This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/doc/intro.html

Parent Directory Parent Directory | Revision Log Revision Log

Revision 34 - (hide annotations)
Mon Oct 8 16:21:17 2007 UTC (13 years, 3 months ago) by dpavlin
File MIME type: text/html
File size: 21443 byte(s)
++ trunk/HISTORY	(local)
$Id: HISTORY,v 1.1480 2007/02/19 01:34:42 debug Exp $
20061029	Changing usleep(1) calls in the debugger to usleep(10000)
20061107	Adding a new disk image option (-d o...) which sets the ISO9660
		filesystem base offset; also making some other hacks to allow
		NetBSD/dreamcast and homebrew demos/games to boot directly
		from a filesystem image.
		Moving Dreamcast-specific stuff in the documentation to its
		own page (dreamcast.html).
		Adding a border to the Dreamcast PVR framebuffer.
20061108	Adding a -T command line option (again?), for halting the
		emulator on unimplemented memory accesses.
20061109	Continuing on various SH4 and Dreamcast related things.
		The emulator should now halt on more unimplemented device
		accesses, instead of just printing a warning, forcing me to
		actually implement missing stuff :)
20061111	Continuing on SH4 and Dreamcast stuff.
		Adding a bogus Landisk (SH4) machine mode.
20061112	Implementing some parts of the Dreamcast GDROM device. With
		some ugly hacks, NetBSD can (barely) mount an ISO image.
20061113	NetBSD/dreamcast now starts booting from the Live CD image,
		but crashes randomly quite early on in the boot process.
20061122	Beginning on a skeleton interrupt.h and interrupt.c for the
		new interrupt subsystem.
20061124	Continuing on the new interrupt system; taking the first steps
		to attempt to connect CPUs (SuperH and MIPS) and devices
		(dev_cons and SH4 timer interrupts) to it. Many things will
		probably break from now on.
20061125	Converting dev_ns16550, dev_8253 to the new interrupt system.
		Attempting to begin to convert the ISA bus.
20061130	Incorporating a patch from Brian Foley for the configure
		script, which checks for X11 libs in /usr/X11R6/lib64 (which
		is used on some Linux systems).
20061227	Adding a note in the man page about booting from Dreamcast
		CDROM images (i.e. that no external kernel is needed).
20061229	Continuing on the interrupt system rewrite: beginning to
		convert more devices, adding abort() calls for legacy interrupt
		system calls so that everything now _has_ to be rewritten!
		Almost all machine modes are now completely broken.
20061230	More progress on removing old interrupt code, mostly related
		to the ISA bus + devices, the LCA bus (on AlphaBook1), and
		the Footbridge bus (for CATS). And some minor PCI stuff.
		Connecting the ARM cpu to the new interrupt system.
		The CATS, NetWinder, and QEMU_MIPS machine modes now work with
		the new interrupt system :)
20061231	Connecting PowerPC CPUs to the new interrupt system.
		Making PReP machines (IBM 6050) work again.
		Beginning to convert the GT PCI controller (for e.g. Malta
		and Cobalt emulation). Some things work, but not everything.
		Updating Copyright notices for 2007.
20070101	Converting dev_kn02 from legacy style to devinit; the 3max
		machine mode now works with the new interrupt system :-]
20070105	Beginning to convert the SGI O2 machine to the new interrupt
		system; finally converting O2 (IP32) devices to devinit, etc.
20070106	Continuing on the interrupt system redesign/rewrite; KN01
		(PMAX), KN230, and Dreamcast ASIC interrupts should work again,
		moving out stuff from machine.h and devices.h into the
		corresponding devices, beginning the rewrite of i80321
		interrupts, etc.
20070107	Beginning on the rewrite of Eagle interrupt stuff (PReP, etc).
20070117	Beginning the rewrite of Algor (V3) interrupts (finally
		changing dev_v3 into devinit style).
20070118	Removing the "bus" registry concept from machine.h, because
		it was practically meaningless.
		Continuing on the rewrite of Algor V3 ISA interrupts.
20070121	More work on Algor interrupts; they are now working again,
		well enough to run NetBSD/algor. :-)
20070122	Converting VR41xx (HPCmips) interrupts. NetBSD/hpcmips
		can be installed using the new interrupt system :-)
20070123	Making the testmips mode work with the new interrupt system.
20070127	Beginning to convert DEC5800 devices to devinit, and to the
		new interrupt system.
