Return to ChangeLog CVS log

Up to [UNIX Amiga Emulator] / uae / src

Annotation of uae/src/ChangeLog, revision 1.1.1.3

1.1.1.3 ! root        1: 980508 Fix compilation problem on ELF systems introduced by BSD patches.
        !             2:        configure.in, aclocal.m4, Makefile.in: Use AM_PATH_GTK from gtk
        !             3:        distribution.
        !             4:        Don't disable GGI if we can't find svgalib.
        !             5: 980504 xwin.c: Call XAllocColorCells for dithered modes. Don't set image_mem
        !             6:        in get_image, instead arrange for it to store the address in a pointer
        !             7:        chosen by the caller.
        !             8:        BSD patches by Krister Walfridsson.
        !             9: 980425 picasso96.c: In BlitPattern, correctly handle the pattern offsets.
        !            10: 980421 BeOS update (Christian Bauer)
        !            11: 980419 In pfield_doline_unaligned_h, don't increment dataptr.
        !            12: 980418 gtkui.c: Don't crash if -G option is given.
        !            13: 980416 custom.c: Don't use bplhires in do_modulos.
        !            14: 
        !            15: Release 0.7.0
        !            16: 980401 gengenblitter.c: Complete rewrite. Much smaller, much faster, much
        !            17:        better results.
        !            18:        Check for -lXi to support newer versions of gtk+.
        !            19: 980316 gencpu.c (read_counts): Two silly bugs fixed.
        !            20:        main.c (save_options): Don't write out 'a' if cpu_level < 2
        !            21:        configure.in: Changes to try to be compatible with Solaris /bin/sh.
        !            22: 980315 New UAE-Control version from Tauno & Cloanto.
        !            23:        Make configure set UNALIGNED_PROFITABLE for x86.
        !            24:        Get rid of "-o", add save command to tui.c and gtkui.c.
        !            25:        Fix some bugs in save_options.
        !            26: 980314 Many improvements in the gtk UI. Made CPU type changeable at run-time.
        !            27:        Make DGA && ! VIDMODE configuration compile. Kludge around the fact
        !            28:        that DGA misses colormap changes. Call XWarpPointer to try to avoid
        !            29:        the click-moves-framebuffer bug.
        !            30:        Make sure p2c.[ch] don't get deleted.
        !            31: 980313 Move all CPU_LEVEL handling from build68k.c to gencpu.c. In gencpu.c,
        !            32:        ignore all parameters and don't write to standard output, instead
        !            33:        hardcode all filenames and generate them all at once. Arrange to
        !            34:        generate code for all CPU emulator variants and to select them at
        !            35:        run-time.
        !            36:        Get rid of copper cheat mode, re-use "-C" option for CPU selection.
        !            37: 980312 In COPxLCL, mask out LSB (Axel Wefers).
        !            38:        newcpu.h: Always compile in both get_disp_ea versions with a different
        !            39:        name. Delete USE_POINTER option; always use pointer. Don't fill
        !            40:        prefetch in m68k_setpc, do it in callers where necessary. Compile in
        !            41:        two versions of get_ixxxx, with different names for prefetch and
        !            42:        non-prefetch. Change gencpu.c to use these appropriately.
        !            43:        Don't generate 16 small CPU emulation source files. With a decent
        !            44:        machine it's no problem to compile all the code in one file. Also turn
        !            45:        off -O3 for gcc, -O2 does fine since all the necessary __inline__s are
        !            46:        there, and the compiler takes less memory that way.
        !            47:        Don't include <linux/joystick.h>, duplicate some definitions from
        !            48:        there instead. Makes configure.in simpler, too.
        !            49:        Fix Makefile.in/configure.in and a couple of other places to support
        !            50:        building in a separate directory (sort of; not complete yet).
        !            51:        Remove p2c.[ch] and genp2c when doing make clean.
        !            52:        Delete code to support Tcl/Tk and XForms GUIs.
        !            53: 971215 More Amiga and BeOS patches from Sam and Christian.
        !            54: 971205 Move some more code from custom.c to audio.c. Optimize sample handlers
        !            55:        a bit.
        !            56:        Make RDTSC asm statement volatile.
        !            57: 971203 Latest GGI patch from Christian Schmitt.
        !            58:        Fix configure to enable zfile decompression for GGI target.
        !            59:        Fix mouse button mapping in ggi.c.
        !            60: 971128 Fix double definition of gfxvidinfo in custom.c.
        !            61:        Fix impossible register constraints in asm statement in newcpu.c
        !            62:        Sprinkle casts and rename some variables to make C++ compilers happy.
        !            63:        Add BeOS patches by Christian Bauer.
        !            64: 971125 More Amiga patches from Sam.
        !            65: 971120 Get rid of DONT_WANT_SOUND.
        !            66: 971119 Support USS sound not only on Linux (Douglas Carmichael)
        !            67:        Update Win32 and DOS ports.
        !            68:        Add Polish documentation by Tomasz Sienicki.
        !            69: 971116 Add a missing return statement in filesys.c (Brian King)
        !            70: 971112 Fix silly error in xwin.c.
        !            71: 971108 Add some more decompressors to zfile.c. Set FLOPPY_SPEED to 4.
        !            72: 971107 Make ggi.c compile again, fix a few bugs and add Picasso support.
        !            73: 971106 Fill prefetch queue after executing an encrypted ROM illegal insn.
        !            74: 971105 Initialize the sprites during customreset now that init_hardware_frame
        !            75:        no longer does it.
        !            76: 971104 Implement action_set_file_size as required by the specs.
        !            77:        SetPanning/picasso96_state: XOffset and YOffset are signed.
        !            78: 971103 xwin.c: Offer 640x512 and 320x256 resolutions in windowed Picasso mode.
        !            79:        Add a few more memory banks in memory.c (A3000 stuff - nonfunctional)
        !            80:        Don't map kickmem at address 0 if using replacement Kickstart.
        !            81:        Must fill prefetch buffer when calling calltraps or ersatz functions.
        !            82:        Write out hardfile parameters in write_filesys_config().
        !            83:        xwin.c: Support 15 bit displays.
        !            84: 971101 Add latest patches by Samuel Devulder. Improve hardfile support in
        !            85:        tui.c.
        !            86: 971031 Offer "-U" only ifdef PICASSO96.
        !            87:        Add support for mounting multiple hardfiles in any order.
        !            88:        Don't reset sprite state in init_hardware_frame. Instead, allow SPRxPT
        !            89:        functions to set the state to SPR_restart in the vertical blanking
        !            90:        area.
        !            91: 971030 Fix 8MB Chip RAM.
        !            92:        filesys_dev_storeinfo: New function and calltrap. Use it to store
        !            93:        per-unit data (which now includes geometry information) in the
        !            94:        parampacket. Modify filesys.asm to use it instead of trying to find
        !            95:        the values by itself.
        !            96: 971028 "AMIROMTYPE1" has 11 characters, not 0.
        !            97: 971027 Implement encrypted ROM features. Comment out non-working InvertRect
        !            98:        optimization.
        !            99: 971026 Fix some problems in PlanarToDirect and CopyColorIndexMappingA2U.
        !           100: 971022 Remove fixed paths in genp2c.c (oops...)
        !           101:        Some more XF86VidMode hacks.
        !           102:        Include target.h in sysdeps.h. Define PICASSO96_SUPPORTED for X11,
        !           103:        SVGAlib and Win32. Don't compile in Picasso support if this macro isn't
        !           104:        defined.
        !           105: 971021 xwin.c: Picasso code wants to use bit_unit, not bitdepth. Also correct
        !           106:        byteorder (it's B8G8R8A8 from the Amiga's point of view).
        !           107:        Fix Picasso InvertRect function.
        !           108:        Half-hearted attempt at 24 bit support in the X11 version, with some
        !           109:        awful code in custom.c.
        !           110:        Make the X11/VidMode version try to resize the screen for Picasso mode
        !           111:        switches.
        !           112: 971020 Delete option -5, make -4 take an argument. Disable z3fastmem and
        !           113:        Picasso emulation when using a 24 bit address space.
        !           114: 971019 record_color_change: No need to remember a color table if the current
        !           115:        line hasn't been decided yet.
        !           116:        Add XF86VidMode extension support.
        !           117:        Integrate Stefan's Zorro III support, add support for up to 64MB fast.
        !           118:        xwin.c: Refresh the whole Picasso screen if there are Expose events.
        !           119:        Fix module length calculation for *Tracker modules.
        !           120:        Use a 32 bit address space by default if CPU_LEVEL > 1. New option -5.
        !           121: 971018 Apparently it's not possible to turn on bitplane DMA in the middle of
        !           122:        the line, so call post_decide_line only when the number of planes is
        !           123:        modified, not in DMACON.
        !           124:        genp2c.c: New file, generates planar to chunky conversion routines.
        !           125:        Remove those which are now generated from custom.c.
        !           126: 971017 Make row_map an array, not a pointer to one.
        !           127:        Audio periods may not be 0, so initialize them to 65535 in customreset.
        !           128:        Implement "S" command in the debugger. Add a bunch of module detection
        !           129:        routines.
        !           130: 971015 od-linux/sound.c: If FRAME_RATE_HACK, reduce vsynctime a bit to give
        !           131:        the sound emulation more room to breathe.
        !           132: 971013 Merge with Brian's latest Picasso code.
        !           133: 971012 Fix a few more portability problems that showed up when compiling with
        !           134:        Watcom C. Also remove superfluous includes from some files.
        !           135:        Add md-i386-watcom directory. Update od-win32 with latest code.
        !           136:        Remove fast_memcmp routine.
        !           137:        Include penguin.h before xwin.h everywhere. Declare a few semaphores
        !           138:        for multithreaded setups in xwin.h. Move inhibit_frame stuff to xwin.h.
        !           139: 971011 gencpu.c: Don't generate labels without following statement. Don't
        !           140:        directly generate assembly statements if X86_ASSEMBLY, instead generate
        !           141:        code that uses macros defined in machdep/m68k.h.
        !           142: 971008 xwin.c, svga.c, custom.c, uaelib.c: Add picasso96 code. Add Picasso
        !           143:        selection to configure.
        !           144:        Move common variables (mouse stuff, gfxvidinfo, xcolors) from graphics
        !           145:        files to custom.c.
        !           146:        memory.c: custom chip area isn't supposed to be mapped at 0x200000. Use
        !           147:        dummy bank for now.
        !           148:        od-linux/joystick.c: Open different devices for different joysticks.
        !           149:        From Win32 sources: set vsyncmintime in m68k_go.
        !           150: 971005 genblitter.c: blitzero is initialized to 1, not to 0, so we need to
        !           151:        clear, not set it.
        !           152: 970928 Add another user interface for X11 using libgtk+.
        !           153: 970925 audio.c: For stereo sound, double all lookup table values.
        !           154:        Added (preliminary) win32.c.
        !           155: 970923 DISK_check_change: Should call gui_filename when ejecting.
        !           156: 970922 Fix od-linux/joystick.c compilation problem with recent kernels.
        !           157:        Merge results of the UAE meeting:
        !           158:        New serial code (Stefan Reinauer)
        !           159:        Expansion "test card" support (Stefan Reinauer)
        !           160:        New GGI code (Christian Schmitt), also configure/Makefile changes
        !           161:        to make it compile.
        !           162:        Fix 68020/FPU code.
        !           163:        Call do_mousehack before every read of the mouse position to get more
        !           164:        accurate values.
        !           165:        Add sd-file directory; add support for it in configure.
        !           166:        Add stereo support in audio.c, od-linux/sound.[ch], and main.c.
        !           167:        Fix copper problem: call decide_line in calc_copcomp_true; and if the
        !           168:        copper wouldn't wait normally (position reached), but the hardware is
        !           169:        fetching data in Hires 4 plane mode, delay until ddf stop.
        !           170:        t-ascii.h: Asciiart target understands "-x" parameter.
        !           171: 970823 gencpu.c [X86_ASSEMBLY]: Turn input constraints for scratch into output
        !           172:        constraints.
        !           173: 970626 Fix 24 bit address space logic (Mathias Ortmann).
        !           174:        Raise exception 3 for Bcc.L insns if CPU_LEVEL < 1. Reported by Stefan
        !           175:        Ropke.
        !           176:        Remove USER_PROGRAMS_BEHAVE for normal emulator. Remove hacks to
        !           177:        generate expanded CPU functions for specific opcodes. Replace with code
        !           178:        that reads an instruction count file and sorts gencpu output by
        !           179:        frequency of the called opcodes.
        !           180: 970625 Calltraps (at least those with call_m68k calls in them) were broken by
        !           181:        the CPU emulator changes. Now use Axxx opcodes as calltraps.
        !           182:        Include pOS port by Samuel Devulder.
        !           183: 970624 Solaris sound patches by Manfred Thole.
        !           184:        gencpu.c: The generated code shouldn't just check for odd addresses,
        !           185:        it should also call Exception...
        !           186: 970622 gencpu.c: don't emit return statements; make code jump to a function
        !           187:        end label instead. Make genamode() check for odd addresses, also reduce
        !           188:        code duplication. Interpret getv parameter so that data is fetched if
        !           189:        getv == 1, and the address is checked if getv != 0. Remove exception 3
        !           190:        checks from get_long() etc. Get rid of buserr.
        !           191:        cpuopti.c: Make a few more checks whether optimization is possible. Add
        !           192:        code to undo GCC's epilogue scheduling.
        !           193:        newcpu.c: Split all the specialflag handling into new function.
        !           194:        Implement -w option as in Win32 port. tui.c: add it to misc menu.
        !           195: 907621 Fix yesterday's CPU emulator changes. Make cpuopti aware of functions
        !           196:        with two RET insns. Make sure that NO_EXCEPTION_3 and
        !           197:        NO_PREFETCH_BUFFER are on when emulating a 68020.
        !           198:        Add usage information for src/dst operands to table68k.
        !           199:        Implement CIA A OVL bit properly.
        !           200: 970620 New NeXTwin.m from Ian. Fix a number of prefetch bugs in gencpu.c.
        !           201:        newcpu.h: m68k_incpc(), get_i{byte,word,long}: New function/macros.
        !           202:        Also add oldpc parameter to m68k_do_[bj]sr(). gencpu.c: use all this.
        !           203:        CPU functions now don't use nextiword; they keep the PC constant and
        !           204:        increment it with the right value at the end.
        !           205:        If chipmem_size > 2MB, disable fastmem.
        !           206:        In configure.in and od-linux/joystick.c, use ljoystick.h as name for
        !           207:        the replacement file for a broken <linux/joystick.h>
        !           208:        configure.in: Check for post-gcc-2.7.2 option "-mno-schedule-prologue"
        !           209:        and use it for compiling the optimized CPU files if it exists.
        !           210: 970618 Rename hardfile.device to uaehf.device to avoid name clashes. Suggested
        !           211:        by George Nicolaidis.
