1 |
<html><head><title>GXemul documentation: Technical details</title> |
2 |
<meta name="robots" content="noarchive,nofollow,noindex"></head> |
3 |
<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 |
<b>GXemul documentation:</b></font> |
8 |
<font color="#000000" size="6"><b>Technical details</b> |
9 |
</font></td></tr></table></td></tr></table><p> |
10 |
|
11 |
<!-- |
12 |
|
13 |
$Id: technical.html,v 1.51 2005/06/04 22:47:49 debug Exp $ |
14 |
|
15 |
Copyright (C) 2004-2005 Anders Gavare. All rights reserved. |
16 |
|
17 |
Redistribution and use in source and binary forms, with or without |
18 |
modification, are permitted provided that the following conditions are met: |
19 |
|
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. |
27 |
|
28 |
THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND |
29 |
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
30 |
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
31 |
ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE |
32 |
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
33 |
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
34 |
OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
35 |
HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
36 |
LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
37 |
OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
38 |
SUCH DAMAGE. |
39 |
|
40 |
--> |
41 |
|
42 |
|
43 |
<a href="./">Back to the index</a> |
44 |
|
45 |
<p><br> |
46 |
<h2>Technical details</h2> |
47 |
|
48 |
<p> |
49 |
This page describes some of the internals of GXemul. |
50 |
|
51 |
<p> |
52 |
<font color="#e00000"><b>NOTE: This page is probably not |
53 |
very up-to-date by now.</b></font> |
54 |
|
55 |
<p> |
56 |
<ul> |
57 |
<li><a href="#overview">Overview</a> |
58 |
<li><a href="#speed">Speed</a> |
59 |
<li><a href="#net">Networking</a> |
60 |
<li><a href="#devices">Emulation of hardware devices</a> |
61 |
<li><a href="#regtest">Regression tests</a> |
62 |
</ul> |
63 |
|
64 |
|
65 |
|
66 |
|
67 |
<p><br> |
68 |
<a name="overview"></a> |
69 |
<h3>Overview</h3> |
70 |
|
71 |
In simple terms, GXemul is just a simple fetch-and-execute |
72 |
loop; an instruction is fetched from memory, and executed. |
73 |
|
74 |
<p> |
75 |
In reality, a lot of things need to be handled. Before each instruction is |
76 |
executed, the emulator checks to see if any interrupts are asserted which |
77 |
are not masked away. If so, then an INT exception is generated. Exceptions |
78 |
cause the program counter to be set to a specific value, and some of the |
79 |
system coprocessor's registers to be set to values signifying what kind of |
80 |
exception it was (an interrupt exception in this case). |
81 |
|
82 |
<p> |
83 |
Reading instructions from memory is done through a TLB, a translation |
84 |
lookaside buffer. The TLB on MIPS is software controlled, which means that |
85 |
the program running inside the emulator (for example an operating system |
86 |
kernel) has to take care of manually updating the TLB. Some memory |
87 |
addresses are translated into physical addresses directly, some are |
88 |
translated into valid physical addresses via the TLB, and some memory |
89 |
references are not valid. Invalid memory references cause exceptions. |
90 |
|
91 |
<p> |
92 |
After an instruction has been read from memory, the emulator checks which |
93 |
opcode it contains and executes the instruction. Executing an instruction |
94 |
usually involves reading some register and writing some register, or perhaps a |
95 |
load from memory (or a store to memory). The program counter is increased |
96 |
for every instruction. |
97 |
|
98 |
<p> |
99 |
Some memory references point to physical addresses which are not in the |
100 |
normal RAM address space. They may point to hardware devices. If that is |
101 |
the case, then loads and stores are converted into calls to a device |
102 |
access function. The device access function is then responsible for |
103 |
handling these reads and writes. For example, a graphical framebuffer |
104 |
