grammar::me::cpu::gasm -
ME assembler
package require grammar::me::cpu::gasm ? 0.1 ?
::grammar::me::cpu::gasm::begin g n ? mode ? ? note ?
::grammar::me::cpu::gasm::done --> t
::grammar::me::cpu::gasm::state
::grammar::me::cpu::gasm::state! s
::grammar::me::cpu::gasm::lift t dst = src
::grammar::me::cpu::gasm::Inline t node label
::grammar::me::cpu::gasm::Cmd cmd ? arg... ?
::grammar::me::cpu::gasm::Bra
::grammar::me::cpu::gasm::Nop text
::grammar::me::cpu::gasm::Note text
::grammar::me::cpu::gasm::Jmp label
::grammar::me::cpu::gasm::Exit
::grammar::me::cpu::gasm::Who label
::grammar::me::cpu::gasm::/Label name
::grammar::me::cpu::gasm::/Clear
::grammar::me::cpu::gasm::/Ok
::grammar::me::cpu::gasm::/Fail
::grammar::me::cpu::gasm::/At name
::grammar::me::cpu::gasm::/CloseLoop
This package provides a simple in-memory assembler. Its origin is that
of a support package for use by packages converting PEG and other
grammars into a corresponding matcher based on the ME virtual machine,
like
page::compiler::peg::mecpu. Despite that it is actually
mostly agnostic regarding the instructions, users can choose any
instruction set they like.
The program under construction is held in a graph structure (See
package struct::graph) during assembly and subsequent
manipulation, with instructions represented by nodes, and the flow of
execution between instructions explicitly encoded in the arcs between
them.
In this model jumps are not encoded explicitly, they are implicit in
the arcs. The generation of explicit jumps is left to any code
converting the graph structure into a more conventional
representation. The same goes for branches. They are implicitly
encoded by all instructions which have two outgoing arcs, whereas all
other instructions have only one outgoing arc. Their conditonality is
handled by tagging their outgoing arcs with information about the
conditions under which they are taken.
While the graph the assembler operates on is supplied from the
outside, i.e. external, it does manage some internal state, namely:
- The handle of the graph node most assembler operations will
work on, the anchor.
- A mapping from arbitrary strings to instructions. I.e. it is
possible to label an instruction during assembly, and later
recall that instruction by its label.
- The condition code to use when creating arcs between
instructions, which is one of always, ok, and
fail.
- The current operation mode, one of halt,
okfail, and !okfail.
- The name of a node in a tree. This, and the operation mode
above are the parts most heavily influenced by the needs of a grammar
compiler, as they assume some basic program structures (selected
through the operation mode), and intertwine the graph with a tree,
like the AST for the grammar to be compiled.
As the graph the assembler is operating on, and the tree it is
intertwined with, are supplied to the assembler from the outside it is
necessary to specify the API expected from them, and to describe the
structures expected and/or generated by the assembler in either.
- Any graph object command used by the assembler has to provide
the API as specified in the documentation for the package
struct::graph.
- Any tree object command used by the assembler has to provide
the API as specified in the documentation for the package
struct::tree.
- Any instruction (node) generated by the assembler in a graph
will have at least two, and at most three attributes:
- instruction
- The value of this attribute is the name of
the instruction. The only names currently defined by the assembler are
the three pseudo-instructions
- NOP
- This instruction does nothing. Useful for fixed
framework nodes, unchanging jump destinations, and the like. No
arguments.
- C
- A .NOP to allow the insertion of arbitrary comments
into the instruction stream, i.e. a comment node. One argument, the
text of the comment.
- BRA
- A .NOP serving as explicitly coded conditional
branch. No arguments.
However we reserve the space of all instructions whose names begin
with a "." (dot) for future use by the assembler.
- arguments
- The value of this attribute is a list of
strings, the arguments of the instruction. The contents are dependent
on the actual instruction and the assembler doesn't know or care about
them. This means for example that it has no builtin knowledge about
what instruction need which arguments and thus doesn't perform any
type of checking.
- expr
- This attribute is optional. When it is present its
value is the name of a node in the tree intertwined with the graph.
- Any arc between two instructions will have one attribute:
- condition
- The value of this attribute determines under which
condition execution will take this arc. It is one of always,
ok, and fail. The first condition is used for all arcs
which are the single outgoing arc of an instruction. The other two are
used for the two outgoing arcs of an instruction which implicitly
encode a branch.
- A tree node given to the assembler for cross-referencing will
be written to and given the following attributes, some fixed, some
dependent on the operation mode. All values will be references to
nodes in the instruction graph. Some of the instruction will expect
some or specific sets of these attributes.
- gas::entry
- Always written.
- gas::exit
- Written for all modes but okfail.
- gas::exit::ok
- Written for mode okfail.
- gas::exit::fail
- Written for mode okfail.
-
::grammar::me::cpu::gasm::begin g n ? mode ? ? note ?
-
This command starts the assembly of an instruction sequence, and
(re)initializes the state of the assembler. After completion of the
instruction sequence use ::grammar::me::cpu::gasm::done to
finalize the assembler.
It will operate on the graph g in the specified mode
(Default is okfail). As part of the initialization it will
always create a standard .NOP instruction and label it "entry". The
creation of the remaining standard instructions is
mode-dependent:
- halt
- An "icf_halt" instruction labeled "exit/return".
- !okfail
- An "icf_ntreturn" instruction labeled "exit/return".
