code{return}); the sort step is necessary to ensure that prefixes
precede larger superinstructions.

Now you can create a version of mini with superinstructions by just
saying @samp{make}

@c ***************************************************************
@node Input File Format, Error messages, Example, Top
@chapter Input File Format
@cindex input file format
@cindex format, input file

Vmgen takes as input a file containing specifications of virtual machine
instructions.  This file usually has a name ending in @file{.vmg}.

Most examples are taken from the example in @file{vmgen-ex}.

* Input File Grammar::          
* Simple instructions::         
* Superinstructions::           
* Store Optimization::          
* Register Machines::           How to define register VM instructions
@end menu

@c --------------------------------------------------------------------
@node Input File Grammar, Simple instructions, Input File Format, Input File Format
@section Input File Grammar
@cindex grammar, input file
@cindex input file grammar

The grammar is in EBNF format, with @code{@var{a}|@var{b}} meaning
``@var{a} or @var{b}'', @code{@{@var{c}@}} meaning 0 or more repetitions
of @var{c} and @code{[@var{d}]} meaning 0 or 1 repetitions of @var{d}.

@cindex free-format, not
@cindex newlines, significance in syntax
Vmgen input is not free-format, so you have to take care where you put
newlines (and, in a few cases, white space).

description: @{instruction|comment|eval-escape|c-escape@}

instruction: simple-inst|superinst

simple-inst: ident '(' stack-effect ')' newline c-code newline newline

stack-effect: @{ident@} '--' @{ident@}

super-inst: ident '=' ident @{ident@}  

comment:      '\ '  text newline

eval-escape:  '\E ' text newline

c-escape:     '\C ' text newline
@end example
@c \+ \- \g \f \c

Note that the @code{\}s in this grammar are meant literally, not as
C-style encodings for non-printable characters.

There are two ways to delimit the C code in @code{simple-inst}:

@itemize @bullet

If you start it with a @samp{@{} at the start of a line (i.e., not even
white space before it), you have to end it with a @samp{@}} at the start
of a line (followed by a newline).  In this case you may have empty
lines within the C code (typically used between variable definitions and

You do not start it with @samp{@{}.  Then the C code ends at the first
empty line, so you cannot have empty lines within this code.

@end itemize

The text in @code{comment}, @code{eval-escape} and @code{c-escape} must
not contain a newline.  @code{Ident} must conform to the usual
conventions of C identifiers (otherwise the C compiler would choke on
the Vmgen output), except that idents in @code{stack-effect} may have a
stack prefix (for stack prefix syntax, @pxref{Eval escapes}).

@cindex C escape
@cindex @code{\C}
@cindex conditional compilation of Vmgen output
The @code{c-escape} passes the text through to each output file (without
the @samp{\C}).  This is useful mainly for conditional compilation
(i.e., you write @samp{\C #if ...} etc.).

@cindex sync lines
@cindex @code{#line}
In addition to the syntax given in the grammer, Vmgen also processes
sync lines (lines starting with @samp{#line}), as produced by @samp{m4
-s} (@pxref{Invoking m4, , Invoking m4,, GNU m4}) and similar
tools.  This allows associating C compiler error messages with the
original source of the C code.

Vmgen understands a few extensions beyond the grammar given here, but
these extensions are only useful for building Gforth.  You can find a
description of the format used for Gforth in @file{prim}.

* Eval escapes::                what follows \E
@end menu

@node Eval escapes,  , Input File Grammar, Input File Grammar
@subsection Eval escapes
@cindex escape to Forth
@cindex eval escape
@cindex @code{\E}

@c woanders?
The text in @code{eval-escape} is Forth code that is evaluated when
Vmgen reads the line.  You will normally use this feature to define
stacks and types.

If you do not know (and do not want to learn) Forth, you can build the
text according to the following grammar; these rules are normally all
Forth you need for using Vmgen:

text: stack-decl|type-prefix-decl|stack-prefix-decl|set-flag

stack-decl: 'stack ' ident ident ident
    's" ' string '" ' ('single'|'double') ident 'type-prefix' ident
stack-prefix-decl:  ident 'stack-prefix' string
set-flag: ('store-optimization'|'include-skipped-insts') ('on'|'off')
@end example

Note that the syntax of this code is not checked thoroughly (there are
many other Forth program fragments that could be written in an

A stack prefix can contain letters, digits, or @samp{:}, and may start
with an @samp{#}; e.g., in Gforth the return stack has the stack prefix
@samp{R:}.  This restriction is not checked during the stack prefix
definition, but it is enforced by the parsing rules for stack items

If you know Forth, the stack effects of the non-standard words involved
@findex stack
@findex type-prefix
@findex single
@findex double
@findex stack-prefix
@findex store-optimization
stack                 ( "name" "pointer" "type" -- )
                      ( name execution: -- stack )
type-prefix           ( addr u item-size stack "prefix" -- )
single                ( -- item-size )
double                ( -- item-size )
stack-prefix          ( stack "prefix" -- )
store-optimization    ( -- addr )
include-skipped-insts ( -- addr )
@end example

An @var{item-size} takes three cells on the stack.

@c --------------------------------------------------------------------
@node Simple instructions, Superinstructions, Input File Grammar, Input File Format
@section Simple instructions
@cindex simple VM instruction
@cindex instruction, simple VM

We will use the following simple VM instruction description as example:

sub ( i1 i2 -- i )
i = i1-i2;
@end example

The first line specifies the name of the VM instruction (@code{sub}) and
its stack effect (@code{i1 i2 -- i}).  The rest of the description is
just plain C code.

