![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to manual.html CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [HATARI the Atari ST Emulator] / hatari / src / gui-osx / French.lproj / AideHatari|
Return to manual.html CVS log | Up to [HATARI the Atari ST Emulator] / hatari / src / gui-osx / French.lproj / AideHatari |
1.1 root 1: <!DOCTYPE html>
2: <html>
3: <head>
4: <title>Hatari User's Manual</title>
5: <meta name="description"
6: content="User's manual for the Atari ST emulator Hatari" />
7: <meta name="author" content="Hatari development team" />
8: <meta name="keywords" content="hatari, documentation, manual, linux" />
9: <meta name="viewport" content="width=device-width; initial-scale=1.0;" />
10: <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-15" />
11: <style type="text/css">
12: <!--
13: body { background:#FFFFFF;
14: color:#000000;
15: margin-left:10px;
16: margin-right:10px;
17: font-family:Verdana,Arial,Helvetica,sans-serif;
18: }
19:
20: h2 { border-bottom:solid thin black;}
21: h4.gui { clear:right }
22: h5 { margin-bottom:2px; margin-left:1em; }
23: hr { width: 100%; height: 2px; margin-top:4ex; margin-bottom: 2ex; }
24:
25: pre { color: black;
26: background:#eeeeee;
27: margin: 0px 20px 8px 20px;
28: padding: 2px 8px 1px 8px;
29: border: solid thin #ccaa88;
30: }
31:
32: th { text-align:center; }
33: td { font-family:Verdana,Arial,Helvetica,sans-serif; }
34:
35: a:link { color:#000099; background:#ffffff; text-decoration:none; }
36: a:visited { color:#cc0000; background:#ffffff; text-decoration:none;}
37: a:hover { color:#0000ff; background:#ffffff; text-decoration:none; }
38: a:active { color:#993399; background:#ffffff; text-decoration:none; }
39:
40: .pageheader { text-align:center; }
41: .commandline { font-family:Courier,monospace; font-size:90% }
42: .file { color: #000088;}
43: .button { color:#000000; background:#c0c0c0; border:outset thin gray; font-family:Courier,monospace; padding-left:1em; padding-right:1em;}
44: .key { color:#550000; font-family:Courier,monospace; font-size:90% }
45: .backdropped { background:#ffffee; }
46: .image {margin-left: 5px; margin-right: 5px; border-width:2px; border-style:solid; border-color:#eeeeff; padding:1cm; text-align:center; }
47: .floatimage { clear:right; float:right; margin-left:0.5cm; border-width:2px; border-style:solid; border-color:#eeeeff; padding:0.5cm; }
48: .clearboth { clear:both; }
49:
50: table.keytable { border:1px solid; }
51: table.keytable tbody { text-align:center; }
52: table.keytable th { border:1px solid; padding:3px; }
53: table.keytable td { border:1px solid; padding:3px; }
54:
55: p.parameter { margin: 1px 1em 1px 11%; font-weight:bold; }
56: p.paramdesc { margin: 1px 1em 1px 22%; }
57: -->
58: </style>
59: <script type="text/javascript" src="toc.js"></script>
60: </head>
61:
62: <body>
63:
64: <h1 class="pageheader">Hatari User's Manual</h1>
65:
66: <p class="pageheader">
67: Version 1.7, June 2013
68: </p>
69: <p class="pageheader">
70: Manual written by: <strong>Thomas Huth</strong>, <strong>Matthias Arndt</strong>
71: & <strong>Eero Tamminen</strong>
72: </p>
73: <p class="pageheader">
74: Hatari on the WWW:
75: <strong>
76: <a href="http://hatari.tuxfamily.org/">http://hatari.tuxfamily.org/</a>
77: </strong>
78: </p>
79:
80: <h2 class="no-TOC">Index</h2>
81:
82: <div id="generated-toc">
83: <!-- The TOC is generated automatically via JavaScript -->
84: </div>
85:
86: <h2>Introduction</h2>
87:
88: <h3>General description</h3>
89: <p>
90: Hatari is an Atari ST, STE, TT and Falcon emulator for Linux, OSX,
91: Windows and other Systems which are supported by the SDL library.
92: The emulator is open source software and is distributed under the terms of the
93: <a href="http://www.gnu.org/licenses/old-licenses/gpl-2.0.html">GNU General
94: Public License (GPL)</a>.
95: </p>
96: <p>
97: The Atari ST was a 16/32 bit computer system which was first released by Atari
98: in 1985. Using the Motorola 68000 CPU, it was a very popular computer having
99: quite a lot of CPU power at that time. See Appendix B for details on emulation
100: in general.
101: </p>
102: <p>
103: Unlike many other Atari ST emulators which try to give you a good
104: environment for running GEM applications, Hatari tries to emulate the hardware
105: of a ST as close as possible so that it is able to run most of the old ST games
106: and demos. Of course you can run normal GEM applications with Hatari, too.
107: Recent versions of Hatari even feature STE, Falcon and basic TT emulation.
108: </p>
109:
110: <h3>Features</h3>
111: <ul>
112: <li>68000 - 68040 emulation via the UAE CPU core
113: (MMU emulation only with the WinUAE CPU core)</li>
114: <li>ST RAM size variable (from 512kiB up to 14MiB are possible)</li>
115: <li>optional cartridge images for the ST ROM port</li>
116: <li>most of the ST specific hardware</li>
117: <li>ST Shifter with ST-High, ST-Medium and ST-Low resolutions,
118: overscan effects for all borders in color resolutions</li>
119: <li>512 color ST palette</li>
120: <li>Spec512 mode support for low and medium resolutions</li>
121: <li>many raster effects </li>
122: <li>scaling of low resolutions by factor two</li>
123: <li>interleaved lines rendering of ST-medium and (scaled) ST-low
124: resolutions for the TV "monitor type"</li>
125: <li>Blitter chip emulation</li>
126: <li>PSG YM2149 emulation (soundchip) including STFM samples</li>
127: <li>Printer port emulation on hardware level (print to file)</li>
128: <li>RS232 emulation</li>
129: <li>MIDI input/output/through emulation</li>
130: <li>Mega ST real time clock</li>
131: <li>IKBD emulation (keyboard, mouse and joystick) with custom
132: keyboard mapping</li>
133: <li>joystick emulation via cursor keys and joystick emulation via a
134: connected PC joystick</li>
135: <li>FDC emulation using floppy disk images in standard formats (*.ST,
136: *.MSA and *.DIM)</li>
137: <li>support for packed disk images (PkZip and Gzip)</li>
138: <li>optional write-protection for floppy disk images</li>
139: <li>partial ACSI emulation for harddisk support</li>
140: <li>GEMDOS interface driver to mount directories as harddrives
141: with optional write-protection</li>
142: <li>support for memory snapshots (save whole system state)</li>
143: <li>driver for extended VDI resolutions</li>
144: <li>recording of sound as .WAV and .YM files</li>
145: <li>screenshots in PNG or BMP format</li>
146: <li>AVI animation capturing with sound</li>
147: <li>TOS versions 1.00, 1.02, 1.04 and 2.06 (and EmuTOS) can be used in ST mode.</li>
148: </ul>
149:
150: <h4>STE hardware emulation</h4>
151: <p>There is support for following additional STE features:</p>
152: <ul>
153: <li>horizontal and vertical hardware fine scrolling</li>
154: <li>split screen techniques / in-screen video address manipulations</li>
155: <li>(STE specific) left border opening</li>
156: <li>4096 colors STE palette</li>
157: <li>Stereo DMA sample sound</li>
158: <li>Microwire/LMC1992 emulation</li>
159: <li>STE joypads</li>
160: <li>TOS versions 1.06, 1.62, 2.05 and 2.06 (and EmuTOS) can be used in STE mode.</li>
161: </ul>
162:
163: <h4>Experimental TT hardware emulation</h4>
164: <p>There is support for following additional TT features:</p>
165: <ul>
166: <li>TT low/med/high resolution support</li>
167: <li>ST/TT palette switching and video shifter</li>
168: <li>RAM upto 14MiB (ST-RAM only, there is no support for TT-RAM yet)</li>
169: <li>Only TOS version 3.06 (and EmuTOS) can be used in TT mode.</li>
170: </ul>
171:
172: <h4>Falcon hardware emulation</h4>
173: <p>There is support for following additional Falcon features:</p>
174: <ul>
175: <li>Partial Videl and Videl borders emulation for all Falcon screen modes</li>
176: <li>Aspect correction and scaling of small resolutions by an integer factor</li>
177: <li>STE/Falcon palette switching and shifter</li>
178: <li>Mono/RGB/VGA/TV monitor types</li>
179: <li>DSP co-processor emulation</li>
180: <li>Experimental microphone (jack) emulation</li>
181: <li>Experimental Crossbar sound matrix (ADC (mic & PSG), DAC, DMA, DSP)
182: interconnect emulation + support for the additional DMA sound
183: sample rates</li>
184: <li>Experimental IDE master and slave emulation for harddisk support</li>
185: <li>TOS versions 4.00, 4.02, 4.04 and 4.92 (and EmuTOS) can be used in Falcon mode.</li>
186: </ul>
187:
188: <p>See the developers' <span class="file">doc/todo.txt</span> file
189: (included with Hatari sources) for the details on the few remaining
190: emulation gaps and the <a href="compatibility.html">Hatari Atari
191: Software Compatibility List</a> for which Atari programs are known
192: to be affected by them.</p>
193:
194:
195: <h3>System requirements</h3>
196:
197: <p> Hatari currently has the following minimum system requirements:</p>
198: <ul>
199: <li>a fast PC (>500MHz, for Falcon and TT emulation
200: <a href="#Performance">even faster</a>)</li>
201: <li>some sort of Unix (preferable <a href="http://www.linux.org/">GNU/Linux</a>)
202: </li>
203: <li>the SDL library (<a href="http://www.libsdl.org/">http://www.libsdl.org/</a>)</li>
204: <li>the zLib (<a href="http://www.gzip.org/zlib/">http://www.gzip.org/zlib/</a>)
205: for support of ZIP-packed disk images (*.zip and *.gz)</li>
206: </ul>
207:
208: <p>
209: In the course of time Hatari has successfully been tested by various people on
210: the following systems:
211: </p>
212: <ul>
213: <li>Linux/i86 with Kernel 2.4.x and 2.6.x</li>
214: <li>Linux/PPC with Kernel 2.4.x and 2.6.x</li>
215: <li>BeOS/i86</li>
216: <li>Apple Mac OS X on PowerPC and i86</li>
217: <li>NetBSD 1.6 on i86</li>
218: <li>NetBSD on a Digital Alpha</li>
219: <li>FreeBSD 4.1 on an i486, FreeBSD 4.8 on a Pentium 4 and FreeBSD 5.1</li>
220: <li>OpenBSD 3.5 and 5.0</li>
221: <li>Solaris 8 on a SUN UltraSparc 1</li>
222: <li>Linux/ARM (oabi) on Sharp Zaurus SL-C760 PDA</li>
223: <li>Linux/ARM (eabi) on Nokia Maemo Internet Tablets and N900 phone</li>
224: <li>Windows XP</li>
225: </ul>
226:
227: <h2>Compiling and running</h2>
228:
229: <h3>Compiling Hatari</h3>
230:
231: <p>Required:</p>
232: <ul>
233: <li>A C compiler. Preferably GCC, but others have worked too.</li>
234: <li>A working CMake installation. See
235: <a href="http://www.cmake.org/">http://www.cmake.org/</a> for details.
236: <li>The SDL library v1.2.10 or newer. You can get it from
237: <a href="http://www.libsdl.org/">http://www.libsdl.org/</a>.
238: </li>
239: <li>The zLib compression library. You can get it from
240: <a href="http://www.gzip.org/zlib/">http://www.gzip.org/zlib/</a>.
241: </li>
242: </ul>
243:
244: <p>Optional:</p>
245: <ul>
246: <li>The PNG image library for PNG format screenshots and to
247: decrease AVI video recording file sizes. You can get it from
248: <a href="http://www.libpng.org/">http://www.libpng.org/</a>.</li>
249: <li>The GNU Readline library for Hatari debugger command line editing.</li>
250: <li>The Xlib library to support Hatari Python UI window embedding
251: on systems with the X window system (Linux and other unixes).</li>
252: <li>The portaudio library for Falcon microphone recording support</li>
253: </ul>
254: <p>
255: The versions available in your Linux distribution will be sufficient
256: in most cases, but make sure you have also the header files installed
257: for the libraries as well! Typically they're in a corresponding -dev
258: package.
259: </p>
260:
261: <p>
262: After you've verified that you have the required libraries and their
263: development files, change to the <span class="file">hatari/</span>
264: directory. Create a <span class="file">build/</span> directory under
265: it and configure the build system for your environment:
266: <pre>
267: mkdir -p build
268: cd build
269: cmake ..
270: </pre>
271: <p>
272: Then compile Hatari by typing <span class="commandline">make</span>.
273: If all works fine, you'll get the executable <span class="commandline">hatari</span>
274: in the src/ subdirectory.
275: </p>
276: <p>
277: Note: Instead of calling CMake directly, you can also use the supplied
278: configure script to run CMake and to give the arguments (like install
279: prefix) in a format familiar from GNU Autotools using programs. Type
280: "<span class="commandline">./configure --help</span>"
281: to see all the options supported by this script.
282: </p>
283:
284: <h3>Installation of a TOS ROM</h3>
285:
286: <p>
287: Before you can start Hatari, you have to copy a TOS ROM image to the data
288: directory (<span class="file"><prefix>/share/hatari/</span>, by
289: default <span class="file">/usr/local/share/hatari/</span>) and
290: rename it to <span class="commandline">tos.img</span>, or use the
291: <span class="commandline">--tos</span> command line option to tell
292: Hatari where to find a TOS ROM.
293: Hatari needs a TOS ROM image because this contains the operating system
294: of the emulated Atari.
295: </p>
296: <p>
297: Unfortunately it is not possible to ship an original ROM
298: image with the Hatari package since these images are still copyrighted.
299: But you can easily create an image with a real ST and one of those various
300: ROM-image programs for the ST (search for "TOSDUMP" with your
301: favourite internet search engine). If your old ST does not work anymore, you
302: can also try to search the internet directly for corresponding TOS ROM image,
303: but don't ask the Hatari team where to get one. </p>
304: <p> Another solution is EmuTOS, which is also shipped with the official
305: release versions of Hatari. EmuTOS is an open-source TOS clone. You can find
306: it at:
307: <a href="http://emutos.sourceforge.net/">http://emutos.sourceforge.net/</a>.
308: It is not the best solution for playing games or running other old software
309: due to compatibility issues (see <span class="file">emutos.txt</span> for
310: more details), but it's free and compatible with Hatari.</p>
311: <p>If you do not specify a TOS image on the commandline nor can Hatari
312: find a suitable TOS image in the default dir, you'll get the chance to
313: select a TOS image file from the GUI. </p>
314:
315: <h3>Installation of the binary</h3>
316:
317: <p> Type <span class="commandline">make install</span> as "root" user to
318: do a systemwide installation.</p>
319: <p>Assuming you didn't change the default installation prefix and that
320: <span class="file">/usr/local/bin/</span> is in your PATH, you should
321: be now able to start the Hatari executable from anywhere.</p>
322: <p> When you finally have got a TOS image, try starting Hatari with the
323: option <span class="commandline">--help</span> to find out more about
324: its command line parameters. </p>
325:
326: <h3>Running Hatari for the first time</h3>
327:
328: <p> Now type <span class="commandline">hatari</span> to run the
329: emulator for the first time. If all goes
330: well, you should now be presented with a window showing you the
331: familiar
332: little green desktop of the Atari ST. Press <span class="key">F12</span>
333: to turn on the GUI to
334: configure Hatari to suit your needs, press <span class="key">F11</span>
335: to toggle windowed and fullscreen mode. </p>
336:
337: <h3>Configuration options precedence</h3>
338:
339: <p>Hatari settings can come from several sources, with later ones
340: overriding the earlier given ones:
341: <ul>
342: <li>Builtin Hatari default options (which are different for old UAE and WinUAE
343: CPU core builds, former defaults to ST, latter to Falcon)</li>
344: <li>Global <span class="commandline">/etc/hatari.cfg</span>
345: (or <span class="commandline">/usr/local/etc/hatari.cfg</span>)
346: configuration file</li>
347: <li>User specific <span class="commandline">~/.hatari/hatari.cfg</span>
348: configuration file</li>
349: <li>Command line arguments
350: <li>Option changes done at run-time in Hatari options GUI, with debugger "setopt"
351: command or through the (optionally enabled) Hatari control socket.
352: </ul>
353:
354: <p>Some of the run-time changes require emulation to be reseted for them
355: to take effect.</p>
356:
357:
358: <h2>Command line options and arguments</h2>
359:
360: <p>Usage:</p>
361: <pre>
362: hatari [options] [disk image | directory | Atari program ]
363: </pre>
364:
365: <p>As an argument one can give either a name of:</p>
366: <ul>
367: <li>A floppy disk image,
368: <li>A directory that should be emulated as a virtual GEMDOS hard disk, or</li>
369: <li>An Atari program that should be autostarted. In this case
370: the program's directory will be used as the C: drive from
371: where this program will be started.
372: (Note that autostarting a program might not work if you've also
373: specified a floppy image for drive A: on command line or in config
374: file which contains a desktop.inf/newdesk.inf/emutos.inf file on
375: it.)</li>
376: </ul>
377:
378: <p>Booting will be done from the disk image or directory that's given
379: last on the command line as an option or the argument (and which
380: corresponds to A: or C:).</p>
381:
382: <p>Hatari command line options are split into several categories:</p>
383:
384: <!--
385: Generated from hatari.1 options section by changing subheaders to h3
386: and removing extra paragraphs:
387: groff -man -Thtml hatari.1 | tidy | awk '
388: /h2.OPTIONS/ { out = 1; next }
389: /COMMANDS/ { out = 0; next }
390: { if(out) print }' \
391: | sed -e 's/h2/h3/g' -e 's/<b>//' -e 's/<\/b>//' \
392: -e 's/style="margin-left:11%;[^"]*"/class="parameter"/g' \
393: -e 's/style="margin-left:22%;"/class="paramdesc"/g'
394: -->
395:
396: <h3>General options</h3>
397: <p class="parameter">−h,
398: −−help</p>
399: <p class="paramdesc">Print command line options and
400: terminate</p>
401: <p class="parameter">−v,
402: −−version</p>
403: <p class="paramdesc">Print version information and
404: terminate</p>
405: <p class="parameter">−−confirm-quit
406: <bool></p>
407: <p class="paramdesc">Whether Hatari confirms quitting</p>
408: <p class="parameter">−c, −−configfile
409: <filename></p>
410: <p class="paramdesc">Read additional configuration values from
411: <file>, these override values read from the global and
412: user configuration files
413: </p>
414: <p class="parameter">−k, −−keymap
415: <file></p>
416: <p class="paramdesc">load keyboard mapping from
417: <file></p>
418: <p class="parameter">−−fast-forward
419: <bool></p>
420: <p class="paramdesc">On fast machine helps skipping (fast
421: forwarding) Hatari output</p>
422:
423: <h3>Common display options</h3>
424: <p class="parameter">−m,
425: −−mono</p>
426: <p class="paramdesc">Start in monochrome mode instead of
427: color</p>
428: <p class="parameter">−−monitor
429: <x></p>
430: <p class="paramdesc">Select monitor type (x =
431: mono/rgb/vga/tv)</p>
432: <p class="parameter">−f,
433: −−fullscreen</p>
434: <p class="paramdesc">Start the emulator in fullscreen
435: mode</p>
436: <p class="parameter">−w,
437: −−window</p>
438: <p class="paramdesc">Start the emulator in window mode</p>
439: <p class="parameter">−−grab</p>
440: <p class="paramdesc">Grab mouse (also) in window mode</p>
441: <p class="parameter">−−frameskips
442: <x></p>
443: <p class="paramdesc">Skip <x> frames after each
444: displayed frame to accelerate emulation (0=disabled, >4 uses
445: automatic frameskip with given value as maximum)</p>
446: <p class="parameter">−−statusbar
447: <bool></p>
448: <p class="paramdesc">Show statusbar (with floppy leds etc
449: etc)</p>
450: <p class="parameter">−−drive-led
451: <bool></p>
452: <p class="paramdesc">Show overlay drive led when statusbar
453: isn’t shown</p>
454: <p class="parameter">−−bpp
455: <bool></p>
456: <p class="paramdesc">Force internal bitdepth (x =
457: 8/15/16/32, 0=disable)</p>
458:
459: <h3>ST/STE specific display options</h3>
460: <p class="parameter">
461: −−borders <bool></p>
462: <p class="paramdesc">Show ST/STE screen borders (for low/med
463: resolution overscan demos)</p>
464: <p class="parameter">−−desktop-st
465: <bool></p>
466: <p class="paramdesc">Whether fullscreen mode uses desktop
467: resolution to avoid: messing multi-screen setups, several seconds
468: delay needed by LCD monitors resolution switching and the resulting
469: sound break. As Hatari ST/E display code doesn’t support
470: zooming (except low-rez doubling), it doesn’t get scaled (by
471: Hatari or monitor) when this is enabled. Therefore this is mainly
472: useful only if you suffer from the described effects, but still
473: want to grab mouse and remove other distractions from the screen
474: just by toggling fullscreen mode. (disabled by default)</p>
475: <p class="parameter">−−spec512
476: <x></p>
477: <p class="paramdesc">Hatari uses this threshold to decide
478: when to render a screen with the slower but more accurate
479: Spectrum512 screen conversion functions (0 <= x <= 512,
480: 0=disable)</p>
481: <p class="parameter">−z, −−zoom
482: <x></p>
483: <p class="paramdesc">Zoom (double) low resolution (1=no,
484: 2=yes)</p>
485:
486: <h3>Falcon/TT specific display options</h3>
487: <p class="parameter">Zooming to sizes
488: specified below is internally done using integer scaling factors.
