OK, here are the steps to make a installer/recovery disk and install F13
on Yeeloong netbook.
First, get the debian6 Loongson recovery disk from bbs.lemote.org.
For example, 20100627.tar.bz2, untar it to the root of a USB disk.
Second, get Fai Wong's F13 rootfs. For example,
fedora-13-n32-rootfs-20100710.tar.gz. Rename it to
loongson2-fedora-13-n32-rootfs-20100710.tar.gz and put to the root of
the same USB disk as above. For saving space,
loongson2-debian6-*.tar.lzma can be deleted.
Then you get the recovery/installer disk for Yeeloong. Plug the disk
into yeeloong before it powers on, and press Tab on PMON loading, then
follow the screen instructions, the recovery procedure starts. Caution:
all data on the disk of Yeeloong will lost during the recovery.
After the installation is completed, the Yeeloong is rebooted. Press DEL
when PMON is loading to get the PMON> prompt. enter the following:
g console=tty root=/dev/hda2
Hold your breath to wait for the welcome message =)
After Fedora is booted, login to it, and then modify /boot/boot.cfg as
title Fedora 13
args console=tty root=/dev/hda2
Save and reboot. Done.
It's a preliminary installer, I know, far more from perfect. However can
be a start for interested people to work with Fedora-MIPS. Next, provide
them a Yum repository, then we can expect contributors to appear.
 Debian6 quick installer: http://www.bjlx.org.cn/node/751 by Liu
 Fai's rootfs: http://lazyfai.dyndns.org/Fedora-13-rootfs-MIPS/
Thank you for the hard work.
it has been a while since I posted an update to this list. One of the
reasons is that I have moved to China. This is a big change, but also
a big chance. I still need to find my bearings and employment, but now
I also had to time to meet up with Wolfgang from Qi Hardware and
Qi Hardware is a project to do copyleft hardware, meaning that all the
schematics and software are open sourced. One of their platforms is
the Nanonote, which is a small PDA-form factor device running on a
Ingenic X-burst processor. This is a MIPS-compatible processor and has
proven to be able to run Linux. The current model provides too little
memory to host a full Fedora distribution, but later models will use
an improved Ingenic processor. We are investigating if it is possible
to port our distribution to this next processor/platform.
I also had the opportunity to meet up with the people at Loongson as
they are based in Beijing. We had a very open minded talk about our
Fedora vision and what the vision of Loongson is. They have offered us
some machines to aid us in our port and hope we can collaborate on
other levels in the future by providing a stable distribution and
build the Fedora brand in China.
This leaves us with talking about the actual distribution. We had
several test builds (F11 and F12) since we started around March and
are now focusing on getting out F13 first. We hope this won't be a
lengthy process. Currently this is still done on a single machine, but
we will have a rootfs with the F13 build toolchain for the other
machines to speed up the rest of the process. As you can see, we are
working on our build infrastructure and streamlining our build
process. When the test release is done, we will be submitting our
We still have an issue to which we would like your advice. We are
currently building our distribution as N32 for MIPS Little Endian
(mipsel), but this is not reflected in our package naming. Maybe over
time we could consider creating packages for the other ABIs, like O32
and N64, or even creating a Big Endian release. We would need to be
able to indicate this in our generated packages. Just let us hear what
your take on this is. Feel free to forward this message to people you
think can help us in making a good decision.
We would also like to know how people are using the current release
and if they had any good (or maybe even bad) experience with it.
Looking forward to this.
Gerard and the MIPS team
Gerard Braad — 吉拉德
below you can find an email which I received related to creating
system images for a Fedora embedded-like system. It would be a good
idea to discuss this possibility as I am sure we can all benefit for a
standardize process (complete or partly).
Gerard Braad — 吉拉德
Open Source/Linux Consultant in Beijing
---------- Forwarded message ----------
From: Federico Simoncelli <federico.simoncelli(a)gmail.com>
Date: Tue, Jul 6, 2010 at 5:12 PM
Subject: Set of tools for embedded systems
my plan is to build a set of tools to generate and write a system
rootfs to a device (SD, CF, etc...).
The main features I'm focusing on are:
* It should work for any architecture (x86, arm, mips, etc...) to
gather in one project all the experience from each group of developers
* Once you have the rootfs, it must be fast (imagine you have to
produce the largest amount of SD/CF devices possible)
Right now I'm dealing with x86 processors so I'll explain what my
script does for me and I hope in some feedback about what you need to
make it work on your architecture:
1 Partition the device
2 Format the partitions
3 Mount the filesystems
4 Explode the rootfs tar.gz
5 Replace some files (and tuning), eg: fstab (tmpfs, noatime),
sysctl.conf (swappiness), inittab (agetty)
6 Relabel (if selinux is needed) - not yet fully working, any selinux
expert out there?
7 Unmount the partitions
8 Install the bootloader (only grub for now)
Things we should discuss about:
* Can we all decide to use qemu to install the system on a file device
and standardize the generation of the rootfs tar.gz too?
* In task 5 there are a lot of things that are common to everybody
(eg: swappiness, tmpfs) and many things that are hardware related (eg:
ifcfg-ethX), we should split them
* Relabeling is currently not working, I need help
* Do we need to support an other bootloader?
* Can we build a system-restore feature?
Thank you for you attention,