Making FirstRib bootable with own initramfs

[rockedge]

very interesting stuff.....I am about to run the latest scripts and work some with what I have set up and running.

I am continuing to use the BIonic64-v8 for these experiments

also a 32 bit system is running what has been created

64 bit machine has 8 gigs RAM and the 32 bit has 4 gigs of RAM

[wiak]

.

[wiak]

.

[wiak]

.

[sc0ttman]

Hi wiak,

if you are good with wpa_supplicant, and want a non X way to setup wifi for users, this post may be of interest to you:

ncurses wifi tools:

http://murga-linux.com/puppy/viewtopic. ... 68#1030568

I kinda got them working in Puppy, but they need a little love..

[rockedge]

Hi wiak

here is my ethernet which is similar on the machines I am using to run FirstRib

Code: Select all

▶—— Network Devices ——◀

Ethernet controller [0200]: Intel Corporation 82578DM Gigabit Network Connection [8086:10ef] (rev 05)
• Kernel Driver: e1000e
• Memory Used: 153.82 KB
• Path: /lib/modules/4.9.96/kernel/drivers/net/ethernet/intel/e1000e/e1000e.ko
• Driver Version: 3.2.6-k
• Description: Intel(R) PRO/1000 Network Driver

I also came close to getting the X server to run and load jwm but some problems still in the way.

I am experimenting full force today and will try out what you get going....once the ethernet connection is up it is going to be pretty nice to set some stuff up

I am running into modprobe errors and can not change into a directory 4.14.rt37 at boot

[wiak]

.

[wiak]

.

[wiak]

.

[rockedge]

I am reading and experimenting and a full report will follow soon.

I found a simple way to modify between boot attempts is to fire up the FirstRib with the mount script or simply add stuff directly to the file system then I run the build initramfs script and then do a reboot. all the components are in the directory which I have as /mnt/sdb1/FirstRib-boot-2.

so I can add by xbps-install or use the Puppy host system to add the files directly and run the OS in an Xnest display to see if the OS woks still.

So far so good....getting the X server up is sooooo close down to one error Can't find screens...which is related to missing firmware I think. But I have done so many things if it starts to work I won't know why exactly.

I have the network components installed as well and now have the Puppy zdrv sfs renamed and in place....lets see what happens

[s243a]

Maybe add an fdrv for the firmware. Of course firstrib needs to be a able to find the fdrv. In puppy you do this by modifying distrospecs and putting the modified distro specs in the Initrd.

[wiak]

Maybe add an fdrv for the firmware. Of course firstrib needs to be a able to find the fdrv. In puppy you do this by modifying distrospecs and putting the modified distro specs in the Initrd.

FirstRib already loads full firmware/modules sfs, and can be connected to Internet (will add network manager later). No woof-CE puppy-type distro specs type complications are required. Builds both 32bit and 64bit system.

wiak

[wiak]

I found a simple way to modify between boot attempts is to fire up the FirstRib with the mount script or simply add stuff directly to the file system then I run the build initramfs script and then do a reboot. all the components are in the directory which I have as /mnt/sdb1/FirstRib-boot-2.

Yes, that's also what I do prior to finally writing new build_firstrib_rootfs script for anything extra I want permanently in firstrib_rootfs. Only the minimum firstrib_rootfs is needed, then mount and add what you want using xbps package manager and then, as you say, build the initramfs.gz using build_firstrib_initramfs script (the initramfs/init produced also loads full firmware_modules sfs, or small firmware_modules sfs if you wish).

I purposively keep build_firstrib_rootfs and build_firstrib_initramfs as seperate scripts. After all, for run-in-host-chroot mode you don't need an initramfs at all. And the initramfs(with purpose-made-to-be-modular-init), to a large extent, can be used with any root filesystem, no matter how it is constructed.

Prior to writing each new build_firstrib_initramfs script I sometimes cd into existing firstrib_rootfs and manually edit the init inside there until I have it working as I wish, then write the new build script. However, it usually proves easier for me to simply modify existing init creation part of build_firstrib_initramfs script (inside the script) and then simply run the new build script to create my new initramfs.gz to test functionality of the new init. It's a simple and smooth modular workflow. Simplicity is key.

I also purposively avoid complicated configuration scripts (since they have a language-like-format of their own which would also have to be learned). No need for that. What I am providing, when useful, is a simple plug-in extra code facility (such as used in wex program). These are just alternative or extra shell script fragments (so no new language/format to learn) - allowing great modularity and flexibility.