device may put a pixel on the screen when a value is written to it, or a |
105 |
serial controller device may output a character to stdout when written to. |
106 |
|
107 |
|
108 |
|
109 |
|
110 |
<p><br> |
111 |
<a name="speed"></a> |
112 |
<h3>Speed</h3> |
113 |
|
114 |
There are two modes in which the emulator can run, <b>a</b>) a straight forward |
115 |
loop which fetches one instruction from emulated RAM and executes it |
116 |
(described in the previous section), and <b>b</b>) |
117 |
using dynamic binary translation. |
118 |
|
119 |
<p> |
120 |
Mode <b>a</b> is very slow. On a 2.8 GHz Intel Xeon host the resulting |
121 |
emulated machine is rougly equal to a 7 MHz R3000 (or a 3.5 MHz R4000). |
122 |
The actual performance varies a lot, maybe between 5 and 10 million |
123 |
instructions per second, depending on workload. |
124 |
|
125 |
<p> |
126 |
Mode <b>b</b> ("bintrans") is still to be considered experimental, but |
127 |
gives higher performance than mode <b>a</b>. It translates MIPS machine |
128 |
code into machine code that can be executed on the host machine |
129 |
on-the-fly. The translation itself obviously takes some time, but this is |
130 |
usually made up for by the fact that the translated code chunks are |
131 |
executed multiple times. |
132 |
To run the emulator with binary translation enabled, just add |
133 |
<tt><b>-b</b></tt> to the command line. |
134 |
|
135 |
<p> |
136 |
Only small pieces of MIPS machine code are translated, usually the size of |
137 |
a function, or less. There is no "intermediate representation" code, so |
138 |
all translations are done directly from MIPS to host machine code. |
139 |
|
140 |
<p> |
141 |
The default bintrans cache size is 16 MB, but you can change this by adding |
142 |
<tt>-DDEFAULT_BINTRANS_SIZE_IN_MB=<i>xx</i></tt> to your CFLAGS environment |
143 |
variable before running the configure script, or by using the |
144 |
<tt>bintrans_size()</tt> configuration file option when running the emulator. |
145 |
|
146 |
<p> |
147 |
By default, an emulated OS running under DECstation emulation which listens to |
148 |
interrupts from the mc146818 clock will get interrupts that are close to the |
149 |
host's clock. That is, if the emulated OS says it wants 100 interrupts per |
150 |
second, it will get approximately 100 interrupts per real second. |
151 |
|
152 |
<p> |
153 |
There is however a <tt><b>-I</b></tt> option, which sets the number of |
154 |
emulated cycles per seconds to a fixed value. Let's say you wish to make the |
155 |
emulated OS think it is running on a 40 MHz DECstation, and not a 7 MHz one, |
156 |
then you can add <tt><b>-I 40000000</b></tt> to the command line. This will not |
157 |
make the emulation faster, of course. It might even make it seem slower; for |
158 |
example, if NetBSD/pmax waits 2 seconds for SCSI devices to settle during |
159 |
bootup, those 2 seconds will take 2*40000000 cycles (which will take more |
160 |
time than 2*7000000). |
161 |
|
162 |
<p> |
163 |
The <b><tt>-I</tt></b> option is also necessary if you want to run |
164 |
deterministic experiments, if a mc146818 (or similar) device is present. |
165 |
|
166 |
<p> |
167 |
Some emulators make claims such as "x times slowdown," but in the case of |
168 |
GXemul, the host is often not a MIPS-based machine, and hence comparing |
169 |
one MIPS instruction to a host instruction doesn't work. Performance depends on |
170 |
a lot of factors, including (but not limited to) host architecture, host speed, |
171 |
which compiler and compiler flags were used to build GXemul, what the |
172 |
workload is, and so on. For example, if an emulated operating system tries |
173 |
to read a block from disk, from its point of view the read was instantaneous |
174 |
(no waiting). So 1 MIPS in an emulated OS might have taken more than one |
175 |
million instructions on a real machine. Because of this, imho it is best |
176 |
to measure performance as the actual (real-world) time it takes to perform |
177 |
a task with the emulator. |
178 |
|
179 |
|
180 |
|
181 |
|
182 |
<p><br> |
183 |