489: This means that different Atari resolutions may show up with
490: different sizes, but they are never blurry. <br />
491: −−desktop <bool></p>
492: <p class="paramdesc">Whether to use desktop resolution on
493: fullscreen to avoid issues related to resolution switching.
494: Otherwise fullscreen will use a resolution that is closest to the
495: Hatari window size. (enabled by default)</p>
496: <p class="parameter">−−max-width
497: <x></p>
498: <p class="paramdesc">Preferred / maximum window width</p>
499: <p class="parameter">−−max-height
500: <x></p>
501: <p class="paramdesc">Preferred / maximum window height</p>
502: <p class="parameter">−−force-max
503: <bool></p>
504: <p class="paramdesc">Hatari window size is forced to
505: specified maximum size and black borders used when Atari resolution
506: doesn’t scale evenly to it. This is most useful when
507: recording videos of Falcon demos that change their resolution.
508: (disabled by default)</p>
509: <p class="parameter">−−aspect
510: <bool></p>
511: <p class="paramdesc">Whether to do monitor aspect ratio
512: correction (enabled by default)</p>
513:
514: <h3>VDI options</h3>
515: <p class="parameter">−−vdi
516: <bool></p>
517: <p class="paramdesc">Whether to use VDI screen mode</p>
518: <p class="parameter">−−vdi−planes
519: <x></p>
520: <p class="paramdesc">Use extended VDI resolution with bit
521: depth <x> (x = 1, 2 or 4)</p>
522: <p class="parameter">−−vdi−width
523: <w></p>
524: <p class="paramdesc">Use extended VDI resolution with width
525: <w> (320 < w <= 1280)</p>
526: <p class="parameter">−−vdi−height
527: <h></p>
528: <p class="paramdesc">Use extended VDI resolution with height
529: <h> (200 < h <= 960)</p>
530:
531: <h3>Screen capture options</h3>
532: <p class="parameter">−−crop
533: <bool></p>
534: <p class="paramdesc">Remove statusbar from the screen
535: captures</p>
536: <p class="parameter">−−avirecord</p>
537: <p class="paramdesc">Start AVI recording</p>
538: <p class="parameter">−−avi-vcodec
539: <x></p>
540: <p class="paramdesc">Select avi video codec (x =
541: bmp/png)</p>
542: <p class="parameter">−−avi-fps
543: <x></p>
544: <p class="paramdesc">Force avi frame rate (x =
545: 50/60/71/...)</p>
546: <p class="parameter">−−avi-file
547: <file></p>
548: <p class="paramdesc">Use <file> to record avi</p>
549:
550: <h3>Devices options</h3>
551: <p class="parameter">−j,
552: −−joystick <port></p>
553: <p class="paramdesc">Emulate joystick with cursor keys in
554: given port (0-5)</p>
555: <p class="parameter">−−joy<port>
556: <type></p>
557: <p class="paramdesc">Set joystick type (none/keys/real) for
558: given port</p>
559: <p class="parameter">−−printer
560: <file></p>
561: <p class="paramdesc">Enable printer support and write data
562: to <file></p>
563: <p class="parameter">−−midi-in
564: <filename></p>
565: <p class="paramdesc">Enable MIDI support and write MIDI data
566: to <file></p>
567: <p class="parameter">−−midi-out
568: <filename></p>
569: <p class="paramdesc">Enable MIDI support and read MIDI data
570: from <file></p>
571: <p class="parameter">−−rs232-in
572: <filename></p>
573: <p class="paramdesc">Enable serial port support and use
574: <file> as the input device</p>
575: <p class="parameter">−−rs232-out
576: <filename></p>
577: <p class="paramdesc">Enable serial port support and use
578: <file> as the output device</p>
579:
580: <h3>Disk options</h3>
581: <p class="parameter">−−disk-a
582: <file></p>
583: <p class="paramdesc">Set disk image for floppy drive A</p>
584: <p class="parameter">−−disk-b
585: <file></p>
586: <p class="paramdesc">Set disk image for floppy drive B</p>
587: <p class="parameter">−−protect-floppy
588: <x></p>
589: <p class="paramdesc">Write protect floppy image contents
590: (on/off/auto). With "auto" option write protection is according to
591: the disk image file attributes.</p>
592: <p class="parameter">−−protect-hd
593: <x></p>
594: <p class="paramdesc">Write protect harddrive <dir>
595: contents (on/off/auto). With "auto" option the protection can be
596: controlled by setting individual files attributes as it disables
597: the file attribute modifications for the GEMDOS hard disk
598: emulation.</p>
599: <p class="parameter">−−gemdos-case <x></p>
600: <p class="paramdesc">Specify whether new dir/filenames are forced to be
601: in upper or lower case with the GEMDOS HD emulation. Off by default.
602: </p>
603: <p class="parameter">−d, −−harddrive
604: <dir></p>
605: <p class="paramdesc">Emulate harddrive partition(s) with
606: <dir> contents. If directory contains only single letter
607: (C-Z) subdirectories, each of these subdirectories will be treated
608: as a separate partition, otherwise the given directory itself will
609: be assigned to drive "C:". In the multiple partition case, the
610: letters used as the subdirectory names will determine to which
611: drives/partitions they’re assigned.</p>
612: <p class="parameter">−−acsi
613: <file></p>
614: <p class="paramdesc">Emulate an ACSI hard disk with an image
615: <file></p>
616: <p class="parameter">−−ide−master
617: <file></p>
618: <p class="paramdesc">Emulate an IDE master hard disk with an
619: image <file></p>
620: <p class="parameter">−−ide−slave
621: <file></p>
622: <p class="paramdesc">Emulate an IDE slave hard disk with an
623: image <file></p>
624: <p class="parameter">−−fastfdc
625: <bool></p>
626: <p class="paramdesc">speed up FDC emulation (can cause
627: incompatibilities)</p>
628:
629: <h3>Memory options</h3>
630: <p class="parameter">
631: −−memstate <file></p>
632: <p class="paramdesc">Load memory snap-shot <file></p>
633: <p class="parameter">−s, −−memsize
634: <x></p>
635: <p class="paramdesc">Set amount of emulated RAM, x = 1 to 14
636: MiB, or 0 for 512 KiB</p>
637:
638: <h3>ROM options</h3>
639: <p class="parameter">−t,
640: −−tos <imagefile></p>
641: <p class="paramdesc">Specify TOS ROM image to use</p>
642: <p class="parameter">−−patch-tos
643: <bool></p>
644: <p class="paramdesc">Use this option to enable/disable TOS
645: ROM patching. Experts only! Leave this enabled unless you know what
646: you are doing!</p>
647: <p class="parameter">−−cartridge
648: <imagefile></p>
649: <p class="paramdesc">Use ROM cartridge image <file>
650: (only works if GEMDOS HD emulation and extended VDI resolution are
651: disabled)</p>
652:
653: <h3>CPU options</h3>
654: <p class="parameter">
655: −−cpulevel <x></p>
656: <p class="paramdesc">Specify CPU (680x0) to use (use x >=
657: 1 with EmuTOS or TOS >= 2.06 only!)</p>
658: <p class="parameter">−−cpuclock
659: <x></p>
660: <p class="paramdesc">Set the CPU clock (8, 16 or 32 Mhz)</p>
661: <p class="parameter">−−compatible
662: <bool></p>
663: <p class="paramdesc">Use a more compatible, but slower 68000
664: CPU mode with better prefetch accuracy and cycle counting</p>
665:
666: <h3>Misc system options</h3>
667: <p class="parameter">
668: −−machine <x></p>
669: <p class="paramdesc">Select machine type (x = st, ste, tt or
670: falcon)</p>
671: <p class="parameter">−−blitter
672: <bool></p>
673: <p class="paramdesc">Enable blitter emulation (ST only)</p>
674: <p class="parameter">−−dsp <x></p>
675: <p class="paramdesc">Falcon DSP emulation (x = none, dummy
676: or emu, Falcon only)</p>
677: <p class="parameter">−−timer-d
678: <bool></p>
679: <p class="paramdesc">Patch redundantly high Timer-D
680: frequency set by TOS. This about doubles Hatari speed (for ST/e
681: emulation) as the original Timer-D frequency causes most of the
682: interrupts.</p>
683: <p class="parameter">−−fast-boot
684: <bool></p>
685: <p class="paramdesc">Patch TOS and initialize the so-called
686: "memvalid" system variables to by-pass the memory test of TOS, so
687: that the system boots faster.</p>
688: <p class="parameter">−−rtc
689: <bool></p>
690: <p class="paramdesc">Enable real-time clock</p>
691:
692: <h3>Sound options</h3>
693: <p class="parameter">−−mic
694: <bool></p>
695: <p class="paramdesc">Enable/disable (Falcon only)
696: microphone</p>
697: <p class="parameter">−−sound
698: <x></p>
699: <p class="paramdesc">Sound frequency: 6000-50066. "off"
700: disables the sound and speeds up the emulation. To prevent extra
701: sound artifacts, the frequency should be selected so that it either
702: matches evenly with the STE/TT/Falcon sound DMA (6258, 12517,
703: 250033, 50066 Hz) or your sound card frequencies (11025, 22050,
704: 44100 or 6000...48000 Hz). Check what your sound card supports.</p>
705: <p class="parameter">−−sound-buffer-size
706: <x></p>
707: <p class="paramdesc">SDL’s sound buffer size: 10-100,
708: or 0 to use default buffer size. By default Hatari uses an SDL
709: buffer size of 1024 samples, which gives approximatively 20-30 ms
710: of sound depending on the chosen sound frequency. Under some OS or
711: with not fully supported sound card, this default setting can cause
712: a bigger delay at lower frequency (nearly 0.5 sec). In that case,
713: you can use this option to force the size of the sound buffer to a
714: fixed number of milliseconds of sound (using 20 is often a good
715: choice if you have such problems). Most users will not need this
716: option.</p>
717: <p class="parameter">−−sound-sync
718: <bool></p>
719: <p class="paramdesc">The emulation rate is nudged by +100 or
720: 0 or -100 micro-seconds on occasion. This prevents the sound buffer
721: from overflowing (long latency and lost samples) or underflowing
722: (short latency and repeated samples). The emulation rate smoothly
723: deviates by a maximum of 0.58% until synchronized, while the
724: emulator continuously generates every sound sample and the crystal
725: controlled sound system consumes every sample.<br />
726: (on|off, off=default)</p>
727: <p class="parameter">−−ym-mixing
728: <x></p>
729: <p class="paramdesc">Select a method for mixing the three
730: YM2149 voice volumes together. "model" uses a mathematical model of
731: the YM voices, "table" uses a lookup table of audio output voltage
732: values measured on STF and "linear" just averages the 3 YM
733: voices.</p>
734:
735: <h3>Debug options</h3>
736: <p class="parameter">−D,
737: −−debug</p>
738: <p class="paramdesc">Toggle whether CPU exceptions invoke
739: the debugger</p>
740: <p class="parameter">−−bios-intercept</p>
741: <p class="paramdesc">Toggle Bios/XBios call interception needed for
742: Bios/XBios call tracing. Allows Atari programs to modify Hatari state
743: through XBios 255 calls which are processed as Hatari commandline
744: arguments. Atari printscreen call takes also Hatari screenshot.</p>
745: <p class="parameter">−−conout <device></p>
746: <p class="paramdesc">Enable console (xconout vector functions) output
747: redirection for given <device> to host terminal. Device 2 is for
748: the (CON:) VT52 console, which vector function catches also EmuTOS panic
749: messages and MiNT console output, not just normal BIOS console output.</p>
750: <p class="parameter">−−disasm <x></p>
751: <p class="paramdesc">Set disassembly options. 'uae' and 'ext' select
752: the dissasembly engine to use, bitmask sets output options for the
753: external disassembly engine and 'help' lists them.</p>
754: <p class="parameter">−−natfeats <bool></p>
755: <p class="paramdesc">Enable/disable (basic) Native Features support.
756: E.g. EmuTOS uses it for debug output.</p>
757: <p class="parameter">−−trace
758: <trace1,...></p>
759: <p class="paramdesc">Activate debug traces, see
760: −−trace help for available tracing options</p>
761: <p class="parameter">−−trace-file
762: <file></p>
763: <p class="paramdesc">Save trace output to <file>
764: (default=stderr)</p>
765: <p class="parameter">−−parse
766: <file></p>
767: <p class="paramdesc">Parse/execute debugger commands from
768: <file></p>
769: <p class="parameter">−−saveconfig</p>
770: <p class="paramdesc">Save Hatari configuration and exit.
771: Hatari UI needs Hatari configuration file to start, this can be
772: used to create it automatically.</p>
773: <p class="parameter">−−no-parachute</p>
774: <p class="paramdesc">Disable SDL parachute to get Hatari
775: core dumps. SDL parachute is enabled by default to restore video
776: mode in case Hatari terminates abnormally while using non-standard
777: screen resolution.</p>
778: <p class="parameter">−−control-socket
779: <file></p>
780: <p class="paramdesc">Hatari reads options from given socket
781: at run-time</p>
782: <p class="parameter">−−log-file
783: <file></p>
784: <p class="paramdesc">Save log output to <file>
785: (default=stderr)</p>
786: <p class="parameter">−−log-level
787: <x></p>
788: <p class="paramdesc">Log output level
789: (x=debug/todo/info/warn/error/fatal)</p>
790: <p class="parameter">−−alert-level
791: <x></p>
792: <p class="paramdesc">Show dialog for log messages above
793: given level</p>
794: <p class="parameter">−−run-vbls
795: <x></p>
796: <p class="paramdesc">Exit after X VBLs</p>
797:
798: <p>Type <span class="commandline">hatari --help</span> to list all
799: the command line options supported by a given version of Hatari.</p>
800:
801:
802: <h2>Using the emulated system</h2>
803:
804: <p> Once you've started Hatari succesfully, you can use the emulator as
805: an almost complete Atari ST computer system. </p>
806:
807: <h3>The GUI</h3>
808:
809: <p>Press <span class="key">F12</span> to enter the GUI. Navigate it
810: with the mouse.
811: The GUI is rather self explanatory.</p>
812:
813: <h4 class="gui">The Main Menu</h4>
814:
815: <div class="floatimage">
816: <img src="images/main.png" width="500" height="304"
817: alt="Hatari's GUI - the main menu" />
818: </div>
819:
820: <p>
821: You can reach the other setup dialogs from the main menu by clicking on
822: the appropriate buttons.
823: </p>
824: <p>
825: You can load the current settings from a configuration file by clicking
826: on <span class="button">Load config.</span> and you can save
827: the current settings to a configuration file by clicking on
828: <span class="button">Save config.</span>.
829: </p>
830: <p>
831: Click <span class="button">OK</span> to go back and continue the emulation.
832: All changed options will be applied.
833: </p>
834: <p>
835: Select the <span class="button">Reset machine</span> option if you
836: want the emulated machine to perform a cold reset. This is equal to
837: switching the power off and on again on a real Atari machine.
838: </p>
839: <p>
840: Click <span class="button">Quit</span> to terminate Hatari
841: and return to the host OS.
842: </p>
843: <p>
844: Click <span class="button">Cancel</span> to abandon any
845: changes that you have made.
846: </p>
847:
848:
849: <h4 class="gui">The File Selector Dialog</h4>
850:
851: <div class="floatimage">
852: <img src="images/fileselector.png" width="640" height="399"
853: alt="Hatari's GUI - the fileselector" />
854: </div>
855:
856: <p>
857: The file selector dialog appears whenever you are prompted to choose a file
858: or folder.
859: </p>
860: <p>
861: To enter a folder or choose a file, simply click on the entry in the
862: main box of the dialog. To navigate in the file list, you can use the
863: scrollbar on the right with mouse, or use keyboard up + down arrow,
864: page up + down, Home and End keys.
865: </p>
866: <p>
867: You can use the three buttons in the upper right corner for additional folder
868: navigation. Click the <span class="button">..</span> button to go up one level
869: in the directory tree. Click the <span class="button">~</span> button to return
870: to your home directory. The <span class="button">/</span> button can be clicked
871: to go to the root directory of the file system.
872: </p>
873:
874:
875: <h4 class="gui">The System Dialog</h4>
876:
877: <div class="floatimage">
878: <img src="images/system.png" width="360" height="384"
879: alt="Hatari's GUI - the system dialog" />
880: </div>
881:
882: <p>
883: First you can select the CPU type here. Here are some important hints for
884: choosing the correct CPU type:
885: </p>
886: <ul>
887: <li>
888: Atari ST and STE have only been shipped with a 68000 CPU, so for best
889: compatibility with old programs, choose this CPU type.
890: </li>
891: <li>
892: Atari TT and Falcon computers were using the 68030 CPU, so you should switch
893: use 68EC030+FPU (Hatari doesn't support 030 with MMU and some programs don't
894: work with 68020 so this is the best choice).
895: </li>
896: <li>
897: TOS 1.0x only works with 68000, while TOS 3.0x and 4.0x work only with a CPU
898: >= 68020.
899: </li>
900: <li>
901: 68010 and 68040 have never been used in official Atari computers, so don't
902: use these CPU types unless you've got some good reasons.
903: </li>
904: </ul>
905:
906: <p>
907: Beside the CPU type, you can also choose the machine type to emulate.
908: The ST was the very first 16/32-bit computer from Atari. Most older games
909: and demos require an ST. The STE was introduced some years later and had
910: some more advanced hardware features. There are not that many demos or
911: games that really require an STE but since most normal ST games/demos also
912: work with an STE, it's normally safe to always work in STE mode.
913: <br />
914: TT and Falcon are more advanced, but they are not as compatible to the ST as
915: the STE was. Therefore many old games and demos do not work with these machine
916: types anymore. There were only very few programs that were made for the TT
917: exclusively, while there were some interesting games and demos specially made
918: for the Falcon.
919: <em>Note:</em> TT and Falcon emulation are incomplete. They may not work
920: very well.
921: </p>
922: <p>
923: For STE emulation a STE compatible TOS image, e.q. version 1.06, 1.62 or
924: 2.x, is strongly recommended. For TT emulation you need TOS 3.0x and for Falcon
925: emulation you need TOS 4.0x. EmuTOS can be used on all machine types.
926: </p>
927: <p>
928: Select the CPU clock you want to use. 8Mhz is ST standard and the most
929: compatible. Use 16MHz for Mega STE and Falcon emulation.
930: The CPU in the TT was clocked with 32 MHz.
931: </p>
932: <p>With the "Slower but more compatible CPU" option, you can enable
933: the emulation of 68k address errors and the CPU prefetch buffer. This is needed
934: for best compatibility, but it slows down emulation a little bit so you can
935: disable it if you don't need it.</p>
936: <p>
937: For Falcon mode, you can choose whether you want to enable DSP emulation,
938: fake it or completely disable it. Most Falcon programs only play sound or work
939: correctly when you enable the DSP emulation, but it needs a lot of host CPU
940: power (more than 2 GHz). So if you have a slow host CPU, you can try if your
941: Falcon program also runs with DSP disabled or in "dummy" mode.
942: Note that you can not change this option while the DSP based program already
943: runs.
944: </p>
945: <p>You can also enable/disable Blitter emulation here. The Blitter is a custom
946: chip that accelerates some graphical operations. This switch only toggles the
947: Blitter in plain ST mode. In STE mode, the Blitter is always enabled (since all
948: STEs have been sold with a Blitter chip).</p>
949: <p>If you enable the "Real time clock emulation" switch, a RTC will
950: be emulated based on the time of the host computer. Note that you need at least
951: TOS 1.02 for proper RTC emulation, TOS 1.00 does not support this.</p>
952: <p>The Timer-D patch changes the Timer-D initialization from TOS. TOS uses
953: the MFP timer D as a baudrate generator for RS232. However, the TOS default
954: value slows down the emulation. The patch gives you a better performance.
955: It is normally safe to enable the patch, but if you encounter a program that
956: does not work, you can try to disable the patch to see if it works then.</p>
957: <p><em>NOTE:</em> The emulated Atari is very very sensitive to these options
958: and it is strongly recommended to reset the emulation after changing
959: them (for most things that's done automatically). The correct
960: CPU type and clock are automatically selected when one uses the
961: <span class="commandline">--machine</span> command line option.</p>
962:
963:
964: <h4 class="gui">The Floppy Disks Dialog</h4>
965:
966: <div class="floatimage">
967: <img src="images/floppydisks.png" width="640" height="320"
968: alt="Hatari's GUI - the floppy disks dialog" />
969: </div>
970:
971: <p>
972: This dialog can be used to choose which floppy disks should be emulated
973: in the disk drives. You can use most standard Atari ST disk image files.
974: You may select and browse also zipped disk images. See the chapter
975: <a href="#Floppy_disk_images">"Floppy disk images"</a> for details.
976: </p>
977: <p>
978: Click on the button <span class="button">Browse</span> next to the
979: A: and B: option to go to the fileselector to choose a disk image for the
980: corresponding drive.