@cindex stack effect
@cindex effect, stack
The stack effect specifies that @code{sub} pulls two integers from the
data stack and puts them in the C variables @code{i1} and @code{i2}
(with the rightmost item (@code{i2}) taken from the top of stack;
intuition: if you push @code{i1}, then @code{i2} on the stack, the
resulting stack picture is @code{i1 i2}) and later pushes one integer
(@code{i}) on the data stack (the rightmost item is on the top

@cindex prefix, type
@cindex type prefix
@cindex default stack of a type prefix
How do we know the type and stack of the stack items?  Vmgen uses
prefixes, similar to Fortran; in contrast to Fortran, you have to
define the prefix first:

\E s" Cell"   single data-stack type-prefix i
@end example

This defines the prefix @code{i} to refer to the type @code{Cell}
(defined as @code{long} in @file{mini.h}) and, by default, to the
@code{data-stack}.  It also specifies that this type takes one stack
item (@code{single}).  The type prefix is part of the variable name.

@cindex stack definition
@cindex defining a stack
Before we can use @code{data-stack} in this way, we have to define it:

\E stack data-stack sp Cell
@end example
@c !! use something other than Cell

@cindex stack basic type
@cindex basic type of a stack
@cindex type of a stack, basic
This line defines the stack @code{data-stack}, which uses the stack
pointer @code{sp}, and each item has the basic type @code{Cell}; other
types have to fit into one or two @code{Cell}s (depending on whether the
type is @code{single} or @code{double} wide), and are cast from and to
Cells on accessing the @code{data-stack} with type cast macros
(@pxref{VM engine}).  By default, stacks grow towards lower addresses in
Vmgen-erated interpreters (@pxref{Stack growth direction}).

@cindex stack prefix
@cindex prefix, stack
We can override the default stack of a stack item by using a stack
prefix.  E.g., consider the following instruction:

lit ( #i -- i )
@end example

The VM instruction @code{lit} takes the item @code{i} from the
instruction stream (indicated by the prefix @code{#}), and pushes it on
the (default) data stack.  The stack prefix is not part of the variable
name.  Stack prefixes are defined like this:

\E inst-stream stack-prefix #
@end example

This definition defines that the stack prefix @code{#} specifies the
``stack'' @code{inst-stream}.  Since the instruction stream behaves a
little differently than an ordinary stack, it is predefined, and you do
not need to define it.

@cindex instruction stream
The instruction stream contains instructions and their immediate
arguments, so specifying that an argument comes from the instruction
stream indicates an immediate argument.  Of course, instruction stream
arguments can only appear to the left of @code{--} in the stack effect.
If there are multiple instruction stream arguments, the leftmost is the
first one (just as the intuition suggests).

* C Code Macros::               Macros recognized by Vmgen
* C Code restrictions::         Vmgen makes assumptions about C code
* Stack growth direction::      is configurable per stack
@end menu

@c --------------------------------------------------------------------
@node C Code Macros, C Code restrictions, Simple instructions, Simple instructions
@subsection C Code Macros
@cindex macros recognized by Vmgen
@cindex basic block, VM level

Vmgen recognizes the following strings in the C code part of simple

@table @code

@item SET_IP
@findex SET_IP
As far as Vmgen is concerned, a VM instruction containing this ends a VM
basic block (used in profiling to delimit profiled sequences).  On the C
level, this also sets the instruction pointer.

@findex SUPER_END
This ends a basic block (for profiling), even if the instruction
contains no @code{SET_IP}.

@item INST_TAIL;
@findex INST_TAIL;
Vmgen replaces @samp{INST_TAIL;} with code for ending a VM instruction and
dispatching the next VM instruction.  Even without a @samp{INST_TAIL;} this
happens automatically when control reaches the end of the C code.  If
you want to have this in the middle of the C code, you need to use
@samp{INST_TAIL;}.  A typical example is a conditional VM branch:

if (branch_condition) @{
  SET_IP(target); INST_TAIL;
/* implicit tail follows here */
@end example

In this example, @samp{INST_TAIL;} is not strictly necessary, because there
is another one implicitly after the if-statement, but using it improves
branch prediction accuracy slightly and allows other optimizations.

This indicates that the implicit tail at the end of the VM instruction
dispatches the sequentially next VM instruction even if there is a
@code{SET_IP} in the VM instruction.  This enables an optimization that
is not yet implemented in the vmgen-ex code (but in Gforth).  The
typical application is in conditional VM branches:

if (branch_condition) @{
  SET_IP(target); INST_TAIL; /* now this INST_TAIL is necessary */
@end example

@end table

Note that Vmgen is not smart about C-level tokenization, comments,
strings, or conditional compilation, so it will interpret even a
commented-out SUPER_END as ending a basic block (or, e.g.,
@samp{RESET_IP;} as @samp{SET_IP;}).  Conversely, Vmgen requires the literal
presence of these strings; Vmgen will not see them if they are hiding in
a C preprocessor macro.

@c --------------------------------------------------------------------
@node C Code restrictions, Stack growth direction, C Code Macros, Simple instructions
@subsection C Code restrictions
@cindex C code restrictions
@cindex restrictions on C code
@cindex assumptions about C code

@cindex accessing stack (pointer)
@cindex stack pointer, access
@cindex instruction pointer, access
Vmgen generates code and performs some optimizations under the
assumption that the user-supplied C code does not access the stack
pointers or stack items, and that accesses to the instruction pointer
only occur through special macros.  In general you should heed these
restrictions.  However, if you need to break these restrictions, read
the following.

