Print

[NobodyHere]

QUOTE(joesnose @ Jun 17 2011, 02:11 PM)

Finally TCL would be superb on the xbox, I think if it was available for a native install I would be willing to lose my xbmc dash to be quite honest. A dual boot kernel would be special, I too am wondering why this has never been achieved.

I would actually like to install Linux natively to an ext2/3/4 partition, but retain my fatx partitions for Xbox gaming/homebrew.

If there was some way to chainload Grub4Dos, or Syslinux from a Xbox executable (xbe), you should be able to pass any kernel options you need to boot from Linux partitions.

I have a feeling that their might some type of bios issues involved in implementing this though.
http://forums.xbox-scene.com/index.php?showtopic=535384

[ldotsfan]

An update on my attempts with MicroCore Linux:
1. I managed to compile a kernel with DHCP networking working, USB working ( keyboard and flash drive single partition formatted to ext2). The ALSA sound modules were incorporated into initramfs as well. Telnetd was started manually. A native install to sda1 aka the usb flash drive was possible by modifying linuxboot.cfg. Yes, this leaves the FATX partitions alone. MicroCore recognises the persistent storage at the flash drive's ext2 partition (connected at usb 1.1).
2. I attempted to get X going following the TCL wiki and got stuck. TCL/MicroCore uses Squashfs4.0 to compress its application extensions and SquashFS 4.0 isn't supported on a kernel of version 2.6.22. Yes, SquashFS 3.4 and 4.0 aren't inter-compatible due to its limitations of inclusion into mainline kernel tree. Puppy Linux went through the route of the conversion AFAIK.
3. I then attempted to include Squashfs 3.4 patch inside the xbox linux kernel. I manually uncompressed a test XVesa.tcz app with Squash4.0 's unsquashfs, recompressed with squashfs 3.4 and a mount -t squashfs -o loop on the resultant file yielded an empty directory. Unless I can figure out what went wrong, I cannot use any of TCL's app from its repository and this leaves a barebone initramfs which is entirely busybox only. This is still useful for stuff like chimp2618/lpartx/LUD though.

I will look at Slitaz instead as it supports cromfs which is FUSE (userland) based and doesn't have to use squashfs 4.X's patches in-kernel which is messy with an old kernel like 2.6.22. Alternatively Puppy Wary 5. But even Puppy Wary uses a more recent kernel than 2.6.22.

I will post updates if I make any progress - and if I can find time to work on this



 

[NobodyHere]

QUOTE(ldotsfan @ Jun 20 2011, 02:41 PM)

A native install to sda1 aka the usb flash drive was possible by modifying linuxboot.cfg. Yes, this leaves the FATX partitions alone. MicroCore recognises the persistent storage at the flash drive's ext2 partition (connected at usb 1.1).

Nice. I've only tried the Cromwell loader xbe included with the Xdsl distro, and I haven't flashed it to my bios. I assumed it would work with booting any of the other Xbox distros if there were installed frugally, or through loopback, to one of the fatx partitions. Does this mean I can use it to boot from Ext2/3/4 partitions on the internal Xbox hdd? When I used the xboxhd softmod install method it automatically used all the free space for drive "f:". Can I resize/recreate f: somehow, and then create a Linux partition for install on my Xbox's hdd instead of an usb drive?

QUOTE(ldotsfan @ Jun 20 2011, 02:41 PM)

I attempted to get X going following the TCL wiki and got stuck. TCL/MicroCore uses Squashfs4.0 to compress its application extensions and SquashFS 4.0 isn't supported on a kernel of version 2.6.22. Yes, SquashFS 3.4 and 4.0 aren't inter-compatible due to its limitations of inclusion into mainline kernel tree. Puppy Linux went through the route of the conversion AFAIK.

That sucks, but sound about right. Good luck with the SliTaz testing. I realize you're doing all the heavy lifting, but I'm sure any of the people who've responded in this thread would be more than happy to help in any way we can.

Peace.

[ldotsfan]

I'm using Gentoox Loader which is Cromwell derived but it shouldn't matter. I'm using the XBE version as well so there's no need to flash any bios.

In my x-dsl remake thread at http://forums.xbox-scene.com/index.php?showtopic=713355, I attempted to make x-dsl work with the xbpartitioner partitions after F/G, ie hda57+. I cannot recall the details now but http://web.archive.org/web/20081121045826/...ki/Boot_Process and http://web.archive.org/web/20081011162714/...ki/Partitioning should get things going.

Also take a look at this archived post, http://blog.gmane.org/gmane.linux.ports.xb.../month=20061201

CODE

Darren Wilkinson | 19 Dec 16:46
sector 0 and sector 1 on the hard disk (esp for Torne Wuff)
Darren Wilkinson chipsugar.plus.com>
2006-12-19 15:46:10 GMT

I am thinking of using the hda=remap option in my linuxboot.cfg in order
to obtain that wonderous thing: co-existing ext2/3 and fatx F: partition
tables using a 2.6.x kernel on my 80gb drive.