        !           212:        Reimplement CPU prefetch by making gencpu aware of it so it generates
        !           213:        code that fetches data from the right places automatically. Also reduce
        !           214:        prefetch length to 4 bytes.
        !           215: 970612 memory.h: Include osdep/memory.h. missing.c: implement xmalloc().
        !           216:        Throughout: use xmalloc instead of malloc.
        !           217:        Apply changes by Sam Devulder.
        !           218: 970608 Fix ncurses.c so that it links. Fix configure so that it can be enabled
        !           219:        Get rid of spixstate by implementing a new sprite drawing method.
        !           220: 970607 do_copper_cheat(): New function, Implement inaccurate, line-based
        !           221:        copper as an option (-C). calc_copcomp_true(): Fix possible hr 4 plane
        !           222:        bug (split the wait into several pieces in this mode).
        !           223:        Allow 0 bytes for fast and bogomem (to override .uaerc)
        !           224: 970606 Fix some bugs in transdisk, and make it compile with SAS C (Joanne Dow)
        !           225:        Stupid compilation fixes: Get rid of "==" in shell scripts, check that
        !           226:        B300 and friends are defined in serial.c before using them.
        !           227:        Implement blitter prefetch. Fix silly typo in main.c that prevented
        !           228:        immediate blits.
        !           229:        init_ham_decoding(): new function, handles case where the first few
        !           230:        pixels of a HAM display are outside the window.
        !           231: 970605 disk.c: Get rid of need_read. Save cylinder/track number for the data
        !           232:        in the MFM buffer for each drive. Force rereads on disk insert and
        !           233:        write operations. In drive_fill_bigbuf(), do nothing if saved data
        !           234:        matches current. Fix disk change logic: give the program time to read
        !           235:        the change bit a couple of times before inserting a new disk.
        !           236:        custom.c: Fix really subtle bug: when deciding a line as border,
        !           237:        plfstrt/plflinelen were not set up properly, which could confuse
        !           238:        post_decide_line(), and eventually there would be lines decided as
        !           239:        playfield with plfstrt == 0, plflinelen == -32 ==> Bang. A related
        !           240:        bug was that setting DDFSTRT to 0 would result in both plfstrt and
        !           241:        plfstop having value 0.
        !           242:        gfxutil.c: Fix bug that could cause color values to go out of bounds.
        !           243:        xwin.c, svga.c: When dithering, color values were not doubled properly.
        !           244: 970604 filesys.c: when deleting an a_ino, remove it from hash table (Mathias
        !           245:        Ortmann). In do_find(), if creating a file failed and we didn't have
        !           246:        an a_ino before, call delete_aino on the one we created.
        !           247:        More configure changes: Add $X_PRE_LIBS and $X_EXTRA_LIBS in some
        !           248:        places. See if MIT-SHM support for X will compile and link. Always
        !           249:        compile it in if it will. Turn DONT_WANT_SHM and LOW_BANDWIDTH into
        !           250:        run-time options (-T, -L). Use single line buffer when dithering and
        !           251:        in low-bandwidth mode (but in all cases there's a full buffer for the
        !           252:        X image).
        !           253:        memory.c: Don't map in custom regs at 0x200000 if chipmem_size > 2MB.
        !           254:        Split up init_sound(): new function setup_sound(). Make "-B" accept
        !           255:        minimum and maximum values as in Gustavo's latest patch. Add code to
        !           256:        tui.c to support changing buffer size, number of bits and frequency.
        !           257: 970603 If using compiler, define NO_EXCEPTION_3 and NO_PREFETCH_BUFFER.
        !           258:        (bugreport by Raphael Poss).
        !           259:        GGI support from Christian Schmitt/Stefan Reinauer. More serial patches
        !           260:        from Stefan. 3 button joystick patch from Mathias.
        !           261:        Add linemem field to gfxvidinfo, for systems that don't want to keep
        !           262:        a buffer of the full screen. Adjust custom.c to use it instead of
        !           263:        bufmem if it's not NULL (mostly by Mathias). svga.c, xwin.c: Use single
        !           264:        line buffer in some cases. xwin.c: If USE_DGA_EXTENSION, undef
        !           265:        LOW_BANDWIDTH. Move test of DONT_WANT_SHM after inclusion of config.h
        !           266:        Make configure.in a bit more structured.
        !           267:        Move colormodes array to main.c.
        !           268: 970602 gfxlib.c: Start implementing layers workalike.
        !           269:        md-i386-gcc/support.c: Don't mix sleep and alarm. (Petter Schau)
        !           270: 970601 Remove may_run_compiled. Print error and abort if m68k_go is called
        !           271:        twice.
        !           272:        Fix yesterdays bugfix in svga.c.
        !           273:        Change FRAME_RATE_HACK so it tries to average the times a bit.
        !           274:        Re-implement a call_m68k function in autoconf.c. The new version
        !           275:        works by doing some heavy setjmp/longjmp magic and switching stacks.
        !           276:        With this, it's even possible to call things that Wait(), so use
        !           277:        the new functionality in gfxlib.c to call LockLayer() in WritePixel().
        !           278: 970531 Turn some inline functions in memory.h into macros. Get rid of
        !           279:        get_along and get_aword. Implement CIA A OVL bit.
        !           280:        main.c: Add missing close_sound() call.
        !           281:        Remove INTEL_FLAG_OPT, use X86_ASSEMBLY for everything.
        !           282:        tui.c/svgancui.c: Implement scrollable menus. Add code to select
        !           283:        video mode from a given list. Add "graphics_setup" call to graphics
        !           284:        specific files. Use it in svga.c to build a video mode menu.
        !           285:        Fix bug in svga.c that would crash non-linear modes.
        !           286:        disk.c needs to include gensound.h and sounddep/sound.h.
        !           287:        Fix bug in fuzzy_memset. In pfield_init_linetoscr, always adjust ddf
        !           288:        values if lores hack is being used.
        !           289: 970530 Apply some patches by Samuel Devulder.
        !           290: 970529 filesys.c: In "Houston, we have a problem" case, do not only delete
        !           291:        the oldest ExKey, but also use it instead of random memory for the
        !           292:        new one. Attempt to reduce the number of lost exkeys by allocating
        !           293:        them in action_examine_next instead of action_examine_object.
        !           294:        Initialize the "uniq" field of unit->rootnode.
        !           295:        Attempt to speed up sprite drawing loop.
        !           296:        Implement sprite<->sprite collisions.
        !           297:        Get rid of -Wtraditional.
        !           298: 970524 New config option NO_PREFETCH_BUFFER. Emulate 16 byte prefetch if
        !           299:        unset.
        !           300:        For exception 3, the stack frame seems to be:
        !           301:        (saved SR).w, (faulting address).l, (faulting instruction).w, 0.w,
        !           302:        (pc at fault time).l. Make everything that could generate this fault
        !           303:        handle it properly. That got Cybernoid II working!
        !           304: 970523 Makefile.in: Make sure $(DEBUGFLAGS) is used for CPU emulation files.
        !           305:        memory.c: Don't map chipmem at 0x200000, use custom chip bank for now.
        !           306: 970507 configure.in: Mutate ac_cv_c_inline=no to ac_cv_c_inline=
        !           307:        Check for tcgetattr, <sys/termios.h>, cfmakeraw, <sys/ioctl.h>
        !           308:        Check for sem_init in -lposix4 (Marc Zimmermann)
        !           309:        serial.c, cia.c, cia.h, serial.c: changes (Stefan Reinauer)
        !           310:        Make disk gap 360 (Michael Krause)
        !           311: 970404 In configure, check for features.h before including it in test program.
        !           312: 970401 New option -4. Bugreport by Mirko Viviani.
        !           313: 970324 Petter Schau says the exception frame for Bus Error and Address Error
        !           314:        is 14 bytes, not 6, on the 68000. Change Exception() accordingly.
        !           315:        It's *** OUCH *** day! Make hardfile beginio return the correct value. 
        !           316:        Also load execbase before calling ReplyMsg().
        !           317:        It's *** OUCH *** day again! The filesystem asm code was an odd number
        !           318:        of bytes, leaving the hardfile code unaligned.
        !           319: 970318 custom.c: Rename spron to sprst, and use new variable spron to indicate
        !           320:        for which sprites the sprite pointer registers have been written. Don't
        !           321:        do sprite DMA if spron==0. (The problem was that some programs write
        !           322:        SPRxCTL and then SPRxPT, which used to set the sprite state to
        !           323:        SPR_restart, but it must be SPR_waiting_start.)
        !           324:        Turn a few m68k_reset()s into uae_reset()s.
        !           325:        New variable nr_joysticks. Delete joystickpresent. New arg "nr" for
        !           326:        getjoystate() and read_joystick().
        !           327:        New extra keycode AK_RCTRL. Handle multiple joysticks & different key
        !           328:        replacements in keybuf.c/custom.c. Make tui.c handle all the settings.
        !           329: 970310 od-linux/joystick.c now compiles again without the joystick header.
        !           330:        Fix some problems in configure; also check whether [n]curses.h defines
        !           331:        attr_t. Keep count of warnings and errors during configuration and
        !           332:        print a hopefully helpful message at the end.
        !           333:        disk.c sanitized by Mathias Ortmann.
        !           334:        Change the replacement getopt around a bit.
        !           335:        execlib.c, timerdev.c: Files removed. This will never be compatible
        !           336:        enough. Also remove related code (several SPCFLAGs, etc.)
        !           337:        Make main.c a bit nicer by introducing additional defines for the t-*
        !           338:        files. Also canonicalize the .uaerc file on all systems, eliminating
        !           339:        lines starting with '#' in the process.
        !           340:        Remove code in autoconf.c to set up 8MB chip; move it to filesys.asm.
        !           341:        Try doing the copper end-of-line hack in copper_memory_cycles().
        !           342:        Disable the copper speedup hack.
        !           343: 970309 Add a small, dumb hash table to struct _unit to speed up repeated
        !           344:        aino lookups. Delete aino in action_delete_object(). Move children in
        !           345:        action_rename_object(). In action_examine_next(), call new function
        !           346:        lookup_child_aino_for_exnext() which behaves more or less like
        !           347:        lookup_child_aino(), except it uses new function get_aname() to
        !           348:        translate filenames.
        !           349:        More portability fixes. New links set up in configure: sounddep,
        !           350:        target.h. All machine-dependent and joystick code removed
        !           351:        from os.c, renamed to audio.c. os.h renamed to audio.h, joystick stuff
        !           352:        moved to joystick.h. New directory targets/.
        !           353:        missing.c: New file.
        !           354: 970308 Rewrite filesystem. Now it keeps one unique "a_inode" structure for
        !           355:        every object in the filesystem. This contains Amiga name, native
        !           356:        name, Amiga permissions, lock counts and several links for building
        !           357:        a tree structure. Keys are now made only for files which need a fd
        !           358:        and a file position.
        !           359:        Treat invalid locking modes als shared.
        !           360:        Comment out the code using readdir_r because that function appears
        !           361:        to be broken in glibc-2.0.1.
        !           362:        Clear RES2 in all packets before handling them.
        !           363:        Give mousehack the same task priority as the input.device.
        !           364:        Fix (?) some silly bugs in disk.c.
        !           365: 970307 Include some Acorn port code from Peter Teichmann.
        !           366:        Yet Another Serial Diff from Stefan.
        !           367: 970306 filesys.c: Don't just crash if lookup_key() has no success.
        !           368:        action_find_write() is supposed to create the file if it does not
        !           369:        exist.
        !           370: 970305 More DOS patches. Make the Makefile parallel-make safe.
        !           371:        svgancui.c: Do cyclic search if character is typed in file selector.
        !           372:        my_alphasort(): New function (work around glibc bug)
        !           373:        custom.c: Fix error in yesterdays change.
        !           374:        filesys.h: New file.
        !           375: 970304 New serial patch from Stefan and Christian.
        !           376:        Remove the assembler code to generate inputevents. Instead, add a new
        !           377:        Amiga C program ("mousehack") which does the trick. New function
        !           378:        mousehack_helper() in custom.c, activated by calltrap 0xF0FF70.
        !           379:        Mouse default is now always normal_mouse, not follow_mouse.
        !           380:        Do disk change checks only every 1/2 second. disk.c: Reverse
        !           381:        nodskchange to dskchange, and make sure there has been a step with the
        !           382:        drive empty before calling drive_insert() from DISK_check_change() (but
        !           383:        this _still_ doesn't work. Arrrghh).
        !           384: 970303 filesys.c: Initialize _all_ of struct _unit in startup_handler().
        !           385:        Remove message_lock and friends and use new variables cmds_sent,
        !           386:        cmds_acked and cmds_completed to determine when to send dummy messages.
        !           387:        Changed examinekey handling so that "dir all" on a large tree doesn't
        !           388:        crash anymore (it used to reuse the top-level exkeys which were still
        !           389:        in use after a wraparound). Don't allocate examinekeys if we're
        !           390:        examining files.
        !           391:        Check for readdir_r in configure. Use a wrapper my_readdir() around
        !           392:        either readdir or readdir_r.
        !           393:        Call filesys_reset() also from filesys_diagentry().
        !           394:        Get rid of most of uaelib.c. Now there isn't a library, just a calltrap
        !           395:        at 0xF0FF60. Changed uae_pragmas.h accordingly and recompiled uaectrl.
        !           396:        Added a bunch of "volatile"s in various places in filesys.c and
        !           397:        penguin.h.
        !           398:        Make the filesystem reset-safe in the threaded case (paranoia changes).
        !           399:        Check disk changes only every 1/10 second in vsync_handler().
        !           400:        Add assembler code to generate InputEvents once every vblank to
        !           401:        filesys.asm. (Doesn't work yet.)
        !           402: 970302 filesys.c: Get rid of DosPacket type, pass pointers into Amiga memory
        !           403:        instead. Use GET_PCK_* and PUT_PCK_* macros to access fields.
        !           404:        Start making filesystem code thread safe & multithreaded. Move static
        !           405:        variables from some functions into struct _unit, likewise the key
        !           406:        and exkey variables. Add code to filesys.asm to queue commands which
        !           407:        did not complete and to reply to them later. Do multithreading if
        !           408:        compiled with SUPPORT_THREADS. See source for details.
        !           409:        Change hardfile and uaelib code not to call 68k routines. Make emulator
        !           410:        traps more efficient (no need to save regs, don't need SPCFLAG_EMULTRAP
        !           411:        anymore).
        !           412: 970301 Fixed ncurses.c so it works with currprefs.
        !           413:        filesys_reset(): New function, called from customreset().
        !           414:        Changed source layout. Break up machdep into machdep, osdep and
        !           415:        threaddep.
        !           416:        penguin.h: Now supports arbitrary types in pipes.
        !           417:        Changed filesys.c so that no 68k code is called from native C code.
