xref: /src/doc/roadmaps/ports

$NetBSD: ports,v 1.2 2017/01/13 13:40:44 reinoud Exp $

NetBSD Ports Roadmap
====================

This roadmap covers ports and port-specific issues, and also bus-level
material even if it's not strictly port-specific.

The following elements, projects, and goals are considered strategic
priorities for the project:

 1. EFI boot for x86
 2. xhci support
 3. Get arm64/aarch64 working

The following elements, projects, and goals are not strategic
priorities but are still important undertakings worth doing:

 4. USER_LDT for amd64
 5. riscv and/or or1k ports
 6. cheri port

The following elements, projects, and goals are perhaps less pressing;
this doesn't mean one shouldn't work on them but the expected payoff
is perhaps less than for other things:

 7. pdp10/risc36/odd-bitsize ports


Explanations
============


 1. EFI boot for x86

EFI boot is now often required for new x86 hardware. This is
effectively a mandatory item for -8. Fortunately, nonaka has most of
it done, though it's not yet committed.

 - As of January 2017 nobody is known to be working on this.
 - There is currently no clear timeframe or release target.
 - Contact agc for further information.


 2. xhci support

xhci is also critical for new x86 hardware. Enough has been committed
to be able to use USB on xhci machines; but (AIUI) the USB 3.0
specific features mostly aren't implemented yet. We would like to get
this pulled into -7

 - As of January 2017 ... who is working on this? XXX
 - Contact jakllsch (?) for further information.


 3. Get arm64/aarch64 working

We have some arm64 code but apparently it doesn't really work yet.

 - As of January 2017 nobody is known to be actively working on this.
 - There is currently no clear timeframe or release target.
 - Contact: ? (XXX)


 4. USER_LDT for amd64

The amd64 port is lacking the USER_LDT bits needed to be able to run
Wine. Adding these bits does not seem to be a particularly large job
(and some of the bits are in place already) but it persistently
doesn't get done. Money's been offered in the past, without result.

 - As of January 2017 nobody is known to be working on this.
 - There is currently no clear timeframe or release target.
 - Contact ? (XXX) for further information.


 5. riscv and/or or1k ports

We have some riscv code and a bit of or1k code, but neither is done.

 - As of January 2017 nobody is known to be working on this.
 - There is currently no clear timeframe or release target.
 - Contact matt@ for further information.


 6. cheri port

http://cheri-cpu.org
There are a number of reasons to tackle this; it will serve as a code
quality lever. Also there's already a FreeBSD port to steal from.


 7. pdp10/risc36/odd-bitsize ports

There's been a fair amount of loose talk over the years about doing a
port to a machine that's got 9-bit bytes, or is word-addressed, or
both. The PDP-10 is one such target; it's also been observed that a
more modern architecture would probably be more likely to allow a
vaguely performant FPGA implementation, and something tentatively
called "risc36" was conceived.

This is both a quixotic retrocomputing project and also a quixotic
code quality project: making the NetBSD code base work on either
word-addressed machines or 9-bit/36-bit machines or both would be good
for it. However, it's also a rather large undertaking.