981: </p>
982: <p>Click on <span class="button">Eject</span> to eject a disk image
983: from the emulated drive. The emulated ST will act as if had no floppy
984: disk in its drive.</p>
985: <p>You can specify a default directory where Hatari will start to
986: browse the filesystem.</p>
987: <p>
988: Check the "Auto insert B" option if you want Hatari to be smart and
989: insert the second disk of a two disk game automatically.
990: Some games then use the second drive automatically.
991: In the case that a game is not able to find the disk in the second drive,
992: you have to insert the second disk in drive A: manually when prompted.
993: <br />
994: <em>NOTE:</em> This option only works properly if the file name of the
995: first disks ends with an 'a' before the extension and the second disk name
996: ends with a 'b'.
997: </p>
998: <p>
999: Select if you want to use fast FDC (Floppy Disk Controller) emulation.
1000: "Fast floppy access" will speed up disk accesses, but this could give
1001: some incompatibilities with some programs that expect correct delays.
1002: (some games/demos don't expect data to be read too fast from the disk)
1003: </p>
1004: <p>
1005: You can choose if you want Hatari to write-protect your disks. Atari ST virii
1006: can spread on disk images too so it might be a good idea to enable the write
1007: protection option. However you can't save highscores or games to your disk
1008: images in that case.
1009: </p>
1010:
1011: <div class="floatimage">
1012: <img src="images/newfloppy.png" width="290" height="224"
1013: alt="Hatari's GUI - the new floppy dialog" />
1014: </div>
1015:
1016: <p>
1017: If you need to create a new blank disk image, click on
1018: <span class="button">Create blank image</span>.
1019: Parameters for the new image can be set in the following dialog.
1020: HD and ED disk sector counts are for non-Atari disk sizes, but such
1021: disks are useful for programs that don't work with GEMDOS emulation.
1022: Click on <span class="button">Create</span> to save the new image or on
1023: <span class="button">Back</span> to return to the disk dialog.
1024: </p>
1025: <p>
1026: After clicking <span class="button">Create</span>, a fileselector
1027: appears. You can browse the filesystem now. Select the target directory,
1028: click beside "File:" and type in a name for the new disk image.
1029: The name should terminate with .st or .msa.
1030: </p>
1031: <p>
1032: Hatari can currently create plain .ST and .MSA disk images exclusively.
1033: <span class="commandline">hmsa</span> command line utility can be used
1034: to convert disk images between .ST and .MSA formats.
1035: </p>
1036:
1037:
1038: <h4 class="gui">The Hard Disks Dialog</h4>
1039:
1040: <div class="floatimage">
1041: <img src="images/harddisks.png" width="640" height="304"
1042: alt="Hatari's GUI - the hard disks dialog" />
1043: </div>
1044:
1045: <p>
1046: This dialog can be used to change the harddisk settings.
1047: </p>
1048: <p>
1049: You can select a harddrive image for ACSI, IDE master or slave hard drive
1050: emulation via image file here or you may select a directory of your local
1051: filesystem to be emulated as the harddrive of the emulated system.
1052: </p>
1053: <p>
1054: Check "Boot from HD" if you want Hatari to execute the AUTO folder
1055: on the harddrive.
1056: This option is checked by default if you specify a harddrive image or
1057: a directory via the command line.
1058: </p>
1059: <p>
1060: Removing the check from the "Allow GEMDOS drive modification" option
1061: will prevent Atari programs from modifying the files in GEMDOS HDD
1062: emulation directory or creating new files under it.
1063: </p>
1064: <p>
1065: Note that for IDE hard disk emulation you also need a TOS version >= 2.05.
1066: And ACSI hard disk emulation does not work with TOS 4.0x in Falcon mode.
1067: </p>
1068:
1069:
1070: <h4 class="gui">The Memory Dialog</h4>
1071:
1072: <div class="floatimage">
1073: <img src="images/memory.png" width="398" height="349"
1074: alt="Hatari's GUI - the memory dialog" />
1075: </div>
1076:
1077: <p>You can select the amount of RAM for the emulated ST here. Only
1078: amounts that were valid on a real unmodified STFM can be selected.</p>
1079: <p><em>Note:</em> This option is critical and you are strongly advised
1080: to reset the emulated ST
1081: when changing this option.</p>
1082: <p>Here you will find the options to save memory snapshots as well.</p>
1083: <p>Click on <span class="button">Save</span> to save a memory snapshot
1084: to file. You can select a new filename here.</p>
1085: <p>Click on <span class="button">Restore</span> to restore a memory
1086: snapshot from a file. Use the fileselector to select the snapshot to be
1087: restored.</p>
1088: <p><em>NOTE:</em> Memory snapshots are not interchangeable between
1089: different versions of Hatari. E.q. if you compile a newer Hatari, you
1090: cannot load your old memory snapshots back.</p>
1091:
1092:
1093: <h4 class="gui">The ROM Dialog</h4>
1094:
1095: <div class="floatimage">
1096: <img src="images/tos.png" width="519" height="367"
1097: alt="Hatari's GUI - the ROM dialog" />
1098: </div>
1099:
1100: <p>Here you can select the TOS image to use. Click on <span
1101: class="button">Browse</span> to select it via the fileselector.
1102: You can also select an optional cartridge image to use. Click on <span
1103: class="button">Browse</span> to select one via the fileselector. Click on <span
1104: class="button">Eject</span> to disconnect the custom cartridge image.
1105: </p>
1106: <p>
1107: For ST mode, use TOS 1.00, 1.02, 1.04 or 2.06.
1108: For STE mode, use TOS 1.06, 1.62, 2.05 or 2.06.
1109: If you want to use the TT mode, you must specify a TOS 3.06 image here.
1110: And in Falcon mode, you have to use either TOS 4.00, 4.02, 4.04 or 4.92.
1111: However, you should always use TOS 4.04 for Falcon mode, it's the most common one.
1112: Also note that TOS 4.92 can not be booted from a boot disk (like it's done on a
1113: real Falcon), you have to specify it directly in the TOS ROM setup dialog here.
1114: </p>
1115: <p>
1116: Keep in mind that any custom cartridge image will not work together with
1117: GEMDOS hard disk emulation or the VDI extended resolution emulation
1118: since some additional driver code will be used in the cartridge memory
1119: space for these emulations.
1120: </p>
1121: <p>
1122: <em>Note:</em> These options are critical and you are strongly
1123: advised to reset the emulated ST
1124: when changing one of these option.
1125: </p>
1126:
1127:
1128: <h4 class="gui">The Joystick Dialog</h4>
1129:
1130: <div class="floatimage">
1131: <img src="images/joystick.png" width="320" height="288"
1132: alt="Hatari's GUI - the joystick dialog" />
1133: </div>
1134:
1135: <p>In this dialog, you can configure the emulated joysticks.
1136: With the upper two arrows, you can choose the joystick which you want to
1137: configure.</p>
1138: <p>Joystick 1 is the normal ST joystick port and 99.9% of all ST games
1139: use this port.
1140: Joystick 0 emulates a joystick plugged into the ST mouse port
1141: and is often used in games for two players.</p>
1142: <p>With STE joypad A and B, you can enable the emulation of Jaguar joypads
1143: which are plugged in the enhanced joystick ports of the Atari STE.
1144: Only very few STE games support these joypads, so you often won't need this.</p>
1145: <p>Finally, Hatari also emulates joysticks which were plugged on the parallel
1146: port with a special adapter on a real ST. These were used in some few
1147: multi-player games like "Gauntlet 2".</p>
1148: <p>For each ST joystick, choose whether you want to disable it,
1149: use the keyboard for emulation or use a real PC joystick.</p>
1150: <p>For keyboard emulation, you can select the keys by pressing the
1151: <span class="button">Define keys</span> button. You will be prompted to press
1152: the keys for up, down, left, right and fire.</p>
1153: <p>If you want to use a real PC joystick for the emulation, you should connect
1154: it to your PC before you start Hatari. Then you can choose the joystick with
1155: the two lower arrows.</p>
1156: <p>Check the "Enable autofire" option if you are too lazy to pound
1157: on the fire button in shoot'em-up games. However, this option only works with
1158: certain games. In some other games, it gets worse if you enable this option.</p>
1159: <p>See also the chapter "Emulated Joystick" for details.</p>
1160:
1161:
1162: <h4 class="gui">The Atari Monitor Dialog</h4>
1163:
1164: <div class="floatimage">
1165: <img src="images/monitor.png" width="340" height="304"
1166: alt="Hatari's GUI - the Atari monitor dialog" />
1167: </div>
1168:
1169: <p>
1170: Here you control the video output of the emulated Atari.
1171: </p>
1172: <p>
1173: You can select which sort of monitor to use. This option depends on
1174: the machine type which you have selected in the "System options"
1175: dialog. In ST and STE mode, you can choose between monochrome mode
1176: (select "Mono") and color mode (select one of the other monitor types).
1177: Note that when you select "TV" and use zoomed low resolution or
1178: switch to ST medium resolution, you will get a TV-like screen rendering
1179: which is a little bit faster but darker compared to the normal "RGB"
1180: monitor mode. Switching between mono and a color monitor acts like a monitor
1181: switch on a real ST - so beware, this will reboot your emulated system!<br />
1182: In TT mode, you can only choose between TT-high resolution ("Mono")
1183: and normal modes (select one of the other monitor types).
1184: Finally the Falcon mode supports all four types of monitors. Note that most
1185: Falcon demos/games require a RGB or TV mode and do not work with VGA.
1186: </p>
1187: <p>
1188: "Show ST/STE borders" toggles the displaying of the borders around the ST /
1189: STE screen. Some demos and games use the screen borders for displaying
1190: additional graphics. As enabling this option increases CPU computing time,
1191: don't enable it if you have a very slow computer.
1192: This option affects only the ST and STE modes, TT and Falcon modes are
1193: always displayed without borders.
1194: </p>
1195: <p>
1196: Extended VDI resolutions will emulate a sort of extended graphics
1197: card in the emulated ST which give you larger resolutions with a higher
1198: colordepth for GEM. Select a resolution and color depth. Check to
1199: activate. It will disable all other video options mentioned above.
1200: Uncheck to get back to a normal ST behaviour.<br />
1201: <em>Note:</em> Using an extended resolution will only work with GEM
1202: conformant applications. 99% of all games and demos will not run if you
1203: activate any extended resolution here.
1204: </p>
1205:
1206:
1207: <h4 class="gui">The Hatari Screen Dialog</h4>
1208:
1209: <div class="floatimage">
1210: <img src="images/screen.png" width="520" height="320"
1211: alt="Hatari's GUI - the Hatari screen dialog" />
1212: </div>
1213:
1214: <p>
1215: Here you control how the video output of the emulated Atari appears
1216: on your screen.
1217: </p>
1218:
1219: <p>
1220: Check "Fullscreen" to run Hatari in fullscreen. By default Hatari
1221: runs in windowed mode.
1222: </p>
1223: <p>
1224: The "Frame Skip" option can be used to speed up the emulator
1225: if it is running too slow on your system. Disable frame-skip if you have
1226: a fast computer. When selecting 1, 2 or 4, drawing of corresponding number
1227: of frames will be skipped after each frame actually shown by Hatari.
1228: Select "Auto" to let the emulator to decide whether, and
1229: how many frames will be skipped.<br />
1230: <em>Note:</em> The frameskip option also affects the frame rate of the
1231: screen animation recording!
1232: </p>
1233: <p>
1234: Indicators that you can have on the Hatari window:
1235: </p>
1236: <ul>
1237: <li>"Statusbar" at the bottom of the screen.
1238: The statusbar shows the floppy drive LEDs, the current frameskip value,
1239: the machine type including TOS version and memory size, and whether
1240: recording is currently active.</li>
1241: <li>"Drive led" is a colored rectangle shown on top of the Hatari window
1242: contents. It will show any disk (floppy or hard disk) activity.</li>
1243: <li>"None" turns both of above options off.</li>
1244: </ul>
1245: <p>
1246: "Keep desktop resolution" option will use your desktop resolution
1247: for fullscreen to avoid issues related to resolution switching,
1248: especially on LCD monitors (they're slow). If this isn't enabled,
1249: values from the "Max zoomed win" option are used in selecting
1250: a suitable resolution.
1251: <p>
1252: "Max zoomed win" option controls up to which size Hatari tries to scale
1253: the Atari resolutions and how much of the borders (enabled in Atari
1254: Monitor dialog) will be shown. Note that there are several limitations
1255: in this and the "Keep desktop resolution" option, partly because Hatari
1256: has different implementations for different video modes:
1257: </p>
1258: <ul>
1259: <li>VDI resolutions (selectable in Atari Monitor dialog) aren't scaled.</li>
1260: <li>ST and STE video emulation supports only doubling of the ST-low
1261: resolution.</li>
1262: <li>Hatari doesn't support downscaling. If the original Atari resolution
1263: is larger than the specified size (e.g. TT-high), the Hatari screen
1264: size will also be larger than requested. Hatari Falcon/TT window size
1265: will be limited to the Desktop size though.</li>
1266: <li>TT and Falcon resolutions support only <em>integer</em> scaling ratios.
1267: If the scaling ratio cannot match the requested size exactly, Hatari
1268: will use a ratio that will produce smaller size closest to the
1269: requested one.</li>
1270: </ul>
1271: <p>
1272: You should set these values to a size that suits best your monitor
1273: resolution. It's intended to help in getting Hatari to best use your
1274: monitor space on a windowed mode and in fullscreen avoiding "fuzzy"
1275: scaling done by your LCD monitor.
1276: </p>
1277: <p>
1278: Giving "-z 2" option on command line will reset max zoomed size to
1279: default values and "-z 1" will disable all zooming.
1280: Note that zooming takes additional CPU computing time and should
1281: not be enabled on very slow computers.
1282: </p>
1283: <p>Click the <span class="button">Screenshot</span> button to create
1284: a screenshot in PNG (or BMP) format to the current working directory
1285: or click the <span class="button">Record AVI</span> button to
1286: record an AVI format video of Hatari screen (and audio) output.
1287: </p>
1288: <p>
1289: Selecting "Crop statusbar" option will leave statusbar out from
1290: the screenshots and recorded videos.
1291: </p>
1292:
1293: <h4 class="gui">The Keyboard Dialog</h4>
1294:
1295: <div class="floatimage">
1296: <img src="images/keyboard.png" width="459" height="223"
1297: alt="Hatari's GUI - the keyboard dialog" />
1298: </div>
1299:
1300: <p>Here you can select the keyboard mapping to use. Two different mappings
1301: called "Symbolic" and "Scancode" are predefined.</p>
1302: <p>"Symbolic" tries to map the symbolic values of your PC keys
1303: to the ST keys. It should be working pretty good on all systems as long
1304: as your keyboard layout looks close to the standard english keyboard
1305: layout. However, you might experience some problems with special keys like
1306: brackets etc.</p>
1307: <p>"Scancode" uses the scancode values of your PC keys for keyboard
1308: mapping. This only works on certain architectures like Linux where the
1309: scancodes are similar to the ST scancodes (e.g. it does not work on Mac OS X).
1310: If it works on your system, this often gives better results than the symbolic
1311: mapping. Note that you also need a TOS version with the right language
1312: (e.g. use a French TOS if you are using a French keyboard).</p>
1313: <p>You can also load a custom keyboard mapping file here if you wish. Please
1314: note that the custom keyboard mapping will use the "symbolic"
1315: mapping for all keys that are not defined by your map file. Have a look
1316: at the supplied example mapfile (keymap-sample.txt) to see how to create
1317: your own keyboard mapping.</p>
1318: <p>
1319: When the emulator runs in fast forward mode, and you want to type text,
1320: it can be annoying that the emulated system detects multiple key events
1321: due to the key repetition of the emulated system. To avoid this you can
1322: disable the key repetition in fast forward mode here.
1323: </p>
1324:
1325:
1326: <h4 class="gui">The Sound Dialog</h4>
1327:
1328: <div class="floatimage">
1329: <img src="images/sound.png" width="400" height="400"
1330: alt="Hatari's GUI - the sound dialog" />
1331: </div>
1332:
1333: <p>Here you can control the sound subsystem.</p>
1334: <p>Check "Enabled" if you want emulated sound at all. Emulation is faster if
1335: sound emulation is turned off.</p>
1336: <p>If you experiment latency issues with your OS audio's output, you
1337: can check the "Synchronize" option to adjust Hatari's video emulation to match
1338: your OS audio.</p>
1339: <p>
1340: Nine frequencies from low to high quality are available. Experiment a
1341: little bit to find out which fits best for your setup.
1342: For most modern computers, 44100 Hz or 48000 Hz should be fine.
1343: For older or slower host systems, you should use a lower frequency.
1344: 12517, 250033 and 50066 Hz are frequencies supported by
1345: the STE/TT/Falcon sound DMA.
1346: </p>
1347: <p>
1348: YM voices volume mixing "ST table" method uses a lookup table of audio output
1349: voltage values measured on STF, "Math model" uses a complex model to mix the
1350: 3 YM voices and "Linear" just averages the 3 YM voices. Use "ST table" or "Math model"
1351: for accurate sound's emulation.
1352: </p>
1353: <p>
1354: You can select to record a piece of sound here.
1355: Use the <span class="button">Browse</span> button to choose a file.
1356: The file name extension that you use (.WAV or .YM) determines in which format
1357: the sound is recorded in. The <span class="button">Record sound</span> button
1358: is a toggle so you will need to return to the GUI to switch sound recording off
1359: again (or to use the keyboard shortcut for that).
1360: </p>
1361:
1362:
1363: <h4 class="gui">The Devices Dialog</h4>
1364:
1365: <div class="floatimage">
1366: <img src="images/devices.png" width="520" height="383"
1367: alt="Hatari's GUI - the device dialog" />
1368: </div>
1369:
1370: <p>Check the first checkmark to enable printer support.
1371: See the <a href="#Emulated_printer">Emulated printer</a> section for
1372: details.</p>
1373:
1374: <p>As Hatari currently only supports printing to file, click on <span
1375: class="button">Browse</span> to select the file to print to. You can
1376: enter a new filename as well.</p>
1377: <p>Check the second checkmark to enable RS232 support.
1378: The RS232 device is configured according to the settings of
1379: the emulated RS232 of the Atari ST. This means Hatari will
1380: automatically use baudrate and handshaking as configured for the
1381: emulated ST.</p>
1382: <p>Click on <span class="button">Browse</span> to select suitable
1383: device files for serial input and output. On Linux a good choice is
1384: /dev/ttyS0 or /dev/ttyS1.
1385: </p>
1386: <p>Check the third checkmark to enable MIDI support.
1387: Click on <span class="button">Browse</span> to select a suitable
1388: MIDI device files for MIDI input and output.</p>
1389: <p><span class="file">midi-linux.txt</span> file explains how to
1390: select the correct MIDI device file, how to set up software sound
1391: synthetizing on Linux (using Alsa) if your sound card/driver doesn't
1392: support MIDI, and how to set up MIDI networking e.g. between multiple
1393: Hatari instances.
1394: </p>
1395:
1396:
1397: <h3 class="clearboth">Keyboard shortcuts</h3>
1398:
1399: <p> While the emulator is running, you can activate or toggle various
1400: features via Hatari keyboard shortcuts. Below are listed the default
1401: shortcut key bindings:</p>
1402: <table border="1" class="keytable">
1403: <thead>
1404: <tr class="backdropped">
1405: <th>Shortcut</th>
1406: <th>Purpose</th>
1407: </tr>
1408: </thead>
1409: <tbody>
1410: <tr>
1411: <td><span class="key">ALTGR+a</span></td>
1412: <td>record animation</td>
1413: </tr>
1414: <tr>
1415: <td><span class="key">ALTGR+g</span></td>
1416: <td>grab a screenshot</td>
1417: </tr>
1418: <tr>
1419: <td><span class="key">ALTGR+i</span></td>
1420: <td>boss key: leave full screen mode, pause Hatari
1421: and iconify its window</td>
1422: </tr>
1423: <tr>
1424: <td><span class="key">ALTGR+j</span></td>
1425: <td>toggle joystick emulation via cursor keys
1426: on/off between ports 0 and 1</td>
1427: </tr>
1428: <tr>
1429: <td><span class="key">ALTGR+m</span></td>
1430: <td>(un-)lock the mouse into the window</td>
1431: </tr>
1432: <tr>
1433: <td><span class="key">ALTGR+r</span></td>
1434: <td>(warm) reset the ST</td>
1435: </tr>
1436: <tr>
1437: <td><span class="key">ALTGR+c</span></td>
1438: <td>coldreset the ST (same as the original power switch)</td>
1439: </tr>
1440: <tr>
1441: <td><span class="key">ALTGR+d</span></td>
1442: <td>open dialog to select/change disk A</td>
1443: </tr>
1444: <tr>
1445: <td><span class="key">ALTGR+s</span></td>
1446: <td>enable/disable sound</td>
1447: </tr>
1448: <tr>
1449: <td><span class="key">ALTGR+q</span></td>
1450: <td>quit the emulator</td>
1451: </tr>
1452: <tr>
1453: <td><span class="key">ALTGR+x</span></td>
1454: <td>toggle normal speed/fast forward</td>
1455: </tr>
1456: <tr>
1457: <td><span class="key">ALTGR+y</span></td>
1458: <td>enable/disable sound recording</td>
1459: </tr>
1460: <tr>
1461: <td><span class="key">ALTGR+k</span></td>
1462: <td>save memory snapshot</td>
1463: </tr>
1464: <tr>
1465: <td><span class="key">ALTGR+l</span></td>
1466: <td>load memory snapshot</td>
1467: </tr>
1468: <tr>
1469: <td><span class="key">ALTGR+f or F11</span></td>
1470: <td>toggle between fullscreen and windowed mode</td>
1471: </tr>
1472: <tr>
1473: <td><span class="key">ALTGR+o or F12</span></td>
1474: <td>activate the options GUI</td>
1475: </tr>
1476: <tr>
1477: <td><span class="key">PAUSE</span></td>
1478: <td>pause emulation</td>
1479: </tr>
1480: <tr>
1481: <td><span class="key">AltGr+PAUSE</span></td>
1482: <td>invoke the internal Hatari debugger</td>
1483: </tr>
1484: </tbody>
1485: </table>
1486:
1487: <p>You can change the key bindings from the Hatari configuration file.