        !           418:          filesys.asm, filesys.sh: New files. filesys.asm contains some
        !           419:         startup code and the main loop for the filesystem. Several locks
        !           420:         are allocated by the assembler code for use by the C code.
        !           421:         DOS_AllocMem(), DOS_FreeMem(), make_dev(), build_parmpacket(),
        !           422:         filesys_init(): Functions removed.
        !           423: 970228 ncurses.c: Fix problem in b/w modes.
        !           424:        finish_drawing_frame(): Don't need to draw lines in multipenguin mode.
        !           425:        Move all options into new struct uae_prefs. New variables currprefs,
        !           426:        changed_prefs.
        !           427:        svgancui.c, tui_filereq(): If user presses a key, start search at
        !           428:        current position, not at the beginning.
        !           429:        DISK_check_change(): new function, compares currprefs with
        !           430:        changed_prefs and handles diskchanges if they differ. Called from
        !           431:        vsync_handler().
        !           432:        DISK_validate_filename(): new function, checks whether a file exists.
        !           433:        Modify svga.c and tui.c so that F11 now brings up a limited UI.
        !           434: 970227 Reimplement Mathias Ortmann's frame rate hack: New option HAVE_RDTSC
        !           435:        in config.h. New typedef frame_time_t. New variables vsynctime and
        !           436:        vsyncmintime. New function machdep_init() in md-*/support.c which
        !           437:        sets up vsynctime in the md-i386-gcc and md-i386-linux cases. Use
        !           438:        all of those in events.h, do_cycles() to determine whether to spend
        !           439:        more time on the CPU emulation.
        !           440:        Add missing "int i;" to xwin.c, init_colors().
        !           441:        Unroll some loops in sample_*handler().
        !           442:        Improve quitting/resetting behaviour: Move quit_program from GUI
        !           443:        specific files to main.c. New functions uae_reset() and uae_quit(),
        !           444:        quit_program is now handled in vsync_handler() and m68k_go().
        !           445:        New variable sample_evtime, used in customreset() and set up in
        !           446:        init_sound().
        !           447: 970226 Solaris sound update from Manfred Thole.
        !           448:        keybuf.c: Decided that it doesn't need locking after all.
        !           449:        Add a bunch of Gustavo's latest diffs. (I hate CR characters!)
        !           450:        Move gfxvidinfo from all the target-specific files to custom.c
        !           451:        More Win32 stuff: New function notice_new_xcolors().
        !           452: 970225 configure.in: In shell string comparisons, always put spaces around =
        !           453:        Add warning about standard SVGAlib being incompatible with
        !           454:        LinuxThreads.
        !           455:        os.c, SOLARIS_SOUND: Remove reference to obsolete variable smplcnt.
        !           456:        Two new variables sound8_base and sound16_base.
        !           457:        sysdeps.h: Split off some stuff into md-amiga/amiga-kludges.h. Also
        !           458:        make the main types now uae_u32 and variants instead of ULONG, etc.
        !           459:        Convert some files to use the new types.
        !           460:        cia.c: use int/long instead of ULONG/UWORD/etc for most static vars.
        !           461:        xdebug.c: Some cleanup.
        !           462:        md-i386-linux/penguin.h: Fix bug. Add "chunks" parameter to
        !           463:        init_comm_pipe. Use it to decide in write routine whether to signal
        !           464:        the other thread if it's waiting. Add no_buffer parameter to
        !           465:        write_comm_pipe_int.
        !           466:        Start adding Mathias Ortmann's Win32 patches.
        !           467:          md-i386-win32: New directory
        !           468:         include/sysconfig.h.win32: New file
        !           469:         custom.c: is_lastline: new variable, set in hsync_handler().
        !           470:         sysdeps.h: Make it possible for system specific code to say that
        !           471:           POSIX/ANSI stdio/malloc functions aren't present and are defined
        !           472:           by a system-specific file instead. In filesys.c, don't reference
        !           473:           errno if DONT_HAVE_POSIX is defined.
        !           474:         readcpu.h: Kludge around the fact that Messysoft C apparently doesn't
        !           475:           handle enums properly.
        !           476:         newcpu.c/sysdeps.h: Preprocessor magic for 64 bit constants to get
        !           477:           around a problem with MS C.
        !           478:         md-*/penguin.h: Add dummy definitions for semaphores.
        !           479:         keybuf.c: Use semaphores to synchronize access.
        !           480:         os.c, win32 sound stuff: add memcpy avoidance code.
        !           481: 970221 X86.S: Don't use .type ifdef USE_UNDERSCORE.
        !           482:        Move thread support configuration to configure from config.h. Change
        !           483:        pipe implementation to use POSIX semaphore functions.
        !           484: 970217 From Samuel Devulder: Fix off-by-one error in LINE_TO_SCR, EHB case.
        !           485: 970215 customreset(): Initialize current_colors.color_regs to invalid values
        !           486:        [reported by Samuel Devulder]. config.h: Uncomment ECS_DENISE [reported
        !           487:        by Mathias Ortmann]. Add ami-win.c and custom.c patch from Samuel
        !           488:        Devulder.
        !           489: 970213 disk.c: Set FLOPPY_SPEED to 3
        !           490:        custom.c: SPRxCTL_1(): if spron was SPR_restart set it to
        !           491:        SPR_waiting_start. do_sprites(): don't do that.
        !           492:        SPRxPT{H,L}: Call decide_line().
        !           493: 970210 custom.c, X86.S: Do LORES_HACK for dual playfield screens. Plug memory
        !           494:        leak in custom_reset(). Fix pfield_draw_line() not to do so many
        !           495:        multiplications. Make line_changed a char array. In remember_ctable(),
        !           496:        remember comparison results also if negative.
        !           497:        In 8 and 16 bit graphics modes, use the high parts of the color values
        !           498:        to duplicate the actual values. This saves a lot of code in the
        !           499:        hdoubling routines. Add a special fill_line in the X86_ASSEMBLY case.
        !           500:        Check for glibc-2 and libpthread in configure.
        !           501:        options.h: In fuzzy_memset(), do rounding slightly more conservative.
        !           502:        Add tons of new keycodes to svga.c
        !           503:        In record_color_change(), also check if vertically outside DIW.
        !           504:        New variable hdiwstate, to represent the horizontal DIW state machine
        !           505:        (thanks Petter!) Handled in reset_decisions(), finish_decisions() and
        !           506:        decide_diw().
        !           507: 970209 Change ncurses.c not to do so many multiplications/divisions. Add bold
        !           508:        character support. Add color support. Make configure recognize
        !           509:        --without-svgalib and enable ncurses.c in that case.
        !           510: 970208 X86.S, custom.c: Do LORES_HACK for EHB screens also.
        !           511:        New function write_log(); convert some fprintf(stderr, ...) calls.
        !           512: 970207 Argh, Linux/i386/ELF check in configure got broken by latest sh-utils.
        !           513:        In disk.c, drive_reset(), clear nodskchange flag.
        !           514: 970206 Sprite fix from Mathias Ortmann: _really_ get rid of SPR_vtop, and
        !           515:        replace all uses by SPR_stop.
1.1.1.2   root      516: 970204 Disk change fix from Mathias Ortmann.
                    517: 970203 New file ncurses.c, from Samuel Devulder.
                    518:        Disabled Exec stuff completely. Now it isn't even compiled in anymore.
                    519: 970129 Changes from Samuel Devulder for Amiga and Christian Bauer for BeBox.
                    520: 970122 Reorder the set_planes_unaligned functions so that better code is
                    521:        generated. Remove the or_planes_unaligned functions, add a parameter
                    522:        do_or to the set_planes functions and rely on the compiler to optimize
                    523:        the additional if away.
                    524:        post_decide_line() can now turn DMA off early.
                    525:        Make struct flag_struct 64 bits large for i386 systems, so we can use
                    526:        long move insns without clobbering the X flag.
                    527:        "make clean" should delete machdep/support.o.
                    528:        Maybe we do need SPR_vtop after all - undo 970120 change. Instead,
                    529:        comment out the code that clears the pos and ctl words at vsync time.
                    530:        In xwin.c, restrict window height to 300 or 600 depending on linedbl
                    531:        mode.
                    532: 970121 Get rid of SPCFLAG_BLIT, call do_blitter() directly.
                    533:        In finish_decisions(), don't check for sprite draws in the border.
                    534:        Reimplement the SMART_UPDATE define so that the memcmpy() overhead can
                    535:        be avoided by not setting it (but this usually loses).
                    536:        Detect 7 lores planes in BPLCON0() and change to 6 to avoid memory
                    537:        clobbers.
                    538:        In pfield_draw_line(), do memcpy linedoubling after handling the line
                    539:        to be doubled, not when we see a LINE_AS_PREVIOUS. Get rid of prev_ypos
                    540:        parameter.
                    541:        Add special code for bpldelay1 == bpldelay2 to the UNALIGNED_PROFITABLE
                    542:        code.
                    543:        In pfield_init_linetoscr(), when delays 1 and 2 differ, clear the areas
                    544:        which are covered only by half of the planes at the start and end of
                    545:        the DDF area.
                    546:        Replace FASTER_CPU and LINUX_SOUND_SLOW_MACHINE with M68K_SPEED.
                    547:        Replace fast_unaligned_memset macros with fuzzy_memset; use this also
                    548:        in pfield_init_linetoscr().
                    549:        Use real (PAL) vblank end in do_sprites().
                    550: 970120 Mask out high byte in DDF{STRT,STOP}
                    551:        Oops, forgot about long frames while calculating the bounds of all the 
                    552:        maxvpos*2-sized arrays. Increment them all by two.
                    553:        Try to detect empty sprites before recording them for the drawing code,
                    554:        but be careful not to hurt attached sprites.
                    555:        From Paul Kienitz: hack for keyboard code in cia.c to work better with
                    556:        newer Kickstarts; small cleanup in cpuopti.c.
                    557:        Undo yesterday's change in copper_memory_cycles(), instead call
                    558:        decide_line() each iteration and test whether copper_cycle_time == -1
                    559:        to determine whether we're vertically in the DIW.
                    560:        calculate_copper_cycle_time() could get astronomical plane counts due
                    561:        to missing masking operation.
                    562:        No need to put the copper to sleep for one memory cycle after a WAIT,
                    563:        merge that with the two cycles for the next instruction.
                    564:        Get rid of SPR_vtop, no longer needed (it was unnatural anyway)
                    565: 970119 Implemented alternate pfield_doline functions that access the apixels
                    566:        array at unaligned addresses; conditional on UNALIGNED_PROFITABLE.
                    567:        Move LORES_HACK from config.h to custom.c.
                    568:        Fix loop exit test for pfield_linetoscr_hdouble_asm8(). Clear the
                    569:        right parts of apixels in pfield_init_linetoscr() when LORES_HACK is
                    570:        enabled.
                    571:        Get rid of bordercol array: the decision-making code should handle
                    572:        lines in the border correctly by setting line_changed.
                    573:        Disable bitplane DMA if more than 4 hires bitplanes are set in BPLCON0.
                    574:        Fix some more copper timing inaccuracies: used to test
                    575:        thisline_decision.which without calling decide_line() before.
                    576: 970118 Now remember BPLCON1 changes during data fetch. New functions
                    577:        decide_delay(), pfield_adjust_delay(); new fields in struct drawinfo
                    578:        and new variables next_delay_change, delay_changes.
                    579:        Forgot to set copstate to COP_morewait in one case in do_copper().
                    580:        Add a kludge that checks for copper position 0xDC and changes it to 
                    581:        0xDE.
                    582:        Restore equality test for vertical sprite positions. Disable sprite
                    583:        DMA before the end of the vertical blank area.
                    584:        Mask not only the positions, but also comparison values in 
                    585:        calc_copcomp_true().
                    586: 970117 Change custom_bput to put the byte value in each half of the word that
                    587:        is put.
                    588: 970116 disk.c: Set the gap between sector 11 and 0 to 380 words. Andromeda's
                    589:        "Seven Seas" demo expects at least 320. Make the KILLEHB bit
                    590:        conditional on AGA_CHIPSET. Decide BPLCON1 (scrolling delays) in
                    591:        decide_diw(). Fix off-by-one errors in the very_broken_program case
                    592:        of decide_plane(). Fix aspect calculation when horizontal resolutions
                    593:        differ. Simplify copper horizontal comparison logic slightly.
                    594:        Fix divide by zero errors when reading from empty floppies.
                    595: 970115 pfield_draw_line(): Check for frame_redraw_necessary also for border
                    596:        lines. New global variable pfield_last_drawn_line, used to implement
                    597:        clever Y centering. Make the KILLEHB test conditional on ECS_AGNUS.
                    598:        Move joystick and sound initialization before gui startup. Adapt tui.c
                    599:        so it understands the gfx_requested_* options.
                    600:        We can't always avoid clearing pixdata.apixels if DIW is larger than
                    601:        DDF. There may be sprites in the area outside DDF, but inside the DIW,
                    602:        so we can't just adjust linetoscr_diw_{start,end} blindly in
                    603:        pfield_init_linetoscr() [Sanity Arte demo].
                    604: 970114 Bebox diffs from Christian Bauer.
                    605:        Support for Symmetric Multi Penguins under Linux (start second thread
                    606:        with clone and use some pipes for communication - rather simpleminded
                    607:        for now). Rework do_flush_line() and do_flush_screen() so that X
                    608:        libraries are only called from the first thread.
                    609:        Eliminate valid field from struct decision (set which to -2 instead of
                    610:        valid = 0)
                    611:        Fix definition for gui_filename() in xdebug.c to use const char * for
                    612:        the second parameter. Remove CRs from disk.c (now how did they get in
                    613:        there??). Likewise for cpuopti.c.
                    614: 970113 When post_decide_line() detects mid-line DDF change, must adjust
                    615:        bitplane pointers in decide_plane() also. Make plfstrt the earlier a
                    616:        position in this case, and adjust bitplane pointers/modulos
                    617:        accordingly.
                    618:        current_hpos() isn't what GCC calls a constant function.
                    619:        Replace line_drawinfo[current_change_set] with new variable
                    620:        curr_drawinfo throughout. Likewise for new variables prev_drawinfo,
                    621:        {curr,prev}_{color_{change,table}s,sprite_positions}; all initialized 
                    622:        in init_drawing_frame (). Turn line_data into a 2dim. array, do some
                    623:        CSE/SR on line_data[lineno] in pfield_doline_{l,h}() and
                    624:        decide_plane(). Make some custom registers unsigned int instead of
                    625:        UWORD.
                    626:        calc_copcomp_true(): Masking with 0xF000 isn't likely to give 0xC0000.
                    627:        Comment out bplcon{3,4} from struct decision.
                    628: 970112 filesys.c, action_set_protect(): if __unix is defined, check whether
                    629:        user wants to change directory protection, and if so, force rwx to be
                    630:        the new protection.