As I've said, FirstRib by design uses simple non-complicated short scripts that are self-documented via comments. That design philosophy hopefully facilities understanding and use and makes it easy to fork and add complications such as complex logic/config scripts/do-everything-in-one-big-script-logic/whatever (if so desired).

Of course the individual scripts to build rootfs and initramfs can be joined together using seperate simple glue script. Not required at all though.

EDIT: A distro built using the combination of build_firstrib_rootfs and build_firstrib_initramfs falls into the family of distros I've called WeeDog (though WeeDog can be as full-featured/full-sized as desired). And I intend several flavours of FirstRib (not just void linux repo/pkg-manager using). I don't like restrictive concept of 'official' distro and 'unofficial' distro used by Puppy community, so there will never be an 'official' WeeDog, though there is a WeeDog home website under construction and github dev sites for both FirstRib components. Whatever bootable WeeDog or WeeDogs I produce will be available, in some shape or form, via the WeeDog homesite, but I will also add links/info regarding WeeDog variants constructed by others.

My next priority (following completing net connect manager) will be switch_root version, which will use very small initial firstrib_rootfs followed by switch_root into big-as-you-like root filesystem. Aside from that, FirstRib can already be built into any shape and form (mini-distro/full-distro) the user wishes. I will eventually put together my own FirstRib-based distro recipe.

wiak

[wiak]

But I have done so many things if it starts to work I won't know why exactly.

Yes, it is always hard to keep track of things done when experimenting. That's why making build scripts is tricky since hacking system together isn't easily repeatable but until we hack something to work lots of what we hack isn't relevant in the end...

However, doesn't really matter. If build scripts kept modular, simple to understand, and to the basics, individuals can add whatever facilities they manage, and eventually a community of knowledgable users can help others with the likes of starting X and so on. HowTos/Tips/Recipes are very useful (and to be encouraged) of course, and pre-built isos of final builds can be usefully published too.

We don't need no gate-keepers, stewarts, or special access status/authority for FirstRib. That's another part of the FirstRib/WeeDog philosophy.

wiak

[wiak]

I'm working on the addition of ethernet connectivity (usually eth0) to initramfs script (along with getting wifi connected).

With busybox udhcpc this is pretty easy and there are many howto's for that available, if you want to try it yourself, such as:

https://mohammadthalif.wordpress.com/20 ... -in-linux/

EDIT: Forgot to say, you can find busybox udhcpc example scripts here:
https://git.busybox.net/busybox/tree/examples/udhcp

A full Void Linux install, however, doesn't use udhcpc but uses dhcpcd instead as decribed here:

https://wiki.voidlinux.org/Network_Conf ... mmand_line

Give either a go! I'll publish scripted version later.

wiak

[rockedge]

I was able to start an X server and JWM with xterm with startx command.
So I built firstrib..mounted it and added Xorg and util-linux and JWM, ROX
umount then used the build initramfs.gz and booted.

but I have no keyboard or mouse after X starts...working on that then I will go for ethernet connection to the router

I sometimes have problems when booting to mount everything correctly
and the path after boot completes I am in /layers.

I must do a complete rebuild from scratch and make a recipe for getting the X server going. Eventually with better understanding the OS can be further optimized in size and knowing what "modules" to add to get this work.

So far success in 64 bit but struggling to get the same going in 32 bit.....but a fresh start in the morning will help!

[wiak]

.

[rockedge]

moving forward on the 64 bit but attempting the same set up on 32 bit the boot process is stopping at the message "Booting the kernel"

so I am concentrating on the 64 bit firstrib right now.

I am not sure why after successfully starting the X server I can not get the keyboard or mouse to work. I am trying all kinds of things but as usual I am missing some small detail.

anyway the desktop looks great and I would be working on connecting with the ethernet if I could get the keyboard and or mouse input. I have to do a complete hard shutdown mechanically then reboot to get out of the X server.

I think the set up I am aiming for can be achieved by using the firstrib.plug file to let the script install the Xorg and JWM, ROX and base-system when building firstrib then go directly into the script building the initramfs.gz

Once I finally get this xorg thing going I can go on ahead to trying out the most minimal set ups to the most complete

Good stuff.

[wiak]

.

[wiak]

.