1488: The required key values can be seen in the SDL_keysym.h include file
1489: (usually in /usr/include/SDL/).</p>
1490:
1491:
1492: <h3>Emulated Atari ST keyboard</h3>
1493:
1494: <p> All other keys on the keyboard act as the normal Atari ST keys so
1495: pressing SPACE on your PC will result in an emulated press of the SPACE
1496: key on the ST. The following keys have special meanings: </p>
1497:
1498: <table class="keytable">
1499: <thead>
1500: <tr class="backdropped">
1501: <th>Key</th>
1502: <th>Meaning</th>
1503: </tr>
1504: </thead>
1505: <tbody>
1506: <tr>
1507: <td><span class="key">Alt</span></td>
1508: <td>will act as the ST's ALTERNATE key</td>
1509: </tr>
1510: <tr>
1511: <td><span class="key">left CTRL</span></td>
1512: <td>will act as the ST's CONTROL key</td>
1513: </tr>
1514: <tr>
1515: <td><span class="key">Print Screen</span></td>
1516: <td>will emulate the ST's HELP key</td>
1517: </tr>
1518: <tr>
1519: <td><span class="key">Scroll Lock</span></td>
1520: <td>will emulate the ST's UNDO key</td>
1521: </tr>
1522: <tr>
1523: <td><span class="key">Page Up</span></td>
1524: <td>will emulate the ST's ( key in the keypad</td>
1525: </tr>
1526: <tr>
1527: <td><span class="key">Page Down</span></td>
1528: <td>will emulate the ST's ) in the keypad</td>
1529: </tr>
1530: </tbody>
1531: </table>
1532:
1533: <p>If joystick emulation via keyboard is enabled, by default cursor keys
1534: are used for the directions and <span class="key">right CTRL</span> key
1535: as the fire button. Otherwise they act as corresponding keys of the emulated
1536: Atari ST.</p>
1537:
1538: <p>NOTE: Problems with simultenous keypresses most likely aren't an
1539: issue in Hatari as many modern keyboards report/support only three
1540: simultenous key presses (or even just two depending on which keys
1541: are in question). Expensive gaming keyboards support more.</p>
1542:
1543:
1544: <h3>Emulated mouse</h3>
1545:
1546: <p>For obvious reasons your PC mouse will act as the emulated Atari ST
1547: mouse. In fullscreen mode it will act as expected, directly controlling
1548: the ST mouse pointer. </p>
1549: <p>However it is a little bit different in windowed mode as
1550: mouse cursor positions between host and emulated Atari can get
1551: out of sync. This can be worked around by constraining the mouse
1552: to the Hatari window. Pressing the <span class="key">ALTGR+m</span>
1553: hotkey combination or starting Hatari with the
1554: <span class="commandline">--grab</span> command line option
1555: grabs the mouse i.e. locks its movements to the Hatari window.
1556: Press the shortcut key (again) to go back to normal mouse behaviour
1557: which allows you to move mouse outside outside the Hatari window while
1558: Hatari is up and running. Note: pausing the emulation will also
1559: (temporarily) release the mouse grab.</p>
1560: <p>Mouse scrollwheel will act as cursor up and down keys. </p>
1561:
1562: <h3>Emulated joystick</h3>
1563:
1564: <p>The Atari ST joysticks are emulated ofcourse allowing you to play
1565: your favourite games with Hatari. </p>
1566: <p>The default mode is to use a connected PC joystick. You can use any
1567: joystick that is supported by your kernel. If your joystick works with
1568: other applications, it will likely work with Hatari as well. Make sure
1569: it is calibrated and then off you go. Move the stick to point into the
1570: desired direction. Please note that Hatari will not detect analogue
1571: movement as the Atari ST only had digital joysticks. The first
1572: firebutton will act as the normal firebutton on the Atari ST while the
1573: second
1574: firebutton will emulate a keypress of the <span class="key">SPACE</span>
1575: key on the ST as many ST
1576: games utilize the SPACE bar for secondary game functions. (Xenon for
1577: example)</p>
1578: <p>If you do not have a PC joystick or joypad, then you do not need to
1579: desperate. You can emulate one of the two Atari ST joysticks via the
1580: cursor keys. Just activate it in the GUI. Then the cursor keys will act
1581: as the joystick directions, the right CTRL key will act as the
1582: firebutton. You can still use the cursor keys as the ST's
1583: cursorkeys in this mode as long as you press <span class="key">SHIFT</span>
1584: along with the cursorkeys. You can also configure these keys from the
1585: joystick options.</p>
1586:
1587: <h3>Emulated video</h3>
1588:
1589: <p>Hatari emulates all screen modes of the original machine.</p>
1590: <p>
1591: ST/STE shifter overscan effects are emulated, but due to the fact
1592: that these effects are achieved by using quirks and glitches in the
1593: original chips to do things beyond their specification, emulation is
1594: a bit tricky for these effects. As a result, some demos using these
1595: techniques might not be displayed correctly in Hatari, known ones are
1596: listed in the <span class="file">compatibility.html</span> file.
1597: </p>
1598: <p>Beside that you can setup extended VDI modes. These only work with
1599: GEM-compliant applications and they are equal to fitting a videocard
1600: into your Mega ST.</p>
1601: <p>Make sure to disable extended VDI modes for playing games as 99% of
1602: all ST games will not be able to make use of higher resolutions.</p>
1603:
1604: <h3 id="Emulated_printer">Emulated printer</h3>
1605:
1606: <p>Due to the fact that printer handling is different on Atari and
1607: current machines, emulation of the printer is achieved by writing all
1608: printer output to a file.</p>
1609: <p>The file will contain a sequence of data, the same that would appear
1610: on the data pins of the Atari ST printer port.
1611: That would include control characters and commands for graphic
1612: printing. Clicking "Print desktop" on the GEM desktop would result
1613: in a messy data dump in the printer output.</p>
1614: <p>Printer emulation works best for plain text files or programs that
1615: do not format the output for a specific printer.
1616: The file contents can be used with your favourite text editor for
1617: further processing and printing to a real printer.</p>
1618: <p>To get real direct printing out of Hatari you may set up a suitable
1619: (e.g. PostScript) GDOS or NVDI printer driver on the emulated Atari and
1620: set your printer device file as Hatari's printer output.<br />
1621: <em>NOTE:</em> If the driver doesn't match or there's some other problem,
1622: this can cause your printer to print out hundreds of pages of garbage.</p>
1623:
1624: <h3>Emulated RS232</h3>
1625:
1626: <p>Serial communications in Hatari is designed to directly use a serial
1627: port on your PC.</p>
1628: <p>Communications parameters are set automatically upon the settings of
1629: the emulated ST. This means all you do is to set
1630: the communication parameters like baudrate from your ST communications
1631: software. Hatari will do the rest and handle
1632: the serial input and output for you.</p>
1633:
1634: <h2 id="Floppy_disk_images">Floppy disk images</h2>
1635:
1636: <p>Hatari does not use floppy disks directly but disk images due to
1637: differences between the floppy disk controllers of the ST and the PC.
1638: Three types of disk images are currently supported: The raw "ST" type,
1639: the similar "DIM" type and
1640: the compressed "MSA" (Magic-Shadow-Archiver) type. </p>
1641: <p> The raw type (file suffix should be "*.st") is simply a sector by
1642: sector image of a real floppy disk. You can easily create such an image
1643: with the <span class="commandline">dd</span> program which should
1644: normally be pre-installed on every
1645: Unix-like system. Simply type something like <span class="commandline">dd
1646: if=/dev/fd0 of=myimage.st</span> to create a disk image. Of course you
1647: need access to
1648: /dev/fd0, and depending on your system and the type of floppy disk you
1649: might have to use another device name here (for example I use
1650: /dev/fd0u720 for 720kB disks). However, if the disk is copy-protected
1651: or
1652: doesn't use a MSDOS compatible file system, this might fail. So be very
1653: careful if you are not sure about the disk format. </p>
1654: <p> The other possibility is to image the disk on a real Atari ST.
1655: There
1656: are programs like the Magic Shadow Archiver for this task. Hatari
1657: supports this slightly compressed MSA disk images, too. Note that
1658: Hatari
1659: only supports the "old" MSA format, there are some Magic Shadow
1660: Archiver
1661: clones (like Jay-MSA) that create better compressed but
1662: Hatari-incompatible disk images. However, if you have got such a MSA
1663: disk and want to use it with Hatari, you can still run the
1664: corresponding
1665: MSA program within Hatari to extract the incompatible disk image to a
1666: normal floppy disk image. </p>
1667: <p> While *.ST and *.MSA are more or less the "standard" types of Atari
1668: disk images, you might sometimes also find STT or ADF images on the
1669: internet. These currently do not work with Hatari. </p>
1670: <p>Hatari can now also utilize *.DIM images just as *.ST ones without
1671: any problems.
1672: Note that DIM images are nearly the same as the raw ST images
1673: (they only have an additional 32 bytes header), so you can easily
1674: transform
1675: the DIM images into ST images by stripping the header from the files.
1676: For example try something like:
1677: <span class="commandline">dd if=input.dim of=output.st bs=32 skip=1</span>
1678: </p>
1679: <p> If you've got a disk image that has been created with the old ST
1680: emulator PaCifiST (for DOS) or with early versions of the program
1681: Makedisk, and the disk image does not work with Hatari, then the disk
1682: probably suffers from the "PaCifiST bootsector bug" (Hatari will
1683: display a
1684: warning message then). In this case, the bootsector of the disk
1685: contains some illegal data, so that the disk even does not work on a
1686: real ST any more. However, if it is a .ST and not a .MSA disk, you can
1687: easily fix it by using a hex-editor to change the byte at offset $D
1688: (13)
1689: from 0 to 1 (don't forget to backup your disk image first, since you
1690: can also easily destroy your disk image when changing a wrong byte
1691: there). If the disk contains a bootsector program, you probably have to
1692: adjust the boot sector check sum, too (it can be found at offset $1FE +
1693: $1FF). </p>
1694: <p>Hatari supports disk images that are compressed with (Pk-)ZIP
1695: (file suffix must be ".zip") or GZip (file suffix must be ".st.gz" or
1696: ".msa.gz"), so you can archive your disk images into zip archives.
1697: You can also directly run the zip archives you may download from the
1698: net as long as the archive contains a disk image in .ST or .MSA format.</p>
1699: <p><em>Note:</em> Hatari does not save disk images back to *.ZIP files
1700: so
1701: your highscores and savegames are lost if you load the game from such
1702: a zipped disk image.</p>
1703:
1704:
1705: <h2>Hard disk support</h2>
1706:
1707: <p>
1708: Hatari supports three ways of emulating Atari hard drives: The low-level
1709: ACSI and IDE hard disk emulation and a GEMDOS based drive emulation.
1710: In most cases the GEMDOS based hard disk emulation is best as it allows
1711: exchanging files easily between the emulated and the host environment.
1712: </p>
1713: <p>
1714: Please note that changing the HD-image or the GEMDOS HD-folder will reset
1715: the emulated Atari since it is not possible to switch the hard disk
1716: while the emulator is running.
1717: </p>
1718: <p>
1719: On a 32-bit host system, the size of a hard disk image is limited to 2 GB.
1720: On 64-bit host systems, bigger images might be possible but the support
1721: for bigger images is not tested very well yet. How large partition sizes
1722: are supported inside the hard disk (images) depends on the TOS version.
1723: TOS 1.0x supports up to 256MB partitions and TOS 4.0x up to 1GB ones.
1724: </p>
1725:
1726:
1727: <h3>GEMDOS based hard disk emulation</h3>
1728: <p>
1729: With the GEMDOS based drive emulation, you can easily "mount" a
1730: folder from the host file system to a drive of the emulated Atari.
1731: </p>
1732: <p>
1733: If you provide Hatari a directory containing only single letter (C-Z)
1734: subdirectories, each of these subdirectories will be treated as a
1735: separate partition, otherwise the given directory itself will be
1736: assigned to drive "C:". In the multiple partition case, the letters
1737: used as the subdirectory names will determine to which
1738: drives/partitions they're assigned.
1739: </p>
1740: <p>
1741: GEMDOS drive emulation is an easy way to share files between the
1742: host system and the emulated Atari, but there are also some known
1743: limitations which are due to the way the GEMDOS drive emulation is
1744: implemented:
1745: </p>
1746: <ul>
1747: <li>Directory entries are returned in a (case-insensitively) sorted
1748: order, for consistency. E.g. moving files to a different directory and
1749: back (without changing their names) like AUTOSORT does, doesn't change
1750: that order. You need to rename the files.</li>
1751: <li>Names which aren't valid TOS directory or file names, are converted
1752: to a valid format. If there are multiple files which converted
1753: names are identical in TOS-format, you see only one of those.</li>
1754: <li>It is not possible to use a cartridge image at the same time
1755: with the GEMDOS drive emulation (Hatari has it's own cartridge code
1756: that is used for GEMDOS emulation).</li>
1757: <li>Anything that installs its own GEMDOS handler, like MiNT, doesn't work
1758: with the GEMDOS drive emulation. Such things need to be run from a real
1759: hard disk image.</li>
1760: <li>GEMDOS drive emulation conflicts with the ACSI and IDE harddisk images.
1761: If you want to use GEMDOS HD emulation and ACSI/IDE disk images together,
1762: use a multiple partition GEMDOS emulation setup and select the partition
1763: subdirectories (letters) so that they don't conflict with the ACSI/IDE
1764: drive partitions (letters). With HD Driver you have also another option,
1765: see <a href="#Using_HD_Driver_with_GEMDOS_partitions">Using HD Driver
1766: with GEMDOS partitions</a>.</li>
1767: <li><em>The GEMDOS drive emulation does not work (very well) with TOS
1768: 1.00 and 1.02</em>. Use at least TOS 1.04 if you want the GEMDOS drive
1769: emulation to work properly.</li>
1770: </ul>
1771: <p>
1772: So, if your programs complain that they could not find/read/write
1773: files on the GEMDOS emulated drive, you should try to copy them to a
1774: floppy disk image or a real hard disk image!
1775: </p>
1776:
1777: <h3 id="ACSI_hard_disk_emulation">ACSI hard disk emulation</h3>
1778: <p>
1779: To use the ACSI hard disk emulation, you need a hard disk image file
1780: with a pre-installed HD driver in it. You can try to get an image of
1781: your old ST hard disk or grab one from the internet (e.g. from the
1782: Hatari website).
1783: </p>
1784: <p>
1785: To create a <em>new</em> ACSI hard disk image, you can start with an empty
1786: image that you have created for example with the following command:
1787: <span class="commandline">dd if=/dev/zero of=hd.img bs=512 count=xxx</span>
1788: (where 'xxx' is size in 512 byte blocks). Copy the complete AHDI 5.0
1789: package to a floppy disk image, then boot Hatari with this floppy disk
1790: image and the fresh hard disk image like this:
1791: <span class="commandline">--acsi hd.img ahdi.st</span>.
1792: Then start HDX.PRG from the floppy disk and format + partition the hard
1793: disk image with it.
1794: </p>
1795: <p>
1796: Formatting and partitioning works currently only with AHDI 5, but you
1797: can install the AHDI 6 driver to the hard disk after it's formatted.
1798: Restart the emulated system, run AHDI.PRG from the floppy disk to access
1799: the hard disk image from the emulated Atari and then run HINSTALL.PRG.
1800: After installing the hard disk driver to the fresh HD image with
1801: HINSTALL.PRG, you can boot directly from the hard disk image.
1802: </p>
1803:
1804: <h3>IDE hard disk emulation</h3>
1805: <p>
1806: <p>
1807: As the IDE disk format (little endian) differs from the ACSI disk format
1808: (big endian), you need separate disk images for them. Hatari doesn't
1809: currently support formatting IDE disks with AHDI, but you can do it with
1810: <em>Cecile</em>.
1811: </p>
1812: <p>
1813: First create an empty image file with the size of your choice with:
1814: <span class="commandline">dd if=/dev/zero of=hd.img bs=1k count=xxx</span>.
1815: Then get the Cecile hard disk driver from
1816: <a href="http://centek.free.fr/atari/softs/s_cecile.htm">http://centek.free.fr/atari/softs/s_cecile.htm</a>
1817: and put it on a floppy disk image (e.g. to one named "cecile.st" using:
1818: <span class="commandline">zip2st.sh cecile.zip</span>).
1819: </p>
1820: <p>
1821: Run Hatari with
1822: <span class="commandline">hatari --machine falcon --tos tos404.rom
1823: --ide-master hd.img cecile.st</span>, switch to larger color resolution
1824: and warm up your French language skills. Then start the Cecile hard
1825: disk driver CECILE.PRG and run CC_TOOLS.APP to partition your hard
1826: disk image. Click the "Partition" button, select "Hatari IDE disk" and set
1827: suitable partition size with the arrows (below type field). Then click
1828: "Valider".
1829: </p>
1830: <p>
1831: If you only want to use your HD image in Falcon mode, you can install
1832: the Cecile hard disk driver to the image from the Cecile CC_TOOLS.APP:
1833: Click the "Installer" button and save the Cecile driver to the
1834: 1st partition on "Hatari IDE disk". If you want to also use your HD
1835: image in ST/STE mode, you need to get and install AHDI 6 driver on it
1836: instead (see <a href="#ACSI_hard_disk_emulation">ASCI hard disk
1837: emulation</a> section).
1838: </p>
1839: <p>
1840: Then you can boot from your hard disk image by simply specifying it
1841: with the <span class="commandline">--ide-master</span> parameter.
1842: </p>
1843:
1844:
1845: <h2>Moving files to/from hard disk images</h2>
1846:
1847: <p>Moving files to and from Atari hard disk images can be done
1848: either through GEMDOS partitions (host directories mounted inside
1849: Hatari emulation) or accessing the images directly on the host
1850: (outside the emulation). Both have their own limitations.</p>
1851:
1852: <p>If it's fine for the IDE/ACSI partitions to be first, you can use
1853: hard disk images with AHDI or Cecile driver and a multipartition
1854: GEMDOS setup as described in above sections. If you want to boot from
1855: a GEMDOS partition i.e. such to be before hard disk image partitions,
1856: and still to be able to access all the IDE/ACSI partitions, you need to
1857: use HD Driver.</p>
1858:
1859: <h3 id="Using_HD_Driver_with_GEMDOS_partitions">Using HD Driver with GEMDOS partitions</h3>
1860:
1861: <p>Uwe Seimet's <a href="http://www.seimet.de/atari/en/hddriver.html">HD
1862: Driver</a> works fine with both the Hatari GEMDOS partitions and normal
1863: hard disk images. However, it doesn't work with EmuTOS so you need
1864: real TOS (at least version v1.04).
1865: </p>
1866:
1867: <p>First copy the HDDRIVER.PRG binary into your GEMDOS drive emulation
1868: directory AUTO folder. Then start the HDDRUTIL.APP configuration utility,
1869: locate HDDRIVER.PRG, open the
1870: <a href="http://www.seimet.de/atari/en/hddriverscreenshots.html">"Devices
1871: and Partitions" dialog</a> and select the "Preserve Existing Partitions"
1872: option. Then you can just start Hatari with your hard disk image and
1873: this GEMDOS directory, for example like this:
1874: "<span class="commandline">hatari --harddisk gemdos-hd/ --ide-master
1875: ide-hd.image</span>".</p>
1876:
1877: <p>If you're using the <em>demo</em> version of HD Driver, you can
1878: write files only to the C: partition, i.e. in above case only copy
1879: files from the hard disk image partition to the GEMDOS partition (with
1880: some write slowndowns included into the demo version). If you want to
1881: copy files to the hard disk image with the <em>demo</em> version of
1882: the HD Driver, you need to set the hard disk image as drive C:.</p>
1883:
1884: <p>To accomplish this, set the GEMDOS partitions to be from D: forward,
1885: i.e. have a directory which contains only single letter subdirectories,
1886: starting from "D" like in "<span class="commandline">mkdir gemdos-hd;
1887: mkdir gemdos-hd/D</span>". Then give Hatari (as the last parameter)
1888: a boot floppy image containing the demo version of HDDRIVER.PRG in
1889: its AUTO folder, like this: "<span class="commandline">hatari
1890: --ide-master ide-hd.image --harddisk gemdos-hd/ hd-driver-floppy.st</span>".
1891: </p>
1892:
1893:
1894: <h3>Accessing HDD image partitions outside of Hatari</h3>
1895:
1896: <p>
1897: If you want to access the harddisk image partitions also outside
1898: the emulation, the disk image needs to have a DOS partition table.