                    631:        Force sane window sizes in xwin.c, and set width/height from mode specs
                    632:        in svga.c.
                    633:        Implement LORES_HACK compile-time option.
                    634:        Whee, there _are_ broken programs that enable bitplane DMA in the
                    635:        middle of the data fetch area, draw half a picture on the first line
                    636:        and the full picture on all subsequent lines. New function
                    637:        post_decide_line(), called from BPLCON0() and DMACON() which tries to
                    638:        detect this case and turns the decision from border to playfield if
                    639:        necessary.
                    640:        New arrays, sprvstart and sprvstop, calculated in SPRxPOS and SPRxCTL,
                    641:        and used by do_sprites. Handle sprites with vstart == vstop correctly.
                    642:        Sprite vpos comparisons are now done with <= , not ==.
                    643:        _Now_ the X86-specific linetoscr routines do their masking correctly.
                    644:        configure: If we have GCC, #define GCCCONSTFUNC as __attribute__ 
                    645:        ((const))
                    646: 970111 Eliminate can_draw from struct decision.
                    647:        Change vertical DIW code; use enum of two states (waiting for start and
                    648:        for stop) which is updated in decide_line(). Lines during the wait for
                    649:        stop phase are decided as border. No data is fetched outside the
                    650:        vertical DIW.
                    651:        Optimize copper memory cycle calculation a little: the new variable
                    652:        copper_cycle_time is set to a time >= 0 if this time is constant
                    653:        (i.e. 2). Calculated whenever DMA, BPLCON0, or DIW state changes.
                    654:        blitter.c, genblitter.c, custom.c: For BLTBDAT, follow the HRM which
                    655:        says that immediate data is shifted as soon as it is written. This 
                    656:        seems to be wrong for the A channel, though. Also, don't change prevb
                    657:        during blitter loops if DMA B is off (Sanity Arte). Turn FAST_BLITTER
                    658:        stuff into run-time option "-n", and eliminate the ancient blitting
                    659:        code (former FAST_BLITTER == 0).
                    660:        Big OOPS: Copper move checked copcon, but there was no COPCON function
                    661:        so it was always zero -> copper-controlled blitter never worked. As a
                    662:        speed optimization, recognize 00010000 copper commands and put the
                    663:        copper to sleep until the blit is finished.
                    664:        Slow down the blitter in decide_line() if bitplane DMA steals blitter/
                    665:        copper/CPU cycles (but this code is commented out for now).
                    666: 970110 New function pfield_init_linetoscr(). Move calculation of diw_end and
                    667:        diw_start values there from x86-specific code. Also use DDF values to
                    668:        determine where data in pixdata.apixels is valid. The x86-specific
                    669:        linetoscr routines now only do the alignment masking (and they now do
                    670:        it correctly, too).
                    671:        Got rid of the display-related contents of struct draw_info, make 
                    672:        static variables and them initialize from pfield_draw_line() only when 
                    673:        necessary in function pfield_expand_dp_bplcon (). Get rid of the four
                    674:        saved_* arrays, remember line_drawinfo instead now that it's smaller.       
                    675:        Color compare avoidance didn't work because color_src_match didn't get
                    676:        set when a compare was successful. Fixed.
                    677:        Reorder parts of the decision-making code so that less redundant checks
                    678:        are performed.
                    679:        Add clever x centering. Change "-O" option parameter format.
                    680:        In md-i386-linux/support.c, memcmpy(), do memcmp() first, then switch
                    681:        to a plain memcpy routine once we find the first difference.
                    682:        Patches from Stefan Reinauer/Christian Schmitt for 
                    683:        xdebug.c/configure.in
                    684:        New uae-ui.tk from Volker Zota.
                    685:        Add --enable-profiling to configure.in.
                    686:        Check for period==0 in AUDxPER() was accidentally commented out -
                    687:        revert.
                    688: 970109 Some tweaks in X86.S, plus a new option X86_PPRO_OPT to use movzbl
                    689:        instead of xorl/movb when possible.
                    690:        Fix maxline calculation in finish_drawing_frame().
                    691:        Make next_lineno always a valid array index. Decrease minfirstline to
                    692:        21 (NTSC vblank end). New option -O to specify display parameters,
                    693:        supersedes -d and -C which will be removed eventually. Differentiate
                    694:        between line doubling and aspect correction. Put aspect correction code
                    695:        in custom.c.
                    696:        Remove some more unused stuff from custom.c.
                    697:        Fix Makefile problems when cpuopti isn't used.
                    698: 970108 use_lores isn't initialized before custom_init(), setup lores_frame and
                    699:        friends in customreset(). Optimize spixstate clearing. Optimize border
                    700:        drawing with copper color changes (use fill_line instead of linetoscr)
                    701:        Fix the sprite DMA logic, it now works with the Interference demo.
                    702:        Silly bug: Sprite 7 never got displayed (introduced sometime last week)
                    703:        Implement USER_PROGRAMS_BEHAVE for normal (non-compiler) CPU emulation.
                    704:        Add most of Gustavo's last patch (new dir md-i386-dos)
                    705: 970107 Finish rewriting most of the drawing code in custom.c/X86.S. See
                    706:        README.PROGRAMMERS and source for details. Improvements include better
                    707:        copper precision, better interlace support, hopefully much less bugs
                    708:        and not too many new ones.
                    709:        Remove calc_adjustment() and gfxvidinfo.x_adjust; replace with code in
                    710:        custom.c that looks at gfxvidinfo.follow_diwstop (new field)
                    711:        Some operations (like EHB playfields) now have a higher penalty than
                    712:        before, the assumption being that these are rare and the normal case
                    713:        runs faster if it doesn't have to care too much about exceptions.
                    714:        Fix sign mismatch bug in BPL1MOD() and BPL2MOD().
                    715: 970104 Move x_ShowEA() from xdebug.c in newcpu.c, rename as ShowEA().
                    716: 970102 Add check for XK_W to decode_es() in xwin.c.
                    717:        New uae-ui.tk from Volker Zota.
                    718: 961218 Add a patch by Bernie Solomon to overcome a problem with K&R compilers. 
                    719:        Change X86.S in a way that may make it either faster or slower.
                    720:        Fix ADF_EXT1.
                    721: 961217 #ifdef XK_dead_*, since Solaris doesn't have it.
                    722: 961202 Fixed joystick.h problem in configure. X11/DGA support from Kai
                    723:        Kollmorgen.
                    724: 961117 Some fixes in svgancui.c.
                    725: 961112 Gustavo's latest patch, with GUS support for DOS by Michael Sontheimer.
                    726:        Add menu browsing to tui.c/svgancui.c
                    727: 961107 Fix m68k_areg and m68k_dreg macros (put args in parentheses)
                    728:        Initialize gfxvidinfo.maxlinetoscr in xwin.c (found by Sam Devulder)
                    729: 961104 svgancui.c tui_refresh() routine tweaked to work with the non-standard
                    730:        ncurses that came with my new SuSE-Linux CD.
                    731: 961031 Added Christian Schmitt's xdebug.c (disabled for this release)
                    732: 961024 Module search function from Michael Krause.
                    733: 961019 Add Samuel Devulder's Amiga patches.
                    734: 961018 Don't use -lm on NeXTStep.
                    735: 961017 configure checks for -lforms.
                    736: 961016 Ian Stephenson's new NeXTwin.m. Make CLXDAT return 0 (no collisions)
                    737:        transdisk and transrom patches from Rodney Hester. Don't warn about
                    738:        256K Kickstarts. Check diskfile sizes and try to determine whether it's
                    739:        a normal or a HD diskfile. No HD drive identification yet (is it
                    740:        different?)
                    741: 961013 Corrected lots of bugs in the return address stack. Fix some errors in
                    742:        the compiler memory management. Use the "locked" flag for hash entries
                    743:        differently, and don't enforce it on a cache flush. Compile JMP to a
                    744:        constant address.
                    745: 961012 Remove regs from cache before sign extending them for ADDA, MOVEA and
                    746:        similar things. Compile NOPs to nothing. Implement variable shifts/
                    747:        rotates. Most shifts clear the V flag: Handle this in the compiler by
                    748:        some extra magic sequences after shifts or in the CC_AFTER_RO case in
                    749:        flush_cc_cache().
                    750:        Implement a stack containing return addresses and their hash entries,
                    751:        and compile JSR (to constant addresses)/BSR/RTS instructions inline.
                    752: 961011 compiler.c: Several bugs introduced in the last two days fixed.
                    753:        New function compile_note_modify() which is used when an ea is
                    754:        modified. This replaces calls to storeea with eaino_s == eaino_d.
                    755:        Bah, the 386 doesn't set flags on a NOT. Use CC_TEST_REG again.
                    756:        Do flag life analysis until no more changes are found.
                    757:        Make sure all "data" parts of EAs are locked exclusively (esp. for 
                    758:        LEA, PEA)
                    759: 961010 compiler.c: Finish yesterdays changes. Get rid of reg/offset
                    760:        bookkeeping in m68k_compile_block(), keep the information in struct
                    761:        ea_info instead. Rewrite all instructions to use the new fetch and
                    762:        store routines.
                    763:        Don't force a byteorder conversion in sync_reg() if the reg doesn't need
                    764:        to be written back. Don't do a SAHF or equivalent in compile_condbranch
                    765:        if we are doing a BRA or BF.
                    766:        Don't use CC_TEST_REG for NEG and NOT (this was inefficient for NOT and
                    767:        wrong for NEG).
                    768: 961009 compiler.c: Two types of locks for registers; shared and exclusive.
                    769:        Implement better EA fetching and storing routines. For pdi addressing
                    770:        modes, lock the necessary registers shared before calling prepareea.
                    771:        Unlock those registers again in generate_possible_exit().
                    772: 961008 MOVEM instructions in the compiler to an absolute address could have
                    773:        caused random memory corruption.
                    774: 961007 Spanish keyboard support for X by Ruben Martinez. Some configure fixes
                    775:        (don't assume current dir is in path, don't assume ncurses.h exists if
                    776:        libncurses does).
                    777:        If we're calling 68k code from an emulator function, don't use compiled
                    778:        code.
                    779:        Add a compilation option USER_PROGRAMS_BEHAVE (for compiler). Make
                    780:        compiler hash table larger.
                    781: 961004 Fix typo in configure.in.
                    782:        Modify the compiler so it doesn't ever throw away code, unless it's
                    783:        necessary for a cache flush.
                    784:        Didn't quite squash the compiler flag bug yesterday. Try again.
                    785:        In filesys.c, do_find(), add a parameter fallback that indicates that
                    786:        the open should be retried with O_RDONLY if it failed with O_RDWR.
                    787:        Use this in find_input(). This fixes problems with r/o files on r/w
                    788:        volumes.
                    789: 961003 Play around with the x86 specific memory access functions. Problem:
                    790:        __attribute__((regparam)) isn't reliable, and it seems that that's a
                    791:        rather hard to fix bug in GCC. Add some alternatives to
                    792:        md-i386-linux/memory.h; these aren't used yet though.
                    793:        Huh, the compiler was completely wrong about the flag needs of
                    794:        conditional instructions. I wonder how it could work at all...
                    795:        Remove a and d arrays from struct regstruct; replace with array regs
                    796:        with 16 elements. This allows better code for get_disp_ea()
                    797: 961002 BeBox diffs from Christian Bauer. Move all code from main() to new
                    798:        function real_main().
                    799: 961001 Use do_get_mem_*() functions for nextiword(), nextilong() and new
                    800:        function nextibyte().
                    801: 960930 genblitter optimizations: Don't generate expressions in genblitter, use
                    802:        a second program called gengenblitter to generate a table of
                    803:        expressions for genblitter to use. gengenblitter knows about a lot of
                    804:        rules to simplify expressions. Some other tweaks to make GCC output
                    805:        better assembly for blitfunc.c.
                    806: 960927 Turn off LED() in xwin.c (may fix SunOS joystick emulation problems)
                    807:        Call XFlush() and XSync() when exiting.
                    808:        Add some DOS patches.
                    809:        Add hooks in filesys.c to support filesystem units in a GUI. Use them
                    810:        in tui.c. Bugfixes in svgancui.c. Change the way tui_refresh() works.
                    811:        New option "-A".
                    812: 960925 For read-only filesystems (CD...), use O_RDONLY in action_find_input.
                    813: 960923 Save options file even if rename fails.
                    814:        Mirror address space every 16MB like we used to (got turned off by
                    815:        accident).
                    816:        Move some checks out of options.h into configure. Set up md-*
                    817:        directories with machine-dependent stuff in them; configure creates
                    818:        a link called machdep to the appropriate one.
                    819: 960918 Solaris sound update from Manfred Thole.
                    820: 960917 Add a timer.device replacement (buggy).
                    821: 960916 Tristan Cavelier's french keyboard patch.
                    822:        Apply ReadTheSpecsBeforeYouWrite(tm) technology to AddPort().
                    823:        Implement RemPort(), GetCC().
                    824: 960915 Tighten up some of the return values in Exec.
                    825: 960914 EXEC: Implement RemTask(), ObtainSemaphoreList() and
                    826:        ReleaseSemaphoreList(). Set up more fields of the ExecBase properly
                    827:        and Enqueue it in the library list. Make InitResident call AddLibrary
                    828:        when appropriate (otherwise, intuition is never put into the library
                    829:        list). Fix major braino in Obtain/ReleaseSemaphore() (tried to Enqueue
                    830:        tasks into two lists). Make OpenLibrary() honor the version parameter.
                    831: 960913 EXEC: Idle task now runs in user mode (abuse 0xFF0D as second STOP
                    832:        instruction). Make EXEC_IntTrap perform the RTE before trying to
                    833:        schedule.
                    834: 960912 Solaris sound didn't compile. So much for making blind changes without
                    835:        being able to test them.
                    836: 960911 As soon as you compile it under Solaris, it bombs... Now check that the
                    837:        hardfile exists before determining the size in main.c.
                    838:        execlib fixes: Set master DMA enable bit at startup, fix interrupt
                    839:        code to survive an interrupt that occurs before the proper SetIntVector
                    840:        call. Implement RawDoFmt(), CloseDevice(), CloseLibrary(),
                    841:        FindResident().
1.1       root      842: 960905 Lower 8 bits of POTGOR should be 0.
                    843: 960904 Implement MULU, MULS, ADDX and SUBX in compiler.
                    844: 960903 ncurses interface works. DOS people: please use the new portable tui.c
                    845:        and turn dos-ui.c into a frontend for conio.h, similar to svgancui.c
                    846:        In xui.c, unlink the temporary files ASAP.
                    847:        Two new command-line options, "-o" and "-G".
                    848:        Get rid of sz_unknown.
                    849:        compiler.h, compiler.c: New files.
                    850:        filesys.c, action_read(): Added a hack to test whether the first four
                    851:        bytes read from a file a file contain 0x000003F3 (HUNK_HEADER), in
                    852:        which case there's probably a LoadSeg() call going on and we ought to
                    853:        flush the cache. Also, use the fact that memory is now linear.