Accessing a stack or stack pointer directly can be a problem for several
@cindex stack caching, restriction on C code
@cindex superinstructions, restrictions on components

@itemize @bullet

Vmgen optionally supports caching the top-of-stack item in a local
variable (that is allocated to a register).  This is the most frequent
source of trouble.  You can deal with it either by not using
top-of-stack caching (slowdown factor 1-1.4, depending on machine), or
by inserting flushing code (e.g., @samp{IF_spTOS(sp[...] = spTOS);}) at
the start and reloading code (e.g., @samp{IF_spTOS(spTOS = sp[0])}) at
the end of problematic C code.  Vmgen inserts a stack pointer update
before the start of the user-supplied C code, so the flushing code has
to use an index that corrects for that.  In the future, this flushing
may be done automatically by mentioning a special string in the C code.
@c sometimes flushing and/or reloading unnecessary

The Vmgen-erated code loads the stack items from stack-pointer-indexed
memory into variables before the user-supplied C code, and stores them
from variables to stack-pointer-indexed memory afterwards.  If you do
any writes to the stack through its stack pointer in your C code, it
will not affect the variables, and your write may be overwritten by the
stores after the C code.  Similarly, a read from a stack using a stack
pointer will not reflect computations of stack items in the same VM

Superinstructions keep stack items in variables across the whole
superinstruction.  So you should not include VM instructions, that
access a stack or stack pointer, as components of superinstructions
(@pxref{VM profiler}).

@end itemize

You should access the instruction pointer only through its special
macros (@samp{IP}, @samp{SET_IP}, @samp{IPTOS}); this ensure that these
macros can be implemented in several ways for best performance.
@samp{IP} points to the next instruction, and @samp{IPTOS} is its

@c --------------------------------------------------------------------
@node Stack growth direction,  , C Code restrictions, Simple instructions
@subsection Stack growth direction
@cindex stack growth direction

@cindex @code{stack-access-transform}
By default, the stacks grow towards lower addresses.  You can change
this for a stack by setting the @code{stack-access-transform} field of
the stack to an xt @code{( itemnum -- index )} that performs the
appropriate index transformation.

E.g., if you want to let @code{data-stack} grow towards higher
addresses, with the stack pointer always pointing just beyond the
top-of-stack, use this right after defining @code{data-stack}:

\E : sp-access-transform ( itemnum -- index ) negate 1- ;
\E ' sp-access-transform ' data-stack >body stack-access-transform !
@end example

This means that @code{sp-access-transform} will be used to generate
indexes for accessing @code{data-stack}.  The definition of
@code{sp-access-transform} above transforms n into -n-1, e.g, 1 into -2.
This will access the 0th data-stack element (top-of-stack) at sp[-1],
the 1st at sp[-2], etc., which is the typical way upward-growing
stacks are used.  If you need a different transform and do not know
enough Forth to program it, let me know.

@c --------------------------------------------------------------------
@node Superinstructions, Store Optimization, Simple instructions, Input File Format
@section Superinstructions
@cindex superinstructions, defining
@cindex defining superinstructions

Note: don't invest too much work in (static) superinstructions; a future
version of Vmgen will support dynamic superinstructions (see Ian
Piumarta and Fabio Riccardi, @cite{Optimizing Direct Threaded Code by
Selective Inlining}, PLDI'98), and static superinstructions have much
less benefit in that context (preliminary results indicate only a factor
1.1 speedup).

Here is an example of a superinstruction definition:

lit_sub = lit sub
@end example

@code{lit_sub} is the name of the superinstruction, and @code{lit} and
@code{sub} are its components.  This superinstruction performs the same
action as the sequence @code{lit} and @code{sub}.  It is generated
automatically by the VM code generation functions whenever that sequence
occurs, so if you want to use this superinstruction, you just need to
add this definition (and even that can be partially automatized,
@pxref{VM profiler}).

@cindex prefixes of superinstructions
Vmgen requires that the component instructions are simple instructions
defined before superinstructions using the components.  Currently, Vmgen
also requires that all the subsequences at the start of a
superinstruction (prefixes) must be defined as superinstruction before
the superinstruction.  I.e., if you want to define a superinstruction

foo4 = load add sub mul
@end example

you first have to define @code{load}, @code{add}, @code{sub} and
@code{mul}, plus

foo2 = load add
foo3 = load add sub
@end example

Here, @code{sumof4} is the longest prefix of @code{sumof5}, and @code{sumof3}
is the longest prefix of @code{sumof4}.

Note that Vmgen assumes that only the code it generates accesses stack
pointers, the instruction pointer, and various stack items, and it
performs optimizations based on this assumption.  Therefore, VM
instructions where your C code changes the instruction pointer should
only be used as last component; a VM instruction where your C code
accesses a stack pointer should not be used as component at all.  Vmgen
does not check these restrictions, they just result in bugs in your

@cindex include-skipped-insts
The Vmgen flag @code{include-skipped-insts} influences superinstruction
code generation.  Currently there is no support in the peephole
optimizer for both variations, so leave this flag alone for now.

@c -------------------------------------------------------------------
@node  Store Optimization, Register Machines, Superinstructions, Input File Format
@section Store Optimization
@cindex store optimization
@cindex optimization, stack stores
@cindex stack stores, optimization
@cindex eliminating stack stores

This minor optimization (0.6\%--0.8\% reduction in executed instructions
for Gforth) puts additional requirements on the instruction descriptions
and is therefore disabled by default.

What does it do?  Consider an instruction like

dup ( n -- n n )
@end example

For simplicity, also assume that we are not caching the top-of-stack in
a register.  Now, the C code for dup first loads @code{n} from the
stack, and then stores it twice to the stack, one time to the address
where it came from; that time is unnecessary, but gcc does not optimize
it away, so vmgen can do it instead (if you turn on the store

Vmgen uses the stack item's name to determine if the stack item contains
the same value as it did at the start.  Therefore, if you use the store
optimization, you have to ensure that stack items that have the same
name on input and output also have the same value, and are not changed
in the C code you supply.  I.e., the following code could fail if you
turn on the store optimization:

add1 ( n -- n )
@end example

Instead, you have to use different names, i.e.:

add1 ( n1 -- n1 )
@end example

Similarly, the store optimization assumes that the stack pointer is only
changed by Vmgen-erated code.  If your C code changes the stack pointer,
use different names in input and output stack items to avoid a (probably
wrong) store optimization, or turn the store optimization off for this
VM instruction.