1899: The <span class="commandline">atari-hd-image</span> script included
1900: with Hatari can be used to create such an image.
1901: </p>
1902: <p>
1903: Inside the Hatari emulator, EmuTOS can access partition(s) on these
1904: kind of images directly without any driver software. Of the Atari HD
1905: drivers mentioned above, Centek's Cecile and Uwe Seimet's HD Driver
1906: (demo) work fine with these partitions. E.g. AHDI and CBHD don't.
1907: </p>
1908: <p>
1909: Note that plain EmuTOS supports only ACSI and the listed HD drivers
1910: support only IDE (emulation). Cecile needs TT or Falcon and HD Driver
1911: doesn't work with EmuTOS. To summarise; if ASCI emulation and
1912: EmuTOS are enough, use those. Otherwise, if you want to use TT or
1913: Falcon emulation, use Cecile (or full HD Driver version if you have
1914: it), otherwise use HD Driver (demo).
1915: </p>
1916: <p>
1917: To access the content of the partitions on Linux host, there are two
1918: possibilities:
1919:
1920: <h4>Using Mtools</h4>
1921: <p>
1922: For this you need to add an entry for the hard disk
1923: image to your <span class="commandline">~/.mtoolsrc</span> and
1924: specify which partition you want to access from the image. For
1925: an image created with the above mentioned script, the line in
1926: the configuration file should look something like this:
1927: </p>
1928: <pre>
1929: MTOOLS_NO_VFAT=1
1930: drive c: file="/home/user/hatari/hd.img" partition=1
1931: </pre>
1932: <p>
1933: Note that Mtools is instructed to use FAT compatibility mode because
1934: EmuTOS cannot deal properly with VFAT file information. If you don't
1935: want this setting for all your Mtools drives, you can set it also via
1936: the environment like this ("::" refers to the drive image given with
1937: the "-i" option):
1938: </p>
1939: <pre>
1940: MTOOLS_NO_VFAT=1 mcopy -spmv -i hd.img files/* ::
1941: </pre>
1942:
1943: <h4>Using a loopback device</h4>
1944: <p>
1945: This is recommended even by Mtools documentation, but it's less
1946: convenient as it requires root rights. First you need to "loop"
1947: mount the image:
1948: </p>
1949: <pre>
1950: $ su
1951: # image="hd.img"; mountdir="hd"
1952: # start=$(parted $image unit s print | awk '/ 1 /{print $2}' | tr -d s)
1953: # losetup -f $image -o $((512*$start))
1954: # loop=$(losetup -a | tail -1 | cut -d: -f1)
1955: # mkdir -p $mountdir
1956: # mount -t msdos $loop $mountdir
1957: </pre>
1958: <p>
1959: This uses <span class="commandline">parted</span> to find out the first
1960: partition offset in sectors and then tells <span class="commandline">losetup</span>
1961: to bind the first free loop device to a corresponding offset from
1962: the <span class="commandline">hd.img</span> image.
1963: <span class="commandline">mount</span> is then used to mount the file system
1964: from the loop device on top of the "hd" directory.
1965: </p>
1966: <p>
1967: After you've copied the relevant files to the "hd" directory, you need
1968: unmount the file system and remove the loop device binding before using
1969: the disk image from Hatari:
1970: </p>
1971: <pre>
1972: # umount $mountdir
1973: # losetup -d $loop
1974: </pre>
1975:
1976:
1977: <h2 id="The_debugger">The debugger</h2>
1978:
1979: <p>
1980: Hatari has a built-in debugging interface which can be used for
1981: analyzing code that runs in the emulated system. To invoke the
1982: debugger, press the <span class="key">AltGr + Pause</span>
1983: key combination.
1984: </p>
1985:
1986: <p>
1987: If you start Hatari with the "-D" command line option, certain m68k
1988: exceptions will automatically drop Hatari into the debugger. You can
1989: toggle this also later from the debugger with the "setopt -D"
1990: command.
1991: </p>
1992:
1993: <p>
1994: To run debugger commands at Hatari startup, one can use the "--parse
1995: <file>" command line option. This is useful e.g. for debugging
1996: TOS or some demo startup code, or if you always want to use some
1997: specific debugger setup (breakpoints etc).
1998: </p>
1999:
2000: <p>
2001: The debugger uses Hatari's parent console window, so make sure you run
2002: Hatari from the command line when you want to use the debugger. On
2003: Linux you can add for example an icon to your desktop that does it
2004: with something like this (replace "xterm" with your favorite terminal
2005: program):
2006: </p>
2007: <pre>
2008: xterm -T "Hatari debug window" -e hatari
2009: </pre>
2010:
2011:
2012: <h3>General debugger use</h3>
2013:
2014: <p>
2015: At the debugger prompt, type "help" to get a list of all
2016: the available commands and their shortcuts:
2017: </p>
2018: <pre>
2019: Generic commands:
2020: cd ( ) : change directory
2021: evaluate ( e) : evaluate an expression
2022: help ( h) : print help
2023: history (hi) : show last CPU & DSP PC values & executed instructions
2024: info ( i) : show machine/OS information
2025: lock ( ) : specify information to show on entering the debugger
2026: logfile ( f) : open or close log file
2027: parse ( p) : get debugger commands from file
2028: setopt ( o) : set Hatari command line and debugger options
2029: stateload ( ) : restore emulation state
2030: statesave ( ) : save emulation state
2031: trace ( t) : select Hatari tracing settings
2032: quit ( q) : quit emulator
2033:
2034: CPU commands:
2035: address ( a) : set CPU PC address breakpoints
2036: breakpoint ( b) : set/remove/list conditional CPU breakpoints
2037: disasm ( d) : disassemble from PC, or given address
2038: profile ( ) : profile CPU code
2039: cpureg ( r) : dump register values or set register to value
2040: memdump ( m) : dump memory
2041: memwrite ( w) : write bytes to memory
2042: loadbin ( l) : load a file into memory
2043: savebin ( ) : save memory to a file
2044: symbols ( ) : load CPU symbols & their addresses
2045: step ( s) : single-step CPU
2046: next ( n) : step CPU, proceeding through subroutine calls
2047: cont ( c) : continue emulation / CPU single-stepping
2048:
2049: DSP commands:
2050: dspaddress (da) : set DSP PC address breakpoints
2051: dspbreak (db) : set/remove/list conditional DSP breakpoints
2052: dspdisasm (dd) : disassemble DSP code
2053: dspmemdump (dm) : dump DSP memory
2054: dspsymbols ( ) : load DSP symbols & their addresses
2055: dspprofile (dp) : profile DSP code
2056: dspreg (dr) : read/write DSP registers
2057: dspstep (ds) : single-step DSP
2058: dspnext (dn) : step DSP, proceeding through subroutine calls
2059: dspcont (dc) : continue emulation / DSP single-stepping
2060: </pre>
2061:
2062:
2063: <h4 id="Entering_arguments_to_debugger_commands">Entering arguments to debugger commands</h4>
2064:
2065: <p>
2066: After writing (with TAB completion) one of the above command names,
2067: pressing TAB will (for most commands) show all the available subcommands.
2068: </p>
2069:
2070: <p>
2071: If you want to give numbers in other number bases
2072: than the default/selected one, they need to be prefixed with a
2073: character indicating this. For decimals this prefix is "#" (#15),
2074: for hexadecimals "$" ($F), and for binary values it's "%" (%1111).
2075: </p>
2076:
2077: <p>
2078: By default debugger expects all numbers without a prefix to be
2079: decimals, but you can change the default number base with the "setopt"
2080: command, just give it the desired default number base (bin/dec/hex).
2081: <em>When using the hexadecimal number base, remember still to prefix
2082: hexadecimal numbers with '$' if they could be confused with register
2083: names (a0-7, d0-7)!</em> Otherwise results from expressions and
2084: conditional breakpoints can be unexpected.
2085: </p>
2086:
2087:
2088: <h4>Calculations and immediate evaluation</h4>
2089:
2090: <p>
2091: Instead of a number, you can also use an arithmetic expression, by
2092: surrounding it with quotes (""). An expression can contain
2093: calculations with CPU and DSP register, symbol and Hatari variable
2094: values in addition to numbers. For example to give a sum of A0 and
2095: D0 register values to a command, use "a0+d0".
2096: </p>
2097:
2098: <p>
2099: Within arithmetic expressions parenthesis are used both to change
2100: the order of precendence <em>and</em> to indicate indirect addressing.
2101: Unlike with conditional breakpoint expressions (explained below), you
2102: cannot give size for the indirect addressing, a long value is always
2103: read from the RAM address given within parenthesis. For example to
2104: get a long value pointed by stack pointer + 2, use "(a7+2)".
2105: </p>
2106:
2107: <p>
2108: Values of arithmetic expressions are always evaluated before being
2109: given to a command. Except for "evaluate" and "address" commands,
2110: they always need to be marked with quotes (""). Besides arithmetics,
2111: this can be used also to give symbol/register/variable values to
2112: commands that don't otherwise interpret them. If command complains
2113: that it didn't recognize e.g. a register name, just put it to quotes
2114: and it will be "evaluated" before being given to the command.
2115: </p>
2116:
2117: <p>
2118: With command argument completion (see <a href="#Build_notes">build
2119: notes</a>), result from the last "evaluate" command can be inserted
2120: by typing '$' and pressing TAB.
2121: </p>
2122:
2123:
2124: <h3 id="Inspecting_emulation_state">Inspecting emulation state</h3>
2125:
2126: <p>
2127: In the beginning, probably the most interesting commands are "m" and "d"
2128: for dumping and disassembling memory regions. You can use "dm" and "dd"
2129: commands to do the same for the DSP.
2130: </p>
2131: <pre>
2132: > help memdump
2133: 'memdump' or 'm' - dump memory
2134: Usage: m [start address-[end address]]
2135: dump memory at address or continue dump from previous address.
2136: </pre>
2137: <pre>
2138: > help disasm
2139: 'disasm' or 'd' - disassemble from PC, or given address
2140: Usage: d [start address-[end address]]
2141: If no address is given, this command disassembles from the last
2142: position or from current PC if no last position is available.
2143: </pre>
2144: <pre>
2145: > disasm pc
2146: $00aa6e : 2f08 move.l a0,-(sp)
2147: $00aa70 : 0241 0fff andi.w #$fff,d1
2148: $00aa74 : 207c 00fe 78c0 movea.l #$fe78c0,a0
2149: $00aa7a : 2070 1000 movea.l (a0,d1.w),a0
2150: $00aa7e : 4ed0 jmp (a0)
2151: </pre>
2152:
2153: <p>
2154: Both commands accept in addition to numeric addresses also register
2155: and symbol names, like in above example. If you don't specify an
2156: address, the commands continue showing from an address that comes
2157: after the previously shown data. "disasm" command default address
2158: will be reseted to PC address everytime you re-enter the debugger.
2159: </p>
2160:
2161: <p>
2162: Use "setopt --disasm help" if you want to set options controlling
2163: the disassembly output.
2164: </p>
2165:
2166: <p>
2167: You can use the "info" command to see state of specific sets of HW
2168: registers (e.g. "info videl") and Atari OS structures (e.g. "info gemdos").
2169: </p>
2170:
2171:
2172: <h4>Selecting what information is shown on entering the debugger</h4>
2173:
2174: <p>
2175: By using the "lock" command, you can ask Hatari to show specific
2176: information whenever you enter the debugger / hit a breakpoint. For
2177: example to see disassembly from current PC address, use "lock disasm".
2178: </p>
2179:
2180: <p>
2181: With the "regaddr" subcommand, you see disassembly or memory
2182: dump of an address pointed by a given register ("lock regaddr disasm
2183: a0"). Of the DSP registers, only Rx ones are valid for this
2184: subcommand.
2185: </p>
2186:
2187: <p>
2188: "file" subcommand can be used to get (arbitrary number of) commands
2189: parsed and executed from a given debugger input file whenever debugger
2190: is entered. With this you can output any information you need:
2191: </p>
2192: <pre>
2193: lock file debugger.ini
2194: </pre>
2195:
2196: <p>
2197: To disable showing of this extra information, use "lock default".
2198: Without arguments "lock" command will show the available options
2199: (like the "info" command does).
2200: </p>
2201:
2202:
2203: <h3>Debug symbols</h3>
2204:
2205: <p>
2206: You can load debugging symbols to the debugger with the "symbols"
2207: command (and with "dspsymbols" for DSP). These symbolic names can be
2208: used in arithmetic expressions and conditional breakpoint expressions.
2209: They also show up in the "disasm" command output and you can trace
2210: calls to them with "trace cpu_symbols" (and DSP symbols with "trace
2211: dsp_symbols").
2212: </p>
2213:
2214:
2215: <h4>For a program under GEMDOS HD emulation</h4>
2216:
2217: <p>
2218: If you're using GEMDOS HD emulation, and your program contains symbol
2219: table in DRI/GST format, you can load its symbol names/addresses to
2220: the debugger with the following command, after program has been loaded
2221: to the memory by TOS (see <a href="#Breakpoint_variables">setting
2222: breakpoint at program startup</a>):
2223: </p>
2224: <pre>
2225: symbols prg
2226: </pre>
2227: <p>
2228:
2229: <p>
2230: The options you need to add suitable symbol table to your programs,
2231: depend on which toolchain you use to build it:
2232: </p>
2233: <dl>
2234: <dt><em>Devpac</em>:</dt>
2235: <dd>"OPT D+,X+"</dd>
2236: <dt><em>AHCC</em>:</dt>
2237: <dd>"-g", and "-l" option for local symbols, both for linking</dd>
2238: <dt><em>GCC</em>:</dt>
2239: <dd>"-Wl,--traditional-format" option for linking,
2240: and "-g" for compilation to get local symbols</dd>
2241: <dt><em>VBCC</em>:</dt>
2242: <dd>"-g" (can only be used at linking phase), <em>when VBCC
2243: configuration file uses "-bataritos" option for
2244: the linker</em></dd>
2245: </dl>
2246:
2247: <p>You can view the generated symbols (and convert them to debugger
2248: ASCII format) with tool installed with Hatari:</p>
2249: <pre>
2250: $ gst2ascii -l -o program.tos > program.sym
2251: </pre>
2252: (Options -l and -o are used to exclude useless symbols from the output.)
2253:
2254:
2255: <h4>For a program on a (disk) image</h4>
2256:
2257: <p>
2258: If the program isn't run from a GEMDOS HD emulated drive, but from a
2259: cartridge, floppy or HD image, you need to have the corresponding
2260: program also as normal host file which location you can give to the
2261: debugger:
2262: </p>
2263: <pre>
2264: symbols /path/to/the/program.tos
2265: </pre>
2266:
2267:
2268: <h4>ASCII debug symbol files</h4>
2269:
2270: <p>
2271: If Hatari complains that your program doesn't have DRI/GST format
2272: symbol table, or its symbols are in some other format, and you
2273: cannot re-compile it to have them, you have two options:
2274: </p>
2275: <ul>
2276: <li>Convert the symbols to ASCII format understood by the Hatari debugger.
2277: Writing converters for other ASCII formats is easy, and Hatari already
2278: contains covertors for DSP LOD files, <span class="commandline">nm</span>
2279: output for MiNT/a.out binaries and AHCC map files.
2280: <li>Create the ASCII symbols file by hand as you're debugging a program.
2281: </ul>
2282:
2283: <p>ASCII symbols file format is following:</p>
2284: <pre>
2285: e01034 T random
2286: e01076 T kbdvbase
2287: e0107e T supexec
2288: </pre>
2289: <p>
2290: Where 'T' means text (code), 'D' means data and 'B' means BSS section
2291: type of address. The hexadecimal address, address type letter and the
2292: symbol name are separated by white space. Empty lines and lines
2293: starting with '#' (comments) are ignored.
2294: </p>
2295:
2296: <p>
2297: Debugger will automatically "relocate" the symbol addresses when it
2298: loads them from a program binary, but with ASCII symbol files you need
2299: to give the relocation offset(s) separately, unless the symbol names
2300: are for fixed adresses (like is the case e.g. with EmuTOS):
2301: </p>
2302: <pre>
2303: symbols program.sym TEXT DATA BSS
2304: </pre>
2305: <p>
2306: If you're interested only about code symbols, you can leave DATA and
2307: BSS offsets out (the values of the above virtual debugger variables
2308: like TEXT come from the currently loaded program's basepage, they're
2309: set after the program is loaded by TOS, see "info basepage" output).
2310: </p>
2311:
2312:
2313: <h3>Breakpoints</h3>
2314:
2315: <p>
2316: There are two ways to specify breakpoints for Hatari. First, there are
2317: the simple address breakpoints which trigger when the CPU (or DSP)
2318: program counter hits a given address. Use "a" (or "da" for the DSP)
2319: to create them, for example:
2320: </p>
2321: <pre>
2322: a $e01034
2323: a some_symbol
2324: </pre>
2325:
2326: <p>
2327: Note that address breakpoints are just wrappers for conditional
2328: breakpoints so you need to use "b" command to remove or list them.
2329: </p>
2330:
2331: <p>
2332: Then there are the conditional breakpoints which can handle much more
2333: complex break condition expressions; they can track changes to
2334: register and memory values with bitmasks, include multiple conditions
2335: for triggering a breakpoint and so on. Use "b" (or "db" for the DSP)
2336: to manage them.
2337: </p>
2338:
2339: <p>Help explains the general syntax:</p>
2340: <pre>
2341: > help b
2342: 'breakpoint' or 'b' - set/remove/list conditional CPU breakpoints
2343: Usage: b <condition> [&& <condition> ...] [:<option>] | <index> | help | all
2344:
2345: Set breakpoint with given <conditions>, remove breakpoint with
2346: given <index>, remove all breakpoints with 'all' or output
2347: breakpoint condition syntax with 'help'. Without arguments,
2348: lists currently active breakpoints.
2349: </pre>
2350:
2351: <p>
2352: Unless you give breakpoint one of the pre-defined subcommands ('all',
2353: 'help'), index for a breakpoint to remove or no arguments (to list
2354: breakpoints), the arguments are interpreted as a new breakpoint
2355: definition.
2356: </p>
2357:
2358: <p>
2359: Each conditional breakpoint can have (currently up to 4) conditions
2360: which are separated by "&&". All of the breakpoint's
2361: conditions need to be true for a breakpoint to trigger.
2362: </p>
2363:
2364:
2365: <h4 id="Breakpoint_options">Breakpoint options</h4>
2366:
2367: <p>
2368: Normally when a breakpoint is triggered, emulation is stopped and you
2369: get to the debugger. Breakpoint options can be used to affect what
2370: happens when a breakpoint is triggered. These options are given after
2371: the conditions and are prefixed with ':'.
2372: </p>
2373:
2374: <dl>
2375: <dt><em><count></em></dt>
2376: <dd>Break only on every <count> hit. For example, to stop
2377: on every other time PC is at given address, use:
2378: <pre>
2379: a $1234 :2
2380: </pre>
2381: </dd>
2382:
2383: <dt><em>once</em></dt>
2384: <dd>
2385: Delete the breakpoint when it's hit i.e. trigger it only once. It may
2386: be useful if you just want to get a specific address. Or if you're on
2387: an instruction that jumps back to a start of the loop and you want to
2388: finish the loop, you could use:
2389: <pre>
2390: b pc > "pc" :once
2391: continue
2392: </pre>
2393: </dd>
2394:
2395: <dt><em>trace</em></dt>
2396: <dd>
2397: Continue emulation without stopping after printing the value that
2398: triggered the breakpoint and doing other possible option actions.
2399: This is most useful when investigating memory or register value
2400: changes (explained below).
2401: </dd>
2402:
2403: <dt><em>lock</em></dt>
2404: <dd>
2405: Show the same information on breakpoint hit as you see when entering
2406: the debugger (see the "lock" command in
2407: <a href="#Inspecting_emulation_state">Inspecting emulation state</a>
2408: above). This enables also trace option as you would anyway see this
2409: information if debugger would be entered.
2410: </dd>
2411:
2412: <dt><em>file <file></em></dt>
2413: <dd>
2414: Execute debugger commands from given <file> when this breakpoint
2415: is hit. With this you have complete control over what information is
2416: show when the debugger is hit, you can even chain breakpoints (as
2417: explained in
2418: <a href="#Chaining_breakpoints">Chaining breakpoints</a> later on)
2419: etc. Use this if "lock" option isn't enough or you want different
2420: information show on breakpoints and when entering the debugger.
2421: </dd>
2422:
2423: <dt><em>noinit</em></dt>
2424: <dd>
2425: Hitting breakpoint doesn't re-initialize debugger which would e.g.
2426: cause profiling data to be reset. This implies trace option as
2427: entering debugger would also re-initialize debugger state. This option
2428: is mainly intended for breakpoints that use :file option to show
2429: backtraces with "profile stack" command during
2430: <a href="#Profiling">profiling</a>. See
2431: <a href="#Usage_examples">Usage examples</a> section for an example.
2432: </dd>
2433: </dl>
2434:
2435: <p>
2436: Note: you can give multiple options for conditional breakpoints, but
2437: for address breakpoints you can give only one these options. And
2438: "file" option is supported only for conditional breakpoints.
2439: </p>
2440:
2441:
2442: <h4>Breakpoint conditions</h4>
2443:
2444: <p>
2445: "b help" explains very briefly the breakpoint condition syntax:
2446: </p>
2447: <pre>
2448: > b help
2449: condition = <value>[.mode] [& <mask>] <comparison> <value>[.mode]
2450:
2451: where:
2452: value = [(] <register/symbol/variable name | number> [)]
2453: number/mask = [#|$|%]<digits>
2454: comparison = '<' | '>' | '=' | '!'
