Fixed the MODE SELECT (6) command for CD-ROM and SCSI disk;
Implemented REQUEST SENSE with DESC bit set for SCSI disk;
Implemented the SCSI RESERVE, SCSI RELEASE, and SEND DIAGNOSTIC commands for SCSI disk.
Removed the Mach64 ISA until someone finds a properly working BIOS;
Changed the Mach64 VLB to use the BIOS that actually works;
Moved the standalone TI CF62011 graphics card to the Dev branch until someone finds a BIOS for it;
Removed an accidental temporary change in win_dialog.c;
Reorganized the graphics card ID's to allow simplification of the MDA and VGA #define's in video.h;
Added a workaround for the TVGA8900D DAC sense, so the card is in color again.
The UMC UM8669F's reset function now make sure the FDC pointer does not get lost, should fix the segmentation fault on reset;
The Winbond W83877F's reset function now zeroes the registers before setting the default values.
The settings dialog now keeps track of MFM/XTIDE/ESDI/IDE channels and SCSI ID's and LUN's;
The IBM PS/2 Model 50 now installs a Model 80-style memory expansion card (MCA ID FCFF) when more than 2 MB RAM is selected, and now supports a maximum of 10 MB RAM.
Merged floppy.c and fdd.c and renamed floppy_*.c (the floppy image format handlers) to fdd_*.c;
Reading the AT or PS/2 keyboard controller status no longer clears the transmit timeout bit, fixes error 8601 (mouse error) on the IBM PS/2 Model 80;
MMU translate and DMA physical reads and writes now go through _mem_exec instead of directly to ram[], should fix the last remaining problems with remapped mappings;
Implemented the Sound gain dialog;
Added the resource for the "New floppy image" dialog and the needed functions for the functionality of exporting the currently mounted floppy image as 86F, both of which should be finished in the next commit;
Applied the CD-ROM fixes from the PCem commit;
Added the "Keep ratio" option for full screen stretch.
Fixed Sound Blaster 16 and earlier CD-ROM volume registers (and made them no longer default to 0);
Serial mouse now resets the FIFO before sending the ID, fixes mouse detection after clipping and moving the mouse;
Rewritten CD-ROM and SCSI disk Mode Select handling and changed things regarding Mode Sense as well, also fixes emulator segmentation faults when anything attempts to do a Mode Select;
Fixed the Microsoft InPort mouse emulation (used the Bochs code as reference for the polling/state update code), now it actually moves;
The PCjr initialization now sets the XT scan code table, makes the keyboard actually work on the PCjr again;
Several bugfixes regarding the Read CD commands, fixes CD Audio in digital mode on Windows ME;
Finally properly fixed the regular CD Audio MSF/LBA conversion mess;
Applied all applicable PCem commits.
Moved the two 440FX board along with the Pentium Pro and Pentium II CPU's to the Dev branch;
Applied the PCem commit that fixed PIIX IDE Bus Master initialization.
Fixed the behavior of the CD-ROM GET CONFIGURATION command when unimplemented features are requested;
Fixed the behavior of the CD-ROM READ DVD STRUCTURE command in some situations and made it correctly report 05/30/02 for incompatible format;
Reworked the PS/2 Model 80 Type 2 memory handling a bit;
The emulator now allocates the few MB of space needed for pages for the entire 4 GB RAM space at the startup and only memset's it to 0 on hard reset - should make sure DMA page reads from/writes to memory-mapped devices no longer crash the emulator on invalidating the memory range;
Applied app applicable PCem patches;
The PS/1 Model 2133 now also applies PS/2-style NMI mask handling - fixes the 486 recompiler on this machine;
Added the missing #include of "cpu/cpu.h" in io.c, fixes compiling when I/O tracing is enabled.
Applied all applicable PCem commits;
The PS/1 and PS/2 machines now have the correct graphics cards - fixes the reference diskettes;
The open dialog code now changes directory to usr_path after the dialog is closed;
Disabled excess logging in the TI SVGA code.
Re-worked the system pathnames (pc.c), renamed cfg_path to usr_path.
Other small things here and there.
Logitech bus mouse re-worked, should be OK now.
If no VM path is specified or the specified VM path is relative, it is now converted to absolute using exe_path;
Fixed saving path for screenshots when using the DirectDraw renderer.