		Converting Playstation 2 devices to devinit, and converting
		the interrupt system. Also fixing a severe bug: the interrupt
		mask register on Playstation 2 is bitwise _toggled_ on writes.
20070128	Removing the dummy NetGear machine mode and the 8250 device
		(which was only used by the NetGear machine).
		Beginning to convert the MacPPC GC (Grand Central) interrupt
		controller to the new interrupt system.
		Converting Jazz interrupts (PICA61 etc.) to the new interrupt
		system. NetBSD/arc can be installed again :-)
		Fixing the JAZZ timer (hardcoding it at 100 Hz, works with
		NetBSD and it is better than a completely dummy timer as it
		was before).
		Converting dev_mp to the new interrupt system, although I
		haven't had time to actually test it yet.
		Completely removing src/machines/interrupts.c, cpu_interrupt
		and cpu_interrupt_ack in src/cpu.c, and
		src/include/machine_interrupts.h! Adding fatal error messages
		+ abort() in the few places that are left to fix.
		Converting dev_z8530 to the new interrupt system.
		FINALLY removing the md_int struct completely from the
		machine struct.
		SH4 fixes (adding a PADDR invalidation in the ITLB replacement
		code in memory_sh.c); the NetBSD/dreamcast LiveCD now runs
		all the way to the login prompt, and can be interacted with :-)
		Converting the CPC700 controller (PCI and interrupt controller
		for PM/PPC) to the new interrupt system.
20070129	Fixing MACE ISA interrupts (SGI IP32 emulation). Both NetBSD/
		sgimips' and OpenBSD/sgi's ramdisk kernels can now be
		interacted with again.
20070130	Moving out the MIPS multi_lw and _sw instruction combinations
		so that they are auto-generated at compile time instead.
20070131	Adding detection of amd64/x86_64 hosts in the configure script,
		for doing initial experiments (again :-) with native code
		Adding a -k command line option to set the size of the dyntrans
		cache, and a -B command line option to disable native code
		generation, even if GXemul was compiled with support for
		native code generation for the specific host CPU architecture.
20070201	Experimenting with a skeleton for native code generation.
		Changing the default behaviour, so that native code generation
		is now disabled by default, and has to be enabled by using
		-b on the command line.
20070202	Continuing the native code generation experiments.
		Making PCI interrupts work for Footbridge again.
20070203	More native code generation experiments.
		Removing most of the native code generation experimental code,
		it does not make sense to include any quick hacks like this.
		Minor cleanup/removal of some more legacy MIPS interrupt code.
20070204	Making i80321 interrupts work again (for NetBSD/evbarm etc.),
		and fixing the timer at 100 Hz.
20070206	Experimenting with removing the wdc interrupt slowness hack.
20070207	Lowering the number of dyntrans TLB entries for MIPS from
		192 to 128, resulting in a minor speed improvement.
		Minor optimization to the code invalidation routine in
20070208	Increasing (experimentally) the nr of dyntrans instructions per
		loop from 60 to 120.
20070210	Commenting out (experimentally) the dyntrans_device_danger
		detection in memory_rw.c.
		Changing the testmips and baremips machines to use a revision 2
		MIPS64 CPU by default, instead of revision 1.
		Removing the dummy i960, IA64, x86, AVR32, and HP PA-RISC
		files, the PC bios emulation, and the Olivetti M700 (ARC) and
		db64360 emulation modes.
20070211	Adding an "mp" demo to the demos directory, which tests the
		SMP functionality of the testmips machine.
		Fixing PReP interrupts some more. NetBSD/prep now boots again.
20070216	Adding a "nop workaround" for booting Mach/PMAX to the
		documentation; thanks to Artur Bujdoso for the values.
		Converting more of the MacPPC interrupt stuff to the new
		Beginning to convert BeBox interrupts to the new system.
		PPC603e should NOT have the PPC_NO_DEC flag! Removing it.
		Correcting BeBox clock speed (it was set to 100 in the NetBSD
		bootinfo block, but should be 33000000/4), allowing NetBSD
		to start without using the (incorrect) PPC_NO_DEC hack.
20070217	Implementing (slow) AltiVec vector loads and stores, allowing
		NetBSD/macppc to finally boot using the GENERIC kernel :-)
		Updating the documentation with install instructions for
20070218-19	Regression testing for the release.