<a name="net"></a> |
184 |
<h3>Networking</h3> |
185 |
|
186 |
Running an entire operating system under emulation is very interesting in |
187 |
itself, but for several reasons, running a modern OS without access to |
188 |
TCP/IP networking is a bit akward. Hence, I feel the need to implement TCP/IP |
189 |
(networking) support in the emulator. |
190 |
|
191 |
<p> |
192 |
As far as I have understood it, there seems to be two different ways to go: |
193 |
|
194 |
<ol> |
195 |
<li>Forward ethernet packets from the emulated ethernet controller to |
196 |
the host machine's ethernet controller, and capture incoming |
197 |
packets on the host's controller, giving them back to the |
198 |
emulated OS. Characteristics are: |
199 |
<ul> |
200 |
<li>Requires <i>direct</i> access to the host's NIC, which |
201 |
means on most platforms that the emulator cannot be |
202 |
run as a normal user! |
203 |
<li>Reduced portability, as not every host operating system |
204 |
uses the same programming interface for dealing with |
205 |
hardware ethernet controllers directly. |
206 |
<li>When run on a switched network, it might be problematic to |
207 |
connect from the emulated OS to the OS running on the |
208 |
host, as packets sent out on the host's NIC are not |
209 |
received by itself. (?) |
210 |
<li>All specific networking protocols will be handled by the |
211 |
physical network. |
212 |
</ul> |
213 |
<p> |
214 |
or |
215 |
<p> |
216 |
<li>Whenever the emulated ethernet controller wishes to send a packet, |
217 |
the emulator looks at the packet and creates a response. Packets |
218 |
that can have an immediate response never go outside the emulator, |
219 |
other packet types have to be converted into suitable other |
220 |
connection types (UDP, TCP, etc). Characteristics: |
221 |
<ul> |
222 |
<li>Each packet type sent out on the emulated NIC must be handled. |
223 |
This means that I have to do a lot of coding. |
224 |
(I like this, because it gives me an opportunity to |
225 |
learn about networking protocols.) |
226 |
<li>By not relying on access to the host's NIC directly, |
227 |
portability is maintained. (It would be sad if the networking |
228 |
portion of a portable emulator isn't as portable as the |
229 |
rest of the emulator.) |
230 |
<li>The emulator can be run as a normal user process, does |
231 |
not require root privilegies. |
232 |
<li>Connecting from the emulated OS to the host's OS should |
233 |
not be problematic. |
234 |
<li>The emulated OS will experience the network just as a single |
235 |
machine behind a NAT gateway/firewall would. The emulated |
236 |
OS is thus automatically protected from the outside world. |
237 |
</ul> |
238 |
</ol> |
239 |
|
240 |
<p> |
241 |
Some emulators/simulators use the first approach, while others use the |
242 |
second. I think that SIMH and QEMU are examples of emulators using the |
243 |
first and second approach, respectively. |
244 |
|
245 |
<p> |
246 |
Since I have choosen the second kind of implementation, I have to write |
247 |
support explicitly for any kind of network protocol that should be |
248 |
supported. As of 2004-07-09, the following has been implemented and seems |
249 |
to work under at least NetBSD/pmax and OpenBSD/pmax under DECstation 5000/200 |
250 |
emulation (-E dec -e 3max): |
251 |
|
252 |
<p> |
253 |
<ul> |
254 |
<li>ARP requests sent out from the emulated NIC are interpreted, |
255 |
and converted to ARP responses. (This is used by the emulated OS |
256 |
to find out the MAC address of the gateway.) |
257 |
<li>ICMP echo requests (that is the kind of packet produced by the |
258 |
<b><tt>ping</tt></b> program) are interpreted and converted to ICMP echo |
259 |
replies, <i>regardless of the IP address</i>. This means that |
260 |
running ping from within the emulated OS will <i>always</i> |
261 |
receive a response. The ping packets never leave the emulated |
262 |
environment. |
263 |
<li>UDP packets are interpreted and passed along to the outside world. |
264 |
If the emulator receives an UDP packet from the outside world, it |
265 |