2455: addressing mode (width) = 'b' | 'w' | 'l'
2456: addressing mode (space) = 'p' | 'x' | 'y'
2457: </pre>
2458:
2459: <p>
2460: For CPU breakpoints, mode is the address width; it can be byte ("b"),
2461: word ("w") or long ("l", default). For DSP breakpoints, mode specifies
2462: the address space: "P", "X" or "Y". Note that on DSP only R0-R7
2463: registers can be used for memory addressing. For example;
2464: <pre>
2465: db (r0).x = 1 && (r0).y = 2
2466: </pre>
2467:
2468: <p>
2469: If the value is in parenthesis like in '($ff820)' or '(a0)', then the
2470: used value will be read from the memory address pointed by it. Note
2471: that this conditional breakpoint expression value is checked at
2472: run-time whereas quoted arithmetic expressions (mentioned in
2473: <a href="#Entering_arguments_to_debugger_commands">Entering arguments
2474: to debugger commands</a> above) are evaluated already when
2475: adding a breakpoint. For example, to break when a value in an address
2476: (later) pointed by A0 matches the value <em>currently</em> in D0, one
2477: would use:
2478: </p>
2479: <pre>
2480: b (a0) = "d0"
2481: </pre>
2482:
2483: <p>
2484: If you're interested only on certain bits in the value, you can use
2485: '&' and a numeric mask on either side of comparison operator to
2486: mask the coresponding value, like this:
2487: <pre>
2488: b ($ff820).w & 3 = (a0) && (a1) = d0 & %1100
2489: </pre>
2490:
2491: <p>
2492: Comparison operators should be familiar and obvious, except for '!'
2493: which indicates inequality ("is not") comparison. For example:
2494: </p>
2495: <pre>
2496: b d0 > $20 && d0 < $40 && d0 ! $30
2497: </pre>
2498:
2499:
2500: <h5>Tracking breakpoint conditions</h5>
2501:
2502: <p>
2503: As a convenience, if the both sides of the comparison are exactly the
2504: same (i.e. condition is redundant as it's always either true or
2505: false), the <em>right side</em> of the comparison is replaced with
2506: its current value. This way you can give something like this:
2507: </p>
2508: <pre>
2509: b pc > "pc"
2510: </pre>
2511: <p>As:</p>
2512: <pre>
2513: b pc > pc
2514: </pre>
2515:
2516: <p>
2517: That in itself isn't so useful, but for inequality ('!') comparison,
2518: conditional breakpoint will additionally track and output all further
2519: changes for the given address/register expression. This can be used
2520: for example to find out all value changes in a given memory address,
2521: like this:
2522: </p>
2523: <pre>
2524: b ($ffff9202).w ! ($ffff9202).w :trace
2525: </pre>
2526: <p>
2527: Because tracking breakpoint conditions will print the evaluated
2528: value when it changes, they're typically used with the trace option
2529: to track changes e.g. to some IO register.
2530: </p>
2531:
2532:
2533: <h5>Breakpoint condition notes</h5>
2534:
2535: <ul>
2536: <li>
2537: Any '!' condition should be given as the first condition. Because
2538: breakpoint evaluation is stopped ("short-circuited") when any of the
2539: conditions fails, the tracked value would not be updated correctly
2540: unless tracking condition is given as the first one.
2541: </li>
2542:
2543: <li>
2544: Hatari will internally update some register values without immediately
2545: updating the corresponding IO address range memory addresses. For
2546: example the Busy bit for the internal Blitter control register is
2547: (internally) cleared when Blitter activity stops, but the actual IO
2548: address for that control register gets updated only when something
2549: actually writes or reads that IO address. Many HW registers behave
2550: (status registers in FDC, ACIA, MFP, Blitter...).
2551: <br>
2552: For breakpoints that track just a single IO register memory address, or
2553: multiple ones of which <strong>none</strong> are modified by Hatari,
2554: only by emulated code, this is not a problem, they get triggered as
2555: expected.
2556: <br>
2557: However, if you have a breakpoint that tracks multiple IO registers
2558: where some of them are updated by Hatari, for example to check that
2559: other Blitter registers aren't updated while control register
2560: indicates Blitter to be active (busy), things don't work as expected!
2561: </li>
2562: </ul>
2563:
2564:
2565: <h4>Breakpoint variables</h4>
2566:
2567: <p>
2568: In addition to loaded symbols, the debugger supports also setting
2569: conditional breakpoints on values of some "virtual" variables listed
2570: by "b help". For example:
2571: </p>
2572: <ul>
2573: <li>If you want the emulation to stop on the first instruction of
2574: next program; after TOS desktop is up, set a breakpoint on
2575: the TEXT segment address given in a program basepage:
2576: <pre>
2577: b pc = TEXT :once
2578: </pre>
2579: Note1: It's better to trigger it only once because if you'd leave it on,
2580: during reboot you would get a warning for every instruction until TOS sets
2581: a valid basepage.
2582: <br />
2583: Note2: you cannot use an address breakpoint for this because value of
2584: a variable given to address breakpoint is evaluated when it's set, not
2585: at run-time, so it cannot get the new value that the TEXT variable
2586: gets when you start a program.
2587: </li>
2588: <li>To find out current program DATA and BSS segment contents,
2589: use the corresponding variables:
2590: <pre>
2591: m DATA
2592: m BSS
2593: </pre>
2594: </li>
2595: <li>If you want to stop at a specific cycle within a frame (that is,
2596: PC relative to the current VBL/HBL in cycles), set breakpoints to
2597: specific "HBL" and "FrameCycles" variable values. If you for
2598: example want to break after 20 HBLs, use:
2599: <pre>
2600: b HBL = "HBL+20"
2601: </pre>
2602: </li>
2603: <li>Aes/Bios/Gemdos/LineA/LineF/Vdi/XbiosOpcode variables can be used
2604: to catch AES, BIOS, GEMDOS, Line-A, Line-F, VDI and XBIOS OS-calls.
2605: By default they contain the 0xffff value, so to trace e.g. all AES
2606: calls, instead of a specific one, one needs to use something like this:
2607: <pre>
2608: b AesOpcode ! AesOpcode && AesOpcode < 0xffff :trace
2609: </pre>
2610: </li>
2611: </ul>
2612:
2613: <p>
2614: Hint: "info" command "aes", "bios", "gemdos", "vdi" and "xbios"
2615: subcommands for can be used to list the corresponding OS-call opcodes.
2616: For example, to see the GEMDOS opcodes, use:</p>
2617: <pre>
2618: info gemdos 1
2619: </pre>
2620:
2621:
2622: <h4 id="Chaining_breakpoints">Chaining breakpoints and other actions</h4>
2623:
2624: <p>
2625: As the file pointed by the breakpoint ":file" option (see
2626: <a href="#Breakpoint_options">Breakpoint options</a>) can contain any
2627: debugger commands, it can also be used to do automatic "chaining" of
2628: debugger and breakpoint actions so that after one breakpoint is hit,
2629: another one is set.
2630: </p>
2631:
2632: <p>For example if you have these input files:</p>
2633: <ul>
2634: <li>"break.ini":
2635: <pre>
2636: b GemdosOpcode = 0x3D :trace :file program.ini
2637: </pre>
2638: </li>
2639: <li>"program.ini":
2640: <pre>
2641: b pc = TEXT :trace :file trace.ini
2642: </pre>
2643: </li>
2644: <li>"trace.ini":
2645: <pre>
2646: symbols prg
2647: trace gemdos,cpu_symbols
2648: b VBL = "VBL+4" :trace :file disable.ini
2649: </pre>
2650: </li>
2651: <li>"disable.ini":
2652: <pre>
2653: trace none
2654: b all
2655: </pre>
2656: </li>
2657: </ul>
2658:
2659: <p>
2660: And then start Hatari with the first debugger input file and a GEMDOS
2661: harddisk directory containing "desktop.inf" file:
2662: </p>
2663: <pre>
2664: hatari --parse break.ini /path/to/your/program.tos
2665: </pre>
2666:
2667: <ol>
2668: <li>"break.ini" input file will break when TOS opens
2669: the "desktop.inf" file (it's the first Fopen() i.e. GEMDOS call
2670: 0x3D done by TOS at boot) and the breakpoint will run
2671: the debugger commands from the "symbols.ini" file
2672: <li>"program.ini" will setup breakpoint to program startup
2673: (because TEXT variable cannot be used before TOS has booted)
2674: <li>"trace.ini" input file loads symbols for the run program, sets Hatari
2675: to trace several things (see <a href="#Tracing">Tracing</a> section
2676: below) in the emulated system for few VBLs until breakpoint runs
2677: commands from the "disable.ini" file
2678: <li>"disable.ini" input file will disable tracing and remove
2679: all (remaining) breakpoints
2680: </ol>
2681:
2682: <p>
2683: <em>Note:</em> because debugger input files cannot "continue"
2684: emulation, ":trace" option needs to be used for the breakpoint(s)
2685: if you want emulation to continue after the breakpoint action(s).
2686: </p>
2687:
2688: <p>
2689: Hint: It's better to test each input file separate before testing the
2690: whole chain. Besides the ":file" breakpoint option, these debugger
2691: input files can be also read with the debugger "file" command, "lock"
2692: command "file" option and with the Hatari "--parse" command line
2693: option.
2694: </p>
2695:
2696:
2697: <h3>Tracing</h3>
2698:
2699: <p>
2700: After analyzing the emulation state and/or setting new breakpoints,
2701: you can continue the emulation with the "c" command. You can continue
2702: for a given number of CPU instructions (or DSP instructions when "dc"
2703: is used), or you can continue forever (until a non-tracing breakpoint
2704: triggers) if you omit the instruction count.
2705: </p>
2706:
2707: <p>
2708: If you want to continue just to the next instruction, use "s" (step)
2709: command to continue for exactly one instruction, or "n" (next), if you
2710: want to skip subroutine calls.
2711: </p>
2712:
2713: <p>
2714: If you want to continue with real-time disassembling, you can enable
2715: it with "trace cpu_disasm" (or "trace dsp_disasm" for DSP) at the
2716: debugger prompt before continuing.
2717: </p>
2718: <p>
2719: Disable tracing with "trace none" when you enter the debugger again.
2720: "trace help" (or TAB) can be used to list all the (over 40) supported
2721: traceable things, from HW events to OS functions.
2722: </p>
2723: <p>
2724: Because Hatari normally emulates things at the hardware level, tracing
2725: certain Atari OS Traps requires setting additional Hatari options
2726: which tell it to intercept this higher level functionality:
2727: <ul>
2728: <li>BIOS and XBIOS tracing require enabling of the BIOS intercepting with
2729: the "--bios-intercept" option (which has also some other side-effects)
2730: </li>
2731: <li>GEMDOS tracing requires enabling the GEMDOS harddisk emulation
2732: (which doesn't work under MiNT because it re-implements GEMDOS)</li>
2733: </ul>
2734: <p>
2735: Unlike with VDI tracing, AES, BIOS, GEMDOS and XBIOS traces show arguments
2736: for most of the calls.
2737: </p>
2738: <p>
2739: Tracing options can be set even from a program within the emulation,
2740: if you enable the Hatari "--bios-intercept" option and call XBios 255
2741: from the program with a suitable Hatari command line string.
2742: </p>
2743: <p>
2744: Note that the trace output file can be set only when Hatari starts,
2745: it cannot be changed from within the debugger (or emulation).
2746: </p>
2747: <p>
2748: If tracing isn't possible for the things you'd like to track, you can
2749: use the OS call opcode breakpoints explained above with the ":trace"
2750: breakpoint option.
2751: </p>
2752:
2753:
2754: <h3>Profiling</h3>
2755:
2756: <p>
2757: Profiling tells where the emulated code spends most of its (emulated)
2758: time. It can be used to find out where a program is (apparently)
2759: stuck, or what are the largest performance bottlenecks for a program.
2760: </p>
2761:
2762: <h4>Collecting the profile data</h4>
2763:
2764: <p>
2765: Profiling is used by first enabling the profiler (use "dp" for DSP):
2766: </p>
2767: <pre>
2768: > profile on
2769: Profiling enabled.
2770: </pre>
2771: <p>
2772: And profiling will start once you continue the emulation:
2773: </p>
2774: <pre>
2775: > c
2776: Returning to emulation...
2777: Allocated CPU profile buffer (27 MB).
2778: </pre>
2779:
2780: <p>
2781: When you get back to the debugger, the collected profiling information
2782: is processed and a summary of in which parts of memory the execution
2783: happened, and how long it took, is shown:
2784: </p>
2785: <pre>
2786: Allocated CPU profile address buffer (57 KB).
2787: ROM TOS (0xE00000-0xE80000):
2788: - active address range:
2789: 0xe00030-0xe611a4
2790: - active instruction addresses:
2791: 14240 (100.00% of all)
2792: - executed instructions:
2793: 4589668 (100.00% of all)
2794: - used cycles:
2795: 56898472 (100.00% of all)
2796: = 7.09347s
2797: Cartridge ROM (0xFA0000-0xFC0000):
2798: - no activity
2799:
2800: = 7.09347s
2801: </pre>
2802: <p>
2803: (DSP RAM will be shown only as single area in profile information.)
2804: </p>
2805:
2806:
2807: <h4>Investigating the profile data</h4>
2808:
2809: <p>
2810: When you're back in debugger, you can inspect the collected profile data:
2811: </p>
2812: <pre>
2813: > h profile
2814: 'profile' - profile CPU code
2815: Usage: profile <on|off|stats|counts|cycles|misses|symbols|callers|stack|addresses|save> [count|address|file]
2816: 'on' & 'off' enable and disable profiling. Data is collected
2817: until debugger is entered again at which point you get profiling
2818: statistics ('stats') summary.
2819:
2820: Then you can ask for list of the PC addresses, sorted either by
2821: execution 'counts', used 'cycles' or cache 'misses'. First can
2822: be limited just to named addresses with 'symbols'. Optional
2823: count will limit how many items will be shown.
2824:
2825: 'addresses' lists the profiled addresses in order, with the
2826: instructions (currently) residing at them. By default this
2827: starts from the first executed instruction, or you can
2828: specify the starting address.
2829:
2830: 'callers' shows (raw) caller information for addresses which
2831: had symbol(s) associated with them. 'stack' shows the currect
2832: profile stack, this is useful only with :noinit breakpoints.
2833:
2834: Profile information can be saved with 'save'.
2835: </pre>
2836:
2837: <p>For example, to see which memory addresses were executed most
2838: and what instructions those have at the end of profiling, use:</p>
2839: <pre>
2840: > profile counts 8
2841: addr: count:
2842: 0xe06f10 12.11% 555724 move.l $4ba,d1
2843: 0xe06f16 12.11% 555724 cmp.l d1,d0
2844: 0xe06f18 12.11% 555724 bgt.s $e06f06
2845: 0xe06f06 12.11% 555708 move.b $fffffa01.w,d1
2846: 0xe06f0a 12.11% 555708 btst #5,d1
2847: 0xe06f0e 12.11% 555708 beq.s $e06f1e
2848: 0xe00ed8 1.66% 76001 subq.l #1,d0
2849: 0xe00eda 1.66% 76001 bpl.s $e00ed8
2850: 8 CPU addresses listed.
2851: </pre>
2852:
2853: <p>
2854: Then, to see what the executed code and its costs look like
2855: around top addresses:
2856: <pre>
2857: > profile addresses 0xe06f04
2858: # disassembly with profile data:
2859: # <instructions percentage>% (<sum of instructions>, <sum of cycles>, <sum of i-cache misses>)
2860: $e06f04 : bra.s $e06f10 0.00% (48, 576, 0)
2861: $e06f06 : move.b $fffffa01.w,d1 12.11% (555708, 8902068, 0)
2862: $e06f0a : btst #5,d1 12.11% (555708, 6685268, 0)
2863: $e06f0e : beq.s $e06f1e 12.11% (555708, 4457312, 0)
2864: $e06f10 : move.l $4ba,d1 12.11% (555724, 11125668, 0)
2865: $e06f16 : cmp.l d1,d0 12.11% (555724, 4461708, 0)
2866: $e06f18 : bgt.s $e06f06 12.11% (555724, 4455040, 0)
2867: $e06f1a : moveq #1,d0 0.00% (16, 64, 0)
2868: Disassembled 8 (of active 14240) CPU addresses.
2869: </pre>
2870: <p>
2871: Unlike normal disassembly, "profile addresses" command shows only
2872: memory addresses which instructions were executed during profiling.
2873: You get instruction cache misses only when using cycle-accurate 030
2874: emulation with a Hatari version configured to use WinUAE CPU core.
2875: <p>
2876: If you have loaded symbol information, symbol names are shown above
2877: the corresponding addresses. With the "profile symbols" command you
2878: get a list of how many times the code execution passed through the
2879: defined symbol addresses.
2880: </p>
2881:
2882:
2883: <h4>Profile data accuracy</h4>
2884:
2885: <p>Profile data accuracy depends on Hatari emulation accuracy.
2886: Profile data accuracy from most to least accurate when Hatari's
2887: default emulation options are used is following:</p>
2888: <ul>
2889: <li>Executed CPU and DSP instruction counts are accurate.</li>
2890: <li>DSP cycles counts (and their variance information) should be accurate.
2891: </li>
2892: <li>030 CPU instruction cache miss information (provided by WinUAE CPU core)
2893: is assumed to be accurate.</li>
2894: <li>Cycles used by a given CPU instruction depend to some extent on what
2895: instruction(s), and data in case of 030, was processed before it.
2896: While Hatari has some (instruction pairing) heuristics to take
2897: that into account for 68000, in general instruction cycles are
2898: averages. For 68000, cycles (provided by OldUAE CPU core) should
2899: be fairly accurate, for 68030 they aren't (yet) not very accurate.</li>
2900: </ul>
2901:
2902:
2903: <h4>Caller information</h4>
2904:
2905: <p>
2906: If you have loaded symbols (see <a href="#Debug_symbols">Debug symbols</a>)
2907: before continuing emulation/profiling, additional caller information
2908: will be collected for all the code symbol addresses which are called
2909: as subroutines. This information includes callstack, call counts,
2910: calling instruction type (subroutine call, branch, return etc), and
2911: costs for those calls, both including costs for further subroutine
2912: calls and without them.
2913: </p>
2914:
2915: <p>When debugger is re-entered, current callstack is output before
2916: profiling information:</p>
2917: <pre>
2918: > a <em>_P_LineAttack</em>
2919: CPU condition breakpoint 1 with 1 condition(s) added:
2920: pc = $30f44
2921: $030f44 : 48e7 3820 movem.l d2-d4/a2,-(sp)
2922: > c
2923: ...
2924: CPU breakpoint condition(s) matched 1 times.
2925: pc = $30f44
2926: Finalizing costs for 12 non-returned functions:
2927: - 0x32a3c: _P_GunShot (return = 0x32b7e)
2928: - 0x32b18: _A_FireShotgun (return = 0x3229a)
2929: - 0x3223a: _P_SetPsprite (return = 0x32e86)
2930: - 0x32e4e: _P_MovePsprites (return = 0x38070)
2931: - 0x37f44: _P_PlayerThink (return = 0x36ea0)
2932: - 0x36e44: _P_Ticker (return = 0x260e0)
2933: - 0x25dcc: _G_Ticker (return = 0x1e4c6)
2934: - 0x1e29e: _TryRunTics (return = 0x239fa)
2935: - 0x238e8: _D_DoomLoop (return = 0x2556a)
2936: - 0x24d7a: _D_DoomMain (return = 0x44346)
2937: ...
2938: </pre>
2939:
2940: <p>("profile stack" command can be used in breakpoints with :noinit
2941: option to show backtraces during caller profiling.)</p>
2942:
2943: <p>Other information collected during profiling is shown with
2944: following command:</p>
2945: <pre>
2946: > profile callers
2947: # <callee>: <caller1> = <calls> <types>[ <inclusive/totals>[ <exclusive/totals>]], <caller2> ..., <callee name>
2948: # types: s = subroutine call, r = return from subroutine, e = exception, x = return from exception,
2949: # b = branch/jump, n = PC moved to next instruction, u = unknown PC change
2950: # totals: calls/instructions/cycles/misses
2951: 0xe00030: 0xffffff = 1 e, _main
2952: 0xe000fe: 0xe00a0c = 1 b, memdone
2953: 0xe0010a: 0xe04e34 = 1 s 1/5/72 1/5/72, _run_cartridge_applications
2954: 0xe00144: 0xe04dbe = 1 s 4/118/1512 1/27/444, _init_acia_vecs
2955: 0xe001ea: 0xe00ec6 = 1 b, _int_acia
2956: 0xe0038c: 0xe04c28 = 1 s 1/191/2052 1/191/2052, _init_exc_vec
2957: 0xe003a6: 0xe04c2e = 1 s 1/388/4656 1/388/4656, _init_user_vec
2958: ...