==============  RELEASE 0.4.4  ==============

1 dpavlin 12 <html><head><title>Gavare's eXperimental Emulator:&nbsp;&nbsp;&nbsp;Introduction</title>
2     <meta name="robots" content="noarchive,nofollow,noindex"></head>
3 dpavlin 4 <body bgcolor="#f8f8f8" text="#000000" link="#4040f0" vlink="#404040" alink="#ff0000">
4     <table border=0 width=100% bgcolor="#d0d0d0"><tr>
5     <td width=100% align=center valign=center><table border=0 width=100%><tr>
6     <td align="left" valign=center bgcolor="#d0efff"><font color="#6060e0" size="6">
7 dpavlin 22 <b>Gavare's eXperimental Emulator:</b></font><br>
8 dpavlin 4 <font color="#000000" size="6"><b>Introduction</b>
9     </font></td></tr></table></td></tr></table><p>
10 dpavlin 2
11     <!--
13 dpavlin 34 $Id: intro.html,v 1.106 2007/02/18 09:19:47 debug Exp $
14 dpavlin 2
15 dpavlin 34 Copyright (C) 2003-2007 Anders Gavare. All rights reserved.
16 dpavlin 2
17     Redistribution and use in source and binary forms, with or without
18     modification, are permitted provided that the following conditions are met:
20     1. Redistributions of source code must retain the above copyright
21     notice, this list of conditions and the following disclaimer.
22     2. Redistributions in binary form must reproduce the above copyright
23     notice, this list of conditions and the following disclaimer in the
24     documentation and/or other materials provided with the distribution.
25     3. The name of the author may not be used to endorse or promote products
26     derived from this software without specific prior written permission.
40     -->
42     <a href="./">Back to the index</a>
44     <p><br>
45     <h2>Introduction</h2>
47     <p>
48 dpavlin 22 <table border="0" width="99%"><tr><td valign="top" align="left">
49 dpavlin 2 <ul>
50     <li><a href="#overview">Overview</a>
51 dpavlin 4 <li><a href="#free">Is GXemul Free software?</a>
52 dpavlin 2 <li><a href="#build">How to compile/build the emulator</a>
53 dpavlin 6 <li><a href="#run">How to run the emulator</a>
54 dpavlin 22 <li><a href="#cpus">Which processor architectures does GXemul emulate?</a>
55 dpavlin 24 <li><a href="#hosts">Which host architectures are supported?</a>
56     <li><a href="#translation">What kind of translation does GXemul use?</a>
57 dpavlin 2 <li><a href="#accuracy">Emulation accuracy</a>
58     <li><a href="#emulmodes">Which machines does GXemul emulate?</a>
59     </ul>
60 dpavlin 22 </td><td valign="center" align="center">
61     <a href="20050317-example.png"><img src="20050317-example_small.png"></a>
62     <p>NetBSD/pmax 1.6.2 with X11<br>running in GXemul</td></tr></table>
63 dpavlin 2
67     <p><br>
68     <a name="overview"></a>
69     <h3>Overview:</h3>
71 dpavlin 14 GXemul is an experimental instruction-level machine emulator. Several
72     emulation modes are available. In some modes, processors and surrounding
73     hardware components are emulated well enough to let unmodified operating
74     systems (e.g. NetBSD) run as if they were running on a real machine.
75 dpavlin 2
76 dpavlin 32 <p>Devices and processors are not simulated with 100% accuracy. They are
77     only ``faked'' well enough to allow guest operating systems to run without
78     complaining too much. Still, the emulator could be of interest for
79     academic research and experiments, such as when learning how to write
80     operating system code.
81 dpavlin 12
82 dpavlin 22 <p>The emulator is written in C, does not depend on third-party libraries,
83     and should compile and run on most 64-bit and 32-bit Unix-like systems.
84 dpavlin 2
85 dpavlin 10 <p>The emulator contains code which tries to emulate the workings of CPUs
86     and surrounding hardware found in real machines, but it does not contain
87     any ROM code. You will need some form of program (in binary form) to run
88     in the emulator. For many emulation modes, PROM calls are handled by the
89 dpavlin 2 emulator itself, so you do not need to use any ROM image at all.
91 dpavlin 10 <p>You can use pre-compiled kernels (for example NetBSD kernels, or
92     Linux), or other programs that are in binary format, and in some cases
93     even actual ROM images. A couple of different file formats are supported
94     (ELF, a.out, ECOFF, SREC, and raw binaries).