To turn on the store optimization, write

\E store-optimization on
@end example

at the start of the file.  You can turn this optimization on or off
between any two VM instruction descriptions.  For turning it off again,
you can use

\E store-optimization off
@end example

@c -------------------------------------------------------------------
@node Register Machines,  , Store Optimization, Input File Format
@section Register Machines
@cindex Register VM
@cindex Superinstructions for register VMs
@cindex tracing of register VMs

If you want to implement a register VM rather than a stack VM with
Vmgen, there are two ways to do it: Directly and through

If you use the direct way, you define instructions that take the
register numbers as immediate arguments, like this:

add3 ( #src1 #src2 #dest -- )
reg[dest] = reg[src1]+reg[src2];
@end example

A disadvantage of this method is that during tracing you only see the
register numbers, but not the register contents.  Actually, with an
appropriate definition of @code{printarg_src} (@pxref{VM engine}), you
can print the values of the source registers on entry, but you cannot
print the value of the destination register on exit.

If you use superinstructions to define a register VM, you define simple
instructions that use a stack, and then define superinstructions that
have no overall stack effect, like this:

loadreg ( #src -- n )
n = reg[src];

storereg ( n #dest -- )
reg[dest] = n;

adds ( n1 n2 -- n )
n = n1+n2;

add3 = loadreg loadreg adds storereg
@end example

An advantage of this method is that you see the values and not just the
register numbers in tracing.  A disadvantage of this method is that
currently you cannot generate superinstructions directly, but only
through generating a sequence of simple instructions (we might change
this in the future if there is demand).

Could the register VM support be improved, apart from the issues
mentioned above?  It is hard to see how to do it in a general way,
because there are a number of different designs that different people
mean when they use the term @emph{register machine} in connection with
VM interpreters.  However, if you have ideas or requests in that
direction, please let me know (@pxref{Contact}).

@c ********************************************************************
@node Error messages, Using the generated code, Input File Format, Top
@chapter Error messages
@cindex error messages

These error messages are created by Vmgen:

@table @code

@cindex @code{# can only be on the input side} error
@item # can only be on the input side
You have used an instruction-stream prefix (usually @samp{#}) after the
@samp{--} (the output side); you can only use it before (the input

@cindex @code{prefix for this combination must be defined earlier} error
@item the prefix for this superinstruction must be defined earlier
You have defined a superinstruction (e.g. @code{abc = a b c}) without
defining its direct prefix (e.g., @code{ab = a b}),

@cindex @code{sync line syntax} error
@item sync line syntax
If you are using a preprocessor (e.g., @command{m4}) to generate Vmgen
input code, you may want to create @code{#line} directives (aka sync
lines).  This error indicates that such a line is not in th syntax
expected by Vmgen (this should not happen; please report the offending
line in a bug report).

@cindex @code{syntax error, wrong char} error
@item syntax error, wrong char
A syntax error.  If you do not see right away where the error is, it may
be helpful to check the following: Did you put an empty line in a VM
instruction where the C code is not delimited by braces (then the empty
line ends the VM instruction)?  If you used brace-delimited C code, did
you put the delimiting braces (and only those) at the start of the line,
without preceding white space?  Did you forget a delimiting brace?

@cindex @code{too many stacks} error
@item too many stacks
Vmgen currently supports 3 stacks (plus the instruction stream); if you
need more, let us know.

@cindex @code{unknown prefix} error
@item unknown prefix
The stack item does not match any defined type prefix (after stripping
away any stack prefix).  You should either declare the type prefix you
want for that stack item, or use a different type prefix

@cindex @code{unknown primitive} error
@item unknown primitive
You have used the name of a simple VM instruction in a superinstruction
definition without defining the simple VM instruction first.

@end table

In addition, the C compiler can produce errors due to code produced by
Vmgen; e.g., you need to define type cast functions.

@c ********************************************************************
@node Using the generated code, Hints, Error messages, Top
@chapter Using the generated code
@cindex generated code, usage
@cindex Using vmgen-erated code

The easiest way to create a working VM interpreter with Vmgen is
probably to start with @file{vmgen-ex}, and modify it for your purposes.
This chapter explains what the various wrapper and generated files do.
It also contains reference-manual style descriptions of the macros,
variables etc. used by the generated code, and you can skip that on
first reading.

* VM engine::                   Executing VM code
* VM instruction table::        
* VM code generation::          Creating VM code (in the front-end)
* Peephole optimization::       Creating VM superinstructions
* VM disassembler::             for debugging the front end
* VM profiler::                 for finding worthwhile superinstructions
@end menu

@c --------------------------------------------------------------------
@node VM engine, VM instruction table, Using the generated code, Using the generated code
@section VM engine
@cindex VM instruction execution
@cindex engine
@cindex executing VM code
@cindex @file{engine.c}
@cindex @file{-vm.i} output file

The VM engine is the VM interpreter that executes the VM code.  It is
essential for an interpretive system.

Vmgen supports two methods of VM instruction dispatch: @emph{threaded
code} (fast, but gcc-specific), and @emph{switch dispatch} (slow, but
portable across C compilers); you can use conditional compilation
(@samp{defined(__GNUC__)}) to choose between these methods, and our
example does so.

For both methods, the VM engine is contained in a C-level function.
Vmgen generates most of the contents of the function for you
(@file{@var{name}-vm.i}), but you have to define this function, and
macros and variables used in the engine, and initialize the variables.
In our example the engine function also includes
@file{@var{name}-labels.i} (@pxref{VM instruction table}).

@cindex tracing VM code
@cindex superinstructions and tracing
In addition to executing the code, the VM engine can optionally also
print out a trace of the executed instructions, their arguments and
results.  For superinstructions it prints the trace as if only component
instructions were executed; this allows to introduce new
superinstructions while keeping the traces comparable to old ones
(important for regression tests).

It costs significant performance to check in each instruction whether to
print tracing code, so we recommend producing two copies of the engine:
one for fast execution, and one for tracing.  See the rules for
@file{engine.o} and @file{engine-debug.o} in @file{vmgen-ex/Makefile}
for an example.