                    854: 960902 Struggle a bit with dos-ui.c, rename it to tui.c (text-based ui), add
                    855:        new file svgancui.c (ncurses ui) and try to make these two understand
                    856:        each other.
                    857: 960829 New transdisk from Stefan Ropke (can write images)
                    858:        Apply Patrick Hanevolds BeBox patches.
                    859: 96082x many changes to Exec. Still not booting (DMACON seems not to be set up
                    860:        properly, and sometimes Signal is called with task==0).
                    861: 960821 Solaris sound from Manfred Thole.
                    862: 960820 In filesys.c, get_fileinfo(), return volume name for top-level 
                    863:        directory.
                    864: 960819 Reimplement WritePixel(). Layer locking is still missing.
                    865: 960818 Reverse meaning of -x flag for X11. More Exec code.
                    866:        Several changes to the main CPU loop and the way call traps work.
                    867: 960817 Implement some simple Exec functions (Lists, AllocMem & friends)
                    868: 960816 filesys.c fixes from David Varley.
                    869: 960815 Implement graphics library region functions.
                    870: 960814 Modify memory layout: All memory is linear now; get_real_address()
                    871:        returns a (UBYTE *).
                    872: 960810 Add Gustavo's latest DOS code.
                    873:        In configure, include <sys/param.h> earlier for the sake of some *BSD.
                    874:        Add three options for sound code (bits, frequency, buffer size).
                    875:        Make -S option take a parameter; remove SPEED_UP_SIDMON_MODULES.
                    876:        New version of transrom (return 0, duh).
                    877: 960805 Fixed the blitter line mode emulation (this time, really! I hope.)
                    878:        Thanks to Fabio Ciucci for a detailed description of the correct
                    879:        behaviour.
                    880: 960730 Spent the last three days trying to make the filesystem emulation
                    881:        bootable by adding an expansion device for it. A fair bit of trial and
                    882:        error, but now it works! Even the hardfile now works with Kick 1.3, but
                    883:        you have to mount it manually after booting.
                    884:        hardfile.c, filesys.c, expansion.c and autoconf.c now are an
                    885:        entangled mess which badly needs to be cleaned up.
                    886:        We can now mount more than one filesystem even with Kick 1.3: we
                    887:        recognize them by param 2 of the startup packet, which has the same
                    888:        value as the dn_Startup field of the devicenode.
                    889:        Make Kickstart replacement work again (didn't like the CPU emulator
                    890:        changes)
                    891:        Include preliminary serial support from Stefan Reinauer.
                    892: 960728 Fixed the scancodes for <> and #' keys in svga.c
                    893:        Apparently, CIA timer latches ought to be 0xFFFF after a reset
                    894:        (prevents Eagleplayer from crashing occasionally). #ifdeffed out a
1.1.1.2   root      895:        slightly weird call to INTREQ from RethinkICRB()
1.1       root      896: 960727 Sound code rewritten again. Now uses events (five new ones). All 
                    897:        modules seem to play correctly now. It's a bit slower, though.
                    898:        AF sound should work again, patches from Marcus Sundberg.
                    899: 960723 ... and more CPU emulation bugs fixed by Herman.
                    900:        Added uae.library from Tauno Taipaleenmaki, plus support files in the
                    901:        "amiga" directory.
                    902:        Added some files for the XFree86/OS2 port from Krister Bergman.
                    903:        Reorganized all the directories.
                    904: 960717 Add latest patch from Herman.
                    905:        More FAST_BLITTER options: 4 and 5. These finish the blit immediately.
                    906: 960713 Call XAutoRepeatXXX() only if we know we want to change the state.
                    907: 960709 Work around a few AIX bugs (ughhh... what a broken OS). New test for
                    908:        sys/select.h.
                    909:        In xui.c, don't try to communicate with a dead process.
                    910: 960704 More bugfixes from Herman.
                    911:        Get rid of dummy_xlate(), use default_xlate() to abort with error.
                    912:        In svga.c, set correct_aspect to 0 if using lores.
                    913:        Change the Bcc exception 3 handling code a bit.
                    914:        Add new program "cpuopti" that takes gcc-generated assembly files and
                    915:        optimizes away unnecessary pushl/popl instructions on the x86.
                    916: 960701 68020 exception handling & various bugfixes from Herman ten Brugge.
                    917:        In ersatz.c, use 0xFF0D instead of 0xF00D as trap opcode.
                    918:        Do MMU "emulation" (no-op MMU instructions, probably many cases decoded
                    919:        incorrectly).
                    920: 960627 New file LIESMICH, with minimal German instructions. Somebody want to
                    921:        translate the docs to French, Spanish, Italian, all the Scandinavian
                    922:        languages, Hebrew, Polish, Dutch, Russian, Welsh, Gaelic and something
                    923:        [email protected] understands? (No offense meant - I'm sure there are
                    924:        brillant minds using AOL. There must be.)
                    925: 960626 In zfile.c, make sure we delete the temporary file even if file
                    926:        decompression fails.
                    927:        Reimplemented sound support, this time following the HRM closely. Seems
                    928:        to work fine most of the time, but occasionally loses drum beats in
                    929:        TFMX modules. do_sound() is now unconditional, it calls
                    930:        flush_sound_buffer(), which is system-dependent. This breaks everything
                    931:        except Linux sound for now :(
                    932:        Add Herman ten Brugge's 68020/68881 emulation code.
                    933:        FreeBSD chokes if sys/mount.h is included without sys/types.h. Fix the
                    934:        configure script.
                    935: 960622 More work on AGA code.
                    936:        Removed 2-bit display routines from custom.c. Ian: please use dithering
                    937:        code.
                    938:        Make slowmem and chipmem sizes run-time configurable.
                    939:        Mask out the high bits in BLT?PTH().
                    940:        Add exception 3 checks to Bcc/BSR instructions. Seems to be not quite
                    941:        correct yet.
                    942:        Completely rewrote the disk emulation. We now create a big track buffer
                    943:        full of MFM data, used in a cyclic fashion by DSKBYTR() and friends. 
                    944:        The current position in the buffer is determined by the cycle counter.
                    945:        DMA reads are now done in one piece. Two new event types: ev_diskblk 
                    946:        and ev_diskindex.
                    947:        ARRGGHH! The RSI demo has _completely_ broken disk loading routines.
                    948:        They start disk DMA on a block while the blitter is still working on 
                    949:        the same memory region.
                    950:        In zfile.c, unlink() the temporary file directly after opening it so
                    951:        it will be freed by the OS as soon as we close it (or crash... :)
                    952: 960621 Merge pfield_sprite_h() and pfield_sprite_l() into one function.
                    953:        Split off new files blitter.c and blitter.h from custom.c.
                    954:        Fixed null-pointer dereference in zfile.c when filename has no
                    955:        extension.
                    956:        Add new field maxline to gfxvidinfo that prevents writing past the
                    957:        screen buffer if it has less lines than expected.
                    958:        Modify the 256 color allocation routine to set nb_cols to the same
                    959:        value first for r,g and b; allocate everything as usual and fill the
                    960:        lost colors afterwards with those that have the greatest difference.
                    961: 960620 For some extremely weird reason, it is faster to simply call the
                    962:        assembly-optimized pfield_linetoscr_full routines _twice_ to double the
                    963:        line than calling the new (also assembly-optimized)
                    964:        pfield_linetoscr_full_double routines, at least when writing directly 
                    965:        into the framebuffer. If someone has a good theory why, let me know 
                    966:        about it.
                    967:        Drawing into memory is faster with the _double routines, but not
                    968:        noticeably (about 1%).
                    969:        Consequence: Removed the new pfield_linetoscr_full_double assembly
                    970:        routines again.
                    971:        Add a "--disable-gui" option to configure.
                    972:        Add Samuel Devulder's dithering code. Provide support for mono and
                    973:        16 color X servers.
                    974:        Put an assembly version of DitherLine() in X86.S.
                    975:        New color mode (3) available, tells UAE to use 256 colors and dither
                    976:        the image. This is meant for people who have only a 256 color X server
                    977:        available, but it works with the SVGA version, too. It's much slower
                    978:        than normal 256 color mode, though.
                    979:        Use the color allocation algorithm from the dither code in
                    980:        alloc_colors256(), too.
                    981: 960619 New files: zfile.c and zfile.h, from Samuel Devulder. These provide
                    982:        transparent decompression of various compressed file formats. Enabled
                    983:        by configure only for Unix systems.
                    984:        Eliminate dont_want_aspect. New option "-C", sets correct_aspect. Make
                    985:        video mode 3 640x480, video mode 4 800x600. Honor correct_aspect in
                    986:        320x400, 640x480 and 800x600 mode with SVGAlib.
                    987:        Add fields maxlinetoscr and x_adjust to gfxvidinfo.
                    988:        Rename pfield_slow_linetoscr() to pfield_do_linetoscr(). Use new fields
                    989:        in gfxvidinfo to enable writing directly to the framebuffer in
                    990:        resolutions with <800 horizontal pixels.
                    991:        New function pfield_do_linetoscr_full(), called when sure that full
                    992:        line update is being done.
                    993:        Move some of the usage() information into some target_specific_usage() 
                    994:        functions, defined in the various system-specific files.
                    995:        Support lower resolutions with LOW_BANDWIDTH, too.
                    996:        memcpy() within screen memory is dog slow. Prevent some common cases:
                    997:        1. Use fill_line() to double a line in the border.
                    998:        2. Add a parameter to pfield_do_linetoscr_full() to indicate whether 
                    999:           the line has to be doubled. New functions
                   1000:           pfield_linetoscr_*_double_slow() that simply do a memcpy(), plus 
                   1001:          optimized ones in X86.S that are much faster by avoiding the copy.
                   1002:        Smart update code could get confused by lores modes when
                   1003:        prev_max_diwstop changed from one frame to the next. New variable
                   1004:        frame_redraw_necessary, which acts as a counter. It's initialized to
                   1005:        2 in interlace mode to handle that case properly, too.
                   1006:        Move sprite priority checks into pfield_sprite_?(), move HAM
                   1007:        calculations into new function decode_ham6() to simplify the various
                   1008:        linetoscr functions.
                   1009:        Fixed a bug that would cause sprites to get deformed in some cases
                   1010:        when they overlapped (never seen it).
                   1011: 960618 New file: dos-snd.c, from Gustavo.
                   1012: 
                   1013: 960617 Release 0.6.0
                   1014: 960617 Fix memory clobber in X86.S when diwfirstword==0. Eliminate double test
                   1015:        for HAVE_SYS_MOUNT_H in sysdeps.h.
                   1016: 960616 Handle programs that don't wait for the blitter by finishing a pending
                   1017:        blit as soon as a blitter register is written to (regardless of DMACON,
                   1018:        to be safe).
                   1019:        Call gui_led only if the LED changed.
                   1020: 960615 Fixed dual playfield sprite priorities. Optimized dual playfield code
                   1021:        in X86.S. More work on AGA code. Get joystick status only in
                   1022:        vsync_handler to speed things up.
                   1023:        Make blitter fill more intelligent with table lookups instead of loop.
                   1024: 960614 DOS update from Gustavo. Pass real bitplane pointer to
                   1025:        bpl_data_differs(), don't calculate it again.
                   1026: 960610 Fixed silly INTEL_FLAG_OPT bug: The new __asm__ expressions use
                   1027:        regflags, which is of course _regflags on a.out...
                   1028:        More BeBox fixes from Christian.
                   1029:        Make the CIA bank bigger (A0..BF) and relax the CIA A/B selection
                   1030:        checks to make broken software (Kingsoft's Emerald Mine, reportedly)
                   1031:        happy.
                   1032:        Default framerate is now 1.
                   1033: 960603 New BeBox code from Christian. Fix test for ELF in configure.in.
                   1034: 960529 genblitter.c: generate blitfunc as a function taking ULONGs and
                   1035:        returning an ULONG. This makes custom.o about 4500 bytes smaller on the
                   1036:        x86 by removing worthless size prefixes, and should increase
                   1037:        performance (better for the cache, and the P6 reportedly doesn't like 
                   1038:        16 bit instructions anyway).
                   1039:        memory.c: Load Kickstart in one block and swap bytes after that if
                   1040:        necessary (eliminates the need for the special bebox load routine)
                   1041:        custom.c: Some preliminary work for AGA support.
                   1042: 960528 Saw a real awful piece of code that detects the right mouse button by
                   1043:        reading a value from POT0DAT, counting up to 25, reading from POT0DAT
                   1044:        again, and deciding the button is pressed if the value changed. Urgl!
                   1045:        Added POT0DAT() in custom.c ...
                   1046:        Gustavo's latest patches.
                   1047: 960525 Do index pulse handling in CIA_hsync_handler(). Make sure bit 12 in 
                   1048:        INTREQ gets set on an index pulse. Don't call DISK_Index() for now,
                   1049:        since this breaks the Workbench.
                   1050:        The CIA code should _probably_ not be resetting the bits in INTREQ.
                   1051:        Disable this as an experiment, let's see what breaks.
                   1052: 960524 Z flag handling for ABCD/SBCD/ADDX/SUBX was wrong (must be clear if not
                   1053:        zero, was set if zero). Some other flag bugs in these functions fixed.
                   1054: 960523 gfxlib.c update from Christian Schmitt. Some new Mac code from Ernesto,
                   1055:        still not all of it, though. Italian keyboard from Alessandro Bissacco.
                   1056: 960522 Fix the CHK instruction. Implement middle mouse button.
                   1057: 960515 Fix some Nextstep problems in configure.in.
                   1058: 960513 New Makefile target "streifenfrei" to remove all output from configure.
                   1059:        Use autoconf 2.7; 2.9 is buggy.
                   1060: 960512 Remove BIG_ALIGNMENT hack. Remove superfluous pfield_fetchdata() call
                   1061:        in pfield_doline_slow_l() that resulted in garbage on the screen.
                   1062:        Move blttbl array from custom.h to genblitter.c
                   1063:        Use new lookup tables imm8_table and areg_byteinc in CPU emulation
                   1064:        instead of ?: expressions.
                   1065:        GCC appears to generate better code if the opcode is passed as ULONG to
                   1066:        the CPU emulation functions instead of as UWORD.
                   1067:        Modify some of the flag emulation __asm__ statements so that GCC uses
                   1068:        less registers for them.
                   1069: 960510 Cast malloc() results to char *, for C++ compilers.
                   1070: 960509 Use test for GCC and __i386__ in options.h to decide whether to use
                   1071:        intel flag optimization.
                   1072:        Use test for wish4.0 to decide whether to use the GUI.
                   1073: 960507 New bebox.cpp from Christian. AF sound patches from Marcus Sundberg.