95 dpavlin 2
96 dpavlin 10 <p>If you do not have a kernel as a separate file, but you have a bootable
97 dpavlin 6 disk image, then it is sometimes possible to boot directly from that
98 dpavlin 34 image. (This works for example with DECstation emulation, Dreamcast
99     emulation, or when booting from generic ISO9660 CDROM images if the
100     kernel is included in the image as a plain file.)
101 dpavlin 2
102 dpavlin 32 <p>Thanks to (in no specific order) Joachim Buss, Olivier Houchard, Juli
103     Mallett, Juan Romero Pardines, Alec Voropay, Göran Weinholt, Alexander
104     Yurchenko, and everyone else who has provided me with feedback.
105 dpavlin 2
108 dpavlin 6
110 dpavlin 10
112 dpavlin 2 <p><br>
113     <a name="free"></a>
114 dpavlin 4 <h3>Is GXemul Free software?</h3>
115 dpavlin 2
116 dpavlin 6 Yes. I have released GXemul under a Free license. The code in GXemul is
117     Copyrighted software, it is <i>not</i> public domain. (If this is
118     confusing to you, you might want to read up on the definitions of the
119     four freedoms associated with Free software, <a
120     href="http://www.gnu.org/philosophy/free-sw.html">http://www.gnu.org/philosophy/free-sw.html</a>.)
121 dpavlin 2
122 dpavlin 12 <p>The code I have written is released under a 3-clause BSD-style license
123     (or "revised BSD-style" if one wants to use <a
124     href="http://www.gnu.org/philosophy/bsd.html">GNU jargon</a>). Apart from
125     the code I have written, some files are copied from other sources such as
126     NetBSD, for example header files containing symbolic names of bitfields in
127     device registers. They are also covered by similar licenses, but with some
128     additional clauses. The main point, however, is that the licenses require
129     that the original Copyright and license terms are included when you make a
130     copy or modification.
131 dpavlin 2
132 dpavlin 12 <p>If you plan to redistribute GXemul <i>without</i> supplying the source
133     code, then you need to comply with each individual source file some other
134     way, for example by writing additional documentation containing copyright
135     notes. I have not done this, since I do not plan on making distributions
136     without source code. You need to check all individual files for details.
137     The "easiest way out" if you plan to redistribute code from GXemul is, of
138     course, to let it remain open source and simply supply the source code.
139 dpavlin 2
140 dpavlin 22 <p>In case you want to reuse parts of GXemul, but you need to do that
141     under a different license (e.g. the GPL), then contact me and I might
142     re-license/dual-license files on a case-by-case basis.
143 dpavlin 2
147 dpavlin 12
148 dpavlin 2 <p><br>
149     <a name="build"></a>
150     <h3>How to compile/build the emulator:</h3>
152     Uncompress the .tar.gz distribution file, and run
153     <pre>
154     $ <b>./configure</b>
155     $ <b>make</b>
156     </pre>
158 dpavlin 22 <p>This should work on most Unix-like systems. GXemul does not require any
159     specific libraries to build, however, if you build on a system which does
160     not have X11 libraries installed, some functionality will be lost.
161 dpavlin 2
162 dpavlin 12 <p>The emulator's performance is highly dependent on both runtime settings
163 dpavlin 2 and on compiler settings, so you might want to experiment with different
164 dpavlin 20 CC and CFLAGS environment variable values. For example, on an AMD Athlon
165 dpavlin 24 host, you might want to try setting <tt>CFLAGS</tt> to <tt>-march=athlon</tt>
166     before running <tt>configure</tt>.
167 dpavlin 2
169 dpavlin 6
174     <p><br>
175     <a name="run"></a>
176     <h3>How to run the emulator:</h3>
178     Once you have built GXemul, running it should be rather straight-forward.
179     Running <tt><b>gxemul</b></tt> without arguments (or with the
180     <b><tt>-h</tt></b> or <b><tt>-H</tt></b> command line options) will
181     display a help message.
183 dpavlin 2 <p>
184 dpavlin 6 To get some ideas about what is possible to run in the emulator, please
185     read the section about <a href="guestoses.html">installing "guest"
186     operating systems</a>. If you are interested in using the emulator to
187     develop code on your own, then you should also read the section about
188     <a href="experiments.html#hello">Hello World</a>.