The following macros and variables are used in @file{@var{name}-vm.i}:

@table @code

@findex LABEL
@item LABEL(@var{inst_name})
This is used just before each VM instruction to provide a jump or
@code{switch} label (the @samp{:} is provided by Vmgen).  For switch
dispatch this should expand to @samp{case @var{label}:}; for
threaded-code dispatch this should just expand to @samp{@var{label}:}.
In either case @var{label} is usually the @var{inst_name} with some
prefix or suffix to avoid naming conflicts.

@findex LABEL2
@item LABEL2(@var{inst_name})
This will be used for dynamic superinstructions; at the moment, this
should expand to nothing.

@findex NAME
@item NAME(@var{inst_name_string})
Called on entering a VM instruction with a string containing the name of
the VM instruction as parameter.  In normal execution this should be
expand to nothing, but for tracing this usually prints the name, and
possibly other information (several VM registers in our example).

@findex DEF_CA
@item DEF_CA
Usually empty.  Called just inside a new scope at the start of a VM
instruction.  Can be used to define variables that should be visible
during every VM instruction.  If you define this macro as non-empty, you
have to provide the finishing @samp{;} in the macro.

@findex NEXT_P0
@findex NEXT_P1
@findex NEXT_P2
The three parts of instruction dispatch.  They can be defined in
different ways for best performance on various processors (see
@file{engine.c} in the example or @file{engine/threaded.h} in Gforth).
@samp{NEXT_P0} is invoked right at the start of the VM instruction (but
after @samp{DEF_CA}), @samp{NEXT_P1} right after the user-supplied C
code, and @samp{NEXT_P2} at the end.  The actual jump has to be
performed by @samp{NEXT_P2} (if you would do it earlier, important parts
of the VM instruction would not be executed).

The simplest variant is if @samp{NEXT_P2} does everything and the other
macros do nothing.  Then also related macros like @samp{IP},
@samp{SET_IP}, @samp{IP}, @samp{INC_IP} and @samp{IPTOS} are very
straightforward to define.  For switch dispatch this code consists just
of a jump to the dispatch code (@samp{goto next_inst;} in our example);
for direct threaded code it consists of something like
@samp{(@{cfa=*ip++; goto *cfa;@})}.

Pulling code (usually the @samp{cfa=*ip++;}) up into @samp{NEXT_P1}
usually does not cause problems, but pulling things up into
@samp{NEXT_P0} usually requires changing the other macros (and, at least
for Gforth on Alpha, it does not buy much, because the compiler often
manages to schedule the relevant stuff up by itself).  An even more
extreme variant is to pull code up even further, into, e.g., NEXT_P1 of
the previous VM instruction (prefetching, useful on PowerPCs).

@findex INC_IP
@item INC_IP(@var{n})
This increments @code{IP} by @var{n}.

@findex SET_IP
@item SET_IP(@var{target})
This sets @code{IP} to @var{target}.

@cindex type cast macro
@findex vm_@var{A}2@var{B}
@item vm_@var{A}2@var{B}(a,b)
Type casting macro that assigns @samp{a} (of type @var{A}) to @samp{b}
(of type @var{B}).  This is mainly used for getting stack items into
variables and back.  So you need to define macros for every combination
of stack basic type (@code{Cell} in our example) and type-prefix types
used with that stack (in both directions).  For the type-prefix type,
you use the type-prefix (not the C type string) as type name (e.g.,
@samp{vm_Cell2i}, not @samp{vm_Cell2Cell}).  In addition, you have to
define a vm_@var{X}2@var{X} macro for the stack's basic type @var{X}
(used in superinstructions).

@cindex instruction stream, basic type
The stack basic type for the predefined @samp{inst-stream} is
@samp{Cell}.  If you want a stack with the same item size, making its
basic type @samp{Cell} usually reduces the number of macros you have to

@cindex unions in type cast macros
@cindex casts in type cast macros
@cindex type casting between floats and integers
Here our examples differ a lot: @file{vmgen-ex} uses casts in these
macros, whereas @file{vmgen-ex2} uses union-field selection (or
assignment to union fields).  Note that casting floats into integers and
vice versa changes the bit pattern (and you do not want that).  In this
case your options are to use a (temporary) union, or to take the address
of the value, cast the pointer, and dereference that (not always
possible, and sometimes expensive).

@findex vm_two@var{A}2@var{B}
@findex vm_@var{B}2two@var{A}
@item vm_two@var{A}2@var{B}(a1,a2,b)
@item vm_@var{B}2two@var{A}(b,a1,a2)
Type casting between two stack items (@code{a1}, @code{a2}) and a
variable @code{b} of a type that takes two stack items.  This does not
occur in our small examples, but you can look at Gforth for examples
(see @code{vm_twoCell2d} in @file{engine/forth.h}).

@cindex stack pointer definition
@cindex instruction pointer definition
@item @var{stackpointer}
For each stack used, the stackpointer name given in the stack
declaration is used.  For a regular stack this must be an l-expression;
typically it is a variable declared as a pointer to the stack's basic
type.  For @samp{inst-stream}, the name is @samp{IP}, and it can be a
plain r-value; typically it is a macro that abstracts away the
differences between the various implementations of @code{NEXT_P*}.

@cindex IMM_ARG
@findex IMM_ARG
@item IMM_ARG(access,value)
Define this to expland to ``(access)''.  This is just a placeholder for
future extensions.

@cindex top of stack caching
@cindex stack caching
@cindex TOS
@findex IPTOS
@item @var{stackpointer}TOS
The top-of-stack for the stack pointed to by @var{stackpointer}.  If you
are using top-of-stack caching for that stack, this should be defined as
variable; if you are not using top-of-stack caching for that stack, this
should be a macro expanding to @samp{@var{stackpointer}[0]}.  The stack
pointer for the predefined @samp{inst-stream} is called @samp{IP}, so
the top-of-stack is called @samp{IPTOS}.

