xref: /src/sys/arch/mvme68k/docs/TODO

$NetBSD: TODO,v 1.1 2000/12/14 20:30:14 scw Exp $

NetBSD/mvme68k's "ToDo" List (In no particular order)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

These are `difficulty' rated on a scale of 1-10, where `1' is easy.


1.    Flesh out the "Memory Controller" (memc) driver (3)

2.    Enhance the PCC2/MC2 driver to support its "Memory Controller"
      functionality on older MVME1x2 boards. (2)

3.    Support for "Industry Pack" modules (4)
      (Need to come up with an MI interface for this, since various
      other non-mvme68k SBCs have IP sites too)

4.    Switch over to the MI mk48txx TOD-chip driver (1)
      (I already have most of this done)

5.    Support for the NVRAM part of the above chips (2)
      (With a userland tool to read/write configuration details?)

6.    Related to #5, add code to locore.s to verify the configuration
      area of NVRAM is kosher (using a checksum) before using any values
      from it (2)
      (eg. at this time, bogus Offboard RAM details cause crashes)

7.    Support for FLASH memory on MVME1x2 boards (3)
      (Will probably need a userland tool of some sort)

8.    Add support for the in-kernel FP emulator, so we can run on MVME
      boards with the MC680LCx0 CPU variant (4)
      (Easy enough to do; slightly harder to test)

9.    Add "Bounce Buffers" to the VMEbus slave code (8)

10.   Support the MVME boards as a slave in A16 address space (8)
      (To be useful, this really needs "Bounce Buffer" support)

11.   Enhance the MI VMEbus interface to provide generic GCSR access (6)
      (Speak with Matthais Drochner <drochner (a] netbsd.org> about this one)

12.   Enhance the VMEChip/VMEChip2 drivers to support #11 (3)

13.   Fix the case where VMEbus RAM is mapped through a translated
      address space on MVME boards with the VMEChip2 (2)

14.   Add/Flesh out evcnt(9) support where useful (1)

15.   The siop driver *badly* needs to be bus_space(9)'d, and made MI (5)
      (Apparently, Manuel Bouyer (bouyer (a] netbsd.org) is working on this)

16.   The sbic driver *badly* needs re-written from scratch, and made MI (5)
      (This will be done when Frank van der Linden <fvdl (a] netbsd.org>
      imports the PC98 code into the tree; it comes with a very nice
      wd33c93 driver ;-)

17.   Switch the install tools over to sysinst (3)
      (Should be fairly easy, although sysinst doesn't do `tapes', so
      there's a bunch of code to write)

18.   Use hardware-assisted softints (3)
      (What to do on MVME1x2 boards with no VMEChip2?)

19.   Make the ZS driver *properly* MI (4)
      (I already have patches for this. Unfortunately, it affects just
      about every NetBSD architecture...)

20.   Implement the MD parts of __HAVE_SYSCALL_INTERN (2)