                   1074: 960505 Fix the configuration stuff so it works. Fix some bugs in gfxlib.c
                   1075: 960504 Use GNU autoconf scripts to generate Makefile and sysconfig.h. Add new
                   1076:        files sysdeps.h that includes the right system header files. Remove
                   1077:        amiga.h and Makefile.
                   1078: 960502 More Next code from Ian (new Uae.app directory, 2 bit drawing for
                   1079:        custom.c).
                   1080: 960501 Allow the sound driver to set the frequency to a value that is slightly
                   1081:        different from the requested one.
                   1082: 960430 Added AmigaOS port from Olaf Barthel.
                   1083: 960427 Fixed single playfield sprite priorities in X86.S.
                   1084: 960426 Added new NeXTwin.m from Ian.
                   1085:        Turn off the blitter delay in line mode.
                   1086:        Added a CPU optimization from Olivier Raoul: gencpu now reads a file
                   1087:        "insncount" that can be generated by UAE during execution and generates
                   1088:        specific functions for the most common opcodes. Not documented yet.
                   1089: 960414 Reimplemented the fast disk emulation in a more sane way. Also it's
                   1090:        only done for reads now. (Plus, it no longer works :-/ Need to work
                   1091:        some more on this).
                   1092:        In blitter emulation, defer not only the setting of the bltfinish bit,
                   1093:        but actually defer the whole blit for the number of cycles it will
                   1094:        take. This makes programs the clear the screen while it is still being
                   1095:        displayed look much better.
                   1096:        Changed supervisor mode handling around a bit (was messy).
                   1097:        New DOS code from Gustavo.
                   1098: 960412 Removed smart update method 2. Added Alessandro Bissacco's blitter
                   1099:        optimization patch.
                   1100:        Fixed some sprite problems: during hysnc, clear sprpos and sprctl, and
                   1101:        don't do_sprites() in line 0.
                   1102: 960409 Be a little more intelligent about when to call pfield_linetoscr() from
                   1103:        pfield_doline_slow(). It's only necessary if a color register changed.
                   1104:        If it can be deferred to the end of the line, use the the _full
                   1105:        variants.
                   1106:        Some CIA TOD fixes.
                   1107:        Better transdisk, accepts first/last track parameters.
                   1108: 960405 New function memcmpy(), inline assembler for the i386. Compares two
                   1109:        memory blocks and copies them at the same time. Used in
                   1110:        bpl_data_differs.
                   1111: 960404 French keyboard from Tristan Cavelier. Parallel port emulation from
                   1112:        Stefan Reinauer.
                   1113:        Turn bpl?pt and r_bpl?pt into arrays, simplify some code using them.
                   1114: 960401 Some parts of Ernesto's new Mac sources.
                   1115: 960331 filesys.c: Close filedescriptors in free_key(). get_path() sometimes
                   1116:        read uninitialized bytes.
                   1117:        Minor cleanups all over the place.
                   1118:        unchain_mnn() leaves sibling intact, but clobbers next. customreset()
                   1119:        didn't care about that and crashed.
                   1120:        Use the __asm__("intel_flag_lookup") magic not only for DOS.
                   1121:        Support 320x300 for X (without LOW_BANDWIDTH).
                   1122:        Add new function fast_memcmp(), defined as memcmp if __i386__ is
                   1123:        undefined, otherwise a slightly optimized inline assembly function.
                   1124: 960329 Some bug fixes in gencpu.c
                   1125: 960328 Stupid programs write odd playfield modulos into the hw registers...
                   1126:        Eek! Sometimes, linedescr[] was indexed with a negative value... Add a
                   1127:        vpos >= minfirstline check to pfield_may_need_update(). I really need
                   1128:        to rewrite this in Oberon...
                   1129:        Don't turn off SHM when LOW_BANDWIDTH is selected.
                   1130:        Keep the nextpos and linepos variables from pfield_doline_slow_?() as
                   1131:        globals and initialize them in pfield_may_need_udpate().
                   1132:        Modified alloc_colors256() some more.
                   1133:        Try to make sprite DMA follow the Amiga DMA slots more closely by
                   1134:        adding a currhp parameter to do_sprites() and calling it at the end
                   1135:        of a scanline or in pfield_may_need_update().
                   1136: 960327 Added Gustavo's latest DOS code, as far as possible. Modified X86.S to
                   1137:        use a macro SYM() to create symbol names either with underscore or
                   1138:        without.
                   1139:        Fixed some very silly SVGAlib lores bugs. Move screen 8 pixels up in
                   1140:        low resolution.
                   1141:        Use diwstop instead of diwstart to caclulate the DIW for 320x200 modes.
                   1142:        Reset it not only in calcdiw(), but also in init_frame().
                   1143:        Speed up 256 color allocation a little bit.
                   1144: 960326 New function flush_block() in xwin.c and svga.c. Called from
                   1145:        do_flush_line() and (new function) do_flush_screen(). Modify struct
                   1146:        vidbuf_description, new member maxblocklines that records the maximum
                   1147:        number of scanlines that do_flush_line() should collect before calling
                   1148:        flush_block(). If it is zero, then flush_line() is used as before.
                   1149:        This makes the special case stuff for SHM unnecessary, so it's removed.
                   1150:        Fixed a stupid bug in events code: Put ev_hsync in front of ev_copper,
                   1151:        so that do_copper gets called with hpos == 0 instead of hpos == 227 at
                   1152:        the beginning of the line.
                   1153:        New function init_frame() that sets up some variables, called from 
                   1154:        customreset() and vsync_handler().
                   1155:        Move smart update code to pfield_doline() from pfield_doline_h() and 
                   1156:        ..._l(). Clean up these two with a macro DO_ONE_PLANE, removing
                   1157:        duplicate code.
                   1158:        Implement big blits and ECS chip ids. New configuration options
                   1159:        ECS_AGNUS and ECS_DENISE.
                   1160:        Implement MOVEP.
                   1161: 960325 Before calling pfield_may_need_update(), make sure the value that was
                   1162:        written to a register actually differs from the previous one.
                   1163:        New expansion code from Stefan Reinauer.
                   1164:        Modify struct mem_notify_node to contain a pointer to a valid bit 
                   1165:        instead of the valid bit itself, so that the playfield code has to
                   1166:        check only one flag to decide whether to draw the line (SU method 2)
                   1167: 960324 The chip_flags approach has several problems. A flag may cover an area
                   1168:        in memory that spans two scanlines, or several scanlines may use the
                   1169:        same area in memory.
                   1170:        Provide two other, different methods for the smart update:
                   1171:        1. Save all the bitplane data for each line and compare the chipmem
                   1172:           with the saved data to decide whether an update is necessary.
                   1173:        2. Use a chain of memory_notify_node structs instead of a single bit
                   1174:           for flags. Each scanline owns a set of these structs and chains them
                   1175:          into the right place, and the chipmem_?put routines walk the chains
                   1176:          and set the modified bit for each element.
                   1177:        Unfortunately, both methods are slower than the old one. Especially the
                   1178:        first one has some advantages though: it speeds up programs that clear
                   1179:        the screen each update and write the same contents back, and it
                   1180:        can speed up double-buffering programs at full framerate, since it does
                   1181:        not have to check whether the bitplane pointers have changed.
                   1182:        Remember the first and last drawn line for each frame, and pass them to
                   1183:        flush_screen. Use this in xwin.c for SHM. Modify SHM logic: if smart
                   1184:        update is selected, call XShmPutImage() from flush_line() instead of 
                   1185:        from flush_screen(). But gather blocks of consecutive lines to avoid
                   1186:        calling XShmPutImage() too often.
                   1187: 960323 "Smart update" method, based very loosely on Alessandro Bissacco's
                   1188:        work, implemented:
                   1189:        Add an array chip_flags that records any change made to specific words
                   1190:        in chip memory.
                   1191:        Move all the hardware registers relevant to playfield display into a
                   1192:        new structure bplinfo. Remember the contents of this structure in
                   1193:        struct line_description for every line on the screen. Avoid redrawing
                   1194:        in pfield_doline if the screen memory has not been modified for the
                   1195:        line to be drawn and if the information from the last frame matches
                   1196:        that in the current frame.
                   1197: 960322 Duplicate all functions responsible for drawing (pfield_doline,
                   1198:        pfield_doline_slow, pfield_sprite, pfield_*word_*res) and append _h
                   1199:        or _l (ah, the golden C++ days, when we had templates... which never
                   1200:        worked). Modify the *_l functions to support drawing in low resolution
                   1201:        (i.e. 320x200). Change the options -d and -H to support selecting color
                   1202:        mode and resolution, thereby getting rid of some SVGALIB_* options.
                   1203:        Get rid of the old pfield_*word_*res functions.
                   1204:        Collect information about the minimum diwstart value during the
                   1205:        previous frame and use that in case 320x200 was selected to fit the
                   1206:        image on the screen.
                   1207:        Get rid of HAVE_JOYSTICK (breaks the "-J" option)
                   1208: 960321 Give up in op_illg() if opcode is 0x4e7b and the exception handler for
                   1209:        it is NULL. In that case, it's a 68020 Kickstart.
                   1210: 960319 New Next code from Ian. SunOS target was still screwed (two CFLAGS).
                   1211:        Turn -x option into a "hair cursor" option. Support multiple mounts,
                   1212:        but warn if more than one is seen (still does not work with Kick 1.3).
                   1213:        Integrated Samuel Devulder's new LOW_BANDWIDTH patch.
                   1214: 960318 Integrated the start of Markus Gietzen's gfx library replacement.
                   1215:        Changed the interface between custom.c and the graphics code. There has
                   1216:        to be a buffer holding the data of all the lines on the screen. For
                   1217:        linear SVGAlib modes, this is the video memory. Eliminated -x option,
                   1218:        prepare_line(), DrawPixel() and the X_?BIT_SCREEN options, replace with
                   1219:        3 versions of pfield_linetoscr for different datatypes. Implement
                   1220:        proper handling of the borders above and below the DIW in custom.c
                   1221:        instead of in flush_screen().
                   1222:        Move common code (color allocation) from all the graphics files to new 
                   1223:        file gfxutil.c.
                   1224:        Swedish keyboard from Per Olofsson.
                   1225: 
                   1226: 960315 Release 0.5.0
                   1227: 960315 Newest DOS port from Gustavo. New sgi-gcc target.
                   1228: 960314 Add -lXext to osf target. Use "rb" for fopen() in main.c
                   1229: 960310 Play a bit with the __asm__ constraints. Implement CHK.
                   1230: 960307 Fix some stupid Makefile errors. Call parse_cmdline() if no init file
                   1231:        is found.
                   1232:        First round of NextStep patches; new file NeXTwin.m
                   1233:        Stupid bug in custom.c: The #define is called NO_FAST_DISK, not
                   1234:        NO_FAST_FLOPPY.
                   1235:        Fix one of the problems the Deform demo had with the blitter: The
                   1236:        line texture wasn't rotated to the right position in blitter_init().
                   1237: 960306 Use some black GCC __asm__ magic to calculate flags more quickly
                   1238:        (config INTEL_FLAG_OPT)
                   1239:        HAM was broken by pfield_linetoscr changes. Fixed.
                   1240:        LOW_BANDWIDTH sometimes got confused when the display window changed.
                   1241: 960305 BCD instructions work well enough for Locomotive Basic on CPE.
                   1242:        New function parse_cmdline_and_init_file() that reads ~/.uaerc or
                   1243:        ./.uaerc and adds the options found in those files to those in argv[].
                   1244:        Patch from Denis Sablic to make disk image filenames runtime 
                   1245:        configurable. Extend it for the ROM filename.
                   1246:        Remove some of the run-time configurable stuff from config.h.
                   1247:        More X keysym fixes (I finally got a proper .Xmodmap ...)
                   1248:        for SVGAlib: Delete key, and PgUp/PgDn as Amiga keys.
                   1249:        The sound driver can now handle 22050 Hz and/or 8 bit samples. Both of
                   1250:        these are the default with LINUX_SOUND_SLOW_MACHINE.
                   1251:        Also use the SNDCTL_DSP_GETBLKSIZE ioctl and write the buffer when 
                   1252:        there is enough data.
                   1253:        Abuse keybuf.c to provide a fake joystick emulation with the numeric
                   1254:        pad (enable with -J)
                   1255: 960304 bebox.cpp from Christian.
                   1256:        Rename pfield_linetoscr_full to pfield_linetoscr_full8.
                   1257:        More x86 assembly: pfield_linetoscr_full16.
                   1258: 960303 SunOS target from Dirk Vangestel. Put some not-so-superfluous includes
                   1259:        back in. Added the DOS specific patches to filesys.c.
                   1260:        Started x86 assembly optimizations. New file X86.S, includes an
                   1261:        optimized version of pfield_linetoscr (for 8 bit screens only).
                   1262:        Don't use SA_RESTART if it isn't defined.
                   1263:        prepare_line() wasn't called often enough (again...), leading to memory
                   1264:        corruption and "hit the reset button" bugs when using SVGAlib.
                   1265: 960226 Get rid of bool, true and false in amiga.h and everywhere else.
                   1266:        Copy yesterdays code to AF_SOUND code, too.
                   1267:        Include first round of Christian Bauer's BeBox patches. Add AIX fix
                   1268:        from Samuel Mackrill.
                   1269:        Add #include <string.h> and #include <stdlib.h> in some more places.
                   1270:        Modify gencpu.c and readcpu.c so that immediate types j and J (0..15
                   1271:        and 1..8) can be merged. Replace HAVE_ONE_GIG... option with a new
                   1272:        CPU_EMU_SIZE option that can be set to a value in the range 0..5,
                   1273:        where 0 is minimum size and 5 is equivalent to HAVE_ONE_GIG...
                   1274: 960225 Remove multiplications in sound code, use a big lookup table.
                   1275: 960222 Go back to having only one memory bank table of type addrbank instead
                   1276:        of several function pointer arrays. Eliminate the HAVE_BROKEN_SOFTWARE 
                   1277:        define by extending the memory bank table to 65536 entries.
                   1278:        Remove remaining DUALCPU code. Allow BTST src,imm instructions in
                   1279:        table68k.
                   1280: 960220 Back out filesys.c change from Stefan Reinauer: don't call V36 specific
                   1281:        functions in expansion.library.
                   1282:        Better fix for the filesys.c assign problems by Oliver Moehrke.
                   1283:        Make new playfield code conditional on EXPERIMENTAL_SCREEN_UPDATE.
                   1284:        Improve the 64bit variant of said code. Try to get alignment for
                   1285:        apixels by putting it into a union with a long.
                   1286:        Kill the child in gui_exit(). Make graphics_init() return a value and
                   1287:        check it in main.c to see whether graphics were initialized. If not,
                   1288:        call gui_exit() anyway.
                   1289:        Use new 256 color allocation routine from svga.c in xwin.c, too.
                   1290:        Make LOW_BANDWIDTH compile _and_ work (wow...).
