Differences between revisions 10 and 11
Revision 10 as of 2011-07-26 20:15:56
Size: 2326
Comment:
Revision 11 as of 2012-01-25 18:09:39
Size: 3013
Comment: a lot has been unbroken; update a bit on remaining brokennesses I know of
Deletions are marked like this. Additions are marked like this.
Line 23: Line 23:
We’re using gcc-4.6 for almost everything now. The eglibc source package manually selects gcc-4.4, and the Linux kernel is being switched to gcc-4.6 at the moment.
Line 26: Line 28:

Problems:
 * http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49856
Line 36: Line 35:
Ported; not tested much yet. gcj-4.6 [[http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49847|FTBFS]]es. Ported, default compiler. gcj-4.6 [[http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49847|FTBFS]]es, but the bugfix is on the 4.6 branch and will probably make it into the package RSN.
Line 42: Line 41:
It’s probably time for someone to make a patch against the gnat-4.6 package with Mikael’s findings right now; Ludovic says he’s not monitoring it but will gladly accept patches. (Maybe wait until the gcc/gcj-4.6 situation has settled down, though.)
Line 46: Line 47:
All outstanding issues are apparently fixed now. Most outstanding issues are apparently fixed now. We still seem to have [[http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=636286|invalid locale files]] leading to lots of segfaults building some d-i package.
Line 55: Line 56:

== aptitude ==

broken, use apt-get and pbuilder-satisfydepends-classic instead.

This page attempts to track the current issues of the Debian/m68k Linux port. Mailing list is at debian-68k@lists.debian.org and more information can be found in this Wiki under M68k and M68k/Porting.

Installing

Get an Aranym/Quick image (sid from July 2011) and update it. Stephen Marenka’s sid ISO installer image is reported to work as well. The status of everything else is unknown, however, you can probably install the etch-m68k snapshot. But if you wanted to dig out that you wouldn’t be here, right?

Once you have it running, grab a /var/cache/pbuilder/base.cow at M68k/Cowbuilder (already included on the dc11m68k.tgz image).

buildd

Thorsten tries to get an ARAnyM one up. Ingo is upgrading Vivaldi (Amiga). None are running at the moment yet, though (builds are still done manually).

installer

Wouter is working to get d-i into shape again.

packages

Some packages are in debian-ports.org unreleased, not in unstable, because they need m68k-local patches. For some, this is because maintainers keep unstable frozen during the squeeze/testing freeze in order to be able to upload fixes via unstable. For some, this is a collection of patches or workarounds that are not, or will not be, included in the proper Debian packages. Once installed, however, it’s fine to work from unstable only (unless you want to hack the packages from unreleased).

Compilers

We’re using gcc-4.6 for almost everything now. The eglibc source package manually selects gcc-4.4, and the Linux kernel is being switched to gcc-4.6 at the moment.

gcc-4.4

Works, has TLS backport.

gcc-4.5

Not ported. Has TLS upstream. Will not be ported, we’ll be switching from gcc-4.4 directly to gcc-4.6 when the time is right; wheezy will not be released with gcc-4.5 anyway.

gcc-4.6

Ported, default compiler. gcj-4.6 FTBFSes, but the bugfix is on the 4.6 branch and will probably make it into the package RSN.

gnat

Working on it: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48835

It’s probably time for someone to make a patch against the gnat-4.6 package with Mikael’s findings right now; Ludovic says he’s not monitoring it but will gladly accept patches. (Maybe wait until the gcc/gcj-4.6 situation has settled down, though.)

core system

libc

Most outstanding issues are apparently fixed now. We still seem to have invalid locale files leading to lots of segfaults building some d-i package.

klibc

Patches submitted; may or may not be enough (will be revisited).

rsyslog

rsyslog seems broken, but sysklogd works.

aptitude

broken, use apt-get and pbuilder-satisfydepends-classic instead.