190     <p>
191 dpavlin 2 To exit the emulator, type CTRL-C to enter the
192 dpavlin 6 single-step debugger, and then type <tt><b>quit</b></tt>.
193 dpavlin 2
194 dpavlin 4 <p>
195     If you are starting an emulation by entering settings directly on the
196 dpavlin 6 command line, and you are not using the <tt><b>-x</b></tt> option, then all
197 dpavlin 4 terminal input and output will go to the main controlling terminal.
198     CTRL-C is used to break into the debugger, so in order to send CTRL-C to
199     the running (emulated) program, you may use CTRL-B.
200 dpavlin 6 (This should be a reasonable compromise to allow the emulator to be usable
201     even on systems without X Windows.)
202 dpavlin 2
203 dpavlin 4 <p>
204 dpavlin 6 There is no way to send an actual CTRL-B to the emulated program, when
205     typing in the main controlling terminal window. The solution is to either
206     use <a href="configfiles.html">configuration files</a>, or use
207     <tt><b>-x</b></tt>. Both these solutions cause new xterms to be opened for
208     each emulated serial port that is written to. CTRL-B and CTRL-C both have
209     their original meaning in those xterm windows.
210 dpavlin 2
214 dpavlin 4
215 dpavlin 2 <p><br>
216     <a name="cpus"></a>
217 dpavlin 22 <h3>Which processor architectures does GXemul emulate?</h3>
218 dpavlin 2
219 dpavlin 24 The architectures that are emulated well enough to let at least one
220 dpavlin 32 guest operating system run (per architecture) are ARM, MIPS, PowerPC,
221     and SuperH.
222 dpavlin 2
223 dpavlin 34 <p>Please read the page about <a href="guestoses.html">guest operating
224     systems</a> for more information about the machines and operating systems
225     that can be considered "working" in the emulator.
226 dpavlin 2
229 dpavlin 14
230 dpavlin 32
231 dpavlin 34
232 dpavlin 24 <p><br>
233     <a name="hosts"></a>
234     <h3>Which host architectures are supported?</h3>
235 dpavlin 2
236 dpavlin 32 GXemul should compile and run on any modern host architecture (64-bit or
237     32-bit word-length).
238 dpavlin 2
239 dpavlin 34 <p>Note: The dynamic translation engine does <i>not</i> require backends
240     for native code generation to be written for each individual host
241     architecture; the "intermediate representation" that the dyntrans system
242     uses can be executed on any host architecture.
243 dpavlin 6
244 dpavlin 2
247 dpavlin 32
248 dpavlin 24 <p><br>
249     <a name="translation"></a>
250     <h3>What kind of translation does GXemul use?</h3>
251 dpavlin 2
252 dpavlin 24 <b>Static vs. dynamic:</b>
254     <p>In order to support guest operating systems, which can overwrite old
255     code pages in memory with new code, it is necessary to translate code
256     dynamically. It is not possible to do a "one-pass" (static) translation.
257     Self-modifying code and Just-in-Time compilers running inside
258     the emulator are other things that would not work with a static
259     translator. GXemul is a dynamic translator. However, it does not
260     necessarily translate into native code, like many other emulators.
262     <p><b>"Runnable" Intermediate Representation:</b>
264     <p>Dynamic translators usually translate from the emulated architecture
265     (e.g. MIPS) into a kind of <i>intermediate representation</i> (IR), and then
266     to native code (e.g. AMD64 or x86 code). Since one of my main goals for
267     GXemul is to keep everything as portable as possible, I have tried to make
268     sure that the IR is something which can be executed regardless of whether
269     the final step (translation from IR to native code) has been implemented
270     or not.
272     <p>The IR in GXemul consists of arrays of pointers to functions, and a few
273     arguments which are passed along to those functions. The functions are
274     implemented in either manually hand-coded C, or automatically generated C.
275     In any case, this is all statically linked into the GXemul binary at link
276     time.
278     <p>Here is a simplified diagram of how these arrays work.
280     <p><center><img src="simplified_dyntrans.png"></center>
282     <p>There is one instruction call slot for every possible program counter
283     location. In the MIPS case, instruction words are 32 bits in length,
284     and pages are (usually) 4 KB large, resulting in 1024 instruction call
285     slots. After the last of these instruction calls, there is an additional
286     call to a special "end of page" function (which doesn't count as an executed
287     instruction). This function switches to the first instruction
288     on the next virtual page (which might cause exceptions, etc).