@findex IF_@var{stackpointer}TOS
@item IF_@var{stackpointer}TOS(@var{expr})
Macro for executing @var{expr}, if top-of-stack caching is used for the
@var{stackpointer} stack.  I.e., this should do @var{expr} if there is
top-of-stack caching for @var{stackpointer}; otherwise it should do

@findex SUPER_END
This is used by the VM profiler (@pxref{VM profiler}); it should not do
anything in normal operation, and call @code{vm_count_block(IP)} for

This is just a hint to Vmgen and does nothing at the C level.

@findex VM_DEBUG
@item VM_DEBUG
If this is defined, the tracing code will be compiled in (slower
interpretation, but better debugging).  Our example compiles two
versions of the engine, a fast-running one that cannot trace, and one
with potential tracing and profiling.

@findex vm_debug
@item vm_debug
Needed only if @samp{VM_DEBUG} is defined.  If this variable contains
true, the VM instructions produce trace output.  It can be turned on or
off at any time.

@findex vm_out
@item vm_out
Needed only if @samp{VM_DEBUG} is defined.  Specifies the file on which
to print the trace output (type @samp{FILE *}).

@findex printarg_@var{type}
@item printarg_@var{type}(@var{value})
Needed only if @samp{VM_DEBUG} is defined.  Macro or function for
printing @var{value} in a way appropriate for the @var{type}.  This is
used for printing the values of stack items during tracing.  @var{Type}
is normally the type prefix specified in a @code{type-prefix} definition
(e.g., @samp{printarg_i}); in superinstructions it is currently the
basic type of the stack.

@end table

@c --------------------------------------------------------------------
@node VM instruction table, VM code generation, VM engine, Using the generated code
@section VM instruction table
@cindex instruction table
@cindex opcode definition
@cindex labels for threaded code
@cindex @code{vm_prim}, definition
@cindex @file{-labels.i} output file

For threaded code we also need to produce a table containing the labels
of all VM instructions.  This is needed for VM code generation
(@pxref{VM code generation}), and it has to be done in the engine
function, because the labels are not visible outside.  It then has to be
passed outside the function (and assigned to @samp{vm_prim}), to be used
by the VM code generation functions.

This means that the engine function has to be called first to produce
the VM instruction table, and later, after generating VM code, it has to
be called again to execute the generated VM code (yes, this is ugly).
In our example program, these two modes of calling the engine function
are differentiated by the value of the parameter ip0 (if it equals 0,
then the table is passed out, otherwise the VM code is executed); in our
example, we pass the table out by assigning it to @samp{vm_prim} and
returning from @samp{engine}.

In our example (@file{vmgen-ex/engine.c}), we also build such a table for
switch dispatch; this is mainly done for uniformity.

For switch dispatch, we also need to define the VM instruction opcodes
used as case labels in an @code{enum}.

For both purposes (VM instruction table, and enum), the file
@file{@var{name}-labels.i} is generated by Vmgen.  You have to define
the following macro used in this file:

@table @code

@findex INST_ADDR
@item INST_ADDR(@var{inst_name})
For switch dispatch, this is just the name of the switch label (the same
name as used in @samp{LABEL(@var{inst_name})}), for both uses of
@file{@var{name}-labels.i}.  For threaded-code dispatch, this is the
address of the label defined in @samp{LABEL(@var{inst_name})}); the
address is taken with @samp{&&} (@pxref{Labels as Values, , Labels as
Values,, GNU C Manual}).

@end table

@c --------------------------------------------------------------------
@node VM code generation, Peephole optimization, VM instruction table, Using the generated code
@section VM code generation
@cindex VM code generation
@cindex code generation, VM
@cindex @file{-gen.i} output file

Vmgen generates VM code generation functions in @file{@var{name}-gen.i}
that the front end can call to generate VM code.  This is essential for
an interpretive system.

@findex gen_@var{inst}
For a VM instruction @samp{x ( #a b #c -- d )}, Vmgen generates a
function with the prototype

void gen_x(Inst **ctp, a_type a, c_type c)
@end example

The @code{ctp} argument points to a pointer to the next instruction.
@code{*ctp} is increased by the generation functions; i.e., you should
allocate memory for the code to be generated beforehand, and start with
*ctp set at the start of this memory area.  Before running out of
memory, allocate a new area, and generate a VM-level jump to the new
area (this overflow handling is not implemented in our examples).

@cindex immediate arguments, VM code generation
The other arguments correspond to the immediate arguments of the VM
instruction (with their appropriate types as defined in the
@code{type_prefix} declaration.

The following types, variables, and functions are used in

@table @code

@findex Inst
@item Inst
The type of the VM instruction; if you use threaded code, this is
@code{void *}; for switch dispatch this is an integer type.

@cindex @code{vm_prim}, use
@item vm_prim
The VM instruction table (type: @code{Inst *}, @pxref{VM instruction table}).

@findex gen_inst
@item gen_inst(Inst **ctp, Inst i)
This function compiles the instruction @code{i}.  Take a look at it in
@file{vmgen-ex/peephole.c}.  It is trivial when you don't want to use
superinstructions (just the last two lines of the example function), and
slightly more complicated in the example due to its ability to use
superinstructions (@pxref{Peephole optimization}).

@findex genarg_@var{type_prefix}
@item genarg_@var{type_prefix}(Inst **ctp, @var{type} @var{type_prefix})
This compiles an immediate argument of @var{type} (as defined in a
@code{type-prefix} definition).  These functions are trivial to define
(see @file{vmgen-ex/support.c}).  You need one of these functions for
every type that you use as immediate argument.

@end table

In addition to using these functions to generate code, you should call
@code{BB_BOUNDARY} at every basic block entry point if you ever want to
use superinstructions (or if you want to use the profiling supported by
Vmgen; but this support is also useful mainly for selecting
superinstructions).  If you use @code{BB_BOUNDARY}, you should also
define it (take a look at its definition in @file{vmgen-ex/mini.y}).