AFAIK the remap option just maps anything on sector 1 to where it would
be on sector 0. This leaves sector 0 free to hold the LBA48 patched
partition table used by xbpartitioner which I can create a "dead" space
where the linux partition is and use the rest for fatx. Everywhere I
read tells me the xbox ms kernel doesn't use sector 0 but nowhere tells
me if it uses sector 1 in any way.

I don't want to ruin my current xbox games setup though so I want to
know does the xbox use sector 1 for any purpose whatsoever?

For Torne Wuff - due to your knowledge about oz_paulb's partition table
and your interest in porting it to kernel 2.6.x you seemed the person to
ask. I hope you don't mind.

Thanks to anyone for any help.

Any resizing of f is destructive on the data in the partition.

I'll deposit a file dump to sourceforge of my work at the next logical milestone if anybody else wants to have a go at it as well.

[NobodyHere]

QUOTE(ldotsfan @ Jun 20 2011, 03:47 PM)

In my x-dsl remake thread at http://forums.xbox-scene.com/index.php?showtopic=713355, I attempted to make x-dsl work with the xbpartitioner partitions after F/G, ie hda57+.

Yeah. I did find the latest versions of Xdsl were from you, but they wouldn't allow me to do a loopback install along with some other minor issues I can't remember at the moment. That's why I was testing some things out with stevewal's last version.

QUOTE(ldotsfan @ Jun 20 2011, 03:47 PM)

I cannot recall the details now but http://web.archive.org/web/20081121045826/...ki/Boot_Process and http://web.archive.org/web/20081011162714/...ki/Partitioning should get things going.

Thanks for the links, and here's another for those interested.
http://web.archive.org/web/20090212143638/...Cromwell_Manual

I personally use Grub4Dos (itself derived from legacy GRUB) on my PC to dual-boot, so I suppose that the cromwell xbe loader automatically uses the equivalent of grub4dos's "find --set-root". Can I also assume that means that while I can't manually specify "extn" partitions to boot from, the cromwell loader can find kernels located on them?

QUOTE(ldotsfan @ Jun 20 2011, 03:47 PM)

Also take a look at this archived post, http://blog.gmane.org/gmane.linux.ports.xb.../month=20061201

Okay this is pretty much what I was referring to, but I don't see if he reached any type of resolution in his research.

QUOTE(ldotsfan @ Jun 20 2011, 03:47 PM)

Any resizing of f is destructive on the data in the partition.

Heh. I suppose it would have been to much to hope for a Gparted with Fatx support.  (IMG:style_emoticons/default/ph34r.gif)  I'll probably look into the XBpartitioner program then after ftp'ing my F contents out.

QUOTE(ldotsfan @ Jun 20 2011, 03:47 PM)

I'll deposit a file dump to sourceforge of my work at the next logical milestone if anybody else wants to have a go at it as well.

Cool.

This post has been edited by NobodyHere: Jun 20 2011, 06:03 PM

[ldotsfan]

Lpartx/chimp2618 derives the partitioning logic from xbpartitioner. Source code is available at sourceforge.

Yes, cromwell should be able to load the kernel file from ext2 native partition based on the code but I haven't tried it myself.

[sender604]

hi havve you seen this

http://wiki.openwrt.org/toh/xbox/info#runn....with.linux-2.6

kernel 2.6.22.7

[NobodyHere]

I'm not quite sure what Xbox Linux needs is an updated kernel with fatx support. What would be better would be not needing a hard-drive of over 127GB in size to install a native Linux to the 'G:\' drive or above. I should be able to make the 'F:\' drive whatever size I want (say 4GB), and install Linux to the G:\ partition which takes up the rest of my 120GB hard-drive.

I'd say the biggest failure of Xbox Linux right now isn't lack of fatx support in the kernel 2.6.n, but not having a competent bios/bootloader. People should be able to partition whatever size F:\ they want, and install Linux (natively) to their G:\, H:\, I:\, J:\, K:\, etc. drives. The bios/bootloader should be able to boot from a ext2, or ext4 (with noatime), partition eliminating the need for fatx support.

[luther349]

127gb is the limits of fatx but hackers enable g to get passed that. fatx is important being most have duel boot boxes.  the issue is we are patching old kernels and modern distros move away in 6 months ago to the next big thing. lets not forget they all use nevu now and need to make a new video driver.

[megapapo]

Hello

I came across this thread while investigating the current state of the xbox
linux project because my xbox router really needs an upgrade. I am still running
some distribution derived from debian sarge, running 2.4.30. May I ask what are
the primary goals of your efforts? By the way, I have an experimental but apparently
working implementation of a FATX module for recent kernels.
   
Regards

Matt

[luther349]

these have more power and would make a better server then a xbox and use 5 watts of power and use less space. https://www.globalsc...r-standard.aspx

ou can use a usb to vga adaptor so you get a display. or get the version with built in outputs.