2959: </pre>
2960:
2961: <p>
2962: For example, if you don't know all the places from which a certain
2963: function is called, or in what context a certain interrupt handler can
2964: be called during the period you're profiling, profile caller
2965: information will tell you:
2966: </p>
2967: <pre>
2968: callee: caller: calls: calltype:
2969: | | | /
2970: 0x379: 0x155 = 144 r, 0x283 = 112 b, 0x2ef = 112 b, 0x378 = 72 s
2971: 583236/359708265/1631189180 72/4419020/19123430, dsp_interrupt
2972: | | |
2973: inclusive costs exclusive costs callee name
2974: (of calls from 0x378)
2975:
2976: Calltypes:
2977: - b: jump/branch
2978: - n: PC just moved to next address
2979: - r: subroutine return
2980: - s: subroutine call
2981: </pre>
2982: <p>
2983: (Most "calls" to "dsp_interrupt" were subroutine call returns (=r)
2984: to it from address 0x155.)
2985: </p>
2986:
2987: <p>
2988: With the execution counts in normal profiling data, caller information
2989: can actually be used to have complete picture of what exactly the code
2990: did during profiling. Main/overview work for this analysis is best done
2991: automatically, by the profiler data post-processor (documented below).
2992: </p>
2993:
2994:
2995: <h4>Caller data accuracy</h4>
2996:
2997: <p>Everything about profile data accuracy applies also to caller costs,
2998: but there are additional things to take into account, mainly because
2999: profiler cannot determine when exceptions are being handling:</p>
3000: <ul>
3001: <li>If there are exception(s) during a subroutine call, costs for
3002: the exception handling will also be accounted for that subroutine.
3003: This shouldn't be a problem unless those costs are very large,
3004: i.e. check how much CPU your exception handlers take.</li>
3005: <li>Indicated exception handler call type can be incorrect.</li>
3006: <li>Profiled code doing return address related stack manipulations
3007: confuses call tracking and produces incorrect results (profiler
3008: has special code to handle EmuTOS AES switcher because of this).
3009: Typically this produces large list of functions that are finalized
3010: at profile end, so it should be easy to detect.</li>
3011: <li>Compilicated recursive calls seem to sometimes cause inclusive
3012: costs (ones including costs of further subroutine calls) to be
3013: incorrect. Sometimes this can be noticed by them being even
3014: >100%.</li>
3015: <li>On DSP, profiler heuristics assume (for speed reasons) that
3016: <em>conditional</em> subroutine calls never call the very next
3017: instruction (as that would be very bad/inefficient code).</li>
3018: </ul>
3019:
3020:
3021: <h4>Saving profile data to a file</h4>
3022:
3023: <p>It's useful to save the profile data to a file:
3024: <pre>
3025: > profile save program-profile.txt
3026: </pre>
3027:
3028: <p>With the saved profile disassembly (and optional caller information)
3029: you can more easily investigate what your program did during
3030: profiling, search symbols & addresses in it, and compare the
3031: results to profiles you've saved from earlier versions of your code.</p>
3032:
3033: <p>You may even create your own post-processing tools for
3034: investigating the profiling data more closely, e.g. to
3035: <a href="http://www.atari-forum.com/viewtopic.php?f=68&t=24561&start=75#p226505">find
3036: CPU/DSP communication bottlenecks</a>.</p>
3037:
3038:
3039: <h3>Profile data post-processing</h3>
3040:
3041: <p>Saved profile data can be post-processed with (Python) script
3042: installed by Hatari, to:</p>
3043: <ul>
3044: <li>Get lists of functions/symbols with highest costs.</li>
3045: <li>Get callgraphs of what functions/symbols cause those
3046: costs and what kind of call hiearchy the profiled code
3047: has.</li>
3048: <li>Export profile data in Valgrind's
3049: <a href="http://valgrind.org/docs/manual/cl-format.html">Callgrind format</a>
3050: for viewing it in
3051: <a href="http://kcachegrind.sourceforge.net/">Kcachegrind</a>
3052: GUI.</li>
3053: </ul>
3054:
3055:
3056: <h4>Providing symbols for the post-processor</h4>
3057:
3058: <p>When the data is post-processed, you should always provide
3059: the post-processor symbols for the profile code! Relying just on the
3060: symbol in the profile data can cause costs to be asssigned to wrong
3061: symbol, if symbol's code wasn't called through symbol's own address,
3062: but by jumping inside its code.</p>
3063:
3064: <p>If your code is in fixed location, you should tell
3065: post-processor to handle symbol addresses as absolute (-a):</p>
3066: <pre>
3067: $ hatari_profile.py <b>-a</b> etos512k.sym emutos-profile.txt
3068: </pre>
3069:
3070: <p>Normal programs are relocated and you should instead give
3071: the symbols as TEXT (code) section relative ones (-r):</p>
3072: <pre>
3073: $ hatari_profile.py <b>-r</b> program.sym program-profile.txt
3074: </pre>
3075:
3076: <p>If symbols are included to your binary in DRI/GST format, first they
3077: need to be extracted to <a href="#Debug_symbols">the ASCII format</a>
3078: understood by the post-processor:</p>
3079: <pre>
3080: $ gst2ascii -l -o program.prg > program.sym
3081: </pre>
3082:
3083: <p>If there are some extra symbols that you don't want to see
3084: separately in profiles, because they aren't real functions,
3085: but e.g. loop labels, you can either remove them manually
3086: from the ASCII *.sym file, or filter them out with grep:
3087: </p>
3088: <pre>
3089: $ gst2ascii -l -o program.prg | grep -v -e useless1 -e useless2 | > program.sym
3090: </pre>
3091:
3092: <p>Above post-processor examples just parse + verify the given data
3093: and produce output like this:</p>
3094: <pre>
3095: Hatari profile data processor
3096:
3097: Parsing TEXT relative symbol address information from program.sym...
3098: [...]
3099: 3237 lines with 1550 code symbols/addresses parsed, 0 unknown.
3100:
3101: Parsing profile information from program-profile.txt...
3102: [...]
3103: 9575 lines processed with 368 functions.
3104:
3105: CPU profile information from 'program-profile.txt':
3106: - Hatari v1.6.2+ (May 4 2013), WinUAE CPU core
3107: </pre>
3108:
3109:
3110: <h4>Post-processor provided statistics</h4>
3111:
3112: <p>To get statistics (-s) and list of top (-t) CPU users in profile,
3113: add "-st" option:</p>
3114: <pre>
3115: $ hatari_profile.py <b>-st</b> -r program.sym program-profile.txt
3116: [...]
3117: CPU profile information from 'program-profile.txt':
3118: - Hatari v1.6.2+ (May 4 2013), WinUAE CPU core
3119:
3120: Time spent in profile = 34.49539s.
3121:
3122: Calls:
3123: - max = 187738, in __toupper at 0x52b88, on line 8286
3124: - 1585901 in total
3125: Executed instructions:
3126: - max = 1900544, in flat_remap_mips+14 at 0x47654, on line 7020
3127: - 64499351 in total
3128: Used cycles:
3129: - max = 15224620, in flat_remap_mips+18 at 0x47658, on line 7022
3130: - 553392132 in total
3131: Instruction cache misses:
3132: - max = 184308, in _BM_T_GetTicks at 0x43b90, on line 4772
3133: - 4941307 in total
3134:
3135: Calls:
3136: 11.84% 187698 __toupper
3137: 11.48% 182105 _BM_T_GetTicks
3138: 11.48% 182019 _I_GetTime
3139: [...]
3140: Executed instructions:
3141: 34.83% 22462729 flat_generate_mips
3142: 14.08% 9080215 flat_remap_mips
3143: 8.55% 5515945 render_patch_direct
3144: 5.09% 3283328 _TryRunTics
3145: [...]
3146: Used cycles:
3147: 23.62% 130702768 flat_generate_mips
3148: 12.42% 68735832 flat_remap_mips
3149: 9.77% 54041148 _TryRunTics
3150: 5.80% 32111536 correct_element
3151: [...]
3152: Instruction cache misses:
3153: 37.03% 1829764 _TryRunTics
3154: 11.20% 553314 _BM_T_GetTicks
3155: 9.44% 466319 _NetUpdate
3156: 9.27% 457899 _HGetPacket
3157: [...]
3158: </pre>
3159:
3160: <p>If you want to see also symbol addresses and what is per call
3161: cost, add -i option:<p>
3162: <pre>
3163: $ hatari_profile.py -st <b>-i</b> -r program.sym program-profile.txt
3164: [...]
3165: Executed instructions:
3166: 34.83% 22462729 flat_generate_mips (0x04778a, 774576 / call)
3167: 14.08% 9080215 flat_remap_mips (0x047646, 313110 / call)
3168: 8.55% 5515945 render_patch_direct (0x047382, 29977 / call)
3169: 5.09% 3283328 _TryRunTics (0x042356, 19660 / call)
3170: [...]
3171: Used cycles:
3172: 23.62% 8.14728s 130702768 flat_generate_mips (0x04778a, 0.28094s / call)
3173: 12.42% 4.28461s 68735832 flat_remap_mips (0x047646, 0.14775s / call)
3174: 9.77% 3.36863s 54041148 _TryRunTics (0x042356, 0.02017s / call)
3175: 5.80% 2.00165s 32111536 correct_element (0x04a658, 0.00001s / call)
3176: [...]
3177: Instruction cache misses:
3178: 37.03% 1829764 _TryRunTics (0x042356, 10956 / call)
3179: 11.20% 553314 _BM_T_GetTicks (0x043b90, 3 / call)
3180: 9.44% 466319 _NetUpdate (0x041bcc, 5 / call)
3181: 9.27% 457899 _HGetPacket (0x041754, 5 / call)
3182: [...]
3183: </pre>
3184:
3185: <p>(For cycles the "per call" information is in seconds, not as
3186: a cost count.)</p>
3187:
3188: <p>If your profile file contains caller information, you should
3189: add -p option to see it, as that will also help in detecting symbol
3190: issues (see <a href="#Interpreting_the_numbers">Interpreting
3191: the numbers</a>):<p>
3192: <pre>
3193: $ hatari_profile.py -st <b>-p</b> -r program.sym program-profile.txt
3194: [...]
3195: 9575 lines processed with 368 functions.
3196: [...]
3197: Of all 1570498 switches, ignored 581 for type(s) ['r', 'u', 'x'].
3198:
3199: CPU profile information from 'badmood-level-load-CPU.txt':
3200: - Hatari v1.6.2+ (May 4 2013), WinUAE CPU core
3201: [...]
3202: Calls:
3203: 11.84% 11.84% 187698 187698 __toupper
3204: 11.48% 11.48% 182105 182105 _BM_T_GetTicks
3205: 11.48% 22.95% 182019 364038 _I_GetTime
3206: [...]
3207: Executed instructions:
3208: 34.83% 34.86% 34.86% 22462729 22484024 22484024 flat_generate_mips
3209: 14.08% 14.10% 14.10% 9080215 9091270 9091676 flat_remap_mips
3210: 8.55% 5515945 render_patch_direct
3211: 5.09% 5.11% 94.96% 3283328 3294022 61247717 _TryRunTics
3212: [...]
3213: Used cycles:
3214: 23.62% 23.69% 23.69% 130702768 131100604 131100604 flat_generate_mips
3215: 12.42% 12.46% 12.46% 68735832 68928816 68930904 flat_remap_mips
3216: 9.77% 9.80% 95.66% 54041148 54238744 529368824 _TryRunTics
3217: 5.80% 5.82% 5.82% 32111536 32193664 32193664 correct_element
3218: [...]
3219: Instruction cache misses:
3220: 37.03% 37.14% 98.57% 1829764 1835261 4870573 _TryRunTics
3221: 11.20% 11.24% 11.24% 553314 555191 555191 _BM_T_GetTicks
3222: 9.44% 9.49% 29.13% 466319 468782 1439340 _NetUpdate
3223: 9.27% 9.29% 9.37% 457899 459197 463217 _HGetPacket
3224: [...]
3225: </pre>
3226:
3227: <p>Now there's a message telling that some of the calls were ignored as
3228: based on "call type" they were determined to be actually returns from
3229: exceptions, not real calls (this is mainly important for callgraph
3230: generation, discussed below).</p>
3231:
3232:
3233: <h4>Interpreting the results</h4>
3234:
3235: <p>In addition to accuracy issues mentioned in previous Profiling
3236: sections, function/symbol level costs have gotchas of their own.</p>
3237:
3238: <p>The first cost percentage and count columns are <em>sums for all
3239: the instructions</em> that were in profile data file <em>between
3240: the indicated symbol's address and the address of the next symbol</em>
3241: (= "between-symbols" cost).</p>
3242:
3243: <p><strong>NOTE:</strong> If your symbol file doesn't contain addresses
3244: for all the relevant symbols, results from this can be misleading because
3245: instructions costs get assigned to <em>whatever</em> symbol's address
3246: happened to preceed those instructions. And you don't see which
3247: caller is causing it from caller info or callgraph either, as entry
3248: point for that time sink lacking a symbol means profiler hadn't
3249: tracked calls to it...</p>
3250:
3251: <p>The next two cost percentage and count columns are for <em>subroutine
3252: calls costs</em>, first one for exclusive and latter for inclusive cost
3253: i.e. including costs for further subroutine calls. Values are based on
3254: caller information documented above.</p>
3255:
3256: <p>Reasons why between-symbol costs, and subroutine call costs can
3257: differ are following:</p>
3258: <ul>
3259: <li>Subroutine terminates before next symbol address: exclusive
3260: cost is smaller than in-between cost <em>because of missing
3261: symbol information</em>
3262: (these are indicated with '*' in statistics).</li>
3263: <li>Subroutine is called more through jumps/branches than through
3264: subroutine calls: inclusive call count may be smaller than
3265: in-between call count which includes branches/jumps.</li>
3266: <li>Subroutine jumps/branches to another function instead of
3267: using subroutine call, or function contains additional
3268: (non-function) labels: exclusive cost is larger than
3269: in-between cost.</li>
3270: <li>Exception happening during subroutine call: exclusive cost is
3271: (slightly) larger than in-between cost.</li>
3272: </ul>
3273:
3274: <p>In the first case, you should check the profile data to find out
3275: whether there are missing symbols for executed function entry points.
3276: You can notice function entry points as address gap and/or code
3277: retrieving arguments from stack, and exit points from RTS instruction.</p>
3278:
3279: <p>Second case can also be seen from the profile data. Call count
3280: is same as count for how many times first instruction is executed
3281: (worst case: large loop on subroutine's first instruction).</p>
3282:
3283: <p>While subroutine costs should be more accurate and relevant, due to
3284: code optimizations many of the functions are not called as subroutines
3285: (on m68k, using JSR/BSR), but just jumped or branced to. Because of
3286: this, it's useful to compare both subroutine and between-symbols
3287: costs. One should be able to see from the profile disassembly which
3288: of the above cases is cause for the discrepancy in the values.</p>
3289:
3290: <p><strong>NOTE:</strong> Before starting to do any serious source
3291: level optimizations, you should <em>always</em> verify from profile
3292: data (disassembly) where exactly the costs are in a function, to make
3293: sure your optimization efforts can actually help the performance.</p>
3294:
3295:
3296: <h4>Generating and viewing callgraphs</h4>
3297:
3298: <p>Callgraphs require saved profile data to contain caller information,
3299: i.e. symbols should have been loaded before starting profiling (normally
3300: done with "symbols prg" command in Hatari debugger).</p>
3301:
3302: <p>Separate callgraphs will be created for each of the costs
3303: (0=calls, 1=instructions, 2=cycles) with the -g option:</p>
3304: <pre>
3305: $ hatari_profile.py <b>-p -g</b> -r program.sym program-profile.txt
3306: [...]
3307: Generating 'program-profile-0.dot' DOT callgraph file...
3308:
3309: Generating 'program-profile-1.dot' DOT callgraph file...
3310:
3311: Generating 'program-profile-2.dot' DOT callgraph file...
3312: [...]
3313: </pre>
3314:
3315: <p>Callgraphs are saved in <a href="http://www.graphviz.org/">GraphViz</a>
3316: "dot" format. Dot files can be viewed:</p>
3317: <ul>
3318: <li>With "dotty" program included with GraphViz</li>
3319: <li>With <a href="http://code.google.com/p/jrfonseca/wiki/XDot">XDot</a>
3320: Python GUI (best option on Linux), or some platform specific viewer</li>
3321: <li>By converting dot file to PostScript or SVG format before
3322: viewing it with viewers for those:
3323: <pre>
3324: $ dot -Tsvg program-profile-1.dot > program-profile-1.svg
3325: </pre>
3326: (problem with most PS/PDF and SVG viewers is that either they
3327: don't allow zooming large callgraphs enough or they use huge
3328: amounts of memory and get very slow)
3329: </li>
3330: </ul>
3331:
3332: <p>Produced callgraph will look like this:</p>
3333: <div style="text-align:center">
3334: <a href="images/callgraph.svg">
3335: <img src="images/callgraph.png" width="953" height="589"
3336: alt="Part of callgraph" />
3337: </a>
3338: </div>
3339:
3340: <p>Interpreting the callgraph:</p>
3341: <ul>
3342: <li>Diamond shaped nodes are symbols called as subroutines.
3343: Values listed in them are subroutine call costs; inclusive
3344: (total) cost with exclusive (own) cost in parenthesis,
3345: followed by inclusive cost count. Exclusive cost is
3346: shown only if it differs from inclusive one.</li>
3347: <li>Ellipse shaped nodes are for other symbols (functions
3348: called using jumps/branches, loop labels etc). Values
3349: listed in them are between-symbols costs, i.e. normally
3350: they're included to inclusive (total) costs shown in
3351: subroutine call node somewhere higher in call hierarchy.</li>
3352: <li>Nodes which exclusive (own) or between-symbols costs
3353: exceed default or explicitly given threshold value,
3354: have gray background.</li>
3355: <li>Both nodes which inclusive or between-symbols cost exceeds
3356: the threshold value, and the arrows to & from them
3357: are marked red.
3358: <li>Arrow types indicate call types; normal arrows subroutine
3359: calls, circles branches/jumps, backarrows returns.
3360: Exception calls and returns are indicated with dashed lines,
3361: unknown calls with dotted lines.</li>
3362: <li>Arrow text tells from which address the call orignated
3363: (inside the calling function). If symbol had multiple
3364: callers, text includes count of calls from that particular
3365: address, and its percentage of all calls done to that
3366: symbol.</li>
3367: </ul>
3368:
3369:
3370: <h4>Making large callgraphs readable</h4>
3371:
3372: <p>If profile is for larger and more varied amount of code
3373: (e.g. program startup), the resulting callgraph can be so
3374: huge it's unreadable.</p>
3375:
3376: <p>If your code has interrupt handlers, they can get called
3377: at any point, which can show in callgraph as "explicit" calls
3378: from the interrupted functions. To get rid of such incorrect
3379: calls, give interrupt handler names to --ignore-to option:</p>
3380: <pre>
3381: $ hatari_profile.py -p -g <b>--ignore-to handler1,handler2</b> -r program.sym program-profile.txt
3382: </pre>
3383:
3384: <p>In large callgraph most of the functions aren't really interesting,
3385: because their contribution to the cost is insignificant. You can
3386: remove large number of them with --no-leafs and --no-intermediate
3387: options, they act <em>only</em> on on nodes which costs are below
3388: given threshold. Leaf nodes are ones which don't have any parents
3389: and/or children and intermediate ones which have only single parent
3390: and children (node calling itself is not taken into account).
3391:
3392: <p>Threshold for this is given with the --limit (-l) option. With
3393: that it typically makes also sense to change the node emphasis
3394: threshold with --emph-limit (-e) option:</p>
3395: <pre>
3396: $ hatari_profile.py -p -g <b>-l 0.5 -e 2.0</b> -r program.sym program-profile.txt
3397: </pre>
3398:
3399: <p>If you're not interested in from how many different addresses
3400: a given function calls another function, use --compact option. If you
3401: still see multiple calls between two nodes with it, the reason is that
3402: they happened through different call paths which were removed from
3403: the callgraph after --compact option was applied:</p>
3404: <pre>
3405: $ hatari_profile.py -p -g -l 1.0 -e 2.0 <b>--no-leafs --no-intermediate --compact</b> -r program.sym program-profile.txt
3406: </pre>
3407:
3408: <p>If even this doesn't help, you can remove all nodes below
3409: the given cost threshold limit with --no-limited option, but this
3410: often doesn't leave much of a call hiararchy. Instead you may
3411: consider removing all nodes except for subroutine call ones with
3412: --only-subroutines option.</p>
3413:
3414: <p>If you have trouble locating nodes you're specially interested
3415: about, you can either color them differently with the --mark option,
3416: or exclude everything else from the callgraph except those nodes and
3417: their immediate callers & callees, with the --only option:</p>
3418: <pre>
3419: $ hatari_profile.py -p -g <b>--only func1,func2</b> -r program.sym program-profile.txt
3420: </pre>
3421:
3422: <p>Last option for reading the callgraph is using -k option to
3423: export the data for use in (Linux) Kcachegrind UI. It generates
3424: callgraphs on the fly, and just for the area around the function
3425: you selected, so navigating in callgraph may be easier. It also
3426: shows the related profile disassembly, which can make verifying
3427: matters easier:</p>
3428: <pre>
3429: $ hatari_profile.py <b>-p -k</b> -r program.sym program-profile.txt
3430: [...]
3431: Generating callgrind file 'program-profile.cg'...
3432: [...]
