- Forgot to release a mutex

- Also added IPv6 support to TCP
- Changed IPv4 checksum back to being byte-based

- Padded with zero, instead of Undefined
- Fixed ping to return the ping time, instead of the timestamp

- Mutex protected current TX descriptor to stop two threads getting
  the same TX.
- Changed in use bitfield to a mutex to stop collisions
- Also fixed a spelling error

- Switched to using the PIT for timekeeping
- Fixed multiple IRQ errors

- Just a hack driver really, used as a fallback when no VESA is avaliable

- Also slightly updated spec to confirm how Axis callbacks happen

- Part of some mods to get UDIref ported (not in git)

- Now can do stuff at the CLIShell prompt
- Fixed some loading issues
- Fixed debug in Kernel/VFS/io.c

- Threads waiting on semaphores can now be woken by other threads
- Changed ping to not base its timestamp on after the packet is sent
 > Fixes extra-long timeouts

- Made Log_KernelPanic actually panic
- Allowe BochsGA driver to supprt 0xB0C0
- x86_64 module configuration

Working on IRC Client
- Fixed various networking bugs (firewall failure, etc)
- Updating VTerm implementation to support alternate buffers and
  scroll regions.
 > Scroll is currently broken, doesn't clear the last line.
- Implemented atexit() in libc

- PCI: Cleaned up (don't need to allocate ports, PCI bios does that for us)
- Fixed Endian flipping issues
- Added some debugging and catches to IPStack
- Fixed IPv4 dropping all packets (due to firewall returning -1)
 > Need to create stub rules for INPUT/OUTPUT/FORWARD
- Huge changes in RTL8136 driver
 > Now tested, but not fully

- Implemented (untested) a RTL8139 driver
- Debugging in IPv4/NE2000, trying to trace lack of ARP replies

- Moved IPStack to end of module list (ensures network drivers are inited earlier)
- Debugging in IPStack (ARP)
- Fixing little bugs exposed by verbose errors

- Ends up being far simpler too
- Untester, but should work

- Mouse driver implemented, only had very basic testing
- Decided to use select() and sockets for the AxWin IPC framework
 > Allows the use of select() on the keyboard and mouse for input

- Also fixed some logging bugs (Heap dump and VFS)
- Fixed a bug in the AxWin2 shell include statement

Bug Description:
The last non-idle thread to be put to sleep didn't get the ->CurCPU field
reset to -1. This probably caused the scheduler to think that the process
was still in use, and hence it never got re-scheduled. The other VTs did
not suffer this problem because they were not the last to sleep.

- Moved PCI into two files, one core (with the scanning etc) and one per
  architecture for the IO commands.
- Addded time/pci to the ARM7 tree
- Moved drv/vga.c to a module
- Misc fiddling with ARM port
- Removed Big/LittleEndian16/32 functions, now in acess.h as macros
- Misc changes elsewhere

- Adding a PS2 mouse driver too

- IPv6 fiddling
- Fixed some const-ness