tuxonice support (again...)

[Tim Gardner]

Matt Price wrote:
> (CC:ed to a couple of active ubuntu/tuxonice users)
> 
> Dear Kernel Team,
> 
> Now that jaunty is practically out the door, I imagine you're starting
> to think about the karmic kernel.  I would like to urge you once again
> to consider including a tuxonice kernel, perhaps as an experimental
> flavor (though of course I'd love to see it in -generic).  I know it's
> been rejected in the past, but the situation continues to change, so I'd
> like to rehearse the arguments in its favor one more time:
> 
> 1.  Tuxonice is _much_ faster.  For most of the intrepid cycle I used
> unmodified -generic kernel packages, and resume times on my Dell D820
> (Intel CoreDuo T2400, 1 Gig RAM) were abysmal -- over 2 minutes (cold
> boot is also quite slow on this system).  I switched last week to
> tuxonice-patched kernels from nigel's ppa, and the resume from GRUB
> prompt to a highly responsive system with working wireless, running
> evolution, firefox with ca. 30 tabs, amarok, and emacs, now takes just
> over 30 seconds.  Even if karmic boot times are much faster than
> jaunty's, those times would be hard to match without something like
> tuxonice.  
> 2. Tuxonice is now a drop-in replacement for swsusp, and it's supported
> by pm-utils.  When installing the tuxonice kernel, the only changes I
> had to make under intrepid were some minor modifications of the
> initramfs.  
> 3.  Tuxonice no longer relies on the LZF compression algorithm, and
> instead uses LZO, which is already in the kernel.  
> 4.  Though there were some delays earlier in the year, Nigel has now
> released Tuxonice 3.0 and now 3.0.1, fixing the vast majority of bugs
> reported on the tuxonice list.  He is very close to submitting a series
> of patches to linux-next, and there's reason to believe that at least
> some of them might be accepted.  
> 5.  Nigel now maintains not only several ubuntu git branches, but a PPA
> with ubuntu kernels; hopefully this reduces the burden on the kernel
> team when it comes to building and testing tuxonice kernels.  
> 
> The main point is that including tuxonice in ubuntu is now much less
> work than it would have been even six months ago, and also less
> disruptive of non-tuxonice suspend and hibernate protocols.  So I guess
> my question is, what would have to happen to begin to have tuxonice
> included?  And is there anything I or other users could do to make
> things easier?  Though this is a small part of the user experience, it
> can also be a very important one.  
> 
> Thanks for your help,
> Matt
> 
> Some References:
> Nigel's PPA:
> https://launchpad.net/~tuxonice/+archive/ppa
> Using the PPA under Intrepid:
> http://lists.tuxonice.net/lurker/message/20090409.181125.d20e0bbe.en.html 
> Tuxonice Website, with announcements of 3.0 and 3.0.1:
> http://www.tuxonice.net/ 
> Nigel's jaunty tree:
> http://kernel.ubuntu.com/git?p=nigelc/ubuntu-jaunty%
> 2Btuxonice.git;a=summary 
> 
> 
> 

While I appreciate the intent and functionality of Tux on ICE, there is
_no_ chance of including it in this distro kernel until it is upstream
in Linus' tree. I'm all for faster hibernate/resume, but I cannot
support such a large divergence from the upstream kernel source. We
tried that in Hardy (xen, rt, openvz), and its still causing maintenance
issues.

Fortunately, as long as Nigel keeps his PPA up to date, adventurous
users have alternatives.

rtg
-- 
Tim Gardner tim.gardner at canonical.com