                   1291: 960219 Fix bug in memory.c: When loading 256K Kickstart, copy 256K, not 128K.
                   1292: 960218 Finished the rest of the UI: reset, debug and quit buttons as well as
                   1293:        drive LEDs. New target linux-gui; all others use dummy file nogui.c
                   1294:        for now.
                   1295: 960217 The GUI now supports inserting and ejecting diskfiles and displays
                   1296:        their names. Two pipes are used now, one for reading and one for
                   1297:        writing. Properly wait for the process to start up.
                   1298:        Use sigaction() instead of signal() in debug.c.
                   1299:        More fixes for the new display routines: Need to do a final call to
                   1300:        pfield_*word_*res with the data shifted right from the last fetched
                   1301:        word.
                   1302: 960216 Copper & sprite bugfixes from Marco Nelissen. Better X keyboard support
                   1303:        from Bruno Coste.
                   1304:        Move my_strdup() from filesys.c to options.h
                   1305:        New files xui.c, uae-ui (shell script) and uae-ui.tk (wish script).
                   1306:        These create a GUI for the X version of the emulator. So far, the
                   1307:        power LED works.
                   1308: 960215 Don't rely on SVGAlib headers defining any keycodes at all.
                   1309:        Include Thorsten Frueauf's NetBSD patch. Apparently, the devname
                   1310:        variable in hardfile.c caused a namespace collision, so rename it to
                   1311:        uaedevname. Fastmem fixes, debugger help, and ConfigDev allocation in
                   1312:        hardfile and filesys from Stefan Reinauer.
                   1313:        xwin.c: Give the window a title.
                   1314:        Half-implemented the DIVU.L, DIVS.L, MULU.L and MULS.L instructions.
                   1315: 960213 AF sound system support from Marcus Sundberg. Win95 keyboard support
                   1316:        for SVGAlib from Thierry Lescot. Added most of Gustavo Goedert's MS-DOS
                   1317:        port.
                   1318:        When using the old pfield_*word_*res functions, the gen_pfield_tables()
                   1319:        function was undefined.
                   1320:        Hmmm, /dev/brain must have been unmounted when I wrote the new pfield_*
                   1321:        functions. Now they make sure they access the apixels array only with
                   1322:        addresses that are multiples of 16 bytes.
                   1323: 960212 Fix copy&paste bug in cia.c found by Marco Nelissen: Need to call
                   1324:        RethinkICRB() instead of RethinkICRA() for CIA B timers.
                   1325:        Added Stefan Reinauers autoconfig area/fastmem support. Added run-time
                   1326:        options for use_slow_mem and use_fast_mem.
                   1327:        Oops, the pfield_*word_*res() optimizations break completely on HPs and
                   1328:        Sparcs. Now use this optimization only if we are using GCC 2.7.
                   1329:        HPUX fixes in the Makefile. New target hpux-gcc.
                   1330:        The code for clearing the areas above and below the DIW was trying to
                   1331:        be a little too clever selecting the color. Just use the current 
                   1332:        background color.
                   1333:        Added channel attachment to the sound code. Untested.
                   1334: 960211 Found some flag setting bugs in the rotate/shift instructions and in
                   1335:        addx/subx (which I thought I had already fixed) by reimplementing 
                   1336:        DUALCPU mode to work with Ed's e68k.
                   1337:        Argh! The copper is completely disabled during DDF in Hires 4 bitplane
                   1338:        mode - so copper positions 0x30 and 0x34 are about 640 pixels apart.
                   1339:        The Kickstart uses FF39FFFE to wait for the end of line 256 instead of
                   1340:        FFDFFFFE as in all the other modes. This is implemented now, but
                   1341:        probably not 100% correct.
                   1342:        Added check whether SVGAlib doesn't support our mode.
                   1343:        If using GCC >= 2.7, define __attribute__((aligned(16))) as
                   1344:        BIG_ALIGNMENT in config.h. Use it for the apixels array to prevent
                   1345:        potential problems on the Alpha.
                   1346:        Tried to make the color selection routine produce better results in 256
                   1347:        color mode.
                   1348: 960210 readdisk didn't work at all for large files - fixed.
                   1349:        In pfield_doline(), clear the apixels array if playfield DMA is off.
                   1350:        In SPRxCTL(), reset sprite mode when sprpos==sprctl==0
                   1351:        Rewrite pfield_orword_hires() and friends to use lookup tables. No
                   1352:        noticeable improvement on my machine, but might be different on other
                   1353:        machines. Two versions of this code: one using 64bit longs on the Alpha.
                   1354:        New targets linux-debug and linux-prof.
                   1355:        Use 1MB slow mem.
                   1356:        -funroll-loops is a little excessive. Remove it.
                   1357:        Add support for linear addressing modes for SVGAlib version. Speedup
                   1358:        is 20-30% on my old 486 VLB. Now, would someone please fix SVGAlib to
                   1359:        support that for the S3-864 as well?
                   1360:        Set SVGAlib keyboard mode to DONT_CATCH_CTRLC.
                   1361:        Change custom.c, xwin.c and svga.c to keep track of the display window 
                   1362:        and background color and to clear the areas above and below the display
                   1363:        window if necessary to remove garbage images from previous frames.
                   1364:        Set the LOF bit in the Kickstart replacement init routine. Also
                   1365:        initialize CIAs to Kick 1.3-like values.
                   1366: 960209 Some more -DHAVE_xxx_H options in the Makefile and filesys.c. DEC Alpha
                   1367:        fixes for filesys.c from Ed and Marcus. HPUX fixes.
                   1368:        New field in table68k for CPU type, parsed by readcpu and compared to
                   1369:        CPU_LEVEL which is defined in config.h. New instructions MOVE2C, MOVEC2
                   1370:        and CAS. Also a new field for priviledge level that is used by gencpu to
                   1371:        generate tests for regs.s and Exception(8) calls if necessary.
                   1372:        Move parsing of table68k into a new file build68k.c. Generate a file
                   1373:        called cpudefs.c containing the information from table68k in 
                   1374:        machine-readable form. Change readcpu.c to use the array found in 
                   1375:        cpudefs.c.
                   1376:        Don't generate the d8(An,Dn) addressing mode in-line. Generate a call to
                   1377:        get_disp_ea(), which is an inline function in newcpu.h if UAE is
                   1378:        configured for 68000 emulation; for 68020 emulation, it is defined in
                   1379:        newcpu.c.
                   1380:        Debugger could crash on illegals because there was no mnemonic defined
                   1381:        in lookuptab[].
                   1382:        Return zero for SERDATR.
                   1383: 960207 It seems that the copper stops completely if an attempt to write to a
                   1384:        specific register fails because of COPDANG. Apparently, the Kickstarts
                   1385:        use this to stop the copper with a 0x00000000 longword. This leaves the
                   1386:        question why they don't do 0xFFFFFFFE like everybody else.
                   1387:        Enable 512K slow memory by default.
                   1388:        Integrate Stefan Reinauers battclock patch that makes it work with
                   1389:        newer Kickstarts.
                   1390:        In cia.c, do the little index pulse hack even if dskdmaen==1, because
                   1391:        DMA is only enabled when dskdmaen==2 (Helps the "Mad Decision" demo).
                   1392:        Add solaris entry to Makefile (GCC, Openwin, -DHAVE_STATFS_H)
                   1393:        Define HAVE_STRINGS_H for SGIs, change filesys.c accordingly.
                   1394: 960206 Build a smaller table instead of cputbl.c. Use readcpu and the smaller
                   1395:        table to build the cpufunctbl array at run-time.
                   1396:        Nuked the other half of the old CPU emulator. Reimplemented debugger to
                   1397:        use the information generated by readcpu.
                   1398:        Define HAVE_STATFS_H for SGIs in the Makefile, and use that as a test
                   1399:        in filesys.c instead of #ifndef __linux.
                   1400:        Hack filesys.c, get_path() to support the assign command.
                   1401:        Add -fstrength-reduce and -funroll-loops in Makefile to make people
                   1402:        nervous.
                   1403:        Rewrite Sam Devulder's LOW_BANDWIDTH patch and add it to xwin.c.
                   1404: 960205 Forgot some PC-relative bit instructions in table68k.
                   1405:        Nuked half of the old CPU emulator.
                   1406:        Split gencpu.c into two parts, new one is called readcpu.c and just
                   1407:        parses the table68k file. Link it to the main program.
                   1408: 
                   1409: 960205 Release 0.4
                   1410: 960203 filesys.c, action_read(): Slightly more efficient code (translate Amiga
                   1411:        address to real pointer).
                   1412:        Moved some common code in the generate_* functions in gencpu.c to a
                   1413:        separate function.
                   1414: 960202 Added an experimental fast disk option. Currently turned off by
                   1415:        default (it's not such a big win).
                   1416:        Attached sprite fixes (overlapping att. sprites looked bad, Katakis).
                   1417:        Add sleep(1) before resetting the console to text mode when using
                   1418:        SVGAlib: this might fix some screen corruption problems.
                   1419:        Add sprite/playfield priority checking to the most important case
                   1420:        (single playfield, no HAM).
                   1421:        In filesys.c, do_find(): open() returns -1 on error, not zero.
                   1422:        Return ERROR_OBJECT_WRONG_TYPE if do_find() is called for a directory
                   1423:        (fixes Champions of Krynn harddisk installation).
                   1424: 960201 Don't abort if sound driver not present, just set produce_sound to 0.
                   1425:        New files keybuf.c and keybuf.h to record keypresses in the right
                   1426:        order and without losing any. In cia.c, force 15 scanlines between
                   1427:        keypresses, just to be sure.
                   1428:        unixfs.device _does_ work with Kick 1.3: just don't trust what Kick 1.3 
                   1429:        sends in the startup packet. For now, disable more than one mount per 
                   1430:        command line.
                   1431:        Started integrating Ernesto's new Mac sources.
                   1432:        Remove superfluous includes from some files.
                   1433: 960131 Added Ed's unixfs.device (great stuff).
                   1434:        Adding ULONGs to pointers is a bad idea on the Alpha if the ULONG value
                   1435:        really is signed. Add some casts to LONG in (pc_p + src) expressions
                   1436:        in genpu.c.
                   1437:        If DMACON is written and copper DMA is enabled, do a COPJMP1 at once.
                   1438:        Helps the "Interference" demo.
                   1439: 960129 More SGI fixes from Ed. Bugfixes and transdisk improvements from Marcus 
                   1440:        Sundberg.
                   1441:        Remove EXTRA_DEFINES from Makefile. Breaks some systems.
                   1442:        Move common sprite code from pfield_doline() and pfield_doline_slow()
                   1443:        to new function pfield_sprite(). The same sprite may appear more than
                   1444:        once on the same line, so don't shift out the bits of sprdata[] and
                   1445:        sprdatb[] while displaying it (Turrican I).
                   1446:        In xwin.c and svga.c, barf if LINUX_SVGALIB doesn't match the file 
                   1447:        being compiled.
                   1448:        Make all .o files depend on config.h in the Makefile.
                   1449:        No need to exit if sound driver unavailable, but -S given.
                   1450:        Small debugger fix: Missing space in output.
                   1451:        Fix for the sprite logic: Specifically, use a state variable indicating
                   1452:        whether the sprite has been restarted after a VSYNC. Fixes most
                   1453:        Turrican problems.
                   1454: 960124 Added Denis Sablic's patch for sound run-time option.
                   1455:        Added Ed Hanway's patch for better Makefile, X mouse cursor blanking 
                   1456:        and more SGI compilation fixes.
                   1457: 960123 Include options.h everywhere.
                   1458:        Handle 8 bit GrayScale visuals like PseudoColor.
                   1459:        Remove C++ leftovers from joystick code.
                   1460: 960122 When using the joystick driver, the button test must come after
                   1461:        handle_events() in vsync_handler().
                   1462: 960118 Removed all the remaining C++ comments. Changed all inline keywords to
                   1463:        __inline__. Define __inline__ if not using gcc.
                   1464:        Make proper prototypes for everything. Compile with maximum warnings +
                   1465:        -ansi + -pedantic.
                   1466:        Remove CIA_cycle(), obsolete.
                   1467:        Reimplemented the STOP optimization in newcpu.c. Removed DualCPU
                   1468:        support in CPU emulator.
                   1469:        Real nasty bug in pfield_doline() fixed: sprxpos could be evaluated as
                   1470:        negative, with not-so-amusing results. (Need to rewrite this in 
                   1471:        Oberon to get array bounds checking :-)       
                   1472: 960117 Heroic effort: Rewrote the thing in C. This might help fix some
                   1473:        problems with users being unable to compile it.
                   1474:        Fixed a problem in hsync_handler(): Only call flush_line() for lines
                   1475:        in the display window, i.e. when we did a prepare_line() before.
                   1476:        Better code for relative branches: Don't use setpc(getpc()+x) calls,
                   1477:        increment regs.pc_p instead.
                   1478: 960116 Reimplemented the function to load the Kickstart ROM. Use stdio instead
                   1479:        of fstreams since this apparently does not work on the Mac. Detect 256K
                   1480:        Kickstarts. Detect corrupt ROM images (calculate checksum).
                   1481:        Added Ernesto Corvi's Mac port. Changed it around a bit, so it
                   1482:        probably won't compile.
                   1483: 960115 Reinstate config.h options for X screen depth, so that DrawPixel() can
                   1484:        be inlined in custom.cc for speed. xlinebuffer is now incremented in
                   1485:        each call to DrawPixel() (for both X and SVGAlib) to get rid of some
                   1486:        address calculations.
                   1487: 960114 Fixed X generic pixel drawing routine for SHM.
                   1488:        Still trying to fix the harddisk emulation.
                   1489:        uae.device no longer breaks the debugger (can step through uae.device
                   1490:        functions now)
                   1491:        Bugs affecting performance: SPCFLAG_STOP never got reset, and DSKLEN() 
                   1492:        would set SPCFLAG_DISK even if DMA was being turned off.
                   1493:        Made slow memory a run-time option.
                   1494:        Defer interrupts by one CPU instruction to give programs a chance to
                   1495:        read INTREQR ("Seeing is Believing" and "Substance" demos)
                   1496:        Added ScrollLock hack for X, too.
                   1497: 960113 SVGAlib version compiles again. Fixed SVGAlib mouse bug.
                   1498:        Fixed SHM bug: Maximum scanline is 313, not 312.
                   1499:        Sometimes, disk.cc missed a side change and would read the wrong data.
                   1500:        Fixed. Apparently, this was the worst compatibility problem.
                   1501:        Implemented trace mode.
                   1502: 960112 Changed layout of class amigamemory a little so that gcc can generate
                   1503:        better addressing modes.
                   1504:        Finally wrote functions in gencpu to generate MOVEMs.
                   1505: 960109 Integrated Ed Hanway's patches for better X support and run-time
                   1506:        configuration of some options. 
