squashfs-lzma “linux-live-script”

[TraperJohn]

Hi I'm new to TC Linux and I have a number of questions.
First I come from Slax and I see that you have some posts about squashfs-lzma format and I was wondering if that works on TC yet?
Second I'd like to know if TC uses “linux-live-script” like Slax, Backtrack and Puppy do?
If the above are true then I can provide a wide number of scripts to make the most of them.
I know that other people may have asked this before but I didn't see any defined answer.

[roberts]

No. We don't use the linux-live scripts from SLAX.
We also rejected squashfs-lzma as too big an impact on boot times, especially with older and/or embedded systems. We use the busybox/initramfs as the boot structure. As described at the busybox and embedded linux sites. Everything else is either tarred over or mounted & symlinked into place. We support ziofs/cramfs and soon, next release, squashfs. A very simple structure results in very fast boot times and with great extensibility.

[TraperJohn]

OK simple solution and response but I believe that squashfs is already on the mainline kernel so we could use “.sfs” at least if we wonted.
I know that mounting squashfs-lzma files takes longer then tar, initramfs and ziofs/cramfs do but if we only used the  squashfs-lzma with the activation script off linux-live-script then boot time would not be affected only the individual application mount time would be.
I think that the benefit of having such a small size for applications/modules to be great enough to give it a try.
I'd definitely suffer for an additional 2 or 3 seconded per application of extra waiting time if I had even only 25% more application to choose from.
As for linux-live-script needed to use activation/deactivation I don't see it having that much affect on boot time, I have set it up under other distros before and its not even as complexed to build as the kernel patches for lzma,
So if anyone can get TC to use squashfs-lzma I will gladly build linux-live-script into a TC-core iso.

[Jason W]

I built squashfs-lzma for TC 1.x, and it could be integrated into a person remaster to use squash-lzma extensions.  It is in the extension area.  Same as regular squashfs for TC 1.x, all one would need to do include the module in the iso and modprobe it in tc-config.  As stated, squashfs will be a supported tcz format along with cramfs and zisofs in TC 2.x since squashfs is in base TC 2.x.
However, also as stated, squashfs-lzma will not be a supported extension format as it incurs a performance loss that is not found with standard squashfs. 

[TraperJohn]

As stated, squashfs-lzma will not be a supported extension format as it incurs a performance loss that is not found with standard squashfs.

Not supported, but will it work?

[Jason W]

An extension could be made to be used on one's own system, simply by replacing the standard squashfs module with the squashfs-lzma one.  Then one can convert their persoanl collection of tcz's in to squashfs-lzma tcz's. 

[TraperJohn]

simply by replacing the standard squashfs module with the squashfs-lzma one.

OK where do I find the standard squashfs.tcz module in the iso or do you refer to a kernel module?

[Jason W]

squashfs-2.6.26.tcem is the kernel module and squashfs-tools.tce is the tools, they are both in the 1.x tce extension area for TC 1.x. 

squashfs-tools-4.0.tce is the tools for TC 2.x, and the kernel module is included in the base system.

[TraperJohn]

So replace squashfs-tools-4.0.tce with the squashfs-lzma one, modprobe it in tc-config then start converting tcz's in to squashfs-lzma tcz's.
Sounds good.

[Jason W]

Actually there is not a squashfs-lzma-tools.tce for TC 2.x, only for 1.x.  And the 1.x version will not work in 2.x. 

[TraperJohn]

All right so how do we build one for TC 2.x?

[Jason W]

Squashfs-lzma is not supported for squashfs 4.0, but when lzma makes it into the kernel there may be patches made for it:

http://www.nabble.com/squashfs-4.0-lzma-support-td22269948.html

Squashfs 4.0 is the version in the kernel and the one used by TC.  A squashfs-lzma 3.4 module could be built, but that would preclude the use of regular squashfs-4.0 tcz's, which is now a supported extension format, once that module is loaded. So it would be better not to have a squashfs-lzma extension in the download area.

Of course, even when or if lzma makes it into the mainline kernel gzip will still be the compression method preferred by TC.

[TraperJohn]

Any other high compression algorithms that will work with squashfs 4.0 in a simullar fashen to lzma with 7z like results?

[jpeters]

Interesting comparison study

http://odzangba.wordpress.com/2009/03/25/gzip-vs-bzip2-vs-lzma/

My vote is for speed, although bzip2 seems like a good compromise. lzma is in a league all it's own for
slowness.     

[TraperJohn]

http://odzangba.wordpress.com/2009/03/25/gzip-vs-bzip2-vs-lzma/ is right but We wouldn't need to compress applications more then once as we only are mounting them with activate.
O and by the way in my own mount tests with zip to lzma I see only a small diff about 3-4 seconds on my last gen P4 system.

I found some 7zip compression algorithms written to work with gzip on sourceforge.net but 0 that exist in the kernel.
I'm not sure as to how to search this one out.
Any one know what to look for?