You do not need to call @code{BB_BOUNDARY} after branches, because you
will not define superinstructions that contain branches in the middle
(and if you did, and it would work, there would be no reason to end the
superinstruction at the branch), and because the branches announce
themselves to the profiler.

@c --------------------------------------------------------------------
@node Peephole optimization, VM disassembler, VM code generation, Using the generated code
@section Peephole optimization
@cindex peephole optimization
@cindex superinstructions, generating
@cindex @file{peephole.c}
@cindex @file{-peephole.i} output file

You need peephole optimization only if you want to use
superinstructions.  But having the code for it does not hurt much if you
do not use superinstructions.

A simple greedy peephole optimization algorithm is used for
superinstruction selection: every time @code{gen_inst} compiles a VM
instruction, it checks if it can combine it with the last VM instruction
(which may also be a superinstruction resulting from a previous peephole
optimization); if so, it changes the last instruction to the combined
instruction instead of laying down @code{i} at the current @samp{*ctp}.

The code for peephole optimization is in @file{vmgen-ex/peephole.c}.
You can use this file almost verbatim.  Vmgen generates
@file{@var{file}-peephole.i} which contains data for the peephoile

@findex init_peeptable
You have to call @samp{init_peeptable()} after initializing
@samp{vm_prim}, and before compiling any VM code to initialize data
structures for peephole optimization.  After that, compiling with the VM
code generation functions will automatically combine VM instructions
into superinstructions.  Since you do not want to combine instructions
across VM branch targets (otherwise there will not be a proper VM
instruction to branch to), you have to call @code{BB_BOUNDARY}
(@pxref{VM code generation}) at branch targets.

@c --------------------------------------------------------------------
@node VM disassembler, VM profiler, Peephole optimization, Using the generated code
@section VM disassembler
@cindex VM disassembler
@cindex disassembler, VM code
@cindex @file{disasm.c}
@cindex @file{-disasm.i} output file

A VM code disassembler is optional for an interpretive system, but
highly recommended during its development and maintenance, because it is
very useful for detecting bugs in the front end (and for distinguishing
them from VM interpreter bugs).

Vmgen supports VM code disassembling by generating
@file{@var{file}-disasm.i}.  This code has to be wrapped into a
function, as is done in @file{vmgen-ex/disasm.c}.  You can use this file
almost verbatim.  In addition to @samp{vm_@var{A}2@var{B}(a,b)},
@samp{vm_out}, @samp{printarg_@var{type}(@var{value})}, which are
explained above, the following macros and variables are used in
@file{@var{file}-disasm.i} (and you have to define them):

@table @code

@item ip
This variable points to the opcode of the current VM instruction.

@cindex @code{IP}, @code{IPTOS} in disassmbler
@item IP IPTOS
@samp{IPTOS} is the first argument of the current VM instruction, and
@samp{IP} points to it; this is just as in the engine, but here
@samp{ip} points to the opcode of the VM instruction (in contrast to the
engine, where @samp{ip} points to the next cell, or even one further).

@findex VM_IS_INST
@item VM_IS_INST(Inst i, int n)
Tests if the opcode @samp{i} is the same as the @samp{n}th entry in the
VM instruction table.

@end table

@c --------------------------------------------------------------------
@node VM profiler,  , VM disassembler, Using the generated code
@section VM profiler
@cindex VM profiler
@cindex profiling for selecting superinstructions
@cindex superinstructions and profiling
@cindex @file{profile.c}
@cindex @file{-profile.i} output file

The VM profiler is designed for getting execution and occurence counts
for VM instruction sequences, and these counts can then be used for
selecting sequences as superinstructions.  The VM profiler is probably
not useful as profiling tool for the interpretive system.  I.e., the VM
profiler is useful for the developers, but not the users of the
interpretive system.

The output of the profiler is: for each basic block (executed at least
once), it produces the dynamic execution count of that basic block and
all its subsequences; e.g.,

       9227465  lit storelocal 
       9227465  storelocal branch 
       9227465  lit storelocal branch 
@end example

I.e., a basic block consisting of @samp{lit storelocal branch} is
executed 9227465 times.

@cindex @file{stat.awk}
@cindex @file{seq2rule.awk}
This output can be combined in various ways.  E.g.,
@file{vmgen-ex/stat.awk} adds up the occurences of a given sequence wrt
dynamic execution, static occurence, and per-program occurence.  E.g.,

      2      16        36910041 loadlocal lit 
@end example

indicates that the sequence @samp{loadlocal lit} occurs in 2 programs,
in 16 places, and has been executed 36910041 times.  Now you can select
superinstructions in any way you like (note that compile time and space
typically limit the number of superinstructions to 100--1000).  After
you have done that, @file{vmgen/seq2rule.awk} turns lines of the form
above into rules for inclusion in a Vmgen input file.  Note that this
script does not ensure that all prefixes are defined, so you have to do
that in other ways.  So, an overall script for turning profiles into
superinstructions can look like this:

awk -f stat.awk|
awk '$3>=10000'|                #select sequences
fgrep -v -f peephole-blacklist| #eliminate wrong instructions
awk -f seq2rule.awk|            #turn into superinstructions
sort -k 3 >mini-super.vmg       #sort sequences
@end example

Here the dynamic count is used for selecting sequences (preliminary
results indicate that the static count gives better results, though);
the third line eliminates sequences containing instructions that must not
occur in a superinstruction, because they access a stack directly.  The
dynamic count selection ensures that all subsequences (including
prefixes) of longer sequences occur (because subsequences have at least
the same count as the longer sequences); the sort in the last line
ensures that longer superinstructions occur after their prefixes.

But before using this, you have to have the profiler.  Vmgen supports its
creation by generating @file{@var{file}-profile.i}; you also need the
wrapper file @file{vmgen-ex/profile.c} that you can use almost verbatim.