is converted into an UDP packet for the emulated OS. (This is not |
266 |
implemented very well yet, but seems to be enough for nameserver |
267 |
lookups, tftp file transfers, and NFS mounts using UDP.) |
268 |
<li>TCP packets are interpreted one at a time, similar to how UDP |
269 |
packets are handled (but more state is kept for each connection). |
270 |
<font color="#ff0000">NOTE: Much of the TCP handling code is very |
271 |
ugly and hardcoded.</font> |
272 |
<!-- |
273 |
<li>RARP is not implemented yet. (I haven't needed it so far.) |
274 |
--> |
275 |
</ul> |
276 |
|
277 |
<p> |
278 |
The gateway machine, which is the only "other" machine that the emulated |
279 |
OS sees on its emulated network, works as a NAT-style firewall/gateway. It |
280 |
usually has a fixed IPv4 address of <tt>10.0.0.254</tt>. An OS running in |
281 |
the emulator would usually have an address of the form <tt>10.x.x.x</tt>; |
282 |
a typical choice would be <tt>10.0.0.1</tt>. |
283 |
|
284 |
<p> |
285 |
Inside emulated NetBSD/pmax or OpenBSD/pmax, running the following |
286 |
commands should configure the emulated NIC: |
287 |
<pre> |
288 |
# <b>ifconfig le0 10.0.0.1</b> |
289 |
# <b>route add default 10.0.0.254</b> |
290 |
add net default: gateway 10.0.0.254 |
291 |
</pre> |
292 |
|
293 |
<p> |
294 |
If you want nameserver lookups to work, you need a valid /etc/resolv.conf |
295 |
as well: |
296 |
<pre> |
297 |
# <b>echo nameserver 129.16.1.3 > /etc/resolv.conf</b> |
298 |
</pre> |
299 |
(But replace <tt>129.16.1.3</tt> with the actual real-world IP address of |
300 |
your nearest nameserver.) |
301 |
|
302 |
<p> |
303 |
Now, host lookups should work: |
304 |
<pre> |
305 |
# <b>host -a www.netbsd.org</b> |
306 |
Trying null domain |
307 |
rcode = 0 (Success), ancount=2 |
308 |
The following answer is not authoritative: |
309 |
The following answer is not verified as authentic by the server: |
310 |
www.netbsd.org 86400 IN AAAA 2001:4f8:4:7:290:27ff:feab:19a7 |
311 |
www.netbsd.org 86400 IN A 204.152.184.116 |
312 |
For authoritative answers, see: |
313 |
netbsd.org 83627 IN NS uucp-gw-2.pa.dec.com |
314 |
netbsd.org 83627 IN NS ns.netbsd.org |
315 |
netbsd.org 83627 IN NS adns1.berkeley.edu |
316 |
netbsd.org 83627 IN NS adns2.berkeley.edu |
317 |
netbsd.org 83627 IN NS uucp-gw-1.pa.dec.com |
318 |
Additional information: |
319 |
ns.netbsd.org 83627 IN A 204.152.184.164 |
320 |
uucp-gw-1.pa.dec.com 172799 IN A 204.123.2.18 |
321 |
uucp-gw-2.pa.dec.com 172799 IN A 204.123.2.19 |
322 |
</pre> |
323 |
|
324 |
<p> |
325 |
At this point, UDP and TCP should (mostly) work. |
326 |
|
327 |
<p> |
328 |
Here is an example of how to configure a server machine and an emulated |
329 |
client machine for sharing files via NFS: |
330 |
|
331 |
<p> |
332 |
(This is very useful if you want to share entire directory trees |
333 |
between the emulated environment and another machine. These instruction |
334 |
will work for FreeBSD, if you are running something else, use your |
335 |
imagination to modify them.) |
336 |
|
337 |
<p> |
338 |
<ul> |
339 |
<li>On the server, add a line to your /etc/exports file, exporting |
340 |
the files you wish to use in the emulator:<pre> |
341 |
<b>/tftpboot -mapall=nobody -ro 123.11.22.33</b> |
342 |
</pre> |
343 |
where 123.11.22.33 is the IP address of the machine running the |
344 |
emulator process, as seen from the outside world. |
345 |
<p> |
346 |
<li>Then start up the programs needed to serve NFS via UDP. Note the |
347 |
-n argument to mountd. This is needed to tell mountd to accept |
348 |
connections from unprivileged ports (because the emulator does |
349 |
not need to run as root).<pre> |
350 |
# <b>portmap</b> |
351 |
# <b>nfsd -u</b> <--- u for UDP |
352 |
# <b>mountd -n</b> |
353 |
</pre> |
354 |
<li>In the guest OS in the emulator, once you have ethernet and IPv4 |
355 |
configured so that you can use UDP, mounting the filesystem |
356 |
should now be possible: (this example is for NetBSD/pmax |
357 |
or OpenBSD/pmax)<pre> |
358 |
# <b>mount -o ro,-r=1024,-w=1024,-U,-3 my.server.com:/tftpboot /mnt</b> |
359 |
