LiveCD without XFCE

Alexander E. Patrakov patrakov at ums.usu.ru
Fri Apr 14 20:17:30 PDT 2006


Jeremy Huntwork wrote:
> Alexander E. Patrakov wrote:
>> Brian Burns wrote:
>>> Also, I thought I'd edit the makefiles for squashfs and the kernel to
>>> upgrade squashfs to 3.0.
>>> Anyone see any possible problems with this?
>>
>> Unionfs/squashfs patches are no longer welcome because both unionfs 
>> and squashfs will go away today (replaced by zisofs and device-mapper).
>>
> 
> Alexander,
> 
> Using the phrase 'no longer welcome' in such a way sounds overly harsh.

Please assume "no longer needed" instead.

> Also, I realize that earlier I said 'let's move to dm' when we were 
> working out the 'Starting init...' bug, but that was if the bug was 
> indeed unionfs which it seemed to be at the time. Since then, I have not 
> heard whether this is unconditionally the case. Has there been a 
> resolution of the problem that I missed? I'd appreciate hearing any 
> explanations you have before the changes are actually made.

There was no resolution of the i586 problem. It is indeed reproducible without 
unionfs, but the systematic analysis isn't done because the laptop doesn't read 
CD-RW disks. And there's virtually no overlap in time when Justin and I are 
present on #livecd.

The problem is definitely hardware-specific and/or compiler-specific (cmov?) 
because it is not reproducible in bochs compiled for i586 emulation, and there 
is a report about a working Pentium.

So far, the layout is:

6.1-3: works, kernel 2.6.11 compiled by gcc-3.4
6.1.1-1: doesn't work, kernel 2.6.12 compiled by gcc-3.4
6.1.1-x: doesn't work, kernel 2.6.11 compiled by gcc-3.4
6.1.1-4 with the kernel config from 6.1-3: doesn't work
6.1.1-4 with initramfs from 6.1-3: discussing in #livecd now

As for the move away from unionfs and squashfs: there is no unionfs for 
linux-2.6.16 except CVS snapshots. I am not going to test these myself and watch 
the CVS commit list for bugfixes (I just expect this to be too much work). I 
won't object if someone else does this, but I will treat any kernel with unionfs 
as tainted until the release is made.

-- 
Alexander E. Patrakov



More information about the livecd mailing list