The CD-ROM IOCTL direct pass through code now does sanity check on the requested data size before passing the command - fixes crashes with some DMA-only host DVD drives;
The network poller is now in its own thread;
The hack is back in the emulation of the National Semiconductors PC87306 Super I/O Chip - it's the only way right now to have serial working on that board;
Fixed a part of the code that was still using NukedOPL even when OPL 3 was set to DOSBox OPL;
Applied all mainline PCem commits.
SCSI DMA writes (for MODE SELECT) now correctly use InitLength instead of request_length when bus type is SCSI;
Fixed MODE SELECT page saving;
CD Audio volume is now correctly initialized on card initialization for all Sound Blaster variants, fixes CD Audio being muted when booting without the Sound Blaster initialization program;
Brought the Sound Blaster variants back to a sane sound volume but still louder than the old quietness;
Moved SCSI adapter initialization to before sound card initialization, fixes DOS SCSI CD-ROM driver hangs under some conditions.
Attempt to fix sound fluttering in waiting for a better fix from mainline PCem;
Attempt to make the Sound Blaster louder;
Removed excess logging from cdrom_ioctl.c.
Applied all mainline PCem commits;
CD-ROM sound thread is now disabled if all CD-ROM drives are either disabled or have audio disabled;
Fixed the displayed incorrect zero cylinders, head, and sectors per cylinder when loading an already existing HDI image;
Fixed the CD-ROM audio menu item not checking/unchecking correctly.
Floppy disk controller is now COMPLETELY reset on hard reset;
NukedOPL is now optional (but enabled by default) so people on older hardware can gain a few % of performance by going back to DOSBox OPL;
*.CPP files now get all the correct optimization flags applied;
Added NEC PowerMate V emulation - you can get into CMOS SETUP but it hangs before booting, and PCI graphics card don't work (yet).
Fixed a link-breaking bug in cdrom.c;
The AHA-154x now supports commands 0x28 (GET EXTENDED BIOS INFO) and 0x29 (UNLOCK MAILBOX) and no longer reports a firmware revision with scatter/gather bug;
The BusLogic now supports command 0x20 (RESET);
Fixed the GPIO register mapping on the National Semiconductors PC87306 Super I/O chip.
Improved the BusLogic incoming mailbox code;
The BusLogic callback is now three-phases and outgoing mailbox interrupts are emulated correctly;
Fixed the CD-ROM command READ DISC INFORMATION (0x51), fixes NetBSD crashing 86Box with segmentation fault when using the AHA-154x;
Added the CD-ROM command PAUSE/RESUME ALT (0xC2).
Previous CD-ROM sense is now correctly cleared when a REQUEST SENSE command is performed, fixes detection of disc change after setting the drive to empty first regardless of what bus is used for the CD-ROM.
ATAPI CD-ROM word reads no longer incorrectly discard the upper 16 bits, fixes the CD-ROM problems;
Significantly rewrote parts of the AHA-154x and BusLogic emulation, NT 3.1 RTM and MS-DOS drivers now works correctly;
The function in cdrom.c to read data blocks from the disc now correctly advances the position of the sector to read in non-passthrough mode, fixes multiblock reads from directly mounted ISO's;
Both ATAPI and SCSI now always check for Unit Attention and Not Ready on command execution.
- Crash dumps now attempt to be created in the directory with 86Box.exe
- Crash dumps now attempt to find the full path + filename of the module
where the crash occured, and includes it if found