or |
360 |
# <b>mount my.server.com:/tftpboot /mnt</b> |
361 |
</pre> |
362 |
If you don't supply the read and write sizes, there is a risk |
363 |
that the default values are too large. The emulator currently |
364 |
does not handle fragmentation/defragmentation of <i>outgoing</i> |
365 |
packets, so going above the ethernet frame size (1518) is a very |
366 |
bad idea. Incoming packets (reading from nfs) should work, though, |
367 |
for example during an NFS install. |
368 |
</ul> |
369 |
|
370 |
The example above uses read-only mounts. That is enough for things like |
371 |
letting NetBSD/pmax or OpenBSD/pmax install via NFS, without the need for |
372 |
a CDROM ISO image. You can use a read-write mount if you wish to share |
373 |
files in both directions, but then you should be aware of the |
374 |
fragmentation issue mentioned above. |
375 |
|
376 |
|
377 |
|
378 |
|
379 |
|
380 |
<p><br> |
381 |
<a name="devices"></a> |
382 |
<h3>Emulation of hardware devices</h3> |
383 |
|
384 |
Each file in the device/ directory is responsible for one hardware device. |
385 |
These are used from src/machine.c, when initializing which hardware a |
386 |
particular machine model will be using, or when adding devices to a |
387 |
machine using the <b>device()</b> command in configuration files. |
388 |
|
389 |
<p> |
390 |
<font color="#ff0000">NOTE: 2005-02-26: I'm currently rewriting the |
391 |
device registry subsystem.</font> |
392 |
|
393 |
<p> |
394 |
(I'll be using the name 'foo' as the name of the device in all these |
395 |
examples. This is pseudo code, it might need some modification to |
396 |
actually compile and run.) |
397 |
|
398 |
<p> |
399 |
Each device should have the following: |
400 |
|
401 |
<p> |
402 |
<ul> |
403 |
<li>A devinit function in dev_foo.c. It would typically look |
404 |
something like this: |
405 |
<pre> |
406 |
/* |
407 |
* devinit_foo(): |
408 |
*/ |
409 |
int devinit_foo(struct devinit *devinit) |
410 |
{ |
411 |
struct foo_data *d = malloc(sizeof(struct foo_data)); |
412 |
|
413 |
if (d == NULL) { |
414 |
fprintf(stderr, "out of memory\n"); |
415 |
exit(1); |
416 |
} |
417 |
memset(d, 0, sizeof(struct foon_data)); |
418 |
|
419 |
/* |
420 |
* Set up stuff here, for example fill d with useful |
421 |
* data. devinit contains settings like address, irq_nr, |
422 |
* and other things. |
423 |
* |
424 |
* ... |
425 |
*/ |
426 |
|
427 |
memory_device_register(devinit->machine->memory, devinit->name, |
428 |
devinit->addr, DEV_FOO_LENGTH, |
429 |
dev_foo_access, (void *)d, MEM_DEFAULT, NULL); |
430 |
|
431 |
/* This should only be here if the device |
432 |
has a tick function: */ |
433 |
machine_add_tickfunction(machine, dev_foo_tick, d, |
434 |
FOO_TICKSHIFT); |
435 |
|
436 |
/* Return 1 if the device was successfully added. */ |
437 |
return 1; |
438 |
} |
439 |
</pre><br> |
440 |
|
441 |
<li>At the top of dev_foo.c, the foo_data struct should be defined. |
442 |
<pre> |
443 |
struct foo_data { |
444 |
int irq_nr; |
445 |
/* ... */ |
446 |
} |
447 |
</pre><br> |
448 |
|
449 |
<li>If foo has a tick function (that is, something that needs to be |
450 |
run at regular intervals) then FOO_TICKSHIFT and a tick function |
451 |
need to be defined as well: |
452 |
<pre> |
453 |
#define FOO_TICKSHIFT 10 |
454 |
|
455 |
void dev_foo_tick(struct cpu *cpu, void *extra) |
456 |
{ |
457 |
struct foo_data *d = (struct foo_data *) extra; |
458 |
|
459 |
if (.....) |
460 |
cpu_interrupt(cpu, d->irq_nr); |
461 |
else |
462 |
cpu_interrupt_ack(cpu, d->irq_nr); |
463 |
} |
464 |
</pre><br> |
465 |
|
466 |
<li>And last but not least, the device should have an access function. |
467 |
The access function is called whenever there is a load or store |
468 |
to an address which is in the device' memory mapped region. |
469 |
<pre> |
470 |
int dev_foo_access(struct cpu *cpu, struct memory *mem, |
471 |
uint64_t relative_addr, unsigned char *data, size_t len, |
472 |
int writeflag, void *extra) |
473 |
{ |
474 |
struct foo_data *d = extra; |
475 |
uint64_t idata = 0, odata = 0; |
476 |
|
477 |
