96 lines
4.2 KiB
Text
96 lines
4.2 KiB
Text
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" [
|
|
<!ENTITY base CDATA "../..">
|
|
<!ENTITY date "$FreeBSD: www/en/platforms/ia64/todo.sgml,v 1.5 2003/06/14 03:31:10 marcel Exp $">
|
|
<!ENTITY title "FreeBSD/ia64 Project -- to do">
|
|
<!ENTITY email 'freebsd-ia64'>
|
|
<!ENTITY % includes SYSTEM "../../includes.sgml"> %includes;
|
|
]>
|
|
|
|
<html>
|
|
&header;
|
|
|
|
<hr>
|
|
|
|
<form action="http://www.freebsd.org/cgi/query-pr-summary.cgi" method="get">
|
|
<center>Search the FreeBSD/ia64 PR database:
|
|
<input type="hidden" name="category" value="ia64">
|
|
<input type="hidden" name="sort" value="none">
|
|
<input type="text" name="text">
|
|
<input type="submit" value="Go">
|
|
</center>
|
|
</form>
|
|
|
|
<hr>
|
|
|
|
<h2>What needs to be done.</h2>
|
|
|
|
<p>This page tries to be the starting point for people trying to find
|
|
anything that can be done. The order of the items on this page are
|
|
not strictly an indication of priority, but it is a good indication.
|
|
There are in all likelyhood tasks that are not mentioned here, but
|
|
that should be done nonetheless. A typical example is the maintenance
|
|
of the ia64 web pages... unfortunately.</p>
|
|
|
|
<h3>Becoming a tier 1 platform.</h3>
|
|
|
|
<p>With two releases as a tier 2 platform, it is time to work towards
|
|
becoming a tier 1 platform. This involves tasks as varied as:
|
|
<ul>
|
|
<li>Improve the installation process to take into account that
|
|
there is already a GPT with an EFI partition, including other
|
|
operating systems. The ability to add a FreeBSD entry to the
|
|
EFI boot menu is also a nice thing.</li>
|
|
<li>Port the GNU debugger. It is sorely missed on a development
|
|
machine and required on tier 1 platforms.</li>
|
|
<li>Port the X server (ports/x11/XFree86-4-Server). Not really
|
|
required for tier 1 status, but one cannot truely do without
|
|
if one wants to use ia64 as a desktop machine.</li>
|
|
</ul>
|
|
|
|
<h3>Ports and packages,</h3>
|
|
|
|
<p>A very important task for the success of FreeBSD on ia64 is making
|
|
sure that users have something to run besides ls(1). Our huge ports
|
|
collection has been targeting ia32 for the most part, so it is not
|
|
surprising that there are a lot of ports that do not build or do not
|
|
work on ia64. Look
|
|
<a href="http://bento.freebsd.org/errorlogs/ia64-5-latest/">here</a>
|
|
for the most up-to-date list of ports that fail to build for some
|
|
reason or another. Note that if there are ports depending on one or
|
|
more ports that fail, those are not built and are not counted. A good
|
|
way to help out here is to work on those ports that have a lot of
|
|
ports depending on it (see the "Aff." column in the table).</p>
|
|
|
|
<h3>Sharpening the saw.</h3>
|
|
|
|
<p>There are plenty functions (especially assembly routines) that
|
|
have been written to provide the missing functionality without any
|
|
consideration for speed and/or robustness. Reviewing those functions
|
|
and replacing them if necessary is a good task that can be done
|
|
concurrently and independently from other activity and does not
|
|
necessarily require huge amounts of knowledge and/or experience.</p>
|
|
|
|
<h3>Core development.</h3>
|
|
|
|
<p>On top of the high-level things that do not work or do not exist,
|
|
there is also some rather involved rewriting to be done at the
|
|
foundation and can potentionally affect all other platforms as well.
|
|
This includes:
|
|
<ul>
|
|
<li>Improve UP and SMP stability by revamping the PMAP module.
|
|
The low-level handling of VM translations needs to be improved.
|
|
This involves both correctness as well as performance.</li>
|
|
<li>Basic device drivers such as sio(4) and syscons(4) do not
|
|
work on ia64 machines that do not have support for legacy
|
|
devices. This is a rather big issue, because it affects all
|
|
platforms and may involve rewriting (big) parts of certain
|
|
subsystems. Clearly a task that needs wholesale support and
|
|
coordination.</li>
|
|
<li>Better handling of sparse (physical) memory configurations
|
|
by not creating VM tables that span the whole address space,
|
|
but rather cover the memory "chunks" that are present. We are
|
|
currently forced to ignore memory because of this.</li>
|
|
</ul>
|
|
|
|
&footer;
|
|
</html>
|