290     <p>The complexity of individual instructions vary. A simple example of
291     what an instruction can look like is the MIPS <tt>addiu</tt> instruction:
292     <pre>
293     X(addiu)
294     {
295     reg(ic->arg[1]) = (int32_t)
296     ((int32_t)reg(ic->arg[0]) + (int32_t)ic->arg[2]);
297     }
298     </pre>
300     <p>It stores the result of a 32-bit addition of the register at arg[0]
301     with the immediate value arg[2] (treating both as signed 32-bit
302     integers) into register arg[1]. If the emulated CPU is a 64-bit CPU,
303     then this will store a correctly sign-extended value into arg[1].
304     If it is a 32-bit CPU, then only the lowest 32 bits will be stored,
305     and the high part ignored. <tt>X(addiu)</tt> is expanded to
306     <tt>mips_instr_addiu</tt> in the 64-bit case, and <tt>mips32_instr_addiu</tt>
307     in the 32-bit case. Both are compiled into the GXemul executable; no code
308     is created during run-time.
310     <p>Here are examples of what the <tt>addiu</tt> instruction actually
311     looks like when it is compiled, on various host architectures:
313     <p><center><table border="0">
314     <tr><td><b>GCC 4.0.1 on Alpha:</b></td>
315     <td width="35"></td><td></td>
316     <tr>
317     <td valign="top">
318     <pre>mips_instr_addiu:
319     ldq t1,8(a1)
320     ldq t2,24(a1)
321     ldq t3,16(a1)
322     ldq t0,0(t1)
323     addl t0,t2,t0
324     stq t0,0(t3)
325     ret</pre>
326     </td>
327     <td></td>
328     <td valign="top">
329     <pre>mips32_instr_addiu:
330     ldq t2,8(a1)
331     ldq t0,24(a1)
332     ldq t3,16(a1)
333     ldl t1,0(t2)
334     addq t0,t1,t0
335     stl t0,0(t3)
336     ret</pre>
337     </td>
338     </tr>
340     <tr><td><b><br>GCC 3.4.4 on AMD64:</b></td>
341     <tr>
342     <td valign="top">
343     <pre>mips_instr_addiu:
344     mov 0x8(%rsi),%rdx
345     mov 0x18(%rsi),%rax
346     mov 0x10(%rsi),%rcx
347     add (%rdx),%eax
348     cltq
349     mov %rax,(%rcx)
350     retq</pre>
351     </td>
352     <td></td>
353     <td valign="top">
354     <pre>mips32_instr_addiu:
355     mov 0x8(%rsi),%rcx
356     mov 0x10(%rsi),%rdx
357     mov (%rcx),%eax
358     add 0x18(%rsi),%eax
359     mov %eax,(%rdx)
360     retq</pre>
361     </td>
362     </tr>
364     <tr><td><b><br>GCC 4.0.1 on i386:</b></td>
365     <tr>
366     <td valign="top">
367     <pre>mips_instr_addiu:
368     mov 0x8(%esp),%eax
369     mov 0x8(%eax),%ecx
370     mov 0x4(%eax),%edx
371     mov 0xc(%eax),%eax
372     add (%edx),%eax
373     mov %eax,(%ecx)
374     cltd
375     mov %edx,0x4(%ecx)
376     ret</pre>
377     </td>
378     <td></td>
379     <td valign="top">
380     <pre>mips32_instr_addiu:
381     mov 0x8(%esp),%eax
382     mov 0x8(%eax),%ecx
383     mov 0x4(%eax),%edx
384     mov 0xc(%eax),%eax
385     add (%edx),%eax
386     mov %eax,(%ecx)
387     ret</pre>
388     </td>
389     </tr>
390     </table></center>
392     <p>On 64-bit hosts, there is not much difference, but on 32-bit hosts (and
393     to some extent on AMD64), the difference is enough to make it worthwhile.
396     <p><b>Performance:</b>
398     <p>The performance of using this kind of runnable IR is obviously lower
399     than what can be achieved by emulators using native code generation, but
400     can be significantly higher than using a naive fetch-decode-execute
401     interpretation loop. In my opinion, using a runnable IR is an interesting
402     compromise.