idata = memory_readmax64(cpu, data, len); |
478 |
switch (relative_addr) { |
479 |
/* .... */ |
480 |
} |
481 |
|
482 |
if (writeflag == MEM_READ) |
483 |
memory_writemax64(cpu, data, len, odata); |
484 |
|
485 |
/* Perhaps interrupts need to be asserted or |
486 |
deasserted: */ |
487 |
dev_foo_tick(cpu, extra); |
488 |
|
489 |
/* Return successfully. */ |
490 |
return 1; |
491 |
} |
492 |
</pre><br> |
493 |
</ul> |
494 |
|
495 |
<p> |
496 |
The return value of the access function has until 2004-07-02 been a |
497 |
true/false value; 1 for success, or 0 for device access failure. A device |
498 |
access failure (on MIPS) will result in a DBE exception. |
499 |
|
500 |
<p> |
501 |
Some devices are converted to support arbitrary memory latency |
502 |
values. The return value is the number of cycles that the read or |
503 |
write access took. A value of 1 means one cycle, a value of 10 means 10 |
504 |
cycles. Negative values are used for device access failures, and the |
505 |
absolute value of the value is then the number of cycles; a value of -5 |
506 |
means that the access failed, and took 5 cycles. |
507 |
|
508 |
<p> |
509 |
To be compatible with pre-20040702 devices, a return value of 0 is treated |
510 |
by the caller (in <tt>src/memory_rw.c</tt>) as a value of -1. |
511 |
|
512 |
|
513 |
|
514 |
|
515 |
|
516 |
<p><br> |
517 |
<a name="regtest"></a> |
518 |
<h3>Regression tests</h3> |
519 |
|
520 |
In order to make sure that the emulator actually works like it is supposed |
521 |
to, it must be tested. For this purpose, there is a simple regression |
522 |
testing framework in the <tt>tests/</tt> directory. |
523 |
|
524 |
<p> |
525 |
<i>NOTE: The regression testing framework is basically just a skeleton so far. |
526 |
Regression tests are very good to have. However, the fact that complete |
527 |
operating systems can run in the emulator indicate that the emulation is |
528 |
probably not too incorrect. This makes it less of a priority to write |
529 |
regression tests.</i> |
530 |
|
531 |
<p> |
532 |
To run all the regression tests, type <tt>make regtest</tt>. Each assembly |
533 |
language file matching the pattern <tt>test_*.S</tt> will be compiled and |
534 |
linked into a 64-bit MIPS ELF (using a gcc cross compiler), and run in the |
535 |
emulator. If everything goes well, you should see something like this: |
536 |
|
537 |
<pre> |
538 |
$ make regtest |
539 |
cd tests; make run_tests; cd .. |
540 |
gcc33 -Wall -fomit-frame-pointer -fmove-all-movables -fpeephole -O2 |
541 |
-mcpu=ev5 -I/usr/X11R6/include -lm -L/usr/X11R6/lib -lX11 do_tests.c |
542 |
-o do_tests |
543 |
do_tests.c: In function `main': |
544 |
do_tests.c:173: warning: unused variable `s' |
545 |
/var/tmp//ccFOupvD.o: In function `do_tests': |
546 |
/var/tmp//ccFOupvD.o(.text+0x3a8): warning: tmpnam() possibly used |
547 |
unsafely; consider using mkstemp() |
548 |
mips64-unknown-elf-gcc -g -O3 -fno-builtin -fschedule-insns -mips64 |
549 |
-mabi=64 test_common.c -c -o test_common.o |
550 |
./do_tests "mips64-unknown-elf-gcc -g -O3 -fno-builtin -fschedule-insns |
551 |
-mips64 -mabi=64" "mips64-unknown-elf-as -mabi=64 -mips64" |
552 |
"mips64-unknown-elf-ld -Ttext 0xa800000000030000 -e main |
553 |
--oformat=elf64-bigmips" "../gxemul" |
554 |
|
555 |
Starting tests: |
556 |
test_addu.S (-a) |
557 |
test_addu.S (-a -b) |
558 |
test_clo_clz.S (-a) |
559 |
test_clo_clz.S (-a -b) |
560 |
.. |
561 |
test_unaligned.S (-a) |
562 |
test_unaligned.S (-a -b) |
563 |
|
564 |
Done. (12 tests done) |
565 |
PASS: 12 |
566 |
FAIL: 0 |
567 |
|
568 |
---------------- |
569 |
|
570 |
All tests OK |
571 |
|
572 |
---------------- |
573 |
</pre> |
574 |
|
575 |
<p> |
576 |
Each test writes output to stdout, and there is a <tt>test_*.good</tt> for |
577 |
each <tt>.S</tt> file which contains the wanted output. If the actual |
578 |
output matches the <tt>.good</tt> file, then the test passes, otherwise it |
579 |
fails. |
580 |
|
581 |
<p> |
582 |
Read <tt>tests/README</tt> for more information. |
583 |
|
584 |
|
585 |
|
586 |
|
587 |
</body> |
588 |
</html> |