@cindex @code{SUPER_END} in profiling
@cindex @code{BB_BOUNDARY} in profiling
The profiler works by recording the targets of all VM control flow
changes (through @code{SUPER_END} during execution, and through
@code{BB_BOUNDARY} in the front end), and counting (through
@code{SUPER_END}) how often they were targeted.  After the program run,
the numbers are corrected such that each VM basic block has the correct
count (entering a block without executing a branch does not increase the
count, and the correction fixes that), then the subsequences of all
basic blocks are printed.  To get all this, you just have to define
@code{SUPER_END} (and @code{BB_BOUNDARY}) appropriately, and call
@code{vm_print_profile(FILE *file)} when you want to output the profile
on @code{file}.

@cindex @code{VM_IS_INST} in profiling
The @file{@var{file}-profile.i} is similar to the disassembler file, and
it uses variables and functions defined in @file{vmgen-ex/profile.c},
plus @code{VM_IS_INST} already defined for the VM disassembler
(@pxref{VM disassembler}).

@c **********************************************************
@node Hints, The future, Using the generated code, Top
@chapter Hints
@cindex hints

* Floating point::              and stacks
@end menu

@c --------------------------------------------------------------------
@node Floating point,  , Hints, Hints
@section Floating point

How should you deal with floating point values?  Should you use the same
stack as for integers/pointers, or a different one?  This section
discusses this issue with a view on execution speed.

The simpler approach is to use a separate floating-point stack.  This
allows you to choose FP value size without considering the size of the
integers/pointers, and you avoid a number of performance problems.  The
main downside is that this needs an FP stack pointer (and that may not
fit in the register file on the 386 arhitecture, costing some
performance, but comparatively little if you take the other option into
account).  If you use a separate FP stack (with stack pointer @code{fp}),
using an fpTOS is helpful on most machines, but some spill the fpTOS
register into memory, and fpTOS should not be used there.

The other approach is to share one stack (pointed to by, say, @code{sp})
between integer/pointer and floating-point values.  This is ok if you do
not use @code{spTOS}.  If you do use @code{spTOS}, the compiler has to
decide whether to put that variable into an integer or a floating point
register, and the other type of operation becomes quite expensive on
most machines (because moving values between integer and FP registers is
quite expensive).  If a value of one type has to be synthesized out of
two values of the other type (@code{double} types), things are even more

One way around this problem would be to not use the @code{spTOS}
supported by Vmgen, but to use explicit top-of-stack variables (one for
integers, one for FP values), and having a kind of accumulator+stack
architecture (e.g., Ocaml bytecode uses this approach); however, this is
a major change, and it's ramifications are not completely clear.

@c **********************************************************
@node The future, Changes, Hints, Top
@chapter The future
@cindex future ideas

We have a number of ideas for future versions of Vmgen.  However, there
are so many possible things to do that we would like some feedback from
you.  What are you doing with Vmgen, what features are you missing, and

One idea we are thinking about is to generate just one @file{.c} file
instead of letting you copy and adapt all the wrapper files (you would
still have to define stuff like the type-specific macros, and stack
pointers etc. somewhere).  The advantage would be that, if we change the
wrapper files between versions, you would not need to integrate your
changes and our changes to them; Vmgen would also be easier to use for
beginners.  The main disadvantage of that is that it would reduce the
flexibility of Vmgen a little (well, those who like flexibility could
still patch the resulting @file{.c} file, like they are now doing for
the wrapper files).  In any case, if you are doing things to the wrapper
files that would cause problems in a generated-@file{.c}-file approach,
please let us know.

@c **********************************************************
@node Changes, Contact, The future, Top
@chapter Changes
@cindex Changes from old versions

User-visible changes between 0.5.9-20020822 and 0.5.9-20020901:

The store optimization is now disabled by default, but can be enabled by
the user (@pxref{Store Optimization}).  Documentation for this
optimization is also new.

User-visible changes between 0.5.9-20010501 and 0.5.9-20020822:

There is now a manual (in info, HTML, Postscript, or plain text format).

There is the vmgen-ex2 variant of the vmgen-ex example; the new
variant uses a union type instead of lots of casting.

Both variants of the example can now be compiled with an ANSI C compiler
(using switch dispatch and losing quite a bit of performance); tested
with @command{lcc}.

Users of the gforth-0.5.9-20010501 version of Vmgen need to change
several things in their source code to use the current version.  I
recommend keeping the gforth-0.5.9-20010501 version until you have
completed the change (note that you can have several versions of Gforth
installed at the same time).  I hope to avoid such incompatible changes
in the future.

The required changes are:

@table @code

@cindex @code{TAIL;}, changes
@item TAIL;
has been renamed into @code{INST_TAIL;} (less chance of an accidental

@cindex @code{vm_@var{A}2@var{B}}, changes
@item vm_@var{A}2@var{B}
now takes two arguments.

@cindex @code{vm_two@var{A}2@var{B}}, changes
@item vm_two@var{A}2@var{B}(b,a1,a2);
changed to vm_two@var{A}2@var{B}(a1,a2,b) (note the absence of the @samp{;}).

@end table

Also some new macros have to be defined, e.g., @code{INST_ADDR}, and
@code{LABEL}; some macros have to be defined in new contexts, e.g.,
@code{VM_IS_INST} is now also needed in the disassembler.

@c *********************************************************
@node Contact, Copying This Manual, Changes, Top
@chapter Contact

To report a bug, use

For discussion on Vmgen (e.g., how to use it), use the mailing list
@email{} (use
@url{} to subscribe).

You can find vmgen information at

@c ***********************************************************
@node Copying This Manual, Index, Contact, Top
@appendix Copying This Manual

* GNU Free Documentation License::  License for copying this manual.
@end menu

@include fdl.texi

@node Index,  , Copying This Manual, Top
@unnumbered Index

@printindex cp


FreeBSD-CVSweb <>