3433: $ kcachegrind program-profile.cg
3434: </pre>
3435: <div style="text-align:center">
3436: <img src="images/kcachegrind.png" width="887" height="442"
3437: alt="Kcachegrind screenshot" />
3438: </div>
3439:
3440:
3441: <h3>Usage examples</h3>
3442:
3443: <p>
3444: Here's a list of some common debugging tasks and how to do them
3445: with the Hatari debugger:
3446: </p>
3447:
3448: <dl>
3449: <dt><em>Stopping on program startup and examining its data</em></dt>
3450: <dd>Please see <a href="#Breakpoint_variables">Breakpoint variables</a>
3451: and <a href="#Inspecting_emulation_state">Inspecting emulation state</a>
3452: sections.
3453: </dd>
3454:
3455: <dt><em>Tracing specific things in the system</em></dt>
3456: <dd>To trace e.g. all GEMDOS calls and IO operations, use:
3457: <pre>
3458: trace gemdos,io_all
3459: </pre>
3460: Please see <a href="#Tracing">Tracing</a> section for more information
3461: on tracing, what's possible with it and what are its limitations.
3462: </dd>
3463:
3464: <dt><em>Stopping when certain PC address is passed Nth time</em></dt>
3465: <dd>To stop e.g. after function/subroutine at $12345 is called for
3466: the 6th time:
3467: <pre>
3468: a $12345 :6
3469: </pre>
3470: </dd>
3471:
3472: <dt><em>Stopping when specific exception happens</em></dt>
3473: <dd>Hatari's -D option doesn't invoke debugger on all exceptions and
3474: doesn't allow invoking debugger just for specific exceptions. To
3475: stop at specific exception, one can check when it's called.
3476: At the start of memory is the CPU exception table for exception
3477: handler addresses, so to stop e.g. at bus error with some extra
3478: information, one can use following:
3479: <pre>
3480: history on
3481: b pc=($8)
3482: </pre>
3483: After bus error invokes debugger, 'history' command can then be used
3484: to see (executed memory addresses with their current) instructions
3485: leading to the error. The most interesting vector addresses are:
3486: $8 (Bus error), $C (Address error), $10 (Illegal instruction),
3487: $14 (Division by zero).
3488: </dd>
3489:
3490: <dt><em>Stopping when register has a specific value</em></dt>
3491: <dd>To stop when e.g. D1 register contains value 5, set a breakpoint on:
3492: <pre>
3493: b d1 = 5
3494: </pre>
3495: </dd>
3496:
3497: <dt><em>Stopping when a register value changes</em></dt>
3498: <dd>To stop when e.g. D1 register value changes, set a breakpoint on:
3499: <pre>
3500: b d1 ! d1
3501: </pre>
3502: </dd>
3503:
3504: <dt><em>Stopping when register value is within some range</em></dt>
3505: <dd>To stop when e.g. D1 register value is within range of 10-30,
3506: set a breakpoint on:
3507: <pre>
3508: b d1 > 9 && d1 < 31
3509: </pre>
3510: </dd>
3511:
3512: <dt><em>Stopping when memory location has a specific value</em></dt>
3513: <dd>To stop when e.g. bit 1 of the Video Shifter Sync Mode byte at
3514: IO address $ff820a is set i.e. video frequency is 60Hz, set
3515: a breakpoint on:
3516: <pre>
3517: b ($ff820a).b & 2 = 2
3518: </pre>
3519: </dd>
3520:
3521: <dt><em>Stopping when a memory value changes</em></dt>
3522: <dd>To stop when above bit changes, set a breakpoint on its value
3523: being different from the current value ('!' compares for inequality):
3524: <pre>
3525: b ($ff820a).b & 2 ! ($ff820a).b & 2
3526: </pre>
3527: </dd>
3528:
3529: <dt><em>Tracing all changes in specific memory location</em></dt>
3530: <dd>To see the new values and continue without stopping, add
3531: the ":trace" breakpoint option:
3532: <pre>
3533: b ($ff820a).b & 2 ! ($ff820a).b & 2 :trace
3534: </pre>
3535: </dd>
3536:
3537: <dt><em>Stopping at specific screen position</em></dt>
3538: <dd>To stop e.g. when VBL is 100, HBL is 40 and line cycles is 5,
3539: use the corresponding debugger variables:
3540: <pre>
3541: b VBL = 100 && HBL = 40 && FrameCycles = 5
3542: </pre>
3543: </dd>
3544:
3545: <dt><em>Stopping after value increases/decreases by certain amount</em></dt>
3546: <dd>To stop e.g. after D0 value has increased by 10, set breakpoint on:
3547: <pre>
3548: b d0 = "d0 + 10"
3549: </pre>
3550: </dd>
3551:
3552: <dt><em>Examining specific system call return value</em></dt>
3553: <dd>To check e.g. what's the Fopen() GEMDOS call return value,
3554: check with "info gemdos 1" its opcode, set a breakpoint for that
3555: and step to next (n) instruction from the trap call when breakpoint
3556: is hit. GEMDOS call return value is then in register D0:
3557: <pre>
3558: > trace gemdos
3559: > b GemdosOpcode = $3D
3560: > c
3561: [...continue until breakpoint...]
3562: 1. CPU breakpoint condition(s) matched 1 times.
3563: GemdosOpcode = $3D
3564: > n
3565: GEMDOS 0x3D Fopen("TEST.TXT", read-only)
3566: > e d0
3567: = %1000000 (bin), #64 (dec), $40 (hex)
3568: </pre>
3569: </dd>
3570:
3571: <dt><em>Seeing code leading to a breakpoint</em></dt>
3572: <dd>To see CPU instructions executed before debugger was entered,
3573: you need to enabled history tracking <em>before</em> it. Whenever
3574: debugger is entered, you can then request given number (here 16) of
3575: past instructions to be shown:
3576: <pre>
3577: history cpu
3578: c
3579: [breakpoint is hit and debugger entered]
3580: history 16
3581: </pre>
3582: </dd>
3583:
3584: <dt><em>Getting instruction execution history for every breakpoint</em></dt>
3585: <dd>
3586: To see last 16 instructions for both CPU and DSP whenever
3587: (a normal or tracing) breakpoint is hit:
3588: <pre>
3589: history on
3590: lock history 16
3591: c
3592: </pre>
3593: </dd>
3594:
3595: <dt><em>Single stepping so that new register values are shown after each step</em></dt>
3596: <dd>
3597: <pre>
3598: lock registers
3599: s
3600: [new register values]
3601: s
3602: [new register values]
3603: ...
3604: </pre>
3605: </dd>
3606:
3607: <dt><em>Showing current stack contents</em></dt>
3608: <dd>To see first 64 bytes on top of the stack, use:
3609: <pre>
3610: m "a7-64"-a7
3611: </pre>
3612: </dd>
3613:
3614: <dt><em>Seeing specific information each time debugger is entered</em></dt>
3615: <dd>To see above information whenever some breakpoint is hit,
3616: you enter debugger manually etc, write that command to e.g.
3617: <span class="file">stack.ini</span> file and then use:
3618: <pre>
3619: lock file stack.ini
3620: </pre>
3621: Please see also <a href="#Chaining_breakpoints">Chaining breakpoints</a>
3622: section for more examples on what you can do with the debugger input files.
3623: </dd>
3624:
3625: <dt><em>Finding where a program or the OS is stuck</em></dt>
3626: <dd>Profiling tells from which addresses CPU is executing the instructions:
3627: <pre>
3628: profile on
3629: c
3630: [after a while, use AltGr+Pause to get back to debugger]
3631: profile counts
3632: </pre>
3633: Please see <a href="#Profiling">Profiling</a> section for more info.
3634: </dd>
3635:
3636: <dt><em>Seeing program callstack when breakpoint is hit</em></dt>
3637: <dd><a href="#Caller_information">Profiler caller data</a> includes
3638: callstack information (with some limitations).
3639: </dd>
3640:
3641: <dt><em>Seeing call backtraces whenever given function is called</em></dt>
3642: <dd>Enable profiling, load symbols for the program and set breakpoint
3643: for the function you're interested about, in the following way:
3644: <pre>
3645: profile on
3646: symbols prg
3647: b pc = _my_function :quiet :noinit :file showstack.ini
3648: </pre>
3649: I.e. whenever 'my_function' address is called, quietly trigger a
3650: breakpoint without reseting profiling (callstack) information and run
3651: debugger command(s) from the 'showstack.ini' debugger script file,
3652: which contains following command:
3653: <pre>
3654: profile stack
3655: </pre>
3656: </dd>
3657:
3658: <dt><em>Seeing how program functions/symbols call each other</em></dt>
3659: <dd><a href="#Profile_data_post-processing">Profile data
3660: post-processing</a> can provide execution callgraphs.
3661: </dd>
3662:
3663: </dl>
3664:
3665: <p>
3666: Hint: for most of the above commands, one just needs to prefix them with
3667: "d" (or "dsp" when using full command names) to do similar operation on
3668: the DSP.
3669: </p>
3670:
3671:
3672: <h3 id="Build_notes">Build notes</h3>
3673:
3674: <p>
3675: Lastly, the debugger is much nicer to use with the command line
3676: history, editing and especially the completion support for the
3677: command, command argument and symbol names.
3678: </p>
3679: <p>
3680: If you're building Hatari yourself, please make sure that you have the
3681: GNU readline development files installed (on Debian / Ubuntu these
3682: come from the libreadline5-dev package). Otherwise the name completion
3683: and other features don't get enabled when you configure Hatari.
3684: </p>
3685: <p>
3686: ENABLE_TRACING define needs to be set for tracing to work.
3687: By default it should be enabled.
3688: </p>
3689:
3690:
3691: <h2 id="Performance">Performance</h2>
3692:
3693: <p>Hatari performance varies between Atari programs, depending on what
3694: features Hatari needs to emulate for them. Less accurate Atari
3695: emulators may be faster as emulation accuracy has a performance
3696: overhead.</p>
3697:
3698: <p>The operating system and libraries below Hatari can also sometimes
3699: have a noticeable effect on performance.</p>
3700:
3701:
3702: <h3>Improving Hatari performance</h3>
3703:
3704: <p>
3705: Hatari currently runs best in 16 or 32 bits per pixel color depth
3706: mode, so try to avoid 24 bits per pixel display modes if possible.
3707: 16-bit mode is fastest.
3708: </p>
3709:
3710: <p>
3711: <em>On OSX, frame skipping, zooming and drive LED options (listed below)
3712: seem to have a large effect on performance in the windowed mode</em>.
3713: This is apparently due to issues in the SDL OSX backend and how OSX
3714: itself composites non-fullscreen window contents. OSX uses always
3715: 32-bit mode.
3716: </p>
3717:
3718: <p>
3719: Unless you've disabled compiler optimizations (like GCC's -O2 or -O3
3720: options) in the Hatari build, the extra optimization flags (like GCC's
3721: "-mtune=i686") don't seem to have very large effect on Hatari
3722: performance. Using GCC -O3 option instead of -O2 can give minor
3723: (5-10%) performance improvements for things (demos) that use very
3724: heavily interrupts.
3725: </p>
3726:
3727: <p>
3728: However, Hatari can be sped up considerably by giving up some
3729: emulation or emulator accuracy. Except for DSP, these options
3730: should be needed only on very slow devices like handhelds. See below.
3731: </p>
3732:
3733: <p>
3734: If nothing else helps, try an earlier Hatari version. More accurate
3735: emulation or emulator output in newer Hatari versions means that they
3736: can be slower despite optimizations.
3737: </p>
3738:
3739:
3740: <h3>Emulation options</h3>
3741:
3742: <p>
3743: Emulation options have the largest impact on performance.
3744: These options can be changed from the Hatari GUI System dialog and
3745: the emulation needs to be rebooted for any of these changes to take
3746: an effect! They're enabled by default.
3747: </p>
3748:
3749: <h4>DSP</h4>
3750: <p>
3751: Emulating the Falcon DSP is performance-wise several times more demanding
3752: than emulating the m68k; DSP runs at higher frequency, executes many
3753: instructions for each m68k instruction and emulation isn't as mature
3754: and optimized. Unless some Falcon program needs DSP, <em>none</em> or
3755: <em>dummy</em> DSP emulation mode could be used. Even of the programs
3756: that do use DSP, many use it only for background music and work
3757: fine without the real DSP emulation.
3758: </p>
3759:
3760: <h4>Timer-D</h4>
3761: <p>
3762: The single largest factor contributing to general Hatari emulation
3763: performance is the handling of interrupts. Enabling Timer-D patching
3764: option (about) doubles Hatari ST/STE emulation performance as it
3765: significantly reduces the number of interrupts generated by the emulated
3766: Atari machine. Using this has adverse effect only for very rare programs.
3767: </p>
3768:
3769: <h4>FDC</h4>
3770: <p>
3771: While accurate FDC emulation doesn't take that much CPU, it slows down
3772: floppy image accesses (and Hatari startup) a lot. Only <em>very</em>
3773: few demos and games require accurate FDC emulation for their copy protection,
3774: so enabling fast floppy access is fairly safe.
3775: </p>
3776:
3777: <h4>Compatible CPU</h4>
3778: <p>
3779: After the DSP and interrupts, m68k emulation takes most time.
3780: Disabling the "Slower but more compatible CPU" option will speed up
3781: the emulation a lot, but it won't anymore be cycle accurate. This can
3782: be fine for many games and other programs, but won't work e.g. for demos
3783: using overscan or rasters.
3784: </p>
3785:
3786: <p>
3787: Roughly speaking, for DSP emulation, one needs at least 2Ghz machine.
3788: For normal (unpatched) Timer-D frequency on some specific cases (like
3789: demos with overscan 512 color animations) one may need over 1GHz
3790: machine, but some rare ST/STE demos may require over 1GHz machine even
3791: with Timer-D patching. For "Compatible CPU" one needs at least 1/2Ghz
3792: machine.
3793: </p>
3794:
3795: <p>
3796: <strong>NOTE</strong>: Above options may cause some programs to work in correctly.
3797: The <a href="compatibility.html">Hatari Software Compatibility List</a>
3798: lists programs known to need real real Falcon DSP emulation, Timer-D
3799: frequency or accurate FDC timings. Disabling "Compatible CPU" option
3800: is recommended only as a last resort.
3801: </p>
3802:
3803:
3804: <h3>Emulator options</h3>
3805:
3806: <p>
3807: Emulator options don't usually have as large effect on performance as
3808: emulation options, but they don't affect the emulated programs at all,
3809: just the quality of the emulation "output". These options can also
3810: be toggled at run-time without rebooting the emulation.
3811: </p>
3812:
3813: <h4>Sound</h4>
3814: <p>
3815: Internal Hatari sound handling and the SDL_mixer sound thread
3816: libALSA sound processing can account up to 1/3 of the Hatari CPU usage
3817: in normal ST/STE emulation. Disabling sound will get rid of that.
3818: Using low sound frequency or one matching your sound card may also help.
3819: Best is if you disable also background music from the programs you run
3820: in Hatari as this can significantly reduce the number of generated
3821: interrupts.
3822: </p>
3823:
3824: <h4>Frame skipping</h4>
3825: <p>
3826: Screen rendering can take noticeable amount of CPU time. The default
3827: Hatari "auto" frame skipping should be used unless there's a good
3828: reason not to. It will skip converting and showing some of the frames
3829: if there's not enough time for them.
3830: </p>
3831: <p>
3832: Also, if your monitor refresh frequency is lower than the selected
3833: Hatari monitor frequency (e.g. LCD monitors usually use 60Hz whereas
3834: Atari monochrome monitor uses 71Hz), you should use frameskip of one.
3835: The reason is that if your SDL library uses VSync to synchronize the
3836: output to screen (like OSX one?), with zero frame skip that forces the
3837: emulation to run slower than a real Atari. If SDL doesn't use VSync,
3838: Hatari does redundant work to convert frames you can't see.
3839: </p>
3840:
3841: <h4>Zooming</h4>
3842: <p>
3843: If you are not using frame skip, disabling zooming can have
3844: noticeable improvement on performance. You can do this by specifying
3845: suitably low "Max zoomed" resolution (<span class="commandline">--zoom
3846: 1</span> command line option sets it to 320x200). If you still want to
3847: have a nice fullscreen mode, you should rather add the right resolution
3848: mode-lines (e.g. "320x200") to your xorg.conf file. If you still want
3849: to use zooming, disabling borders may help a bit.
3850: </p>
3851:
3852: <h4>Spec512 color handling</h4>
3853: <p>
3854: Handling Spec512 color modes which change the ST/e palette constantly
3855: takes some extra CPU. If you have problems with CPU usage in such
3856: screens and you care more e.g. from the sound quality than visuals, you
3857: can either increase the threshold or disable the Spec512 mode handling
3858: completely by zeroing the threshold for that with the
3859: <span class="commandline">--spec512 0</span> option.
3860: </p>
3861:
3862: <h4>Statusbar and drive LED</h4>
3863: <p>
3864: If your version of the SDL library uses VSync to synchronize the screen
3865: output, drawing of the statusbar or the drive LED may have some minor
3866: impact on performance too. Normally they shouldn't.
3867: </p>
3868:
3869:
3870: <h3>Measuring the performance</h3>
3871:
3872: <p>
3873: There are a couple of ways to monitor and measure Hatari performance.
3874: </p>
3875: <p>
3876: By default Hatari has Statusbar visible and automatic frameskip
3877: enabled. When Hatari has enough time that it can sleep a little each
3878: frame, the statusbar frame skip ("FS") value keeps at zero. If Hatari
3879: is completely busy, it will increase to the maximum specified
3880: (automatic) frame skip value.
3881: </p>
3882: <p>
3883: Hatari has also a facility to measure FPS i.e. Frames Per Second.
3884: Just enable the <span class="commandline">--fast-forward</span> option
3885: on command line (or use the corresponding keyboard shortcut), and
3886: after a while, press the "Pause" key. Whenever Hatari emulation is
3887: paused, Hatari will output on console how many VBLs it could show per
3888: second along with some other numbers.
3889: </p>
3890: <p>
3891: It depends on what you want to measure, but usually it's best to
3892: disable sound and set high frame skip like
3893: <span class="commandline">--sound off --frameskips 60</span> so that
3894: the associated external overheads are minimized. E.g. video output
3895: can on some platforms do VSync and measurements would then show your
3896: monitor refresh frequency instead of the actual Hatari performance.
3897: </p>
3898: <p>
3899: On Unix systems with <span class="commandline">times()</span> function
3900: call, only the time spent by the Hatari process itself is measured.
3901: On other systems, much less accurate SDL "wall clock" timings are
3902: used. To make latter more accurate you could use also
3903: <span class="commandline">--run-vbls</span> option to specify how many
3904: VBLs Hatari should run before it exits. In this case it's best to
3905: either have the test-case run automatically from the AUTO-folder or
3906: given as memory snapshot to Hatari with the frame skip set equal to
3907: the VBL count.
3908: </p>
3909: <p>
3910: Note that these numbers can fluctuate quite a bit, <em>especially</em>
3911: when the SDL timings are used, so for (statistically) reliable numbers
3912: you may need to repeat the measurement several times. You should of
3913: course make also sure that the system doesn't have any other activity
3914: at the same time you're making the measurements.
3915: </p>
3916:
3917:
3918: <h2>Appendix</h2>
3919:
3920: <h3>Copying</h3>
3921:
3922: <div class="backdropped">
3923: <p>This program is free software; you can redistribute it and/or modify
3924: it under the terms of the GNU General Public License as published by
3925: the
3926: Free Software Foundation; either version 2 of the License, or (at your
3927: option) any later version. </p>
3928: <p>This program is distributed in the hope that it will be useful, but <em>WITHOUT
3929: ANY WARRANTY</em>; without even the implied warranty of <em>MERCHANTABILITY</em>
3930: or <em>FITNESS FOR A PARTICULAR PURPOSE</em>. See the GNU General
3931: Public License for more details. </p>
3932: <p>
3933: You should have received a copy of the GNU General Public License
3934: along with this program; if not, write to the Free Software Foundation,
3935: Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
3936: </p>
3937: </div>
3938: <p><a href="http://www.gnu.org/">The GNU Project and the Free Software
3939: Foundation</a> | <a href="http://www.fsf.org/licenses/gpl.html">The
3940: GNU General Public License</a></p>
3941:
3942: <h3>Introduction to Emulation</h3>
3943:
3944: <p>Emulation via software is an art and Hatari is an example of this.</p>
3945: <p>Emulation is to make a computer behave like a (probably) completely
3946: different machine on the lowest possible niveau.
3947: This includes CPU and custom chip emulation allowing software written
3948: for the emulated machine to be run without notice.
3949: A good emulator will run most of the software intended for the emulated
3950: platform without trouble.
3951: </p>
3952: <p>
3953: The key to emulation is to simply do those things with a software
3954: program, the emulator, that normally chips would perform.
3955: So you have an CPU emulator that basically consists of a large loop
3956: that does exactly what the real thing would do:
3957: </p>
3958: <ul>
3959: <li>fetch an instruction from virtual memory</li>
3960: <li>interpret this instruction</li>
3961: <li>fetch operands from the emulated registers and memory</li>
3962: <li>perform the operation like addition or changing the program
3963: counter on a jump instruction</li>
3964: <li>writes results back into the intended registers or memory
3965: locations</li>
3966: <li>increment of the program counter and loop</li>
3967: </ul>
3968: <p>
3969: The typical von-Neumann CPU can be emulated very fast, stable and
3970: error-free using such a simple loop system.
3971: </p>
3972: <p>
3973: But in most cases the CPU emulation is the simplest part. Correct
3974: emulation of the various custom chips and hardware
3975: parts of the emulated system is much trickier.
3976: </p>
3977:
3978: <hr>
3979:
3980: </body>
3981: </html>
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.