3 The Mono runtime will implement the JIT engine (and a byte
4 code interpreter for quickly porting to new systems), the
5 class loader, the garbage collector, threading system and
6 metadata access libraries.
8 Currently the runtime has an image loader and metadata access
9 entry points. The runtime comes with a simple interpreter
10 that can execute very simple programs.
12 ** Executing MSIL/CIL images
14 The code will load an executable and map the references to
15 external assemblies to our own version of the assemblies on
18 Our roadmap looks like this, this has been updated as of
23 * Milestone 1: <b>Done</b> Fully read and parse all CIL byte-codes
24 and metadata tokens (ie, a disassembler).
26 * Milestone 2: <b>Done</b> Complete an interpreter for CIL byte
27 codes. This interpreter can be used temporarly to
28 run CIL byte code on a system where no JIT is
31 * Milestone 3: <b>Done</b>Define an <i>lburg</i>-like
32 instruction selector for the JITer for Intel.
33 Although slower at JITing than a streaming JITer, it
34 generates better code. The same grammar can later
35 be used for the stream jitter.
37 * Milestone 4: Implement JITer. This is where our
38 current efforts are focused on, the JITer is 60% ready.
40 * Milestone 5: Port of the JITer to non IA32 systems.
43 A setup similar to the Kaffe JIT engine will be used to
44 layout the code to support non-IA32 architectures. Our work
45 will be focused on getting a IA32 version running first.
47 The JIT engine should work on Linux and Win32, although you
48 will need to install the CygWin32 development tools to get a
49 Unix-like compilation environment.
51 ** JIT Engine (<b>updated, Nov 16th, 2001</b>)
53 The JIT engine uses a code-generator generator approach for
54 compilation. Given the properties of CIL byte codes, we can
55 take full advantage of a real instruction selector for our
58 There are a couple of books that deal with this technique: "A
59 Retargetable C Compiler" and "Advanced Compiler Design and
60 Implementation" are good references. You can also get a
61 technical description of <a
62 href="http://research.microsoft.com/copyright/accept.asp?path=http://www.research.microsoft.com/~drh/pubs/iburg.pdf&pub=ACM">lbrug</a>.
64 A few papers that describe the instruction selector:
67 * <a href="http://research.microsoft.com/copyright/accept.asp?path=http://www.research.microsoft.com/~drh/pubs/interface.pdf&pub=wiley">A code generation interface for ANSI C</a>
70 * <a href="http://research.microsoft.com/copyright/accept.asp?path=http://www.research.microsoft.com/~drh/pubs/iburg.pdf&pub=ACM">Engineering efficient code generators using tree matching and dynamic programming.</a>
76 We have decided to implement a generational tracing garbage
77 collector, which is very similar to the one being used by
78 .NET. For an introduction to the garbage collection system
79 used by Microsoft's CLR implementation, you can read this book
81 href="http://www.amazon.com/exec/obidos/ASIN/0471941484/o/qid=992556433/sr=2-1/ref=aps_sr_b_1_1/103-5866388-0492603">Garbage
84 Another consideration is to use the same interface that ORP
85 uses to its Garbage Collection system and reuse that GC system
86 instead of rolling our own, as the ORP system is pretty advanced
87 and is independent of the rest of ORP.
89 Although using a conservative garbage collector like Bohem's
90 would work, all the type information is available at runtime,
91 so we can actually implement a better collector than a
92 conservative collector.
95 * Garbage collection list and FAQ:<br>
96 <a href="http://www.iecc.com/gclist/">http://www.iecc.com/gclist/</a>
98 * "GC points in a Threaded Environment":<br>
99 <a href="http://research.sun.com/techrep/1998/abstract-70.html">
100 http://research.sun.com/techrep/1998/abstract-70.html</a>
102 * "A Generational Mostly-concurrent Garbage Collector":
103 <a href="http://research.sun.com/techrep/2000/abstract-88.html">
104 http://research.sun.com/techrep/2000/abstract-88.html</a>
106 * Details on The Microsoft .NET Garbage Collection Implementation:<br>
107 <a href="http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnmag00/html/GCI.asp">http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnmag00/html/GCI.asp</a>
108 <a href="http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnmag00/html/GCI2.asp">http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnmag00/html/GCI2.asp</a>
113 The ECMA runtime and the .NET runtime assume an IO model and a
114 threading model that is very similar to the Win32 API. Dick
115 Porter has been working on the Mono abstraction layer that allows
116 our runtime to execute code that depend on this behaviour.
120 Paolo Molaro found a few interesting links:
123 * On compilation of stack-based languages:<br>
124 <a href="http://www.complang.tuwien.ac.at/projects/rafts.html">
125 http://www.complang.tuwien.ac.at/projects/rafts.html</a>
127 * A paper on fast JIT compilation of a stack-based language:<br>
128 <a href="http://www.research.microsoft.com/~cwfraser/pldi99codegen.pdf">
129 http://www.research.microsoft.com/~cwfraser/pldi99codegen.pdf</a>
131 * Vmgen generates much of the code for efficient virtual machine (VM)
132 interpreters from simple descriptions of the VM instructions:<br>
133 <a href="http://www.complang.tuwien.ac.at/anton/vmgen/">
134 http://www.complang.tuwien.ac.at/anton/vmgen</a>
139 PInvoke is the mechanism we are using to wrap Unix API calls
140 as well as talking to system libraries.
142 We hvae implemented PInvoke through libffi, but we are likely
143 going to roll our own system as the runtime matures, specially
144 as the interpreter is approaching completion, and we move into