404     <p>The overhead per emulated instruction is usually around or below
405     approximately 10 host instructions. This is very much dependent on your
406     host architecture and what compiler and compiler switches you are using.
407     Added to this instruction count is (of course) also the C code used to
408     implement each specific instruction.
410     <p><b>Instruction Combinations:</b>
412     <p>Short, common instruction sequences can sometimes be replaced by a
413     "compound" instruction. An example could be a compare instruction followed
414     by a conditional branch instruction. The advantages of instruction
415     combinations are that
416     <ul>
417     <li>the amortized overhead per instruction is slightly reduced, and
418     <p>
419     <li>the host's compiler can make a good job at optimizing the common
420     instruction sequence.
421     </ul>
423     <p>The special cases where instruction combinations give the most gain
424     are in the cores of string/memory manipulation functions such as
425     <tt>memset()</tt> or <tt>strlen()</tt>. The core loop can then (at least
426     to some extent) be replaced by a native call to the equivalent function.
428     <p>The implementations of compound instructions still keep track of the
429     number of executed instructions, etc. When single-stepping, these
430     translations are invalidated, and replaced by normal instruction calls
431     (one per emulated instruction).
433 dpavlin 34 <p><b>Native Code Back-ends:</b>
434 dpavlin 24
435 dpavlin 34 <p>In theory, it will be possible to implement native code generation,
436     similar to what is used in high-performance emulators such as QEMU,
437     as long as that generated code abides to the C ABI on the host.
438 dpavlin 24
439 dpavlin 34 <p>However, since I wanted to make sure that GXemul works without such
440     native code back-ends, there are no implemented backends in this release.
441 dpavlin 24
442 dpavlin 34 <p>(There is a place-holder in the source code for native code generation,
443     which can be used for experiments, but it does not contain any working
444     code at the moment.)
445 dpavlin 24
449 dpavlin 34
451 dpavlin 2 <p><br>
452     <a name="accuracy"></a>
453     <h3>Emulation accuracy:</h3>
455 dpavlin 6 GXemul is an instruction-level emulator; things that would happen in
456 dpavlin 24 several steps within a real CPU are not taken into account (e.g. pipe-line
457 dpavlin 6 stalls or out-of-order execution). Still, instruction-level accuracy seems
458     to be enough to be able to run complete guest operating systems inside the
459 dpavlin 2 emulator.
461 dpavlin 24 <p>The existance of instruction and data caches is "faked" to let
462     operating systems think that they are there, but for all practical
463     purposes, these caches are non-working.
464 dpavlin 2
465 dpavlin 32 <p>The emulator is in general <i>not</i> timing-accurate, neither at the
466     instruction level nor on any higher level. An attempt is made to let
467     emulated clocks run at the same speed as the host (i.e. an emulated timer
468     running at 100 Hz will interrupt around 100 times per real second), but
469     since the host speed may vary, e.g. because of other running processes,
470     there is no guarantee as to how many instructions will be executed in
471     each of these 100 Hz cycles.
472 dpavlin 2
473 dpavlin 32 <p>If the host is very slow, the emulated clocks might even lag behind
474     the real-world clock.