                   1507:        Got rid of the direct VGA memory access. (Need to do this differently).
                   1508:        Changed the method of drawing lines: custom.cc now tells the graphics
                   1509:        code the line number and whether it needs to be doubleed before drawing
                   1510:        it.
                   1511:        Added Andre Beck's MIT-SHM patch.
                   1512:        Remove warnings for newcpu.cc.
                   1513: 960108 Fixed exceptions in op_illg(): Need to decrement PC.
                   1514: 960107 Added an "uae.device" resident module at 0xF00000. This emulates a hard
                   1515:        disk (fixed size 8MB for now).
                   1516: 960106 Moved some common code from pfield_doline() and pfield_doline_slow() to
                   1517:        a separate function. This fixes a potential HAM bug (two static vars
                   1518:        for the same value).
                   1519:        Sound support for Linux. Works only with graphics off and the CPU
                   1520:        slowed down.
                   1521:        Better SVGAlib keyboard support.
                   1522: 960105 Added AvailMem(), AllocMem(), AllocAbs() and FreeMem() dummies.
                   1523:        The Hardwired demo times the multiplication instructions and prints
                   1524:        "This demo don't like Axel" if they are too fast. Apparently, Axel has
                   1525:        a 68040. Added a WANT_SLOW_MULTIPLY option to config.h.
                   1526:        Fixed the fast blitter emulation (seems to work now).
                   1527: 960104 Fixed all the ChangeLog entries from 95 that said 96 (oops??!)
                   1528:        pfield_may_need_update() should check whether bitplane DMA is on.
                   1529:        Added ersatz.cc and ersatz.h. The purpose of these files is to
                   1530:        implement one or two Kickstart functions that are commonly called from
                   1531:        bootblocks. This should help support some games and demos that only use
                   1532:        the Kickstart as an initial track loader. So far, it's only good enough
                   1533:        for one program.
                   1534: 951223 More intelligent event handling in the CPU emulator. Slightly faster.
                   1535: 951222 Optimize CPU emulation by inlining cctrue(). Also, the real PC no
                   1536:        longer needs to be incremented each instruction. The real PC value
                   1537:        now has to be fetched by m68k_getpc().
                   1538:        Added direct screen access for SVGAlib, but it didn't help much. I'll
                   1539:        probably remove it again.
                   1540:        The gencpu executable is 2M smaller if it allocates memory
                   1541:        dynamically.
                   1542: 951216 custom_bput() enhanced a little. Now remembers the value that was 
                   1543:        written in the other half of the register.
                   1544:        Apparently, the USEx bits of BLTCON0 are ignored in line draw mode.
                   1545:        (Silents-Demo)
                   1546: 
                   1547: 951212 Release 0.3
                   1548: 951212 Included Alessandro's battclock patch.
                   1549:        Fixed mouse autocalibration for wrong-aspect screens.
                   1550: 951210 Joystick support for Linux.
                   1551:        Better color matching for 256 color modes.
                   1552:        Rewrote most of the memory management in plain C, mainly to avoid the
                   1553:        various template bugs in some compilers, but also because gcc can't
                   1554:        handle the regparm attribute for virtual functions yet.
                   1555: 951209 Added Alessandro's patches for automatic mouse calibration.
                   1556:        Tried to optimize the blitter loop a bit. Doesn't work.
                   1557: 951205 Added Hannu's patches for diskfile write support.
                   1558: 951201 More portability fixes, especially for weird compilers that workstation
                   1559:        vendors ship.
                   1560: 951125 More keys for X, some keyboard support for SVGAlib (slightly 
                   1561:        dangerous); beginnings of US keyboard support.
                   1562: 951124 256 color mode support for SVGAlib. FAST!
                   1563: 951123 Tiny tweak to central loops in playfield code results in much better
                   1564:        assembly. Apparently, GCC does not feel too good about optimizing 
                   1565:        reference variables.
                   1566: 951113 Bug fix in the events code (nextevent must not be changed while events
                   1567:        are being processed).
                   1568: 951112 The playfield can now be drawn with the right aspect (i.e., every line
                   1569:        twice). Slower, but looks nicer and supports interlaced mode
                   1570:        (flickerfree!). The first 29 lines (blanking interval) are omitted
                   1571:        now to fit everything in 800x600.
                   1572:        Oops, disk drive identification did not work. Fixed, all four drives
                   1573:        appear now.
                   1574:        Oops, CIA_reset() was there all the time but never got called.
                   1575:        Optimized the planar->chunky conversion in pfield_doline() a little.
                   1576:        This really is time critical.
                   1577: 951111 If colors change on a line, we might still be able to use the full 
                   1578:        line update, if the change occurred outside the DIW. Have to be a 
                   1579:        little careful, though.
                   1580:        Playfield code now uses real pointers again, but makes sure it can
                   1581:        read enough bytes to complete the current line without segfaulting.
                   1582:        Oops, my "better estimate" (see below) was a mistake. Restored the
                   1583:        original 4 hw cycles per instruction relation. This slows down the
                   1584:        emulator while booting (more screen refreshs), but speeds up many
                   1585:        demos that otherwise spend most of their time waiting for a specific
                   1586:        raster position.
                   1587:        Playfield modulos must be added immediately when DDF stops. (WOC demo)
                   1588:        Portability fixes in disk.cc (null pointers).
                   1589:        Implemented average frame rate calculation ('c' in the debugger).
                   1590:        Debugger fixes.
                   1591:        Sprite fixes (height was wrong, wrong priorities in fast update)
                   1592:        Added an AutoConfig expansion RAM device. The OS seems to recognize
                   1593:        it, but breaks sometime later (seems to be the same bug that kills
                   1594:        the emulator when there's RAM at 0xC00000).
                   1595:        SVGAlib now has mouse support, too.
                   1596:        HAM mode implemented.
                   1597: 951110 Implemented fast complete line playfield update. Switches back if
                   1598:        necessary. Fixed attached sprites bug.
                   1599:        Copper wait no longer slows down the CPU emulation. (there were
                   1600:        lots of new copper bugs because of this, but I hope I fixed all
                   1601:        of them).
                   1602:        Changed cia?t? from UWORD to ULONG, since there were overflows in
                   1603:        the assertion (because of 65535+1 == 0).
                   1604: 951108 The DIVS remainder could have the wrong sign.
                   1605:        New HAVE_BROKEN_SOFTWARE define in config.h that makes the emulator
                   1606:        mask out the high byte of all addresses.
                   1607:        Wrote genblitter.cc and replaced the awful code in blitter_blit()
                   1608:        and blitter_line() with calls to the generated function.
                   1609: 951107 Implemented ?BCD. These don't really work yet.
                   1610: 
                   1611: 951106 Release 0.2
                   1612: 951105 Fixed a bug in the new CIA timer code.
                   1613:        Fixed flag settings for some cases of ADDX/SUBX
                   1614: 951104 Changed the main loop a bit - call all the *cycle() functions only if
                   1615:        it is really necessary. I hope everything still works - at least it is
                   1616:        twice as fast now :-D
                   1617:        Improved mouse emulation - avoid overflows, permit "wraparound".
                   1618:        Added debugger 'g' command.
                   1619:        Removed some useless casts from the CPU emulator.
                   1620:        Added a configuration option to prevent checks for odd addresses. Only
                   1621:        very bad programs need this check.
                   1622: 951103 The sprite emulation sometimes did not work too well, especially with
                   1623:        dual playfield mode. Hopefully fixed.
                   1624: 951102 The 10 cycle counter is more like a 5 cycle counter in this
                   1625:        implementation. (Fine, that makes floppy loading even faster :-)
                   1626:        Implemented mouse movement. Totally out of sync with the X pointer,
                   1627:        but at least something.
                   1628:        Found out about the XAutoRepeatOff() function. Real useful.
                   1629:        DualCPU mode works again.
                   1630:        Corrected flag settings for CMPA.W.
                   1631:        Fixed DIV instructions: Don't change dest in case of overflow.
                   1632: 951031 The PC is now converted to a real pointer. This is safe (much safer
                   1633:        than converting bitplane pointers, since the PC always has to be valid)
                   1634:        and should speed things up a little.
                   1635: 951029 Some primitive event handling: 2 mouse buttons and 1 joystick button
                   1636:        are emulated, as well as parts of the keyboard (only german mapping).
                   1637:        Fixes: Delay values and sprite widths are always LoRes. Playfield 2
                   1638:        uses colors 9..15.
                   1639:        Implemented a frame rate configuration option that can speed up the
                   1640:        emulation quite a bit.
                   1641: 951028 A-ha. The 68000 has one strange feature after all. MOVEM.W to Dn sign 
                   1642:        extends the value to a longword.
                   1643:        Katakis expects that accessing odd addresses causes exception 3, so 
                   1644:        we have to emulate this. *Sigh*
                   1645:        Do at least something with the index pulse bit. It's more or less a 
                   1646:        toggle now.
                   1647:        Implemented sprites & dual playfields, both with small glitches.
                   1648: 951027 When LOF is set, the frame is one line longer.
                   1649: 951026 Implemented a preliminary diskfile type for special formats. Turrican
                   1650:        boots up correctly (crashes if I want to play a game, though).
                   1651:        Implemented EHB and added a few missing masking operations in the color
                   1652:        handling.
                   1653:        Fixed the absw addressing mode - need to sign extend here as well.
                   1654: 951025 Fixed another extremely stupid bug, this time in the playfield code.
                   1655:        Each pixel was calculated four times...
                   1656:        Increased the size of the apixels array - this was being overrun.
                   1657: 951024 Fixed an extremely stupid bug in flag settings for byte and word
                   1658:        operations.
                   1659:        The "sectors until gap" field must be in the range 1..11 instead of
                   1660:        0..10, or very bad things will happen if it is used to calculate a
                   1661:        blit size.
                   1662: 951021 Don't use real pointers to fetch bitplane data - too many segfaults.
                   1663:        Delay the copper after executing an instruction, or wait sequences like
                   1664:        FFDF FFFE 0200 FFFE won't work.
                   1665:        Some programs expect that the CPU is completely turned off if the
                   1666:        blitter nasty bit is set. So be it.
                   1667:        The RSI Megademo chokes when disk DMA fetches two syncs. Leave one out.
                   1668:        The LOF bit wasn't set correctly.
                   1669: 951020 Blitter fixes: Load fill carry from FCI each line; keep the masked
                   1670:        value of bltadat in bltolda
                   1671:        Give the CPU twice as many "cycles" - this still is not even remotely
                   1672:        correct, but a better estimate.
                   1673: 951017 Added SVGAlib support. Unfortunately, my graphics card isn't fully
                   1674:        supported by the current SVGAlib version.
                   1675: 951015 More small optimizations to memory.h and custom.cc.
                   1676:        Added code for 256 color screens. The type of screen has to be
                   1677:        selected at compile time.
                   1678: 951014 Modified the memory management to be (hopefully) a little more
                   1679:        efficient.
                   1680:        Changed gencpu.cc to generate functions and a big function pointer
                   1681:        array instead of the old case labels. The generated C++ files now
                   1682:        compile a lot faster, and the code should be better as well.
                   1683: 951013 Implemented pseudo POTGOR to keep the RMB from being detected as
                   1684:        pressed all the time.
                   1685:        GenerateDecTab() generated TRAP incorrectly.
                   1686:        Made some more instructions (MOVE.B A0,D0 etc) illegal in gencpu.cc.
                   1687:        Fixed X colors. Fixed data fetching for HiRes screens.
                   1688: 951011 EXT.WL instructions set the flags.
                   1689:        Implemented second CPU emulator, plus "DualCPU" option. The new CPU
                   1690:        emulator is generated automatically from a file called table68k by
                   1691:        a program called gencpu.
                   1692: 951002 Removed version information from the file headers.
                   1693:        Removed some sillinesses in custom.cc. 
                   1694:        DSKLEN bug fixed (mask out 0xC000).
                   1695:        Playfield code rewritten. Now supports horizontal scrolling. Implemented
                   1696:        some primitive color support.
                   1697: 950927 Fixed predecrement/postincrement addressing with byte size for the
                   1698:        stack pointer.
                   1699: 950926 Fixes to the disk code: Terminate disk DMA, flip disk side, read more
                   1700:        than just two sectors during disk DMA, fix checksum calculation.
                   1701:        Fixed copper comparison bugs.
                   1702:        
                   1703: 950901 Release 0.1
                   1704: 950826 bugfix in GenerateDecTab(): LEA.W instructions were generated.
                   1705: 950312 New file disk.cc, for (guess what) disk support. Doesn't work yet.
                   1706: 950311 Multiplication instructions were generating 16 bit results.
                   1707:        Sign extension was wrong in some places.
                   1708:        Removed bugs in the blitter (one word width: ALWM & AFWM and line mode)
                   1709:        The Kickstart logo looks perfect now.
                   1710:        Playfield routine should check BPLCON0 instead of BLTCON0
                   1711: 950309 ASR D8,D0 is not a valid instruction.
                   1712:        Bitplane pointers were not incremented.
                   1713:        Copper comparison didn't work. COP2LC and bitplane pointers were written
                   1714:        incorrectly. Fixed.
                   1715:        Something that bears a striking resemblance to the familiar Kickstart
                   1716:        logo now appears in my X window!
                   1717: 950306 ROR used to set the X flag. Bug fixes in GenerateDecTab.
                   1718:        More bug fixes in several 68000 instructions.
                   1719:        Improved blitter. ?????? KS1.3 doesn't seem to set BLTDMOD on a 
                   1720:        line draw. What is this??
                   1721:        Fixed stupid bugs in playfield code.
                   1722: 950305 fixed bugs in opcode generation and interrupt handling. Implemented 
                   1723:        STOP. Fixed flag settings: Result wasn't masked. Ooops!!
                   1724:        ADDQ #imm,An doesn't set flags.
                   1725:        One-shot timer mode didn't work.
                   1726:        Implemented blitter, doesn't work.
                   1727:        Some primitive and slow playfield support, only to make the things that
                   1728:        hopefully will soon appear visible.
                   1729: 950304 CIA timers ought to work OK. Copper is emulated.
                   1730:        Emulator loops infinitely when trying to initialize the console.device
                   1731: 950303 experiments with X, figured out how to open a window and draw pixels.
                   1732:        No graphics emulation yet, though.
                   1733: 950301 Nearly all instructions work.
                   1734:        Fixed supervisor mode stack.
                   1735:        The memory manager now handles CIA chips.
                   1736: 950228 Bug fixes in 68000 code, exceptions and more instructions implemented
                   1737:        Memory management extended to handle custom chips as well.
                   1738:        Debugger improved.
                   1739:        Kickstart passes memory & CPU type test, builds up some system
                   1740:        structures.
                   1741: 950227 MC68000 & little debugger for tracing. It can execute the first couple
                   1742:        of KS1.3 instructions, but doesn't get past the memory test.
                   1743: 950225 Work begun

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.