- okfail
- Two .NOP instructions labeled "exit/ok" and
"exit/fail" respectively.
The note, if specified (default is not), is given to the "entry" .NOP instruction.
The node reference n is simply stored for use by
::grammar::me::cpu::gasm::done. It has to refer to a node in the
tree t argument of that command.
After the initialization is done the "entry" instruction will be the
anchor, and the condition code will be set to always.
The command returns the empy string as its result.
-
::grammar::me::cpu::gasm::done --> t
-
This command finalizes the creation of an instruction sequence and
then clears the state of the assembler.
NOTE that this does not delete any of the created
instructions. They can be made available to future begin/done cycles.
Further assembly will be possible only after reinitialization of the
system via ::grammar::me::cpu::gasm::begin.
Before the state is cleared selected references to selected
instructions will be written to attributes of the node n in the
tree t.
Which instructions are saved is mode-dependent. Both mode
and the destination node n were specified during invokation of
::grammar::me::cpu::gasm::begin.
Independent of the mode a reference to the instruction labeled "entry"
will be saved to the attribute gas::entry of n. The
reference to the node n will further be saved into the attribute
"expr" of the "entry" instruction. Beyond that
- halt
- A reference to the instruction labeled
"exit/return" will be saved to the attribute gas::exit of
n.
- okfail
- See halt.
- !okfail
- Reference to the two instructions labeled
"exit/ok" and "exit/fail" will be saved to the attributes
gas::exit::ok and gas::exit::fail of n
respectively.
The command returns the empy string as its result.
-
::grammar::me::cpu::gasm::state
-
This command returns the current state of the assembler. Its format is
not documented and considered to be internal to the package.
-
::grammar::me::cpu::gasm::state! s
-
This command takes a serialized assembler state s as returned by
::grammar::me::cpu::gasm::state and makes it the current state
of the assembler.
Note that this may overwrite label definitions, however all
non-conflicting label definitions in the state before are not touched
and merged with s.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::lift t dst = src
-
This command operates on the tree t. It copies the contents of
the attributes gas::entry, gas::exit::ok and
gas::exit::fail from the node src to the node dst.
It returns the empty string as its result.
-
::grammar::me::cpu::gasm::Inline t node label
-
This command links an instruction sequence created by an earlier
begin/done pair into the current instruction sequence.
To this end it
- reads the instruction references from the attributes
gas::entry, gas::exit::ok, and gas::exit::fail
from the node n of the tree t and makes them available to
assembler und the labels label/entry, label/exit::ok, and
label/exit::fail respectively.
- Creates an arc from the anchor to the node labeled
label/entry, and tags it with the current condition code.
- Makes the node labeled label/exit/ok the new anchor.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::Cmd cmd ? arg... ?
-
This is the basic command to add instructions to the graph.
It creates a new instruction of type cmd with the given
arguments arg...
If the anchor was defined it will also create an arc from the
anchor to the new instruction using the current condition code.
After the call the new instruction will be the anchor and the
current condition code will be set to always.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::Bra
-
This is a convenience command to create a .BRA pseudo-instruction. It
uses ::grammar::me::cpu::gasm::Cmd to actually create the
instruction and inherits its behaviour.
-
::grammar::me::cpu::gasm::Nop text
-
This is a convenience command to create a .NOP pseudo-instruction. It
uses ::grammar::me::cpu::gasm::Cmd to actually create the
instruction and inherits its behaviour.
The text will be saved as the first and only argument of the new
instruction.
-
::grammar::me::cpu::gasm::Note text
-
This is a convenience command to create a .C pseudo-instruction,
i.e. a comment. It uses ::grammar::me::cpu::gasm::Cmd to
actually create the instruction and inherits its behaviour.
The text will be saved as the first and only argument of the new
instruction.
-
::grammar::me::cpu::gasm::Jmp label
-
This command creates an arc from the anchor to the instruction
labeled with label, and tags with the the current condition
code.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::Exit
-
This command creates an arc from the anchor to one of the exit
instructions, based on the operation mode (see
::grammar::me::cpu::gasm::begin), and tags it with current
condition code.
For mode okfail it links to the instruction labeled either
"exit/ok" or "exit/fail", depending on the current condition code, and
tagging it with the current condition code
For the other two modes it links to the instruction labeled
"exit/return", tagging it condition code always, independent
the current condition code.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::Who label
-
This command returns a reference to the instruction labeled with
label.
-
::grammar::me::cpu::gasm::/Label name
-
This command labels the anchor with name.
Note that an instruction can have more than one label.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::/Clear
-
This command clears the anchor, leaving it undefined, and
further resets the current condition code to always.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::/Ok
-
This command sets the current condition code to ok.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::/Fail
-
This command sets the current condition code to fail.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::/At name
-
This command sets the anchor to the instruction labeled with
name, and further resets the current condition code to
always.
The command returns the empty string as its result.
-
::grammar::me::cpu::gasm::/CloseLoop
-
This command marks the anchor as the last instruction in a loop
body, by creating the attribute LOOP.
The command returns the empty string as its result.
This document, and the package it describes, will undoubtedly contain
bugs and other problems.
Please report such in the category
grammar_me of the
http://sourceforge.net/tracker/?group_id=12883.
Please also report any ideas for enhancements you may have for either
package and/or documentation.
virtual machine, parsing, grammar, assembler, graph, tree