475 dpavlin 2
477 dpavlin 6
479 dpavlin 20
481 dpavlin 2 <p><br>
482     <a name="emulmodes"></a>
483     <h3>Which machines does GXemul emulate?</h3>
485 dpavlin 4 A few different machine types are emulated. The following machine types
486     are emulated well enough to run at least one "guest OS":
487 dpavlin 2
488     <p>
489     <ul>
490 dpavlin 24 <li><b><u>ARM</u></b>
491 dpavlin 14 <ul>
492 dpavlin 28 <li><b>CATS</b> (<a href="guestoses.html#netbsdcatsinstall">NetBSD/cats</a>,
493     <a href="guestoses.html#openbsdcatsinstall">OpenBSD/cats</a>)
494     <li><b>IQ80321</b> (<a href="guestoses.html#netbsdevbarminstall">NetBSD/evbarm</a>)
495 dpavlin 30 <li><b>NetWinder</b> (<a href="guestoses.html#netbsdnetwinderinstall">NetBSD/netwinder</a>)
496 dpavlin 14 </ul>
497     <p>
498 dpavlin 24 <li><b><u>MIPS</u></b>
499 dpavlin 14 <ul>
500 dpavlin 28 <li><b>DECstation 5000/200</b> (<a href="guestoses.html#netbsdpmaxinstall">NetBSD/pmax</a>,
501     <a href="guestoses.html#openbsdpmaxinstall">OpenBSD/pmax</a>,
502     <a href="guestoses.html#ultrixinstall">Ultrix</a>,
503     <a href="guestoses.html#declinux">Linux/DECstation</a>,
504     <a href="guestoses.html#sprite">Sprite</a>)
505     <li><b>Acer Pica-61</b> (<a href="guestoses.html#netbsdarcinstall">NetBSD/arc</a>)
506 dpavlin 32 <li><b>NEC MobilePro 770, 780, 800, 880</b> (<a href="guestoses.html#netbsdhpcmipsinstall">NetBSD/hpcmips</a>)
507 dpavlin 28 <li><b>Cobalt</b> (<a href="guestoses.html#netbsdcobaltinstall">NetBSD/cobalt</a>)
508 dpavlin 34 <li><b>Malta</b> (<a href="guestoses.html#netbsdevbmipsinstall">NetBSD/evbmips</a>, Linux/Malta <font color="#0000e0">(<super>*1</super>)</font>)
509 dpavlin 30 <li><b>Algorithmics P5064</b> (<a href="guestoses.html#netbsdalgorinstall">NetBSD/algor</a>)
510 dpavlin 34 <li><b>SGI O2 (aka IP32)</b> <font color="#0000e0">(<super>*2</super>)</font>
511 dpavlin 28 (<a href="guestoses.html#netbsdsgimips">NetBSD/sgi</a>)
512 dpavlin 14 </ul>
513 dpavlin 20 <p>
514     <li><b><u>PowerPC</u></b>
515     <ul>
516 dpavlin 28 <li><b>IBM 6050/6070 (PReP, PowerPC Reference Platform)</b> (<a href="guestoses.html#netbsdprepinstall">NetBSD/prep</a>)
517 dpavlin 34 <li><b>MacPPC (generic "G4" Macintosh)</b> (<a href="guestoses.html#netbsdmacppcinstall">NetBSD/macppc</a>)
518 dpavlin 20 </ul>
519 dpavlin 32 <p>
520     <li><b><u>SuperH</u></b>
521     <ul>
522 dpavlin 34 <li><b>Sega Dreamcast</b> (<a href="dreamcast.html#netbsd_generic_md">NetBSD/dreamcast</a>)
523 dpavlin 32 </ul>
524 dpavlin 2 </ul>
526 dpavlin 32 <p>
527     <small><font color="#0000e0">(<super>*1</super>)</font> =
528 dpavlin 34 Linux/Malta may be run as a guest OS, however I have not yet found any stable
529     URL to pre-compiled Linux/Malta kernels. Thus, Linux/Malta emulation is not
530     tested for every release of the emulator; sometimes it works, sometimes
531     it doesn't.</small>
533 dpavlin 32 <br><small><font color="#0000e0">(<super>*2</super>)</font> =
534 dpavlin 34 SGI O2 emulation is enough for root-on-nfs, but not for disk boot.</small>
535 dpavlin 22
536 dpavlin 34
537 dpavlin 10 <p>There is code in GXemul for emulation of many other machine types; the
538     degree to which these work range from almost being able to run a complete
539     OS, to almost completely unsupported (perhaps just enough support to
540     output a few boot messages via serial console).
541 dpavlin 2
542 dpavlin 10 <p>In addition to emulating real machines, there is also a "test-machine".
543     A test-machine consists of one or more CPUs and a few experimental devices
544     such as:
545 dpavlin 2
546     <p>
547     <ul>
548     <li>a console I/O device (putchar() and getchar()...)
549     <li>an inter-processor communication device, for SMP experiments
550     <li>a very simple linear framebuffer device (for graphics output)
551 dpavlin 32 <li>a simple disk controller
552 dpavlin 12 <li>a simple ethernet controller
553 dpavlin 32 <li>a real-time clock device
554 dpavlin 2 </ul>
556 dpavlin 10 <p>This mode is useful if you wish to run experimental code, but do not
557 dpavlin 2 wish to target any specific real-world machine type, for example for
558     educational purposes.
560 dpavlin 10 <p>You can read more about these experimental devices <a
561     href="experiments.html#expdevices">here</a>.
562 dpavlin 2
568     </body>
569     </html>

  ViewVC Help
Powered by ViewVC 1.1.26